Ecology of Plants PDF

PAGE PROOF: 2ND PASS
C H A P T E R
1 The Science of Plant Ecology
T
he biological science of ecology is the study of the relationships between
living organisms and their environments, the interactions of organisms
with one another, and the patterns and causes of the abundance and dis-
tribution of organisms in nature. In this book we consider ecology from the
perspective of terrestrial plants. Plant ecology is both a subset of the discipline
of ecology and a mirror for the entire field. In The Ecology of Plants, we cover
some of the same topics that you might find in a general ecology textbook,
while concentrating on the interactions between plants and their environments
over a range of scales. We also include some subjects that are unique to plants,
such as photosynthesis and the ecology of plant-soil interactions, and others
that have unique aspects in the case of plants, such as the acquisition of
resources and mates.
Ecology as a Science
Ecologists study the function of organisms in nature and the systems they
inhabit. Some ecologists are concerned in particular with the application of
ecological principles to practical environmental problems. Sometimes the
distinction between basic and applied ecology becomes blurred, as when the
solution to a particular applied problem reveals fundamental knowledge about
ecological systems. In both basic and applied ecology, the rules and protocols
of the sciences must be rigorously followed.
What ecology is not is environmental or political activism, although ecol-
ogists are sometimes environmental activists in their personal lives, and envi-
ronmental activists may rely on ecological research. Ecology is not about one’s
feelings about nature, although ecologists may have strong feelings about what
they study. Ecological systems are complex things, with a great many parts,
each of which contributes to the whole in different ways. Nevertheless, ecolo-
gy is indeed a science, and it works like other scientific disciplines.
How do we know whether something is true? Science is one way of know-
ing about the world—not the only way, but a spectacularly successful one. In
contrast to some of the other ways of knowing that are part of our lives, the
legitimacy of science is based not on authority, or opinion, or democratic prin-
ciples, but on the weight of credible, repeatable evidence.
2 Chapter 1 PAGE PROOF: 2ND PASS
The Genesis of Scientific Knowledge

Observations
Throughout this book we examine how ecolo- and data
gists have come to their current knowledge and
understanding of organisms and systems in
nature. Ecology has both a strong and a rich the- Speculate. Apply inductive
oretical basis and has developed from a foun- and deductive reasoning
to observations. Compare
dation based on an enormous collective store- with current theories.
house of information about natural history.
Ecology, like all of science, is built on a tri-
pod of pattern, process, and theory. Patterns con- Formulate hypothesis
sist of the relationships between pieces or enti- (often phrased as a question)
ties of the natural world. Processes are the Reevaluate
observations
causes of those patterns. Theories are the expla- and theory
nations of those causes. When ecologists carry Predict results assuming
out original scientific research, they seek to hypothesis to be correct
document patterns, understand processes,
and ultimately to put together theories that More predictions,
explain what they have found out. further experiments, Design experiment(s) to test
further development validity of predicted results
There is a distinction between the kind of theory
of research a scientist does and the kind
of research done for a term paper, or by
Results support Results do not
any member of the public trying to gather hypothesis (predictions support hypothesis
information about a topic using library books or confirmed) (“null hypothesis”)
material posted on Web sites. Although there are
exceptions, research carried out by students or
the general public is usually secondary research: Independent verification
gathering data or confirming facts that are of results by other researchers:
“reproducible results”
already known. This sort of research is not only
useful, it is essential: every scientific study must
begin by assessing what is already known. But the heart Figure 1.1
of what research scientists do is primary research: gath- The scientific method. The cycle of speculation, hypothesis,
and experimentation is circular, as new questions constantly
ering information or finding out facts that no one has ever emerge from the answers scientists obtain.
known before. That experience of discovery is what
makes doing science so incredibly exciting and fun.
Scientists gain knowledge by using the scientific
method. They carry out a series of steps, although not The word “theory” has a very different meaning in
always in a fixed order (Figure 1.1). In ecology, these science than it does in common usage. A scientific the-
steps can be summarized as follows: observation, ory is a broad, comprehensive explanation of a large
description, quantification, posing hypotheses, testing body of information that, over time, must be support-
those hypotheses using experiments (in a broad sense of ed and ultimately confirmed (or rejected) by the accu-
the word, as discussed below), and verification, rejec- mulation of a wide range of different kinds of evidence
tion, or revision of the hypotheses, followed by retest- (Table 1.1). In popular usage, the word “theory” usual-
ing of the new or modified hypotheses. Throughout this ly refers to a limited, specific conjecture or supposition,
process, ecologists gather various kinds of information, or even a guess or “hunch.” Equating the meaning of a
look for patterns or regularities in their data and pro- scientific theory with “a guess” has caused no end of
pose processes that might be responsible for those pat- mischief in the popular press and in public debates on
terns. They often put together some sort of model to help politically charged issues.
in advancing their understanding. Eventually, they con- When a theory is buttressed over many years with
struct theories, using assumptions, data, models, and the strong evidence, with new findings consistently sup-
results of many tests of hypotheses, among other things. porting and amplifying the theory while producing no
The building of comprehensive scientific theories pro- serious contradictory evidence, it becomes an accept-
ceeds simultaneously from multiple directions by ed framework or pattern of scientific thought from
numerous people, sometimes working in synchrony and which new speculation can spring. This is what has
sometimes at cross-purposes. Science in operation can occurred with Newton’s theory of gravity, Darwin’s the-
be a messy and chaotic process, but out of this chaos ory of evolution, and Einstein’s theory of relativity.
comes our understanding of nature. Scientists use such overarching theories to organize
PAGE PROOF: 2ND PASS The Science of Plant Ecology 3
One reason the models often

Table 1.1 The components of a theory rely on mathematics is that
ecologists are so often con-
Component Description cerned with the numbers of
Domain The scope in space, time, and phenomena addressed by the theory things (for instance, is a species’
Assumptions Conditions or structures needed to build the theory population size so small that it
Concepts Labeled regularities in phenomena is becoming endangered? How
Definitions Conventions and prescriptions necessary for the theory to work rapidly is an invasive species
with clarity spreading? How many species
Facts Confirmable records of phenomena can coexist in a community,
Confirmed generalizations Condensations and abstractions from a body of facts that have and how does this change as
been tested conditions change?). Mathe-
Laws Conditional statements of relationship or causation, or statements matical models offer well-de-
of process that hold within a universe of discourse fined methods for addressing
Models Conceptual constructs that represent or simplify the natural world questions in both qualitative
Translation modes Procedures and concepts needed to move from the abstractions and quantitative terms.
of a theory to the specifics of application or test
All models are necessari-
Hypotheses Testable statements derived from or representing various
components of the theory ly based on simplifications
Framework Nested causal or logical structure of the theory and rest on a set of assump-
tions. Those (implicit and ex-
Source: Pickett et al. 1994.
plicit) simplifications and
assumptions are critical to
their thinking and derive additional predictions about recognize because they can alert you to the limitations
nature. of the model, and because faulty assumptions and unjus-
A scientific hypothesis is a possible explanation for tified simplifications can sink even the most widely
a particular observation or set of observations. A hypoth- accepted or elegant model.
esis is smaller in scope than a fully developed theory.
Hypotheses must be testable: they must contain a predic- Objectivity, Subjectivity, Choice, and Chance
tion or statement that can be verified or rejected using sci- in Scientific Research
entific evidence. Experiments are the heart of science, and When you read a typical scientific paper, it may at first
we discuss their design and use in more detail later in this seem arcane and dull. The format follows a rigid proto-
chapter. A crucial characteristic of science is the need to col, designed for efficiently conveying essential infor-
revise or reject a hypothesis if the evidence does not sup- mation to other scientists. Ideas are tightly packaged,
port it. Science does not accept hypotheses on faith. with a clear logical line running from start to finish. It
Some of the most important tools in the scientist’s may seem as if the researchers knew exactly what they
toolkit are models. A model is an abstraction or simpli- would find even before they began. We will let you in
fication that expresses structures or relationships. Mod- on an open secret: That is not usually how real science
els are one of the ways in which the human mind works. The justifications for the research presented in a
attempts to understand complex structures and rela- paper’s introduction may have been thought up or dis-
tionships, whether in science or in everyday life. Build- covered long after the research project began, or even
ing a model airplane from a kit can tell you a lot about after the work was finished. Because of serendipitous
the basic form of an airplane; civil engineers often build discoveries, laboratory or field disasters, or unusual nat-
small models of structures such as bridges or buildings ural occurrences, the original purpose of a research proj-
(either physical models or three-dimensional images on ect is sometimes modified or, occasionally, entirely dis-
a computer) before construction is begun. You have no carded and replaced with something else.
doubt seen models of DNA and of chemical reactions, Ideas in science, especially in ecology, come from a
and you may have heard about global climate models, variety of sources. While everyone knows that science
which we will discuss at length in Chapter 22. is objective and rational, that is only half the story. In
Models can be abstract or tangible, made of plastic order to reach a genuinely new understanding, subjec-
or words. They can be diagrams on paper, sets of equa- tivity and creativity must also come into play. While one
tions, or a complex computer program. In science, mod- must be objective in, for example, examining the weight
els are used to define patterns, summarize processes, and of evidence in support of a hypothesis, subjectivity plays
generate hypotheses. One of the most valuable uses of a subtle but important role throughout all of scientific
models is to make predictions. Ecologists deal almost research. What one chooses to study, where one choos-
exclusively with abstract models that can range from a es to look for answers, and what is or is not a valid topic
simple verbal argument to a set of mathematical equations. for scientific research are all subjective decisions scien-
tists must routinely make. To a large extent, these choic- If an experiment is conducted in a field setting, it is
es depend on the questions one asks. While determin- more realistic or more natural, but now many factors
ing the answers must be objective, choosing what ques- may vary in an uncontrolled fashion. In a field experi-
tions to ask, and how to ask them, is largely subjective. ment, the only factors that are controlled are the ones
Many scientific endeavors are highly creative as being studied. Instead of attempting to control all vari-
well. Coming up with a good experiment, looking at a ation, variation due to factors other than the experi-
seemingly intractable problem from a new perspective, mental ones is randomized among replicates, and con-
switching gears after a disastrous laboratory failure to clusions are based on the use of statistical inference (see
extract a successful outcome from the jaws of catastro- the Appendix). Such experiments can be carried out in
phe, and pulling a large number of disparate facts many settings and are not restricted to the field. This sort
together to build a comprehensive theory are all highly of experiment was first developed by R. A. Fisher in
creative activities. the early twentieth century. Fisherian experiments are a
Many scientific discoveries start with casual obser- mainstay of ecology and evolutionary biology as well as
vations, such as Newton’s proverbial apple. Or an idea the social sciences. They are typically less narrowly
may arise as a “what if” thought: What if the world defined than Baconian experiments, and thus their
works in a particular way? Or a previous experiment results may be more readily generalized. Where along
may have raised new questions. What makes a scien- this continuum of control versus realism the ecologist
tist successful is the ability to recognize the worth of carries out her experiment depends on both her scien-
these casual observations, what-if thoughts, and new tific goals and practical considerations.
questions. From these sources, an ecologist constructs Experiments are usually designed as tests of hypothe-
hypotheses and designs experiments to test them. ses. If the hypothesis is partially or wholly falsified, the sci-
entist goes back, revises his ideas, and tries again. If the
Experiments: The Heart of Research hypothesis is not falsified by the outcome of the experi-
A cornerstone of the scientific process is the experiment. ment, the scientist gains confidence that his hypothesis
Ecologists in particular use a wide variety of types of might be correct. Sometimes, however, experiments are
experiments. We use the term “experiment” here in its done without an explicit hypothesis in mind. These are
broadest sense: a test of an idea. Ecological experiments “poke-at-it-and-see-what-happens” experiments designed
can be classified into three broad types: manipulative, nat- to find out more about the world. Such experiments are
ural, and observational. Manipulative, or controlled, common throughout the biological sciences, including
experiments are what most of us think of as experiments: ecology. Scientists have studied in detail only a few hun-
A person manipulates the world in some way and looks dred of the quarter-million terrestrial plant species; of
for a pattern in the response. For example, an ecologist these, only a few [Zea mays (maize), Arabidopsis thaliana,
might be interested in the effects of different amounts of and possibly Oryza spp. (rice)] approach being “thor-
nutrients on the growth of a particular plant species. She oughly studied.” An ecologist beginning the study of a
can grow several groups of plants, giving each of them a new species, community, or ecosystem must do many of
different nutrient treatment, and measure such things as these general types of experiments. Of course, he is guid-
their time to maturity and their final size. This experiment ed by his knowledge of other similar species and ecosys-
could be done in a controlled environment such as a tems. Each species or ecosystem is unique, however, which
growth chamber, in a greenhouse, in an experimental gar- is why each study expands our ecological knowledge.
den, or in a natural community in a field setting. Manipulative experiments are powerful tools for two
This range of potential settings for the experiment major reasons: first, because the scientist can control which
comes with a set of trade-offs. If the experiment is con- parts of the natural world will be altered to study their
ducted in a laboratory or growth chamber, the ecolo- effects, and second, because she can separate factors that
gist is able to control most of the possible sources of vari- typically occur together to test them individually. Such
ation so that the differences among treatments can be experiments have limitations, however. Sometimes manip-
clearly attributed to the factors being studied in the ulative experiments are plagued by artifacts—outcomes
experiment. These sorts of controlled experiments exem- caused by a side effect of the experimental manipulation
plify the scientific method as it was first laid out by itself rather than being a response to the experimental
Frances Bacon in the seventeenth century. Baconian treatment being tested. Good experiments avoid artifacts
experiments are the mainstay of most of molecular and or take them into account in evaluating the results.
cellular biology as well as the physical sciences. By Another limitation is that of scale. Ecology is con-
working in a controlled environment, however, the ecol- cerned with patterns and processes that occur across
ogist sacrifices something. The controlled environment large scales of space and time—for example, the causes
is highly artificial so that it compromises realism, and of differences in the numbers of species on different con-
it is also narrow in scope (the results apply only to a lim- tinents, or the responses of populations to climate
ited range of conditions), sacrificing generality. change over the next two centuries. We cannot do
(A) (B) (C)
(D) N
0 1 2 3
km
AL
3C
AL 3A
Ungrazed 1A
A
HQ HQB 20A 3B
Grazed by bison
AL
HQC
Grazed by cattle 4A 2A
20A
4A
Agricultural land 1A 4B
4B
2A 1B
HQ Headquarters 1B
area (small
20B
experimental plots) 4C
1C
1, 2, 3, 4, 10, 20 = Number 2B
of years
between 4D 1B
burns 1D 2D
4A 4D
20D
A, B, C, D, H = Replicate
indicators 4A 4B
10A 20B SuB
4F
20A
Season of burn treatment: WA WH 4H
SpB FH
W, Winter; 2A SpA 20C
Sp, Spring; 1B 2C
1A SuA
Su, Summer;
F, Fall 1C
2B 10B FA
Figure 1.2 pared with ungrazed areas and with plots subjected to cattle
Large-scale manipulative experiments are being carried out grazing. The experimental patches (D), which are watershed
at the Konza Prairie Research Natural Area (A). Prescribed units, vary in size from approximately 3 to 200 hectares. In
burns (B) are done at various intervals to investigate the this map, each patch is designated by a code indicating the
effects of fire and fire frequency on prairie communities. In fire treatment. (After Knapp et al. 1998. Photographs courtesy
addition, areas grazed by bison (C) are studied and com- of A. Knapp, Konza Prairie Biological Station, and S. Collins.)
manipulative experiments at these great scales of time into a series of large patches, which are subjected to dif-
and space. Ecologists are, however, increasingly making ferent combinations of controlled burning at various
use of longer-term and larger-scale manipulative exper- time intervals and grazing by bison or cattle (Figure 1.2).
iments (see Box 12C). One example of such a study is the Large-scale manipulative experiments are often lim-
long-term study of prairie ecology at the Konza Prairie ited by the range of possible treatment, however; for
Research Natural Area in Kansas, begun by Lloyd Hul- example, at Konza Prairie, almost all of the controlled
bert in 1981 (Knapp et al. 1998). The reserve is divided burning is carried out in the spring. Because we do not
have firm data on prairie fire regimes prior to European ecologist cannot tell for sure whether the increase in her-
settlement, we do not know how this spring-burning bivores is a result of increased plant numbers and pro-
treatment compares to “natural” fire regimes. ductivity, or whether the increased productivity is a
Some types of manipulative experiments would be result of increased herbivory.
unethical to carry out. For example, we would not cause As with natural experiments, observational experi-
the extinction of a species just to study the effects of such ments repeated in space or time add confidence to our
an event. In such cases, ecologists must rely on two other conclusions (Figure 1.3). Other sciences, notably geolo-
types of studies. These are natural and observational gy and astronomy, also rely largely or exclusively on
studies, which may be thought of as different kinds of observational experiments because of the spatial or tem-
experiments. poral scales of their studies or because direct manipu-
Natural experiments are “manipulations” caused lation is impossible.
by some natural occurrence. For example, a species may Ecological knowledge comes from combining infor-
go extinct in a region, a volcanic eruption may denude mation gained from many different sources and many
an area, or a flash flood may scour a streambed. Natur- different kinds of experiments. The ecologist’s use of this
al and manipulative experiments represent a trade-off complex variety of information makes ecology a chal-
between realism and precision, similar to the trade-off lenging and exciting science.
between laboratory and field experiments. Just as with
a manipulative experiment, the ecologist compares the Testing Theories
altered system either with the same system before the The testing of scientific theories, especially ecological
change or with a similar, unchanged system. ones, is a more subtle, nuanced, and complicated endeav-
The major limitation of natural experiments is that or than nonscientists often realize. The popular image of
there is never just a single difference before and after a the scientific method portrays it as a process of falsifying
change or between systems being compared. There are hypotheses. This approach was codified by the German
no guarantees, for instance, that the altered and unal- philosopher of science Karl Popper (1959). In this frame-
tered systems were identical prior to the event. For work, we are taught that we can never prove a scientific
example, if we are comparing areas burned in a major hypothesis or theory. Rather, we propose a hypothesis
fire with others that remained unburned, the unburned and test it; the outcome of the test either falsifies or fails
areas might have been wetter, might have had a differ- to falsify the hypothesis. While hypothesis testing and
ent site history or different vegetation before the fire, and falsification is an important part of theory testing, it is
so on. In other words, there are many other potential not the whole story, for two reasons.
sources of difference besides the fire. Therefore, it can be First, this approach fails to recognize knowledge
difficult to determine the cause of any change. accumulation. In a strict Popperian framework, all the-
The best natural experiments are ones that repeat ories are held to be potentially false. We never prove any-
themselves in space or time. If an ecologist finds similar thing to be true; we merely disprove ideas that are false.
changes each time, then she gains confidence about the This assumption goes against our own experience and
causes of those changes. Another approach is to com- the history of the accumulation of scientific under-
bine natural experiments with manipulative experi- standing. Today we know that the Earth revolves around
ments. For example, the patches subjected to grazing the sun, even though this was once just a hypothesis. We
and fire treatments at Konza Prairie are being compared know that the universe is approximately 15 billion years
with patches elsewhere that are not subjected to exper- old (give or take a few billion) and began with the Big
imental manipulation. Bang, even if we still do not know the details of that
Observational experiments consist of the system- event. We know that life began and assumed its pres-
atic study of natural variation. Such observations or ent shape through the process of evolution. We know
measurements are experiments if an ecologist starts with that many diseases are caused by microbial infections,
one or more hypotheses (predictions) to test. For exam- not by “humours,” and that hereditary traits are con-
ple, one could measure patterns of species diversity veyed by DNA, not by blood. While we may acknowl-
across a continent to test hypotheses about the relation- edge that all of this knowledge has not, in a strictly
ship between the number of plant species number and philosophical sense, been proved to be true, but has only
productivity (see Chapter 20). Again, the limitation of failed thus far to be falsified, we also recognize that some
this type of experiment is the potential for multiple fac- knowledge is so firmly established and bolstered by so
tors to vary together. If several factors are tightly corre- many facts that the chance that we are wrong is very
lated, it becomes difficult to determine which factor is much less than the chance of winning the lottery sever-
the cause of the observed pattern. For example, if the al times in a row. The school of philosophy of science
number of herbivores is also observed to increase as the called realism recognizes this progressive accumulation
number and productivity of plant species increases, the of knowledge (Mayo 1996).
Colder 1 Year 10 Years Figure 1.3

130 130 Repeated observations over space or time can reveal infor-
mation that is not apparent from one or a few observations.
As an example, records of the duration of ice cover on Lake
Duration of ice covert (days)

Duration of ice cover (days)
110 110 Mendota, Wisconsin, have been kept for more than 142
years. The information for a single year (A) is fairly mean-
ingless, but expanding the context with increasing numbers
90 90 of observations over time shows that (B) 1998 was the
warmest winter in 10 years; (C) there is a cycle of warmer
winters recurring every few years (now known to be the
70 70
result of the El Niño Southern Oscillation; see Chapter 18);
and (D) overall, winters in Wisconsin are warmer now than
50 50 they were 142 years ago. (After Magnuson et al. 2001.)
30 30
1998 1988 1998 processes of competition and herbivory each
Warmer
contribute to shaping this community?” So,
when we are building our theories about
Colder 50 Years
130 plant community structure, our activities are
more akin to assembling a complex model
than to falsifying a set of propositions.
110
Falsification does play a role in science,
but a more limited one than Popper envis-
90 aged. Theory construction is like assembling
a jigsaw puzzle from a pile of pieces from
70 more than one box. We can ask whether a
particular piece belongs in this spot, yes or
no, by erecting a hypothesis and falsifying
50
it. We may even conclude that this particu-
lar piece does not belong in this puzzle. Less
30
1950 1960 1970 1980 1990 2000
often are we attempting to completely
Warmer throw away the piece, saying that it does
not belong to any puzzle.
Colder 142 Years
Controversy also plays an important
170 part in ecology, as it does in all scientific
fields. During the process of amassing evi-
dence regarding the validity of a theory, dif-

130 ferent interpretations of experimental data,
and different weight given to different pieces
of evidence, will lead different scientists to
differing opinions. These opinions may be
90
passionately held and forcefully argued, and
discussion can sometimes become heated.
As the evidence supporting a theory accu-
50
mulates, some scientists will be willing to
accept it sooner, while others will wait until
30 the bulk of the evidence is greater.
1860 1880 1900 1920 1940 1960 1980 2000
Warmer
If the issue under debate has political
or economic implications, nonscientists will
Second, and more importantly, the Popperian frame- also contribute to the debate and may be able to offer
work fails to account for a second type of question that valuable insight, judgment, and perspective to the dis-
we very commonly ask in ecology. Often the issue is not cussion. But when the evidence in favor of a scientific
one of falsifying a hypothesis. Rather, we ask about the theory becomes overwhelming and the vast majority
relative importance of different processes. When we of scientists knowledgeable in that field are convinced
examine the structure of a plant community, we do not of its validity, then the matter becomes settled (unless
ask, “Is it true or false that competition is occurring?” startling new evidence or a new, broader theory forces
Instead, we ask, “How much, and in what ways, do the a reevaluation). Ultimately, it is the judgment of scien-
tists that must decide the answers to scientific questions. ogy. One approach to resolving this tension is to see how
When a scientific consensus has been reached on a sci- the outcome of a particular experiment fits into the
entific theory, it is unreasonable to consider that theory workings of existing models, and whether it supports or
to be just another guess or opinion and to hold that rejects the predictions of those models. Another
everyone’s opinion is equally valid. That may work for approach is to use methods for the quantitative synthe-
the democratic process, but not in science. Opinions not sis of the results of independent experiments. These
supported by evidence are not the same as those sup- methods, known collectively as meta-analysis, can be
ported by the weight of a great deal of evidence; giving used to evaluate where the outcome of a particular
them equal weight would be contrary to the way science experiment fits in with—or differs from—the results of
works. other similar experiments conducted on different organ-
This does not mean that scientists should decide isms at different places and times. This approach has
issues of public policy. For example, if scientists are in been used to evaluate the broad body of experimental
strong agreement about something—say, that if more evidence for a number of important ecological questions
than x% of its remaining habitat is lost, then plant species in recent years (Gurevitch et al. 2001).
Y has a 90% chance of extinction within the next 20
years—that does not necessarily dictate any particular Scale and Heterogeneity
public policy. Policy decisions depend on how impor-
tant people think it is to save species Y, and on what costs A great deal of recent interest in ecology has been gen-
they are willing to pay to do so. While we personally erated by consideration of how ecological patterns and
hope this would never happen, we recognize that some- processes vary as a function of the scale at which they
one who wanted (for whatever reason) to exterminate operate and are studied (Figure 1.4). The same phe-
species Y could use the scientific conclusions for their nomenon can be seen very differently when studied
own ends, just as we could use those same scientific con- within a small local area and at the scale of a landscape
clusions to promote conservation. or region—that is, at different spatial scales. Likewise,
one’s perspective can change dramatically when study-
Specific Results versus General Understanding ing an ecological process over a single growing season
Because ecologists work at such a variety of scales and of a few months or over a period of decades or centuries
on such a diversity of organisms and systems, the ques- (see Figure 1.3). Different kinds of things may be going
tion arises as to how far one can extend the conclusions on at different scales, and expanding one’s focus to more
of a particular study to other organisms or places. In the than one scale can be richly rewarding. In a study of a
fields of chemistry or physics, the results of an experi- local community, for example, we might see that com-
ment are considered to be absolutely true for all times petitive interactions keep individual plants of a partic-
and places: an atom of helium is made up of two pro- ular species at a distance from one another. At a larger
tons and two neutrons, which in turn are made up of scale, we might notice that the plants are grouped
quarks, with no qualifications needed. This is the pop- together across the landscape because individuals that
ular image of scientific theories. are too far apart from any others never become polli-
Ecology is different. Do the results of a field exper- nated, and fail to leave descendents; or because the seeds
iment on competition between two plant species extend have limited ability to disperse. At a regional or conti-
to other seasons or locations, or to other pairs of species nental scale, the plants may exist in several large but sep-
within the same families or functional groups? Experi- arated enclaves, determined by patterns of glaciation
ments involving helium deal with a universal entity, the and species migration thousands of years in the past.
helium atom. In contrast, in experiments on plant com- We often refer to these scale changes in terms of a
petition, the exact composition of the entities changes hierarchy, and one can move up and down many dif-
(e.g., the individual plants used each time are not genet- ferent kinds of hierarchies in ecology. For instance, one
ically identical), and the surroundings change as well can move from the level of molecules to tissues to organs
(e.g., the weather is different this year than last year). For to entire organisms. A different kind of hierarchy could
this reason, extremely cautious scientists take the posi- expand from individual organisms to populations to
tion that no conclusion can be extended beyond the par- communities to ecosystems and up to entire biomes; an
ticular conditions that existed when the experiment was alternative hierarchy might move from things that occur
conducted. If that were so, however, there would be no at the level of organisms to those that function at a scale
value and no point in doing any experiments, because of habitats, landscapes, watersheds, regions, and so on
they would tell us nothing beyond something of such up to global-scale phenomena. These levels are not nec-
limited scope as to be worthless. essarily congruent: one might study the individual adap-
The truth is somewhere between these two ex- tations of plants over a range of different environments
tremes, creating a constant and dynamic tension in ecol- across an entire landscape or even a region, for instance,
106 Figure 1.4

H
Ecologists study patterns and processes across a wide range
I
of scales in space and time. The processes of plant physiolo-
104 gy, such as the diffusion of CO2 molecules in a leaf, occur
over the shortest distances (10–4 m) and times (10–7 years).
G
F Moving up the hierarchy, the domain of the whole plant and
102 its birth, growth, reproduction, and death encompasses
Space (m)
slightly longer distances (100 m) and times (100 years). At yet

100 E larger scales (101–102 m and 101–102 years), we enter the
realm of populations and communities and their changes
D over the course of years and decades. Finally, ecologists
10–2 B study patterns that stretch across the entire globe (105 m)
C and thousands or even millions of years (103–106 or more
A years).
10–4
10–6 10–4 10–2 100 102 104 106

(1 year) seed eaters or dispersers, and mutualistic or pathogen-
Time (years) ic fungi or bacteria).
Similarly, the environment varies from moment to
A CO2 transport in leaves F Population moment. There are no specific ecological terms for the
B Nutrient transport in soil G Community components of temporal heterogeneity, but it also exists
C Seeds H Biomes, range shifts at many scales and has major effects on plants. Varia-
tions in conditions from day to night, summer to win-
D Annual plant (growing phase) I Climate change,
paleoecology, ter, across periods of wet years, cold years, or snowy
E Perennial plant (growing phase)
continental drift years, and at a longer scale as climate changes over thou-
sands of years all have important influences on plants.
Depending on the ecological process being studied and
the organisms involved, it may be the small-scale,
or how population interactions at very local scales con- moment-to-moment variation that matters most (such
tribute to the global range limitations of a species. Like- as fluctuations in light levels in a small forest gap on a
wise, one’s interpretation of data collected over a short partially cloudy day), or it may be large-scale, long-term
time period may be completely upended when the same average conditions (such as CO2 concentration in the
data are examined for trends over longer periods of time. atmosphere), or it may be the interplay between process-
One of the reasons that scale is now recognized as es occurring at different scales (such as CO2 flux in a for-
being so important is that the world is a very heteroge- est canopy over the course of a day or a season).
neous place. Even over very small distances, conditions Groups of organisms, such as populations and
can change in ways that may be very important to liv- species, sometimes average these sorts of microenviron-
ing organisms. Environmental conditions are a particu- mental influences over larger areas and over generations
lar concern in plant ecology because plants cannot move. of organisms’ lives. This averaging acts to counter the
Or, at least, mature terrestrial plants generally are firm- effects of heterogeneity, particularly over evolutionary
ly rooted in place, although their offspring may be time. At even larger scales, heterogeneity again becomes
dispersed some distance away. So, the environment critical. As continents are carried apart on tectonic plates
immediately surrounding an individual plant is over- and climates are altered, organisms must either respond
whelmingly important to its survival, growth and repro- to changing conditions by evolving or by changing their
duction. distributions, or else become extinct.
The habitat of a population or species is the kind
of environment it generally inhabits and includes the set The Structure and History of Plant Ecology
of biotic (living) and abiotic (nonliving) factors that
influence it in the places one usually finds it. But the con- Ecology is a very synthetic subject. By that we do not
ditions in the immediate surroundings of an individual mean that it is unnatural or artificial, but that it brings
plant—its microhabitat—may differ considerably from together a very wide range of other fields of science, per-
the average conditions in the general habitat (Figure 1.5). haps to a greater extent than does any other subject.
Factors operating to distinguish a microhabitat from oth- Some of the fields that ecology encompasses or overlaps
ers around it include the composition of the soil, the with include geology, geography, climatology, soil sci-
microclimate of the immediate area, the presence, size, ence, evolutionary biology, genetics, statistics and other
and identity of neighboring plants, and other organisms branches of mathematics, systematics, behavior, physi-
in the immediate surroundings (grazers, pollinators, ology, developmental biology, molecular biology, and
N
Sun
1m Exposed
to wind
South-facing; warmer
(more heat from sun)
North-facing; Cold air

drainage Re-radiation
Drier cooler of heat
ridges Greater herbivory
Water Slope
Groundhog runoff
Frost pockets Dense vegetation,
burrow more competition
Subsurface
drainage Moister soils
Low-lying area
Figure 1.5
The environment in a particular microhabitat can differ in a number of ways from con-
ditions in the surrounding area. Individual plants experience the conditions in their
immediate microhabitat, not the average conditions in the general area. The grass on the
north-facing slope at the left of the diagram experiences cooler temperatures, and per-
haps greater herbivory due to its proximity to the groundhog burrow. The plants in the
low-lying area may experience frost sooner in the fall and later in the spring than the
surrounding areas due to cold air drainage; here, the soils are moister, and competition
for light and soil moisture may be more intense, than on the ridges. Other potential
effects of a microclimate are also illustrated.
biochemistry. We touch on many of these fields through- often carried a ship’s naturalist who cataloged the
out this book, showing you how they fit into the toolk- remarkable range of organisms and environments they
it of an ecologist and how familiarity with them affects encountered. Charles Darwin was one such ship’s nat-
the ways in which ecologists think about and study uralist, and the story of his five-year voyage was pub-
organisms in nature. lished as his Voyage of the Beagle (Figure 1.6). Ecology as
This is not the place to present a detailed and defin- a recognized discipline coalesced in the latter half of the
itive history of plant ecology. Instead, we sketch some nineteenth century. The German biologist Ernst Haeck-
of the major milestones, with an admitted bias toward el coined the term “oecology” in 1866, and by early in
the English-speaking scientific community. Other his- the twentieth century the Ecological Society of Ameri-
torical details are scattered throughout the book as we ca had formed.
discuss particular topics and subfields. While no single Plant ecology as a discipline is made up of a num-
definitive history of plant ecology exists, several books ber of different subdisciplines, some of which have quite
and papers describe parts of its history (McIntosh 1985; distinct traditions and histories. Some early plant ecol-
Westman and Peet 1985; Nicholson 1990; Allen et al. ogists and botanists focused on whole communities,
1993). while others focused on single species and the proper-
The roots of plant ecology go back to prehistoric ties of individuals. The older (now largely archaic) terms
times, when people’s health and survival depended on for these two subfields are synecology and autecology.
their abilities to understand many aspects of the ecolo- Plant community ecologists, in particular, were active in
gy of plants with great accuracy. Ecology as a science the origins of ecology as a discipline in the last part of
began with the Greeks, most notably Aristotle, in the the nineteenth century and dominated plant ecology
fourth and fifth centuries B.C.E. The modern science of during the first half of the twentieth century. A more
plant ecology began as the study of natural history in detailed discussion of the history of plant community
the eighteenth and nineteenth centuries, carried out by ecology and some of the key figures in that history is
professional and amateur naturalists in Europe and given in Chapters 12 and 13.
North America and in their travels throughout the Early studies in plant autecology were especially
world. Ships on journeys of discovery and colonization concerned with understanding unique plant adaptations
Figure 1.6
H.M.S. Beagle sailed from England December 27, 1831 on a 5-year mission to chart the
oceans and collect biological information from around the world. Charles Darwin sailed
with the Beagle as ship’s naturalist; he is pictured here at the age of 24, shortly after com-
pleting the voyage. Darwin collected vast numbers of plant and animal specimens and
recorded copious scientific observations that were instrumental in the creation of his most
famous work, On the Origin of Species. (Images from Science Photo Library/Photo
Researchers, Inc.)
to extreme environments, such as deserts, and a number tion ecology extends back to at least the 1920s (for exam-
of famous studies were concerned with plant perform- ple, with the work of Gause, Pearl, Lotka, and others).
ance in the field. Although some major insights were Plant population ecology drew on these ideas and the-
gained, technological limitations severely hampered the ories as it was developing, as well as other ideas that
development of the field. As instrumentation and originated among plant ecologists. Eventually new the-
methodology became more sophisticated, plant physi- ories were needed as discoveries about the unique
ologists began to carry out most of their research in con- nature of plants made it obvious they could no longer
trolled laboratory environments. be shoehorned into many of the theories constructed for
Beginning in the middle of the twentieth century, animals.
further advances in technology made it possible for Conversely, physiological ecology advanced earlier
physiological studies to come out of the greenhouse and and more rapidly among plant ecologists than it did
into nature, eventually leading to the creation of the among animal ecologists. Undoubtedly this was because
fields of plant physiological and functional ecology. At the characteristics of plants are much easier to measure,
about the same time, autecology began to fission into and their environments easier to characterize, than those
subfields that focused on single individuals and on pop- of animals (for most purposes, one does not have to
ulations. Plant population ecology as a recognizable catch plants!). On the other hand, in the 1980s, animal
subdiscipline had its origins in Great Britain in the physiological ecology joined with evolutionary biolo-
1960s, particularly with John Harper and his students. gy to create the field of evolutionary physiology (Feder
It then spread to North America in the 1970s. This is et al. 1987), a move that plant biologists have not yet
necessarily a very simplified and limited description of clearly made.
events; for instance, a number of individuals in many The gap between the fields of plant and animal ecol-
countries around the world were carrying out studies ogy was bridged in the 1970s, although distinct subfields
that today we would label as plant physiological ecol- continue to this day. Two related developments were
ogy or plant population ecology as far back as the nine- responsible. The first was the rise of studies of plant-ani-
teenth century. mal interactions, especially pollination (see Chapter 8)
For the most part, plant ecology for the first three- and herbivory (see Chapter 11). The second develop-
quarters of the twentieth century developed independ- ment was the burgeoning interest in the evolutionary
ently of animal ecology. Animal community ecology has aspects of ecology in the 1970s and 1980s, which tran-
a long history parallel to that of plant community ecol- scended the traditional separation of the studies of
ogy (Mitman 1992). Substantial work in animal popula- plants and animals.
The most recent changes in the field of plant ecolo- A major trend in contemporary ecology, including
gy have been the rise of landscape ecology and conser- plant ecology, is toward larger, more integrated research
vation ecology as recognized disciplines in the late 1980s. projects that involve many collaborators and examine
Landscape ecologists began their careers from various phenomena across large scales of space and time or
different directions, including fields as diverse as plant across levels of organization. Except in the subdiscipline
community ecology and remote sensing. Conservation of ecosystem ecology, which was undertaking projects
ecologists likewise created their field from backgrounds with large teams of scientists in the 1970s, such multi-
in mathematical modeling and population, community, investigator studies were very rare in ecology until
and ecosystem ecology. Other fields within plant ecolo- recent years. These studies may cover a range from
gy have seen major shifts in emphasis. Plant communi- molecular genetics up through ecosystems and social
ty ecology has seen such a shift in the past quarter cen- systems, and are erasing many of the traditional bound-
tury. Previously it was dominated by questions about aries among subdisciplines. Plant ecology is experienc-
whole-community patterns and processes; now, its main ing exciting times, and we hope you will sense that
focus has shifted to questions closer to population ecol- excitement in this book.
ogy, about interactions within and among species.
Additional Readings
Classic References Mayo, D. G. 1996. Error and the Growth of Experimental Knowledge. Uni-
versity of Chicago Press, Chicago, IL.
Platt, J. R. 1964. Strong inference. Science 146 : 347–353.
Nicholson, M. 1990. Henry Allan Gleason and the individualistic hypoth-
Popper, K. R. 1959. The Logic of Scientific Discovery. Hutchinson & Co.,
esis: The structure of a botanist’s career. Bot. Rev. 56:91–161.
London.
Salt, G. W. 1983. Roles: Their limits and responsibilities in ecological
and evolutionary research. Am. Nat. 122: 697–705. Additional Resources
McIntosh, R. P. 1985. The Background of Ecology. Cambridge University
Press, Cambridge.
Contemporary Research
Pickett, S. T. A., J. Kolasa and C. G. Jones. 1994. Ecological Understand-
Hull, D. L. 1988. Science as a Process. University of Chicago Press, Chica-
ing. Academic Press, San Diego, CA.
go, IL.
PART
I The Individual
and its Environment
2 Photosynthesis
and the Light Environment 15
3 Water Relations and Energy Balance 41
4 Soils, Mineral Nutrition,
and Belowground Interactions 63
C H A P T E R
2 Photosynthesis
and the Light Environment
H
ow does a plant “perceive” and respond to the environment that sur-
rounds it? Why is it that some plants can survive extremes of tem-
perature and drought while others cannot? What enables certain
plants to thrive in the deep shade of the understory of a tropical rainforest, and
others to succeed only in the sunniest habitats?
The functional ecology of plants is concerned with how the biochemistry
and physiology of individual plants determine their responses to their envi-
ronments within the structural context of their anatomy and morphology. Func-
tional ecology is similar to physiological ecology, a subdiscipline of ecology
that focuses on physiological mechanisms underlying whole-plant responses
to the environment. It underlies much about the ecology of plants. Part I of this
book is particularly concerned with functional ecology.
Plants must acquire energy and materials for growth, maintenance, and
reproduction. They must also limit their losses; for example, if a plant loses too
much water, it will wilt and eventually die. Plants must also allocate resources
in ways that maximize their chances for contributing offspring to the next gen-
eration while simultaneously maximizing their chance of surviving to repro-
duce. In this chapter and the next two, we examine how plants capture the ener-
gy of sunlight and incorporate carbon from the atmosphere in photosynthesis,
their adaptations to the light environment, their water relations, and their
uptake and use of mineral nutrients. We also examine the structures in which
some of these processes take place and consider some of the biochemistry
involved.
While we focus here on processes occurring at the small scale of a cell, a leaf,
or an individual plant, it is important not to lose sight of the forest: plants have
evolved and live in an ecological context. Photosynthesis, the biochemical
process by which plants capture energy from sunlight and fix carbon from the
air, is usually carried out not in a laboratory, but in natural environments. The
photosynthetic machinery, and the leaf in which it is housed, are both evolu-
tionarily adapted and acclimated to the environment in which the individual
plant is growing. The temperature and the amount of available light, water, and
nutrients in the environment determine when and how rapidly the leaf can pho-
tosynthesize and the extent to which the plant grows and is likely to survive.
The physical conditions that the plant experiences are determined not only
by the physical features of the environment, but also by other living organisms
16 Chapter 2
in that habitat. The amount of light available for photo-

The Process of Photosynthesis
synthesis may be limited by other plants competing for
that light. A plant’s ability to capture carbon and ener- Photosynthesis is the set of processes by which plants
gy may also be reduced by herbivores eating its leaf tis- acquire energy from sunlight and fix carbon from the
sue. Pathogens and pollutants may limit the plant’s abil- atmosphere. It consists of two major parts: the initial
ity to photosynthesize. The plant responds to all of these capture of light energy and the incorporation of that
aspects of its environment as an integrated unit, energy, together with carbon dioxide, into organic com-
although in textbooks such as this one, we arbitrarily pounds. The organic molecules formed in photo-
separate its responses into categories for convenience, synthesis are used by the plant to create new tissues, reg-
treating them in different chapters. With that perspec- ulate the plant’s metabolic processes, and to supply
tive, we begin our examination of plants’ interactions energy to those metabolic processes. Both the capture of
with their environment with the process by which they energy (the light reactions) and the initial formation of
acquire energy and carbon: photosynthesis. carbohydrates (carbon fixation) take place in the chloro-
plasts.
The light reactions of photosynthesis occur on the
thylakoid membranes (also called lamellae) in the inte-
rior of the chloroplasts. These double membranes exist
(A) in the form of grana stacks alternating with sheets of
interconnecting stroma thylakoids (Figure 2.1). The suc-
cessful capture of light energy depends on the precise
spatial arrangement of these photochemical reactions
Lipid (oil) globule
within the membranes in which they occur. The archi-
Chloroplast envelope tecture of the thylakoid membranes is complex and is
(outer membrane) not yet fully understood, but they appear to consist of
a lipid matrix (which may be partially fluid) into which
enzyme proteins and pigment-protein complexes are
Plasmalemma integrated. These highly organized components exist
(cell membrane) within a dynamic and flexible membrane system.
The pigment molecules responsible for the capture
of light energy form two distinct molecular complexes
in plants, photosystem I and photosystem II (Figure
Stroma thylakoids 2.2). Unicellular eukaryotic algae, such as the Chloro-
phyta, and prokaryotic cyanobacteria also have photo-
Grana stacks (B)
Stroma
Stroma thylakoids
Cell wall
Grana stack
Grana stacks
Figure 2.1
(A) Chloroplast from a leaf of Nicotiana tabacum (tobacco, Solanaceae), showing grana
stacks and stroma thylakoids. (B) A model of grana stacks and stroma thylakoid
1 µm structure within a chloroplast. (After Esau 1977.)
Photosynthesis and the Light Environment 17
Figure 2.2
Chl* The light reactions of photosynthesis in plants.
2e−
Photons of light excite electrons in photosystems I
F and II. The excited electrons from photosystem I
are passed eventually to NADPH, where the ener-
Fd 2e− gy is captured in a high-energy bond. The excited
ATP ADP 2e− electron from PS II is passed through a series of
Chl*
2e− NADP+ electron acceptors, transferring energy in a regulat-
Q
2e− PQ ed fashion to ATP, finally replacing the electron lost
NADPH Cytb/f from PS I with an electron from a water molecule.
PC
Chl
Reaction
2e−
ENERGY STATE
center P700
1/ O
Antenna 2 2
molecules + 2H+
PHOTOSYSTEM I 2e−
Reaction
Chl center P680
H2 O
Antenna
molecules
(accessory
PHOTOSYSTEM II pigments)
Photon of
light
Photon of
light
systems I and II, but certain other photosynthetic bac- In the biochemical reactions of the Calvin cycle (Fig-
teria have only photosystem I. Each photosystem con- ure 2.3), CO2 is taken up from the atmosphere and the
sists of hundreds of pigment molecules, including sev- carbon is ultimately incorporated into organic com-
eral forms of chlorophyll plus accessory pigments. In pounds (“fixed”). These reactions take place in the stro-
terrestrial plants, the accessory pigments are primarily ma, the watery matrix that fills the chloroplast. In C3
carotenoids and xanthophylls, but eukaryotic algae and plants (plants with the most common type of photo-
photosynthetic bacteria also use other pigments. synthetic pathway; see below), large amounts of the
When a photon of light is captured by the tightly enzymes that catalyze the reactions of the Calvin cycle
packed “antenna molecules,” it is passed from one mol- are dissolved in the stroma. Energy captured by the light
ecule to another by the process of resonance transfer reactions is transferred, in the form of high energy bonds
until it reaches the chlorophyll molecule in the reaction in ATP and NADPH+, to the stroma, where it is incor-
center of the photosystem, where it is trapped. The excit- porated, along with CO2, into carbohydrate molecules.
ed chlorophyll molecule then passes a light-excited, The fixation of carbon is thus powered by the light reac-
high-energy electron to an electron acceptor. The ener- tions, and the light energy captured in photosynthesis is
gy in this high-energy electron is ultimately captured in ultimately stored in the chemical bonds of carbohydrates
high-energy bonds in ATP and NADPH+. and other organic molecules.
Water molecules are the ultimate donors of these
electrons. The oxygen that we breathe was released into Photosynthetic Rates
the atmosphere from water molecules that were split to
replace the electrons in photosystem II. Oxygen from The rate at which a leaf can capture light energy and
photosynthesis was first released into the atmosphere fix carbon is determined by several factors. Plants, like
beginning about 2 billion years ago, but was consumed other aerobic organisms, use oxygen and release CO2
by the weathering of iron and other minerals in rocks for in the process of cellular respiration, by which organic
over a billion years after that. Atmospheric oxygen compounds are broken down to release energy. Gross
reached its present levels about 400 million years ago, photosynthesis, or the total amount of carbon captured,
although there have been some fluctuations in that level is reduced by the plant’s respiratory release of CO2. Pho-
since that time. tosynthetic uptake of CO2 by plants is far greater on
18 Chapter 2
To lipids
and proteins
CO2 Energy from
ATP light reactions
ADP
3PGA
(3-C)
RUBISCO NADPH
RuBP
ADP (5-C)
NADP+
G3P
(3-C)
ATP
Other
Ru5P carbon
(5-C) chains
Energy
from light
reactions
Fructose Glucose To sucrose,

starch,
cellulose
Figure 2.3
The Calvin cycle of carbon fixation in plants. CO2 enters the stomata from the air sur-
rounding the leaf. In a reaction catalyzed by the enzyme rubisco, CO2 is joined with
the 5-carbon molecule RuBP to form two 3-carbon molecules (3PGA). Eventually
simple sugars such as fructose and glucose are formed, to be ultimately transformed
into many other organic (carbon-based) molecules.
average than respiratory losses, however, resulting in a

net gain of carbon by plants.
Maximum rate of photosynthesis (Pmax)
Limitations Caused by Light Levels
The most basic factor limiting photosynthesis is the total
amount of light energy that reaches the thylakoid mem-
branes. In darkness, cellular respiration results in a net
Net CO2 exchange
loss of carbon and energy from the plant, as there is no +

photosynthetic capture of either light or carbon (for a
partial exception, see the discussion of CAM photosyn-
thesis below). As the light level increases, plants begin
to take up CO2. At the light compensation point, pho- 0
600 1200 1800
tosynthetic gains exactly match respiratory losses (in
PPFD (µmol m–2 s–1)
other words, net CO2 exchange is zero) (Figure 2.4).
Beyond that point, the more light that is available to be Compensation point
– = light intensity at zero CO2 exchange
captured, the greater the photosynthetic rate, up to a
maximum at which the rate plateaus in most plants.
The light compensation point can differ among plant Figure 2.4
Net CO2 exchange (per unit leaf area) for a typical C3 plant
species living in different parts of the environment or as a function of increasing light levels, showing the light
within a given habitat. It can even differ within individ- compensation point and a plateau at a maximum rate of
ual plants depending on the structure and biochemical photosynthesis. (After Fitter and Hay 1981.)
Table 2.1 Maximum photosynthetic rates (Amax), light compensation points (LCP),
and rubisco levels for three forest understory herbs
Spring Summer Autumn
Parameter Allium Viola Tiarella Viola Tiarella Tiarella
Amax 15.4 ± 0.9 12.1 ± 0.7 6.8 ± 0.7 5.6 ± 0.5 3.9 ± 0.5 5.4 ± 0.3
LCP 21.6 ± 1.4 8.4 ± 1.3 9.0 ± 1.0 4.1 ± 0.9 3.2 ± 0.5 6.5 ± 0.8
Rubisco 2.83 ± 0.21 1.84 ± 0.25 1.47 ± 0.12 0.93 ± 0.07 0.50 ± 0.17 0.78 ± 0.11
Source: Rothstein and Zak (2001)
Note 1: Values are expressed on a per unit leaf area basis. Amax is given in µmol CO2/m2/s; LCP is given as the PPFD
at which net CO2 assimilation is zero, in µmol/m2/s; and rubisco levels are in g/m2. Values are means ± 1 standard error,
with n = 5 plants per measurement.
Note 2: The duration during which each species had green leaves above ground was: Allium tricoccum, about 75 days;
Viola pubescens, about 150 days; and Tiarella cordifolia, about 185 days.
constituents of the leaves. Rothstein and Zak (2001) con- rally arranged leaves that minimize self-shading. A. mar-
trasted the photosynthetic characteristics of three for- tiana was found in locations with somewhat higher light
est-floor herb species within a northern hardwood for- levels, and had a greater number of leaves and a greater
est. In this forest, understory light levels are high in early total leaf area, compared with G. cuneata. As a result, G.
spring before the canopy leafs out, low in midsummer, cuneata had greater efficiency of light interception (the
and higher again in autumn as leaves start to fall. A proportion of incident light intercepted by the plant
spring ephemeral, Allium tricoccum (wild leek, Liliaceae),
had a constant light compensation point (Table 2.1), but
was photosynthetically active only during a short peri-
Epidermis Palisade parenchyma
od in spring. In contrast, a summer-green plant, Viola
pubescens (downy yellow violet, Violaceae), shifted its
light compensation point downward from spring to mid-
summer, while a semi-evergreen species, Tiarella cordifo-
lia (foam-flower, Saxifragaceae), also shifted its light
compensation point downward over that period, but
shifted it upward again in autumn. The spring ephemer-
al appears to be adapted to optimize its photosynthetic
uptake in the high-light environment it experiences in
spring, while the other two species are both better adapt-
ed to photosynthesize under shady conditions, at least
in part due to their ability to shift the light compensation
point.
The quantity of light reaching the thylakoid mem-
branes of a chloroplast can be limited by a number of fac-
tors. The location of the chloroplast within the leaf can
affect the light reaching the thylakoid membranes, as can
the angle at which sunlight hits the leaf. In a typical C3
leaf, photosynthesis takes place in the spongy and pal-
isade parenchyma cells that make up the mesophyll (the
photosynthetic tissue between the upper and lower epi- Stoma Spongy Stoma
parenchyma
dermis of a leaf). There are many chloroplasts in each
photosynthetic cell (Figure 2.5). On a larger scale, self- Figure 2.5
Scanning electron micrograph of a cross section of a leaf of
shading by other leaves on the same plant, or shading by Brassica septiceps (turnip, Brassicaceae), showing the palisade
competitors, can also limit the amount of light available parenchyma and spongy parenchyma cells inside which
to be captured. We will examine some of these factors most chloroplasts are found and in which most of the plant’s
in more detail in other chapters. photosynthesis takes place. Many of the cells in this micro-
Robin Chazdon (1985) studied the efficiency of light graph have been broken open to expose their internal struc-
ture. The upper epidermis is visible, as are several stomata
capture in two understory dwarf palms in the rainforests on the underside of the leaf, along with the substomatal cav-
of Costa Rica, Asterogyne martiana and Geonoma cuneata ities they open into on the inside of the leaf. (Photograph ©
(both in the Arecaceae). Both species have narrow, spi- J. Burgess/Photo Researchers, Inc.)
20 Chapter 2
(A) Figure 2.6

(A) Absorption spectra of (wavelengths of light absorbed by)
1.0
Whole leaf a whole leaf and an extract of all photosynthetic pigments
from that leaf. (B) Absorption spectra of purified extracts of
0.8 chlorophylls a and b. (C) Absorption spectra for two of the
most important accessory pigments, lutein and β-carotene.
Absorbance
0.6 (After Mohr and Schopfer 1995.)
0.4
0.2 leaves to mature plants with many leaves, this shift

Pigment extract
reduced self-shading to a minimal level and optimized
0
340 420 500 580 660 740 light capture for individuals of very different forms and
total leaf areas.
(B) The quality of light, or the wavelengths of light
160
available to be captured, can also limit photosynthetic
120
rates (Figure 2.6). Blue and red wavelengths are prefer-
140
) Molar absorptivity
100 Chlorophyll b ( entially captured by the light reactions. Perhaps para-

120 doxically, given our image of the beautiful green world,
[l•nmol–1•cm–1]
[l•nmol–1• cm–1]
) Molar absorptivity
80 100 green wavelengths are particularly ineffective for pho-

Chlorophyll a
60 80 tosynthesis. We see nature as green because green light
60 is reflected or transmitted—“discarded” rather than
40 used—by plants. The wavelengths of light that can be
40
20 used in photosynthesis are termed photosynthetically
20
(
active radiation, or PAR. The amount of usable light

0 0 energy impinging upon a leaf per unit time is called
340 420 500 580 660 740
the photosynthetic photon flux density (PPFD).
(C)
Limitations on Carbon Uptake
1.0
Plants take up CO2 from the atmosphere as air moves
Lutein through the stomata and into the intercellular spaces sur-
0.8
rounding the photosynthetic cells within a leaf. Carbon
Absorbance
0.6 uptake is driven by a concentration gradient of CO2, set

β-Carotene up by the biochemical reactions in the chloroplasts that
0.4 remove CO2 from the intercellular spaces. The uptake
of CO2 is regulated by the conductance to CO2 diffusion
0.2 on the pathway from the air into the leaf and into the
chloroplast. The concentration of CO2 in the intercellular
0
340 420 500 580 660 740 spaces depends on how rapidly CO2 is removed by being
Wavelength (nm) fixed into organic compounds, and on how readily CO2
UV Violet Blue Green Yellow Red IR comes into the leaf to replace that CO2.
The leaf conductance to CO2 is the rate at which CO2
flows into the leaf for a given concentration difference
canopy, which depends upon leaf arrangement and dis- between ambient and intercellular CO2. The inverse of
play angle), but A. martiana had a greater total capacity conductance is resistance. Low conductance or high
to capture light (where the light interception capacity, or resistance at a particular point in the pathway of CO2
effective leaf area, is the product of total leaf area and movement will limit its movement along that pathway.
light interception efficiency). Takenaka and associates If the overall leaf conductance to CO2 is high and CO2
(Takenaka et al. 2001) followed up this work by analyz- concentrations in the intercellular spaces are being con-
ing the effects of leaf display on light capture efficiency tinually drawn down by the rapid fixation of carbon, then
in another understory palm, Licuala arbuscula (Are- CO2 influx from the air surrounding the leaf will be high.
caceae), which grows in lowland rainforests in southeast The rate of CO2 uptake can be modeled with a flux
Asia. This species has compound, fan-shaped leaves equation. Flux equations are used to model flow rates,
with long petioles. The authors found that the angle at and are of the general form
which the petioles are held changed as the number of
flux = (conductance) × (driving force)
leaves increased. As plants grew from juveniles with few
For CO2 uptake, the driving force is a difference in CO2 carbon under different conditions as morphology, phys-
concentration, and the flux equation can be stated as iology, and other plant characters evolve.
Why would plants ever restrict their uptake of CO2?
CO2 uptake rate = (leaf conductance to CO2 diffusion)
We examine this question in more detail in Chapter 3,
× (difference in CO2 concentration between air and
but briefly, it is largely because CO2 gain is linked inex-
chloroplast)
tricably with the loss of water through the same stom-
or, using conventional symbols, atal openings in the leaf through which CO2 is taken up.
A different formulation for photosynthetic rate is
A = gleaf × (Ca – Ci)
sometimes employed to describe net photosynthesis at
The term A is the assimilation rate (in µmol/m2/s); this light saturation, Asat, the light level at which the maxi-
is the rate at which CO2 is taken up by the leaf. The terms mum photosynthetic rate is reached, when CO2 uptake
(CO2a) Ca and Ci are the ambient and intercellular con- is not limited by stomatal conductance:
centrations of CO2, respectively; that is, the concentra-
Asat = gm × (Ci – Cc)
tions in the surrounding air and at the surface of the pho-
tosynthetic cell. The term gleaf is the total conductance of where Cc is the compensation point for CO2 and gm
the leaf to CO2. is the mesophyll conductance or intracellular conduc-
We can separate leaf conductance into its two major tance, the conductance to CO2 through the leaf meso-
components, gs and ga—the conductances to CO2 phyll cells and cell walls.An enormous amount of air
through the stomata and through the boundary layer of must be processed by the leaf in the course of photo-
air surrounding the leaf, respectively—so that synthesis. To make a single gram of the carbohydrate
glucose, a plant needs 1.47 grams of CO2, which is the
1/gleaf = 1/ga + 1/gs
amount in about 2500 liters of air. Looked at another
Generally, ga is large, since CO2 readily passes through way, the air needed to fill a structure the size of the
the boundary layer, and so does not contribute greatly Houston Astrodome could supply enough CO2 to fix
to the regulation of CO2 flux. The conductance to CO2 about 590 kilograms (1300 lbs) of glucose (Figure 2.7).
through the stomata (gs), however, is highly variable and When the stomata of a leaf are fully open, its con-
is under the control of the plant. Stomatal conductance ductance to CO2 is generally high. The exact value
regulates leaf CO2 flux under most conditions. Thus, depends on the number and size of the stomata, and it
plants are not merely passive recipients of CO2, but reg- varies among species, individual plants, and even leaves
ulate its uptake closely. This regulation occurs over short on the same plant. (We will return to the issue of stom-
time scales (seconds to minutes), as stomata are opened atal number and size in the next chapter.) When the
or closed, and over longer time scales (days to months), stomata are closed, leaf conductance to CO2 approach-
as leaf morphology and chemistry are altered. Over es zero, although sometimes small amounts of CO2 may
much longer time scales (centuries to millennia or “leak” through the cuticle.
longer), natural selection acts to alter the capacity of The stomata of many plants are very dynamic. The
plant populations in different environments to take up guard cells that determine the degree of stomatal open-
Photo to come afternoon Figure 2.7

The Houston Astrodome in its glory
of 5/13/02- days. Given the amount of air that
filled the Astrodome at any one
moment, had its builders used real
David waiting on corbis- Bermuda grass instead of inventing
Astroturf and not painted the dome
to make it opaque, the turf could
have turned the CO2 in that air into
roughly 590 kilograms of glucose—
and that’s without a single player or
spectator breathing into it. The
advent of retractable domes has
returned natural, carbon-fixing turf to
its former prominence at ball fields.
22 Chapter 2
ing are continually in motion, widening and narrow-

ing the stomatal pores to regulate CO2 entering the leaf
and water leaving it. Some of the stomata may begin to
close while others remain open (Figure 2.8). Such patchy
stomatal closure may be more common when plants are
experiencing stress, but this phenomenon is still under
active study (Terashima 1992). The guard cells are under
a complex set of controls that respond to both internal
and external factors.
Variation in Photosynthetic Rates within

and between Habitats
Photosynthetic rates sometimes vary among plants with-
in a habitat, and across habitats, in ways that seem to
make sense because they are correlated with species
composition, habitat preferences, or growth rates. In
other cases, photosynthetic rates may have little role in
Stomatal
determining population processes or species’ distribu- width (µm)
tions. Even growth rates may be minimally related to
photosynthetic rates. The total carbon accumulated by 0–2
a plant depends not only on the rate of photosynthesis 2–4
on a leaf area basis, but also on the total leaf area of the 4–6
plant, as well as on other factors, such as the length of
6–8
time the leaves are maintained and are photosyntheti-
cally active. 8–10
An example of a case in which differences in pho- > 10
tosynthetic rates correlate with species and habitat dif-
ferences comes from a study of dwarf shrubs in a sub-
alpine heath community in the Apennine Mountains of
northern Italy (Gerdol et al. 2002). In sheltered locations, Figure 2.8
two deciduous shrubs, Vaccinium myrtillus (blueberry, Stomatal widths in different parts of a leaf of Commelina com-
Ericaceae) and V. uliginosum (bog-bilberry, Ericaceae) are munis (day-flower, Commelinaceae) at midday. Some stoma-
ta are wide open, while others are partially open or fully
dominant, or at least abundant, and the vegetation is closed. (After Larcher 1995.)
very dense. In exposed habitats, the vegetation is more
open, and a variety of shrubs are dominant, including
V. uliginosum and an evergreen species, Empetrum her-
maphroditum (crowberry, Ericaceae). Renato Gerdol and synthetic rates, and the evergreen Tiarella cordifolia had
colleagues found that the evergreen E. hermaphroditum the lowest rates. In midsummer, when light levels were
had the lowest photosynthetic rates of the three species. lowest, photosynthetic rates declined substantially for
The deciduous V. uliginosum had intermediate rates, both Viola and Tiarella, but Tiarella still had the lower
which did not differ among habitats. The deciduous rates of photosynthesis. In autumn, only Tiarella was
species restricted to the more favorable sites, V. myrtillus, photosynthetically active, and its maximum photosyn-
had the highest photosynthetic rates. The differing pho- thetic rate increased again in the higher light environ-
tosynthetic rates of these plants did not, however, cor- ment. These differences among species and seasons were
respond to their relative growth rates, nor did they positively correlated with the levels of rubisco (the
explain the relative responses of the plants to removal enzyme that catalyzes the initial capture of CO2; see
of competing neighbors or to fertilizer additions. below) in the leaves and patterns of plant growth as well
In the study of northeastern forest understory as with the duration of time that each plant is photo-
species discussed above (Rothstein and Zak 2001), max- synthetic. Allium gained all of its biomass during the
imum photosynthetic rates were correlated with the spring high light period and lost biomass after that time.
growth environment of the three species studied (see Viola and Tiarella also gained biomass rapidly during
Table 2.1). The spring ephemeral, Allium, which grew spring, and continued to increase in biomass during the
only in the highest light period, had the highest maxi- summer. Viola sharply declined in biomass from late
mum photosynthetic rates overall. During spring, the summer through winter, but Tiarella continued to accu-
summer-green Viola pubescens had intermediate photo- mulate biomass through early winter. Only 25% of the
BOX 2A
Photorespiration
CO2 O2
The mitochondria of plants carry out
cellular respiration much as those of PHOTOSYNTHESIS PHOTORESPIRATION
animals do, consuming O2 and releas-
ing energy to be used by the cells. Carboxylase Oxygenase
Plants also carry out another kind of
respiration, called photorespiration.
Like ordinary cellular respiration, pho- CH2O P
torespiration consumes O2 and releas- C O
es CO2, but unlike cellular respiration,
it depends on light. It takes place in CHOH
cells that contain chloroplasts, but
CHOH
involves two additional organelles:
mitochondria and peroxisomes. CH2O P
In photorespiration, rubisco cat- Ribulose-1, 5-
alyzes the binding of O2 to ribulose bis- biphosphate
phosphate (RuBP). This process thus
competes with photosynthesis both for
RuBP, the substrate of both reactions, C3
and for rubisco, the enzyme that cat- COOH pathway COOH
alyzes both reactions. High partial pres-
CHOH CHO P
sures of oxygen, low CO2 concentra- 2-Phosphoglycolate
tions, and warm temperatures favor CH2O P
photorespiration over photosynthesis. 3-Phosphoglycerate
Although photorespiration is often
considered to be disadvantageous be- COOH
cause it competes with photosynthesis,
it may have a protective function. Pho- CH2OH
Glycolate
torespiration may “soak up” excessive
electron flow in bright light, thereby
Chloroplast
protecting photosystem II from damage
when the leaf’s carboxylation capacity
is not capable of keeping up with the
energy captured in the light reactions COOH
(e.g., when drought forces stomata to Methyl groups CH2OH
close partially or fully, limiting or cut-
ting off the supply of CO2). COOH COOH
NH3
CH2NH2 CH2NH2
Glycine
The biochemical reactions of photorespiration
take place in chloroplasts, mitochondria, and Mitochondrion Peroxisome
peroxisomes. CO2
biomass gain for Tiarella came during the low light peri- metabolism). C3 photosynthesis and C4 photosynthesis
od of summer. are named for the three-carbon and four-carbon mole-
cules that are the first stable products of photosynthesis
in these pathways, while CAM is named after the plant
The Three Photosynthetic Pathways family Crassulaceae (the stonecrops), in which it was
Plants fix carbon using one of three different photosyn- first discovered. The vast majority of plants use C3 pho-
thetic pathways: C3, C4, or CAM (crassulacean acid tosynthesis, and C3 plants are found everywhere that
24 Chapter 2
plants exist. C3 photosynthesis was the first pathway in the air. Plants evolved under atmospheric CO2 levels
to evolve and the first to be understood by scientists. very different from those of today, as we will see short-
C4 and CAM photosynthesis are modifications of C3 ly. The current rapid increases in atmospheric concen-
photosynthesis, and evolved from it. trations of CO2 due to human activities may have long-
term consequences for CO2 uptake by plants. We will
C3 Photosynthesis return to this issue in Chapter 22.
In the Calvin cycle of C3 photosynthesis (see Figure 2.3), There is an evolutionary “solution” to the dilemma
CO2 is joined with a five-carbon molecule, RuBP (ribu- posed by photorespiration and the limitations of rubis-
lose bisphosphate), to form a six-carbon compound that co as a catalyst for CO2 uptake. That solution is C4 pho-
instantly separates into two three-carbon molecules tosynthesis.
(phosphoglycerate; 3PGA). In C3 photosynthesis, there-
fore, the first stable product of carbon fixation is a three- C4 Photosynthesis
carbon chain. Like C3 photosynthesis, C4 photosynthesis ultimately
The initial step in which CO2 is captured—the car- depends on the Calvin cycle to convert CO2 into carbo-
boxylation of RuBP—is catalyzed by the enzyme RuBP hydrates. However, C4 photosynthesis contains an addi-
carboxylase/oxygenase (mercifully nicknamed rubisco), tional step that is used for the initial capture of CO2 from
probably the most abundant protein on Earth. Rubisco, the atmosphere (Figure 2.9). In this additional step, a
however, is curiously inept at capturing CO2, which is three-carbon molecule called PEP (phosphoenolpyru-
particularly strange considering how important this task vate) is joined with CO2 to form a four-carbon acid, OAA
is for primary productivity on Earth—one might have (oxaloacetate). Thus, the first product of carbon fixation
expected a more efficient process to have evolved and in C4 photosynthesis is a molecule with four carbons.
replaced it in plants long ago. Not only does rubisco The initial capture of CO2 is catalyzed by the enzyme
have a relatively low affinity for CO2, it also has an alter- PEP carboxylase, which functions only to fix CO2. It has
native function that competes with its role in capturing a much higher affinity for CO2 than does rubisco.
CO2. Besides catalyzing the initial step of photosynthe- Because it does not also catalyze photorespiration, PEP
sis, rubisco can also catalyze a process called photores- carboxylase can maintain high rates of CO2 uptake even
piration, in which oxygen is taken up instead of car- at warm temperatures.
bon dioxide (see Box 2A on the preceding page). At After its formation, the four-carbon molecule is
higher temperatures, rubisco increasingly favors the decarboxylated (the CO2 is removed), and the CO2 is
oxygenation reaction over carboxylation, or photores- then incorporated into organic molecules via the Calvin
piration over photosynthesis. Likewise, the higher the cycle. Rubisco functions to fix this internally liberated
concentration of O2 and the lower the concentration of CO2 molecule in C4 plants, just as it acts to fix CO2 com-
CO2 reaching the chloroplast, the more O2 is taken up in ing in from the external atmosphere in C3 plants. There
preference to CO2. These properties of rubisco limit pho- are three different subtypes of C4 photosynthesis, each
tosynthetic CO2 uptake. with its own enzyme for decarboxylation: NADP-me,
The limitations of rubisco are not especially impor- which uses NADP-malic enzyme, NAD-me, which uses
tant for plants whose leaves are shaded, because in their NAD-malic enzyme, and PEP-ck, which depends on
case photosynthesis is limited mainly by light levels PEP carboxykinase. While it has been suggested that
rather than by the efficiency of CO2 uptake. However, there are ecological differences among these C4 subtypes,
for plants growing in warm, bright environments, the the evidence for this is still ambiguous because these
limitations posed by the properties of rubisco can have physiological differences are confounded with other dif-
major ramifications for photosynthetic rates, and ulti- ferences among the species possessing them.
mately for growth. Even under the best conditions, C3 C4 photosynthesis depends on specialized leaf
plants must maintain large quantities of rubisco to main- anatomy (Figure 2.10). In the typical Kranz (meaning
tain adequate rates of photosynthesis. Rubisco, like all “wreath”) anatomy found in C4 plants, there is a spa-
enzymes, contains a substantial amount of nitrogen. tial separation of the C4 and C3 reactions.The initial cap-
Between 10% and 30% of the total nitrogen in the leaves ture of CO2 from the atmosphere takes place in the mes-
of C3 plants is in rubisco. ophyll cells just under the epidermis and adjacent to the
Because of the limitations of rubisco, photosynthet- substomatal cavities, while its incorporation into carbo-
ic rates are also limited by the concentration of CO2 in hydrates via the Calvin cycle takes place deep inside the
the atmosphere. Consequently, at elevated CO2 concen- leaf in the bundle sheath cells. In C3 plants, the concen-
trations, C3 plants can achieve much higher photosyn- tration of oxygen in the chloroplasts is typically about
thetic rates, all else being equal. Plant growers some- a thousand times greater than the concentration of car-
times make use of this response by growing plants in bon dioxide, resulting in substantial rates of photores-
greenhouses with artificially high concentrations of CO2 piration. In C4 plants, rubisco is located (along with the
Mesophyll cell Bundle sheath cell
NADP+ NADP+
Malate Malate
(4-C) NADPH
NADPH
Aspartate Aspartate
(4-C)
OAA (4-C) +
–NH3 ATP
Cell wall
PEP carboxylase RUBISCO ADP
CO2
CO2 NADPH
RuBP
ADP NADP+
PEP
(3-C)
Alanine Alanine ATP
AMP +
–NH3 To sucrose,
starch,
Pyruvate Pyruvate cellulose
ATP (3-C) Calvin cycle
Figure 2.9
The C4 photosynthetic pathway, with the biochemical steps that take place in the
mesophyll cells shown on the left, and those that take place in the bundle sheath cells
shown on the right
(A) Mesophyll cells
Xylem
Bulliform cells Phloem
Stoma
Stoma
(B)
Xylem
Bulliform cells Bundle sheath
Stoma
Phloem
Mesophyll cells 0.1 mm
Figure 2.10 more loosely packed photosynthetic cells in the C3 leaf.
Anatomy of (A) a leaf of Saccharum officinarum (sugarcane, There are large numbers of chloroplasts in the bundle sheath
Poaceae), a C4 grass, and (B) a leaf of Avena sp. (oats, cells (clustered at the outer edge within each cell) as well as
Poaceae), a C3 grass, both in cross section, showing the dif- in the mesophyll cells of the C4 leaf, in contrast with the
ferences in their architectures. Note the tight packing of the absence of chloroplasts in the bundle sheath cells surround-
mesophyll cells in the outer ring surrounding the bundle ing the vascular bundle in the C3 leaf. The bulliform cells act
sheath cells of the C4 leaf, which themselves tightly encircle as hinges to allow the leaf to roll up during drought (see
the vascular bundle (xylem and phloem), in contrast to the Chapter 3). (From Esau 1977.)
26 Chapter 2
BOX 2B
Stable Isotopes and Photosynthesis
Chemically identical carbon com- nates against 13C). PEP carboxylase is depending on the relative proportion of
pounds fixed via C4 photosynthesis dif- more effective at “mopping up” all of carbon fixed initially by rubisco and by
fer from those fixed via the C3 pathway the CO2 molecules it encounters (13CO2 PEP carboxylase.) Because the ranges of
in a subtle way: their ratios of the two as well as 12CO2), and thus discrimi- the 13C values for C3 and C4 plants are
stable isotopes of carbon, 12C and 13C. nates less against 13C. nonoverlapping, the carbon isotope
Most molecules of CO2 in the air contain The relative proportion of the two ratio provides a convenient way to dis-
12
C, the stable carbon isotope with an stable carbon isotopes in carbon com- tinguish between C3 and C4 plants and
atomic weight of 12. Approximately pounds is expressed in reference to a plant tissues. These carbon isotope
eleven out of every thousand CO2 mol- standard (dolomite from the PeeDee ratios also persist in the food chain. The
ecules instead have the slightly heavier formation): fossilized carbon of herbivores that ate
stable isotope 13C, with an atomic (13 C /12 C) 13 12  × 1000 largely C4 plants, for example, can be
 − ( C / C)standard 
δ13C =  
sample
weight of 13. (14C, which is unstable and distinguished from that of those that ate
13 12
thus radioactive, is much rarer still.) ( C / C)standard C3 plants. This property has been used
When the initial capture of CO2 is For terrestrial C3 plants, the mean by paleontologists and archeologists to
made by rubisco, 13CO2 molecules, 13
C value is –27‰ (range –36‰ to distinguish food sources; for example,
being slightly heavier, are dispropor- –23‰), while C4 plants have a mean of it has been used to determine when
tionately left behind, unfixed, in the –13‰ (range –18‰ to –10‰). (CAM grazers first began to depend heavily
atmosphere (i.e., the process discrimi- plants have a range of 13C values, on C4 grasses.
other enzymes of the Calvin cycle) in the bundle sheath

cells, which are not exposed directly to the external
atmosphere. The four-carbon acid OAA travels directly
through thin strands of living tissue, plasmodesmata,
from mesophyll cells to bundle sheath cells, where it is
decarboxylated. The concentration of CO2 in the bundle
Net photosynthetic CO2 uptake (µmol—2 s—1)
sheath cells of C4 plants is an order of magnitude high-

80
er than its concentration in the photosynthetic cells of C3
plants, and the ratio of oxygen to carbon dioxide is great- Camissonia
ly reduced. Thus, rubisco is ‘fed’ a concentrated stream claviformis (C3)
of CO2 molecules, resulting in the effective elimination 60
of photorespiration in C4 plants.
The consequences of overcoming the limitations of
rubisco are enormous (see Box 2B for a discussion of one 40 Amaranthus
palmeri (C4)
less obvious consequence). C4 plants generally have
higher maximum rates of photosynthesis than do C3
plants. The temperature optimum for photosynthesis 20
is usually much higher for C4 than for C3 species (Fig-
ure 2.11). C3 species typically become light-saturated at
levels well below full sunlight, whereas for C4 species 0
photosynthesis often does not become light-saturated 10 20 30 40 50
even in full sunlight (Figure 2.12) because CO2 uptake is Leaf temperature (¡C)
not limited by the high oxygenase activity of rubisco. C4 Figure 2.11
plants contain only one-third to one-sixth the amount of Photosynthetic rates at different temperatures in a desert C3
rubisco that C3 plants have and yet are able to maintain plant, Camissonia claviformis (Onagraceae), that grows mainly
the same or greater photosynthetic rates, resulting in during the winter and early spring, and in Amaranthus
palmeri (Amaranthaceae), a C4 plant from the same habitat
higher nitrogen use efficiency (maximum photosyn- that grows mainly in the summer. The maximum rate of
thetic rate per gram of nitrogen in the leaf). Because the photosynthesis occurs at about 23°C in the C3 plant and at
C4 pathway makes CO2 available to the Calvin cycle about 42°C in the C4 plant. (After Ehleringer 1985.)
70 with great efficiency, the ratio of CO2 fixed to water lost

(water use efficiency) is greater for a C4 than a C3 leaf
at equal stomatal conductance.
60
The price for these advantages is also substantial:
Net photosynthetic CO2 uptake (µmol m–2 s–1)
it takes additional energy in the form of ATP to run the

50
Pleuraphis rigida C4 pathway. When light levels are high and conditions
(C4 grass)
are optimal, the energy invested is more than compen-
sated by the additional photosynthetic gains. At lower
40 light levels, however, this expensive machinery must
continue to be paid for, resulting in potential disadvan-
tages for plants that possess it.
30
Encelia farinosa
Crassulacean Acid Metabolism (CAM Photosynthesis)
(C3 shrub) Crassulacean acid metabolism (CAM) uses essentially
20 the same biochemistry as C4 photosynthesis to overcome
the limitations of rubisco and eliminate photorespira-
tion. However, CAM plants accomplish this in a very
10
different way than C4 plants (Figure 2.13). In C4 plants,
rubisco is found only in the bundle sheath cells, which
are spatially segregated from external air. In CAM
0 500 1000 1500 2000 plants, rubisco is found in all photosynthetic cells.
PPFD, µmol m–2 s–1 Instead of using spatial separation, CAM plants tem-
Figure 2.12 porally separate the capture of light energy and the
Photosynthetic response to light intensity (PPFD, photosyn- uptake of CO2.
thetic photon flux density) in a C4 desert grass, Pleuraphis
rigida (Poaceae), and in a C3 desert shrub, Encelia farinosa
(brittlebush, Asteraceae), from the western United States. At
full sunlight (approximately 2000 µmol m–2s–1), the leaves of
the C3 shrub are light-saturated, while those of the C4 grass Figure 2.13
are not. (After Nobel 1983.) Biochemical reactions involved in CAM photosynthesis. The
reactions all take place in the same photosynthetic cell, but
the reactions shown on the left take place at night, and those
on the right take place in the daylight.
Light energy
Chloroplast Chloroplast
Vacuole Starch Vacuole

ATP, NADPH
RUBISCO
CO2
RuBP
Calvin cycle
PEP
Stomata OAA (3-C) Stomata
open (4-C) closed
NADPH Malic acid Malic acid
(4-C) (4-C)
NADP+
CO2 PEP
carboxylase
CO2
Glycerate Malate
Malate (3-C) (4-C)
(4-C) Pyruvate
PEP (3-C)
(3-C) Starch Cytosol Cytosol
28 Chapter 2
In C3 and C4 plants, stomata are open in the day, plants because water loss is much lower in CAM plants.
when light capture and carbon uptake and fixation Unlike C3 and C4 plants, in which the photosynthetic
occur. CAM plants open their stomata at night. During pathway is obligate (cannot be turned on or off), some
this time, CO2 is captured by PEP carboxylase, and large CAM plants utilize both nighttime CO2 uptake, with car-
pools of organic acids are accumulated in the vacuoles bon fixation via CAM, and daytime CO2 uptake via the
of the photosynthetic cells. In daylight, the light reac- C3 pathway (Figure 2.14). Some species behave in a fac-
tions occur just as they do in other plants, but with the ultative manner, using CAM photosynthesis when water
stomata closed. The organic acids that accumulated is most limiting, and incorporating CO2 during the day
overnight are decarboxylated, providing CO2 to the using C3 photosynthesis when conditions are more
Calvin cycle. Like C4 photosynthesis, CAM photosyn- favorable so as to achieve higher photosynthetic rates.
thesis allows rubisco to function in a high CO2 and low
O2 environment, in which photosynthesis is favored
over photorespiration. Evolution of the Three Photosynthetic
CAM plants must have thick, succulent photosyn- Pathways
thetic tissues with sufficient physical capacity to accu-
mulate large amounts of organic acids overnight. The Phylogeny of the Photosynthetic Pathways
amount of photosynthate that can be accumulated by C4 photosynthesis and CAM have arisen independent-
CAM photosynthesis over a 24-hour period is limited ly from C3 photosynthesis many times over the course
by the amount of space available in the vacuoles of the of evolution. Approximately 2000 angiosperm species in
photosynthetic cells. The temporal separation of the C4 about 18 families use C4 photosynthesis, including both
and C3 reactions also slows maximum photosynthetic monocot and dicot species. Approximately half the
rates. CAM plants generally cannot accumulate carbon species in the grass family, the Poaceae, are C4 plants.
as rapidly as can either C4 or C3 plants. Other families in which C4 is well represented are the
This photosynthetic pathway excels in one area, and Amaranthaceae, Euphorbiaceae, Cyperaceae, and Por-
that is water use efficiency, the grams of carbon fixed tulaceae. Typically, all of the species in a genus are either
in photosynthesis per gram of water lost in transpira- C3 or C4, but there are several genera with both C3 and
tion. Because stomata are open only at night, when tem- C4 members (for example, Atriplex in the Amaranthaceae
peratures are cool and the air is more humid, water use and Panicum in the Poaceae). A few species are inter-
efficiency is far higher in CAM plants than in C3 or C4 mediate between C3 and C4 in their structure and func-
80
A. haworthii
Dark (CAM) CO2 fixation as a percentage of total CO2 uptake
70
A. glutinosum
60 A. undulatum
50
40
30 A. glandulosum
A. arboreum
20 A. tabulaeforme
Figure 2.14
A. holochrysum Relationship between leaf succulence (grams of
10 water in fresh leaves per unit leaf area) and the
percentage of CO2 uptake that occurs via CAM
A. spathulatum among species in the genus Aeonium (Crassu-
laceae). Plants with thicker, more succulent leaves
0 2 4 6 8 10 12 14 16 exhibit a greater proportion of their total CO2
Degree of succulence [g H2O/dm−2] uptake at night, using CAM. (After Larcher 1995.)
Figure 2.15
Flaveria linearis (Asteraceae), a species with
a photosynthetic pathway that is interme-
diate between the C3 and C4 pathways. It
is shown here growing along a beachfront
on the coast of Florida. (Photographs cour-
tesy of P. Teese.)
Flaveria
linearis
tion. Flaveria (Asteraceae), for example, contains sever- CO2 levels at that time, and for eons during the early
al intermediate species (Figure 2.15), as well as species evolution of land plants, were much greater than they
that are completely C3 and completely C4. are now (see Figure 21.7). The evolution of stomata
The steps that have occurred in the evolution of C4 about 400 million years ago was a milestone in the early
photosynthesis are not definitively known for most history of terrestrial plants because they permitted the
species that have evolved this pathway. In the genus regulation of water loss relative to carbon uptake. Stom-
Flaveria, however, it appears that the first step in the evo- ata appear to have evolved from simple pores in the epi-
lution of C4 photosynthesis may have been the initiation dermis of plants that possessed a cuticle and intercellu-
of biochemical differentiation between the bundle sheath lar gas spaces. Even before the evolution of functional
and mesophyll cells, with recycling of photorespired stomata, these pores would have facilitated higher pho-
CO2 (using rubisco to catalyze the recapture of carbon) tosynthetic rates for the plants occupying a given area
in the bundle sheath cells. The next step may have been (Raven 2002).
an increase in the activity of PEP carboxylase, followed During the Mesozoic era (248 to 65 million years
by increasing activity of the other enzymes of the C4 ago), gymnosperms dominated, dinosaurs arose,
pathway. Finally, full anatomical separation of the C4 became dominant, and went extinct, and flowering
and Calvin cycle pathways was achieved. plants (the angiosperms) first appeared. At this time,
The evolutionary transition to CAM from C3 ances- atmospheric CO2 levels were about 3.5 times greater
tors is apparently much easier than the transition to C4, than they are today (1260 parts per million partial pres-
and more than 20,000 angiosperm species in approxi- sure of CO2). These levels would have resulted in rela-
mately 25 families have evolved this pathway, includ- tively high CO2:O2 ratios and limited photorespiration
ing both monocots and dicots. There are also CAM for C3 plants, even at warm temperatures (Ehleringer
species among the ferns, as well as in the primitive vas- and Monson 1993).
cular plant genus Isoetes (Isoetaceae). It is not unusual In the Eocene epoch (54 to 38 million years ago),
for a plant family to have both C3 and CAM members; when angiosperms began to dominate the landscape,
the genus Euphorbia (Euphorbiaceae) includes C3, C4, CO2 levels in the atmosphere became much lower (per-
and CAM species. There are also species that appear to haps 700 parts per million). By the Recent epoch, atmos-
represent intermediates between fully expressed CAM pheric CO2 had dropped to very low levels, measuring
and more primitive CAM cycling (that is, C3 photosyn- between 180 and 280 parts per million over the past
thesis with the ability to recycle CO2 from ordinary cel- 160,000 years. As CO2 levels dropped, C4 photosynthe-
lular respiration at night). sis would have been increasingly favored, particularly
in warm climates. The first evidence for C4 photosyn-
Photosynthesis Through Evolutionary Time thesis comes from plants that existed 57 million years
Photosynthesis first evolved in bacteria in the Precam- ago, at the end of the Miocene epoch. C4 plants became
brian era, perhaps 2500 million years ago. Atmospheric dominant in some environments at about this time
30 Chapter 2
(Ehleringer and Monson 1993), although climate change C4 grasses dominate, or are major components of,
may have played a more significant role in the rise of grassland ecosystems covering vast portions of the
these plants than the decline in atmospheric CO2 (see globe. Several important crop species are C4 grasses (e.g.,
Chapter 21). Today, as industrialization, technology, and maize, millet, and sugarcane). It is no accident that C4
human population growth push CO2 levels up rapidly photosynthesis is so important in the grasses. Grassland
and drastically (see Chapter 22), some ecologists predict environments are often characterized by warm, bright
that plants with C4 photosynthesis may lose their advan- conditions, and even outside of grasslands, grasses are
tages over C3 plants, perhaps leading disproportion- commonly found in high light environments. These are
ately to the extinction of existing C4 species. precisely the conditions under which C4 photosynthesis
Even among C3 species, there may be marked dif- is favored. While there are many C4 annuals, perennials,
ferences in the responses to elevated CO2. In a field exper- and even shrubs, very few tree species use C4 photo-
iment, Travis Huxman and Stanley Smith (2001) com- synthesis; likewise, while there are C4 species that are
pared photosynthetic responses to elevated CO2 in two native to shady, cool, and aquatic habitats, they are the
Mojave Desert species, both with C3 photosynthesis. One exceptions.
of the species, Bromus madritensis ssp. rubens (red brome, CAM plants are found in two distinct habitat types
Poaceae) is an invasive annual grass; the other, Eriogon- with two distinctive growth forms: as succulent terres-
um inflatum (desert trumpet, Polygonaceae), is an herba- trial plants growing in deserts or other arid habitats, and
ceous native perennial. Bromus responded to elevated as epiphytes growing in the canopies of trees in tropical
CO2 by increasing its net photosynthetic rates through- and subtropical habitats. CAM photosynthesis has
out the growing season, as well as by extending the peri- evolved in many different families in arid environments
od of high photosynthetic activity. In contrast, Eriogon- (Figure 2.16), including cacti (Cactaceae), agaves
um experienced increased photosynthetic rates for only (Agavaceae), species in the Euphorbiaceae from Africa,
a brief period at the peak of photosynthetic activity, and and species in the Crassulaceae. Succulents in this group
may have even curtailed photosynthetic activity more often serve as examples of convergent evolution due to
steeply over the growing season in the enhanced CO2 their similar external morphologies (see Figure 6.11).
treatment. Maximum stomatal conductances over the Epiphytes are plants that rely for support on other
course of the year were sharply reduced in the high CO2 plants, generally living on trees, but depend on miner-
treatment for Eriogonum, but were relatively unaffected als deposited from the atmosphere for nitrogen and
for Bromus. The ultimate effects of rising CO2 levels on other mineral nutrients (Figure 2.17A). Epiphytes have
the persistence or spread of these species is as yet unclear, at least one thing in common with desert plants, being
but differences in responses to elevated CO2 could poten- subjected frequently to severe shortages of water. Even
tially alter species’ distributions and community com- in a tropical rainforest, life high in the canopy can be sur-
position over the next century. prisingly dry (see Figure 19.3B). A number of families
have epiphytic CAM species. Some familiar examples
are Spanish moss (Tillandsia usneoides; Figure 2.17B) in
Growth Form, Phenology, and Distribution the Bromeliaceae and many orchids (the family Orchi-
of C3, C4, and CAM Plants daceae) and ferns.
Perhaps surprisingly, CAM plants are rare in the
Growth Forms and Habitats very hottest, driest deserts, where rainfall is not only lim-
The most striking evidence for the adaptive significance ited, but cannot be counted on in any given year. For
of the three photosynthetic pathways comes from pat- example, there are few CAM plants in Death Valley, Cal-
terns of species distribution and abundance. C3 plants ifornia, or in the Sahara of North Africa. In the Ataca-
are by far the most abundant in terms of both number ma Desert of Peru, where it may be years between rains
of species and total biomass. While there are more CAM in some places, cacti are almost entirely restricted to
species overall than C4 species, CAM plants are much the zone that receives moisture as coastal fog. Why
less abundant than C4 plants in terms of both biomass might CAM plants be kept out of the most arid deserts?
and worldwide distribution. CAM plants are generally Under these extreme conditions, rapid growth during
found in particular habitat types, and their importance, the brief, unpredictable periods when water is available
whether it is measured as the percentage of species in a would be strongly favored (see Figure 18.25), and CAM
regional flora or as the proportion of biomass in the veg- plants do not generally grow rapidly. Another factor
etation, is small in most ecosystems. Each photosyn- may be survival during very prolonged drought. Both
hetic pathway, as we have just seen, offers advantages cellular respiration and water loss would be less in a
under particular conditions, but has costs associated completely dormant plant with no aboveground photo-
with it that put it at a disadvantage under other condi- synthetic parts than in a succulent plant. Even with
tions. closed stomata, tiny amounts of water are lost from any
(A) (B)
(C) Figure 2.16

Some CAM species. All of these succulents display different forms of
fleshy leaves or other tissues. (A) Sempervivum tectorum (“hen-and-
chicks,” (Crassulaceae) (Photograph © David Matherly/Visuals
Unlimited.) (B) Sedum sp. (Crassulaceae) (Photograph by S. Scheiner).
(C) Opuntia humifusa (smooth mountain prickly pear, Cactaceae) (Pho-
tograph courtesy of W. S. Judd.)
(B)
(A)
Figure 2.17
CAM epiphytes. (A) A bromeliad, Guzmania mono-
stachya (Bromeliaceae), growing as an epiphyte on
the trunk of a tree. (Photograph courtesy of W. S.
Judd.) (B) Tillandsia usneoides (Spanish moss, Bromeli-
aceae) is an epiphyte that grows hanging from the
branches of large trees in the southeastern United
States. (Photograph by S. Scheiner.)
32 Chapter 2
green tissue, which may add up to a substantial amount Desert of Arizona and northern Mexico, there are two
over an extended time without any water. Succulents, separate rainy seasons, and annual C3 species, includ-
particularly those with CAM photosynthesis, are thus ing many beautiful wildflowers, dominate in the early
most abundant and diverse in deserts with both rela- spring, while C4 species are found in the summer. The
tively higher levels and greater predictability of rainfall. periods of greatest growth for C3 and C4 perennials in
In somewhat drier deserts with less predictable rainfall, these environments follow a similar pattern.
perennials with either C3 or C4 photosynthesis are more Similarly, the “cool season” perennial grasses and
likely to be dominant, depending on the season in which forbs (broad-leaved herbaceous plants) that predomi-
rain is most abundant and likely (see below). In the dri- nate in spring and again in autumn in the mixed-grass
est deserts, the perennials decline in importance and prairie of South Dakota are largely C3 species, while the
annual species tend to become more prevalent. “warm season” species that grow most actively in mid-
summer have a much greater representation of C4 plants.
Phenology However, in the shortgrass prairie farther south, while
The three photosynthetic types also tend to differ in their there are differences in phenology between the C3 and
phenology (the seasonal timing of life history events, C4 species, the period of maximum growth for the two
such as initiation of growth, flowering, and dorman- photosynthetic types does not differ (Hazlett 1992).
cy). C4 photosynthesis is well represented among annu-
al species that complete most of their growth in the sum- Geographic Distributions
mer; such summer annuals are common in semiarid Various studies have considered the broad geographic
environments with seasonal rainfall, as well as among distributions of C3, C4, and CAM species and the climatic
weedy species that colonize disturbed environments, factors associated with their relative predominance. Not
including roadsides and agricultural fields. In areas with surprisingly, the physiological differences between C3
mediterranean climates (such as those of California, and C4 photosynthesis are associated with differences in
southern France, or western Australia, for instance) distribution between plants of the two types. Across
where most of the rainfall occurs in winter, the native North America, the percentage of C4 grass species is
species are overwhelmingly C3 plants, but the weeds most strongly determined by summer temperatures:
that invade fields irrigated in the summer may be C4 C4 grasses are most common where summers are warm
species. (Figure 2.18).
In the southwestern deserts of the United States, dis- Paruelo and Lauenroth (1996) examined the relative
tinct patterns exist for C3 and C4 annual plants. In the abundances (in terms of biomass rather than number of
deserts of eastern California, where the rainy season is species) of C3 grasses, C4 grasses, and shrubs in the west-
in winter and early spring, the annuals are C3 species ern United States (Figure 2.19). C4 grasses are increas-
and grow when temperatures are cool. In New Mexico ingly dominant where temperature, total rainfall, and
and Texas, where there is a summer rainy season, the the proportion of rain falling during summer are high-
desert annuals are largely C4 species. In the Sonoran er. C3 grasses are favored by cool temperatures and by
0%
0
20%
40%
20
Figure 2.18
40 60%
60 Distribution of C4 grass species as a percent-
80% age of the total grass flora across North Amer-
80
ica. (Data from Teeri and Stowe 1976.)
(A)
C3 grasses C4 grasses Shrubs
(B)
Relative abundance key
100 0
Figure 2.19
C3 grasses, C4 grasses, and shrubs are the domi-
C4 grasses C3 grasses nant functional groups in the central part of
North America, which is dominated by grass-
lands and shrublands. The maps indicate the
relative abundances of each as an approximate
percentage of the total biomass. (A) C4 grasses
are the most abundant species in the southern
plains, while C3 grasses predominate farther to
the west and north. Shrubs are most abundant
0 100 in the far west. (B) The combined abundances of
100 0 C4 grasses, C3 grasses, and shrubs, as indicated
Shrubs by the key. (After Paruelo and Lauenroth 1996.)
34 Chapter 2
50 C4 grasses Other research has documented changes with alti-

tude, from a vegetation dominated by C4 species in the
warm, often dry habitats at low altitudes to a C3-domi-
Absent Present nated flora in the cooler, moister conditions at high alti-
40 tudes, in Kenya, Costa Rica, Hawaii, Arizona, and
Shrubs Argentina. CAM species are, as one might expect from
Latitude (°N)
their superior water use efficiencies, most closely asso-

ciated with aridity in temperate and semiarid regions.
Present Absent These studies of correlations between climate and
photosynthetic type all share a common limitation, how-
Present
ever. It is possible that their results are due, to some
30 Absent
unknown extent, to historical phylogenetic factors that
C3 grasses link the distributions of related species to climate and bio-
geography, regardless of photosynthetic type. C4 photo-
synthesis, for example, is most likely to have evolved in
120 115 110 105 100 95 90 85
families that are found in warm environments, and there-
Longitude (°W)
fore C4 species in those families, like their C3 relatives,
Figure 2.20 should have numerous adaptations to living in those
Results of a model with lines showing the latitude and lon- environments, irrespective of the pathway used to fix
gitude where C3 grasses, C4 grasses, and shrubs are predict- CO2. Tropical C4 grasses do not perform well in cold tem-
ed to be absent or present within the region of North Ameri-
ca. The distributions of C4 grasses and shrubs are largely peratures; neither do tropical C3 grasses. These factors
influenced by longitude, while C3 grasses are influenced by make it somewhat difficult to disentangle the adaptive
latitude. The predictions of this model were largely in significance of photosynthetic type from all of the other
agreeement with the distribution patterns shown in Figure adaptations of species of contrasting types. This is a par-
2.19. (After Paruelo and Lauenroth 1996.) ticularly big problem in disentangling the possible eco-
logical significance of the C4 subtypes, which are almost
totally confounded with phylogenetic groups. We discuss
a greater proportion of rainfall during the winter this problem in a more general context in Chapter 6.
months. Shrubs, with deep roots that increase their Nevertheless, there is abundant evidence, derived
access to water stored deep in the soil, are most abun- from a variety of approaches, that C4 and CAM photo-
dant as total rainfall decreases and as the proportion of synthesis are adaptations to environmental factors, and
winter rainfall increases. Consequently, the southeast- we have a fairly clear idea of what those factors are, as
ern portions of the great North American grasslands well as the mechanistic bases for the adaptations. Keep
(west of 90° longitude) are dominated by C4 grasses. C3 in mind that the above discussion concerns general pat-
grasses increase to the north and to the west, and the terns of association between physiology and climate.
westernmost, driest part of this region is dominated by There are many C3 species that live in hot, dry places,
(largely C3) shrubs (Figure 2.20; see also Figure 18.25). and a few C4 species that live where it is cold most of the
Table 2.2 Contrasts between sun leaves and shade leaves of Fagus sylvatica
(beech, Fagaceae)
Character Sun leaves Shade leaves
2
Stomatal density (number/mm ) 214 ± 26 144 ± 11
Leaf thickness (µm ) 185 ± 12 93 ± 5
Leaf area (cm2) 29 ± 4 49 ± 7
Fresh weight (g) 0.5 ± 0.1 0.4 ± 0.1
Dry weight (g) 0.24 ± 0.03 0.12 ± 0.02
Water content (% fresh weight) 53 ± 4 70 ± 5
Total chlorophyll (mg/g dry weight) 6.6 ± 2 16.1 ± 2
Total chlorophyll on a leaf area basis 5.5 ± 1.8 3.9 ± 0.4
(mg/100 cm2)
Source: Lichtenthaler et al. 1981.
Note: Values are means of 9 leaves ± 1 standard deviation.
(A) Epidermis Palisade cells (B) Epidermis
Vascular bundle
100 mm
Figure 2.21
Spongy
Cross sections of leaves of Thermopsis montana (golden banner,
mesophyll
cells Fabaceae) grown in two different light environments. (A) Leaf
grown in sun. (B) Leaf grown in shade. Note in these scanning
Epidermis electron micrographs that the sun leaf is much thicker and that
its palisade parenchyma cells are much longer than those of the
Vascular bundle shade leaf. (Photographs courtesy of T. Vogelmann.)
time. While general patterns may be very revealing, they example of phenotypic plasticity, the ability of an indi-
cannot explain everything, and photosynthetic type is vidual of a given genotype to produce different struc-
only one of a host of factors that determine how a plant tures (such as leaf tissues) or to function differently
interacts with its environment. under different environmental conditions (see Chapter
5). This flexibility in response can also be called an accli-
mation (a type of plasticity that involves potentially
Adaptations to the Light Environment reversible adjustments to environmental conditions).
Not all acclimation to changes in light levels is
Sun and Shade Leaves through the production of different types of leaves. In
A leaf that functions well in bright sunlight usually will the study of northeastern forest understory species dis-
not perform well in deep shade, and the reverse is also cussed above (Rothstein and Zak 2001), the shifts in light
true. Many species of plants produce different kinds of
leaves in the sun and in the shade (Figure 2.21). Leaves
at the top of a tree, for example, may differ in a variety
of ways from those deep within the canopy. “Sun leaves” 4.0
are typically smaller in size (leaf area) and thicker than
CO2 uptake (mg 100 cm–2 h–1)
“shade leaves,” with greater leaf mass per leaf area 3.0
Sun leaves
(LMA, also called specific leaf mass or weight, dry mass
per unit leaf area; the inverse, specific leaf area, or leaf
2.0
area per gram of dry mass is often reported). They also
tend to have higher concentrations of rubisco, chloro-
phyll, and other key components of both the light reac- 1.0
tions and carbon fixation, per unit leaf area (Table 2.2). Shade leaves
In some plants, sun and shade leaves can also have 0
different shapes, with deeper lobing, for example, in
leaves produced in bright sunlight. As a consequence of –1.0
the greater investments made in the machinery of pho- 0 250 500 750 1000
tosynthesis, sun leaves usually have much higher light PPFD (µmol m–2 s–1)
saturation levels and greater maximum photosynthetic Figure 2.22
rates than shade leaves (Figure 2.22). Shade leaves often Photosynthetic CO2 uptake in response to light intensity for
have lower rates of cellular respiration, perhaps because sun leaves and shade leaves of Fagus sylvatica (beech,
there is less “machinery” to maintain, and they may Fagaceae) on a leaf area basis (mg CO2/100 cm2/h), showing
have higher net rates of photosynthesis at low light lev- the midpoint and range for nine different sun leaves and
nine different shade leaves. The sun leaves become light-sat-
els, with lower light compensation points. urated at much higher light intensities, and the maximum
The ability of individual plants to produce different photosynthetic rates are much higher for sun leaves. (After
kinds of leaves in different light environments is an Lichtenthaler et al. 1981.)
36 Chapter 2
compensation point in Viola pubescens and Tiarella cordi- 2500

folia were due to physical and chemical changes within Solar-tracking leaf
Incident photon flux (400–700 nm) (µmol m–2 s–1)

individual leaves, rather than the plant producing dif-
ferent kinds of leaves in different seasons.
2000
Species’ Adaptations to High and Low Light Habitats
Contrasts between the leaves of plants that are native to
Horizontal leaf
high light environments and the leaves of plants from 1500
shaded habitats are similar in many ways to contrasts
between sun leaves and shade leaves produced by the
same individual. Species adapted to bright habitats fre- 1000
quently have costs or specializations that would put them
Vertical
at a disadvantage in deep shade, and the reverse is also east-west leaf
true; in many cases plants could not survive in environ- Cupped leaf
500
ments with drastically different light regimes. For exam-
ple, the high investments in rubisco and chlorophyll and
higher respiration rates that generally characterize the
leaves of plants from high light environments would not 0
6 8 10 12 14 16 18
be sustainable in deeply shaded habitats. Hour of day
Unlike the more general differences between
sun and shade leaves on the same plant, species native Figure 2.23
The amount of light received over the course of a midsum-
to high light or low light habitats also have a number mer day in the desert by a solar tracking leaf, a leaf held hor-
of unique features. One such character is solar tracking. izontally (in which case the light received in the early morn-
Irwin Forseth and James Ehleringer (1982) found that ing and late afternoon is sharply reduced), a leaf held
many desert annuals and some perennials from warm vertically (in which case the light received during the middle
deserts—a high light environment—have leaves that of the day is sharply reduced), and a cupped leaf. The solar
tracking leaf receives at least 38% more light than the fixed
track the sun across the sky during the day. This strate- horizontal or vertical leaves. (After Ehleringer 1985.)
gy increases the amount of light available to the chloro-
plasts over the course of the day (Figure 2.23), thereby
increasing photosynthetic rates and presumably subse-
quent growth rates.
Why should solar tracking be advantageous in the may, however, make the difference between life and
bright, sunny environment of a desert? In desert envi- death.
ronments, water is available for growth for only a small In some species, a velvety sheen is produced by epi-
part of the year, and desert annuals, in particular, often dermal cells, which act as lenses to focus light onto
germinate, grow, and reproduce in a very short time chloroplasts. Lee (1997) studied the intense, metallic blue
before water becomes unavailable and they die. Solar iridescence found in the leaves of many plants in equa-
trackers are adapted to maximize their light capture in torial rainforests and found that it may serve to help
the short growth period available to them. In North them to absorb light more efficiently. In this environ-
American deserts, the proportion of species that are solar ment, light in the photosynthetically most usable range,
trackers becomes greater as the length of the growing 400–700 nm, is particularly depleted because it has been
season gets shorter. We discuss adaptations to desert absorbed by the plants in the canopy above. Working
environments in more detail in Chapter 3. somewhat like the coating on a camera lens, the color of
Some plants that live in deeply shaded habitats also these leaves may enable them to absorb more light at the
have unique features. The small understory species that longest photosynthetically useful wavelengths, which
live in the extreme shade of humid tropical forests pos- are less depleted than the shorter wavelengths. The blue
sess a number of unique adaptations to the quality and iridescence in such leaves is not produced by blue pig-
quantity of light available to them. Light intensities at ments, but is a structural effect (Box 2C).
the rainforest floor may be less than 1% of the PPFD at All understory environments are not uniformly
the top of the canopy. Characters such as velvety or dark, however. The light environment is changing con-
satiny leaf surfaces and blue iridescence are found in stantly, dim one moment and illuminated by the brilliant
widely divergent species in understory habitats (Lee and flash of a sunfleck the next (Figure 2.24). Shade-adapt-
Graham 1986). Each of these characters appears to ed species appear to be capable of using brief sunflecks
enhance the light available for photosynthesis slightly. with unusual efficiency. Studies by Robin Chazdon and
That slight advantage in such an extreme environment Robert Pearcy (1991) and their colleagues suggest that,
BOX 2C
Leaf Iridescence and Structural Coloration
Structural coloration is caused by the that alternatively transmit or reflect blue jays, the blue rump of the baboon,
way light bounces off physical surfaces particular wavelengths of light. and the blue eyes of some humans and
with particular optical properties. It can Similar optical interference or, alter- cats are all due to these optical phe-
be created by simple or very complex natively, scattering of blue light by nomena rather than to blue pigment.
microscopic structures in both plants small particles is responsible for the Blue colors in plants may also be pro-
and animals. In the case of iridescent structural blues found in many animals duced by blue pigments such as antho-
blue leaves, the optical effect is caused as well. The blues of butterfly wings, cyanins. No blue pigment has ever been
by thin-film interference within the leaf, the brilliant blue colors found in some found in a vertebrate.
which results from multiple layers beetles, the amazing blue feathers of
(usually in the cell walls) of materials peacocks and the more modest blues of
while understory plants in both temperate and tropical fixation of CO2). These plants also exhibit a slow loss of
forests are capable of surviving well in very low light, photosynthetic induction (the necessary start-up time
they depend on the energy captured from occasional for plants to reach maximum photosynthesis after expo-
sunflecks to be able to grow and reproduce. A large pro- sure to bright light), so that once “primed” by a sunfleck,
portion (up to half) of the carbon gained by understory they remain capable of more fully responding to other
plants may come from energy captured from sunflecks. sunflecks.
The ability of these forest floor species to capture the The ability of some shade-adapted species to cap-
energy available in brief, unpredictable episodes of ture and use sunflecks and light coming into the canopy
direct bright light depends on several adaptations that at steep angles early in the morning and in the late after-
make them able to respond quickly and with great effi- noon may be enhanced by their having irregularly ori-
ciency to that light. These adaptations include stomata ented grana stacks, in contrast to the neatly aligned
that open even in dim light and a high electron transport grana of sun leaves, which are stacked perpendicular to
capacity (a function of the light reactions) relative to car- the leaf’s surface. That irregular orientation would mean
boxylation capacity (the capacity to carry out the initial that at least some grana would be directly aligned with
Figure 2.24
Sunflecks on leaves of Geonoma cuneata (Arecaceae) in the understory of a tropi-
cal forest. (Photograph courtesy of R. Chazdon.)
38 Chapter 2
incoming light, regardless of the direction from which

the light came. These plants may also have various adap-
tations that prevent them from being damaged by the
intense sunlight in patches and flecks that may briefly
illuminate the deep shade.
Do Sun and Shade Adaptations Exist within Species?

More controversial than the existence of species adapt-
ed to high and low light conditions is the question of
whether, within a single species, there exist ecotypes
(genetically distinct populations within a single species,
adapted to local conditions; see Chapter 6) adapted to
sunny spots and other ecotypes adapted to shaded habi-
tats. This idea was first proposed in the 1960s based on
comparisons of populations of several species collected
from sunny and shaded habitats, including Solidago vir- Figure 2.25
gaurea (Asteraceae), Rumex acetosa (Polygonaceae), Geum Solanum dulcamara (nightshade, Solanaceae), a European
rivale (Rosaceae), Dactylis glomerata (Poaceae), and species that has become a common weedy vine in the east-
Solanum dulcamara (Solanaceae). These studies report- ern and central United States. Its common name hints at its
ed that genetically distinct populations of these plants poisonous leaves and berries. In England it is also called
snakeberry and witchflower, and was reputed to be helpful
existed, with some adapted to high light environments against witchcraft. (Photograph © J. Wexler/Visuals Unlim-
and others adapted to shade, and went on to describe ited.)
the physiological adaptations of these ecotypes in detail
(for example, see Björkman and Holmgren 1966; Björk-
man 1968). The results of this research became widely
accepted. these researchers did find genetic differences in photo-
However, there were several basic problems with synthetic traits among individual plants, the sun and
the research upon which these results were based. We shade populations did not differ from each other on
discuss these problems here not only because this ques- average, and did not differ at all with regard to traits
tion is inherently interesting, but also because this case associated with hypothesized sun/shade differences.
provides some insight into the scientific process and the Rather, all individuals adjusted their physiology and
role controversy sometimes plays in that process. While morphology to their growth conditions. Research has
the original physiological work was innovative and of continued in the effort to find evidence for the existence
high caliber, not enough attention was paid to the pop- of within-species sun and shade ecotypes in these and
ulation-level aspects of the question. First, only one to other species (for example, Sims and Kelley 1998).
three individuals were sampled from each population.
These sample sizes were so small that it was impossi- Day Length: Responses and Adaptations
ble to know whether the differences found were real bio- Another aspect of the light environment that is impor-
logical differences between sun and shade populations tant to plants is day length. Plants can detect, quite pre-
or were due merely to chance. How reliably did these cisely, not only how long the day (or actually, the night)
few individuals represent the populations from which is, but they can also determine whether day length is
they were drawn? A second limitation of the original increasing (as it does in spring) or decreasing (as it does
studies was that the sun and shade populations sampled in autumn). Day length is often a more reliable indica-
were often located far apart (sometimes by hundreds tor of season than is temperature or other more proxi-
of kilometers) from one another. Were the hypothesized mate cues, and plants can make use of this information
differences due to sunny versus shaded habitats or to to time important phenological events, including flow-
other differences between the habitats? ering, the onset of dormancy, or leaf fall. Even many
To overcome these methodological difficulties, John seeds are capable of detecting day length (usually via
Clough and his colleagues (Clough et al. 1979) sampled the seed coat) and using this information to determine
15 individual plants of S. dulcamara (Figure 2.25) that when to remain dormant and when to begin the process
were growing in sunny sites (100% full sun) and 15 indi- of germination. Day length is often used together with
viduals from shady locations (< 10% of full sun) sepa- other environmental cues, including temperature and
rated by only 25 meters in northern Illinois. Cuttings moisture.
were made from each plant and grown in an array of dif- In some species, genetic variation in both the sensi-
ferent light, water, and temperature treatments. While tivity and the nature of the response to day length exists.
In the widespread arctic/alpine species Oxyria digyna

Summary
(alpine sorrel, Polygonaceae), populations from Cali-
fornia, Wyoming, and Colorado flower when the The process of photosynthesis consists of two parts, the
daylength is about 15 hours. In populations from near light reactions, in which light energy is captured, and
the Canada–United States bor- the Calvin cycle, in which carbon is fixed. The rate at
der, some plants will flower in 15 which photosynthesis proceeds depends on the amount
hour daylength, but many will and quality of the light reaching the chloroplasts, the
not. In populations from north of CO2 concentration gradient between the air and the
the Arctic Circle, flowering chloroplasts, and conductance to CO2.
requires a daylength of over 20 There are three different photosynthetic pathways:
hours (Mooney and Billings 1961; C3, C4, and CAM photosynthesis. C3 photosynthesis is
Billings and Mooney 1968). ancestral to the other types and is the most common.
Thus, the onset of flowering C3 plants rely on the enzyme rubisco to capture CO2
occurs in mid-May both at the from the atmosphere. Because rubisco also catalyzes
Canada–United States border and photorespiration, photosynthetic rates can be severely
Dwight Billings north of the Arctic Circle, although limited for C3 plants under warm, bright conditions. C4
at the border daylengths are 15 and CAM photosynthesis are evolutionary specializa-
hours at this time, while at the Arctic Circle they are 20 tions that overcome the limitations of rubisco in such
hours. The more southern, high elevation populations environments. Each offers advantages under particular
do not reach 15 hour daylengths until mid- to late-June. environmental conditions and is at a disadvantage
These alpine populations may still be under snow in under other conditions.
mid-May, with a high probability of frost, and do not Numerous adaptations exist among species adapt-
flower until about a month later than the more northern ed to high or low light environments. These adaptations
populations. confer advantages in the environment to which the
Although plant sensitivity to day length was first species are adapted, but cause them to function more
discovered in the 1920s, its mechanistic basis is still not poorly in other environments. Some individual plants
fully understood, and depends in part on some can produce leaves with different properties in the sun
unknown biological clock. It is known, however, that and in the shade. The ability to use the light available in
phytochromes—bluish pigments that exist in several dif- brief sunflecks is an important feature of many under-
ferent forms that respond to light—play a major role in story forest species. Day length is another feature of the
the photoperiodic timing mechanism. Phytochromes light environment that plants are often capable of detect-
play a role in a number of the ways in which plants ing and which they may use as a reliable indicator of
detect and respond to various aspects of light in their changing seasons.
environments (see Chapter 5).
Additional Readings
Classic References Wookey, P. A., C. H. Robinson, A. N. Parsons, J. M. Welker, M. C. Press,
T. V. Callaghan and J. A. Lee. 1995. Environmental constraints on the
Kramer, P. J. 1981. Carbon dioxide concentration, photosynthesis, and
growth, photosynthesis and reproductive development of Dryas octopeta-
dry matter production. BioScience 31: 29–33.
la at a high Arctic polar semi-desert, Svalbard. Oecologia 102: 478–489.
Mooney, H. A. and W. D. Billings. 1961. Comparative physiological ecol-
ogy of arctic and alpine populations of Oxyria digyna. Ecol. Monogr. 31:
1–29. Additional References
Von Caemmerer, S. and G. D. Farquhar. 1981. Some relationships between Chazdon, R. L. and R. W. Pearcy. 1991. The importance of sunflecks for
the biochemistry of photosynthesis and the gas exchange of leaves. Plan- forest understory plants. BioScience 41: 760–766.
ta 153: 376–387. Jones, H. G. 1992. Plants and Microclimate: A Quantitative Approach to
Environmental Plant Physiology. Cambridge University Press, Cambridge.
Contemporary Research Lambers, H., T. L. Pons and F. S. Chapin III. 1998. Plant Physiological Ecol-
ogy. Springer, New York.
Cushman, J. C. and A. M. Borland. 2002. Induction of Crassulacean Acid
Metabolism by water limitation. Plant Cell Environ. 25: 295–310. Larcher, W. 1995. Physiological Plant Ecology. 3rd ed. Springer, Berlin.
Thomas, S. C. and W. E. Winner. 2002. Photosynthetic differences between Nobel, P. S. 1999. Physiochemical and Environmental Plant Physiology. Aca-
saplings and adult trees: An integration of field results by meta-analysis. demic Press, New York.
Tree Physiol. 22: 117–122.
C H A P T E R
3 Water Relations
and Energy Balance
P
lants, like everything else alive, need water to live, grow, and reproduce.
Like animals, plants use water as the medium in which all of their bio-
chemical reactions occur; unlike terrestrial animals, most plants also rely
on water to a greater or lesser extent for support of their nonwoody parts and
for maintaining their physical structure. Much of the evolutionary history of
plants on land is the story of increasing success in obtaining water, moving it
to parts of the plant far from the source of the water, and being able to repro-
duce in a very dry atmosphere. We briefly review this history, pointing out
some of the essential structures that evolved during the course of the history
of land plants.
Terrestrial plants not only must obtain water, but are also faced with the
problem of restricting their water losses. Intimately related to this problem is
the challenge of maintaining their leaf temperatures within an acceptable range.
In this chapter, we examine the evolutionary “solutions” to these interrelated
problems and take a look at the physics of how plants exchange water and
energy with their environments.
Adapting to Life on Land

Photosynthetic organisms originated in marine environments, where they were
continuously bathed by water in which mineral nutrients were dissolved. The
ancestors of terrestrial land plants were single-celled aquatic organisms that
depended on having each cell in direct contact with water, obtaining nutrients
from and being kept moist by the water in their immediate surroundings. Like
modern photosynthetic aquatic organisms, these ancestors of terrestrial plants
also depended on water for sexual reproduction, releasing their gametes into
the water.
Early land plants did not differ dramatically from their aquatic ancestors
in many of these characteristics. One of the earliest adaptations of the first
land plants was the evolution of a cuticle, a waxy, nonliving covering over the
exposed epidermal cells that prevents desiccation by the air. In bryophytes such
as mosses, for example, water moves by diffusion from cell to cell. There are no
specialized organs to take up water, nor to transport it to distant parts of the
plant. The process is slow and not very efficient; as a consequence, none of the
cells in a moss plant can get very far away from the substrate (soil or rock, for
instance) that provides the water, so mosses are constrained to be very short.
42 Chapter 3
The dominant generation in bryophytes, including the hormonal regulation of germination in response to
mosses, is the haploid gametophyte (see Chapter 8). environmental conditions, as well as for maternal pro-
Male gametophytes produce male gametes, which must visioning of food for the embryo. Seed plants were,
swim to the female gametes. Because these plants are therefore, able to evolve far greater variation in dor-
terrestrial, this is possible only when the moss shoots are mancy and in seed provisioning than their non-seed
covered with a film of water from rainfall or dew. The ancestors.
sperm swim along this film of water from a male plant
to a female plant, eventually making their way to an Water Potential
archegonium containing an egg. This will work suc-
cessfully only under certain restricted weather condi- The ability of a plant to acquire water, the ability to move
tions, and males and females must grow close to each it to all cells throughout the plant, and the propensity of
other. the plant to lose that water all depend on the water
Early vascular plants, which branched off from the potential of the various parts of the plant and its imme-
ancestors of bryophytes in the Paleozoic era (see Chap- diate environment (Figure 3.2). One way of thinking
ter 21), had an enormous advantage over bryophytes for about water potential is that it represents the difference
living in a terrestrial environment: their vascular sys- in potential energy between pure water (which is defined
tems allowed the transport of water much more rapid- as having a water potential of zero) and the water in
ly and efficiently throughout the plant. Because they had some system, such as that in a plant cell or in the soil.
evolved vascular tissues, they were able to develop spe- When water is bound up in some way, or has mate-
cialized structures to take up water from the soil (roots) rials dissolved in it, the potential energy it possesses is
and woody tissue capable of supporting a trunk and less than that of pure water, and its water potential is
crown. Without vascular tissue to take up water and con- negative. Analyzing the movement of water in plants in
duct it over long distances, such structures could not terms of water potential puts the study of plant water
exist. The early vascular plants also evolved a dominant relations into the unifying context of thermodynamics.
sporophyte (diploid) generation. As the sporophyte It makes it possible to study the movement of water in
became larger and longer-lived over the course of evo- soils, in plants, and between plants and the atmosphere
lution, the gametophyte became smaller and shorter- using the same concepts, terms, and units.
lived. However, the plants in these groups still depend- Water potential (Ψ) in plants and soils can be bro-
ed on motile sperm to swim from the male to the female ken down into four major components,
gametophyte, and had not evolved seeds.
Ψ = Ψπ + ΨP + Ψm + Ψg
The seed plants—seed ferns (now extinct), gym-
nosperms, and angiosperms—represent the greatest where Ψπ is the osmotic potential, ΨP is the pressure
advances in the adaptation of plant life to a dry envi- potential, Ψm is the matric potential, and Ψg the gravita-
ronment. The tissue that conducts water in this group tional potential of the water in the system.
of plants, the xylem, is capable of moving large volumes The osmotic potential is the component of water
of water efficiently over long distances (Figure 3.1). Par- potential that is due to solutes dissolved in the water. It
ticularly in the angiosperms—the flowering plants— results from the difference in potential energy between
there is a great diversity of structure and physiological pure water and water containing dissolved substances.
function related to the ability to withstand dry con- Osmotic potential is either zero (in pure water) or neg-
ditions. ative, because the solutes reduce the capacity of the
The evolution of pollination removes many restric- water to do work. The osmotic potential is capable of
tions on the ability to reproduce in the dry conditions acting as a driving force for water movement such as
that dominate terrestrial life, including the requirement when the movement of the solutes is restricted by a
that male and female gametophytes be located adja- semipermeable membrane that allows water, but not
cent to each other and that reproduction occur only solute molecules, to pass through it. The osmotic poten-
under a very limited set of weather conditions. tial is the major component of water potential in living
The evolution of seeds was another key innovation. cells (such as those in roots and leaves), and thus is the
In non-seed vascular plants, such as ferns, dispersal major driving force by which water moves into those
occurs as haploid spores, produced by the diploid sporo- cells.
phyte plant. Both seeds and spores can disperse from the The pressure potential is a second key component
parent plant and can survive various environmental haz- of plant water potential, and is important both in liv-
ards, including dessication. However, seeds include a ing cells and in functioning xylem tissue (which is non-
number of tissues, such as the seed coat, that allow for living at maturity). It can be negative, zero, or positive
the detection of subtle variation in the environment in value, and is a function of the hydrostatic or pneu-
(such as changes in daylength and temperature) and for matic pressure in the system. When water is enclosed by
Water Relations and Energy Balance 43
Simple perforation plate
Scalariform
perforation plate
(A)
(D)
Simple perforation plate (E)

Bordered
pit
(B) (F)
(C)
Figure 3.1
Water-conducting xylem cells from the wood of a gym- Fagus grandifolia (beech, Fagaceae), with an open end wall
nosperm and several angiosperm trees. (A) A tracheid of (simple perforation plate). (E) Xylem vessel element of Popu-
Pinus lambertiana (sugar pine, Pinaceae), bent in several lus trichocarpa (black cottonwood, Salicaceae). (F) Xylem ves-
places to show its full length. Notice that the ends of the tra- sel element of Ailanthus altissima (tree-of-heaven, Simaru-
cheid are not open. (B) An enlarged view of the central part baceae). Moving from A and B to E, the elements become
of the tracheid in (A), showing pits in the side walls. (C) progressively shorter, wider, and have larger openings in the
Xylem vessel element of Liriodendron tulipifera (tulip tree, end walls, all contributing to facilitating the passage of
Magnoliaceae), showing openings in the end wall (the greater volumes of water more rapidly. Scale bars are all 100
scalariform perforation plate). (D) Xylem vessel element of µm. (From Carpenter and Leney 1952.)
something that restricts its volume, such as a healthy plant cell under normal conditions, in which the action
plant cell enclosed by cell walls, the water presses on the of the water is similar to that of air inflating a balloon.
walls, generating a positive pressure potential. This is The opposite situation occurs when water is pulled
the ordinary situation in a fully hydrated, or turgid, through an open system, such as a garden hose or a
44 Chapter 3
Atmosphere
Plant
Soil
–150 –100 –50 –10 –4 –2 0

Water potential (MPa)
(Note change of scale)
Figure 3.2
Typical values (in MPa) for water potentials in the soil, in a potential acts as the driving force for water movement
plant, and in the atmosphere in a mesic environment. The from the soil, into and through the plant, and out to the
curve at the left side of the shaded region indicates the val- atmosphere through transpiration.
ues for atmosphere, plant, and soil water potentials when
the atmospheric relative humidity is low and the soil is dry.
The energy for moving water is provided without
Under these conditions, soil water potentials range from the expenditure of any energy by the plant itself. Ener-
close to 0 at deeper layers to about –1.8 MPa near the soil gy is needed to lift the water up from the soil to the top
surface, plant water potentials range from about –2.0 to of the plant, and to change the less energetic liquid water
about –4.0 MPa, and atmospheric water potentials range molecules inside of the plant into the more energetic
from about –5.0 to –150.0 MPa. The light gray curve at the
right side of the shaded region indicates the values for moist
molecules of water vapor in the atmosphere. The move-
soils and high atmospheric relative humidity; under these ment of water is driven by the gradient in water poten-
conditions, values for soil, plant, and atmosphere would be tial along this pathway, and the energy comes directly
about 0 MPa, 0–0.75 MPa, and –1.0 to < –2.0 MPa respective- from the sun. In this way, plants are able to move vast
ly. Intermediate conditions would result in values falling amounts of water, sometimes over tremendous distances
within the shaded region. (After Etherington 1982.)
(in large trees), at no direct energy cost to themselves.
In a typical plant growing in moist soil and tran-
spiring on a warm day, the values for water potential and
plant xylem vessel. In that case, the pressure potential is its components might look like those in Figure 3.3A.
negative, and actually pulls the walls of the vessel Water moves from the soil into the plant roots, which have
inward. Both positive and negative pressure potentials more negative Ψ, due primarily to the osmotic potential
exist in plants, and both are important driving forces for of the root cells. Then the water is drawn into the xylem
the movement of water within the plant and between vessels, which have still more negative Ψ values, due in
the soil, the plant, and the atmosphere. this case to the negative pressure potential. Water moves
Gravitational potential results from the pull of into the leaves from the smallest veins, and out into the
gravity on water, and it is largely responsible for water substomatal cavity (Figure 3.4). From there, when the
draining from the largest pores of the soil in the first few stomata are open, water moves out into the atmosphere.
days after a saturating rain (see Chapter 4). It is negative Even in humid air (Figure 3.3B), the atmosphere sur-
when the water is moving downward, since the water rounding the leaf has a highly negative water potential.
is losing potential energy. Matric potential is a result of The liquid water molecules ultimately gain the energy
the cohesive force that binds water to physical objects, needed to move into a vapor state from the energy in sun-
such as cell walls and soil particles; it is always negative light, and the water throughout the rest of the pathway is
because water bound to an object has less potential ener- pulled upward by the transpirational stream escaping
gy than free water. The matric and gravitational poten- from the leaves of the plant through the stomata.
tials are generally much more important in soils than
in plants. Transpiration and the Control
of Water Loss
The Soil-Plant-Atmosphere Continuum
Plants transpire vastly more water than they use direct-
Water always moves from larger to smaller values of ly for plant functions. A large rainforest tree with a plen-
Ψ. In plants and in the soil, this usually means moving tiful water supply might transpire as much as a thou-
from a region of less negative water potential to one of sand liters of water in a day (weighing 1000 kg—a ton
more negative water potential. The gradient in water of water lifted from the soil, transported up the trunk,
(A) Air (50% relative humidity): –9.4 MPa (B)
Leaves: –2.3 MPa
–0.5 to –2.5 MPa
–2.3 –1.9 MPa –2.3
Xylem sap:
– 0.5 to –1.5 MPa
Roots: – 0.2 to – 0.4 MPa

–2.3 –2.3
Moist soil: – 0.1 MPa
Figure 3.3
(A) Typical values for water potential from roots to stem –1.5
to leaves and atmosphere in an herbaceous mesophytic
plant, with moist soil and air at 50% relative humidity.
(B) Water potential values in different parts of a Sequoia-
dendron giganteum (giant sequoia, Cupressaceae) tree
that is transpiring rapidly. Here, the soil is moist and the
air humid. The needles on the periphery of the canopy
have the lowest water potentials, and the trunk has a
higher water potential at the base than high in the
canopy. (After Mohr and Schopfer 1995.)
Bundle sheath Mesophyll Xylem
100 µm Phloem
Figure 3.4
Cross section of a leaf of Hordeum vulgare (barley, Poaceae), showing the upper and
lower epidermis, which are each one cell thick; photosynthetic mesophyll cells, and a
vascular bundle with xylem and phloem surrounded by a single-layered bundle sheath.
Stomata are shown by arrows, and the substomatal cavities are visible inside the leaf,
above the guard cells of the stomata. (From Esau 1977.)
46 Chapter 3
and lost through the leaves each day!). Temperate decid- leaves through the stomata, water moves out through
uous trees may lose up to 140 liters of water, but a typ- transpiration. A plant that keeps its stomata closed to
ical conifer loses only 30 liters in a day. While most restrict water loss eventually faces what is essentially
plants transpire fairly modest amounts of water, in one starvation, as it can no longer take in CO2. Angiosperms
study, Salix fragilis (brittle willow, Salicaceae) growing have evolved a greater diversity of mechanisms for cop-
in moist soil lost a remarkable 463 liters per day (Cer- ing with this dilemma, and for living with drought, than
mak et al. 1984). any other group of plants.
The loss of water through transpiration can be
described by a flux equation, much like the diffusion of Strategies for Coping
CO2 into a leaf during the process of photosynthesis: with Water Availability Conditions
Plants experience drought in a variety of ways, depend-
E = gleaf(wv) × (Ci(wv) – Ca(vw))
ing on both environmental conditions where they live
where E is the transpiration rate in mmol H2O/m2/s. and on their adaptations to those conditions. Drought
The term m2 refers to the surface area of the leaf, usual- may be brief or prolonged, mild or extreme, occasional
ly the area of one side of the blade, but including both and unpredictable or regular in its onset and duration. It
upper and lower sides for leaves with stomata on both may occur when conditions are otherwise favorable for
surfaces. The transpirational flux is driven by the dif- growth, or it may occur when it is too cold for plants to
ference between the concentration of water vapor inside grow.
the leaf, Ci(wv), and that in the external atmosphere sur- Plants have evolved various means of coping with
rounding the leaf, Ca(vw), expressed in mol/m3. Some- limited water availability. These evolutionary solutions,
times the vapor pressure of water is used instead of or sets of coordinated adaptive traits, are known as
vapor concentration in this equation, and the differ- strategies. Ecologists have come up with different terms
ence between water vapor in the leaf and the air is called to categorize some of these sets of traits, but it is worth
the vapor pressure deficit (measured in kilopascals, remembering that these categories are designated for
KPa). The vapor pressure deficit of water is related convenience rather than describing fixed, rigidly defined
directly to the water vapor concentration difference. The entities. In reality, each species has a unique set of char-
leaf conductance to water vapor, gleaf(wv), is a measure acters for coping with the conditions in its environment.
of how readily water vapor moves into or out of the Mesophytes are plants that live in moderately moist
stomata, and is defined as the rate at which water vapor soils and generally experience only an occasional, mild
flows into the leaf for a given difference between ambi- shortage of available water. Mesophytes typically tran-
ent and interc0ellular concentration; it is expressed in spire as long as soil Ψ is greater than about –1.5 MPa,
mmol H2O/m2/s and is the inverse of leaf resistance. although some need soils wetter than –1 MPa, while oth-
The leaf conductance for water vapor flux consists of the ers may be able to continue withdrawing water from the
boundary layer conductance (ga), cuticular conductance soil at water potentials as low as –4 MPa. When the soil
(due to the waxy leaf cutical; gc), and stomatal conduc- becomes drier, mesophytic plants close their stomata and
tance (gs) to water vapor. Boundary layer conductance wait until conditions improve. In a typical mesic (moist)
to water vapor is generally high. The thickness of the habitat, that may be a few hours or a few days. If the
boundary layer of air surrounding a leaf depends on leaf drought extends for many days (a few weeks is all that
size and shape as well as on ambient conditions, so it is most of them can tolerate), many mesophytes will begin
not under the immediate control of the leaf. It has little to die. Most crops, ornamental plants, forest trees, com-
effect on impeding water vapor flux. Cuticular conduc- mon meadow and forest wildflowers, and other famil-
tance is very small and varies little: only limited water iar plants are mesophytes. Agricultural crops are gener-
vapor leaves the leaf through the cuticle. In contrast, ally the least drought-tolerant, and need soils that do not
stomatal conductance is highly variable and is under the dry out too much in order to grow well.
control of the plant: it is high when the stomata are open, Aquatic plants grow in standing water, and hygro-
and very low when the stomata close. Stomata are there- phytes are found in permanently moist soils. Plant ecol-
fore very important in controlling the amount of water ogists and amateur naturalists alike have always been
lost by the plant through transpiration. fascinated with the adaptations of xerophytes, plants
The loss of large volumes of water through tran- that live in regions with frequent or extended drought.
spiration does not have negative effects on the plant Halophytes are plants that live in saline soils and pos-
when soil water is freely available. Transpiration is what sess unique adaptations for tolerating salt, as well as
moves water, along with dissolved mineral nutrients, some characteristics in common with xerophytes. Halo-
from the roots to the leaves without the need for the phytes experience a dilemma similar to that of xero-
plant to expend energy. As the soil becomes drier, how- phytes in that plants in both dry and saline soils must
ever, the plant faces a dilemma. As CO2 moves into the be able to extract water from soils with very negative
water potentials. Some xerophytes continue to transpire conditions. Many desert annuals, particularly the short-
in soils that are phenomenally dry, as low as –6 MPa. The lived desert ephemerals, complete their entire above-
halophyte Atriplex polycarpa (desert saltbush, Amaran- ground lives during the brief rainy season characteris-
thaceae) can continue to grow with leaf water potentials tic of many deserts. These plants spend most of their
as low as –5 MPa. Even under well-watered conditions, lives as seeds, which are able to survive extended dry
its leaf water potentials are still at –2 MPa—a condition periods. The seeds of these plants germinate only after
that would be fatal to most mesophytes. prolonged rains, and the plants photosynthesize rapid-
Adaptation to environmental conditions can occur ly, have high rates of transpirational water loss, grow
by means of either a fixed phenotype (in physiology, rapidly, then set seed and die as the rainy season ends.
morphology, or anatomy) or a plastic phenotype (one By “living fast and dying young,” they experience the
that allows an individual plant to change in response to desert as an environment with plentiful water for most
changing conditions). One type of phenotypic plasticity of their lives as green plants. This strategy has been
is the ability to acclimate (acclimatize) to changes in envi- called drought avoidance.
ronmental conditions, such as drought, with either per- Many desert shrubs lose all or some of their leaves
manent or reversible changes in anatomy, physiology, or during the long dry season and grow new leaves after
morphology. Permanent changes might include the pro- the rains begin again (Figure 3.5). These drought-decid-
duction of leaves with different structural characteristics; uous plants are also practicing a form of drought avoid-
once formed, these leaves would retain their structure. ance, minimizing their water loss through transpiration
Reversible changes might include changes in the form of by reducing their leaf area when water is not available.
enzymes that function optimally at different tempera- Drought-deciduous trees, such as Ceiba pentandra (kapok
tures, or changes in the osmotic potential of leaf cells. The tree, Bombacaceae), are also common in seasonal tropi-
ability to acclimate is important for plants because they cal forests, which also experience predictable dry peri-
are not able to move when conditions change: they can ods alternating with rainy seasons.
neither step into the shade when it is hot nor migrate to Many herbaceous perennials in deserts, arid grass-
a warmer climate when it is cold. lands, and other xeric (dry) environments are likewise
largely dormant during the dry season, when most of
Whole-Plant Adaptations to Low Water Availability the living material is found below ground or at the
A plant experiences its environment very differently ground surface. The aboveground parts of many grass-
than one might expect from a cursory look at average es, for example, die back to the ground during the dry
Figure 3.5
Fouquieria splendens (ocotillo, Fouquieriaceae), a drought-deciduous desert plant, in
Joshua Tree National Park, in the Mojave Desert of eastern California. This species may
produce several sets of leaves during a single year. (Photographs courtesy of P. Curtis.)
48 Chapter 3
season. The advantage of dormancy during the dry sea- Root morphology can be adapted to dry environ-
son is, of course, reduction of water loss, enhancing sur- ments in a variety of ways. Roots can be extensive and
vival during the unfavorable dry season. Its major dis- shallow, as they are in many cactus species, enabling the
advantage is that there is a long period of time during plant to take up water after a brief rain that saturates
which the plant cannot photosynthesize, grow, or repro- only the top layers of soil. Other species have roots that
duce. Drought-deciduous plants generally have fairly extend deep into the soil profile, where the soil is slow
high maximum photosynthetic and growth rates during to dry out. Phreatophytes have roots that extend so
the time when they are active, which help to compen- deeply into the soil that they reach the water table, gain-
sate for the time when they are dormant. Many of these ing relatively permanent access to water—the ultimate
plants also have thick roots and underground stems that drought avoidance mechanism.
are capable of storing food for long periods of time. The ratio of dry mass invested in roots to the mass
These structures enable them to survive extended of aboveground tissue (leaves, stems, etc.), called the
drought periods and allow rapid production of leaves root-to-shoot ratio, varies widely among plants. Xero-
and stems when conditions become favorable again. phytes typically have much higher root-to-shoot ratios
Another way in which some species avoid drought than do mesophytes. Consequently, they are able to take
is to grow only in the wettest places within a dry envi- up more water, and they lose less water through tran-
ronment, such as in low spots where water temporarily spiration. The cost of a high root-to-shoot ratio may be
collects after rains, or along temporary streams. Such a reduction in maximum growth rates, both because of
plants typically lack adaptations for extreme aridity and a reduction in the total leaf area, and therefore the
the seedlings will die if their seeds germinate in drier amount of photosynthetic tissue, and because of the
microhabitats. metabolic cost of maintaining the large root mass.
There are, however, plants that are able to live and The form of the whole plant may also be an adapta-
grow under remarkably dry conditions. These drought- tion to limited water availability. Succulents, such as
tolerant species, or true xerophytes, use a variety of cacti, have the capacity to store large amounts of water
means to “tough it out” in the desert. From the whole in their tissues. The succulent tissue may be located in
plant to the cellular and molecular levels, the morphol- leaves, stems, or other plant parts. Cacti use stems instead
ogy, physiology, and anatomy of these plants are adapt- of leaves as their photosynthetic organs. As a result, they
ed for life in xeric habitats. have reduced surface-to-volume ratios, decreasing the
total surface area capable of water loss and increasing the
volume that retains water. Other desert plants also have
photosynthetic stems. For example, in the genus Parkin-
sonia (palo verde, Fabaceae; Figure 3.6) in the Sonoran
Figure 3.6
Parkinsonia microphyllum (foothill palo verde, Fabaceae) (left)
has extremely small leaflets (right), which reduce water loss
by transpiration. Even these leaves are lost during the dry
season, but photosynthesis continues at low rates in the
green bark. (Photographs by S. Scheiner.)
Desert retain their leaves only during the rainy season,
but remain capable of photosynthesis at low rates all year
through their green bark.
In parts of the world where the soil freezes in win-
ter, it can be difficult for plants to obtain liquid water from
the soil for many months. Plants in these environments
can experience severe drought stress in winter. This is one
reason why deciduous trees and shrubs in the northern
temperate zone and further north lose their leaves in win-
ter, and it is one of the factors that limits the distribution
of broad-leaved evergreen species northward.
Physiological Adaptations
When a plant begins to experience a water deficit, a series
of physiological events is triggered. Hormones are pro-
duced and travel throughout the plant, signaling the
onset of various changes in plant function. Cell growth
and most protein synthesis slows, then ceases. As the
plant experiences more prolonged water deficits, the allo-
cation of materials to roots and shoots is adjusted, the
stomata begin to close, and photosynthesis is inhibited.
Leaves may begin to wilt. In some species, older leaves
are allowed to dry out and die, freeing up available water
to preserve younger leaves for a longer period.
In some species, specific measures for osmoregula-
tion are initiated under dry conditions. These plants syn-
thesize certain soluble compounds (proline and other
low-molecular-weight nitrogen compounds as well as
soluble carbohydrates) during drought. The resulting
increase in solutes lowers the osmotic potential of the Figure 3.7
cells, leading to an influx of water by osmosis, which Selaginella lepidophylla (spike moss, Selaginellaceae) is a “res-
prevents turgor loss and wilting. Osmoregulation is par- urrection plant”. During times of extended drought, the
ticularly important in allowing many halophytes grow- plant has developed adaptations that allow it to virtually
stop its metabolism. Although the plant appears to be dead
ing in saline soils (where soil water has a negative water (top), the advent of rain brings it “miraculously” back to life
potential largely due to a negative osmotic potential (below). (Photographs © W. P. Armstrong.)
caused by dissolved salts) to maintain a favorable water
potential gradient. Other halophytes have the ability to
excrete salt.
Some plants, known as “resurrection plants,” have
highly unusual adaptations that allow them to survive At the other end of the water availability spectrum,
extended complete desiccation (Figure 3.7). These plants adaptation to flooding is critical for survival in some
include many lichens and some mosses, ferns, and other habitats (Figure 3.8). Plants may be subjected to varia-
non-seed plants, as well as a number of angiosperms (in tion in the depth to which they are submerged (from
the Scrophulariaceae, Lamiaceae, Poaceae, and Lili- waterlogged soil to total submergence of the shoot) as
aceae). They are found in various parts of the world, but well as in the frequency, season, and duration of expo-
are most diverse and abundant in southern Africa. All sure to flood conditions. Coastal species, such as those
of these different plants survive cellular dehydration by in salt marshes, and species found in floodplains and
a highly coordinated set of processes. Drought-stable in riparian (river and stream) margins are predictably
proteins are synthesized, phospholipid-stabilizing car- exposed to flooding, but the frequency and severity of
bohydrates are incorporated into the cell membranes, flooding in many such locations have been increasing
and the cytoplasm may become gelated. In the dehy- over time (see Chapter 22), and may surpass the tolera-
drated state, metabolism is brought almost to a halt. tion of these plants. Other habitats, including low-lying
Rehydration when water is again available also occurs areas in forests, meadows, and fields, also expose plants
in a highly coordinated manner in which the cell com- to occasional submersion, particularly in sites with clay
ponents are reconstructed step by step. soils that do not drain readily after heavy rains.
50 Chapter 3
Anatomical and
Morphological Adaptations
A wide variety of anatomical and
morphological adaptations en-
able plants to survive and grow
under very dry or very wet con-
ditions. Among the most impor-
tant are variations in stomatal
numbers, arrangement, size and
behavior, which vary widely
among plants. Stomata are the
gateway for most of the water
leaving a plant. Stomatal con-
ductance varies directly with the
pore width (size of the stomatal
opening). Pore width is con-
trolled by the guard cells on
either side of the opening, which
continually change shape, wi-
dening or narrowing the opening
(Figure 3.9). The movement of
Figure 3.8 the guard cells is controlled hor-
A swamp in southern Illinois dominated by Taxodium distichum (bald cypress, Taxodi- monally by the plant; it is not a
aceae. This species produces “knees” (visible in the foreground) that project from the simple physical consequence of
root system upward above the water, to facilitate root gas exchange. (Photograph by
S. Scheiner.) the guard cells passively wilting
as water is lost, as once was
thought.
When the soil is flooded, plants are faced with a num- The guard cells open and close the stomata primari-
ber of conditions that make it difficult to function nor- ly in response to three factors: light, CO2 concentration,
mally, or even to survive. The greatest problem is lack of and water availability. The action of the guard cells dif-
soil oxygen. In ordinary, well-drained soils, oxygen differs among the three photosynthetic pathways: C3, C4,
fuses through the soil to the roots, which use it for respi- and CAM (see Chapter 2). Light directly causes the stom-
ration. In waterlogged soils, the diffusion of oxygen is ata of C3 and C4 plants to open. The partial pressure of
severely diminished because the soil pores are filled with CO2 in the intercellular leaf space has a major influence
water rather than with air. Longer-term exposure to flood- on the guard cells, signaling the stomata to close when
ing results in compact soils, in which larger soil particles it goes up and to open when it goes down. At night,
collapse, leading to a reduction in soil pore space and soils when photosynthesis ceases in C3 and C4 plants, the CO2
that are harder for growing roots to penetrate. The oxy- concentration in the intercellular leaf space increases,
gen that remains in the soil is quickly consumed by plant and the stomata close. During the day, photosynthetic
roots and soil organisms. Oxygen-dependent microbial capture of CO2 reduces the partial pressure of CO2 in the
processes are severely inhibited, and toxic substances leaf, and the stomata are induced to open. In CAM
from bacterial anaerobic metabolism begin to accumulate. plants (see Chapter 2), the partial pressure of CO2 drops
Plants that have evolved to tolerate flooding exhib- at night as it is captured by PEP carboxylase, and the
it a variety of physiological, morphological, anatomical, stomata open. During the day, malate is decarboxylated
and life history characteristics that allow them to func- (CO2 is removed) and accumulates in the leaf, signaling
tion in flooded environments. One physiological mech- the stomata to close (see Figure 2.13).
anism for surviving for some time with little or no oxy- Plant water relations are critical in determining the
gen available to the roots is an increased reliance on behavior of the stomata. Stomata respond to both leaf
glycolysis. The NADH that is generated by glycolysis water potential and ambient humidity (Figure 3.10). A
can be oxidized to NAD by ethanolic fermentation, declining water potential in the leaf will override other
allowing some ATP production to proceed. The oxygen factors, such as CO2 partial pressure, to close the stoma-
deficiency caused by flooding also stimulates a series of ta; clearly, preventing desiccation is of more immediate
metabolic responses, including the production of concern than is maintaining photosynthetic rate.
polypeptides called anaerobic stress proteins, whose The stomata of different plants vary in their sensi-
function is not yet fully understood. tivity to these three factors. Mesophytes, particularly
(A)
Stomatal pore
Guard cell
(B) Figure 3.9

▼ ▼ (A) Scanning electron micrographs of stomata in a leaf of
Allium sp. (onion, Alliaceae). The left micrograph shows the
outside epidermis of the leaf; the stomatal pore is inserted in
the cuticle. On the right, the view is of the inside epidermis,
and shows a pair of guard cells facing the stomatal cavity.
Guard cells control the width of the stomatal pore, and thus
the amount of water leaving the plant. (Photographs courtesy
of E. Zeiger and N. Burnstein.) (B) Cross section of a leaf of
Dianthus sp. (pink, Caryophyllaceae), a dicot mesophyte. The
micrograph shows a larger vein (at right) and several smaller
▲ veins, and palisade and spongy mesophyll tissue. There are
▲
stomata on both the upper and lower surfaces of the leaf
(arrowheads)—a condition more typical of xerophytes than
of mesophytes such as this species. (From Esau 1977.)
100 agricultural species, often close their stomata during the

middle of the day when water losses are greatest, even
Maximum CO2 uptake (percent)
80
in moist soil. The stomata of xerophytes behave differ-
ently from those of mesophytes when the soil begins to
dry. Most mesophytes quickly close their stomata as the
60 soil dries and leaf water potential drops. In contrast,
S Sc many xerophytes are able to keep taking up water from
M X
40 the soil and photosynthesizing deep into a period of
drought (see Figure 3.10). Mesophytes typically have
much greater maximum pore widths, higher maximum
20
stomatal conductances, and higher maximum rates of
photosynthesis than xerophytes (Table 3.1).
0 In environments where dry conditions alternate
–1 –2 –3 –4 –5 –6 –7 –8
with a predictable wet season, some plants are able to
Leaf water potential (MPa)
photosynthesize and grow rapidly during periods when
Figure 3.10 water is available. By maximizing carbon accumulation
Stomata respond to a decline in leaf water potential by clos-
ing. The graph shows the decline in photosynthetic rate, when water is most available and stomata can be opened
expressed as the percentage of maximum CO2 uptake, with without excessive water loss, the plant gains the most
decreasing leaf water potential in plants with different carbon and energy over the course of the year without
strategies. S, succulents; M, mesophytes; Sc, sclerophyllous adverse consequences. Specializations in leaf anatomy
plants (trees and shrubs of mediterranean and other semi- can help plants to achieve high rates of photosynthetic
arid climates), X, xerophytes (which include herbaceous
dicots, grasses, and shrubs of deserts and other very arid carbon uptake during favorable periods.
regions). Succulents and xerophytes have partially overlap- Many plants native to arid habitats are amphis-
ping values. (After Larcher 1995.) tomatous (have stomata on both sides of the leaf; mes-
52 Chapter 3
Table 3.1 Properties of stomata on the lower leaf surface in plants from mesic and xeric environments
Stomatal density Maximum

(number/mm2 Pore length Maximum (mm) Pore area conductance for CO2
Type of plant leaf area) (mm) pore width (% leaf area) (mmol/m2/s)
Mesic environments
Herbaceous plants 150 (200) 15 (20) 5 0.9 (1.0) 425 (700)
(sunny habitats)
Herbaceous plants 75 (100) 23 (30) 6 1.0 (1.2) 130 (200)
(shaded habitats)
Grasses 75 (100) 25 (30) 3 0.6 (0.7) 270 (460)
Tropical forest trees 400 (600) 18 (25) 8 2.3 (3.0) 100 (300)
Deciduous trees 200 (300) 10 (15) 6 0.9 (1.2) 150 (250)
Xeric environments
Xerophytic desert shrubs 225 (300) 13 (15) — 0.4 (0.5) 155 (260)
Sclerophylls 300 (500) 13 (15) 2 0.4 (0.5) 175 (250)
Succulents 33 (50) ca. 10 10 0.3 (0.4) 95 (130)
Evergreen conifers 80 (120) 18 (20) — 0.7 (1.0) 200 (—)
Source: Larcher 1995.
Note: Medians and average maxima (in parentheses) are shown for each stomatal characteristic. While there are broad overlaps between the values for plants
from different environments, stomatal characteristics reflect environmental adaptations in a number of ways. Grasses are found in a wide range of habitats,
from mesic to xeric. Succulents, many of which use CAM photosynthesis, have very different stomatal properties from other plants of xeric environments.
Evergreen conifers tend to live in habitats that are dry due to reduced rainfall or low soil water holding capacities, but have unique leaves (needles) with
unusual characteristics.
ophytes typically have stomata only on their under- increasing the leaf’s resistance to water loss. Transpira-
sides). Over 90% of species in North American deserts tion is also directly reduced because the water vapor
are amphistomatous. Many desert plants are also isobi- concentration difference between the leaf and the air
lateral (have a distinctive, symmetrical internal leaf immediately outside the stoma (within the crypt) is
architecture with palisade mesophyll tissue on both greatly reduced due to the water vapor retained in the
upper and lower sides of the leaf) (Figure 3.11). These crypt.
two features together are likely to be an adaptation to The waxy cuticle that covers the epidermis of the
the high light levels available to plants growing in arid leaves may be especially thick in xerophytes, increasing
habitats. Because intense sunlight can penetrate more cuticular resistance. Cuticular resistance becomes impor-
deeply into the interior of a leaf than can dimmer light, tant when the stomata are closed. A plant with low cutic-
maintaining these packed layers of photosynthetic cells ular resistance could leak considerable amounts of water
deep inside the leaf allows the plant to achieve high over a prolonged dry period, even with closed stoma-
maximum photosynthetic rates. Having stomata on both ta, and preventing such losses is critical for survival in
sides of the leaf reduces the average distance that CO2 an arid habitat. The cuticle may also have a specialized
must diffuse before reaching chloroplasts, which also construction that reflects much of the light falling on the
enhances photosynthesis. leaf.
While xerophytes generally have a greater number The leaves of many species adapted to dry conditions
of stomata per square millimeter of leaf surface area than are sclerophyllous, or tough and leathery. Various struc-
do mesophytes, the average total pore area as a per- tures make the leaves tough and hard, including thick cell
centage of the total leaf area is about half as great for walls (especially in the epidermis) and increased amounts
xerophytes as for mesophytes (Table 3.1). This may be of fibers and other structural tissues. Stronger, tougher
an adaptation for maintaining tight control of water loss leaves serve several functions, from providing mechani-
when the soil begins to dry, while minimizing the dis- cal strength to reduce wilting to protecting against her-
tance CO2 must travel to reach any photosynthetic cell bivory, pathogen attack, and even damage from wind
much as having amphistomatous leaves does. (which is likely to be stronger in open, arid habitats).
Another feature of some plants from arid environ- Many of the features of xerophytes discussed here are
ments is stomata that are sunken below the leaf surface, also, to some degree, characteristic of the plastic respons-
or even located within deep crypts (Figure 3.12). This es of nonxerophytic plants to sunny, dry conditions.
feature has the effect of lengthening the distance that There is a vast array of morphological differences
water must travel to diffuse away from the leaf surface, among the root systems of different plants. Roots, of
(A) (C)
(B)
Water storage tissue
Palisade tissue
(E)
(D)
(F)
(G)
Palisade tissue
Hypodermis
Figure 3.11
Cross sections of leaves with various xeromorphic features.
(A) Sphaeralcea incana (desert mallow, Malvaceae), with mes-
ophyll consisting only of elongated palisade cells; also note
the numerous trichomes (needlelike hairs; see Figure 11.9).
A large midvein is visible in the center of the illustration. (B,
C) Salsola kali (tumbleweed or Russian thistle, Amaran- middle of the leaf. (F, G) Atriplex canescens (four-wing salt-
thaceae). (B) Overall leaf outline. (C) Anatomy of the succu- bush, Amaranthaceae). (F) Overall leaf outline. (G) Leaf
lent leaf, showing large-celled water-storing tissue sur- anatomy, showing isobilateral mesophyll and large, abun-
rounded by a single layer of palisade parenchyma. (D, E) dant hairs on both the lower and upper epidermis. The large
Nerisyrenia camporum (fanmustard, Brassicaceae). (D) Overall midvein is in the center of the leaf, with a cap of thick-
leaf outline. (E) Leaf anatomy, showing low surface-to-vol- walled fibers between it and the upper epidermis. Note that
ume ratio and mesophyll consisting only of elongated pal- the leaves of all these plants have thickened cuticles. (From
isade cells; several small vascular bundles are visible in the Esau 1977.)
54 Chapter 3
Vascular
bundle
Spongy
parenchyma
Fibers
Closed
stomata
(in crypt)
Cuticle Epidermis Stomatal crypt Open stomata

(sunken in crypt)
Figure 3.12
Stomatal crypts in the leaf of Yucca sp. (yucca, Agavaceae). There are two stomata within
each crypt in this illustration; the ones in the crypt at the left are closed (the substomatal
chambers are to the left and right of the crypt) and the ones in the crypt at the right are
open. Other xerophytic features of this plant include thick leaves, large bundles of
fibers, and a thickened, waxy cuticle. (After Weir et al. 1974.)
course, are what most vascular plants use to obtain inside the plant. Oxygen from the external air, as well as
water, and the root systems of different species vary oxygen released by photosynthesis, diffuses to the roots
according to taxonomic group as well as the environ- through the aerenchyma, while other gases (such as eth-
ment to which the species is adapted and, to some ylene and methane) escape from the roots and out
extent, the environment in which the individual plant through the aboveground parts of the plant. Other plants
finds itself growing. There are a number of different growing in inundated soils, such as some mangroves,
basic kinds of root systems. Fibrous root systems form produce specialized aboveground roots, or roots that
a dense network, capable of exploring a large volume of bend upward and emerge from the soil into the air, pre-
soil for water and nutrients. Most grasses, and many sumably to obtain access to oxygen (Figures 3.15, 3.8).
other monocots, have fibrous root systems (Figure 3.13). Xylem, the tissue that carries water throughout the
Other species, generally dicots, have a single taproot that plant in vascular plants, also differs greatly among
extends deep into the soil, and may also be thickened plants depending on their phylogeny and their ecolog-
and capable of storing food. ical adaptations. In all plants, mature xylem cells are
Roots also possess various anatomical adaptations dead, and the cell contents have been removed to pro-
to the environments in which they grow. Roots have duce hollow vessels through which water is transport-
received less study than leaves, perhaps because they ed. The size, shape, and connections between xylem ves-
are less apparent. One of the most dramatic adaptations sels differ among taxa, and these differences strongly
of root anatomy to habitat conditions is the aerenchyma affect their function. Xylem evolution illustrates the
(aerated tissues) of aquatic plants and other plants adaptation of plants to moisture availability and loss, to
adapted to flooded conditions (Figure 3.14). The for- the predictability of water stress, and to the need for
mation of aerenchyma is a way of avoiding the low oxy- mechanical strength
gen conditions of flooded soil by providing an inter- The water-conducting tissue in earlier-evolved vas-
connected system of air channels for gas movement cular plants (ferns, horsetails, lycopods, and the like)
50 cm
Vascular bundle Air channel
Figure 3.14
This scanning electron micrograph shows the aerenchyma tis-
sue of a species of Nuphar (yellow water lily, Nymphaeaceae).
The air channels allow gases, including essential oxygen, to
move through plants growing in inundated conditions. (Pho-
tograph by J. N. A. Lott/Biological Photo Service.)
100 cm
Gymnosperms also depend on tracheids, although

their tracheids display various advances over those of
the earlier vascular plants. Angiosperms sometimes
have tracheids as well, but they have also evolved anoth-
er kind of cell called xylem vessel elements (see Figure
3.1). These cells make up what are essentially long water-
conducting tubes called xylem vessels (Figure 3.17). In
comparison with tracheids, vessel elements are shorter,
and they may be fully open at the ends because of the
150 cm
Figure 3.13
The fibrous root system of Triticum aestivum
(wheat , Poaceae), a cultivated monocot species.
The shoot of this plant has been chopped off just
above the soil surface. (After Mohr and Schopfer
1995.)
consists of elongated cells called tracheids

(Figure 3.16). These cells are connected to one
another by the thinner parts of their end
walls, through which water diffuses. Water
Figure 3.15
transport in such a system is limited; much Aerial roots of Rhizophora mangle (red mangrove, Rhizophoraceae),
less water is transported, and it moves more which act as props and as a means of channeling air to the sub-
slowly, than in more recently evolved taxa. merged portions of the roots. (Photograph by S. Scheiner.)
56 Chapter 3
Tracheid Vessel elements
Primary
wall
Secondary
wall
Perforation
Figure 3.17
Xylem vessel elements from four different
angiosperms, arranged from the earliest to
evolve (left; similar to Magnolia) to the most
recently evolved (right; similar to Quercus
[oak]). The earliest vessel elements are long
and thin, and closely resemble tracheids (see
Figure 3.16). The evolutionary trend has been
toward shorter, wider elements with thicker
cell walls, which can conduct much larger
Figure 3.16 volumes of water. Because of the greater vol-
Structure of a tracheid (left) compared with that of a vessel umes of water, these wider vessel elements
element (right). The tracheid is not drawn to scale, and is are under greater stress; additional structural
approximately 50 times longer than shown in relation to the support is provided by surrounding fibers.
vessel element. Note that the outer (primary) wall of the tra-
cheid is whole, but that of the vessel element is perforated
and open to water. The secondary wall is laid down inside of
the primary wall, and strengthens it. (After Mauseth 1988.)
tain their leaves at temperatures at which they can carry
out metabolic processes optimally, or at least adequate-
lack of a primary cell wall, allowing more water to move ly, most of the time. How can a plant manage to do this?
rapidly from roots to shoots. The xylem vessel elements One does not ordinarily think of plants as being capa-
of dicots from dry environments tend to be particularly ble of regulating their own temperatures. Animals can
short and narrow, with exceptionally thick walls. This shiver or bask in the sun; they can retreat to the shade
structure offers an adaptive advantage for conducting or to an underground burrow when it is too hot, too
water under the very negative pressure potentials nec- sunny, or too windy—but none of these options is avail-
essary for maintaining a favorable water potential gra- able to plants. What is possible for a plant? To understand
dient when water is being extracted from very dry soils. this better, we need to look at the way plants, particularly
The extreme negative pressure that the xylem is exposed leaves, interact with their energy environment.
to in these xerophytic plants would be enough to col- All physical objects in the universe are continually
lapse the vessels of species not possessing this special- exchanging energy with their surroundings, and plants
ized kind of xylem. are no exception. The factors that determine the tem-
perature of any object, such as a leaf, are best understood
in terms of the energy balance (also called the energy
The Energy Balance of Leaves
budget) of the object. Temperature is a measure of the
The survival of any living thing depends on maintaining average random kinetic energy of the molecules in a sub-
a temperature that is neither too cold nor too warm for stance. Calculating the energy balance of an object
its tissues and organs to function. Clearly, plants must means accounting for the amount of energy absorbed by
avoid freezing or cooking. They must, in addition, main- the object, the amount of energy leaving it, and the
amount of energy stored within it. To do so, we must Araceae; see Box 8A) and Symplocarpus foetidus (skunk
first take a more specific look at the components of the cabbage, Araceae), that are capable of generating sub-
energy balance of an object such as an organism. We take stantial metabolic energy. In such cases, additional terms
an intuitive rather than a mathematical approach to this for heat storage and metabolic heat must be included in
topic (see Campbell 1977 and Nobel 1983 for a detailed the energy balance equation.
look at the physics of the energy balances of organisms). We now explore what these terms mean, and what
In order to get an intuitive feel for the components determines how large or how small they are under dif-
of the energy balance of an organism—a groundhog, a ferent conditions. We also discuss what a plant can do
leaf, or yourself—we begin by describing an ordinary to modify these terms, on an immediate basis or over
experience; precise definitions will follow. Imagine that evolutionary time, to control its exchange of energy with
you are sitting on a rock at the beach on a warm sum- the surrounding environment.
mer day, soaking up the rays. Your skin is warmed by
the sun’s radiant energy. You can also feel heat reach- Radiant Energy
ing your skin by conduction—direct transfer—from the Radiant energy is that energy transferred from one object
rock you are sitting on, which has also been warmed to another by photons, discrete packets of energy that
by the radiant energy of the sun. A breeze cools you by travel at the speed of light. Physical objects are contin-
convection—heat transfer via a fluid medium. ually absorbing and emitting photons. The emission of
Now suppose you jump into the water. The water a photon (by an electron undergoing a transition in ener-
is cooler than you are, so you feel chilly. You do not get gy state) produces electromagnetic radiation at a single
cold as quickly as a smaller animal would, however, wavelength. The photon flux from a source—the amount
because your greater mass stores more heat energy. You of energy emitted—depends on the wavelengths of the
swim vigorously, and the metabolic energy produced photons emitted, with shorter wavelengths having high-
by your muscles burning calories stored from food er energy levels.
warms you up. When you get out of the water, the evap- The radiant energy emitted by the surface of an
oration of water from your skin makes you feel chilled: object, per unit surface area, is given by the Stephan-
you are experiencing latent heat loss. Boltzmann equation,
This scenario summarizes the major terms of the
Φ = εσT4
energy balance equation. An understanding of the val-
ues for these terms can explain many aspects of the way where Φ is the energy emitted (the emitted flux densi-
organisms exchange energy with their environments ty, in W/m2), ε is the emissivity of the surface, σ is the
and their adaptations for maintaining appropriate tem- Stephan-Boltzmann constant (5.67 × 10–8 W/m2/K4), and
peratures. T is the absolute temperature in degrees kelvin (K; equal
Most leaves are too small to store appreciable to the temperature in °C + 273). Emissivity is a measure
amounts of energy, and unlike animals, most plants do of how efficient a body is at emitting energy. If an object
not generate much metabolic heat. Because it is usually absorbs all the radiation that falls on it, and emits, or
unnecessary to account for metabolic energy and ener- radiates, all of the energy possible for an object at its tem-
gy storage, it is relatively simple to calculate the energy perature, it is called a perfect blackbody radiator. The
budget of a typical leaf. The energy balance equation for emissivity of a perfect blackbody radiator is therefore
a leaf at constant temperature can be written as 1.0. No real objects are perfect blackbody radiators over
the entire electromagnetic spectrum, but most natural
Rn – H – λE = 0
objects have emissivities of between 0.90 and 0.98
In other words, Rn, the net amount of radiant heat ener- (although this can vary quite a bit according to the wave-
gy coming into the leaf, minus H, the sensible heat (con- length of the emissions).
duction + convection) loss, and λE, the latent heat loss, Notice that one of the most important quantities in
is equal to zero: the energy gained by the leaf is exactly the Stephan-Boltzmann equation is temperature,
balanced by the energy lost. If there is more energy com- because it is raised to the fourth power. Thus, the sur-
ing in than going out, the temperature of the leaf will face of the sun, which has a temperature of approxi-
rise; if more energy is leaving the leaf than coming in, its mately 6000 K, emits 73,000 W/m2; the sun is close to
temperature will decline. being a perfect blackbody radiator. The energy reaching
While these terms are generally adequate for leaves, the survace of the Earth from the sun is reduced, among
they are not sufficient for determining the energy bal- other things, by the scattering of the photons by parti-
ance of certain massive plant organs, such as the thick cles in the atmosphere (Box 3A; see also Figure 18.1). The
stems of some cacti, which can store considerable heat sun emits the greatest number of photons in the visible
energy. There are also a few unusual plants and plant range of the spectrum, with the peak at yellow light
organs, such as Amorphophallus titanum (voodoo lily, (explaining why the sun looks yellow), although more
58 Chapter 3
BOX 3A
Why the Sky Is Blue and the Setting Sun Is Red
Some of the visible radiant energy from scattered blue light from Earth’s sur- the reds and oranges of a sunset. In
the sun is scattered before it reaches the face, the sky appears blue to us. addition, large-particle scattering of
Earth’s surface. The gas molecules and At day’s end, the angle of incoming light is most effective at longer wave-
small particles of the atmosphere scat- sunlight is much steeper relative to the lengths. That is why volcanic dust, par-
ter light primarily at short wavelengths Earth’s surface, and blue light is refract- ticulate pollution, and other sources of
(Rayleigh scattering). This explains ed (bent) high above the surface by the suspended material in the atmosphere
why the sky is blue: blue wavelengths atmosphere. With the blue wavelengths can result in particularly intense red
are the shortest visible ones common in removed, what remains are the longer sunsets.
the solar spectrum. When we see this (i.e., red) wavelengths, which produce
energy comes from higher-energy, shorter

wavelengths in the ultraviolet (UV) range.
The Earth, in contrast, has an average tem-
perature of about 290 K and emits about 400
W/m2 in the infrared part of the spectrum.
You may be surprised to learn that the
Earth radiates energy. It does, as do all Sun
objects in the universe, including you and the SWdiffuse
leaves of plants. These wavelengths of these
emissions are not in the visible spectrum,
however, because Earth’s temperature (as SWdirect
well as your temperature, and that of a leaf)
LWoutgoing
is too cool to emit visible light.
The net radiation of a leaf (Rn) accounts LWinput
for all of its incoming and outgoing energy
SWoutgoing,
(Figure 3.18). In sunlight, Rn is positive, but reflected
at night it will often be negative as the leaf
radiates more energy than it absorbs from its
LWoutgoing
surroundings. By convention, we divide
incoming radiation into shortwave radiation
SWreflected
(wavelengths < 700 nm, including visible
and ultraviolet wavelengths) and longwave
radiation (infrared and longer wavelengths LWinput
> 700 nm). Shortwave radiation includes
direct sunlight shining on the leaf, light
reflected from the surroundings, and diffuse
shortwave light from various sources,
including the background light from the sky.
A leaf held directly perpendicular to the sun- Figure 3.18
light will receive more shortwave energy The radiant energy flux of a leaf held horizontally on a warm, sunny day.
than one held at a steep angle (Figure 3.18). Longwave radiant energy is indicated by LW, shortwave radiant energy by
Longwave inputs include the radiant energy SW. The leaf absorbs direct and diffuse shortwave radiant energy from the
emitted by the ground, the sky, and other sun, as well as shortwave energy reflected from nearby objects, such as the
ground (or other leaves). It also absorbs longwave radiant energy from the
objects surrounding the leaf. The leaf absorbs ground and the sky. It emits longwave radiant energy upward and down-
only some of the radiant energy impinging ward. The leaf also reflects some of the shortwave radiant energy that
on it, depending especially on its own sur- reaches it. (After Campbell 1977.)
proportional to the organism’s resistance to heat trans-
Shortwave fer. This relationship can be modeled using flux equa-
radiant energy Latent heat
tions. Heat is transferred from warmer objects to cooler
ones. The molecules in a warmer solid object vibrate
faster than those in a cooler one (molecules or atoms in
Latent heat
warmer liquids slide around over one another more rap-
idly, and those in gases bounce around more rapidly in
three-dimensional space). When the molecules of two
Convection
Leaf boundary objects are in contact, the molecules with more kinetic
layer energy speed up those with less. In the process, the faster
molecules slow down a bit, until eventually both objects
Longwave have the same temperature.
radiant energy
Conduction Conduction is the direct transfer of heat energy from
the molecules of a warmer object to the molecules of a
cooler one. It occurs when objects of different tempera-
Longwave
radiant energy tures are in direct contact with each other. Conduction
is a slow and fairly ineffective way of transferring ener-
gy, and it is not usually very important in either water
or the atmosphere. It may be important in the soil, how-
ever, as well as inside leaves, in the boundary layer of
air surrounding a leaf, and sometimes in other parts of
Longwave the plant under certain conditions.
radiant energy
Convection is heat transport by a packet of fluid (air
Conduction
or water) moving as a unit. It is much more rapid and
effective than conduction in transferring heat energy
between organisms and their environment. If your body
is warmer than the surrounding air, you experience con-
vective heat loss as a cooling breeze. That is why wind
chill is important in how you experience the cold: the
greater the wind speed, the more rapidly your body is
cooled by convective heat loss. In unusual environments
Figure 3.19 in which the surrounding air is warmer than your body,
Major components of the energy balance of a grass leaf,
showing the major sources of energy entering and leaving the breeze will feel like a blast of heat from a furnace.
the leaf. The leaf has a net gain of shortwave radiant energy In this case, you are experiencing convective heat gain.
from the sun, and it also gains longwave radiant energy The fur of mammals and the feathers of birds can
from the atmosphere and its surroundings (especially the offer considerable resistance to convective heat exchange,
ground). It loses energy by convection, by conduction (pri- but in general, leaves and other plant parts do not. How-
marily through the boundary layer and underground to the
soil), and by latent heat loss from transpired water, and it ever, leaves are surrounded by a blanket of relatively still
radiates energy out to its surroundings in the longwave part air, called the boundary layer, that surrounds objects
of the spectrum. The boundary layer of the leaf—a blanket in air or water. Heat (and mass) transfer across a bound-
of still air that surrounds it and reduces convective heat ary layer is much less efficient than heat transfer else-
loss—is thickest at the central part of the leaf. where in the fluid. The thicker the boundary layer, the
greater the resistance to heat transfer posed by a leaf.
The boundary layer of a leaf is determined mainly
by wind speed and by the size and shape of the leaf. The
face characteristics. The remainder is reflected back to greater the wind speed, the more eroded the boundary
the sky or toward other objects in the environment. layer becomes. Larger, rounder, more entire leaves carry
a thicker boundary layer than leaves that are smaller,
Conduction and Convection narrower, or have more dissected (broken-up) shapes.
Sensible heat exchange—energy exchange that results The characteristic dimension of a leaf is a measure that
in a change in temperature—includes two major com- takes into account both its size and shape and represents
ponents, conductive and convective heat transfer (Fig- the effective width of the leaf with respect to fluxes of
ure 3.19). The transfer of sensible heat between an organ- energy and mass entering and leaving the leaf. It is
ism and its surroundings is directly proportional to the directly proportional to the boundary layer resistance of
temperature difference between them, and inversely the leaf.
60 Chapter 3
Latent Heat Exchange 50
Latent heat exchange is the transfer of energy that Air temperature = 30 °C
occurs during evaporation—the process of converting Wind speed = 1 m/s
water from a liquid to a gaseous state. (It is called Relative humidity = 20
latent—concealed—because there is no change in tem-
Leaf temperature (°C)

40
perature during this energy transfer process, in contrast /m
30 00 s
with sensible heat exchange, which can be “sensed” due 2 , rs =
W/m
to a change in temperature.) It is equal to the product = 1 400
S W in p u t
of the evaporation rate, E, and λ, the latent heat of vapor- 2 , r = 200 s/m
SWinput = 1400 W/m s
ization of water (2.45 MJ/kg at 20°C). Like convection, 30 SWinput = 0, r = 3
latent heat exchange can be a highly efficient cooling s 000 s/m
SWin
put = 0, r =
mechanism for organisms. The molecules in a substance s 200 s/
m
in the gaseous state have much more energy than those
in the liquid state at the same temperature. At 20°C,
water vapor has about 2450 joules per gram more ener- 20
1 2 5 10 20 50 100
gy than liquid water. In contrast, warming liquid water
Leaf characteristic dimension (mm)
by one degree takes only about 4 joules per gram. Where
does all of the extra energy come from as water is trans- Figure 3.20
formed from liquid to gas in the process of evaporating. Leaf temperatures at a constant air temperature of 30°C,
It is taken from—and thus cools—the surface that is los- with a constant wind speed of 1 m/s (a fairly stiff breeze)
ing the water: the tongue of a panting animal, the skin and relative humidity of 20%. Leaf temperatures were calcu-
of a sweating athlete, or the tissue of a transpiring leaf. lated for leaves of increasing characteristic dimensions
(approximately equal to leaf widths) at different shortwave
radiant energy input levels (SWinput, expressed in W/m2,
Putting It All Together: Leaf Temperature watts per square meter of leaf surface) and stomatal diffu-
The ability of a leaf to survive and function depends on sive resistances (rs) expressed as s/m, seconds per meter).
its ability to maintain its temperature within an accept- Small leaves (those with low values for leaf characteristic
able range. The leaf temperature, in turn, depends on the dimension) remain close to air temperature, regardless of
sunlight intensity and stomatal resistance (which affects
energy balance of the leaf. The large amount of radiant water loss, and thus latent heat loss), while large leaves can
energy input to a leaf in direct sunlight, for example, become much hotter or much cooler than the air, depending
must be balanced by adequate convective and/or latent on shortwave input and stomatal resistance. (After Camp-
heat loss. Leaves that have a large characteristic dimen- bell 1977.)
sion and a high boundary layer resistance have the
potential to become many degrees warmer or cooler
than leaves with a small characteristic dimension (Fig-
ure 3.20). Thus, on a short time scale, plants (at least those
A plant with large leaves in bright sunlight (large with larger leaves) can directly control latent heat loss
radiant energy input), still air (minimal convective heat from their leaves, and thus leaf temperature, by con-
loss), and dry soil is faced with a critical dilemma. Open- trolling transpiration rate. Over long periods of time,
ing the stomata to allow cooling through latent heat adaptation to warm, dry environments can involve the
exchange risks severe wilting and even death. Stomatal evolution of leaves of different sizes and shapes, chang-
closure to restrict water loss will also reduce latent heat ing boundary layer properties and altering convective
loss, potentially causing the leaf to heat to lethal tem- heat exchange. Many desert plants, for example, have
peratures. With adequate water available and even a small or narrow leaves with reduced boundary layer
light breeze, however, latent and convective heat losses resistance to heat transfer, resulting in leaves that remain
can cool a large leaf to well below air temperature. close to air temperature even when the stomata are
Small leaves are likely to remain close to air tem- closed.
perature, even with closed stomata, as the slightest Radiant heat input is also, to some extent, under the
breeze will minimize their boundary layer resistance. control of the plant. The leaves of many desert shrubs
Under windy conditions, leaves of any size will have are held at a steep angle (Figure 3.21), reducing radiant
substantial rates of convective heat exchange. Con- heat gain. There are also plants that change leaf angle in
versely, at equally high stomatal conductances, a warm response to leaf temperature. Leaf coverings such as
leaf will lose a lot more water than a cool leaf because reflective white pubescence (trichomes) and shiny, waxy
water evaporates more quickly from a warm surface (the coatings serve a similar function, reducing radiant heat
warm surface supplies more energy to the evaporating input to the interior of the leaf (at the cost of reducing
water). the light available for photosynthesis).
Summary
Plants are descended from organisms that lived in the
oceans. Their ability to live on land has required the evo-
lution of anatomical, physiological, and other adapta-
tions. The ability of a plant to obtain water, and the
movement of water throughout the plant, depends on
the water potential of plant tissues and of the sur-
rounding soil and atmosphere. Water potential consists
of various components, of which osmotic potential and
pressure potential are generally most important within
the plant. In any system, water tends to move from com-
ponents with less negative water potential to compo-
nents with more negative water potential, so we can
predict the direction of water movement if the water
potential of various components is known.
Plants lose much more water in transpiration than
they use metabolically. Plant strategies for taking up and
coping with the loss of water differ greatly depending on
the species and the environment to which it is adapted.
Mesophytes, aquatic plants, hygrophytes, xerophytes,
halophytes, phreatophytes, and desert ephemerals all use
Figure 3.21 different strategies to obtain water and restrict its loss.
Leaves of Simmondsia chinensis (jojoba, Simmondsiaceae), a
desert shrub, in Joshua Tree National Park, in the Mojave
Desert. The thick, leathery leaves are held at an upright
angle to reduce radiant heat gain. (Photograph courtesy of
P. Curtis.)
Some cacti have dense coverings of highly reflective

spines; while these structures may also act to deter her-
bivores, their primary benefit is reducing radiant heat
input. Encelia farinosa (brittlebush, Asteraceae) is a desert
shrub that produces two different kinds of leaves,
depending on the season. In summer, the plant puts out
densely pubescent and highly reflective leaves. In
spring, when temperatures are cooler and water is more
available, it produces green, nonpubescent leaves, which
absorb much more shortwave radiation and conse-
quently have higher photosynthetic rates than the reflec-
tive summer leaves (Figure 3.22). Other plants also pro-
duce different types of leaves in different seasons, each
of which has morphological or physiological adaptations
for functioning optimally in its own season.
A common adaptation of many plants is to roll their
leaves into narrow cylinders when facing water stress.
Figure 3.22
This change serves to modify the leaf energy balance in Variation in the appearance of leaves of Encelia farinosa (brit-
several ways, minimizing both boundary layer resist- tlebush, Asteraceae). The leaves in the center of the plant are
ance and radiant heat input. Also, rolling up the leaf being produced as the environment is becoming warmer
directly reduces total water loss by reducing the amount and drier; those surrounding it were produced earlier, when
of leaf surface exposed to the dry surrounding air. There conditions were cooler and moister. The central leaves are
thicker, smaller, and covered with a dense white pubescence
are also many whole-plant adaptations for managing (hairs) that reflects sunlight; those toward the outside are
water and energy losses, including partial or total shed- greener because they lack the thick pubescence. (Photograph
ding of leaves in dry or cold seasons. courtesy of S. Schwinning.)
62 Chapter 3
There are a rich variety of physiological, morphologi- of the leaf energy balance and together determine leaf
cal, and anatomical adaptations to the typical range of temperature.
water conditions experienced in an environment. Plants are often perceived as passive receptors of
Plants are continually exchanging energy with their environmental conditions. Yet, the long-term adapta-
surroundings. The temperature of leaves and other plant tions of plant species and the shorter-term functions of
organs depends on their energy balance: the difference individuals modify the actual values they experience for
between the energy taken up and the energy lost. Radi- many critical environmental factors affecting them,
ant energy, sensible heat exchange (conduction and con- including temperature, light, and moisture.
vection), and latent heat loss are the major components
Additional Readings
Classic References Holbrook, N. M. and F. E. Putz. 1996. From epiphyte to tree: Differences
in leaf structure and leaf water relations associated with the transition
Ehleringer, J., O. Björkman and H. A. Mooney. 1976. Leaf pubescence:
in growth form in eight species of hemiepiphytes. Plant, Cell, Environ. 19:
Effects on absorptance and photosynthesis in a desert shrub. Science 192:
631–642.
376–377.
Maximov, N. A. 1929. The Plant in Relation to Water. Allen and Unwin,
London. Additional Resources
Odening, W. R., B. R. Strain and W. C. Oechel. 1974. The effect of decreas- Campbell, G. S. 1977. An Introduction to Environmental Biophysics.
ing water potential on net CO2 exchange of intact desert shrubs. Ecology Springer Verlag, New York.
55: 1086–1094. Lambers, H., F. S. Chapin III and T. L. Pons. 1998. Plant Physiological Ecol-
ogy. Springer-Verlag, New York.
Contemporary Research Kramer, P. J. and J. S. Boyer. 1995. Water Relations of Plants and Soils. Aca-
demic Press, New York.
Borchert, R. 1994. Soil and stem water storage determine phenology
and distribution of tropical dry forest trees. Ecology 75: 1437–1449. Nobel, P. S. 1983. Biophysical Plant Physiology and Ecology. W. H. Free-
man, New York.
C H A P T E R
4 Soils, Mineral Nutrition,

and Belowground Interactions
M
ost terrestrial plants are rooted in the soil and depend on it for sup-
port, water, and mineral nutrients. What is soil, and how does it affect
plants? What are the characteristics and properties of soils? How is
soil created, and how long does it take for soil to be made? What are some of
the differences among soils in different places?
In this chapter, we look down and examine some of the surprisingly com-
plex things going on right under our feet. We look first at the characteristics
and properties of soil, and will see how those properties affect the availabili-
ty of water to plants. Since plants depend on soil for mineral nutrients, we
examine what these minerals are and how the plant uses them. We also intro-
duce the ecologically important underground interactions between plants and
two other group of organisms: nitrogen-fixing bacteria and mycorrhizal fungi.
Soil Composition and Structure

You might think of soil as “dirt,” a sort of ground-up mixture of sand, dust,
and crumbled bits of rock. If so, your image of soil would be very far from real-
ity. Soil is a complex, often highly structured system that is a unique product
of the interaction between living organisms and a physical matrix. Because
soils are formed by the interaction among living things, rocks, air, water, and
other materials, they occur only on Earth and nowhere else in our Solar Sys-
tem (at least as far as we know!).
What materials are found in soil, and what is it made of? To begin with,
there are mineral particles, derived from rock. These particles can range from
very large ones—stones and boulders—to, in progressively smaller sizes, peb-
bles, gravel, sand particles (>2.0–0.02 mm diameter), silt particles (0.02–0.002
mm), and clay particles (< 0.002 mm). (Conventionally, only sand, silt, and clay
are considered to be part of the soil itself.) Besides these rock-derived parti-
cles, there is organic matter in varying states of decomposition. Air and water
containing dissolved minerals are found in the pores between the mineral
particles.
Finally, soils are, to one extent or another, the major environment of vari-
ous kinds of living things: fungi, bacteria, photosynthetic prokaryotes (such as
cyanobacteria), single-celled eukaryotes such as diatoms, protists, visible and
microscopic insects and other arthropods (see Chapter 15), and many other
kinds of small (slugs, earthworms, nematodes) and larger (groundhogs,
(A)
gophers) animals. And, of course, soils are
filled with the roots and rhizomes of plants.
Composition (%) fpo Soil type
The living things in the soil continually alter Sand Silt Clay
its physical and chemical properties. Plants,
for example, take up large volumes of water 92 4 4 Sandy
from the soil, changing the amount and dis-
tribution of water within the soil. The vari-
85 10 5 Loamy sand
ous soil organisms act and interact in a wide
variety of ways. Some benefit one another;
others consume one another or defend 45 40 15 Loam
themselves to avoid being consumed. In the
process of metabolism, soil organisms
20 60 20 Silt loam
respire, altering the chemistry of their envi-
ronment, and produce waste products and
other substances, which become part of the 28 37 35 Clay loam
soil. These activities affect plants in many
ways, from making nutrients available to
causing disease. We examine some of these 25 30 45 Clay
effects in this chapter.
Soil Texture 0 10 20 30 40 50 60 70 80 90 100

Percent composition by dry weight
The properties of soil, and its effects on
plants, depend, first, on soil texture: the rel- (B)
ative proportions of the different particles 100% 0
making up the soil. Soils are composed of Clay
sand, silt, and clay particles. Depending on 10
90
which particle size dominates the character
of a soil, the soil texture is categorized as
sandy, silty, clayey, or loamy (Figure 4.1).
fpo 80 20
The properties of loamy soils are a balance 70 30

between sand, silt, and clay particles, and
are generally considered the most desirable
)
Clay
(%
Sil
60 40
soils for agriculture. Sandy soils, with more tc
nt
on
nte
than about 50% sand particles, have a ten

co
50 50 t (%
coarse texture. These soils drain rapidly Silty
ay
Cl
Sandy clay
)
after a rain, and they hold water and min- 40 clay 60
erals poorly. Water and air penetrate sandy Clay loam Silty clay
soils easily. Because of these characteristics, 30 Sandy clay
loam 70
they warm readily in spring and cool quick- loam
ly in autumn. 20 80
Clay particles have very distinctive Loam Silt loam
Lo
properties, and even a soil with as little as 10 am Sandy loam 90
ys
an Silt
35–40% clay will exhibit those properties (a d
Sand 100%
little clay goes a long way). Clayey soils 0
100% 90 80 70 60 50 40 30 20 10 0 Silt
can hold a large volume of water, and they
Sand
retain water and minerals exceptionally Sand content (%)
well. Similarly, clay-dominated soils retain
Figure 4.1
pesticides, pollutants, and other substances. (A) Examples of soil particle size of several different soil types. (B) The soil
They are much less permeable to air and texture triangle, which shows how soil types are classified according to the
water than are sandy soils, which can result percentage of sand, silt, and clay they contain (by dry weight). The values
in puddling, greater runoff, poor drainage, for clay are drawn parallel to the sand side of the triangle, those for silt are
and poor aeration (due to water filling up parallel to the clay side, and those for sand are parallel to the silt side. The
points on the grid at which the three lines intersect defines the type of soil;
the pore spaces and excluding air). As a for example, the grid lines for a soil with 20% clay, 40% sand, and 40% silt
consequence, clayey soils are very slow to intersect in the region that shows a soil with this particular composition is
warm up in spring and cool more slowly in loam.
PAGE PROOF: 2ND PASS Soils, Mineral Nutrition, and Belowground Interactions 65
(A) Figure 4.2

(A) Sand grains are irregular in size and shape, and are composed largely of quartz
with some other secondary minerals. Silt particles are similar to sand grains in min-
eral composition and shape, but are smaller in size. (After Buckman and Brady 1969.)
(B) Clay particles have a distinctive crystalline structure. Three different kinds of clay
particles are shown here. Of the three, montmorillonite has the greatest cation
exchange capacity (ability to hold cations) because in addition to binding cations on
its surface, it has a large exchange surface available between its plates. Contrast the
highly structured form of these clay particles with the rather formless sand and silt
particles in part A. The clay particles are also far smaller. (After Etherington 1982.)
(B)
autumn; plant growth in these soils may be slow to get
Kaolinite started early in the growing season. Silty soils tend to
Si tetrahedra be intermediate in their characteristics and properties
between sandy and clayey soils. Because soils often pos-
No water between lattice layers
sess characteristics of more than one of these texture
classes, they may be described by a combination of these
7.2Å Al octahedra terms, such as sandy loam, silty clay loam, or sandy clay.
To understand why different soil textures have the
properties they do, we need to look more closely at the
Si tetrahedra soil particles themselves. Sand and silt particles are gen-
erally irregular in shape, ranging from somewhat rec-
tangular or blocky to chunky, spherical shapes (Figure
Montmorillonite 4.2A). In contrast, clay particles (Figure 4.2B) have a
much more specific structure. These particles are made
Si tetrahedra up of plates, whose size, shape, and arrangement
Water and exchangeable cations depends on the minerals they contain and the conditions
between lattice layers under which they were formed. A clay particle is made
up of two or three flat crystalline plates, layered or lam-
Si tetrahedra inated together. These particles can be hexagonal in
14Å shape with distinct edges, or they may form irregularly
Al octahedra shaped flakes or even rods. Soils may also contain par-
(Isomorphous Mg)
ticles similar in size to clays, but without the distinc-
tive crystalline structure of clay particles. For example,
Si tetrahedra an amorphous material called allophane, with particles
about the same tiny size as clay particles, is prevalent in
soils developed from volcanic ash.
Most of the clay particles in soil exist in a colloidal
Illite
state (sand and silt particles are much too large to form
Si tetrahedra colloids). In a colloid, one or more materials in a finely
Potassium ions between divided state are suspended or dispersed throughout a
lattice layers second material. (Other examples of colloids are gelatin,
fog, cytoplasm, blood, milk, and rubber.) Clay particles
Si tetrahedra
have an enormous amount of external surface area
10Å because they have a large surface-to-volume ratio and
Al octahedra
(Isomorphous Fe, Mg)
because there are so many of them in a given volume
of soil. In some kinds of clays, there is additional inter-
nal surface area between the plates (see Figure 4.2B).
Si tetrahedra
Thus, the tremendous surface area that characterizes
clays is due to the fineness of the particles and to their
plate-like structure. The external surface area of a gram
of fine clay is at least 1000 times that of a gram of coarse
sand; the total surface area in the clay present in the
top 10 cm of soil in less than a half hectare of clayey soil, Soil pH

if spread out, would cover the continental United States. The pH of soil—the negative logarithm of the concen-
Because sand particles have a low surface-to-volume tration of H+ ions in the soil solution—varies widely
ratio, sandy soils have large, open pores between the among different soils. The pH scale ranges from 1 to
mineral particles. Water drains easily from these pores 14, where 7.0 is neutral (the pH of pure water); acid soils
because there is nothing to hold the water against the pull have lower pH numbers and higher H+ ion concentra-
of gravity, and air penetrates them readily. Clay-domi- tions. Soils in the United States can range from less than
nated soils have a much larger number of pores than do 3.5 (for example, in the pine barrens of New Jersey) to
sand-dominated soils. The total amount of pore space— as high as 10 (for example, in arid grasslands of the
the total proportion of the soil occupied by air and southwestern United States). For a soil to have a pH
water—is greater in clayey soils (50–60%) than in sandy above 7 (neutral soil), it must be calcareous (containing
soils (35–50%), not only because of the smaller size of clay CaCO3), sodic (containing Na2CO3), or dolomitic (con-
particles, but because of their arrangement. Clay parti- taining CaCO3 • MgCO3). Most crops grow best in slight-
cles (as well as particles of organic matter) tend to clus- ly acidic soils, but native vegetation can be adapted to
ter together, forming porous aggregates, resulting in anything from very acid to neutral to alkaline soils.
much more pore space than in sands, in which the par- Soil pH has enormous effects on plant growth, and
ticles lie close together. (Soils with native vegetation often indeed, in determining what species can survive and
have greater pore space than those that have been culti- grow in the soil. It acts on plants indirectly, however,
vated for crops because tilling damages some of the soil through its strong effects on the availability of mineral
structure.) nutrients and on the activity of some soil organisms
The greater pore space is one factor contributing to (such as bacteria and fungi), changing the conditions for
the greater amount of water that can be held in clay- plant growth in a complex manner. Forest trees can grow
dominated soils, but there is another factor that is also over a range of soil pH, but are especially tolerant of acid
very important. Unlike sand and silt particles, clay par- soils. Conifers and some other tree species tend to
ticles typically bear a strong negative electrochemical increase the acidity of the soil in which they are grow-
charge. Thus, they act as anions in the soil, attracting ing, primarily through the properties of the litter they
cations (positively charged ions, including the major produce (shed needles, etc.). Grasslands tend to be found
plant nutrients), as well as water molecules, to their sur- on relatively alkaline (high pH) soils, but this may be an
faces. The role of clay particles in adsorbing nutrient indirect effect because low rainfall results in soils with
cations—attracting and holding them to their surface— high pH (as we will see below) and also favors grass-
is one of their most important effects for plants. Al- dominated vegetation. There are other characteristic
though there are many kinds of cations that are attract- associations between soil pH and plants. Species in the
ed to clay particles, certain ones are most prominent and Ericaceae, for example, such as heaths and blueberries,
most important for plant growth. In humid regions, tend to grow only in very acid soils, while other taxa,
hydrogen and calcium ions (H+ and Ca2+) are usually such as Larrea tridentata (creosote bush, Zygophyllaceae),
most abundant, followed by magnesium (Mg2+), potas- are typically found in alkaline soils.
sium (K+), and sodium (Na+) ions. In soils of arid Soil pH can strongly affect the availability of cations
regions, hydrogen ions move to last place in this list, and to plants. Cations bind loosely to clay particles, which
sodium ions become more important. These positively are generally negatively charged. Under acid conditions,
charged ions attract numerous molecules of water to the excess H+ ions tend to bind more strongly to the clay
their surfaces, adding to the overall capacity of the clay particles than do the nutrient ions, displacing these nu-
particles to retain moisture in the soil. trients into the soil solution. Thus mild acidity, which
The adsorbed cations on clay particles are partially is characteristic of many soils, promotes nutrient avail-
available to be taken up by plants. There is a continuous, ability. Under extreme acidity (whether natural or due
dynamic interaction between the ions adsorbed on clay to acid precipitation), however, the nutrient cations are
and other colloidal particles and those in the soil solu- so mobile that they are easily leached, and are carried
tion— the water in the soil and its associated dissolved off in groundwater—water found underground in
minerals. Ions that are displaced from their positions on aquifers, rock crevices, and so on.
the surface of a particle enter the soil solution, from What determines the pH of a soil? Two cations,
which they can be taken up by plants, leached (lost from hydrogen and aluminum, tend to increase soil acidity,
the surface soil as water drains), or adsorbed on anoth- while the other cations have the opposite effect, increas-
er particle. These reactions differ greatly among soils of ing soil alkalinity. Hydrogen ions are continually added
different origins, textures, and chemical compositions, to the soil by decaying organic matter, roots, and vari-
and among regions differing in temperature and espe- ous soil organisms. They are actively exchanged
cially in rainfall. between the surfaces of colloidal particles in the soil
(such as clay and organic matter) and the soil solution, Clay loam Clay loam
contributing directly to the acidity of the soil. Aluminum + lime
ions (Al3+) indirectly cause hydrogen ions to be released CEC = 20 cmolc/kg CEC = 20 cmolc/kg
from colloidal particles by reacting with water to form
Al(OH)2+ and Al(OH)2+ plus H+. These hydrogen ions
H+
4
are then added to the H+ ions in the soil solution. High
rainfall levels favor the predominance of aluminum and
10
hydrogen ions because these ions are held very strong- H+and
Al3+
ly by colloidal particles, while the other cations are more
Sandy loam
readily leached and thus lost from the soil.
Most of the other cations, called exchangeable bases, CEC = 10 cmolc/kg
contribute to making the soil more alkaline. The cation H+
2
16
bold exchange capacity (CEC) of a soil is a measure of the total Bases
ability of the soil colloids to adsorb cations (in units of
centimoles of positive charge per kilogram of dry soil
10
Bases
8
[cmolc/kg]), and the percentage base saturation is the Bases
proportion of the CEC that is occupied by exchangeable
bases (Figure 4.3). In arid regions, the bases are not
leached out of the soil by rainfall, so the percentage base Base saturation = 50% 80% 80%
saturation is very high (90–100%), H+ concentration is pH = 5.5 6.5 6.5
low, and the soils tend to be alkaline. In areas with high-
er rainfall, the bases are leached more easily while H+ and Figure 4.3
Al3+ ions are retained, so the percentage base saturation Percentage base saturation in three soils: a clay loam (left),
the same clay loam with agricultural lime (calcium carbon-
is much lower (50–70%), and the soils tend to be acid. ate) added to raise the pH (center), and a sandy loam with
Plants and soil organisms are an important source low cation exchange capacity (right). The CEC for each soil
of soil acidity. When roots or soil organisms respire, they is indicated. CEC is measured in the SI units, centimoles of
generate CO2. In wet soils, this CO2 goes into solution positive charge per kilogram of dry soil (cmolc/kg), indicat-
immediately, creating a weak acid, carbonic acid. This is ing the numer of centimoles (1/100th of a mole) of positive
charge adsorbed per unit mass of the soil; 1 mole of negative
also the reason why rainwater is naturally a bit acid: CO2 charge attracts 1 mole of positive charges whether the
from the atmosphere dissolves into the raindrops. In charges come from H+, K+, Ca2+ or Al3+. See a current soils
tropical rainforests, where conditions favor massive text for further explanation. (After Buckman and Brady
amounts of respiration and there are often few clay par- 1969.)
ticles to bind cations, this added CO2 can help make
the soils quite acid and promotes the mobility of cations.
The consequence is rapid uptake of nutrients by plants highly decayed organic matter. The B horizons, deeper
in undisturbed forests, and rapid loss of those nutri- in the soil, represent the region of maximum illuviation,
ents when forests are cleared. or deposition of minerals and colloidal particles leached
from elsewhere. Clays, iron, and aluminum oxides often
Horizons and Profiles accumulate in the B2 horizon. The C horizon is the unde-
Soils are not homogenous; they contain characteristic veloped mineral material deep in the soil; it may or may
layers, or horizons, that differ from one soil type to not be the same as the material from which the soil devel-
another. If you look at roadcuts or excavations dug for ops. Below that may be bedrock, or just deep accumula-
construction, you can often see the horizons quite clear- tions of mineral material deposited by wind, water, or
ly. The sequence of horizons that characterizes a soil is glaciers.
called the soil profile (Figure 4.4). The horizons are No one soil has all of the horizons pictured in Fig-
grouped under four categories: O, A, B, and C. Subcat- ure 4.5. Some horizons may be much more distinct and
egories of these four categories are numbered according better developed than others. Plowing and the activity
to their particular characteristics (Figure 4.5). of earthworms may obscure the distinction between the
The O horizons consist of organic material formed upper horizons. The soil profile may also not be fully
above the mineral soil, derived from decaying plant developed, and some horizons may be absent or indis-
materials, microbial matter, and the remains and waste tinct, if the soil is relatively young. Upper horizons may
products of animals. The A horizons—the surface layer have been lost through erosion, leaving the deeper hori-
of mineral soil— represent the region of maximum leach- zons exposed at the surface.
ing, or eluviation. The uppermost A horizon, the A1, is Erosion is a particular problem where forests have
often darker than the rest of the soil and may contain been clear-cut, on soils that have been long cultivated
Figure 4.4 Depth Horizon Extent Composition pH

Profile of a forest soil at the edge of the
Adirondack Mountains in northern New 3 cm Litter 4.78
York State (in the Spodosol soil order).
This soil is a stony loam, forested with 12 cm Fermentation zone of 3.48
birch, hemlock, and spruce, and is quite O partially decayed organic
acidic throughout the profile. There is a matter; reddish-brown
thick layer of organic material at the top
of the profile, and a number of distinct 15 cm 3 cm Raw humus; black 3.45
horizons, each with particular properties
and a characteristic appearance. Roots are
shown reaching down to the top of the B3 A2 12 cm Silicious gray layer 4.20
horizon at about 45 cm deep, although the
deeper roots of large trees would certainly
penetrate more deeply into this soil. 30 cm B21 3 cm Precipitated humus; 3.93
(After Buckman and Brady 1969.) dark brown to black
12 cm Compact sesquioxide 4.50

accumulation; yellowish
B22
to reddish-brown sandy
45 cm loam
using poor farming practices, on slopes,
and on lands that have been severly 18 cm Friable yellow sandy 4.55
loam
overgrazed. Soil erosion has been enor- B3
mously important in human history
with both economic and social conse- 60 cm Friable yellow sand 4.80
C
quences, and even precipitated the col-
lapse of several Old and New World civ-
ilizations (Figure 4.6). It is also one of the
most critical (and most widely ignored) environmental
problems today. It occurs in many different parts of the
world and affects many natural and semi-natural ecosys-
tems. Rebuilding soils to replace losses to erosion can
take thousands of years. The consequences of soil ero-
fpo Horizon Properties
sion for our ability to raise food crops and for the sus- O1 Organic; original forms recognized
tainability of natural ecosystems can be devastating.
O2 Organic; original forms not recognized
Soils vary in depth, from thin layers of soil barely
covering a rock substrate (for example, in many alpine
areas) to very deep soils of close to 2 meters (for exam- A1 Mineral mixed with humus; dark colored
ple, in some well-developed prairie soils). Soil depth has
a great deal of influence on vegetation and plant growth: A2 Horizon of maximum eluviation of
The greater the soil depth, the more favorable the soil silicate clays; Fe, Al oxides, etc.
for plant growth, all other things being equal. Deeper
A3 Transition to B; more like A than B
soils can hold more water and nutrients and can retain B1 Transition to A; more like B than A
water for a longer period of time without rainfall, and
they allow greater development of plant root systems. B2 Maximum illuviation of silicate clays;
Fe, Al oxides, some organic matter
B3 Transition to C; more like B than C
Figure 4.5
An abstract, general soil profile showing the major horizons C Zone of least weathering; accumulation
that might be present in particular soils. No one soil is likely of Ca, Mg carbonates; cementation
to have all of the horizons shown, and particular soils may
have greater development of subhorizons than what is
shown. Only the upper part of the C horizon is considered
to be part of the soil proper. The depth of soils varies
tremendously with the location and nature of the soil, but to
gain some perspective, readers might picture this illustrated R Bedrock
profile as being about a meter deep to the top of the bedrock.
(After Buckman and Brady 1969.)
Figure 4.6
During the 1930s, soil erosion in the
midwestern United States led to mas-
DM looking for more up-to-date photo sive dust storms that destroyed farms
to replace this one. and even entire towns. What came to
be known as the “Dust Bowl” dis-
placed thousands of families, who left
the region in search of a new liveli-
hood. This shows the remains of a
farm lot in South Dakota in 1936. (Pho-
tograph courtesy of the U.S. Depart-
ment of Agriculture.)
Origins and Classification years old; an old soil may be 100,000 years old or more.
Soils are formed by the action of weathering on parent As a soil ages, the primary minerals that came from the
material, the upper layers of the heterogeneous mass parent material undergo chemical changes, and second-
that is left over after the action of weathering and other ary minerals are formed. The structure of the soil devel-
forces on rocks. The physical and chemical actions of ops, and then changes, as materials are leached, rede-
temperature, rainfall, and other climatic factors further posited, and lost from the soil, as well as being physically
grind up, move, and leach the parent material until it and chemically altered. The mineral constituents of the
begins to develop into true soil. Parent material is orig- soil shift, as minerals that are most stable at Earth’s sur-
inally derived from rocks, which are classified as face gradually come to dominate. Young soils are found
igneous (of volcanic origin), sedimentary (from the where the parent material is still present, as in areas that
deposition and recementation of material derived from were glaciated relatively recently (such as much of the
other rocks), or metamorphic (changed by the action of northern part of North America) or in areas of recent
great pressures and temperatures on igneous or sedi- alluvial deposits (such as the bottomlands of the world’s
mentary rocks deep underground). great river systems). Old soils are most commonly found
The parent material can be broken up by the in the Tropics and Subtropics. Some of the world’s old-
mechanical action of temperature changes, including est soils are found, for instance, in parts of Africa.
repeated freezing and thawing, by the direct action of Five major factors are responsible for determining
water, wind, and ice in abrading and eroding rock frag- the kinds of soils that develop in an area: climate, the
ments, and by the actions of plants, fungi, and animals. nature of the parent material, the age of the soil, topog-
Chemical reactions occurring in the disintegrating rock raphy (which acts to alter the effects of climate),and liv-
material accelerate the physical breakdown of the frag- ing organisms. These factors do not operate in isolation,
ments and begin the process of decomposition and but rather interact in complex ways with one another.
chemical alteration (Figure 4.7). Large and small frag- Vegetation, with its associated soil organisms, has an
ments of rock and rock-derived debris are moved by ice especially strong influence on soil development, but the
(particularly by glaciers), wind, and water. They are then nature of the vegetation that is present is also depend-
redeposited in the form of glacial till and outwash, loess ent on both soil and climate.
(pronounced “luss”) and other aeolian (wind) deposits, Soils are classified according to a comprehensive
alluvial (river and stream) deposits, and lake and marine taxonomic system developed in the United States,
sediments. These deposits may eventually form soil par- although some soil scientists and ecologists still rely on
ent material. earlier classification systems. The broadest category in
The process of soil formation from parent material the modern system is the soil order. There are ten soil
takes thousands of years. A young soil may be 10,000 orders worldwide (Table 4.1).
(A) Figure 4.7

Development of soil structure from
two different parent materials: (A)
bedrock and (B) loess deposited by
Forests – Natural vegetation wind. These soils are also develop-
ing under two different climatic and
vegetation types. Over time, organic
matter accumulates in the upper
Organic matter accumulation horizon, depending on the type of
Zone of maximum leaching vegetation present. In the lower
horizons, clay, iron oxides, or
Hydrous clay, oxide CaCO3 are deposited and accumu-
Dis
int accumulation late, and characteristic structures
egr
ate develop. The soil orders (ultisol and
dw
eat mollisol) are described in Table 4.1.
her
ed
soi
(After Buckman and Brady 1969.)
Relatively unweathered bedrock lm
ate
ria
l
Fresh Well developed

unweathered ultisol
rock (red-yellow podzolic)
Time
(B)
Grass – Natural vegetation
Organic matter accumulation
Blocky structure
development
Ca
Relatively unweathered loess Ca CO
SO 3 acc
4 ac um
cu u
mu latio
lat n
ion
Fresh loess Well developed

deposit mollisol
(chernozem)
Time
The soil classification categories most commonly local soils. Such data can be highly valuable for many
used by plant ecologists are the soil series and soil type. kinds of ecological studies, particularly at the commu-
These are used to classify local soils. A soil series is usu- nity and ecosystem levels. The availability of informa-
ally named after some local geographic area or feature, tion on local soils in other countries varies widely.
and generally consists of a dominant type with several
associated other types. Soil types are based upon topog- Organic Matter and the Role of Organisms
raphy, parent material, and the vegetation under which So far we have emphasized the role of physical process-
the soils were formed. Soil series occur at regional scales, es in soil development and character. However, we
and soil types characterize landscape-scale features. In began this chapter by emphasizing that soils are the
the United States, soil series are mapped for almost unique product of living things acting on the physical
every county, and detailed maps are available for most environment (and in turn being affected by that envi-
locations showing the soil series and type, and often pro- ronment). What are some of the ways in which organ-
viding extensive information on the characteristics of isms affect soils?
Table 4.1 Soil orders and some of their characteristics
Soil order Development and characteristics Found in

Mollisols Develop under prairie vegetation; many of these soils Great Plains of the United States and Canada;
are under cultivation, and they constitute some large areas in the heartland of Russia, Mongolia,
of the most productive agricultural soils and northern China; northern Argentina,
Paraguay, and Uruguay
Spodosols Generally develop in cold-temperate humid regions Northeastern and northern midwestern
under northern forests; usually strongly acid and United States and adjacent areas in Canada;
of low fertility northern Europe, Siberia; also some soils in
southern South America
Alfisols Develop in humid regions under deciduous forests; Large area from Baltic states to western Russia;
weathered somewhat less than spodosols but more southern half of Africa; eastern Brazil; England,
than inceptisols; contain silicate clay-dominated France, central and parts of southern Europe;
horizon; often cultivated and highly productive Michigan and Wisconsin to Pennsylvania and
agriculturally New York in the United States
Oxisols Usually tropical soils, often supporting tropical rainforest One of the most widespread soil orders globally,
vegetation. The oldest and most highly weathered soil and among the most important in terms of the size
order, with a deep subsurface horizon high in clays and of the human population supported; also supports
other colloidal particles; intense leaching has removed the greatest biodiversity of many groups of plants
most of the silica, leaving a high proportion of oxides and animals. Covers much of Africa and South
of iron and aluminum, often with a red color America
Entisols Recent soils with lack of significant profile development; Sand hills of Nebraska; areas of northern and south-
highly variable in fertility ern Africa; northern Quebec; parts of Siberia and
Tibet; many areas that were covered by the most
recent glaciation
Inceptisols Young soils exhibiting limited weathering, but with All continents—north Africa, eastern China,
greater profile development than entisols; many western Siberia, Spain, central France, central
of these soils are under agricultural production Germany, northern South America, northwestern
United States
Aridisols Develop in arid regions where soils are dry through Southwestern United States and northern Mexico;
most of the year and there is little leaching southern and central Australia; Bobi and Takla
makan Deserts in China; Sahara Desert and also
southwestern Africa; Pakistan; other arid regions
Vertisols High content of certain clays; sticky and plastic when Large regions of India, Sudan, and eastern Australia;
wet and hard when dry, with extensive cracking; small areas in southeastern Texas and eastern
unstable for building and difficult to farm Mississippi in the United States
Ultisols Develop under warm to tropical climates under forest or Southeastern United States; northeastern Australia;
savanna vegetation; generally highly weathered but still Hawaii; southeastern Asia; southern Brazil
retaining some minerals; somewhat acid, with clay
horizons; these are fairly old soils
Histisols Wetland and bog soils; develop in a water-saturated Not widespread, but may be locally important,
environment; very high in organic matter content; as in the Everglades in southern Florida and in
sometimes also high in clay. Important soils not only in many areas of northern Europe. The largest areas
present-day wetlands, but in areas of ancient swamps are in Canada southwest of Hudson and James Bay
and bogs that are currently forested, cultivated, or used and in northwestern Canada into eastern Alaska
for mining peat
Note: These soil orders reflect the modern classification system currently brunizems (mollisols), latosols and lateritic soils (oxisols), lithosols, solon-
used in the United States. Some of the major Great Soil Groups of the chak soils, and desert soils (aridisols), azonal soils (entisols), and
older classification system, with the soil orders with which they roughly wetland/bog soils (histisols).
correspond, are: podzols (spodosols and alfisols), chernozems and
Organic matter is the decaying and decomposed ter to the soil. The activities of various different kinds of
material in soil that comes from living things. It includes microorganisms affect soil properties in many addition-
substances secreted by plants, microorganisms, and ani- al ways, including the recycling of mineral nutrients
mals, products of excretion by animals, parts shed by used by plants (see Chapter 15).
animals and plants, and dead organisms and parts of A property of soils that is of vital importance to plant
organisms. Microscopic organisms, while tiny individ- life (and to animals, fungi, and the other organisms liv-
ually, collectively contribute a great deal of organic mating in soils) is soil structure. Soil structure describes the
physical arrangement of soil particles into

larger clusters, called aggregates (or peds). Transpiration
One of the most important aspects of soil
structure is the porosity of the soil, which
Precipitation
measures the total volume of pores and their
size, shape, and arrangement between and
within the aggregates (the aggregates are not Evaporation
solid, and may be “fluffy” or packed to vary-
fpo
ing degrees). Soil porosity determines how Evaporation
much, and how easily, water and air can be
held by and move around in the soil, and Infiltration Surface runoff
thus is an important factor in the water-hold-
ing capacity of the soil. Deep drainage
Evaporation
Porosity and organic matter content also
Uptake by plants
determine how easily roots or fungal hyphae
can penetrate soil, and how easy it is for ani- Impermeable horizon
mals to burrow and tunnel through it. The River
Groundwater
actions of organisms (from animals to roots
and fungal hyphae) in moving through the
soil create pores and soil structure. Organic Figure 4.8
matter is amorphous (it does not have a well-defined The fate of water falling on the ground from a rainstorm.
structure, unlike clay particles, for example), and parti- (After Kramer 1983.)
cles of organic matter typically have very large surface
areas. Organic materials bind mineral particles together
to stabilize the aggregates so that the soil porosity is Rainwater immediately begins to drain downward
maintained throughout the physical action of wetting, due to the pull of gravity, entering the groundwater and
drying, freezing, thawing, and other processes. draining into the streams, rivers, and lakes of the water-
Organic matter also supplies H+ ions, determining shed. After about a day, this rapid downward movement
soil pH. Furthermore, soil organic matter is critically of water slows, and many of the large macropores in the
important in plant nutrition, both by supplying essential soil refill with air (Figure 4.9A) The tiny micropores, how-
nutrients and by providing physical particles that act, like ever, remain filled with water at this point, and the soil
clay particles, to attract and retain ions because they are is at field capacity. The water potential (see Chapter 3)
negatively charged and have very large surface areas. of the soil is now at –0.01 to –0.05 MPa (Figure 4.9B).
Fresh and decayed organic matter in the soil also contains Water is held in the soil largely by attraction to the
compounds, including various organic acids, that chem- surfaces of soil particles, particularly clays and organic
ically alter essential plant mineral nutrients such as calci- matter. It moves through small pores by capillary action,
um (Ca), iron (Fe), manganese (Mn), copper (Cu), and zinc a process that occurs within narrow tubes or in a surface
(Zn), making them more readily available to plants. film of water (such as within soil pores, the xylem, and
the substomatal chambers of leaves). Water is pulled
upward (or horizontally) by the attraction of the water
Water Movement within Soils
molecules to the charged particle surfaces and to one
Imagine a meadow in the summertime that has been another. As the plants in the hypothetical meadow of our
without rain for some time. A thunderstorm sails in, example transpire, a water potential gradient (see Fig-
offering a solid drenching. (A similar set of events would ures 3.2 and 3.3) is established, and soil water moves
take place in a suburban lawn, a forest, an agricultural toward the roots and is absorbed by them.
field, and even, to some extent, an urban vacant lot.) As the soil continues to become drier, the small
What happens within the soil? Pores that had been filled pores begin to empty of water, filling with air. As the
with air fill with water, first in the upper layers of the small pores empty, the once continuous film of water is
soil profile, then deeper as more rain falls. Eventually broken in many places, and water movement is greatly
almost all of the pores are filled with water. Some water slowed. Water can continue to move as water vapor
runs off the soil surface, the amount depending on the through these emptied pores, but only small amounts of
vegetation, slope, and other factors (Figure 4.8). The soil water can be transported in this way. Each day, as the
is now saturated. Depending on the soil texture, soil plants transpire and the soil progressively dries, the
structure, and soil depth, soils may hold very different water potential of the plants declines (becomes more
amounts of water when they are saturated. negative), and each night, as stomata close, the water
(A) (B)
40 0.4
Field
Water content (cm/cm of soil)

capacity
Volume percent soil water

30 Available water 0.3
fpo
20 0.2
Permanent
10 wilting point 0.1
Saturation Field capacity Wilting point
Unavailable water
Solid space Pore space 0 0
Sand Sandy Loam Silt Clay Clay
100 g 40 g Saturated loam loam loam
Solid Water soil
fpo
100 g 20 g Air Field capacity amount (by weight or volume) of water held in the soil
Solid Water
at any point in this drying curve will vary greatly with
the soil texture and also with the proportion of organic
100 g 10 g Air Wilting matter in the soil (Figure 4.11).
Solid Water point
Water moves upward in the soil as well as down-
ward. As water evaporates from the soil surface, capil-
100 g 8g Air Hygroscopic
lary action pulls water from deeper in the soil upward
Solid Water coefficient against gravity. The roots of transpiring plants pull water
upward from deeper in the soil as well. Sometimes large
Figure 4.9 amounts of water can be transferred from deep horizons
(A) A silt loam soil at saturation, at field capacity, and at the in the soil to dry surface layers by deeply rooted plants.
wilting point, showing the soil particles and pores filled During the day, roots generally have more negative
with water and/or air (above). At saturation, the soil is hold- water potential than soils. At night, the dry upper lay-
ing all the water it can accomodate, and water will drain ers of the soil may have more negative water potential
from the soil due to gravity. At field capacity, a considerable
amount of water has been removed from the soil (shown in than roots, and may be capable of removing water from
the beaker below the soil), and at the wilting point, more the roots. Deeply rooted plants may take up water from
water still has been removed. The relative amounts of solid the moist deeper horizons, from which it travels up the
particles, water, and air in the soil are shown for each of root system and moves out of the upper part of the root
these states below. A further reduction in soil moisture is
reached at the hygroscopic coefficient, when water is held
mostly by the soil colloids, and is completely unavailable to
plants. (From Buckman and Brady 1969.) (B) The relative
amounts of available and unavailable water in soils of vari-
ous textures, expressed both as percentage of soil volume 0
Soil water
Water potential (MPa)
and centimeters of water per centimeters of soil. (After potential

Kramer 1983.) – 0.5
–1.0 Leaf water

potential
potential of the plants rises somewhat as the plants
equilibrate with the soil (Figure 4.10). –1.5
When the soil reaches a water potential of about –1.5 1 2 3 4 5 6 7 8 9
MPa, most mesic plants no longer can extract water from Time from last irrigation (days)
the soil, and exist in a permanently wilted condition,
about to die. This state is called the permanent wilting Figure 4.10
point of the soil, and the soil moisture content at this Daily changes in soil and leaf water potential for a cultivated
plant growing in a clay loam soil over the course of nine
point is called the wilting coefficient of the soil. How- days and nights after receiving irrigation. Dark portions of
ever, plants adapted to dry environments can continue the x-axis indicate nighttime conditions; open portions indi-
to remove water from soils that are much drier. The cate daytime. (After Etherington 1982.)
system into the dry soil in the upper horizons. There, the
Plant Mineral Nutrition
water can become available to other, more shallowly
rooted plants. This phenomenon, called hydraulic lift, What is a plant made of? A large tree is clearly many
may in some cases allow the survival of plants with shal- orders of magnitude larger than the seed from which it
low roots during periods drought or in arid environ- grew. Where did all of that “stuff” in the tree come from?
ments, because deeply-rooted neighbors may increase A man or woman is much larger than a child largely as
the availability of soil moisture closer to the surface. a result of his or her cumulative food intake (of course,
Table 4.2 Elements essential for plant growth and survival
Principal Concentration in plant

form in Average mass Percent
which (g/kg plant or
Element Symbol absorbed dry weight) ppm, range Important functions
Found in all organic compounds

Carbon C CO2 450.0 ~44% Major component of all organic compounds,
including cellulose in cell walls, which
makes up much of the plant’s dry weight;
other carbohydrates (sugars, starches) and
lipids store and transport the energy
captured in photosynthesis
Oxygen O H2O or O2 450.0 ~44% Component of all major organic compounds;
cellular respiration
Hydrogen H H2O 60.0 ~6% Component of all major organic compounds;
essential to all major biochemical reactions
and to acid/base balance
Macronutrients
Nitrogen N NO3– or 15.0 1–4% Essential component of nucleotides, nucleic
NH4– acids, amino acids, proteins (including
structural proteins and enzymes), and
chlorophylls
Potassium K K+ 10.0 0.5–6% Involved in osmosis, ion balance, pH
regulation, and opening and closing
of stomata; activator of many enzymes;
protein synthesis
Calcium Ca Ca2+ 5.0 0.2–3.5% Strengthens cell walls and some plant tissues;
involved in cell division and cell elonga-
tion, membrane permeability, cation-anion
balance; structural component of many
molecules; second messenger in signal
conduction between environment
and plant growth and developmental
responses
Magnesium Mg Mg2+ 2.0 0.1–0.8% Essential component of the chlorophyll
molecule; activator of many enzymes;
involved in cation-anion balance and
regulation of cytoplasm pH
Phosphorus P H2PO4– or 2.0 0.1–0.8% Essential structural component of nucleic
HPO42– acids, proteins, ATP, and NADP+; critical
for energy transfer and storage
Sulfur S SO42– 1.0 0.05–1% Component of some amino acids and pro-
teins, coenzymes (including coenzyme A),
and secondary metabolic products (includ-
ing defensive compounds)
much of the food one eats is used to provide energy, carbon is obtained by fixing atmospheric CO2 during
which is used up along the way). But plants clearly do photosynthesis. The oxygen also comes from the atmos-
not eat. In order to understand where all of the materi- phere, while the hydrogen comes from water taken up
al that makes up a plant comes from, we first have to see by the roots. With one noteworthy exception—the nitro-
what a plant is made of. gen fixed by symbionts of some plants—all the rest of
Most of the material in a plant is made up of com- the elements in a plant are taken up from the soil solu-
pounds of carbon, hydrogen, and oxygen (Table 4.2). The tion by the plant’s roots.
Principal Concentration in plant

form in Average mass Percent
which (g/kg plant or
Element Symbol absorbed dry weight) ppm, range Important functions
Macronutrients
Chlorine Cl Cl– 0.1 100–10,000 ppm Used in stomatal regulation, proton
pumps, and osmoregulation; critical for
splitting water molecules in photosystem
Iron Fe Fe3+, Fe2+ 0.1 25–300 ppm Component of heme proteins (such as
cytochromes, essential molecules in light
reactions, cellular respiration, and nitrate
reduction; leghemoglobin, used in N fixa-
tion) and iron-sulfur proteins; necessary
for chlorophyll synthesis and in the elec-
tron transport chain in light reactions
Manganese Mn Mn2+ 0.05 15–800 ppm Present in several enzymes; needed for
activation of many enzymes
Boron B B(OH)3, 0.02 5–75 ppm Poorly understood; cell wall synthesis,
B(OH)4– nucleic acid synthesis, and plasma mem-
brane integrity
Zinc Zn Zn2+ 0.02 15–100 ppm Present in several enzymes; needed for
activation of many enzymes; protein
synthesis and carbohydrate metabolism;
structural component of ribosomes;
important but not well understood role
in plant hormone metabolism
Copper Cu Cu2+ 0.006 4–30 ppm Present in some proteins and in plastocyanin
(necessary for light reactions); enzyme
activation; pollen formation and ovule
fertilization; lignification of secondary cell
walls (in wood formation)
Nickel Ni Ni2+ ~0.0001 ~0.1 ppm Necessary for function of some enzymes;
nitrogen metabolism
Molybdenum Mo MoO42– 0.0001 0.1–5 ppm Enzyme cofactor for N fixa-
tion and some other reactions; pollen
formation and seed dormancy
Essential to some plants
Sodium Na Na– Trace Osmotic and ionic balance, especially in
some desert and salt-marsh species
Cobalt Co Co2+ Trace Required by nitrogen-fixing microorganisms
Silicon Si SiO3– Variable Not well understood, appears to have a role
in disease resistance; important structural
component of cells in grasses, Equisetum spp.
(horsetails) and other plants; may reduce
leaf water loss
Source: Compiled in part from data in Barbour et al. 1987, Marschner 1995, and Raven et al. 1999.
able in the soil frequently limits the ability of the plant

60
to carry out these functions. Plant form and develop-
ment are also influenced by nitrogen availability.
50 Increased nitrogen supply may delay the senescence of
Water content (%, by volume)
leaves or of the entire plant. It may increase the propor-

40 Clay
tion of biomass allocated to shoots relative to roots and
the proportion allocated to vegetative growth versus sex-
Loam
ual reproduction. Nitrogen availability may also affect
30
the plant’s biochemical composition in a complex man-
ner, and may alter the concentrations and types of defen-
20 sive compounds present. The ratio of carbon to nitrogen
in plant tissues is affected by nitrogen supply rates, and
10
Sand in turn affects many other aspects of plant life, includ-
ing susceptibility to herbivores and litter decomposition
rates.
0 Nitrogen exists in the soil in inorganic and organic
–0.0001 –0.001 –0.01 –0.1 –1.0 –10.0 –100.0 –1000.0
Soil water potential (MPa, log scale)
forms. Plants take up both inorganic nitrate (NO3–) and
ammonium (NH4+) from the soil, but nitrate is general-
Figure 4.11 ly more abundant and more readily taken up. Some
Illustration of general relationships between water content plants are able to take up ammonium, particularly when
(by volume) and soil water potential (MPa) for clay, loam,
and sand. These curves differ depending on whether the soil
they are young, and plants adapted to acid soils or soils
was measured as water content was increasing or decreasing with limited oxygen potential (such as waterlogged
(the direction of change in moisture shown by arrows). soils) preferentially take up ammonium. The costs and
(After Rendig and Taylor 1989.) benefits of nitrate versus ammonium uptake are com-
plex, and depend on their concentrations in the soil as
well as other factors. Nitrate must be reduced to ammo-
nia before it can be incorporated into organic com-
How do plants use these elements? There are broad- pounds in the plant, but as nitrate it is highly mobile,
ly two roles that an element can play in an organism: it and can be transported readily in dissolved form in the
can be part of the material that makes up the organism’s xylem and stored in cell vacuoles. Ammonium is large-
structure, or it can be essential for the operation of the ly incorporated into organic compounds in the roots
organism’s metabolism. All the elements except C, O, directly after it is taken up. Ammonium can be toxic, par-
and H are usually called the essential mineral nutri- ticularly when converted to ammonia (NH3).
ents—those chemical elements that plants need to live,
grow, and reproduce. Table 4.2 lists some of the main Biological Nitrogen Fixation
functions of each of these elements. The essential nutri- Nitrogen is needed in large quantities by plants, and its
ents needed in larger quantities are called macronutri- availability in the soil is almost always low. Ironically,
ents; those needed in very small quantities are known plants are surrounded by an ocean of nitrogen in the
as micronutrients. Another group of minerals is the ben- atmosphere in the form of elemental N2, which they can-
eficial mineral elements; these minerals either are essen- not tap directly. You might wonder why plants cannot
tial only for certain plant species, or they are not essen- use this form of nitrogen. It turns out that some plants
tial but do stimulate growth. Beneficial mineral elements are able to do so—not directly, but by forming symbioses
are sodium, silicon, and cobalt. (Box 4A) with nitrogen-fixing prokaryotes (bacteria).
While plants cannot grow and function without the Only prokaryotes can directly use N2 and “fix” it
essential mineral nutrients, more is not necessarily bet- in biologically usable forms. These nitrogen-fixing
ter. Most of these minerals are necessary in small to mod- organisms are remarkably diverse, and encompass nine-
erate amounts, but toxic in large quantities. Some species teen families of bacteria, including eight families of
have evolved tolerance of high concentrations of min- cyanobacteria (the blue-green bacteria, which are pho-
erals that are toxic to most other species (see Figure 6.1). tosynthetic). Some nitrogen fixers are free-living, some
are symbiotic with terrestrial plants, and some are asso-
Nitrogen in Plants and Soils ciated with the root surfaces and intercellular spaces of
The two most important essential mineral nutrients for plants. The biological reduction of N2 to NH3 is cat-
plants are nitrogen and phosphorus. Plants need large alyzed by an enzyme complex, nitrogenase, which is
amounts of nitrogen to photosynthesize (see Chapter 2), found in all N2-fixing prokaryotic microorganisms, and
grow, and reproduce, and the amount of nitrogen avail- nowhere else in the living world. The reaction requires
BOX 4A
Symbioses and Mutualisms
The word “symbiosis” is rooted in two while the other is unaffected, the rela- Two important symbioses between
Greek words, bios (life) and sym (togeth- tionship is called a commensalism. plants and other organisms are symbi-
er). Symbioses are associations between And if one member benefits while the otic nitrogen fixation by certain species
members of two different species that other is harmed (sometimes even des- of bacteria living in or on plant roots,
live in intimate contact with each other. troyed), it is parasitism. Any of these and the formation of mycorrhizae by
There are a variety of symbiotic associ- three types of symbioses may be obli- plants and certain species of fungi. Both
ations among organisms. gate—necessary for the survival of one of these symbioses are often, but not
A symbiotic association may benefit or both members—or facultative— always, mutualistic. Other common
members of both species involved, or it existing only under some conditions. symbioses include interactions between
may benefit only one member, with no The nature of a particular symbiosis some plants and animals that act as pol-
effect or a negative effect on the other. may even change over time, or as envi- linators or seed dispersers.
If both members benefit, the symbiosis ronmental conditions change.
is a mutualism. If one member benefits
a great deal of energy, and the source of that energy is as symbiotic) nitrogen-fixing cyanobacteria are impor-
the key to understanding the different ecological roles tant sources of nitrogen in flooded rice paddies (see Fig-
played by different nitrogen fixers. ure 15.15) and in some surface soils.
The free-living nitrogen fixers are either photosyn- Symbiotic nitrogen fixation is more important in
thetic cyanobacteria, such as Anabaena (Figure 4.12), or most ecosystems than fixation by free-living organisms.
are heterotrophic and obtain their energy from the There are a great variety of symbiotic associations
decomposition of plant residues in the soil. The vol- between nitrogen-fixing prokaryotes and terrestrial
ume of nitrogen fixation by the heterotrophs is general- plants. There is generally a strong specificity in these
ly low because they are limited by the amount of carbon symbioses, in which particular bacterial species prefer-
and energy they can extract from dead material. In con- entially infect particular plant species. Three of the most
trast, the photosynthetic free-living nitrogen fixers can common general types of nitrogen-fixing symbioses
provide substantial amounts of nitrogen to the systems between prokaryotes and plants are nodulated legumes,
in which they live because they use the far more abun- nodulated non-legume symbioses, and looser associa-
dant energy of sunlight. For example, free-living (as well tions between plants and cyanobacteria. The first is the
most familiar, particularly because of its agricultural
Heterocyst importance, but it is not the only one with ecological
importance.
Symbioses with legumes (plants in the family
Fabaceae, such as peas, beans, clover, acacia trees) are
formed by bacteria belonging to either the fast-growing
species in the genus Rhizobium (Figure 4.13A) or slow-
growing Bradyrhizobium species. Rhizobium and Bradyrhi-
zobium bacteria live in nodules on the roots of legumes
(Figure 4.13B), providing the plant with a source of NH3
and receiving carbohydrates (providing carbon and ener-
gy) from the host plant. The nodules provide the anaer-
obic conditions that the bacteria require to fix nitrogen.
The Fabaceae is one of the largest and most ecolog-
ically diverse plant families. Many important crop
species are members of this family, and its crops are
major protein sources for civilizations worldwide (e.g.,
Figure 4.12
Free-living nitrogen-fixing cyanobacteria such as Anabaena soybeans in eastern Asia, lentils in southern Asia, pigeon
azollae fix nitrogen in specialized cells called heterocysts. peas and peanuts in Africa, pintos and other beans in the
(Photograph © J. R. Waalund/Biological Photo Service.) Americas, and fava beans and garbanzos in the Middle
(A) Figure 4.13
Symbiotic nitrogen-fixing bacteria of the genus Rhizobium
cannot fix nitrogen until they form an association with the
plant’s root cells. (A) The cell on the right has not been
“infected” with Rhizobium. Bacteria have entered the cell on
the left, where they take a nitrogen-fixing form called bac-
teroids. Root cells with bacterioids form nodules. Photo-
graph © E. Newcomb and S.Tandon/Biological Photo Ser-
vice. (B) Root nodules on Glycine max (soybean, Fabaceae).
The anaerobic conditions inside the nodules provide the nec-
essary environment for Rhizobium to fix nitrogen. (Photo-
graph © K. Wagner/Visuals Unlimited.)
fixed by this symbiotic association can be very impor-

Rhizobium tant in these successional ecosystems (see Chapter 13).
(B) In the third type of plant-prokaryote association,
that of nitrogen-fixing cyanobacteria with plant species,
there may be high specificity between the host plant and
the bacterial species, but the bacteria remain external
to the root cells. In these loose associations, the bacteria
obtain energy from root exudates from the plant, but
provide little nitrogen to the plant until they die. These
associations are not always mutualistic, and sometimes
one of the partners in the relationship obtains all or most
of the benefits of the association without providing
much to the other partner.
External energy input is required for N fixation. The
symbiotic bacteria obtain this energy from photosyn-
thate supplied by the plant host, which must be divert-
ed from other plant functions such as growth or repro-
duction. Factors that increase photosynthetic rate tend
to increase nitrogen fixation rates, and factors (such as
East; Figure 4.14). Many tropical trees and

shrubs (such as Acacia spp.) belong to the
Fabaceae as well; these plants can play a crit-
ical role in ecosystem function by providing
nitrogen to soils and by providing high-pro-
tein food for wildlife (particularly seeds and
fruits). There are also members of the Fa-
FPO
baceae that do not form symbioses with nitro-
gen-fixers.
The second major type of nitrogen-fixing
symbiosis is that of non-legumes with actin-
omycete bacteria in the genus Frankia. As in
the legumes, the bacteria live in nodules on
the plants’ roots and obtain carbon and ener-
gy from the plants while providing them
with NH3. There are about 200 plant species
in the Tropics and Subtropics that form such
associations, but they also occur in some tem-
perate systems. Alnus spp. (alder, Betulaceae),
Figure 4.14
for example, are symbiotic with Frankia, and A wide variety of dried beans are for sale in this market stall in Ecuador.
are common early colonizing trees in many Legumes such as beans, members of the family Fabaceae, are crucial
riparian communities; the nitrogen that is sources of protein for people throughout the world.
light and water limitation) that reduce photosynthetic forming them back to available inorganic forms, and
rates tend to decrease nitrogen fixation. releases them back into the soil solution, where they can
The presence of large amounts of mineral nitrogen be taken up by plants once again.
in the soil inhibits nitrogen fixation by suppressing nitro- Phosphorus in soils is derived from apatite miner-
genase activity and severely reducing the number of als in the parent material (see Chapter 15). In natural sys-
nodules formed. The mechanisms by which this sup- tems, most of the phosphorus in plants is recycled by
pression occurs are complex, and beyond the scope of microbial decomposition. While some systems have con-
this book. As nitrogen becomes more readily available siderable reserves of phosphorus in the soil, in others,
in the soil, however, the uptake of soil nitrogen becomes such as many grasslands and some tropical rainforests,
energetically cheaper than providing carbohydrates to essentially all of the available phosphorus exists in liv-
bacterial symbionts. From the perspective of plant fit- ing plant tissue, litter, and decomposing organic matter,
ness, the switch from using symbiotically fixed nitrogen from which it is rapidly recycled or lost. Most of the
to direct uptake from the soil should occur at this point. phosphorus taken up by plants comes from the rhizos-
phere, the microbe-rich environment within about a mil-
Phosphorus limeter of and surrounding the root.
Like nitrogen, phosphorus is needed by plants in larg- It is sometimes possible for plants to directly influ-
er quantities than are generally available in soils. Next ence the availability of nutrients in the rhizosphere.
to nitrogen, phosphorus is likely to be the most limit- Some plants, for example, can create their own locally
ing mineral nutrient for plant growth. Phosphorus is also acidic environment by secreting organic acids that pro-
needed to support symbiotic nitrogen fixation. mote the uptake of cations. Plants of the family Pro-
Soils vary greatly in the amount of phosphorus they teaceae—a prominent group in the old Gondwanan
contain. In the United States, total soil phosphorus tends continents (Australia, South Africa, and South Amer-
to be greatest in the Pacific Northwest, extremely low in ica)—which often grow in very phosphorus-poor soils,
the Southeast, and low to moderate in the Northeast, may form specialized proteoid roots. These dense clus-
Midwest, and Southwest. Soils derived from high-phos- ters of short, fine lateral rootlets produce organic acids
phate limestones are generally highest in phosphorus, or other chelating agents, which act to release phospho-
and those derived from sandstones and acid igneous rus from insoluble complexes with Ca, Fe, and Al. This
rocks lowest. Arid regions with calcareous soils usually ability allows these plants to acquire phosphorus (and
are high in phosphorus because substantial leaching has sometimes other mineral nutrients) in soils where it is
not occurred. Very old soils, particularly in the Tropics unavailable to other plants. Some unrelated plants—
and Subtropics, such as those in Australia and parts of such as a Lupinus sp. (lupine, Fabaceae)—have also been
Africa, tend to be exceptionally low in available phos- observed to form proteoid roots.
phorus (organic forms readily accessible to plants) due
to steady, slow losses by leaching over an extended peri-
Mycorrhizae
od of time (as well as to the slow transformation to less-
available organic forms). While proteoid roots are an unusual and highly special-
However, the total amount of phosphorus in the soil ized adaptation to plant growth in phosphorus-limited
is only a small part of phosphorus limitation. The avail- conditions, plants have another approach to phosphorus
ability of phosphorus to plants is not necessarily related limitation and other low-nutrient conditions that is
to the total amount present in the soil. Unlike nitrogen, extremely common, widespread, and of overwhelming
most phosphorus present in the soil is bound in ways ecological importance. Symbioses between various fungi
that make it unavailable to plants. Soil phosphorus and the roots of terrestrial plants, called mycorrhizae
chemistry is exceedingly complex and depends on pH (singular mycorrhiza; the word refers to the plant-fungal
and many other factors. Phosphorus in the soil exists in symbiosis, not strictly to the fungus), confer a number of
various organic and inorganic forms, and both organic benefits on the plants. These interactions depend on a
and inorganic phosphates can be tightly bound in ways fungal species successfully infecting and living in or on
that make them unavailable to plants (see Figure 15.18). the roots of a plant. Mycorrhizae are often (but not
Phosphorus can be tightly bound to clay particles, or always) mutualistic symbioses (see Box 4.1), in which the
complexed with Ca, Fe, Al, and silicates, or bound up in fungus aids the plant in obtaining certain materials from
organic matter. Because much of the phosphorus in the the soil, and is in turn provided with carbon and energy
soil is tightly bound, it leaches much less readily and in the form of carbohydrates by the plant.
more slowly than does nitrogen. Microbial activity both Mycorrhizae are not only common, they are ubiq-
removes available forms of phosphorus from the soil uitous—they are found on almost all terrestrial plants
solution and transforms them to forms unavailable to and in almost all ecosystems. The ability of plant species
plants (that is, it immobilizes them), as well as trans- to survive in many environments depends on mycor-
rhizae. They are critical components of ecosystem func- endomycorrhizae are subdivided into three groups. The
tion in many environments as well, and play a major role most important of the endomycorrhizae are the vesicu-
in nutrient uptake by plants and in nutrient recycling in lar-arbuscular mycorrhizae (VAM or AM; not all arbus-
soils. About 80% of angiosperms and all gymnosperms cular mycorrhizae make vesicles). Ericoid (specialized
are involved in some form of mycorrhizal symbiosis. for symbioses with plant species such as heathers, blue-
Mycorrhizae occur in all plant divisions, including some berries, and rhododendrons) and orchidaceous mycor-
mosses and ferns. There are also some plants that never rhizae (specialized for symbioses with orchids) are
form symbioses with mycorrhizal fungi; mycorrhizae sometimes classified on their own and sometimes as
are almost never found in plants in the families Brassi- members of the endomycorrhizae. The ectendomycor-
caceae and Amaranthaceae, and are rarely found in the rhizae have characteristics of both the ECM and eri-
Proteaceae (or in other species with proteoid roots). coid groups. The kinds of plants with which each of
The overwhelming importance of mycorrhizae in these groups is associated, and some of the fungal taxa
natural ecosystems and throughout the plant kingdom involved, are listed in Table 4.3. Plants can be hosts to
is often overlooked. Mycorrhizae are sometimes con- one or many fungal species; some plant species have
sidered only within the general context of symbioses both VAM and ECM.
formed between plants and soil organisms. These plant- The most common type of mycorrhizae is the VAM,
fungal symbioses differ in important ways from the sym- which occur in both herbaceous and woody species.
bioses between plants and nitrogen-fixing bacteria. The VAM are most abundant in systems where phosphorus
latter, of course, involve bacteria (prokaryotes), while is limited and in warmer or drier climates. They pre-
the former involve fungi (complex, multicellular eukary- dominate in tropical ecosystems and are especially im-
otes). Mycorrhizae are found on most plants in most portant to many crops. The body of the VAM fungus
places, while nitrogen-fixing symbioses are found only grows branched inside the cortex cells of the roots, form-
on particular plant species, and only in particular places. ing a structure called an arbuscule, and in the intercel-
Nitrogen fixation is the process of transforming ele- lular spaces between root cells, with hyphae extending
mental N2 from the atmosphere into ammonia, and thus out several millimeters (or more) into the soil (Figure
involves a single nutrient, nitrogen. Mycorrhizal sym- 4.15).
bioses, on the other hand, function in a wide variety of Ectomycorrhizae are most commonly found on
ways. They greatly facilitate the availability and absorp- woody plant species, including those in the Pinaceae,
tion of multiple nutrients and soil water as well as pro- Betulaceae, Fagaceae, and Salicaceae, and in some trop-
tecting plants from pathogens. It is possible for a species ical and subtropical trees. ECM are particularly common
to have both symbiotic nitrogen-fixing bacteria and myc- in northern coniferous and temperate deciduous forests
orrhizae. For legumes growing in phosphorus-deficient and in soils where nitrogen is particularly limiting. In
soils, mycorrhizae can increase nodulation and nitrogen fact, coniferous trees cannot grow in nature without these
fixation by the bacteria as well as host plant growth. fungal symbionts. ECM are distinguished by two struc-
tures, one inside the root and one outside it: the Hartig
Major Groups of Mycorrhizae net, a complex of mycelia that grows in between the root
There are two major, common groups of mycorrhizae: cortical cells, enmeshing them, and the mantle, a dense
ectomycorrhizae (ECM) and endomycorrhizae. The network of hyphae that partially or fully ensheaths the
Table 4.3 Major groups of mycorrhizal associations
Mycorrhizal group Plant taxa involved Fungal taxa involved

Ectomycorrhizae (ECM) Dipterocarpaceae (98%), Pinaceae (95%), Basidiomycetes, some ascomycetes,
Fagaceae (94%), Myrtaceae (90), Salicaceae (83%), some zygomycetes
Betulaceae (70%), Leguminosae (16%), and some others
Endomycorrhizae Most plant families
VAM By far most common mycorrhizae in gymnosperms Glomales (common genera are
except Pinaceae, and in angiosperms, except for non- zygomycetes belonging to the Glomus,
mycorrhizal families—Brassicaceae, Portulaceae, Acaulaspora, Gigaspora,
Caryophyllaceae, Proteaceae, etc.) and Sclerocytis)
Ericoid Many species in the Ericales Basidiomycetes and some ascomycetes
Orchidaceous Members of the Orchidaceae (orchids) Basidiomycetes and some ascomycetes
Ectendomycorrhizae Some species in the Ericales; some gymnosperms Basidiomycetes and some ascomycetes
Figure 4.15
Fungal spore Association of a vesicular-arbuscular (VA) endomycorrhizal fun-
gus with a plant root. The fungal hyphae, shown in green, grow in
the intercellular spaces and may penetrate cells of the root’s cortex.
During this process, the fungus can form structures that are either
oval (vesicles) or branched (arbuscules). Fungal mycelium grow-
ing on the outside of the root can form reproductive spores. (After
Mauseth 1988.)
Arbuscule
Corticle cell
of root
The Role of Mycorrhizae
in Plant Phosphorus Nutrition
Vesicle There are several mechanisms by which both VAM and
ECM improve phosphorus uptake. VAM in particular
are highly efficient at scavenging available phosphorus
from the soil, due in part to the phenomenally large sur-
Root hair face area of the many small hyphae that extend into the
soil. In addition, a number of active metabolic process-
es enhance nutrient uptake. For example, both VAM and
External Root ECM produce enzymes (acid phosphatases), which are
mycelium secreted from the external hyphae into the surround-
of fungus ing soil. These enzymes release phosphorus that is tight-
ly bound to organic matter and otherwise unavailable
to the plant. The phosphorus, once it is made available,
outside of the root (Figure 4.16A). Plant species with is taken up by the fungus, and what is not used by the
ECM grow unusual-looking, short, stubby roots (Figure fungus itself is transferred to the plant.
4.16B); this growth form may be controlled hormonally Plants with mycorrhizae can, on average, take up
by the fungus. several times the amount of phosphorus as nonmycor-
The different branched and finely divided structures rhizal plants, and as a consequence, can grow much
of VAM and ECM, including hyphal strands growing more successfully on soils in which phosphorus is low.
into the soil, function to increase nutrient uptake from However, the phosphorus uptake role of mycorrhizae is
the soil. Nutrients are taken up from the soil by the fun- highly variable. Some species of mycorrhizal fungi are
gal mycelium and transferred from the fungal cells to highly effective at taking up phosphorus and transfer-
the root cells of the host plant. ring it to their hosts, while others provide very little
(A) (B)
Cortical cells
of root Figure 4.16
(A) The hyphae of an ectomycor-
rhizal fungus, shown here in
Hartig
green, form an interior Hartig net
net between the cortical cells of a
plant root. In addition, a mantle
of hyphae covers the exterior of
the root. (After Rovira et al. 1983.)
(B) A eucalyptus root mantled in
fungal hyphae. Such a covering
benefits the plant by greatly
increasing the amount of area
available for the absorption of
Mantle water and mineral nutrients.
(Photograph © R. L. Peterson/
Biological Photo Service.)
phosphorus to their hosts. Environmental factors are also toward the root and helping to bind the soil particles
important in determining whether or not phosphorus together, decreasing erosion (which is especially impor-
uptake is enhanced by the fungus. At extremely low lev- tant in sands and sandy soils, as in beach dunes). ECM
els of phosphorus, the fungi are apparently not able to fungi form long, thick mycelial structures called rhi-
function well, and the degree to which roots become zomorphs that can extend several meters into and across
infected with the mycorrhizal symbionts is low. At the soil. These rhizomorphs are capable of carrying
slightly higher phosphorus levels, the fungi colonize water rapidly from the soil to the root.
roots more successfully and become highly effective at One of the most surprising effects of mycorrhizae is
enhancing phosphorus uptake for the plant. At high lev- their protection of the host plant from bacterial and fun-
els of available soil phosphorus, fungal growth inside gal plant diseases, as well as nematode attack. Both ECM
the root is apparently actively suppressed by the plant, and VAM can offer this protection. The mechanisms for
and the roots take up phosphorus directly on their own. this protective function are diverse, particularly in ECM.
The fungus may either secrete pathogen-killing chemi-
Other Functions of Mycorrhizae cals directly into the rhizosphere or stimulate the root to
High levels of nitrogen in the soil may also act to sup- produce such compounds. The fungal sheath of ECM
press infection by mycorrhizal fungi. Other factors, such can directly shield the root from pathogens and plant
as aridity, toxic minerals, and inadequate or excess con- root-consuming invertebrates as well.
centrations of other minerals (such as, for example, too
little boron), may also inhibit the formation and function Orchids and Their Mycorrhizal Associations
of mycorrhizae because these conditions either harm the Orchids may have a special obligate relationship with
fungal partner or function in some more complex way mycorrhizal fungi, depending on them for ordinary
to inhibit the symbiosis. growth and development. Orchid seeds are exception-
In addition to enhancing phosphorus availability, ally tiny, with no stored food reserves. The presence and
VAM increase plant access to some other minerals, such activity of mycorrhizal fungi are necessary for seed ger-
as copper and zinc. ECM tend to have a much more mination and early seedling development and growth.
diverse role. If the fungal mantle of ECM totally encas- Particularly when the plant is young, the fungus may
es the roots, all water and mineral uptake must take supply it with carbon and energy in the form of carbo-
place through the fungus. Because the surface area of the hydrates, rather than receiving them from the plant.
mass of hyphae is far greater than that of the root sys- Some orchids never become photosynthetic and depend
tem, the uptake of water, nitrogen, and other dissolved on mycorrhizae throughout their life span for all of their
minerals is greatly enhanced, Some ECM produce and nutrients, including carbohydrates. It is not known
excrete organic acids, increasing the solubility and where these fungi obtain their carbon, but they may get
uptake of various minerals, somewhat like proteoid it by parasitizing other plants or by breaking down
roots. Some ECM and ericoid mycorrhizae secrete organic matter in the soil (that is, by functioning as
enzymes that degrade proteins, making the nitrogen saprophytes, like many nonmycorrhizal fungi).
locked in these complex organic molecules accessible to Not enough work has been done on orchid-mycor-
their hosts. ECM fungi can enable their hosts to toler- rhizal interrelationships (particularly in the Tropics,
ate toxic heavy metals in the soil, protecting the roots by home to the majority of orchids) to understand the
sequestering and binding heavy metals and other toxic extent and nature of this extreme form of plant depend-
ions in their cell walls. ence. If only because the orchids are the most highly
Plants in the heath family (Ericaceae) and in some diverse group of plants on Earth, orchidaceous mycor-
other families in the Ericales generally live in acidic, rhizae present an interesting and important topic for
extremely nutrient-poor habitats. The formation of eri- future ecological and evolutionary research.
coid mycorrhizae is apparently necessary for them to
survive, grow, and reproduce in their natural habitats, Mutualism or Parasitism?
and they clearly depend on these associations to obtain You may wonder, why would any self-respecting fun-
mineral nutrients from these infertile soils. Few, if any, gus do so many nice things for a plant? Soil fungi are
nonmycorrhizal plants live in these habitats. heterotrophic, like animals, and typically degrade organ-
Mycorrhizae can also exert a substantial effect on ic matter or parasitize living things to make a living.
plant water relations. VAM often decrease the negative (Others are predators, particularly on soil nematodes).
effects of drought on the host plant. It is possible that Mycorrhizal associations are a way in which fungi can
this is an indirect effect of the fungus on plant mor- obtain carbon and energy from a living host plant. Gen-
phology and nutrient status. In addition, the fine hyphae erally, they appear to be mutualistic associations
extending into the soil from VAM may change the soil (although how generally is not yet known), with the
structure itself, increasing the movement of water plants supplying carbohydrates obtained by photosyn-
thesis to the fungi and receiving mineral nutrients and A closely related issue is the potential for such inter-
other benefits from the fungi. connections to alter competitive relationships among
However, like any relationship, this one is not individuals. Mycorrhizal nutrient transfer could poten-
inevitably benign. One partner (either the plant or the tially intensify such competitive interactions or amelio-
fungus) can become parasitic on the other, receiving ben- rate them. It could alter the competitive hierarchy, mak-
efits without supplying them. Sometimes mycorrhizal ing inferior competitors more successful, or it could help
fungi do not provide any minerals or other detectable the dominant competitor. We do not yet know how to
benefits to their hosts, yet continue to receive nutrients. predict the effects of such interconnections on competi-
Sometimes only minor amounts of nutrients are pro- tion, nor do we have any idea how common or how
vided by the fungus, the nutrients made available are important such effects are in nature. Future research on
not needed or do not enhance plant growth, or the ben- mycorrhizae is certain to come up with many surprises.
efits provided are vastly outweighed by the carbon and
energy drain on the plant. Sometimes the plant provides Summary
little benefit to the fungus, yet receives benefits from it,
as in the case of the orchid mycorrhizal association (at Soil is a complex product of the interaction between liv-
least some of the time). ing organisms and their terrestrial substrate. It includes
In some cases, the plant may be capable of discard- physically and chemically altered products derived from
ing the fungus when phosphorus is available, accepting rock and from organic materials, along with air and
the fungus only when it is of benefit to the plant. The fun- water. Soils vary from one region to another, and with-
gus may, on the other hand, be the controlling member in local environments, in many different aspects. There
of the pair, not only determining the association but con- is a tremendous diversity of soil properties and charac-
trolling the plant’s growth, development, and alloca- teristics, and these differences affect the vegetation found
tion patterns by producing hormones that override the growing in the soil as well as affecting the growth of
plant’s own hormonal signals. Nonmycorrhizal plants individual plants.
have been observed to use a variety of mechanisms to Soil texture describes the relative proportions of
actively reject fungal infection. clay, silt, and sand particles in the soil. These particles
differ from one another in size, shape, and mineral com-
Plant Interconnections via Mycorrhizal Fungi position, and they impart different characteristics to
Among the host of strange and interesting phenomena soils. The pH of soil also has an important influence on
associated with mycorrhizae, one that has fascinated plant growth and on what plants are found growing in
many plant ecologists is the potential interconnection the soil. One of the important factors in characterizing
between different individual plants via mycorrhizal local soils is the soil profile, which consists of often dis-
hyphae. Two plants that are infected with the same myc- tinct soil horizons, or layers at different depths. In addi-
orrhizal fungus can be connected by a hyphal strand tion to the physical factors responsible for soil proper-
running through the soil. Even plants of completely ties, soil organic matter is a critical component in
unrelated species or genera can be connected in this way. determining soil structure.
Phosphorus, nitrogen, water, and even carbohydrates Plants depend on the soil to obtain water. Different
have been found to move between plants, apparently soils hold different amounts of water in the macrop-
through these hyphal connections. However, it is diffi- ores and micropores of the soil structure. As a soil dries
cult in practice to distinguish such a “pipeline” effect after a rain, water first drains out of the macropores, and
from that of substances leaking out of the roots or then moves by capillary action through the micropores
hyphae of one plant and being immediately picked up to plant roots. Eventually, as the soil water potential
by the roots or hyphae of another. declines, water becomes less and less available to plants,
The major unresolved question is, to what extent is causing wilting and ultimately death unless more water
this kind of transfer important in natural ecosystems? enters the soil.
Some ecologists have argued that nutrients may be Plants also depend on the soil for the mineral nutri-
transferred from dominant individuals to seedlings, to ents that are essential to their survival and growth.
understory species in forests, or to competitively infe- Nitrogen and phosphorus are generally the most limit-
rior plants, allowing them to survive and persist in the ing mineral nutrients for plants. Both are needed by
plant community. These plants would then be, at least plants in fairly large quantities. Nitrogen availability,
to some extent, parasitizing the plants from which they in particular, depends on bacterial activity, including
were drawing carbon or mineral nutrients. We do not symbiotic nitrogen fixation. The availability of these two
yet know whether this phenomenon is a rare curiosity nutrients varies greatly in soils because nitrogen is eas-
or a common occurrence, or how important it is in deter- ily lost from soils, while phosphorus is often present in
mining plant community composition. unavailable forms.
Plants may depend on a symbiotic relationships natural habitats. Two of the most important types of
with fungi living on or in their roots, called mycorrhizae, mycorrhizae are ectomycorrhizae (ECM) and endomy-
to obtain phosphorus and other minerals from the soil. corrhizae, particularly vesicular-arbuscular mycorrhizae
Plants may obtain other benefits from these associations (VAM). The relationships between plants and mycor-
as well. Most terrestrial plants have mycorrhizae and rhizal fungi are complex and variable, and are impor-
depend on them to survive, grow, and reproduce in their tant in the function of many ecosystems.
Additional Readings
Classic References Newsham, K. K., A. H. Fitter and A. R. Watkinson. 1995. Multi-func-
tionality and biodiversity in arbuscular mycorrhizas. Trends Ecol. Evol. 10:
Lyon, T. L. and H. O. Buckman. 1922. The Nature and Properties of Soils.
407–411.
Macmillan, New York. [Subsequent editions, H. O. Buckman and N. C.
Brady 1952, 1960, 1969.] Fitzhugh, R. D., C. T. Driscoll, P. M. Groffman, G. L. Tierney, T. J. Fahey
and J. P. Hardy. 2001. Effects of soil freezing disturbance on soil solution
Current version:
nitrogen, phosphorus, and carbon chemistry in a northern hardwood
Brady, N. C. and R. R. Weil. 2001. The Nature and Properties of Soils. ecosystem. Biogeochemistry 56: 215–238.
13th ed. Prentice-Hall, Upper Saddle River, NJ.
Additional Resources
Jeffrey, D. W. 1987. Soil-Plant Relationships: An Ecological Approach.
Batty, A. L., K. W. Dixon, M. Brundrett and K. Sivasithamparam. 2001. Croom Helm, London, and Timber Press, Portland, OR.
Constraints to symbiotic germination of terrestrial orchid seed in a
mediterranean bushland. New Phytologist 152: 511–520. Pimentel, D. (ed.). 1993. World Soil Erosion and Conservation. Cambridge
University Press, Cambridge.
Brundrett, M. 1991. Mycorrhizas in natural ecosystems. Adv. Ecol. Res. 21:
171–313. Rendig, V. V. and H. M. Taylor. 1989. Principles of Soil-Plant Interrela-
tionships. McGraw Hill, New York.
PART
II Evolution and
Population Biology
5. Processes of Evolution 87
6. Outcomes of Evolution 101
7. Population Structure, Growth and Decline 117
8. Growth and Reproduction of Individuals 143
9. Life Histories 167
C H A P T E R
5 Processes of Evolution
I
n Part I, as we explored the various ways in which plants interact with
their environments, we looked at the features of plants that permit them to
cope with different environments. All of those features arose through the
process of evolution by natural selection. The most amazing aspect of this process
is how a few simple consequences of biology and natural laws have resulted in
the vast diversity of life. All of the species that compose ecological communi-
ties arose through evolution, and ecological processes provide the context for
evolution. The ecologist G. Evelyn Hutchinson emphasized this intimate rela-
tionship in his seminal book, The Ecological Theater and the Evolutionary Play (1965).
The intertwining of evolution and ecology is greatest in the process of natural
selection.
This chapter and the next summarize the basic principles and processes of
evolution and explore its outcomes. In this chapter, we cover the process of
natural selection as well as other processes that affect evolution. In Chapter 6,
we examine a series of specific examples of evolution by natural selection.
Natural Selection
Natural selection occurs when individuals with differing traits leave different
numbers of descendants because of those differences. Evolution by natural
selection occurs when those differences are heritable (have a genetic basis).
Adaptive traits are ones that have come about through the process of evolu-
tion by natural selection. The suite of traits associated with CAM photosyn-
thesis (see Chapter 2), for example, is adaptive in hot, dry climates because
individuals that have those traits are able to leave more descendants than indi-
viduals that do not. Arctic plants are short-statured because by remaining close
to the ground they remain warmer, grow more, and ultimately leave more off-
spring.
The principles of natural selection were first proposed by Charles Dar-
win in his book On the Origin of Species by Means of Natural Selection, published
in 1859. Natural selection is one of the four central processes of evolution.
The others are mutation, migration, and genetic drift.
Variation and Natural Selection

An important starting point for any discussion of evolution and natural selection
is variation. Variation is ubiquitous in nature. Nearly all natural phenomena vary
88 Chapter 5
at some level. This principle is expressed in the adage that The genotype of an individual is the information
“no two snowflakes are exactly alike.” Living beings are contained in its genome—the sequence of its DNA. This
much more complex than snowflakes, with an even information is expressed through the processes of tran-
greater potential for different kinds and amounts of vari- scription of DNA to RNA and translation of the RNA
ation. Evolution requires two kinds of variation: pheno- into protein. Gene expression, and the development of
typic variation and genetic variation. the plant as a whole, is controlled by the information
Plants, like most living organisms, vary phenotyp- contained within the genome itself through various
ically. The term phenotype refers to all of the physical feedback mechanisms. A seed contains all of the infor-
attributes of an organism. These properties include mation necessary for the growth and development of the
aspects of outward appearance (such as height, leaf size adult plant. Development can be thought of as the
and shape, flower color, or fruit number), behaviors unfolding of the information contained in the genome.
(such as growth of roots toward water), life history char-
acteristics (such as being an annual), internal anatomy, The Components of Natural Selection
and the content of cells (such as protein composition). Natural selection requires three components: phenotypic
Phenotypic variation can be extensive for traits such variation among individuals in some trait; fitness differ-
as seed size, which can vary 20-fold within a single pop- ences (some individuals must leave more descendants
ulation of a wildflower species (Figure 5.1), or biomass, than others as a result of the phenotypic differences), and
which can vary over several orders of magnitude in trees. heritability (the phenotypic differences must have a
Other kinds of traits, such as flower size and solute con- genetic basis). If these three components are present for
centrations in cells, tend to vary much less among indi- a trait within a population, then the frequency of that
viduals in a population. Still other traits, such as petal trait will change in that population from one generation
number or photosynthetic pathway, tend to be invariant to the next: evolution by natural selection will occur.
within a species. However, even these traits vary within Fitness differences include differences in mating
some species (see Chapter 2). Furthermore, the particular ability, fecundity (number of gametes produced), fertil-
pattern of growth and development exhibited by an indi- izing ability, fertility (number of offspring produced),
vidual plant depends on a complex interaction between and survivorship (chance of surviving). What causes
its genes and its environment. All of these differences these fitness differences? It could be chance. But often
among individuals result in phenotypic variation. these differences are associated with differences in some
trait of the plants. Phenotypic selection occurs when
individuals with different trait values have consistent
differences in fitness.
30
A useful and convenient way to study the process
of natural selection is to subdivide it into two parts: that
which occurs within a single generation—phenotypic
25 selection—and that which occurs from one generation
to the next—the genetic response (Endler 1986). Pheno-
typic selection consists of the combination of phenotypic
Percentage of sample
20 variation and fitness differences. This part is what we

often think of as natural selection. But for evolution to
occur, the other part—the genetic response—is equally
15 important. The genetic response is the change in the
Coreopsis lanceolata genetic makeup of the population that occurs from one
generation to the next. The genetic response depends on
10 the heritability of a trait. If a trait is heritable, then if phe-
notypic selection favors that trait in one generation, the
next generation will have a greater proportion of indi-
5 viduals with that trait.
In a population of the grass Danthonia spicata, for
0
example, individuals with longer leaves had higher sur-
0.0 0.4 0.8 1.2 1.6 2.0 vivorship and fecundity than individuals with shorter
Seed mass (mg) leaves (Figure 5.2A). Variation in leaf length is known to
be at least partially under genetic control. So the next
Figure 5.1 Variation in individual seed mass in a popula-
tion of Coreopsis lanceolata (Asteraceae) growing on an inland generation, on average, will be likely to have longer
dune south of Lake Michigan. (Data from Banovetz and leaves. From one generation to the next, the change in
Scheiner 1994.) the population may be small. But if the process of natu-
Processes of Evolution 89
(A) (B)
5 6
4
5
3
4
Relative fitness
Relative fitness
2
1 3
0
2
–1
1
–2
–3 0
0 4 8 12 16 20 –4 –2 0 2 4
Leaf length (cm) Late growth rate
(C) (D)
5 4
Relative fitness
4 3
3 2
Relative fitness
2 1
1 100
80
60 8
0 6
Water use 40
efficiency 4
20
2
–1 0 0 Leaf length (cm)
0 1 2 3
Corn earworm damage
Figure 5.2 Relationships between phenotypic variation in plant traits and fitness differ-
ences result in phenotypic selection. If genetic variation for these traits also exists, then evo-
lutionary change may occur. (A) Directional selection for longer leaf length in the perennial
Danthonia spicata (poverty grass, Poaceae) in a natural population growing in a white pine-
red oak forest in northern lower Michigan. Fitness was measured as the total number of
spikelets produced over five years and was corrected for correlations with other traits. (Data
from Scheiner 1989.) (B) Directional selection for greater growth rate late in the season in the
annual Impatiens capensis (jewelweed, Balsaminaceae) in a natural population growing in a
deciduous forest in Wisconsin. Fitness was measured as the final dry weight of the plant,
which is correlated with the number of seeds produced. The fitness function is curved, but
still monotonically increasing. (After Mitchell-Olds and Bergelson 1990.) (C) Stabilizing selec-
tion for the amount of damage caused by the corn earworm to the annual Ipomoea purpurea
(morning glory, Convolvulaceae) in an experimental population growing in an old field in
North Carolina. Fitness was measured as the number of seeds produced by the end of the
growing season. (Data from Simms 1990.) (D) Stabilizing selection for water use efficiency
and directional selection for smaller leaf size in the annual Cakile endentula (sea-rocket, Bras-
sicaceae) in an experimental population growing on a beach along southern Lake Michigan.
Fitness was measured as the number of fruits produced by the end of the growing season.
The optimal water use efficiency depends on leaf size, which is an example of correlational
selection. (After Dudley 1996.)
90 Chapter 5
ral selection goes on for many generations, the popula- sunny conditions, it is better for a plant to use its
tion may become very different from its ancestor. Much resources to produce more and larger leaves and more
of the striking variation among species is due to such flowers.
long-term evolutionary responses to natural selection. Plants need some way to sense how shady their
Directional selection occurs when individuals with environment is. A key difference between sunny areas
the most extreme value for a trait have the highest fit- and shady areas is the ratio of red light (wavelength =
ness (Figure 5.2A,B). In this case, the population will 665 nm) to far-red light (wavelength = 730 nm). In open
continue to evolve in a single direction over time, if no areas, this ratio is about 1.15, while under a leafy canopy
other processes interfere. The primary change will be in it ranges from 0.15 to 0.97. Plants sense differences in
the mean value of the trait. Stabilizing selection occurs light quality through a group of pigments called phy-
when individuals with intermediate trait values have tochromes (Smith and Whitelam 1990). Phytochrome A
the highest fitness (Figure 5.2C). Under stabilizing selec- is labile, changing its form in the presence of light of dif-
tion, the mean value of the trait will not change, but ferent wavelengths. The Pr form, when exposed to red
the variability in that trait will decline. Under both direc- light, changes to the Pfr form. This form, in turn,
tional selection and stabilizing selection, all additive responds to far-red light, which changes it back to the
genetic variation for the selected trait will eventually dis- Pr form. The other four phytochromes are stable in
appear, unless other processes introduce new variation. form, but interact with phytochrome A. Together, these
Phenotypic variation may remain, however, due to two phytochromes turn different suites of genes on or off,
other factors: plasticity and errors of development. Cor- resulting in the different growth forms.
relational selection occurs when the pattern of selection Phytochromes are also used by plants to determine
on one trait depends on the value of another trait (Fig- the length of the night. The Pfr form of phytochrome
ure 5.2D). Correlational selection can result in highly A slowly, but spontaneously, changes back to the Pr
coordinated suites of traits and complex adaptations. form after some time in the dark. So the Pfr:Pr ratio is
a cue to day length, and thus season, in temperate parts
Phenotypic Variation and Phenotypic Plasticity of the globe. Flowering, seed dormancy, and seed ger-
Phenotypic differences among individuals can result mination are among the physiological responses that
from three types of variation: genetic, environmental, are mediated by phytochromes in many species.
and developmental. The genotypes of individuals usu- Besides affecting gene expression, the environment
ally differ, and this genetic variation often results in phe- can influence plant growth directly. The amount of
notypic variation among individuals. Individuals also water available in the soil, for example, has a direct
experience different environments as they develop and effect on plant size. Opening the stomata to take up CO2
grow. The environment can vary in many ways, even during photosynthesis results in the loss of water (see
over very small distances. Each grass shoot in a seem- Chapter 3). If a plant does not have an adequate supply
ingly uniform meadow, for example, experiences a of water, it will close its stomata. This closure causes a
somewhat different environment. Individuals with dif- reduction in the amount of carbon fixed by the plant
ferent genotypes may respond differently to their envi- and can result in reduced growth. Likewise, in shady
ronment, and their development may proceed differ- environments, the number of photons of light energy
ently. Even two individuals with identical genotypes available to be captured for photosynthesis is limited,
growing in identical environments, however, do not and this limitation affects both photosynthesis and plant
necessarily look alike or function identically. Small, ran- growth.
dom differences in when and how genes are expressed,
called errors of development, can lead to measurable Heritability
differences in the adult plant.
Phenotypic plasticity is the capacity of a genotype Although at times it may seem as if the world consists
to give rise to different phenotypes in response to dif- of an infinite variety of species, that is not the case. It is
ferent environmental conditions. In the species Impa- easy to imagine organisms that could exist but do not,
tiens capensis (jewelweed, Balsaminaceae), for example, such as unicorns or dragons. Why not? One part of the
plants grown in the shade tend to be tall and spindly, answer is that the genetic basis of traits constrains evo-
while those grown in the sun tend to be short and stout lution. For evolution to occur, there must be appropri-
(see Figure 6.6; Schmitt et al. 1995). Shady conditions ate genetic variation. What limits this variation? To
can be due to competition from neighboring plants. answer that question, we must first understand what we
Under those conditions, it is better for a plant to grow mean by genetic variation in an evolutionary context.
taller quickly so that it can shade out its neighbors Then we can explore the ways in which the environment
rather than be shaded itself. On the other hand, in interacts with genes to determine that variation.
Resemblance among Relatives there is a resemblance, but some offspring are taller than
2
Heritability (h ) is the amount of resemblance among their parents, while others are shorter. Negative correla-
relatives that is due to shared genes. Offspring tend to tions are also possible for some traits in some species,
resemble their parents and their siblings because the although they are very unusual.
phenotype of an individual is determined, in part, by its One measure of the heritability of a trait is the slope
genotype, and an individual receives its genes from its of the line of a regression of offspring trait values on
parents and shares those genes with its siblings. Con- parental values. In the above example, because we used
sider a trait such as height in an annual plant at the end information from both parents, the slope is exactly equal
of the growing season. We might do the following exper- to the heritability. If we had measured only one parent,
iment with Brassica campestris (Brassicaceae): We choose we would have information about only half of the genes
pairs of plants, pollinate one individual with the other being contributed to the offspring, and the slope would
in each pair, and cover the flowers to prevent any other be one-half the heritability.
plant from pollinating that individual. When we col- The other common way of measuring heritability is
lect the seeds at the end of the growing season, we thus to measure the correlation among siblings. Imagine if we
know exactly who the parents were. Before the parental took two seeds from each of many plants. We could ger-
plants die, we measure their heights. Then we plant the minate the seeds and grow the pairs of siblings, meas-
seeds, let them grow, and measure their heights at the ure their heights, and construct a graph much like Fig-
end of the next growing season. ure 5.3, except that now the axes would be the heights
We then plot the height of the offspring against the of the two siblings, and each point would represent a
height of their parents (Figure 5.3). In this case, we find sibling pair. Again, the slope would measure heritabili-
that taller parents tend to produce taller offspring. We ty, with the exact relationship depending on whether the
measure this tendency using a statistical technique called plants shared both parents or just one parent. We can do
correlation (see Appendix). If offspring always exactly such an analysis with cousins or with any individuals
matched their parents, the correlation between parental that were related as long as we knew their relationships.
height and offspring height would be 1.0. If there were no Nor are we restricted to using pairs of individuals. Var-
relationship, the correlation would be 0.0. In our exam- ious statistical techniques can be used to measure heri-
ple, the correlation is 0.41 and the slope of the line is 0.21; tability in groups of related plants with different degrees
of relatedness.
There is a critical distinction between the heritabili-
6 ty of a trait and whether that trait has a genetic basis. Her-
itability requires that phenotypic differences among indi-
viduals be due, at least in part, to genetic differences
among those individuals. In Box 5A we describe a case
5
in which height is genetically determined. In that exam-
ple, some individuals have a genotype of AA, some Aa,
and some aa. Instead, imagine that all individuals in the
Offspring height (dm)
4 population have the same genotype, AA. Assume, how-

h2 = 0.21 ever, that height also depends on the amount of nitrogen
in the soil. If the population is growing in a field that
varies in soil nitrogen from spot to spot, then individu-
3
als will differ in height. However, none of those pheno-
typic differences will be due to differences in genotypes.
If we were to measure these plants, collect seeds, and
2 raise the offspring in that same field, the correlation
between parental height and offspring height would be
0, and the heritability of height in that population would
be 0. Yet, there is still a gene that determines height.
1
0 1 2 3 4 5 6 7 This example also demonstrates that the heritabili-
Mean parental height (dm) ty of a trait depends on the frequencies of its alleles in
the population. When the frequency of A is 1.0—all indi-
Figure 5.3 Plot of offspring height against mean height of viduals have the AA genotype—the heritability of the
the parents in the annual Brassica campestris (Brassicaceae)
for plants grown in a greenhouse. The heritability of this trait is 0. Thus, heritability estimates for the same trait
trait (the slope of the line) is 0.21. (Unpublished data cour- can differ among populations, or for the same popula-
tesy of Ann Evans.) tion measured at different times. Heritability estimates
92 Chapter 5
BOX 5A
A Simple Genetic System and the Resemblance of Relatives

This example is based on the work that value of the parents’ phenotypes, 80 their parents. The exact heritability of a
Gregor Mendel did in the nineteenth cm, exactly equals the mean value of (A) Strict additivity
century with garden peas (Pisum sativa, the offspring phenotype. Note that for
100
Fabaceae), although the details have this cross, no parent or offspring is actu-
been modified for illustrative purpos- ally at the mean height; the mean is a
es. Although it is based on a simple, descriptor of the group, not a property
Offspring height (cm)

one-locus system, in fact nearly all traits of any particular individual. A graph of h2 = 1.0
of ecological interest are based on sev- mean parental phenotype against mean
60
eral to many loci. However, the same offspring phenotype (part A of the
principles hold no matter how many accompanying figure) has a slope of 1.0.
loci affect a trait. That is, the heritability of this trait is 1.0,
Consider a simple genetic system in because we can perfectly predict the
which plant height is determined by a average offspring phenotype from our
single diploid locus. We assume that knowledge of the parental phenotypes. 20
individuals with genotype AA are tall
(100 cm) and those with genotype aa Case 2: Dominance 20 60 100
are short (20 cm). We also assume that Now assume that A is dominant to a, Mean parental height (cm)
there are no environmental effects such that Aa individuals are 100 cm tall.
(VE = 0 and VG×E = 0) or errors of devel- In this case, predicting offspring phe- (B) Dominance
opment (Ve = 0). notypes becomes more difficult. If both 100
parents are short, all offspring will be
Case 1: Strict Additivity short. But if both parents are tall, their
If individuals with genotype Aa have a genotypes could both be Aa, or both be Offspring height (cm)
phenotype that is exactly intermediate AA, or one AA and the other Aa. In the h2 = 0.67
between AA and aa individuals, then latter two instances, all offspring will be 60
genetic variation is strictly additive. In 100 cm tall. But if both parents are Aa,
this case, Aa individuals would be then 1/4 of the offspring will be aa and
intermediate in size (60 cm tall). Sup- will be short. The mean offspring phe-
pose we want to predict the phenotypes notype will be 80 cm (3/4 × 100 + 1/4 ×
of the offspring of a cross. Because the 20), even though the mean phenotype 20
effects of the alleles are strictly additive, of the parents was 100 cm. If we assume
we can do so. If both parents are tall, the that both alleles exist with equal fre- 20 60 100
cross will be AA × AA, and all offspring quency in our population, then a graph Mean parental height (cm)
will be tall. If both parents are short, the of mean parental phenotype against
cross will be aa × aa, and all offspring mean offspring phenotype will have a
Plot of offspring phenotype against mean
will be short. If one parent is tall and the slope of 0.67 (see the accompanying fig- parental phenotype for two genetic systems.
other short (AA × aa), all offspring will ure). The heritability of the trait is less The slope of the regression line is the heritabili-
be 60 cm tall (Aa). That is also the height than 1.0 because some of the genetic ty of the trait. (A) Strictly additive; slope = 1.
that we get by averaging the parental variation is nonadditive due to the (B) Complete dominance; slope = 0.67. These
phenotypes; the mean offspring pheno- dominance relationship. In other heritabilities assume equal frequencies of the
type equals the mean value of the par- words, some offspring differ pheno- two alleles in the population.
ents’ phenotypes. If one parent is 100 cm typically from their parents because of
tall and the other is 60 cm tall (AA × Aa), the effects of dominance; if they do not trait in a population depends on both
half the offspring will be 100 cm tall and inherit the dominant allele from either the degree of dominance and allele fre-
half will be 60 cm tall. Again, the mean of their parents, they do not resemble quencies in the population.
are always specific to the population and environment sider the various sources of phenotypic variation
in which they are measured. described above. If we measure height in a population
of annual plants, it will vary due to differences in geno-
Partitioning Phenotypic Variation type, environment, and errors of development. The per-
Another way of thinking about heritability is to con- centage of that variation that is caused by genetic dif-
ferences is heritability. Mathematically, we can express do not depend on their environment. That is, we assume
this idea as follows. First, consider the total phenotypic that if an individual is 10% taller than another when
variation, which we symbolize as VP. By a combination growing in one environment, it will still be 10% taller in
of experimental and statistical techniques, we can deter- a different environment. But what happens when this
mine how much of that variation (statistical variance; assumption does not hold? Suppose, for example, that
see the Appendix) is due to different causes. Breaking when a certain plant species is grown under shady con-
up variation into its components is called variance par- ditions, all individuals are small and about the same size,
titioning. In the simplest case, we might be concerned but when it is grown in a sunny spot, some of those indi-
with partitioning the phenotypic variation only into that viduals are much taller than the others due to genetic
part due to genetic variation (VG) and that part due to differences. In other words, the genetic differences are
all other causes (Ve): VP = VG + Ve. Then, heritability apparent in some environments, but not in others.
would be h2 = VG/VP, the percentage of phenotypic vari- These types of differences in genetic expression as a
ation that is due to genetic differences among individ- function of the environment are referred to as genotype-
uals. This concept of heritability is identical to that of environment interactions. Genotype-environment inter-
heritability defined as the resemblance among relatives, actions are the genetic component of phenotypic plas-
just a different way of measuring it. ticity. Adding environmental effects, a full partition of
Genetic variation can be further partitioned, how- the phenotypic variation of a population is
ever. In a diploid organism, dominance occurs when the
VP = VE + VG×E + VG + Ve
expression of an allele depends on the properties of the
other allele at that locus. In both diploid and haploid where VE is variation due to the environment, VG×E is
organisms, epistasis occurs when gene expression variation due to genotype-environment interactions, and
depends on the properties of alleles at other loci. The Ve now refers only to variation due to errors of devel-
amount of genetic variation that is manifested in a pop- opment (Figure 5.4).
ulation can be a result of differences in the direct expres- The presence of variation resulting from genotype-
sion of an allele (additive variation, VA), or to differences environment interactions can have large effects on her-
in combinations of alleles at each locus (dominance vari- itability. Heritability is not simply a result of the genet-
ation, VD), or to differences in combinations of alleles
at different loci (epistatic variation, VI) (see Box 5A).
Again, by various experimental and statistical proce- 2.0
dures, we can partition this variation into its causes:
VG = VA + VD + VI. If heritability is calculated as just 1.8
the percentage of additive genetic variation (h2 =
VA/VP), then we are speaking of narrow-sense heri- 1.6
tability. If it is calculated as the total genetic variation
Plant height (m)
(VG), then we are speaking of broad-sense heritability. 1.4

This distinction is important because the response of a
trait to natural selection depends on its narrow-sense 1.2
heritability.
Heritability values tell us whether there is genetic
1.0
variation for a trait in a population, and if so, whether
there is just a little variation or a lot of variation. In terms
0.8
of evolution, the amount of genetic variation may
impose a constraint on evolution. If there is no genetic
0.6
variation (h2 = 0), the constraint is strong. No matter how
much natural selection there is on a trait, there will be Low High
no genetic response. If there is a little bit of genetic vari- Light environment
ation, the constraint is weak; there will be a genetic
Figure 5.4 An example of phenotypic plasticity and geno-
response, but it will be small, and evolution will proceed type-environment interaction for plants growing in high
slowly. If there is a lot of genetic variation, there is almost light and low light environments. In this example, cuttings
no constraint. were taken from individual genets and grown in each envi-
ronment. Each shaded line connects the mean height for
Genotype-Environment Interactions each genet in the two environments. The dashed line con-
nects the overall means in each environment and represents
How does the environment influence heritability meas- VE. Variation in average heights among genets represent VG.
ures? Our original definition of heritability assumes that The extent to which the shaded lines are not parallel repre-
differences among individuals with different genotypes sents VG×E.
94 Chapter 5
ic differences among individuals: those genetic differ- most of the seeds will germinate in the shade of the par-
ences must result in phenotypic differences. Some kinds ent plant. Because large plants produce many seeds,
of genetic differences among individuals never result in those seeds will be growing under crowded conditions.
phenotypic differences; for example, some types of vari- Plants that are genetically capable of growing larger (the
ation in noncoding regions of the DNA do not do so. In offspring of the large shrubs) will be growing under
other cases, whether genetic differences result in phe- poorer conditions, so genetic and environmental influ-
notypic differences depends on the environment. When ences will be acting on size in opposite directions . There-
genotype-environment interaction variation is present, fore, a negative covariation might exist.
then the amount of expressed genetic variation may dif- Adding the term for covariation to the equation for
fer among environments. In the example given above, phenotypic variation given above, we end up with our
plants grown in the shade were all of similar size; in final equation describing phenotypic variation in a nat-
other words, phenotypic differences were minimized in ural population:
that environment. If the heritability of size were meas-
VP = VE + VG×E + VG + Cov(G,E) + Ve
ured only in the shade, we would conclude that it was
very low because the amount of phenotypic variation Since VP is the denominator in calculating heritability,
would be low. On the other hand, if heritability were changes in any of its components will affect the heri-
measured only in a sunny environment, it would be larg- tability of the trait.
er. Thus, evolution would be constrained in the shady
environment because of a lack of heritable variation.
Patterns of Adaptation
Gene-Environment Covariation Natural selection can result in three different patterns of
The final term that must be included in accounting for adaptation. First, individuals may become specialized
the total phenotypic variation of a trait in a population to perform best in different environments. Within a pop-
is gene-environment covariation, which is abbreviat- ulation, for example, individuals of one phenotype
ed Cov(G,E). A nonrandom relationship between any might have the highest survival during periods of
two factors is called covariation; such a relationship can drought or in the driest spots, while individuals of
be either positive or negative. Covariation is closely another phenotype might survive best during wetter
related to correlation; one is a mathematical transfor- periods or in microhabitats with moister soil. Second,
mation of the other. In this case, we are interested in individuals might become phenotypically plastic, chang-
covariation between genetic and environmental effects ing their form in different environments so as to match
on a trait. Such covariation is often found in nature the most fit trait value for each set of environmental con-
because individuals are usually not randomly distrib- ditions. In response to changes in water availability, for
uted across environments. This is especially true of example, individuals might produce leaves of different
plants, as seeds often end up close to the maternal par- sizes and shapes. This pattern would allow them to have
ent plant. high survivorship during both wet and dry periods.
Positive covariation between genetic and environ- Finally, all individuals in a population might converge
mental effects occurs when genetic traits are positively on a single, intermediate phenotype that does at least
associated with responses to the environment. For passably well in all environments. This pattern is some-
instance, more vigorous competitors might dominate times referred to as a jack-of-all-trades strategy, as in the
small patches of rich soil. In this case, plants that are saying “a jack-of-all-trades is a master of none.”
already genetically capable of growing more rapidly will Which of these patterns of adaptation will prevail
also be growing under better conditions, while slow depends on a complex combination of factors, including
growers will be relegated to poorer conditions. Thus, how variable the environment is, how that variation is dis-
genetic differences in growth rate are exaggerated by tributed in time and space, and how much genetic varia-
environmental effects. tion exists in the population. If the environment varies over
Nonrandom distribution of genotypes can act short time periods, then phenotypic plasticity is often
instead to minimize genetic differences. Negative favored. Usually we think of phenotypic plasticity with
covariation between genetic and environmental influ- regard to traits that become fixed during development,
ences occurs when genetic and environmental influences such as leaf shape. In aquatic plants from many unrelated
have opposite effects. Consider a population of shrubs families, submerged leaves are characteristically feathery
in which larger individuals produce more seeds. If you and highly dissected, while emergent leaves on the same
took the seeds of several individual shrubs and grew plant may have very different shapes (Figure 5.5). In water,
them all under optimal conditions in a garden, you CO2 diffusion is much slower than in air, and the highly
would find a positive correlation between parental size dissected underwater leaves have increased surface areas
and offspring size. In a natural population, however, and a smaller boundary layer (see Chapter 3), allowing
better CO2 uptake rates. When
Emergent
environmental changes occur over
Ranunculus much longer time spans, however,
trichophyllus Erigeron
heteromorphus phenotypic plasticity is much less
likely to be favored. Instead, spe-
Submerged cialization is often favored, giving
Emergent
rise to patterns such as the con-
trasting leaf shapes among eco-
types of Geranium and Achillea dis-
cussed in Chapter 6.
Emergent
Levels of Selection
1 cm Natural selection usually occurs
at the level of the individual.
Submerged Individuals have different fit-
nesses, and genetic responses to
natural selection are measured by
Submerged looking at differences among the
individuals in a population from
Submerged generation to generation. How-
ever, none of the three compo-
Emergent nents of natural selection requires
that individuals be the only focus
of the process. Plants, in particu-
lar, have growth forms that can
result in other units being the
focus of natural selection.
Nuphar 1 cm
lutea The first question we must
address is, what is an individual?
fpo Sagittaria
In many plant species, an individ-
ual is obvious. There is a central
sagittifolia trunk or stem that grows, repro-
duces, and dies. But many other
Emergent plants are clonal, existing as a set
of genetically identical, possibly
Emergent interconnected, but semi-autono-
mous to fully autonomous units.
Submerged
Many grasses are good examples.
From a single grass seed grows a
Submerged plant that first sends up leaves and
puts down roots. Later, it produces
specialized roots called rhizomes,
which cause the plant either to
grow larger or to spread over a
1 cm Megalodonta
beckii wider area. New leaves and roots
Ondinea are produced at nodes along the
purpurea
1 cm
rhizome called tillers. Water and
nutrients pass along the rhizomes
to different parts of the plant.
However, once the new units are
Figure 5.5 Phenotypic plasticity results in differences in form between emergent and large enough, the connections
submerged leaves of several aquatic plant species which may be adaptive: Sagittaria
sagittifolia (Alismataceae), Erigeron heteromorphus (Asteraceae), Ranunculus trichophyllus between them can be cut and
(Ranunculaceae), Megalodonta beckii (Asteraceae), Nuphar lutea (Nymphaeaceae), and each unit can survive as an inde-
Ondinea purpurea (Nymphaeaceae). pendent plant. Even if the connec-
96 Chapter 5
Genet 6. When trying to determine whether a par-
ticular feature of a plant is a result of natu-
ral selection, we should consider the possi-
bility that the feature is due to selection at
a level other than the individual. Because
evolution occurs over very long stretches of
time, even very slow processes, or ones that
occur very rarely, can be important.
Rhizome
Tiller
Ramet Roots Other Evolutionary Processes
Figure 5.6 Diagram of a grass plant that grows by vegeta- One way of classifying processes that con-
tive spread. The entire plant is a single genet; each tiller and tribute to evolution is to group them into those that add
its roots comprise a single ramet. genetic variation to a population and those that elimi-
nate that variation. Natural selection acts to eliminate
genetic variation as alleles that lead to increased fitness
tions remain, the transfer of water and nutrients among become fixed in a population and other alleles are lost.
different units may be small. An important constraint on evolution is the lack of suf-
Deciding exactly what constitutes an individual, ficient genetic variation for natural selection to act upon.
then, is not simple. The British ecologist John Harper For evolution by natural selection to continue, genetic
proposed the following distinction (Figure 5.6): A genet variation must reenter a population.
is a genetic individual, the product of a single seed. A
ramet is a potentially physiologically independent unit Processes that Increase Variation
of a genet, such as the tillers of a The ultimate source of all genetic variation is mutation:
grass plant. A single genet may change in DNA sequences. Mutation includes not only
consist of many separately func-
tioning ramets. While individual
ramets may come and go, a genet
can exist for a long time. Quaking
aspen (Populus tremuloides, Sali-
caceae) is a tree that can spread by
sending up new trunks from its
roots. While individual trunks
usually live only 50 or 60 years,
John Harper one quaking aspen genet was
estimated to be more than 10,000
years old (Mitton and Grant 1996; Figure 5.7). Other
common examples of plants that grow clonally, or veg-
etatively, are kudzu (Pueraria, Oxalidaceae), strawber-
ries (Fragaria, Rosaceae), and cattails (Typha, Typhaceae).
Natural selection can occur whenever its three com-
ponents—phenotypic variation, fitness differences, and
heritability—are present. While most selection occurs at
the level of the individual, selection at other levels has
been found (Lewontin 1970). Selection can occur below
the level of the individual. A somatic mutation—a muta-
tion in a non-gamete-producing cell—could occur in the
meristems of a grass tiller (see Figure 8.1), resulting in
genetic variation among the ramets within a genet. If
those genetic differences led to phenotypic differences
that affected fitness, then selection could act on the ram-
ets. Selection can also occur above the level of the indi- Figure 5.7 This quaking aspen genet, growing in the Rocky
vidual, among populations or among species. We Mountains of Colorado, is estimated to be more than 10,000
describe an example of this type of selection in Chapter years old. (Photograph courtesy of J. Mitton.)
30
Red maple
ing-over, meiosis, and recombination, is an important
Paper birch source of new genetic variation.
25 White spruce
White pine
Processes that Decrease Variation
Seed density (% of source)
Red pine Several processes can act to eliminate genetic variation

20
from a population. Probably the most important of these,
after natural selection, is genetic drift. Genetic drift
15 refers to changes in gene frequencies due to random
sampling effects (Figure 5.9). We can illustrate genetic
10 drift with a simple example: Imagine a population that
consists of yellow-flowered and white-flowered plants.
5 In this population, flower color has a genetic basis. In
one generation, by chance alone, the more yellow-flow-
ered plants happen to be growing in richer soil. Because
0 200 400 600 800 1000
Distance from source (m)
Figure 5.8 Density of seeds of five tree species found at (A)

varying distances from their source. (After Greene and John- 1.0
son 1995.) 2N = 18
0.8
changes in single base pairs (e.g., A–T → G–C), but also Allele frequency (p)
deletions, insertions, rearrangements of parts of chro-
mosomes, and duplications of part or all of a chromo- 0.6
some. Some mutations, such as third-position codon sub-
stitutions, may have no effect on the phenotype. Other 0.4
mutations, such as deletions, may destroy the function of
a gene. Duplications of whole genes may have no imme-
0.2
diate effect on the phenotype, but may allow the evolu-
tion of a new function in one of the two copies. In order
to produce a permanent change in a population, a muta- 0 4 8 12 16 20
tion must occur in gametes or gamete-producing cells. Generation
Mutations are relatively rare events. Mutations of
(B)
single base pairs occur at a frequency of 10–8 to 10–10 per
base pair per generation. For the average-sized gene, 1.0
2N = 100
mutations occur at a frequency of 10–5 to 10–7 per gene
per generation. On average, in a given population for 0.8
most species, one individual in ten has a new mutation
Allele frequency (p)
somewhere in its genome each generation.

Another source of genetic variation is migration, 0.6
which can introduce new alleles to a population. Migra-
tion in plants occurs primarily by the movement of spores, 0.4
pollen, and seeds. For seed plants, the movement of
pollen is particularly important. Pollen can be carried by
0.2
the wind for tens of kilometers. Seed movement tends to
be more restricted, although studies of wind-dispersed
seeds show that fairly long-distance dispersal is common.
0 4 8 12 16 20
A study of five Canadian tree species, for example, found
Generation
substantial densities of seeds nearly a kilometer from their
source (Figure 5.8). Pollen and seed movement is dis- Figure 5.9 Changes over 20 generations in gene frequen-
cussed in more detail in Chapter 8. cies due to random sampling. (A) An allele frequency of p =
Because gene expression can depend on the pres- 0.5 implies that there are 9 copies of the A allele and 9 copies
of the a allele. (B) An allele frequency of 0.5 implies 50 copies
ence of alleles at other loci, new combinations of alleles of each allele. The much larger population size in (B) results
at different loci can create new variation. For this reason, in smaller oscillations in allele frequencies. As population
the process of sexual reproduction, which includes cross- size decreases, the effects of genetic drift increase.
98 Chapter 5
of this, the yellow-flowered plants, on average, produce 4
slightly more seeds than the white-flowered plants. In 3
the next generation, the frequencies of the two flower 2
colors in the population will have changed, but not
Similarity
1
because of natural selection. Rather, the change in trait 0
frequency is due to random seed establishment.
–1
Genetic drift can occur at many life history stages, Ipomoae
–2 purpurea
including meiosis, pollination, seed germination, and
maturity. In most populations, changes due to genetic –3
drift will be small. Generally, genetic drift is important –4
0 5 10 15
only in populations smaller than 100 individuals; in pop- Distance class
ulations greater than 1000 individuals, its effects are usu-
ally much smaller than those of other evolutionary Figure 5.10 Spatial autocorrelation in clusters of Ipomoea
processes. Thus, population size is a critical parameter purpurea (morning glory, Convolvulaceae) across a field near
Athens, Georgia. Similarity is measured as the likelihood
in determining how evolution will proceed. that a pink-flowered plant will be found next to a blue-flow-
Ipomoea purpurea (morning glory or bindweed, Con- ered plant. At close distances, the similarity is negative, indi-
volvulaceae) provides an example of the effects of pop- cating that pink-flowered plants tend to be found next to
ulation size on genetic drift. This species is an annual other pink-flowered plants, and the same for blue-flowered
weed of agricultural fields in the eastern United States. plants. In contrast, at longer distances, clumps dominated by
blue-flowered plants alternate with those dominated by
Its flowers vary in color; some individuals have blue pink-flowered plants. The point at which the line crosses
flowers and some have pink flowers. Flower color is a zero indicates the size of the average cluster, approximately
simple one-locus trait, and blue is dominant to pink. The 50 meters in this field. (After Epperson and Clegg 1986.)
chief pollinators of this plant are bees, and gene flow
among populations is primarily by pollen. Seeds are pas-
sively dispersed and stay close to the parent plant,
although long-distance seed dispersal probably occurs cluster contained 10 to 15 plants. The pollinators, bees,
in conjunction with agricultural or horticultural activi- tend to travel from one plant to the next nearest plant.
ties. Population densities vary among fields, and dis- So alleles in these populations tend to move very short
tances between plants range from 1 to 6 meters. distances from one generation to the next.
A study of flower color was initiated by Michael Clegg Within fields, variation in flower color was most clear-
and his associates (Ennos and Clegg 1982; Brown and ly due to the combination of genetic drift and migration
Clegg 1984; Epperson and Clegg 1986) to determine the by pollen movement. Random sampling caused gene fre-
evolutionary processes responsible for variation within quencies to become different between clusters. Migration
and among populations. These studies looked at patterns caused nearby clusters to be similar, but was not sufficient
both within fields and among fields across the states of to keep more distant clusters from becoming different.
Georgia, South Carolina, and North Carolina. The per- Spatial autocorrelation (resemblance as a function of dis-
centage of individuals with pink flowers varied greatly tance) is one hallmark of genetic drift, although spatial
from field to field, from 5% in some fields to 55% in oth- autocorrelation can also be caused by other processes.
ers. This variation occurred over a number of different dis- Population size is of particular concern for species
tances from as small as 50 meters (the shortest distance conservation. In a widespread species, even if each pop-
between fields) to as great as 560 kilometers (the farthest ulation is small, genetic variation can still be maintained
distance between samples). Fields that were located close across the species as a whole as long as there is some
to one another were no more similar in their color morph migration between populations. If the entire species is
frequencies than distant fields. This pattern most likely reduced to a single population or a small number of pop-
means that long-distance seed dispersal occurs through ulations, then maintaining genetic variation can become
human activity. difficult. Genetic variation can be lost quickly through
In contrast, distance did matter within fields (Fig- natural selection and genetic drift. New variation
ure 5.10). Within a field, there were clusters of plants through migration will not be available because the few
with different flower colors. Some clusters had only populations will all contain the same alleles. New vari-
plants with pink flowers, some clusters had only plants ation through mutation will be rare because the rate of
with blue flowers, and yet other clusters had some plants new mutations depends on the number of individuals.
with pink and some with blue flowers. On average, each In such populations, if the environment changes such
that new alleles would increase fitness, those alleles may sible evolutionary outcome is that all of the populations
not be present, and the population may be unable to will evolve to the average condition. A third outcome
adapt to the new conditions. Such concerns are even may result if the populations differ in how well they
greater today because of human-caused changes to the each adapt to local conditions. The most successful pop-
environment (see Chapter 22). Global warming may ulation may send out many more migrants than the
cause large changes in local environments in just a few other populations. Eventually, all of the populations will
decades. Small populations, unable to adapt, will be resemble that successful one.
faced with extinction.
Summary
Variation among Populations
Natural selection is the central evolutionary process deter-
So far, we have discussed evolutionary processes that mining the forms of living organisms. It comes about
affect variation within a single population. We can also through the interaction of three components: phenotyp-
consider how evolutionary processes affect variation ic variation in traits, fitness differences linked to those trait
among populations. Natural selection can both increase differences, and heritable variation in those traits. All three
and decrease variation among populations. If natural components are necessary for evolution by natural selec-
selection favors different trait values in different location to occur. The ecology of a plant plays a role in all three
tions, different populations of a species may become components, influencing the phenotype by affecting the
adapted to different conditions. Eventually, the species course of growth and development, determining fitness
may split into two species (see Chapter 6). Conversely, if differences, and influencing the expression of genes.
natural selection favors the same trait values in different Natural selection acts on variation among pheno-
locations, the populations will tend to remain similar. types. Almost all plant traits vary among individuals
Mutation and genetic drift both tend to increase within populations and species. Heritability of variation
variation among populations. It is very unlikely that means that at least some of the difference among indi-
exactly the same mutations will occur in different pop- viduals in a population is due to genetic differences.
ulations. Because genetic drift is random, it is unlikely Heritability is measured by studying resemblances
that different populations will experience exactly the among relatives. Heritability values depend on gene fre-
same random changes. These tendencies for populations quencies within populations and genotype-environment
to diverge will be greatest in small populations. interactions, and they can be affected by the nonrandom
Migration tends to decrease genetic variation among distribution of genotypes. Thus, heritability estimates
populations. Migration brings alleles into a population are specific to the population and environment in which
from nearby populations. These alleles could represent they are measured.
genetic variation previously lost from the population, or Other processes—mutation, migration, and genet-
they could represent a mutation that appeared in anoth- ic drift—are also important in evolution. The first two
er population. If migration rates between two popula- processes are important for supplying genetic variation.
tions are high enough, the populations will have the Genetic drift can act in two ways, decreasing genetic
same alleles at almost the same frequencies. variation in a population and causing evolution in the
All of these processes—natural selection, mutation, absence of natural selection. Not all evolutionary change
genetic drift, and migration—can act together to deter- is due to natural selection, although natural selection is
mine the evolution of a set of populations. Imagine a set the most important evolutionary process and is key to
of populations that initially are genetically the same. If creating adaptations.
the populations are relatively small, they may start to Natural selection is ubiquitous. Substantial amounts
diverge from one another by genetic drift and mutation. of phenotypic selection have been found in nearly every
One result is that new alleles, and new combinations of instance that it was looked for. Additive genetic varia-
alleles, will appear in each population. These changes tion has been found for most traits. Thus, the necessary
can result in several possible evolutionary outcomes. components for natural selection often exist. These
First, natural selection may favor different combinations observations are the basis for our conclusion that natu-
of alleles in each population, resulting in the populations ral selection has shaped the characteristics and nature of
becoming different from one another. Sufficiently high the plants that we see around us, even though that
rates of migration among populations, however, can process occurred long in the past, or is occurring now at
sometimes prevent a population from evolving adap- rates too slow to measure directly. In the next chapter
tations to the local environment. Therefore, another pos- we examine how natural selection has operated.
100 Chapter 5
Additional Readings
Classic References Winn, A. A. and T. E. Miller. 1995. Effect of density on magnitude of direc-
tional selection on seed mass and emergence time in Plantago wrightiana
Darwin, C. 1859. On the Origin of Species by Means of Natural Selection.
Dcne. (Plantaginaceae). Oecologia 103: 365–370.
John Murray, London.
Dobzhansky, Th. 1937. Genetics and the Origin of Species. Columbia Uni-
versity Press, New York.
Futuyma, D. J. 1997. Evolutionary Biology. 3rd ed. Sinauer Associates, Sun-
Stebbins, G. L. 1950. Variation and Evolution in Plants. Columbia Uni- derland, MA.
versity Press, New York.
Hartl, D. L. and A. G. Clark. 1989. Principles of Population Genetics. 2nd
Contemporary Research ed. Sinauer Associates, Sunderland, MA.
Lande, R. and S. J. Arnold. 1983. The measurement of selection on cor- Provine, W. B. 1986. Sewall Wright and Evolutionary Biology. University
related characters. Evolution 37: 1210–1226. of Chicago Press, Chicago, IL.
Mitchell, R. J. 1994. Effects of floral traits, pollinator visitation, and plant Wilson, E. O. and W. H. Bossert. 1971. A Primer of Population Biology. Sin-
size on Ipomopsis aggregata fruit production. Am. Nat. 143: 870–889. auer Associates, Sunderland, MA.
C H A P T E R
6 Outcomes of Evolution
H
ow do we know that a particular plant trait was shaped by natural
selection? Answering this question is not simple. There is no single
method for doing so, and all methods require assembling multiple
kinds of information. Historically, ecologists have sometimes been guilty of
not documenting natural selection. Rather, they have simply assumed that
organisms were optimally adapted to the environments in which they were
found. This attitude was termed “Panglossian” by Stephen Jay Gould and
Richard Lewontin (1979), after the character Dr. Pangloss in the novel Candide
by Voltaire. In that story, the naïve Dr. Pangloss goes through life confidently
declaring his assumption that “All is for the best in this, the best of all possi-
ble worlds.”
Sometimes ecologists have constructed scenarios to explain what process-
es may have led to particular adaptations, and then simply accepted those sce-
narios without substantiation. For example, some plant species produce seeds
with a sticky, mucilaginous coating. Some ecologists originally claimed that
this coating was an adaptation for dispersal; seeds getting stuck on the feet of
ducks was one particular scenario they proposed. Such scenarios are termed
“just-so stories,” after the title of the book in which Rudyard Kipling recounts
fanciful tales about the origins of various animals. Today we understand that
in most species the coating has more to do with water retention by the seed.
The problem with “just-so” scenarios constructed by scientists is not that
they are necessarily wrong; it is that they are based on scant evidence or on
unfounded extrapolation from what is known. For example, while we cannot
justifiably make the leap to sticky seeds having evolved by the process of nat-
ural selection acting to increase the dispersal success of all seeds, it has been
documented with substantial evidence based on careful study that the sticky
seeds of some species, such as Phoradendron californicum, (desert mistletoe, San-
talaceae), are in fact spread by birds.
Thus we must be cautious in either accepting or dismissing such adaptive
speculations. Speculations are an indispensable first step in posing hypothe-
ses to test whether a character has come about by natural selection. As ecolo-
gists are usually knowledgeable about the organisms and systems they study,
such speculations often prove eventually to be correct. But until evidence has
been assembled that firmly supports these suppositions, they remain just spec-
ulations.
102 Chapter 6
In the last chapter we described how natural selec- Antonovics and Bradshaw 1970) began to study local
tion works. In this chapter we explore examples of the adaptation of grasses to differ-
results of natural selection and the process of speciation. ences in soil conditions due to
The examples in this chapter were chosen because they mine waste contamination. In
are based on more than just speculation. Most of them Great Britain, zinc and copper
are based on multiple lines of evidence. As we examine have been mined for centuries.
each, we discuss the reasoning behind the conclusions The soil left over from the dig-
concerning the shaping of the trait by natural selection. gings and ore extraction, known
This brings us back to the question we posed in the as tailings, was simply dumped
first sentence of this chapter: Why is it so difficult to outside the mines. Although most
decide whether a trait was shaped by natural selection? of the ore had been removed,
While natural selection is a primary process in shaping these tailings still contained high
plant form and function, other processes may also be concentrations of the metals, con- A. D. Bradshaw
responsible for shaping particular traits. In some centrations too low to be worth
instances these processes act in concert with natural extracting but still high enough to be toxic to most plants
selection. Mammals have backbones not only because growing on the soil. Some plants, however, managed
a backbone is a handy thing to have, but because they to grow on these tailings—in particular, the grasses
are vertebrates, and they have inherited this trait from Anthoxanthum odoratum and Agrostis tenuis. Although
vertebrate ancestors: all vertebrates share this trait for some of the mines had been abandoned for only a cen-
that reason. Likewise, angiosperm cells exhibit cellular tury—probably fewer than 40 generations for these
respiration not because it is especially advantageous to perennial species—the researchers found that popula-
angiosperms per se, but because the ancestors of tions of these grasses on the mine tailings were tolerant
angiosperms possessed this trait, which they have inher- of the heavy metals, while plants growing in ordinary
ited, and thus share with a much broader group of liv- adjacent pastures were not. The tolerance of heavy met-
ing organisms. als was achieved by the evolution of biochemical mech-
In addition, natural selection is only one of the fac- anisms that prevented the uptake of the toxic ions. These
tors that cause evolutionary change (see Chapter 5). differences in genetic composition occurred over very
Other processes may act in concert with natural selec- short distances—within a meter of the mine boundary
tion, or they may act against it, or they may act inde- (Figure 6.1).
pendently. Mutation, for example, is a necessary source The abrupt nature of the genetic boundary illus-
of the variation that natural selection requires. Migra- trates several aspects of the evolutionary process. First,
tion of individuals from other environments can prevent the alleles for heavy-metal tolerance did not spread out-
adaptation to local conditions in response to natural ward from the population on the mine tailings. Why?
selection. Genetic drift can also cause changes in gene Experiments were performed in a greenhouse in which
frequencies. Often each of these three processes can pro- plants from both the mine tailings and the adjacent pas-
duce results that look like the outcome of natural selec- ture were grown in soil with and without the heavy-
tion. Because we are reconstructing a historical event, metal contaminants. As expected, plants from the pas-
our conclusions must be based on indirect inference, ture population died when grown with the heavy
rather than direct observation. All of these factors make metals, usually as seedlings. However, when grown in
the determination of adaptation by natural selection a the absence of the heavy metals, the pasture plants grew
challenging task. much more quickly than those from the mine tailings
population. Something about the mechanism of toler-
Heavy-Metal Tolerance ance reduced growth under uncontaminated conditions.
This interaction is an example of a trade-off, a reduction
We begin with one of the best-documented cases of nat- in fitness in one feature of an organism due to an
ural selection and local adaptation in plants. This exam- increase in fitness in another feature. In this case, heavy-
ple is also important historically: it was the first demon- metal tolerance and growth in uncontaminated soils are
stration of fine-scale genetic differentiation in response traded off against each other. Thus, different alleles are
to selection by a known factor in the environment. The favored on and off the mine tailings, and the populations
example involves a cline, a gradient in genetic compo- have diverged genetically due to natural selection.
sition, resulting from genetic differentiation and adap- Other features of the populations have also evolved
tation. Clines can occur over very short distances as well as a consequence of this divergence. All grasses are wind
as over larger geographic scales. pollinated, so pollen from the adjacent pastures can eas-
In the 1960s, A. D. Bradshaw and his students (e.g., ily land on the stigmas of plants on the mine tailings, and
Jain and Bradshaw 1966; McNeilly and Antonovics 1968; vice versa. The abruptness of the genetic boundary was
Outcomes of Evolution 103
Pasture Mine tailing These studies clearly demonstrate, for several rea-
60 sons, that the differences among populations were due
Adults to natural selection. First, the investigators demonstrat-
Seeds ed the presence of all three necessary components of nat-
50 ural selection: phenotypic variation, fitness differences,
Mine and heritability (see Chapter 5). Second, they found that
boundary secondary predictions were also met. The change in
Index of copper tolerance
40 flowering time, for example, was a consequence of nat-

ural selection for heavy-metal tolerance. Third, they
found the pattern of adaptation repeated both among
30 populations of the same species at different mines and
among different species. Such repeated patterns of dif-
ferentiation would be very unlikely to have occurred by
20 chance alone. For all of these reasons, we conclude that
the differences in heavy-metal tolerance between popu-
lations on and off the mine tailings were due to evolu-
10
tion by natural selection.
Ecotypes
0
0 10 20 30 40 Plant ecologists and taxonomists have recognized since
Distance (m) before the time of Linnaeus, well over 250 years ago, that
Figure 6.1 the same kind of plant may have a different appearance
Ability of adults and plants grown from seeds of Agrostis
tenuis collected from mine tailings and from adjacent uncont-
depending on where it is growing. Early in the twenti-
aminated pasture to grow in the presence of copper. Some eth century, researchers in Europe and in the United
100 years after the mine was abandoned, the genetic compo- States began careful experimental work to better under-
sition of the grass population differs dramatically at the mine stand the causes of some of these differences. In Sweden,
boundary, where alleles for enzymes that prevent the uptake Turesson (1922) collected plants belonging to several
of toxic ions are suddenly favored. (After McNeilly 1968.)
dozen species from habitats all over Europe and grew
them in a common garden experiment at a single site in
Sweden. He found that many of the differences that he
found to be related to the prevailing wind direction, with noticed when the plants were growing in their natural
more of the “wrong” genotype being found on the down- habitats were retained when they were all grown in a
wind side. If an individual is pollinated by a plant from single environment. Turesson coined the term ecotypes
the other population, then its offspring will be less well to describe populations of a species from different habi-
adapted to the local conditions, as shown by the lower tats or locations that possess genetically based differ-
tolerance of the plants grown from seed in Figure 6.1. The ences in appearance and function. Ecologists usually use
problem of “wrong” pollinations results in selection for the term “ecotype” to refer to such differences that
mechanisms that reduce cross-pollination. In these pop- appear to be adaptive.
ulations, two mechanism were involved in reducing out- At about the same time as Turesson was doing his
crossing (Antonovics 1968; McNeilly and Antonovics experiments in Sweden, Clausen, Keck, and Hiesey
1968). First, individuals on the mine tailings tended to (1940) were doing similar work in California. These
flower earlier than those in the pasture, with the differ- researchers established three common garden sites at low,
ence being greatest for plants just at the boundary of the intermediate, and high altitudes, from the Carnegie Insti-
mine tailings. Second, plants near the boundary tended tute south of San Francisco, where they worked, up into
to self-pollinate more than plants farther away from the the high Sierra (the spectacular Sierra Nevada that John
boundary. Both changes reduced the chance of an indi- Muir, the famous conservationist, called “the range of
vidual receiving the “wrong” pollen from plants adapt- light”). Plants from a large number of species were col-
ed to the alternate conditions (toxic or uncontaminated lected along a transect—a long line along which samples
soils). These changes in mating patterns were a second- are taken—running from the coast of California inland,
ary consequence of the primary selection for heavy-metal up to the highest part of the Sierra Nevada and down the
tolerance. They acted to reinforce the genetic differentia- arid eastern side of the mountains (Figure 6.2). The plants
tion of the populations. Over a long enough period of were vegetatively propagated, and genetically identical
time, such reinforcement can lead to total genetic isola- copies of many different individual plants were grown
tion of the populations, and ultimately to speciation. in each of the three gardens. Clausen, Keck, and Hiesey
104 Chapter 6
Figure 6.2
Part of Clausen, Keck, and Hiesey’s transect across California from the foothills of the
Sierra Nevada to the Great Basin desert, showing (below) the height in altitude of the
sites from which plants were collected and (above) the appearance of Achillea plants
when grown in a common garden at Stanford near sea level. The small graphs show the
variation in height among individual plants from each site, with the specimens pictured
representing plants of average height and the arrows indicating mean plant height.
Note that horizontal map distances are compressed. (From Clausen et al. 1948.)
studied two groups of species most intensively: several from lower-altitude plants were smooth and green, and
closely related species of Achillea (yarrow, Asteraceae), were highly dissected (had a blade divided into small
and Potentilla glandulosa (cinquefoil, Rosaceae). but connected parts, making the leaf look very feathery).
Like Turesson, the researchers found that many of Many years after these original studies, Gurevitch
the differences in morphology (growth form) and phe- and colleagues (Gurevitch 1988, 1992a,b; Gurevitch and
nology (the timing of seasonal events, such as flowering Schuepp 1990) returned to the original sites to carry out
and dormancy) among plants from different sites were further studies on the same populations that Clausen,
still present when those plants were compared in the Keck, and Hiesey had worked on. As anyone who has vis-
common gardens (Figure 6.3). Achillea plants from lower ited or lived in California might expect, sadly, many of
sites were larger than plants from higher altitudes, with the original sites where plants had been collected are now
longer leaves and taller flower stalks. Plants from high- under asphalt. However, the Carnegie Institute still main-
er altitudes generally entered dormancy sooner and tains the sites at Mather and Timberline, where two of the
began growing later than plants from lower, warmer common gardens once were located, and plants from
sites. These alpine (high-mountain) plants also tended to these intermediate- and high-altitude populations were
flower later than those from lower-altitude sites. Plants collected for study. The populations differed genetically
generally had the highest survival rate and best per- in photosynthetic characteristics. There were also genet-
formance in the garden with the conditions that most ic differences between the populations in both leaf size
closely resembled those in which they grew naturally and leaf shape, particularly in the degree to which these
(Clausen et al. 1948). The researchers also noted striking complex leaves were dissected. As the early researchers
differences in the appearance of the leaves from different had noticed, the Mather population had much more feath-
sites along the transect. Leaves from the high-altitude ery, highly dissected leaves, while those from the Tim-
populations were not only smaller, but tended to have a berline plants were more compact (Figure 6.4).
dense gray pubescence (a mat of short hairs on the leaf Could there be an adaptive reason for these differ-
surface), and were much more compact in shape. Leaves ences in leaf shape? Energy budget studies demonstrat-
dry environment at the lower altitude, leaves would
remain relatively cool, while high-altitude plants might
be able to warm up above the chilly air temperatures
common in their environment to maximize photosyn-
thesis and growth.
Another study taking a similar approach was car-
ried out on the widespread European species Geranium
sanguineum (Geraniaceae; Lewis 1969, 1970). This species
grows in a variety of habitats. As Turesson himself and
other researchers had noticed for other species, plants of
this species from the most xeric (dry) sites had the most
highly dissected leaves, with the narrowest lobes, while
those from more mesic sites (with more water available)
had larger, less dissected leaves (Figure 6.5). Lewis found
genetic differences in leaf shape and size among plants
collected from a large number of different sites in
Europe. There was a cline in the degree of leaf dissec-
tion, with the most highly dissected leaves from the dri-
est environments and the more compact leaves from
moister, more oceanic sites. A different geographic cline
in leaf size correlated with the openness of the habitat,
with the largest-leaved plants from the least open, most
tree-shaded habitats. Based on physiological and ener-
gy budget studies, he hypothesized that differences in
leaf energy budgets due to the differences in leaf size and
shape would lead to differences in leaf temperatures.
The more dissected leaves from the drier habitats would
remain closer to air temperature, thus potentially avoid-
ing overheating when drought forces stomatal closure.
Other physiological advantages were also predicted for
leaves of each ecotype in its “home” environment.
Originally, it was believed that there was a strict
dichotomy between clines and ecotypes. Clines repre-
sented gradual differences in genetic characters, respon-
sible for gradual morphological and physiological differ-
ences over geographic space. Ecotypes, in contrast,
represented sharp differences among populations that are
distinct genetically and in various phenological traits.
Most plant ecologists no longer see this distinction as
being very useful. Genetic variation responsible for mor-
Figure 6.3
Appearances of 7 clones from the Mather population of phological, physiological, and phenological differences
Achillea lanulosa (collected at 1400 m. elevation) when grown may occur over many different scales, sometimes with
in the three transplant gardens. Timberline (top; 2100 m.), sharp distinctions among populations and sometimes
Mather (center), and Stanford (bottom; 30 m. elevation). with gradual change. The abruptness of the change
(From Clausen et al. 1948.) depends on the nature of the evolutionary processes act-
ing on these groups of plants: selection, genetic drift, and
migration all shape the rate of genetic change over space.
ed that the highly dissected leaves of the Mather plants We may choose to call distinctive populations “ecotypes”
had a thinner boundary layer with greater heat con- for convenience, or we may wish to emphasize the grad-
ductance (see Chapter 3), making it more likely that they ual nature of change and call the gradient a “cline.” As
would remain close to air temperature. The compact natural habitats grow ever smaller and more fragmented,
leaves of the high-altitude plants had a thicker bound- more and more cases of gradual genetic and phenotypic
ary layer and less heat conductance, and energy budg- change in plant and animal populations will become iso-
et calculations suggested that these leaves could warm lated, thus resembling ecotypes even though their origins
considerably above air temperatures. Thus, in the warm, were as members of a continuously varying cline.
106 Chapter 6
(A) Mather plant (B) Timberline plant
fpo
1 cm
Warm Cool
1 cm
Warm Cool
Figure 6.4
Leaves from the Mather (A) and Timberline (B) populations of Achillea lanulosa when
grown together in common greenhouse environments under warm and cool condi-
tions. An expanded part of the midsection of a Mather leaf is shown to show the
complex shape. The longer Mather leaf is approximately 11.5 cm. long. (After Gure-
Adaptive Plasticity
vitch 1992.) For the most part, plants are not
mobile. This observation, while
obvious, has profound implications for the evolution
and adaptation of plants. If the environment becomes
unfavorable in one spot, a mobile animal can move to a
5 cm more favorable location. But what can a plant do?
Because plants cannot move, they must be able to tol-
erate environmental variation. As discussed in Chapter
5, one solution to this problem is for individuals to be
phenotypically plastic, able to change their form so as
to match the most fit trait value for the environment they
find themselves in (Bradshaw 1965). For example, plants
produce leaves with different characteristics when grown
in the sun than when grown in the shade (see Chapter 2).
Plants in general are far more plastic than animals; for
example, while well-fed animals are taller and heavier
than those with poorer diets, plants grown under opti-
mal conditions can be orders of magnitude larger than
genetically identical plants grown in poor conditions.
Dry, open Moist, shaded
But how do we demonstrate that plasticity itself is
adaptive, or investigate the condi-
Habitat tions under which it is most advan-
Figure 6.5 tageous? Many characteristics of
Differences in leaf shape in Geranium sanguineum collected from the field from a vari- plants vary with the environment.
ety of habitats from northern to central and eastern Europe, then grown in a common While some of this variation is
garden. The sites ranged from dry, open sites (alvar, xeric limestone, left) to interme-
diate sites (steppe, woodsteppe–open woodlands, and coastal clifftops) to moist,
adaptive, some changes in plant
shaded sites (woodland, right). (After Lewis 1972.) appearance and function are mere-
ly unavoidable consequences of
tually becoming shaded by its neighbors? If so, a plant
finding itself in this situation will benefit by elongating
its stem as it grows. If it can detect the fact that it is grow-
ing in an uncrowded environment, it may be better off
not elongating its stem very much. The chlorophyll in
the leaves of neighboring plants absorbs heavily in the
red part of the spectrum. Therefore, a plant that receives
light with a low red:far-red ratio can use that informa-
tion as a signal that it is surrounded by other plants.
Is the elongation response truly adaptive? Schmitt
addressed this question by manipulating plant form and
growing conditions. Seedlings were initially grown in
a greenhouse under two conditions: (1) with filters that
blocked the red part of the light spectrum and (2) with
full-spectrum controls that reduced light by the same
amount without changing the red:far-red ratio. These
treatments created two sets of plants, elongated and sup-
pressed. The experimental plants were then transplant-
ed into a forest at either high or low density. Fitness was
measured as the total number of seed capsules present
at the end of the summer. As predicted by the adaptive
Figure 6.6
Impatiens capensis (Balsaminaceae) grown in the sun (high
plasticity hypothesis, the elongated form had higher fit-
red:far-red ratio) and in the shade (low red:far-red ratio). ness under crowded conditions, while the short, stout
Short, stout plants are favored in low-density conditions, form had higher fitness at low densities (Figure 6.7).
while tall, thin plants are favored in crowded conditions. The pattern of plasticity also differed in an adaptive
(Photograph courtesy of J. Schmitt.) way among populations growing in different environ-
ments. The researchers compared three populations
plant physiology, such as having yellow leaves when

deprived of sufficient nitrogen. Demonstrating that phe- 30
notypic plasticity is adaptive requires, first, that there be
more than one possible response, and second, that there
Fitness (number of seed capsules present)
25
is a consistent relationship between plasticity and fit-
ness. The following study demonstrates how one can
amass evidence on these points. 20
Johanna Schmitt and her colleagues studied popu- Suppressed
lations of Impatiens capensis (jewelweed, Balsaminaceae)
growing in different light conditions (Schmitt 1993; Dud- 15
Elongated
ley and Schmitt 1995, 1996). This species is an annual
that grows in deciduous forests of the northeastern and 10
north-central United States. Jewelweeds grow in con-
ditions from full sun to forest understories, often along
streams, and can reach very high densities. 5
A typical response to crowding in plants is stem
elongation. This strategy allows a plant to overtop its
0
close neighbors, thus gaining the most light (Figure 6.6). High Low
In uncrowded situations, however, plants remain short, Density
because elongation has costs as well as advantages. More Figure 6.7
resources must be put into support structures rather than A test of adaptive plasticity of stem elongation in Impatiens
flowers and seeds, and the elongated stem is thinner and capensis. Elongated plants were initially grown in a green-
in greater danger of falling over. The elongation response house with filters that blocked the red part of the light spec-
is controlled by the ratio of red to far-red light that a trum; suppressed plants were grown with full-spectrum
controls that reduced light by the same amount without
plant receives (see Chapter 5). changing the red:far-red ratio. The two sets of plants were
Is it possible that a plant can tell whether or not it then grown at high or low density. The bars indicate 1 stan-
is growing near other plants, and thus at risk of even- dard error. (After Dudley and Schmitt 1996.)
108 Chapter 6
growing within 1 kilometer of one another, one in a variation in a species can result from some combination
woodland clearing and the other two in the woods. They of plastic and nonplastic genotypes traces to an influ-
grew plants from all three populations together in a ential paper by Bradshaw (1965). However, only in the
greenhouse. The clearing population had a stronger past 20 years has this distinction begun to be more wide-
elongation response than the other two populations. ly recognized (see reviews by Schlichting 1986; Sultan
This result is predicted on adaptive grounds: In the 1987; West-Eberhard 1989; Scheiner 1993; Via et al. 1995;
woods, a plant that grows taller than its neighbors will Pigliucci 2001).
still be in the shade. On the other hand, in a sunny site, We also speak of two versions of the niche, the fun-
a plant that overtops its neighbors will be in bright light. damental niche and the realized niche (Hutchinson
Thus, being more plastic is adaptive for the population 1957). The fundamental niche is the range of conditions
in the clearing. As in our first example, here again the that a species is physiologically capable of growing in.
researchers showed the existence of the three necessary The realized niche is the range it is actually found in.
components of natural selection: plants varied in how Factors such as competition, herbivory, and lack of pol-
plastic they were, differences in plasticity led to fitness linators or seed dispersers all act to reduce a species’
differences, and plasticity was shown to be heritable. realized niche from the potential of its fundamental
Finally, they showed that differences in plasticity among niche; these factors will be covered in more detail in
populations were in the direction expected if plasticity Chapters 8 through 11, and Leibold (1995) offers a com-
were evolving by natural selection. prehensive review of niche concepts.
The term “plasticity” is used in two different con-
texts by ecologists. We have already discussed one con-
Selection in Structured Populations
text, the plasticity of an individual. The other context is
the plasticity of a species, meaning the range of ecolog- The fitness of an individual depends on its environment.
ical conditions that a species will grow in. This range is One component of that environment is population struc-
also called the species’ niche (Grinnell 1917). The niche ture: the phenotypes of the other individuals in the pop-
of a species is determined by the sum of the niches of its ulation and their frequencies. In Chapter 7 we describe
members (Figure 6.8). Those individuals may be plas- two types of population structure, age structure and
tic and may have wide or narrow niches. They may vary stage structure, but there are also other sources of phe-
genetically in their optima. If they vary sufficiently, they notypic variation among the individuals in a population,
may be referred to as ecotypes. So, a species may consist such as the colors or shapes of their flowers. When the
of a combination of generalist and specialist genotypes, fitness of an individual depends on the number and
while at the same time we speak of generalist and spe- kinds of other phenotypes in the population, selection
cialist species. In plant ecology, the idea that the total can take various forms. Two of these forms are frequen-
cy-dependent selection and group selection; here we
explore examples of both types.
Species niche
100 Frequency-Dependent Selection
Frequency-dependent selection occurs when the fitness
Survival probability (%)
of a genotype depends on whether it is common or rare

in a population. An example is seen in the evolution of
mating systems in Lythrum salicaria (purple loosestrife,
Lythraceae; see Figure 14.3). This species is an invasive
weed in wetlands of northern North America. It is still
in the process of invading the continent, which makes it
an ideal species for studying evolution in action. In the
study described here, scientists were able to directly
0 observe populations evolving over a five-year period.
Low High Lythrum salicaria has an unusual mating system,
Temperature tolerance called tristyly. The style is the female reproductive struc-
Figure 6.8 ture that supports the stigma, upon which pollen lands
The niche of a species is the sum of the niches of its mem- to fertilize the ovules (see Figure 8.6). This mating sys-
bers. Here we consider one aspect of a niche, temperature tem gets its name because individual plants in a popu-
tolerance. In this graph, each solid curve is the survival lation have one of three different forms for the style:
probability of an individual as a function of temperature.
The temperature range of the curve is one measure of plas- short, medium, or long (Figure 6.9A). The form of the
ticity of that individual. The dashed curve is, then, the niche style is genetically determined. The style lengths are
of the species. matched by three stamen lengths (the stamen supports
(A) (B)
Style
FPO 0.9
Stamens/
anthers
0.8
Morph 1
Evenness value in year 5

(long style)
0.7
0.6
Morph 2
(medium style)
0.5
0.4 0.5 0.6 0.7 0.8 0.9

Evenness value in year 0
Morph 3
Lythrum salicaria (short style)
(purple loosestrife)
other 99 plants through its pollen). In contrast, each of
Figure 6.9 the long-styled individuals would have offspring
(A) Tristyly in purple loosestrife (Lythrum salicaria, through at most 2 plants (themselves and the short-
Lythraceae). The plant has three different forms, or morphs, styled plant). As a result, the genes of the short-styled
of stamen and style, a system that promotes outcrossing. An plant would increase in frequency. In general, the rarest
individual with a given style length can mate only with the
other two stamen types. (B) Change in style length evenness mating type (the one with the least common style length)
over a five-year period in 24 populations of L. salicaria. An will always enjoy this sort of mating advantage. The
evenness value of 1 would indicate that the three lengths population will therefore continue to evolve until all
were equally frequent. In all but 4 of the populations, the fre- three types are at equal frequency, with one-third of the
quencies were more even in year 5 than in year 1, indicating population having each of the style types.
that the rare morph had increased in frequency. (B after Eck-
ert et al. 1996.) Just such a pattern of evolutionary change was
observed by Christopher Eckert and colleagues (1996),
who surveyed 24 populations of L. salicaria across the
the anthers, which hold the pollen). A single individual Canadian province of Ontario. These populations were
has a style of one length and stamens of the other two newly established as part of the invasion of L. salicaria
lengths. This system is a way of enforcing outcrossing. into North America. During the creation of each popu-
An individual with one style length cannot mate with lation, some style length types had a high frequency
another individual with the same style length; it can while others had a low frequency due to chance alone.
mate only with the other two types. (See Chapter 8 for Sometimes it was the short type that was rare, some-
more details on mating systems and tristyly.) times it was the medium type, and sometimes it was the
Frequency-dependent selection comes about long type. The researchers predicted that over time the
because of this inability to mate with other individuals populations should evolve to an equal frequency of the
of the same type. Imagine a population of 99 long-styled three types. They tested this prediction by returning to
individuals and 1 individual with short styles. That the populations 5 years later. The prediction was met: In
short-styled individual would be able to mate with all all cases, the rarest style length had increased in fre-
the other plants, as it could fertilize all the other plants quency (Figure 6.9B). Most of the populations were still
in the population with its pollen. The long-styled plants not yet at equilibrium, with one-third of each type; that
would be limited to pollinating just the one short-styled will require several more decades. However, the popu-
individual. So in the next generation, the genes of the lations had consistently evolved in the predicted direc-
short-styled individual would end up in the offspring of tion, something that would be highly unlikely to occur
100 plants (itself through its female gametes and the by chance.
110 Chapter 6
This example is an instance in which natural selec- (A)
tion is strong enough that we can observe its effects in 25
Number of cleistogamous flowers

action. It is unusual to be able to measure a consistent
change in a natural population in just 5 years. Part of the 20
strength of this study is that this change was observed
across 24 different populations. Thus, it is highly unlike- 15
ly that the change is due to anything other than natural
selection. So, we can be confident that what we are
10
observing is evolution by natural selection.
Group Selection 5
Group selection occurs when variation in group char-

acteristics causes differences in fitness among groups. 0
–3 –2 –1 0 1 2
We briefly described the basis of group selection in Standardized size
Chapter 5 in the discussion of levels of selection. Group
selection is one term used for selection above the level (B)
of the individual. A common form of group selection is
11
kin selection, which occurs when the groups in ques-
Number of chasmogamous flowers

10
tion consist of closely related individuals.
9
The concept of group selection is somewhat contro-
8
versial among evolutionary biologists. Kin selection is
7
less controversial because the individuals in a group are 6
closely related, and thus share large numbers of genes. 5
No one disputes that group characteristics can affect 4
individuals’ fitness. However, group selection also 3
implies that group-level characteristics can be subject to 2
natural selection. Whether selection on group-level traits 1
has occurred outside of kin groups is a matter of dispute. 0
An example of group-dependent fitness effects was –3 –2 –1 0 1 2
found in populations of Impatiens capensis (Stevens et al. Standardized size
1995). These plants often grow in dense patches, or
Figure 6.10
groups of plants, as we saw in Schmitt’s work on adap- Individual- and group-level selection on size in Impatiens
tive plasticity. Impatiens can produce both closed, self- capensis. Fitness was measured as the number of cleistoga-
pollinated flowers (cleistogamous flowers) and open, mous flowers (A) and chasmogamous flowers (B) produced.
potentially cross-pollinated flowers (chasmogamous The black lines show the relationship between an individ-
flowers) on the same plant. In this study, plant size and ual’s flower production and the individual’s size—a result of
individual-level selection. The green lines show the relation-
fitness were measured; fitness was measured as the ship between an individual’s flower production and the
number of either cleistogamous or chasmogamous flow- average size of the other individuals in its patch—a result of
ers produced on each plant by the end of the growing group-level selection. (After Stevens et al. 1995.)
season. The researchers then compared the flower pro-
duction among groups (patches). Some patches had
mostly small individuals, and thus a small mean plant
size. Other patches had a few large individuals sur- acteristics of the group determined the number of flow-
rounded by small neighbors; in these patches the aver- ers, and thus the fitness of individuals.
age plant size was large, due to the very large size of the How could this occur? Individuals with small
dominant individuals. Both individual-level and group- neighbors had greater numbers of cleistogamous flow-
level selection was found (Figure 6.10). Individual-level ers than those with large neighbors. In patches where
selection occurred within each patch, because larger the mean size of individuals was large, only the largest
individuals within a patch had greater numbers of both individuals were able to produce cleistogamous flow-
kinds of flowers. Group-level selection occurred among ers. These very large individuals depressed the flower
patches, because individuals in patches with a small production of their smaller neighbors by shading them
mean plant size produced, on average, greater numbers and reducing their access to resources. Yet the large indi-
of closed flowers than did individuals in patches with viduals did not completely use up the resources in the
a large mean plant size. (There was no group-level selec- patch; some resources went unused. In contrast, in
tion on the number of open flowers.) That is, the char- patches where the mean size of individuals was small,
more individuals had access to the resources, and in dry habitats may have nothing to do with their pos-
because more of the total resources were used, all indi- session of CAM photosynthesis; this association could
viduals had greater flower production. This effect of just be a coincidence.
asymmetries in size leading to suppression of growth How do we separate such chance associations from
and flowering is discussed in detail in Chapter 10. those due to adaptation by natural selection? As before,
The researchers’ conclusions about the existence of we look for repeated associations. In this instance we
group selection and the mechanisms responsible were look for repetition across different, unrelated groups of
based on a combination of studies of natural populations species. One type of such repetition is convergent evo-
and manipulative experiments involving clipping of lution, the independent evolution of similar features in
plants (Stevens et al. 1995; Kelly 1996). The clipping exper- unrelated taxa. Because plants in many unrelated fami-
iments were especially important for providing informa- lies found in arid environments or microhabitats have
tion on the causative agent of natural selection. Manip- CAM photosynthesis, and because CAM photosynthe-
ulative experiments are an important complement to sis evolved independently in many of them, we begin to
studies of selection in natural populations (see Chapter 1). become convinced that CAM photosynthesis offers
adaptive advantages in these environments. Because we
know the physiological mechanisms involved, we
Convergent Evolution
understand that CAM photosynthesis results in restrict-
One way to determine whether the form of a trait is due ed water use and greater water use efficiency in plants
to adaptation by natural selection is to ask the question, that have this photosynthetic pathway. These multiple
“Do we find the same form repeated in similar envi- lines of evidence make it more believable that CAM pho-
ronments?” Heavy-metal tolerance, for example, has tosynthesis is an evolutionary adaptation.
been found repeatedly in populations from different Another example of convergent evolution is found
mines. But how do we decide about adaptations if an in the morphology of some of the cacti of the Sonoran
entire species, or group of species, shows one form? In Desert of North America and species in the family
the Cactaceae, all species have CAM photosynthesis (see Euphorbiaceae growing in the deserts of southern
Chapter 2). These species also all grow in dry habitats Africa. Consider Lophocereus schottii and Euphorbia stap-
or microhabitats. Is CAM photosynthesis an adaptation fii; these two plants look remarkably alike (Figure 6.11).
to low water availability? One alternative explanation is Both are leafless, have green photosynthetic stems, are
that plants in the Cactaceae all have CAM photosyn- columnar in shape, have spines along the ribs of the
thesis not because it is adaptive, but only because the stem, and have CAM photosynthesis. These species are
ancestors of the Cactaceae had CAM photosynthesis. very distantly related, so their similar features must be
The adaptation of species in the Cactaceae to conditions due to independent evolutionary events. Other mem-
(A) (B)
Figure 6.11
An example of convergent evolution:
two similar-looking desert-dwelling
species. (A) Lophocereus schottii (Cac-
taceae), found in the Sonoran Desert of
North America. (Photograph by Doug
Sokell/Visuals Unlimited.) (B) Euphor-
bia spiralis (Euphorbiaceae), found in
Africa. (Photograph by J. N. Trager/
Visuals Unlimited.)
112 Chapter 6
bers of the genus Euphorbia, far more closely related to parts of the plants, at 2 meters above the ground, stay
Euphorbia stapfii, look very different; the genus includes above freezing due to longwave radiant energy input
plants with a wide variety of growth forms, ranging from the ground (see Chapters 3 and 18). Thus, the sen-
from small herbaceous species to shrubs to large plants sitive flowers and meristems never experience freezing
(the widely cultivated poinsettia is a member of this temperatures. Both of these species evolved from herba-
genus; E. pulcherrima). Another way that we know that ceous, ground-dwelling species growing at lower eleva-
the striking similarities between L. schottii and E. stapfii tions. Again, the similarity of traits in species that are both
are due to convergent evolution is that some of the fea- distantly related and geographically separated, but grow
tures have different developmental origins. The spines in similar environments, is strong evidence that these fea-
on cacti evolved from leaves, which can still be seen on tures are adaptations shaped by natural selection.
primitive cacti. In contrast, the spines on the Euphorbia
evolved from stipules (appendages to petioles). The Speciation
independent evolution of such a large number of dif-
ferent features by chance alone is very unlikely. There- One outcome of evolution and natural selection is spe-
fore, we can be confident in our conclusion that these ciation, the production of new species. Speciation may
features are adaptations to conditions in desert envi- occur when two or more populations of the same species
ronments. become isolated from each other and adapt to different
Another striking example of convergent evolution environmental conditions over a long time. Eventually,
is found among alpine plants in the families Campanu- the populations’ responses to differences in natural selec-
laceae and Asteraceae. Lobelia telekii grows on Mount tion in their disparate environments can result in pop-
Kenya in Africa and Espeletia pycnophylla grows in the ulations that have become so different that they are
Andes of South America. They are both unusual-look- reproductively isolated, meaning they can no longer
ing plants, standing up to 2 meters tall, with a thick stem interbreed. Most biologists consider reproductive iso-
topped by a rosette of leaves (Figure 6.12). Their tall size lation to signify that two populations have become dif-
is an adaptation to large day/night temperature swings. ferent species—which can also occur through process-
At night the temperature near the ground can get below es other than adaptation. Usually, such differentiation
freezing because the ground is losing large amounts of happens in populations that are geographically distant
radiant energy to the very cold night sky. But the upper from each other (allopatric speciation) (Figure 6.13), but
(A) (B)
Figure 6.12
An example of convergent evolution: two similar-looking alpine tundra species.
(A) Lobelia telekii (Campanulaceae), found on Mount Kenya in Africa. (Photograph by
Elizabeth Orians.) (B) Espeletia pycnophylla (Asteraceae), found in the Andes of South
America. (Photograph by P. Jorgensen.)
term to refer to crosses between different strains of a sin-
gle species.) The extent of hybridization varies greatly
among genera. In most genera, hybridization is a very
G. tenuiflora rare event, if it occurs at all. On the other hand, some
genera are notorious for hybridizing frequently, includ-
ing Quercus (oaks, Fagaceae), Crategus (hawthorns,
G. austrooccidentalis
Rosaceae), and Atriplex (saltbush, Amaranthaceae). In
some cases species identities become very difficult to dis-
G. jacens
cern as the boundaries between species become indis-
tinct, and we speak of the group of hybridizing species
as a hybrid swarm. Hybridization appears to be much
more common in plants than in animals. This difference
California
may be due in part to the more passive nature of mating
in plants, which must rely on wind, water, or animals to
Mexico transport pollen from one flower to another. Hybrid
swarms tend to be more common in wind-pollinated
Figure 6.13 genera.
Geographic ranges of three species in the genus Gilia (Pole- Another feature of evolution common to plants is
moniaceae). The most likely explanation for the pattern of polyploidy, the duplication of the entire set of chromo-
adjacent ranges is allopatric speciation of an originally wide-
spread species. somes, resulting in two or more copies of the genome in
each cell. Polyploidy can come about in two ways. First,
consider a diploid (2n) individual that by some error
produces pollen and ovules without meiosis (the reduc-
tion division), resulting in diploid gametes with 2n
it can occur in adjacent populations (parapatric specia- chromosomes instead of haploid gametes with 1n chro-
tion), or even within a single population (sympatric spe- mosomes. If that individual self-pollinated, its offspring
ciation). These modes of speciation are common to both would have four copies of each chromosome instead of
animals and plants; however, plants undergo some the usual two; these offspring would be tetraploid, with
unique speciation processes. 4n chromosomes. Such doubling does not require self-
The biological species concept defines a species pollination; if two separate individuals both produced
as a group of actually or potentially interbreeding organ- unreduced, diploid gametes and they mated with each
isms that are reproductively isolated from other such other, again a tetraploid offspring would result. This
groups (Mayr 1942). A species thus forms an evolution- process is referred to as allopolyploidy if the gametes
ary unit within which populations are linked in their come from individuals of different species in conjunc-
genetic changes through time. In most cases, species tion with hybridization. It is referred to as autopoly-
defined in this fashion are identical to species defined ploidy if both gametes come from individuals of the
by taxonomists. Taxonomic species are generally same species.
defined based on shared morphological characters. One Polyploidy does not have to stop at 4n; further dou-
problem with using morphological criteria to define blings can take place. Hybridization with 2n members
species, however, is the existence of cryptic species, of the same species can result in 6n offspring. In one
individuals that appear to belong to the same species species of saltbush, Atriplex canescens, individuals rang-
and yet are reproductively isolated from each other. In ing from 2n to 20n are known. Ferns, which are a very
plants, cryptic species occur among polyploids (see ancient lineage, often have very large numbers of chro-
below). More common are individuals that appear to be mosomes. The Polypodium vulgare (Polypodiaceae) group
different, especially if they come from distant locations, has individuals with 2n = 74, 148, and 222. The high base
yet are still able to interbreed. If those individuals are number is undoubtedly due to much older polyploidy
connected by others that are morphologically and geo- events. The highest chromosome number known in any
graphically intermediate, they would probably be clas- plant occurs in the fern Ophioglossum reticulatum
sified as members of the same species. (Ophioglossaceae), with 2n = 1260. The modular devel-
The reproductive barriers between species are not opmental system of plants, at least in some taxa, is
always impermeable, however. Hybridization occurs apparently able to handle these ploidy changes without
when members of different species of the same genus difficulty. A typical consequence of polyploidy is an indi-
mate and produce viable offspring (Figure 6.14). vidual that is larger or more vigorous than its parents,
(“Hybridization” as used by evolutionary biologists has but otherwise similar in appearance. The larger size
a different meaning to agronomists, who use the same comes about in part because cell size is proportional to
114 Chapter 6
Genus Clarkia
prostrata 26 purpurea 26 affinis 26 delicata 18 pulchella 12
BIORTIS similis 17
CONNUBIUM CLARKIA
breweri 7
9 deflexa 9
17 9 xantiana 9
epilobioides 9 concinna 7
tenella 17 FIBULA
davyi 17
unguiculata 9 rhomboidea 12
EUCHARIDIUM
exilis 9
cylindrica 9 PHAEOSTOMA
9 8
dudleyana 9 bottae 9 virgata 5
mildrediae 7
lingulata 9 modesta 8
gracilis 14
biloba 8 MYXOCARPA
lassenensis 7
PERIPETASMA
arcuata 7 amoena 7 rubicunda 7
williamsonii 9
nitens 9
speciosa 9 imbricata 8
PRIMIGENIA
GODETIA
Figure 6.14
Hybridization among species in the genus Clarkia (Onagraceae). Names in uppercase
letters are sections of the genus. Names in lowercase letters are species, given with their
haploid chromosome number. Polyploid forms have arisen from hybridization between
species. (After Lewis and Lewis 1995.)
nucleus size; more chromosomes result in a larger nucle- In the genus Hesperis (Brassicaceae), for example, species
us and, thus, a larger cell. are known in which n = 7, 14, 13, and 12. In this instance,
Individuals with odd ploidy numbers (e.g., 3n or 5n) the most likely evolutionary pathway was first a dou-
are also possible. Such individuals cannot produce bling of the number of chromosomes from 7 to 14, then
gametes through typical meiotic processes because the a subsequent loss of first one, then a second chromosome.
chromosomes cannot pair properly. Through a process One result of polyploidy is individuals that are
known as agamospermy, some plants can produce seeds instantly reproductively isolated. A 4n offspring can-
without fertilization or meiosis (see Chapter 8); dande- not mate with its 2n progenitors, except by a new, rare
lions (Taraxacum officinale, Asteraceae) and grapefruit hybridization event. Thus, under the biological species
(Citrus paradisi, Rutaceae) are two such plants. (Some- concept, such a polyploid individual is a member of a
times pollination is required to initiate seed develop- new species. In order for the species to become estab-
ment, even though fertilization does not occur.) Seeds lished, however, the polyploid individual must repro-
produced by agamospermy contain an exact duplicate duce, and the population must grow. It is only because
of the parent’s genome. Agamospermy is not restricted most plants produce both male and female gametes and
to species with odd ploidy numbers. Species with odd are self-compatible that such individuals are likely to
ploidy also can evolve to have even ploidy through chro- reproduce. Atriplex provides an interesting exception; as
mosome doubling or hybridization. mentioned above, this genus contains many hybrids and
Aneuploidy, the gain and loss of individual chro- polyploids, yet it is dioecious (male gametes and female
mosomes, is another common way in which plant line- gametes are produced on different individuals). Despite
ages change their chromosome number. Aneuploidy is the reproductive isolation caused by polyploidy, taxon-
especially common in lineages that are already polyploid. omists tend to classify individuals with different ploidy
levels as members of the same species if their overall action. Phenotypic selection has been measured in
appearance is similar—an instance of taxonomic species dozens of populations and species. By examining
differing from biological species. instances in which selection is strong (e.g., heavy-metal
tolerance), we are able to find evidence for adaptation
by natural selection occurring over just the past few hun-
Summary
dred years. Much of our evidence comes from observ-
Explaining the origin and maintenance of current plant ing current patterns of variation among populations and
traits is not a simple matter. We can look at a plant and its species. These observations range from comparisons
environment and make an educated guess about the evo- among nearby populations of the same species growing
lutionary processes that shaped it. But elevating that guess in different environments, to comparisons among pop-
to the level of scientific confidence can be difficult. To some ulations separated by large geographic distances, to
extent, all traits are shaped by natural selection. Howev- comparisons among species that are distantly related
er, other processes, such as mutation and migration, are and geographically separated.
also important in shaping traits. Sorting out which process- The sum of these evolutionary processes creates the
es are responsible for current traits is difficult because dif- amazing diversity of the living world that we see around
ferent processes can create similar-looking outcomes. Evo- us. This diversity can be seen on scales ranging from
lutionary processes often work very slowly over long individuals to populations to species. Ecologists study
periods of time, making them difficult to observe in the this diversity to see how it fits together in today’s world.
short time span of most scientific studies. Evolution provides the context within which this diver-
Nevertheless, evolutionary biologists have been able sity arose, while ecology provides the context for the
to observe natural selection in action and infer its past operation of evolution in the past, present, and future.
Additional Readings
Classic References Van Tienderen, P. H. and J. Van der Toorn. 1991. Genetic differentiation
between populations of Plantago lanceolata. II. Phenotypic selection in
Clausen, J. D., D. Keck and W. M. Hiesey. 1940. Experimental Studies on
a transplant experiment in three contrasting habitats. J. Ecol. 79: 43–59.
the Nature of Species. I. Effects of Varied Environments on Western North
American Plants. Carnegie Institute of Washington, Washington, DC.
Turesson, G. 1922. The species and the variety as ecological units. Hered- Additional Resources
itas 3: 100–113. Endler, J. A. 1986. Natural Selection in the Wild. Princeton University Press,
Princeton, NJ.
Contemporary Research Primack, R. B. and H. Kang. 1989. Measuring fitness and natural selec-
tion in wild plant populations. Annu. Rev. Ecol. Syst. 20: 367–396.
Kalisz, S. 1986. Variable selection on the timing of germination in Collinsia
verna (Schrophulariaceae). Evolution 40: 479–491. Travis, J. 1989. The role of optimizing selection in natural populations.
Annu. Rev. Ecol. Syst. 20: 279–296.
Lechowicz, M. J. and P. A. Blais. 1988. Assessing the contributions of mul-
tiple interacting traits to plant reproductive success: Environmental
dependence. J. Evol. Biol. 1: 255–273.
C H A P T E R
7 Population Structure,
Growth, and Decline
I
s a particular plant population increasing, decreasing, or staying constant
in number? What is its genetic composition? How are the plants in the pop-
ulation arranged spatially? Questions of this kind are central to both basic
and applied ecology. Foresters may be interested in maximizing the survival
of trees that have been attacked by bark beetles, while conservation ecolo-
gists may seek to prevent the decline of a rare plant species. Evolutionists want
to understand the action of natural selection on plant traits. Land managers
want to control the spread of invasive weeds. All of them are asking questions
that fall into the domain of population ecology—the study of population
growth, composition, and spatial dispersion. Changes in a population, such as
changes in the number of individuals alive, changes in the genetic composi-
tion or age structure of a population, or movement among a population’s indi-
viduals are all part of a population’s dynamics.
In studying population dynamics, we consider the properties of popula-
tions rather than individuals. For example, the wind may blow an individual
seed a certain distance from its parent plant; but an entire plant population has
the property mean seed dispersal distance. Individuals germinate and die on par-
ticular days, but populations have germination and death rates. These exam-
ples may be simple, but the distinction between individual and population
properties can sometimes seem complicated. For example, the net transpira-
tion and the photosynthetic rates of trees in a closed canopy depend both on
the properties of the individual trees and on population properties such as the
number, sizes, and spacing of individuals. Not only do populations have prop-
erties that individuals do not, but population-level properties can be more than
just the sum of the properties of individuals. Thus, we have to study the pop-
ulations themselves.
Plant population ecology dates only from the 1960s. Prior to that time, ecol-
ogists studied plant communities and the physiological ecology of plants,
but did not generally think of plant populations as being composed of indi-
viduals that varied. In this way, they lagged behind animal ecologists, who had
been acquiring demographic data and modeling population growth since the
1940s. John L. Harper, working at the University of Bangor in Wales, first intro-
duced a strong population-level perspective into plant ecology. Harper and his
students published a large number of studies in the 1960s and 1970s examin-
ing aspects of the growth of plant populations. Indeed, the first study to use a
118 Chapter 7
population growth model tailored ferent kinds of phenomena would be to distinguish

for plant populations was pub- between changes in the population that depend on inter-
lished in 1974 by Harper’s student actions among individuals (such as competition) and
Jose Sarukhán, working with changes that do not require such interactions. Because
Madhav Gadgil (Sarukhán and ecologists use the terminology of density to describe pop-
Gadgil 1974). In the decades since ulation change, we will use it here as well.
this early work, there has been an This chapter concerns density-independent changes
explosion of work in plant popu- in population size. We discuss density-dependent
lation ecology; hundreds of changes in populations in Chapters 10 and 11.
papers are now published annu-
José Sarukhán ally. Harper’s students have in Some Issues in the Study of Plant
turn trained many plant ecologists
all over the world. Sarukhán, for example, returned to
Population Growth
Mexico and helped develop the Universidad Nacional What controls the growth and decline of plant popula-
Autónoma de México into an important center for eco- tions? Change in population size is determined by the
logical research, especially on plants. numbers of births and deaths, plus the number of indi-
Ecologists often distinguish between density- viduals immigrating into the population, minus the
dependent and density-independent population growth. number emigrating from the population. Ecologists usu-
Some factors affecting population growth depend on the ally get data on changes in population size by conduct-
density of a population. For example, competition might ing regular censuses—say, by counting individuals
be important in a crowded stand of plants, but much less every year or every week. If we think of the first census
important when there are greater distances between indi- as occurring at time t, and the next census as occurring
viduals. Thus, competition is usually thought of as a den- at time t + 1, then we can describe the change in popu-
sity-dependent factor. By contrast, if 10% of all juveniles lation size as
were killed by a severe storm, regardless of population
density, ecologists would say that the weather is acting nt+1 = nt + Bt – Dt + It – Et (7.1)
as a density-independent factor.
Although this distinction between density depend- Here the n’s are the population sizes at sequential cen-
ence and density independence seems simple enough, suses. For example, if we were tracking a population of
nature is often more complicated. If a severe storm kills Pinus ponderosa (Ponderosa pine, Pinaceae), nt would be
10% of all the individuals not growing in sites protect- the number of individual trees in the first year of our
ed from high winds, and protected sites are in limited census, and nt+1 would be the number of trees in the fol-
supply, is the mortality density-dependent or density- lowing year. Bt is the number of births between the two
independent? Clearly, the distinction between the two censuses, Dt the number of deaths, It the number of in-
types of mortality is not very meaningful in such cases. dividuals arriving (immigrating) from other popula-
Nevertheless, this distinction is often quite useful,* and tions, and Et the number dispersing away (emigrating)
in any case density independence is a good starting from the population.
point for thinking about population growth. The relationship in Equation 7.1 may seem to be a
This distinction must be applied cautiously to plants. truism. Though simple, it is a useful starting point
In discussing the assessment of competition, Harper because it helps us to focus on some major factors affect-
(1977) pointed out that the average density over a whole ing plant population growth. It also points to some major
plant population is not the important quantity to meas- differences between plant and animal populations and
ure. Because plants are sessile, they usually compete only to some major conceptual problems in plant ecology.
with their neighbors, not with more distant plants in Equation 7.1 implies several equivalent ways in
the same population (see Chapter 10). Thus, when assess- which plant population size can be affected. It implies,
ing competition, one needs to think of density depend- for example, that the same population growth can be
ence on a very localized scale. Other spatial scales may assured by increasing the number of births or by decreas-
be important for other phenomena—for example, many ing the number of deaths. As we will see in much of the
pathogens can disperse over considerable distances, so rest of this chapter, life is not generally so simple unless
average density over a larger area may be a useful quan- all of the individuals in a population are demographi-
tity to measure to understand the progress of disease in cally equivalent, and that is rarely the case. Large plants
a population. A more general way to categorize these dif- usually have different effects on the population than
small ones, and old plants usually have different effects
* Many important scientific ideas are like this: valuable in one context but not
in another. That is because such ideas are simplified descriptions of nature— than young ones. Therefore, adding x births is not usu-
useful but limited models of patterns or processes (see Chapter 1). ally equivalent to reducing the number of deaths by x.
Population Structure, Growth, and Decline 119
But Equation 7.1 raises still more issues. What exact- taceae Figure 7.1), for example, lose joints that are then
ly do we mean by “birth” and “death” in plant popula- moved by animals or runoff from storms; these may take
tions? That depends on whether we mean births and root and grow into mature plants. At the other end of
deaths of genetically distinct individuals (genets) or of the moisture availability gradient, while dispersal of
individuals that may be physiologically independent but clonal fragments is rare in marine angiosperms such as
are not necessarily genetically distinct (ramets; see Fig- Zostera spp. (eelgrasses, Zosteraceae; Barrett et al. 1993),
ure 5.8). One useful way to keep this distinction straight it may be quite important in the establishment of new
is to remember that every new genet is also a new ramet, populations. Since most living plant tissues can, at least
but the reverse is not necessarily true: a new (potential- potentially, generate entirely functional new plants, dis-
ly) physically independent individual is not necessarily persal by clonal fragmentation may be widespread.
a new genetic individual. Dispersal by clonal fragmentation has another
New genets are formed by the fertilization of an implication. Seed dispersal means that an individual
ovule and the maturation of the resulting seed (see Chap- immigrating to one population must be emigrating from
ter 8). In plant species that never spread clonally (such as another. Clearly this is not so with dispersal by frag-
most annual herbs), each genet is a single ramet. Plant mentation: while the fragment itself may leave one pop-
ecologists usually think of birth as occurring either when ulation and move to another, the rest of the genet from
the seed is mature (i.e., the embryo is fully formed) or which it was derived may stay put.
when the maternal plant sheds the seed. Seed germina- This does not mean that dispersal by fragmentation
tion is not considered birth because seeds are already is a “free lunch.” Indeed, it has an obvious consequence:
new individuals—physiologically independent and (usu-
ally) genetically distinct from their parents long before
they germinate. Moreover, they often spend a substan-
tial amount of time, sometimes years, living in the soil
before germination. Plant populations whose seeds have
such a dormant phase are said to have a seed bank.
Many plants can spread clonally (also called vege-
tatively) by producing structures that contain tissues
capable of producing an entire new plant, which is
genetically identical to the parent. In these species, each
genet is usually made up of multiple ramets, some of
which may be physiologically independent from one
another. Thus, discussing births and deaths in popula-
tions of these species means that first we must decide
whether we are studying change in the number of genets
or in the number of ramets. Studies of genet number
would use a definition like the one above: newborns
would consist only of new seeds. Studies of ramet num-
ber would count all new individuals, including both
new seeds and new shoots, because all of them are new
ramets. Which definition is best? That depends on the
questions one asks.
This distinction between ramets and genets affects
the way we think about dispersal between populations
(the It and Et terms) as well. Most mature plants are ses-
sile—rooted in place—but (by the above definition)
seeds are certainly plants, and they are usually mobile.
Similarly, pollen grains are plants that are usually mobile
(a pollen grain is the male gametophyte of gym- Figure 7.1
nosperms and angiosperms, and is multicellular and Many Opuntia cacti—especially the chollas, which have
genetically different from the sporophyte plant that pro- cylindrical joints—disperse partly by clonal fragmentation;
duced it; see Chapter 8). It should not be surprising, some of the fragments root and establish new ramets. Opun-
therefore, that seeds and pollen, and their dispersal, have tia fulgida, shown here, is known as “jumping cholla”
because its joints very readily detach from the parent plant;
been and continue to be the subject of much ecological unwary people walking by (including the authors) have
research. But some plants can disperse by clonal frag- been unpleasantly surprised by this plant! (Photograph by S.
mentation as well. Many cholla cacti (Opuntia spp., Cac- Scheiner.)
120 Chapter 7
the fragmenting genet is packaged into smaller ramets.

Since the survival and fecundity of ramets usually depend
at least in part on their size, it is possible that clonal frag-
mentation might eventually affect the survival of the
genet as a whole. It is an open question how the popula-
tion dynamics of ramets and of genets affect one another.
Population Structure
The preceding discussion points to an important con-
clusion: When we study plant populations, we usually
need to pay attention not only to the total number of
individuals in a population, but to the classes of indi-
viduals and the relative frequencies of each class in the
population as well. These classes are usually defined by
age, life history stage (e.g., seed, seedling, sapling,
canopy tree), size (e.g., 0–2 cm, 2–4 cm, >4 cm tall), or
some combination of these categories. The population
structure is a description of the relative frequency of
each class (Figure 7.2). Figure 7.2
Information about population structure is important Stage structure of a population of Pentachlethra macroloba
for two reasons. First, different kinds of individuals have (Fabaceae), a tropical tree species, at Barro Colorado Island,
different effects on population growth. Adding 100 seeds Panama. Here, the stage classes above seed are based on
to a population of the giant Sequoia sempervirens (Cali- size. Classes 2–7 (seedlings; light green) are in increments of
50 cm in height. Subsequent classes are based on diameter at
fornia redwood, Taxodiaceae), for example, would obvi- breast height (DBH): class 8 = 2–5 cm, class 9 = 5–10 cm,
ously have a very different effect on the growth of the class 10 = 10–20 cm, classes 11–14 are in 20 cm increments,
population than adding 100 mature redwood trees. If and class 15 > 100 cm. Thus 88% of the population consists
you were asked to census a tree population, you would of seeds and 9% of seedlings 0–50 cm in height. The remain-
almost certainly count seeds, seedlings, saplings, and ing stages (gray bars) comprise only 3% of the population.
(After Hartshorn 1975.)
mature trees as different classes, and you might make
even finer distinctions. Ecologists need to know what
kinds of individuals are in a population as well as their
numbers. This is generally true for both animal and seed, seedling, juvenile, small individual, large individ-
plant populations. ual—plays a stronger role in determining demographic
The second reason we need to know about popula- performance than age; therefore, plant populations are
tion structure is particular to plants. Because of their generally thought of as stage-structured (see Figure 7.2).
extreme phenotypic plasticity, individual plants can vary This should be obvious in the case of reproductive out-
over orders of magnitude in size, shape, physiological put because the number of flowers and fruits usually
status, and consequently, in their importance for popu- depends on plant size. Survival and subsequent growth
lation growth (see Chapter 5). Animals certainly vary in usually also depend on plant size more than on age.
their size and contribution to population growth as well. Older plants are usually larger, but in most plants there
But we can often predict the vital rates of an animal—its is so much variation in the sizes of individuals of a given
reproductive rate and its chance of surviving—if we age (Figure 7.3) that it is often more useful to use
know only its age. Many animal populations (especial- stage-based methods than age-based methods to study
ly mammals and birds) can thus be thought of as plant demography.
age-structured. To describe such populations, we need That said, we hasten to add that chronological age
to know how many individuals are in each age group. can be important to know in plants for several reasons.
The methods used for the study of these populations are Age is quite important in an evolutionary context; for
age-based, requiring only information on the popula- example, if we want to understand how the genetic
tion’s age structure. Age-structured populations have composition of a population changes over time, we need
another convenient property: they progress in only one information about the ages of individuals. Similarly, if
direction. A bear that is x years old now will either be we want to study the evolution of life history traits (see
dead or age x + 1 next year. Chapter 9), we often need information about the ages of
In most plant populations, stage—a category based individuals. As we will see in several examples in this
on size, physiology, or developmental status, such as chapter, age is sometimes an important determinant of
PAGE PROOF: 2ND PASS Population Structure, Growth, and Decline 121
old seeds have a different chance of germinating than

120 two-year-old seeds, and a different chance of surviving
as viable seeds in the seed bank. Similarly, in Cypripedi-
100 um acaule (pink ladyslipper orchid, Orchidaceae; Figure
Average number per acre
7.5), the survival of corms from year to year depends on

80 their age, and the survival of dormant plants from year
to year depends on the number of years they have been
60
dormant.
While age can be important, it is frequently impossi-
ble to determine a plant’s age. Although many temper-
40
ate-zone trees do produce reliable annual growth rings,
most other plants—both herbs and woody plants—do not.
20
80
Unless we are working with marked individuals of known
60 age, then, we usually do not know the ages of plants (but
0
s)
see the discussion of long-lived plants on page 133).
ar
18 16 40
ye
14
(
12 20
Diamete 10 ge Some population structure issues special to plants
A
r (inche 8
s)
Why are plants so much more variable in their popu-
Figure 7.3
The average relationship between age and size in Pinus
lation structure than animals? A major reason is their
palustris (longleaf pine, Pinaceae) in the southeastern United modular growth form: an individual plant is a system
States. The data show the average number of plants per acre of repeated units (nodes, lateral organs, and internodes:
of a particular size, given the age of the stand. Longleaf pine see Chapters 5 and 8). This modularity means that
frequently grows in such even-aged stands. (After Forbes plants have very flexible growth patterns. It also means
1930.)
that they can lose large areas of their bodies and still
survive. In other words, plants can actually shrink from
year to year. In the ladyslipper orchid, for example,
survivorship (the chance of surviving) and fecundity adults may not appear aboveground in some years (see
(the number of offspring contributed to the next gener- Figure 7.5).
ation). But age alone is not a very good predictor of these Stage structuring makes life complicated for plant
vital rates in plants; stage information is usually need- ecologists. As mentioned above, when dealing with ani-
ed as well. mals, one can be certain that in a year, all individuals now
As recently as the 1980s, there was some
doubt about the relative importance of age
and stage in plant demography. Older text-
books, for example, often discussed the prob- F47
lem of “age- versus stage-based demogra- F44 F46
F45
phy.” It is now clear that when chronological
age is important, it is usually important only
within the context of a particular stage (see Adult4 Adult5 Adult6 Adult7
Caswell 2001). For example, as we can see in
a life cycle graph for Collinsia verna (blue-eyed F16
F15 F17
Mary, Campanulaceae; Figure 7.4), age plays P51
an important role in the seed bank: one-year- P62 P73
Seed1 Seed2 Seed3

P21 P32
P33
Figure 7.4
Life cycle graph of Collinsia verna (blue-eyed Mary, Campan-
ulaceae), an annual plant of eastern North America with an
age-structured seed bank. There are three age classes of
seeds; when these germinate, the resulting vegetative plants
are in different classes. The notation for life cycle graphs is
explained on pages 123–124. (After Kalisz and McPeek 1992;
Collinsia verna photograph courtesy of S. Kalisz.)
122 Chapter 7
Cypripedium acaule
Figure 7.5
Life cycle graph of Cypripedium acaule (pink
ladyslipper, Orchidaceae), a terrestrial orchid
from eastern North America. After germina-
tion, a corm develops underground for at
least two years; subsequently, each plant pro-
duces zero, one, or two leaves every year.
Mature C. acaule can produce additional ram-
ets, treated in this study as new individuals;
this vegetative reproduction is not shown
here. (Photograph by S. Scheiner.)
x years old will be either dead or a year older. Even in to keep track of many stages and possible transitions
many stage-structured animal populations, individuals between them.
can never move to an earlier stage—frogs do not become
tadpoles. There are few such prohibitions in plants. Once Sources of Population Structure
germinated, a plant can never become a seed, but in most Individuals within plant populations vary in many char-
plant species established plants can grow, stay the same acteristics, including size, morphology or developmen-
size, or shrink (Figure 7.6). Even trees, which we often tal status, genotype, and physiological status. Size is the
think of as always growing larger, lose branches and sec- most commonly studied factor structuring plant popu-
tions of their trunks. Consequently, plant ecologists need lations. Size can be measured in several ways: as a quan-
Figure 7.6
Life cycle graph of Argyroxiphium sandwicense (Mauna Kea
silversword, Asteraceae) from Hawaii. These plants are
mainly semelparous, reproducing only once in their lives;
rare individuals, however, revert to vegetative growth after
flowering, or flower repeatedly. (After Powell 1992; photo-
Argyroxiphium sandwicense graph courtesy of R. Robichaux.)
tity such as weight or height or as the number of mod- old. Thus neither very old nor very young trees are
ules (e.g., branches, tillers in a grass, or leaves) a plant reproducing, so a population of 100 very old or very
has. Tree size is often measured as diameter at breast young trees will simply decline in number as some die
height (DBH); curiously, “breast height” is defined as and none reproduce.
1.3 meters in some countries and 1.4 in others. Morpho- Now imagine that year-to-year survival rates are the
logical or developmental status is usually important in same among young and old trees, and compare a pop-
structuring populations: how many individuals are ulation composed entirely of old trees with one com-
seeds, how many seedlings have become established, posed entirely of young trees. In the short run, both pop-
how many plants have become reproductively active? ulations will decline at the same rate. But over time, this
Genetic variation is often important in structuring pop- will not be true, because the surviving trees in the young
ulations, and physiological status is probably often population will eventually reach maturity and begin
important as well. reproducing, so the population will grow. By contrast,
This list of categories may seem a bit overwhelm- the old population will disappear. Therefore, studies of
ing, but categories such as size, morphology, age, geno- population growth and decline in plants must take into
type, and physiological status are important for popu- account the structure of the population and look at both
lation studies only if there are demographic differences long-term and short-term growth.
among the classes. For example, it might be easy to
measure the difference in size between 30-, 60-, and Life Cycle Graphs
90-cm DBH trees of the same species, but it may not be To learn how population structure affects population
useful to use these size differences as size classes if the growth, we need a picture of how individual plants can
trees of each size did not differ from the other sizes in shift from stage to stage between censuses. Life cycle
their chances of survival, growth, or reproductive out- graphs such as those in Figures 7.4 through 7.7 provide
put. useful summaries of information about transitions
Currently, there are real limits to our abilities to between stages. Each node in a life cycle graph repre-
measure this variation, to tell how important it is in sents a stage. Note that some stages are defined by
structuring a particular population, and to perform developmental status (e.g., seeds and dormant individ-
appropriate computations. Consequently, most ecolog- uals), while others are defined by their size or age with-
ical studies today concentrate on only one or a few of in a developmental stage. Arrows between stages
these categories of variation. Put differently, we know describe “transitions” between those stages—survival
that reality is much more complex than our demo- and reproduction—during the interval between cen-
graphic analyses. suses.
Let’s look at the life cycle graph for Coryphantha rob-
binsorum (pincushion cactus, Cactaceae) (Figure 7.7).
Studying Population Growth and Decline
(This endangered cactus species will be used as an exam-
Is a population growing or declining, and at what rate? ple throughout the next few sections of the chapter.) The
This might seem like a simple question to answer: com- arrow from large juveniles to adults represents the prob-
pare the number of plants last year (call it nt) with the ability (P) of a large juvenile at one census becoming
number this year (call it nt+1). The rate of change in pop- an adult by the next census. The self-loops—arrows from
ulation size is then n t+1/nt. If the population is growing, a node to itself—represent the probability of remaining
this ratio is greater than 1; if the population is declining, in the same stage. Thus, the arrow from small juveniles
the ratio is less than 1. to small juveniles in Figure 7.7 represents the probabil-
Unfortunately, life is this simple only when there is ity that a small juvenile at one census will still be a small
no population structure. Certainly it is true that if nt+1 juvenile at the next census.
> nt, there are more plants now than there were last year. The arrow from adults to small juveniles in Figure
But in a structured population—that is, in most plant 7.7 needs a more careful interpretation, which points to
populations—individuals of different classes make dif- an important lesson about life cycle graphs and the
ferent contributions to future population growth. This matrix models that can be derived from them. This
means that their short-term effects can differ from their arrow refers to the number of small juveniles produced
long-term effects on the population, and we need to be by the adults. Biologically, we know that every indi-
able to ask about both kinds of effects. vidual starts as a seed, and that there is generally some
Consider a population of a long-lived tree species. time between seed maturation and germination (there
The oldest trees have a period of senescence that lasts are a few viviparous plant species that germinate right
up to several decades, during which they do not flower on the maternal plant, but Coryphantha is not one of
but slowly lose limbs and experience heartwood rot. them). A complete biological diagram of the life cycle
Young trees cannot flower until they are several decades would include these steps. But we are trying to describe
124 Chapter 7
Coryphantha robbinsorum
Figure 7.7
Life cycle graph of Coryphantha robbinsorum (pincushion cac-
tus, Cactaceae), an endangered cactus from Arizona and
the demography of the life cycle as examined in a real Sonora, Mexico. As this species has no seed bank and plants
were censused annually, the graph does not include a seed
field study. Since this species has no seed bank and the stage. The transition between adults and small juveniles is
censuses are annual, no individuals in the seed class thus the product of (rate of seed production) times (chance
could have been counted unless the census were timed of a seed surviving) times (chance of germination) times
to include them. To represent Coryphantha demography (chance of a seedling surviving to the first census). (Data
as measured, then, we need to estimate fecundity: from Schmalzel et al. 1995; photograph courtesy of U.S. Geo-
logical Survey, Phoenix, AZ.)
F = (rate of seed production) × (chance of seed survival) ×
(chance of germination) × (chance of seedling surviving to
the first census) =
in a life cycle graph or matrix model can greatly affect
(the rate at which adults one year produce small juveniles
most calculations for which the graph or matrix model
the next)
is subsequently used.
If we wanted to study the seed stage per se, we would Once we have a life cycle graph, we can then use
need censuses that were timed differently, and possibly field data to estimate values for the various transitions
more frequent censuses. This may seem like a quibble, (Figures 7.4–7.7; Box 7A). Often it is impossible to
but it turns out that the number of classes we include observe all of the transitions directly using the same indi-
BOX 7A
Demography of an Endangered Cactus, Coryphantha robbinsorum

Coryphantha robbinsorum is a small, a model that does not include a seed Analysis of these matrices reveals
cluster-forming cactus found on lime- stage. The data can be used to create that the population at site C is growing,
stone outcroppings in southern Ari- matrix models for each of the three because λ1 = 1.12. At the other two sites,
zona and adjacent Sonora, Mexico. sites: the population is stable or declining
Schmalzel et al. (1995) marked plants slowly: λ1 = 0.998 at site A and λ1 =
0.67 0 0.56 
at three sites on a hill and followed 0.997 at site B. The stable stage distri-
their growth, reproduction, and sur- A site A= 0.02 0.85 0  butions differ among these plots, as Fig-
vival for a five-year period.  0 0.14 0.87  ure 7.9 shows. Most notable is the pre-
Because adults (plants that had flow- dicted scarcity of large juveniles at sites
ered at least once), small juveniles 0.49 0 0.56 A and B. The plots also differ in the
(plants smaller than 11 mm in diame- rates at which they approach the stable
ter), and large juveniles had signifi- A site B = 0.01 0.73 0 

distribution. Sensitivities are shown in
 0 0.23 0.99
cantly different chances of survival and Table 7.2, elasticities in Figure 7.10, and
reproduction, the authors used these reproductive values in Table 7.1.
and seeds as their stage classes. Since C. 0.43 0 0.56 
robbinsorum has no seed bank (see Fig-
ure 7.7) and the censuses were annual, A site C= 0.33 0.61 0 
 0 0.30 0.96 
we re-analyzed the authors’ data with
viduals. For example, it is difficult to use the same indi- large juvenile becoming an adult, and so on. Using these
viduals to estimate the survival and germination rates symbols, we can write an equation for the number of
of seeds as well as subsequent transitions, because seeds small juveniles in year t + 1:
usually cannot be marked, and even when they can be,
n1(t +1) = P11 n1(t) + F n3(t) (7.2)
accurate censusing usually requires digging them up—
which can certainly bias their future chances of surviv- In other words, the number of small juveniles this year
ing or germinating! is the sum of the individuals that remain small juveniles
In a large study of Collinsia verna, Susan Kalisz and plus new juveniles; that is, the number of small juveniles
Mark McPeek (1992) addressed this problem by esti- last year times the chance that they survived as small
mating seed survival rates destructively but independ- juveniles, plus the number of adults last year times their
ently from their estimates of the vital rates of above- rate of production of small juveniles.
ground plants (see Figure 7.4). They were able to do so Similarly, the number of large juveniles is given by
because C. verna grows in very large populations, mak-
n2(t + 1) = P21 n1(t) + P22 n2(t) (7.3)
ing it possible to randomly select separate areas for
studying aboveground and belowground demography. That is, the number of large juveniles this year is the
On the other hand, in the ladyslipper orchid study number of small juveniles last year times the chance that
(see Figure 7.5), Margaret Cochran and Stephen Ellner they became large juveniles, plus the number of large
(1992) were forced to make several assumptions about juveniles last year times the chance that they remained
mortality rates in dormant plants and seeds. It would large juveniles.
have been more satisfactory to observe the belowground Finally, the number of adults is
events directly, but this would have destroyed study
n3(t + 1) = P32 n2(t) + P33 n3(t) (7.4)
areas, and might have added to the effects of poaching
on these orchids. There are no simple or general solu- The number of adults this year is the number of large
tions to this problem. juveniles last year times the chance that they became
adults, plus the number of adults last year times the
Matrix Models chance that they survived as adults.
How do the rates of survivorship and fecundity affect A transition matrix model is a compact way of writ-
the growth of a population as a whole? Which stages ing the same thing. The matrix arranges the coefficients
have the strongest effects on population growth? Ques- from the life cycle graph (the F’s and P’s, shown as a
tions like these are important in many contexts, includ- matrix in the middle of Equation 7.5). This matrix thus
ing conservation biology (on which stages should we represents the survivorship and fecundity rates. Next a
concentrate protection efforts?), population ecology vector is assembled consisting of the number of indi-
(which stage is most likely to limit population growth?), viduals of each stage; in Equation 7.5, this is the column
and evolution (on which stage can natural selection have on the right-hand side. To obtain a vector of the number
the greatest effect?). of individuals in each stage next year—the column on
One approach to answering such questions is to use the left-hand side of Equation 7.5—we multiply the vec-
estimated survivorship and fecundity rates to model the tor by the matrix. In Box 7B, we explain the rules for
population’s growth rate, using matrix demographic multiplying the vector by the matrix. A matrix model of
methods developed over the past two decades. These the life cycle graph in Figure 7.7 can thus be written as
models can be used to ask how the population growth
rate changes as the specific survivorship and fecundi- n 1(t + 1)  P11 0 F  n1(t) 
ty rates in each class change. All of the information   P P   
needed for this method is contained in the life cycle n 2 (t + 1)  =  21 22 0  × n2 (t)  (7.5)
n 3 (t + 1)  0 P32 P33  n3 (t) 
graph.
To see this, consider the Coryphantha example (see
Figure 7.7). The life cycle graph tells us that small juve- More general rules for manipulating matrices are given
niles are produced by adults at a rate F (as noted above, in Caswell (2001).
F is a composite of several factors). For convenience, we The principal reason for using matrices is that there
number the stages from 1 to 3, so that every symbol we are standard rules for manipulating them, and these
use can be interpreted. Thus n1 means the number of rules make it much easier to draw some important eco-
small juveniles, n2 the number of large juveniles, and logical conclusions about the population. If we assume
n3 the number of adults. Each transition has two sub- (for the moment) that the birth and survival probabili-
scripts—the first refers to the stage this year, and the sec- ties (the coefficients) stay constant (so that each year we
ond to the stage last year. Thus P11 is the probability of multiply the population vector by the same matrix), we
remaining a small juvenile, P32 is the probability of a can find
126 Chapter 7
BOX 7B
Multiplying a Population Vector by a Matrix

To understand how Equation 7.5 can number of small juveniles), multiply To find the second element in next
say the same thing as Equations 7.2 the coefficient in the first row, first col- year’s vector (the number of large juve-
through 7.4, you need to know how to umn (P11), by the first element in the niles), repeat this process, but multiply
multiply matrices and vectors. When vector for this year [n1(t)] to get P11 n1(t). each of the elements of this year’s vec-
you multiply a vector by a matrix, the Then multiply the coefficient in the first tor by the coefficients in the second row
result is a vector. In this case, the ini- row, second column (0), by the second of the matrix.
tial vector is the numbers of individu- element in the vector [n2(t)] to get 0. Another useful way of thinking of
als in each stage this year, the result- Finally, multiply the coefficient in the the matrix is that it describes the tran-
ing vector is the numbers of individuals first row, third column (F), by the third sitions in “from-to” form. The matrix
in each stage next year, and the matrix element in the vector [n3(t)] to get element in the ith row and jth column
is the survivorship and fecundity rates. Fn3(t). Then sum these three products always refers to the transition from
Matrix multiplication is “row-by-col- to get n1(t + 1) = P11 n1(t) + 0 + Fn3(t), the jth stage to the ith stage—from the
umn.” To get the first element in the which is exactly what Equation 7.2 “column-number” stage to the “row-
population vector for next year (the says. number” stage.
• The short- and long-term population growth rates tor attains a stable structure or stable stage distribution,
• The population structure at any time in the future at which point the proportion of individuals in each class
• The reproductive value of each age or stage class. stays constant each generation, although the population
Roughly speaking, the reproductive value of an keeps growing. This observation implies that, at least
individual in class x is its expected contribution to once the population has reached its stable structure, we
future population growth. In other words, repro- can multiply the population vector by a single number
ductive value is a way of evaluating the relative (a scalar) rather than by the entire matrix and get the
demographic importance of the different classes same result as if we were multiplying by the matrix.
• The sensitivity and elasticity of population growth How do we find this number? Mathematically, we
to changes in the specific probabilities of survival are looking for a number, which can be denoted by λ (the
and reproduction. Sensitivity tells us how absolute Greek letter lambda), based on the equation A x = λ x,
changes (e.g., how adding 0.01 to a survivorship where A is the transition matrix and x is a vector. (In this
coefficient) in the survival and fecundity of each chapter we will use standard mathematical notation:
class would affect population growth rates; elastic- capitalized boldfaced symbols, such as A, are matrices,
ity tells us how proportional changes (e.g., how and lowercase boldfaced italic symbols, such as x, are
increasing a survival coefficient by 1%) would vectors.) It is always possible to find such numbers (the
change the population growth rate. values for λ) for transition matrices; this is almost always
• The relationship between the ages of individuals done numerically on a computer. Numbers that can play
and their stages this role are called eigenvalues or characteristic values.
Every such number λ has a corresponding population
Matrix models and the estimates of growth rates, vector x, called an eigenvector or characteristic vector.
reproductive value, sensitivity and elasticity that they For a population with N stages, the transition matrix has
provide are important tools in evolutionary ecology, con- N rows and columns. There are also N eigenvalues. Each
servation biology, and applied ecology. eigenvalue has a corresponding eigenvector. In most
cases, all of the eigenvalues are distinct, although under
Analyzing Matrix Models some circumstances there are pairs of duplicated eigen-
How do we get all of this information from matrix mod- values.
els? In this section we introduce some major ideas used The number λ tells us some important things about
in analyzing these models. The basic method used to a population. If λ > 1, the population is growing, while
analyze matrices depends on an important observation: if λ < 1, it is declining in number. The population
if we repeatedly multiply the population vector—the remains at a constant size only in the special case that
vector of the number of individuals of each stage—by λ = 1. But what are these eigenvalues and eigenvectors?
the transition matrix, after a while the population vec- The eigenvalues are components of the population’s
PAGE PROOF: 2ND PASS Population Structure, Growth, and Decline 127
Population size Component 1 Small where the c’s depend on the initial condi-
15 15 juveniles tions (Caswell 2001), xi is the ith eigenvec-
Number in class
Large tor, and t is the number of time steps. Rais-
Contribution
juveniles ing the eigenvalues to a power (equivalent
10 10
Adult to repeated multiplication by the matrix)
5 5 makes the smaller eigenvalues decline in
relative importance over time. Recent stud-
ies have taken a look at methods for studying popula-
0 2 4 6 8 10 0 2 4 6 8 10
tions that are far from their stable distribution (Fox and
Year Year
Gurevitch 2000).
Component 2 Component 3 But Real Plants Live in Variable Environments
2 2 By now you might reasonably be wondering whether
this approach to plant populations makes much sense.
Contribution
Contribution
1 1
In our discussion of matrix models, we have assumed
0 0 that a population experiences fixed survival and birth
rates. Then we have concentrated on what happens
–1 –1 eventually in a constant environment. Since one of the
–2 –2 more obvious facts about ecology is that things change,
0 2 4 6 8 10 0 2 4 6 8 10 and since factors influencing populations (such as the
Year Year weather) can be quite variable (see Chapter 18), the
Figure 7.8 unrealistic assumptions of these matrix models might
Growth of the three stage classes in the Coryphantha robbinso- seem to render any conclusions based on them invalid
rum population at site C, and convergence to the stable stage (see Chapter 1).
distribution. The panel on the upper left gives the total pop- Matrices and life cycle graphs are nevertheless quite
ulation size for each of the stage classes. The other three pan- useful, but their utility depends on how one interprets
els show the contributions of each pair of eigenvalues/
eigenvectors to the change in population size. By the eighth the results. We can use these models for two very dif-
year, the size change for each class is within 0.01 of λ = 1.12; ferent purposes: to try to predict actual population
only the dominant eigenvalue is making a substantial contri- growth and composition at some point in the future, or
bution at this point. (After Fox and Gurevitch 2000.) to ask what would happen to the population if present
conditions persisted. Only the latter usually makes much
sense.
growth rate, and the eigenvectors are components of the

population’s structure. A very important result is that
one eigenvalue is always larger than the others for this
type of matrix. As a result, over time, the population Small juveniles
Large juveniles
growth rate approaches this value (which is therefore Adults
called the stable growth rate; Figure 7.8), and the pop-
ulation structure approaches the associated eigenvector 0.6 Site A Site B Site C
Frequency in population
(the stable stage distribution; Figure 7.9). The eigenval-

ues are often ranked by their size, so that λ1 usually 0.4
refers to the largest (dominant or leading) eigenvalue
and λN to the smallest. When authors discuss λ with-
0.2
out any subscripts, they usually mean the dominant
eigenvalue.
Even when the population is not yet near its stable 0.0
distribution, its growth can be predicted using the eigen- Figure 7.9
values and eigenvectors (Caswell 2001). The popula- Stable stage distribution for the cactus Coryphantha robbinso-
tion’s size at any time in the future can be written as a rum at three sites. Sites A and B (on hillsides) have similar
weighted sum of the products of the eigenvalues and stable population structures. The stable structure at site C
eigenvectors: (on the hilltop) has many more large juveniles. At this site,
33% of small juveniles become large juveniles, as compared
n with only 1–2% at the other two sites. This is also the princi-
n(t) = c1λt1x1 + c2 λt2 x 2 + L = ∑ ci λti xi pal factor causing λ to be much larger at site C than at the
other two sites (1.12 at site C; just below 1).
i =1
128 Chapter 7
It follows from this that matrix approaches are helping over the whole population, a newborn individual
ful in studying the year-to-year variation experienced can be expected to have lx Fx offspring at age x, and lx+1
by plant populations. For example, in their study of the Fx+1 at age x + 1, and so on. Over its entire life, an indi-
annual Collinsia verna, Kalisz and McPeek (1992) found vidual plant can be expected to have R0 offspring:
that in one year the population grew substantially, with ∞
λ estimated as 1.80, but in the next year it declined, with
λ estimated as 0.41. In another example, hurricanes were
R0 = l1F1 + l 2F2 +K+ l x Fx +K = ∑ l i Fi
i =1 (7.6)
known to cause large-scale variation in the population
growth rates of Pinus palustris (longleaf pine, Pinaceae), R0 is called the net reproductive rate. In general, it is not
but matrix models showed that there is substantial vari- equal to λ, unless the population is at equilibrium (has
ation in growth rates among even “normal,” non-hur- reached a stable stage or age distribution, with λ = 1).
ricane years (Platt and Rathbun 1993; Platt et al. 1988). R0 is obviously a useful quantity to know. Because
Finally, concentrating on the dominant eigenvalue it is based on age, it was not possible to estimate R0 for
and eigenvector does not mean that one is studying stage-structured populations until recent advances in
things that would be important only in the long run, and the analysis of stage-structured matrices (see “Age and
only in a constant environment. As noted above, Stage Revisited,” p. 132).
short-term population growth can be analyzed as a
weighted sum of the eigenvalues and eigenvectors of a Reproductive Value: The Contribution
matrix. The dominant eigenvalues and eigenvectors are of Each Stage to Population Growth
part of this sum, so in studying them, one is always Individuals of different stages do not make equivalent
studying a major component of short-term population contributions to future population growth. In the
growth (Caswell 2001). Many plant population matrices Coryphantha populations described in Box 7A, for exam-
seem to approach the stable stage distribution fairly rap- ple, most small juveniles died, while most adults sur-
idly—usually within five or ten years (see Figure 7.8). vived for a long time. If the manager of a preserve want-
This can occur only if the dominant eigenvalue is much ed to establish a new population, it might be better to
larger than the others, and therefore it is a very impor- introduce n adult plants than n seeds or small juveniles
tant component of short-term population growth. (if enough adults were available, and if they could sur-
Thus, one need not know the entire history of a pop- vive transplanting well enough). But since seeds are gen-
ulation to understand the reasons for its stage structure. erally much easier to work with, one might be tempted
The stage structure will be the result of the average sur- to introduce them instead.
vivorship and fecundity rates even when the matrices Instead of guessing how many individual seeds or
vary randomly from year to year, as they might due to adults to introduce, it would be helpful to have a way of
environmental variation (Tuljapurkar 1990). measuring the effects of different kinds of individuals
on future population growth. The reproductive value of
Lifetime Reproduction: The Net Reproductive Rate the different stages or ages gives us precisely that infor-
How many offspring does an average individual pro- mation. The reproductive value of stage x is the contri-
duce in its lifetime? For age-structured populations, this bution an average individual now in stage x will make
quantity is easily estimated from the data in a transition to the next generation before it dies. For simplicity, we
matrix, using some simple mathematics. Call lx the prob- first discuss reproductive value for age-structured pop-
ability of surviving from the first census to the xth cen- ulations, then extend the concept to stage-structured
sus. The calculation of lx amounts to multiplying all the populations.
individual survival probabilities up to that point. For It might seem at first that reproductive value could
example, the probability of a newborn reaching age 3, l3, be estimated by summing the quantities in R0 over a
is the probability of its reaching age 2, P21, times the shorter interval—for example, for an individual of age
probability of an individual age 2 reaching age 3, P32. It x, one might start the sum at x instead of 1. However,
follows from this definition that l1 = 1. In an age-struc- this does not provide the information we need. Sum-
tured population, lx cannot increase over time: if 1000 ming the items in Equation 7.6 from age x on will always
individuals are censused starting after their birth, at each give a smaller number than summing over an entire life-
subsequent census the number surviving must be the time. Summing from age x on provides only the num-
same as at the prior census, or smaller. ber of offspring a newborn can be expected to have from
We can use Fx to represent the fecundity of an average x on, instead of the number expected over its whole
age individual in age class x.* This means that, averag- life. Moreover, this sum over the shorter period does not
account for the fact that the population will have
*There is some confusion in ecology texts over the notation to use for fecun- changed in size by the time a newborn reaches age x. Its
dity; you may have seen fecundity represented as mx. That symbol is best
reserved for continuous-time models. In this book we use the notation sug- future reproduction will have a different effect on the
gested in Caswell 2001. population than its present reproduction, because a new-
born that begins its life sometime in the future will be

part of a larger (or smaller) population. Calculating the Table 7.1 Reproductive values (vx) of stage classes
of Coryphantha robbinsorum at three sites
effect of individuals of age x on future population
growth therefore requires accounting for the chance of Small Large
surviving to age x, as well as for the change in popula- Site juveniles juveniles Adults
tion size in that interval. A 1 17.82 19.26
This leads us to a second, and more precise, verbal B 1 39.72 45.22
definition of reproductive value (R.V.): C 1 2.06 3.44
Note: These reproductive values are calculated as the dominant left
proportion of future births eigenvectors of the transition matrices in Box 7A, standardized so that
in the population small juveniles have v1 = 1.
to individuals now age x (7.7)
R. V. = proportion of the population now age x
These ideas apply to stage-structured populations as

Both the numerator and denominator of this expression well. We can think of the reproductive value of a tree that
are related to the sum in Equation 7.6, but they also take is 5 m tall, or of a dormant seed in the seed bank. But a
into account the rate at which the population is grow- moment’s thought should suggest a problem: it is diffi-
ing. To see how this works, consider a population with cult to calculate the numerator in Equation 7.7 for a stage-
n individuals, growing at the rate λ. A single offspring structured model. For example, our 5m tree may grow
born to a parent of age x now constitutes 1/n of the pop- continually, stay 5 m tall for years, or shrink for a time.
ulation. A single offspring born in the next interval will There is an alternative way of calculating repro-
constitute 1/(λ n) of the population, one born two ductive value that gets around this problem. The repro-
time-steps later will constitute 1/(λ2 n) of the popula- ductive values for each age or stage can be calculated as
tion, and so on. In a growing population, offspring born the dominant left eigenvector of the matrix model. A
later in life will contribute less to the next generation left eigenvector y of the matrix A is defined so that y A
than offspring born earlier. The reverse is true in a = l y. This definition is parallel to that for the (right)
declining population. eigenvector that we used above in discussing the stable
Thus, to calculate the numerator in Equation 7.7 (the stage distribution. (When biologists refer to an eigen-
proportion of future births to individuals now age x), we vector of a matrix without any further qualification, they
need to sum the expected future reproduction at each usually mean a right eigenvector.)
age (from x to death) as in Equation 7.6, but we also need There are some important differences between
to divide each element liFi by λi: age-structured and stage-structured populations. In
∞
most age-structured populations, reproductive value
lF
numerator = ∑ iλii (7.8)
at birth is very low (because reproduction is delayed for
some time, and many newborns never reach maturity).
i=x
Reproductive value increases to a maximum near the
The denominator in Equation 7.7—the proportion of the age of maturity, and then decreases. In stage-structured
present population that is age x—can be found using populations, in which individuals can remain in the last
similar reasoning. Individuals now age x were first stage for a long time—as in many perennial plant pop-
recorded x – 1 censuses ago, and lx of the original cohort ulations—a decrease may not occur.
is still alive now. Because the population has grown by The Coryphantha data make this clear. The repro-
a factor of λx–1 since then, we have ductive values of each stage at each site are given in
lx Table 7.1. At site A, a plant that makes it to the adult
denominator = x−1 stage contributes roughly 19 times as much, on average,
λ (7.9)
to the next generation as a small juvenile does on aver-
Putting the numerator and denominator together, the age, but only slightly more than an average large juve-
final expression for the reproductive value (the expect- nile. At all three sites, the reproductive value of an adult
ed contribution to future population growth) of an indi- is much larger than that of a small juvenile, but only a
vidual of age x (Goodman 1982) is bit larger than that of a large juvenile. This is because rel-
∞ atively few small juveniles survive to reproduce. Repro-
x−1
v x = λl
lF
x
∑ iλ ii ductive value does not decline in the adult stage because
adults can survive for indefinite periods, as expected
i=x (7.10)
in stage-structured plant populations in which there is
Thus, all of the information we need to calculate repro- no senescence. This information is precisely what would
ductive value is contained in a transition matrix or a life be needed if a manager were planning to introduce
cycle graph. plants to augment this population. For example, if trans-
130 Chapter 7
BOX 7C
Partial Derivatives
If you have forgotten (or never knew) For example, if f = ax + by2 + cxy, then momentarily acting as though y is con-
what a partial derivative is, think of it ∂f stant). Using this logic, you should see
as being like an ordinary derivative = a + cy that the partial derivative of f with
(that is, a rate of change over a very ∂x respect to y is
small change in something else, such as In other words, f changes with respect ∂f
time), but one in which we temporari- to x by a (because a multiplies x in the = 2by + cx
ly hold everything else (all other vari- first term in the sum) plus cy (because ∂y
ables) constant. cy multiplies x in the third term; we are
planting nursery-bred plants is practical, the data sug- matrix element in the ith row and jth column. The sen-
gest that it is nearly as effective to introduce large juve- sitivity of λ to changes in aij is
niles as adults, allowing one to save time, money, and
effort in growing the plants. ∂λ = vixj
∂a ij N
Sensitivity ∑ x kv k (7.11)
How does λ change as the individual matrix elements k =1
change? We already have all the information we need to where the x’s are the elements of the dominant eigen-
answer this question, in the form of the eigenvectors of vector (so that xj is the jth element of the dominant eigen-
A and the reproductive values of each stage. Call aij the vector), the v’s are the reproductive values (so that vi is
the reproductive value of stage i),
and ∂ denotes a partial derivative
(see Box 7C). In other words, a
Table 7.2 Sensitivities of Coryphantha robbinsorum long-term population growth small change in aij (the rate at which
rates (λ) to changes in the matrix elements stage j individuals generate stage i
individuals) causes a change in the
Site Sensitivities
long-term growth rate. The magni-
A Small juveniles Large juveniles Adults tude of this change is proportional
Small juveniles 0.069 0.009 0.040
to the relative importance of stage
j in the stable distribution (xj) times
Large juveniles 1.239 0.152 0.720 the reproductive value (vi) of stage
Adults 1.339 0.164 0.778 i individuals.
The sensitivities of λ to changes
in the matrix elements in the three
B Small juveniles Large juveniles Adults Coryphantha populations are shown
Small juveniles 0.023 0.001 0.021 in Table 7.2. The element in the ith
row and jth column represents the
Large juveniles 0.912 0.043 0.820
rate at which λ changes as that ele-
Adults 1.038 0.049 0.934 ment in the transition matrix A
changes, with all other matrix ele-
Small juveniles Large juveniles Adults ments held constant. The sensitivi-
C
ties at sites A and B were quite sim-
Small juveniles 0.153 0.100 0.186 ilar. In both cases, the largest change
Large juveniles 0.314 0.205 0.383 in λ is expected if there are increas-
es in small juveniles’ chances of
Adults 0.526 0.344 0.642
surviving to either of the larger
classes, and in the survivorship of
Note: Sensitivities and λ are calculated from the transition matrices in Box 7A. Each entry shows the adult plants. In contrast, increasing
change in λ resulting from a unit change in the corresponding element of the transition matrix, while all
other matrix elements are held constant. For example, a unit increase in P11 increases λ at a rate of 0.069 the fecundity term—the rate at
at site A. which adult plants generate new
small juveniles—has little effect on λ in either popula- er terms. A related problem with sensitivities is that
tion. Site C (where the population is clearly growing) is matrix elements may be zero for some basic biological
a bit different: increasing any of the terms has a marked reason, or by definition. In the Coryphantha example, it
effect on λ, although the largest effects are still be is impossible for juveniles to reproduce (by definition)—
achieved by increasing survival rather than fecundity. but the sensitivity analysis still tells us that λ would be
increased by a certain amount if that transition were
Elasticity increased from zero.
There is a potential problem with the use of sensitivities: The constraints of sensitivity analyses can be
they compare things that are often measured on very dif- addressed by using the elasticity, the proportional
ferent scales. For example, the survival terms in a matrix change in λ caused by a proportional change in a matrix
model must be between 0 and 1, but the fecundity terms element. The elasticity of λ with respect to changes in
can sometimes be orders of magnitude larger. Similarly, the ijth matrix element is (Caswell 2001)
the survival terms for small plants are often much small- aij ∂λ
er than those for larger plants. When this is true, a small ∂ ij = ∂ ln(λ) =
∂ ln(aij ) λ ∂a ij (7.12)
increase of, say, 0.01 actually represents a much larger
proportional change in the smaller terms than in the larg- In other words, the elasticities are the sensitivities times
aij/λ. Thus, when a matrix element is zero, its elasticity
will also be zero.
The elasticities for the Coryphantha data are given in
Figure 7.10. At sites A and B, the largest proportional
effects on λ would be achieved by increasing adult sur-
vival. All other changes would have very small effects
Figure 7.10
Elasticities for the three populations of the endangered
cactus, Coryphantha robbinsorum. Each bar gives the elastic-
ity for a particular transition corresponding to the matri-
ces in Box 7A. The elasticities describe the proportional
change in the long-term growth rate (λ) resulting from a
proportional change in each matrix element, holding all
other matrix elements constant. They sum to 1 for any
given analysis, and can therefore be thought of as describ-
ing the proportional importance of each matrix element
for λ. For example, at site A, adult-adult survival (P33)
accounts for nearly 80% of λ. At site C, where the popula-
tion is expected to grow rapidly (λ = 1.12), adult-adult sur-
vival makes only about a 60% contribution to λ.
132 Chapter 7
on λ. At site C, λ would still be most strongly affected by newborn plant to be a small juvenile for one year, a large
increasing adult survival. Increasing the other terms juvenile for one year, and then mature to become an
would have smaller, though perhaps not negligible, adult. The matrix model tells us that the proportion
effects. In any case, it seems clear that at all three sites P21P32 of newborns following this path become adults
an effective plan to protect the population would involve after two years. Another path is for individuals to spend
protecting the established plants—especially adults— two years at each stage before making the transition to
rather than enhancing their fertility. This pattern is com- the next stage; the model tells us that the proportion
mon for organisms with long-lived adults. P11P21P22P32 of newborns follow this path to maturity.
Elasticities have an additional property that makes Cochran and Ellner (1992) developed methods for sum-
them useful: all the elasticities of a matrix sum to 1. Con- ming over all the possible paths and thereby estimat-
sequently, they can be interpreted as the relative con- ing age-based information from a useful-based model.
tribution to λ of the corresponding matrix element Table 7.3 shows some useful age-based data for
(given that all other elements stay constant). This means Coryphantha; estimated survivorship curves for the
that one can directly compare the elasticities of differ- species are shown in Figure 7.11. It is straightforward to
ent matrices based on the same life cycle graph. For calculate these and other quantities. Because the calcu-
example, at Coryphantha site C, adult survival was relations involve many summations over all possible path-
sponsible for about 55% of λ. By contrast, at the sites ways, the mathematical expressions are complex in
where the population is shrinking or just holding its appearance, and we refer interested readers to Cochran
own, adult survival accounted for a much greater pro- and Ellner (1992).
portion of the population growth rates: 76% at site A, This approach also makes it possible to calculate the
and 92% at site B. life table for a stage-structured population. A life table
is a list of the estimated mortality, cumulative survival,
Age and Stage Revisited and related rates for a cohort (a group that germinated,
Saying, as we have, that age alone is usually a poor pre- reached a particular size, or entered a study at about the
dictor of demography in plant populations does not same time). In age-based populations, the survival
imply that age is unimportant. However, apart from
cases in which we know that age plays a direct and
important role in plant demography (e.g., see Figure 7.4),
it has been difficult to relate stage to age. Nevertheless,
it is sometimes important to do so. For example, in man-
aging an endangered population, there are several
important questions we might need to ask: How long
does it take, on average, for a seed to reach maturity?
How old is the average individual in a particular stage
class? What is an individual’s chance of surviving x
years? What is the net reproductive rate of a population?
How long will it take the population to replace itself?
Until recently, questions such as these could not gen-
erally be answered. In some groundbreaking work,
Cochran and Ellner (1992) realized that these questions
have answers that are, at least in principle, simple. A
stage-based matrix model implicitly includes informa-
tion about age because it predicts stage transitions one
time-step into the future. Consider the Coryphantha
example: A small juvenile has a chance P11 of surviving
until next year as a small juvenile, and a chance of P21 of
growing into the large juvenile class and surviving until
next year. Its chance of being alive next year is there-
fore the sum P11 + P21. Figure 7.11
By generalizing this approach—summing over all Estimated survivorship curves for Coryphantha robbinsorum
possible paths between two stages—it turns out that one at three sites, calculated using the methods of Cochran and
can calculate any age-based quantity from stage-based Ellner (1992) to analyze the matrix models in Box 7A. Sur-
vival is much greater at site C than the other two sites,
data, including R0. Even in the relatively simple because 76% of small juveniles (P11 + P12 = 0.76) survive
Coryphantha life cycle graph (Figure 7.7), there are many there; the comparable figures are 69% and 50% at sites A and
possible paths from birth to maturity. One path is for a B, respectively.
Table 7.3 Some age-based quantities calculated from the stage-based Finally, some size-structured
Coryphantha matricesa plant populations may be better
modeled with a method that treats
Probability Average age Net
Site of reaching maturity reproductive Generation size as a continuous variable,
Site position maturity (years) rate (R0) timeb rather than with a matrix approach.
A Northeast 0.05 10.7 0.92 42 Michael Easterling et al. (2000)
exposure have suggested a general approach
B Southwest 0.02 6.7 0.81 72 that is analogous to the matrix
exposure approach, but treats size as a con-
C Hilltop 0.46 5.3 5.84 27 tinuous variable. The mathemat-
a
Quantities calculated using the methods of Cochran and Ellner (1992). ics involved are a bit more compli-
b
Mean age at which a cohort of newborns produces offspring. cated than with matrix approaches
because the method uses integrals,
but these can be estimated numer-
information in a transition matrix is the same as the ically on a computer. Easterling et al. (2000) argue that
information in a life table. But it is not obvious how one their approach is more natural than the matrix approach;
could use the survival data in a stage-based matrix to an added bonus, they suggest, is that one can often use
predict a quantity such as the fraction of newly germi- their method with fewer parameters than are typically
nating plants that will still be alive in 10 years. Fortu- used in a matrix model. There have been few studies to
nately, Cochran and Ellner (1992) have provided for- date using this method, so all of its strengths and draw-
mulas that allow one to estimate the life table for backs are not yet clear, but it is certainly worthy of fur-
stage-structured populations. ther attention.
Other Approaches to Modeling Plant Demography

Demographic Studies of Long-Lived Plants
The matrix approach can be much more sophisticated
than the basics we have outlined here. For example, A small number of plant species reach very great ages.
matrix models can be arranged so that they include both Some Larrea tridentata (creosote bush, Zygophyllaceae)
environmental and demographic stochasticity (Caswell genets in the Mojave Desert are estimated to be about
2001). They can also be made to be density-dependent 14,000 years old. Obviously it is difficult to study the
(Caswell 2001). Doing so requires more advanced math- demography of populations when some individuals
ematics than we will use in this book, and the analysis have been alive for nearly as long as humans have occu-
no longer depends simply on the eigenvalues and eigen- pied North America! Fortunately, most plants do not
vectors, but it is important to know that this method has present quite such difficulties for demographic study.
much broader applications than those we have so far dis- More typically, trees and shrubs live for a few decades
cussed. to a few centuries. But even these life spans are diffi-
There are also situations in which matrix models are cult to study when the average researcher is active for
not the best choice. For example, if the spatial arrange- only a few decades.
ment of individuals proves to be important in popula- Considerable insight into the demography of long-
tion dynamics, a matrix model can be misleading. In such lived plants can be gained by following marked indi-
cases, models that take spatial location into account to viduals. William Platt and his associates (1988, 1993)
simulate the birth, growth, and death of individuals and have been studying populations of Pinus palustris (lon-
their particular interactions with their neighbors can be gleaf pine, Pinaceae) since the 1970s. Year-to-year varia-
useful. There is, of course, a catch: to estimate the param- tion in environmental conditions—sometimes due to
eters of such spatially explicit individual-based models large disturbances such as hurricanes—has been a hall-
accurately, one needs substantially more data than for a mark of these studies, and it appears to play a major role
matrix model. A series of models developed by Stephen in the dynamics of these populations. Similarly, Robert
Pacala and associates (Pacala and Silander 1985; Pacala Peet and Norman Christensen (1987) have studied large
1986a, 1986b, 1987) illustrates both the strengths of this plots of trees established in the early twentieth century
approach (the authors were able to successfully predict in the Duke Forest at Duke University. The results of
many features of the dynamics of their study population) these studies suggest that it is often reasonable to study
and some of its difficulties and limitations (large quan- long-lived plants using such methods. Clearly, such
tities of data were needed to estimate model parameters, studies can have some weaknesses. Measurements of
sophisticated computer programming was necessary to survival are often taken at longer intervals than one
study the models, and analyses such as sensitivity and might like—such as 10 years, or even multiples of 10
elasticity would be difficult to perform). years. Seeds, seedlings, and small saplings are usually
134 Chapter 7
not measured. Because of the long intervals between by counting their rings (in adults) or terminal bud scars
censuses, the variation between years is generally under- (in seedlings and saplings). Then they used the observed
sampled. But the information obtained even with such relationship between age and size to estimate the pop-
poor resolution is still useful. ulation age structure. Hett and Loucks first considered
One approach to studying long-lived plants, used two models for these data: one that assumed the same
widely in forestry and forest ecology, has been to devel- rate of mortality for all ages, and one that assumed
op static or vertical life tables. This method assumes decreasing mortality rates with age. The fit of these
that the population has a stable age distribution, that the models to the data is shown in Figure 7.12. Noting that
population’s transition matrix has been roughly constant the data systematically depart from both regression lines,
for a long period, and that λ = 1. It then asks what pat- Hett and Loucks then fit a more complex model to the
tern of survival and reproduction would have generat- data, in which the pattern of decreasing mortality with
ed a population with the (present) observed structure. age was overlain by a periodic function. The latter model
Joan Hett and Orie Loucks (1976) used this approach appeared to provide the best fit, and Hett and Loucks
to infer survival curves for several populations of Tsuga proposed several hypotheses that might explain the
canadensis (eastern hemlock, Pinaceae) and Abies balsamea apparent oscillation, including population cycling and
(balsam fir, Pinaceae) in Michigan. First, they measured periodic outbreaks of pests.
tree sizes and determined the ages of a sample of trees As this example illustrates, the use of static life tables
requires strong assumptions that cannot usually be jus-
tified. Static life tables require the assumption of con-
(A) stant vital rates, which is unlikely over long periods.
15 Many trees in the hemlock study lived through part of
the Little Ice Age—a period of low solar activity, cold
Number of individuals (x 1012)
temperatures, and glacial advances from 1500 to 1850—

4
as well as relatively warm interludes in 1540–1590 and
1770–1800. But is this not a criticism that would apply
3 equally well to matrix models? A principal use of matrix
models is to study present conditions by asking what
2 their consequences would be if they remained constant.
By contrast, the static life table approach asks what set
Tsuga canadensis of constant conditions would have generated the cur-
1
rent population structure. As Jonathan Silvertown (1987)
points out, many quite different models, assuming con-
sd sp 1 2 3 4 5 6 7 8 9 10 stant or varying conditions, can be fitted to data of this
Size class kind equally well. The consequences of constant condi-
tions can be calculated, but there
(B) (C) is no unique answer to the ques-
8 8 tion asked by static life table
loge y = 2.9927 – 0.0099 x loge y = 5.9777 – 0.9327 loge x approaches.
6 r = 0.61 r = 0.73
This does not mean that to
loge y
6 study a long-lived plant one must

4
begin by marking vast numbers of
loge y
2
4 individuals and hoping for a long
career, along with plentiful
0 40 80 120 160 200 240 280 320 360
descendants to continue the study.
Age (years) 2
There are reasonable inferences
that can be drawn from a popu-
(D) lation’s structure without invok-
8 0 2 4 6
loge x, age (years)
6 T = 126 yr
Figure 7.12
loge y
4 (A) Estimated age structure of a Tsuga canadensis (eastern hemlock,

Pinaceae) population in Michigan. (B–C) Three models attempting to
2 explain the data with static life table approaches. (B) Model assuming equal
mortality rates for all ages. (C) Model of declining mortality with age. (D)
0 40 80 120 160 200 240 280 320 360 Model of declining mortality with age, overlain by a periodic function with
Age (years) period T. (After Hett and Loucks 1976.)
ing the strong assumptions of vertical life tables. If one ment opportunities is spatially patchy. Many forest trees,
finds that a population consists of widely separated age for example, occur in nearly even-aged stands because
classes (as occurs in many forest trees and in saguaro wildfires create patches for recruitment. By construct-
cacti), for example, it is reasonable to suggest that oppor- ing fire histories—inferred from examining the ages of
tunities for successful recruitment to the population occur fire scars on trees and from dating charcoal in the soil—
only rarely. If there is also a tendency for plants of a cer- investigators have gained insight into the frequency and
tain age class to be located near one another, it is rea- severity of these disturbances. Although methods like
sonable to infer that the process creating these recruit- these do not provide estimates for matrix elements, they
produce strong inferences if well
done.
Several researchers have gained
insight into the demography of
long-lived plant populations by
finding creative ways to obtain data
on survivorship retrospectively.
Deborah Goldberg and Raymond
Turner (1986) first used maps of
plots established by Forrest Shreve
in the 1930s to document survival in
several species of long-lived cacti
and desert shrubs. Perhaps the most
unusual studies have been those in
which James Hastings and Ray-
1903 mond Turner (1965; Turner 1990)
used landmarks to match old pho-
tographs (some dating to the nine-
teenth century) with new ones and
identify surviving desert trees and
large cacti (Figure 7.13).
Figure 7.13
Comparison of repeated photographs
has provided important insight into
the demography of large plants and
vegetation change in arid country.
Pioneered by James Hastings and
Raymond Turner (1965), the tech-
nique requires landmarks for identi-
1961 fying the sites of old photographs.
These repeated photographs show
changes in the population of Pachyc-
ereus pringlei (cardón, Cactaceae) on
Isla Melisas in Guaymas Bay, Sonora,
Mexico. (A) This photograph, taken
in 1903, shows a population of old,
many-branched cardón. (B) In this
photograph, taken in 1961, most of
these old individuals have been
replaced by much younger plants,
although several large individuals
can still be seen along the ridgeline.
(C) In this photograph, taken in 1996,
a dense stand of cardón can be seen;
most of the plants are much larger
than in the 1961 photo. Three large,
old individuals are still apparent
along the ridgeline. (Photographs
1996 courtesy of R. Turner.)
136 Chapter 7
Random Variation
fore, it should not surprising, in a small enough popu-
in Population Growth and Decline lation, if all the individuals happen to die.
In the real world, population growth rates vary because Fortunately, many of the tools described above can
of stochastic (randomly varying) factors. Hurricanes and be modified for use in a variable world. First, however,
fires, for example, play important roles in the growth it is important to be clear about how random variation
of many populations of forest trees, and occasional wet affects population growth.
years may be crucial to the persistence of some popula-
tions of desert plants. Furthermore, small populations Causes of Random Variation
of endangered plants may go extinct largely due to There are two ways in which population growth can be
chance: If individuals have a 50% chance of surviving a affected by stochastic factors. First, in the case of envi-
winter, for example, it is likely that the actual fraction ronmental stochasticity, vital rates vary due to environ-
surviving will be different from half—just as repeated mental factors that affect all the individuals in a stage class
coin tosses do not usually result in 50% heads. There- (or in a population) in roughly the same way. Desert envi-
Figure 7.14
The effect of environmental sto-
chasticity on the realized fecundi-
ty of Sonoran Desert winter
annuals. Increased rainfall due to
the El Niño phenomenon
occurred in 1982–1983, 1986–1987,
1990–1991, and 1991–1992—all
years of high realized fecundity
for most species. All species
except Erodium cicutarium are
native to the study area. (After
Venable and Pake 1999.)
ronments, for example, have highly variable rainfall. In a sider a population of eight plants with a probability P =
10-year study of winter annuals in the Sonoran Desert, 0.25 of an individual surviving until next year. There is a
Lawrence Venable and Catherine Pake (Venable and Pake fair chance that next year’s population will not be two,
1999) documented major variation from year to year in the product of P and n (this year’s population). (For fur-
the realized fecundity (the chance of surviving to matu- ther insight, see the discussion of genetic drift in Chap-
rity times the fecundity of survivors) of ten species of ter 5, with which demographic stochasticity shares many
these plants (Figure 7.14). This variation underlies fluc- properties.)
tuations in the population sizes of these species (Figure Some important features of demographic stochas-
7.15). Interestingly, years in which realized fecundity is ticity are shown in Figure 7.16. Figure 7.16A shows the
high appear to be associated with El Niño events (Ven- observed number of survivors in a hypothetical popu-
able and Pake 1999; see Chapter 18). lation in which individuals have a probability of 0.25
Second, in the case of demographic stochasticity, the of surviving. You might notice that there are more
chance variation in the fates of individuals reduces the extremely low values than extremely high values. There
average long-term growth rate of the population. Con- is a reason for this: Even in a population of 1000 plants,
Figure 7.15
Fluctuations in population
size in ten species of Sonoran
Desert winter annuals. (After
Venable and Pake 1999.)
138 Chapter 7
Long-Term Growth Rates

Imagine a population of annual plants (without a seed
bank) that grows, in good years, at the rate λg = 1.01, and
in bad years, at the rate λb = 0.99. To keep things simple,
imagine that good and bad years alternate. It may sur-
prise you to realize that this population is slowly going
extinct! How can this be, when the average rate of
growth is obviously 1? It is because the long-term
growth rate is not an ordinary arithmetic average, but is
equal to the square root of 1.01 × 0.99, which is less than
1. Why? Population growth is a multiplicative process:
seeds produced in good years then germinate in bad
years, and vice versa. The square root thus gives us the
appropriate average over time.
Given an average rate of population growth, varia-
tion in that rate among years reduces the long-term rate
of population growth. Consider a group of unstructured
populations—such as annual plants without seed
banks—having the possible values of λ that are given in
the second column of Table 7.4, and assume that these
values occur with equal probability. The average value
of λ is the same for populations 1–3, but λ varies more
in population 3 than in population 2, and it does not
vary at all in population 1. The average value of λ is
called the arithmetic mean and is symbolized λ . The
arithmetic mean of n numbers is their sum divided by n
(see the Appendix).
Although populations 1–3 all have the same aver-
age growth rate, their long-term growth rates differ,
because population growth is a multiplicative process.
Next year population 1, whose growth rate does not
vary, will be 1.03 times its current size, and in 8 years it
will be 1.038 = 1.267 times its current size. If the two
different values of λ occur for population 2 with equal
probability, that population will grow at some different
Figure 7.16
The effect of demographic stochasticity on survival in a rate, such as 1.05 × 1.01 × 1.01 × 1.01 × 1.05 × 1.01 × 1.05
hypothetical population with a survival probability of 0.25. × 1.05. In this example, population 2 will be about 1.264
(A) The observed number of survivors given a population times its current size in 8 years—it will have grown less
size of N where survivorship rates were determined by the than population 1.
average survival probability plus a deviation due to random The long-term growth rate of an unstructured pop-
chance. The line shows the expected number of survivors:
0.25 times the population size. (B) The relative size of the ulation is given by the geometric mean of the annual λ’s,
departure from the expected value: (number of survivors – symbolized as a in the fifth column of Table 7.4. (The
expected survivors)/expected survivors. geometric mean of n numbers is the nth root of their
product.) The geometric mean has an important prop-
erty: it is guaranteed never to exceed the arithmetic
it is possible that, by chance, no plants will survive—but mean, and is equal to the arithmetic mean only if the n
it is never possible for more than 1000 to survive. Figure numbers are all the same—in other words, if the vari-
7.16B shows the relative departure from the expected ance of λ is 0 (see the Appendix for a discussion of vari-
number of survivors, which tends to be greater in small ance). For a given average growth rate λ , then, increas-
populations. ing the variance of λ always reduces the long-term
Both forms of stochasticity occur in all populations. growth of the population.
Environmental stochasticity can have substantial effects This discussion might be interpreted to mean that
in populations of any size. As the examples above sug- increasing the variance of λ always reduces a, but this is
gest, demographic stochasticity is important mainly in true only if the average growth rate λ stays constant.
small populations. Population 4 has a larger arithmetic mean and a larger
Table 7.4 Relationship between annual and long-term population growth rates
in a variable environment
Values Arithmetic Variance Geometric
Population of λ mean (λ) (σ 2λ) mean a
1 1.03 1.03 0 1.03
2 1.01, 1.05 1.03 0.0008 1.0298
3 1, 1.01, 1.05, 1.06 1.03 0.0009 1.0297
4 1, 1.01, 1.05, 1.07 1.0325 0.0011 1.0321
5 0.9, 1.01, 1.05, 1.17 1.0325 0.0124 1.0280
Note: The long-term growth rate depends on the mean and variance of yearly rates. In unstructured
populations, the long-term rate is the geometric mean of the yearly rates.
variance in λ, and its long-term growth rate is larger than Few of these studies, however, are of sufficient length
any of populations 1–3. On the other hand, population to provide good estimates of variance, and fewer still pro-
5 has still greater variance in λ, and its long-term growth vide estimates of the covariance among matrix elements.
rate is the smallest of the five populations. Even a population with a > 1 can sometimes go
These examples should help to convince you that extinct due to random variation. A population’s extinc-
both the average conditions and their variation are tion probability is estimated as the fraction of replicate
important in determining long-term rates of population populations that can be expected to go extinct. This
growth. There are few studies on this kind of long-term value is estimated by simulation. Extinction probabili-
variation. There are many anecdotal accounts of plants ties for several California species of Calochortus (mari-
in highly variable environments (such as desert annu- posa lily, Liliaceae) are shown in Figure 7.17. Note that
als) that usually have high mortality and low fertility, extinction probabilities increase as the amount of envi-
but in an occasional “good year” produce vast quanti- ronmental stochasticity increases, and that this increase
ties of seed. Thus it is possible for a plant population to is fastest for populations with the smallest value of λ.
have positive growth in the long run, but be declining An idea that is very closely related to extinction
in numbers during most years (Venable and Pake 1999). probability is the minimum viable population (MVP).
These general ideas hold for age- and stage-struc- The MVP is the minimum size necessary to give a pop-
tured populations as well. To study population growth ulation a probability x of surviving t years. Typically the
in a stochastic environment, we use the stochastic probability is taken as 0.95, and the time as 50 or 100
growth rate a rather than λ1. In structured populations, years. Plans for the management of endangered species
a is the long-term average growth rate of the population. increasingly require estimates of MVP. Eric Menges
Generally, a must be estimated by simulation. Because (1992) conducted simulations to estimate the MVP for
matrix multiplication is not commutative, a is not a sim- the Mexican palm Astrocaryum mexicanum (chocho, Are-
ple geometric mean of the estimate annual growth rates caceae), using published matrices from the extensive
(see Caswell 2001 for a discussion of methods for esti- study of Piñero et al. (1984); the results are shown in Fig-
mating a). Just as the geometric mean of a sequence of ure 7.18. Populations subject to only demographic sto-
numbers is smaller than the arithmetic mean, a is less chasticity could begin with only 50 plants and still have
than λ1 unless there is no stochasticity, in which case the a 95% chance of persisting 100 years. Populations sub-
two quantities are equal. Calculations of the long-term ject to environmental stochasticity required larger initial
growth rate are actually done with logarithms analogous sizes to have this chance of persisting. As environmen-
to r = ln(λ). tal stochasticity became large (i.e., there was more year-
to-year variation), the chance of a population going
Studying Variable Population Growth extinct increased rapidly in smaller populations.
To study the effects of environmental stochasticity, one Most demographic studies of plants have lasted
needs the same kinds of data used to estimate average only a few years, and consequently have focused on
population growth rates, recorded over enough years average values of growth rates λ .These are useful quan-
to estimate the variances and covariances of the matrix
elements—the survivorship and fecundity rates. Given tities, but it is important to realize that λ overestimates
such data, one can perform computer simulations of pop- a (just as the arithmetic mean is always larger than the
ulation growth that is subject to random variation. There geometric mean, unless the variance is zero). There have
are many studies reporting variation in matrix elements been few attempts to estimate a in plant populations.
(e.g., Horvitz and Schemske 1990; Bierzychudek 1982). Hal Caswell (2001) estimated a for two populations of
140 Chapter 7
(A)
Figure 7.17
(A) Calochortus howellii (Mariposa lily, Lili-
aceae), which is endemic to southwestern Ore-
gon, especially serpentine areas. (Photograph
courtesy of J. Sainz.) (B) Extinction probabili-
ties of several California populations of Calo-
chortus spp. Environmental stochasticity was
modeled by taking 1% of the variance/
mean for seed production, and 0.01% of the
variance/mean for other vital rates, and then
multiplying these values by the numbers
shown on the horizontal axis. (After Menges
1992.)
Figure 7.18
Results of a simulation conducted to estimate the
minimum viable population (MVP) size for a
population of the Mexican palm Astrocaryum mex-
icanum (chocho, Arecaceae) under different levels
of environmental stochasticity (ES). Environmen-
tal stochasticity was modeled using the methods
described in Figure 7.16. MVP is the smallest
number giving an extinction probability of 5% or
less in 100 years. (After Menges 1992.)
Arisaema triphyllum (jack-in-the-pulpit, Araceae) in New Matrix models are useful tools for studying change
York state studied by Paulette Bierzychudek (1982). In in population size and composition. They make it pos-
this study, a and its 95% confidence intervals were 1.2926 sible to estimate the population’s eventual growth rate
± 0.0025 in the Fall Creek population, and 0.8979 ± 0.0028 and composition—an important way of examining the
in the Brooktondale population. consequences of current demographic conditions.
Reproductive value, which can be calculated from a tran-
sition matrix, is useful in comparing the importance of
Summary
different stage classes for future population growth. It is
Studies of change in plant populations require defining now possible to calculate age-based quantities (such as
an “individual.” In many plants, unlike most animals, the average age of first reproduction) from stage-based
genetic and physiological individuals are different data.
because many genetic individuals can reproduce vege- Plant populations vary in their growth rates. One
tatively as well as sexually. The “right” kind of individ- cause of this variation is random variation in the envi-
ual to study depends on the questions being asked. ronment, which affects all plant populations. Variation
Survival and reproductive rates in plants usually in population growth rates is also caused by random
depend much more on stages (size, life history stage, events involving individuals. This kind of random vari-
physiological status) than on ages. An important goal of ation is most important in small populations. General-
plant population studies is identifying the factors caus- izations of matrix modeling approaches make it possi-
ing variation in survival and reproductive rates. Most ble to estimate long-term population growth rates,
population models use stages to study changes in pop- extinction probabilities, and minimum viable popula-
ulation size. tion size.
Additional Readings
Classic References Kalisz, S. and M. A. McPeek. 1992. Demography of an age-structured
annual: Resampled projection matrices, elasticity analyses, and seed bank
Harper, J. L. 1977. Population Ecology of Plants. Academic Press, Lon-
effects. Ecology 73: 1082–1093.
don.
Hartshorn, G. S. 1975. A matrix model of tree population dynamics. In
Tropical Ecological Systems, F. B. Golley and E. Medin (eds.), pp. 41–51. Additional Resources
Springer-Verlag, New York. Caswell, H. 2001. Matrix Population Models. 2nd ed. Sinauer Associates,
Sunderland, MA.
Contemporary Research Cochran, M. E. and S. Ellner. 1992. Simple methods for calculating
age-based life history parameters for stage-structured populations. Ecol.
Horvitz, C. C. and D. W. Schemske. 1990. Spatiotemporal variation in
Monogr. 62: 345–364.
demographic transitions for a tropical understory herb: Projection matrix
analysis. Ecol. Monogr. 65: 155–192. Easterling, M. R., S. P. Ellner and P. M. Dixon. 2000. Size-specific sensi-
tivity: Applying a new structured population model. Ecology 81: 694–708.
Platt, W. J. and S. L. Rathbun. 1993. Dynamics of an old-growth longleaf
pine population. In The Longleaf Pine Ecosystem: Ecology, Restoration and
Management, S. M. Hermann (ed.), pp. 275–297. Proceedings of the Tall
Timbers Fire Ecology Conference, No. 18.
C H A P T E R
8 Growth and Reproduction

of Individuals
T
he growth and reproduction of individuals are critical components of
the growth (or decline) of populations. Growth and reproduction are
also key ways in which plants interact with other plants, animals, and
the physical environment around them through uptake of nutrients, use of
space, pollination relationships, and the dispersal of fruits and seeds. The
movements of pollen, seeds, and fruits are the main factors responsible for ini-
tiating spatial patterns in plant populations. The evolution of specific repro-
ductive traits marks key episodes in the history of terrestrial plants. We begin
this chapter at the within-plant level of anatomy and physiology, but we then
make the transition to considering growth and reproduction at the level of the
whole plant and of plant populations.
Plant Growth
The growth of plants is modular; that is, plants grow by adding repeated units,
or modules, to their bodies. We can compare growing plants to children’s con-
struction toys in which complex structures are built up by adding simple
repeated units to the original structure. As with such toys, there are only a few
types of repeated units in modular plant growth; all plant structures are iter-
ations of these units.
The basic rules of plant growth have important consequences for the ecol-
ogy of plants. In particular, the ability to add (or lose) individual modules
means that plants have a great deal of plasticity in their size and shape. This
plasticity affects their ability to respond to damage, to capture resources, and
to interact with one another. Many plants can survive the loss of large portions
of their bodies to herbivores. Commonplace as this may seem, it is remarkable:
few animals could survive a similar loss of body parts! Similarly, many plants
regularly lose parts of the shoot—the aboveground part of a plant—to abiot-
ic factors such as freezing temperatures, drought, and fire—and nevertheless
survive.
The addition of new modules depends on the activity of meristems (meris-
tematic tissues). Meristems are collections of undifferentiated cells (cells that
do not yet have specialized functions). At the tip of a growing stem or branch
is an apical meristem (Figure 8.1). The growing stem is created by the pro-
duction of new cells behind the apical meristem. Eventually, these new cells
become differentiated into the many different kinds of cells making up the
144 Chapter 8
Figure 8.1
The basic structures involved in Shoot apical meristem
plant growth. The shoot apical
meristem differentiates, generat-
ing lateral organs such as leaves,
Leaf
as well as axillary meristems. Apical
primordia
The points at which these lateral bud
organs or buds are attached are
Lateral
called nodes. The leaf primordia buds with
expand to become leaves. (After
Purves et al. 2001; photographs Axillary bud
meristem
Internode
fpo axillary
meristems
by J. R. Waaland/Biological
Photo Service.)
Node
Vascular
Cork cambium
cambium
Lateral
root
Meristem
that will form
lateral root
Root
Root hairs apical
meristem
Root
cap
shoot, such as epidermal, xylem, phloem, and paren-

chyma cells. Certain cells within the apical meristem
divide, periodically generating a node—a point of pro-
liferation for the development of leaves or flowers. tion of cells in the axillary meristems; this phenome-
Nodes, therefore, are the places where leaves and flow- non is called apical dominance. If the apical meristem
ers are attached to the stem. In many plants, the stem is is removed—for instance, when an herbivore chomps
visibly swollen at each node, and a primordium (bud) off the tip of the plant—the axillary meristems are freed
forms there for each potential new leaf or flower. from apical dominance, and many of them begin to cre-
Internodes—the stem segments between nodes— ate new branches below the damage. This basic modu-
are created in a separate step. When the apical meristem lar growth response to damage is a major way in which
generates a node, it may create a group of cells called plants recover from herbivory and continue to grow,
intercalary meristems. The internodes grow when these although now in a lower, bushier form.
intercalary meristems are activated. Some plants have A young plant or new shoot begins to increase in
short or even unnoticeable internodes. These plants, size by the process described above, called primary
called rosette plants, include cabbage, dandelions, many growth. All perennial gymnosperms and angiosperm
alpine plants, and many low-growing weedy herba- dicots (and many annuals in these groups) continue their
ceous plants. In many of these species, the onset of flow- growth by another process, called secondary growth.
ering activates the intercalary meristems, and the plant Secondary growth is largely a process of increasing girth
“bolts,” or lengthens due to the expansion of the inter- (thickness) by producing woody tissues (secondary
nodes. They have little in the way of intercalary meris- xylem). In a new stem, the vascular bundles are sepa-
tems, or never activate them. rated from one another, but as the stem matures, the vas-
At each node there is also an axillary meristem in cular bundles grow together and coalesce to form a
the leaf axil, the place where the leaf and stem join. sheath surrounding the inner part of the stem. In the cen-
When the axillary meristems grow, they may become ter of this vascular tissue, between the xylem and the
branches or flowers (which are actually special types of phloem, a new meristem, the vascular cambium, is
branches). The axillary meristems are generally inactive formed just inside the stem, under the phloem and epi-
while the apical meristem is intact because the apical dermis. The vascular cambium forms a thin sheath that
meristem produces hormones that inhibit the prolifera- extends for much of the length of the stem. A similar
Growth and Reproduction of Individuals 145
process occurs in roots. Plants growing under poor resource conditions often
Secondary xylem, or wood, is typically much have a different shape than plants of the same species
stronger and better protected than primary xylem and growing in richer microhabitats. In some species, this
is a major source of the structural strength that enables variation reflects evolutionary adaptation to variation in
perennial plants to survive for more than a short time. the environment; in other species, however, the fact that
In monocots, increase in girth is more complicated. plants have different shapes under poor conditions is a
Palms, for example, have no vascular cambium, but simple reflection of the stress induced by those condi-
early in life they add numerous adventitious roots (roots tions. Not all plasticity is adaptive.
that originate from the stem), leading to an increase in Some forms of plasticity are adaptations that enable
girth. Once this process stops, the width of the palm is individuals to acquire more resources, disperse, or
set. That is why palms stay about the same girth over improve their competitive ability. One of the best-stud-
their entire height. ied examples of such plasticity is tree architecture. In
Root apical meristems control root growth. The api- many forest trees, shaded individuals have a character-
cal meristem of a growing root is located behind, and istic shape that is different from that of the same species
protected by, a cap of loosely held cells that sloughs off when it reaches the forest canopy, or when it grows in
as the growing tip of the root pushes through the abra- a gap or along an edge. Many species have patterns of
sive soil. Roots do not have systems of nodes, intern- branching and growth that vary in response to light
odes, or most of the different meristems characteristic of availability, and light interception is often a key factor
shoots. Roots branch when special differentiated cells affecting plant shape. Forest trees typically shed lower
resume active division, and each branch forms its own branches that are shaded and fall below the light com-
root apical meristem. One important feature of roots is pensation point (see Chapter 2). In some cases, the same
their very fine root hairs. Most uptake of water and min- species retain their lower branches when grown in the
erals occurs through these hairs (or through mycorrhizal open because those branches are above the light com-
structures; see Chapter 4), but the hairs usually function pensation point under those circumstances. Most species
for only a few days. Thus, continued growth of roots is have a characteristic pattern of self-pruning—shedding
often important for uptake of water and nutrients. of branches—that tends to keep branches above the light
compensation point.
Self-pruning, however, is not necessarily just a pas-
Ecology of Growth sive response to shading; it can also be an important part
Plant Architecture and Light Interception

Plants have a great variety of architectures
(arrangements of regenerating parts) and
shapes. The growth forms of plants can reflect
FPO
their environments in at least three ways.
First, plant growth form is in part a result of
evolutionary adaptations to the timing of
unfavorable seasons. A widely used classifi- Buds
cation of perennial growth forms is shown in Buds
Figure 8.2. Each growth form has different
consequences for survival and resource reten-
tion. An excellent example of the evolution of
a diversity of growth forms is the morning
glory genus, Ipomoea (Convolvulaceae). Some
Mexican and Central American species are Perennating
organ
drought-deciduous trees (i.e., phanero-
phytes), while others are vines with succulent Phanerophyte Chamaephytes Hemicryptophyte Cryptophytes
root crowns in which the leaves and stems die Figure 8.2
back during the dry season (i.e., both crypto- The Raunkaier system of perennial growth form classification based on the
phytes and hemicryptophytes). Farther north, position of buds or regenerating parts of a plant. Perennating tissues are
in the United States, Ipomoea species are annu- shown in black; deciduous tissues are unshaded. The four categories are
al vines and other herbaceous forms. phanerophytes, trees or tall shrubs with buds more than 25 cm above the
ground; chamaephytes, shrubs with buds less than 25 cm above the
Variation in plant shape can also be a ground; hemicryptophytes, perennial herbs with buds at the ground sur-
result of phenotypic plasticity (see Chapter face; and cryptophytes, perennial herbs with perennating organs below the
5) in response to environmental conditions. ground surface.
146 Chapter 8
of plant strategies for competition. When plants grow in (A) “Phalanx”
dense stands (as in forests), selection often favors allo-
cation of resources to upper branches rather than to
lower branches because the upper branches not only
gain more light, but also can also overtop neighbors and
deprive them of light. Such competition for light appears
to have driven the evolution of trees. Trees are plants
that have invested most of their biomass in wood, which
allows vertical growth, rather than in leaves or repro- (B) “Guerilla”
duction. The principal selective advantage of producing
a long trunk seems to be competitive superiority,
Soil
surface
fpo
although woody tissue also contributes to the ability to
live for a long time.
Growth of Clonal Plants Rhizomes
In clonal plants—those with numerous ramets—growth Figure 8.3

form can vary in more complex ways. The spatial dis- Spatial patterns of the spread of ramets within a clone are
tribution of ramets can vary, as can the size and shape sometimes described as being like the phalanx of a classic
of each ramet. The two most important ecological fac- army (A), because the ramets are spatially clumped, or like a
guerilla army (B), because the ramets are spread widely in
tors affecting the spatial distribution of ramets are com- space. Although this characterization is dichotomous, the
petition among genets (including individuals of differ- spacing of ramets actually varies continuously. How one
ent species) and spatial variation in the distribution of characterizes their spacing is dependent on scale: the image
resources. These factors are not necessarily independent in part B might be viewed as an example of either extreme,
of each other. If resources (say, available soil nitrogen) depending on whether one views it from near or afar.
are concentrated in distinct patches, individual genets
are subject to selection for their abilities both to acquire
these resources and to prevent their neighbors from get-
ting them. Some of the variation may be caused simply by dif-
One attempt to classify clonal growth forms was ferences in the biological mechanisms of clonal spread-
Lesley Lovett Doust’s (1981) characterization of plants ing. Each method of spread can potentially result in a
as having either “phalanx” or “guerilla” growth. The wide range of patterns of ramet dispersion. Spread by
idea was that some plants spread like a classic advanc- vegetative (nonsexual) connections, for example, can
ing army, with ramets grouped tightly together in a dis- lead to the slow growth of a compact patch of one genet,
tinct front, while others spread like guerilla forces, usu- or it can lead to a more diffuse structure, with a number
ally through isolated stolons or rhizomes penetrating of smaller patches of ramets. The spread of genets by the
their competitors’ turf (Figure 8.3; see also Figure 5.6). dispersal of plantlets—small plants created vegetative-
Although Lovett Doust pointed out that these categories ly—or asexual seeds (described below)—units that obvi-
were only the end points of a continuum, some ecolo- ously are not physiologically integrated—may also cre-
gists devoted considerable effort to typing clonal plants ate a variety of spatial patterns.
as either phalanx or guerilla species. One problem with It is also clear that some variation in clonal form may
such typology is scale: the categories are subjectively be a more or less passive response to spatial variation in
determined based on what individual ecologists see as resource availability. Once a resource-rich (or toxin-rich)
being tightly grouped. If this kind of metaphor is to be patch is encountered, roots and new ramets tend to fol-
truly useful, it must be applied on the scale of the plant low the concentration gradient of increasing resources
and the dispersion of resources. (or avoid the gradient of increasing toxins). Amy Salz-
Why do clonal plants have variable patterns of spa- man (1985) showed that a clonal ragweed, Ambrosia
tial dispersion? Several kinds of explanations, none of psilostachya (Asteraceae), was able to avoid concentra-
which are mutually exclusive, have been investigated. tions of saline soil in this manner.
These explanations involve the consequences of four dif- This kind of clonal growth has been called “forag-
ferent factors: the mechanics of clonal spread, the degree ing” as an analogy to the way in which animals search
to which plants grow by simply following the distribu- for food. Colleen Kelly (1992) made a useful distinction,
tion of resources, the degree to which plants actively however, between plants passively following a resource
explore the environment seeking resource patches (“for- gradient and those actively “seeking” new patches of
aging”), and the degree to which the clone is integrated resources. She suggested that one can distinguish the
as a strategy for minimizing environmental variability. two responses in cases in which the plant growth occurs
(A)
(B) Dodder Does not coil Coils
Figure 8.4
Cuscuta europaea (dodder, Convolvulaceae) is an obligate
parasite. Like the other 400 or so Cuscuta species, C. europaea Finally, clonal plants vary considerably in the extent
can use a number of host species, which vary in quality as to which they remain physiologically integrated. Ram-
resources. To attack a host plant, it coils around the host and ets connected by a root or rhizome can share water, car-
then sinks absorptive tissues into the host’s phloem. The
“decision” to attack a particular host is made before any bohydrates, and other nutrients. Such sharing may ben-
resource uptake, and it is made within minutes of contact efit the plant in several ways. It may allow a genet to
with the host, as shown in this choice experiment. (A) A spread into microhabitats where a seedling could not
strand of dodder tied to a branch (left) of Crateagus monogyne survive. The genet may grow faster by increasing the
(hawthorn, Rosaceae) fails to coil on this host (right); Cratea- number of ramets. If there are very small-scale differ-
gus is sometimes accepted as a host, but dodder grows poor-
ly on it. (B) A strand of dodder tied to Urtica dioica (stinging ences in the environment (for example, greater avail-
nettle, Urticaceae) coils readily; dodder grows well on Urtica. ability of nitrogen in one spot and phosphorus in anoth-
Host choices are predicted well by models of optimal forag- er), the differences can be averaged by the plant,
ing from animal ecology. (Photographs courtesy of C. Kelly.) sometimes leading to improved overall vigor for the
genet.
Plant Reproduction
before any resource is actually taken up. Her studies of
Cuscuta europaea (dodder, Convolvulaceae)—a parasitic What is reproduction in a plant? It may seem surprising
plant that coils around and sinks absorptive tissues into to begin a discussion of reproduction by asking what it
its host before any resource uptake can occur—showed is—after all, we have a pretty good idea of what our own
that this plant does forage in her strict sense. Dodders reproduction involves. But the biology of plants means
attacked different host species at different rates. Remark- that their reproduction is not so clearly defined. Repro-
ably, the rate at which each host was attacked was pre- duction must, by definition, involve the formation of a
dicted well by optimal foraging models derived from new individual. But as we saw in Chapter 5, the concept
animal behavior studies (Figure 8.4). Kelly suggested of an individual is not as well defined in plants as it is
that this kind of strict-sense foraging might occur in in animals (especially vertebrates). A genet may be made
many other plants. up of many ramets. While everyone would include sex-
148 Chapter 8
ual reproduction in the category of reproduction, pro- tion is necessarily asexual. Elodea spread so extensively
duction of new ramets is sometimes best thought of as by clonal fragmentation that British canals became
asexual reproduction and sometimes as growth, depend- severely clogged, and it is thought that this plant may
ing on the ecological context. For example, if we are con- have contributed to the predominance of railways rather
cerned with natural selection on Populus tremuloides than barges in British commerce (Simpson 1984). Some
(quaking aspen, Salicaceae), which forms gigantic clones plants, such as Kalanchoe daigremontianum (mother of
(see Figure 5.7), we look at the entire clone as the indi- thousands, Crassulaceae), produce little plantlets on the
vidual, and reproduction occurs only when new genets edges of leaves. Physiologically independent ramets are
are formed. But if we are interested in the spread of also created in plants that reproduce with bulbils (tiny,
aspen on a mountain, it may be useful to think of aspen bulblike organs vegetatively produced in inflorescences
as reproducing in two ways, through new ramets (asex- or leaf axils), which are formed above ground in mono-
ual or vegetative reproduction) and through new genets cots such as Agave (century plants, Agavaceae), Allium
(sexual reproduction). (onions and garlics, Liliaceae), and Lilium (lilies, Lili-
aceae), dicots such as Polygonum viviparum (Polygo-
Vegetative Reproduction naceae) and Saxifraga cernua (Saxifragaceae), and ferns
Vegetative reproduction (reproduction by vegetative such as Cystopteris bulbifera (Athyriaceae).
growth of a new ramet) is extremely widespread among
plants. It occurs in most herbaceous perennials as well Seeds Produced Asexually
as in many shrubs and a few trees, and it can take a num- Apomixis (a general term for asexual reproduction)
ber of forms. Often the result is a collection of ramets that occurs in many plant taxa. In many cases this asexual
are physiologically integrated with other ramets, at least reproduction occurs as vegetative reproduction. How-
initially. For example, some plants generate new ramets ever, agamospermy, the production of seeds asexually,
by sending out modified shoots. Fragaria (strawberries, also occurs in a number of plant taxa (many people use
Rosaceae) spread by stolons (runners)—branches that the term “apomixis” to refer specifically to this process).
spread at or just above the surface of the soil and gener- There are a number of mechanisms by which asexual
ate ramets at nodes touching the ground. Eichhornia cras- seed production can occur, described by a complex ter-
sipes (water hyacinth, Pontederiaceae), a South American minology (Gustafsson 1946, 1947a,b). Most of these
native, has become a serious pest in many tropical and mechanisms involve partial meiosis without a reduction
subtropical regions after human introduction because of division, but in some cases there is no meiosis at all. In
its ability to spread rapidly through stolons. It is current- most cases, however, recombination is prevented, so that
ly a severe problem in many waterways in Africa, the the new embryos are clones—genetic duplicates—of the
southern United States, and subtropical Australia. Many plants on which they are formed.
grasses (bamboos, for instance), gingers, and irises spread Agamospermy is fairly widespread. It occurs in
by rhizomes, which are underground horizontal stems some familiar plants, including members of the Aster-
growing near the soil surface. The dense rhizomes of aceae such as dandelions (Taraxacum) and hawkweeds
Hedychium gardnerianum (Kahili ginger, Zingiberaceae), a (Hieracium); citrus plants (Rutaceae); a number of plants
Himalayan plant introduced into Hawaii, make it a seri- in the Rosaceae such as raspberries and their relatives
ous pest that crowds out many native species. Other (Rubus) and cinquefoils (Potentilla); some buttercups
species, including familiar plants in the lily family (Lili- (Ranunculus, Ranunculaceae); many grasses (Poaceae);
aceae) such as tulips (Tulipa), onions (Allium), and daf- and nettles (Urtica, Urticaceae).
fodils (Narcissus), spread by division of bulbs, under-
ground rosette stems that store nutrients. Many woody Sexual Life Cycles of Plants
plants spread by the formation of buds on some of their Plant life cycles are fundamentally different from those
near-surface roots, often called suckers. Populus tremu- of animals because all plants have alternation of gen-
loides forms huge clones in this way, spreading over erations: they have a haploid generation that produces
hectares in western and northern North America. haploid gametes, alternating with a diploid generation
Some plants (such as some seagrasses and cholla that produces diploid spores (Figure 8.5). Alternation of
cacti; see Figure 7.1) spread by clonal fragmentation generations set the stage for the diversification and suc-
(pieces of the plant break off and are capable of rooting cess of plants on land (see Chapter 3). Most of the plants
to form new independent plants), and the new ramets that we see around us are the diploid generation,
are thus never physiologically integrated with other because that is what predominates in gymnosperms and
ramets. Elodea canadensis (Canadian pondweed, Hydro- angiosperms, the most abundant plants in the modern
charitaceae), a common aquarium plant, was introduced world. The individuals that actually mate are haploid.
into Great Britain in the 1800s. In almost every part of In some plant phyla, such as the Bryophyta (mosses),
Britain, only male plants were introduced, so reproduc- however, the haploid generation is actually the larger,
longer-lived, dominant generation; the familiar moss of gametes they produce. However, in flowering plants,
plants you see growing in forests are haploid. we sometimes speak loosely of a male or a female plant
The haploid generation is called the gametophyte (sporophyte) if the individual produces spores that result
because it produces gametes. The diploid generation is in only one type of gamete, either male (pollen) or
called the sporophyte because it produces spores (hap- female (embryo sacs). Figure 8.5 gives a general picture
loid somatic cells). Only the gametophytes have a gen- of plant life cycles; consult an introductory biology or
der (are male or female), which is reflected in the type botany text for a more complete discussion.
Figure 8.5 Sporophyte (2n) Flower

Life cycle of an angiosperm. The sporo- Stigma
Style
phyte—the plant form with which we are
familiar—is the diploid multicellular part Petals
of the life cycle. It gives rise to a haploid,
unicellular spore by meiosis, which grows Ovary
into a haploid, multicellular gameto- Anther
phyte—either a pollen grain or an embryo
sac—by mitosis. At fertilization, a diploid, Ovule
unicellular zygote is formed, which grows
by mitosis into a multicellular embryo. The
endosperm is triploid, formed by the
union of a sperm cell with a binucleate Filament
cell. Tissue in the seed coat comes from the
maternal sporophyte.
Ovary
Seed dispersal
fpo wall
and germination
Ovule
Embryo (2n, the Microsporangium
new sporophyte) (2n) Megasporangium
Mature seed (2n)
inside fruit Meiosis
Endosperm
(3n; 2n from
maternal gametophyte,
1n from pollen)
Seed coat (2n, from

maternal sporophyte)
Microspores (1n) Megaspores (1n)
Microgametophyte Mitosis
Ovary (germinates on stigma)
wall Egg
Sperm
Pollen tube
Ovule
Microgametophyte Megagametophyte
(pollen) (embryo sac)
Megagametophyte
Pollination and
fertilization
150 Chapter 8
There are trade-offs between the two plant genera- spores grow mitotically into either pollen (also called
tions (Niklas 1997). If gametophytes live independently microgametophytes) or embryo sacs (megagameto-
(as they do in mosses and ferns), they can live only in phytes). An embryo sac usually consists of seven cells,
moist places. Free-living gametophytes must be large one of which ultimately becomes the egg cell. It devel-
enough to nourish the new sporophyte, at least initially, ops within a sporophytic structure called an ovule,
but must be low enough to the ground to permit the which is located inside the ovary. The spores that
transfer of swimming sperm. Moreover, they must be become pollen grains undergo two further mitotic divi-
near one another if the sperm are to get from plant to sions. One division creates a vegetative cell and a gen-
plant. If sporophytes grow to a large size, they cause the erative cell. The latter divides again, forming two sperm
demise of the parental gametophyte, limiting its size (as cells. These three cells are haploid, and make up the
in ferns). By contrast, gametophytes maturing inside the entire male gametophyte in angiosperms. The pollen
sporophyte (as in the seed plants) can get their nutrition grains develop in a sporophytic structure called an
from the sporophyte. This arrangement allows the game- anther, located at the end of a stalk called the stamen.
tophytes to be much smaller. It also allows reproduction Pollen germinates after landing on a stigma, the
to occur in dry environments, and frees them from the receptive surface of the flower (Figure 8.6A). Upon ger-
constraint of having to be near one another for fertil- mination, the vegetative cell grows into a pollen tube
ization to occur. (see Figure 8.10), which penetrates the stigma and style,
Most familiar plants are the sporophytes of gym- and, ultimately, the ovary and an ovule inside the ovary.
nosperms (such as the conifers) or angiosperms (flow- The sperm cells are transported down the pollen tube,
ering plants). The reproductive structures of gymno- where one of them fertilizes the egg cell, forming a new
sperms are cones or conelike structures. The reproductive diploid zygote. This zygote grows and divides by mito-
structures of angiosperms are flowers. Within either the sis, forming the embryonic tissues of a new sporophyte.
cone or flower the gametophyte is produced and fertil- In almost all angiosperms, the second sperm cell fuses
ization occurs. Both gymnosperms and angiosperms pro- with a specialized cell in the embryo sac that has two
duce seeds in angiosperms; fertilization results in the nuclei. This triploid cell grows to become endosperm,
ovary maturing into a fruit with seeds inside. Thus, which serves as the primary nutritive source for the
although not all angiosperms produce showy flowers or developing embryo. Once fertilized, an ovule is called a
fruit that is edible by humans, they all produce flowers, seed. In angiosperms, this seed grows inside a fruit,
seeds, and fruit as their basic reproductive structures. which is the mature ovary of a flower produced by the
Flowers can be large and conspicuous or small and incon- maternal sporophyte. The tissue in the seed coat comes
spicuous, depending largely on the mechanisms by from the ovule. Thus the tissues in a fruit, and some of
which pollination occurs in a particular plant group. the seed tissues, come from the maternal sporophyte;
Fruits can be large or small, colorful, sweet, and fleshy only the embryo inside the seed is a new genetic indi-
or dry and hard, toxic or nutritious, depending on the vidual. In the other seed plants, the gymnosperms, the
evolution and ecology (especially the seed dispersal basic pattern of gametophyte development and seed
mechanisms) of the plant producing them. The tremen- development are similar, with the key differences being
dous diversity of seed plants and their dominance over that the female gametophyte consists of many more cells,
terrestrial habitats are due largely to these reproductive endosperm is absent, and development usually takes
structures, which permit them a great deal of independ- substantially longer.
ence from water for mating and dispersal.
In seed plants, the new embryo (sporophyte) begins
life in a protected structure, the seed, which can be dis- Pollination Ecology
persed and can sometimes remain dormant for a long
time. Flowers sometimes act to attract certain animals Wind Pollination
(and discourage others) as pollinators. The fruits of When most people think of flowers and pollination, they
angiosperms further enhance the protection and disper- think of showy flowers that are visited by birds or
sal of the seeds enclosed within them. Finally, the inter- insects. But for a large number of plants, pollen is car-
action of several different kinds of tissues—from both ried from flower to flower mainly by wind. There are
gametophyte and sporophyte—has allowed the evolu- two quick ways to convince yourself that wind pollina-
tion of mechanisms for limiting matings within geno- tion is important. First, almost all pollen transfer in such
types. In sum, these characteristics of plant life cycles are important plant groups as grasses, rushes, and most
basic to plant reproductive ecology and to the evolution- temperate-zone trees (including all conifers, oaks, beech-
ary success of the seed plants, especially the angiosperms. es, maples, elms, willows, and many others) is by wind.
When cells in flowers undergo meiosis, the hap- Second, one of the most common human medical com-
loid daughter cells that result are called spores. These plaints (at least in developed countries) is allergy to
(A) Petal
(B)
Stigma
Anther
(micro-
Pistil
Style sporangium) Stamen
(male)
fpo
(female) Ovary Filament
Ovule Sepal
Receptacle
fpo
Figure 8.6 (C)
The form of a flower depends on the
mode of pollination. (A) The parts of a
“typical” flower. (B) A flower of Poa Calyx spur
pratensis (Kentucky bluegrass, Poaceae). Petals
The enlargement shows a single spikelet.
The lower floret is open; the feathery Pedicel
structures are the receptive stigmas. In the
upper floret, the anthers have already
dehisced—opened—and pollen has been
fpo
dispersed. These grass flowers have no
petals or sepals. (C) Delphinium (larkspur,
Ranunculaceae) has a spur in one petal
that contains nectar to attract pollinators.
pollen (“hay fever” is one type)—the overreaction of our striking, a note of caution is in order: In many plants,
immune systems to the foreign proteins in the huge pollen is carried both by wind and by animals. The pres-
amounts of pollen in the air. ence of petals does not prohibit wind pollination, nor
The majority of plant species in the modern world does their absence mean that animal pollination is nec-
are angiosperms, and all angiosperms produce flow- essarily unimportant.
ers. However, while many animal-pollinated species Pollen transfer by wind is much reduced in shel-
produce attractive flowers, wind-pollinated flowers are tered habitats where air movement is reduced. Thus
rarely showy. There is generally no selective advantage wind pollination is most common in plants of open habi-
to producing metabolically costly large petals, scents, tats. Wind-pollinated temperate forest trees usually
nectar, and floral pigments if they are not needed to flower well before their leaves emerge in spring. While
attract pollinators. In fact, there is often a disadvantage this timing has sometimes been explained as a mecha-
to making large petals and other organs that interfere nism to reduce the amount of foreign pollen received on
with pollen transfer by wind. The grass family, for exam- stigmas, it is worth noting that pollen movement is gen-
ple, is almost entirely wind-pollinated, and grass flow- erally greatest when trees are leafless, which also often
ers lack petals and sepals (Figure 8.6B). happens to be the windiest time of the year. Wind can
Wind-pollinated plants produce massive quantities carry pollen very long distances. On June 5, 1998, some
of pollen because they cannot exert much influence over local hunters at Repulse Bay in extreme northern Cana-
where their pollen blows. The quantity of pollen pro- da along the Arctic Ocean noticed large amounts of
duced can be expressed as the ratio of pollen to ovules pollen at the edges of ponds (Campbell et al. 1999). It
produced by an individual or population. Wind-polli- proved to be pollen from Pinus banksiana (jack pine,
nated plants typically have much higher pollen:ovule Pinaceae) and Picea glauca (white spruce, Pinaceae) that
ratios than insect-pollinated species. Many wind-polli- had been blown on June 1 from Quebec, Canada, over
nated species also have very large, feathery stigmas to 3000 kilometers away!
enhance pollen capture. Pollen in these plants is light It is tempting to think that wind pollination is evo-
and rarely sticky, whereas insect-pollinated plants often lutionarily primitive. It seems so simple and inelegant
have heavy, sticky pollen. Although these patterns are when compared with some of the complex adaptations
152 Chapter 8
some plants have for animal pollination. And it is clear sensitive to blue (436 nm), green (546 nm), and red (700
that adaptation to animal pollinators has been a driving nm). Insects, on the other hand, can usually see colors at
force in plant evolution. However, both ancient lineag- shorter wavelengths than humans. Bees see wavelengths
es (such as gymnosperms) and more recently evolved mainly from 300 nm to 700 nm. Thus, bees are less sen-
taxa (such as grasses) have wind pollination. Some sitive than we are to long-wave light and have difficul-
wind-pollinated taxa have given rise to animal-polli- ty discriminating among shades that we perceive as red,
nated taxa, while others have been derived from animal- but there are colors that they can see—those between 300
pollinated ancestors. and 380 nm—that are invisible to us. Birds’ eyes are most
sensitive to colors in the middle and red parts of the
Attracting Animal Visitors: Visual Displays spectrum. Thus, flowers that are visually attractive to
Flowers exist for one reason only: sex. Everything about bees may not be as noticeable, attractive, or easily dis-
them has to do with sex. This is obvious with the floral tinguishable to birds, and vice versa. Many bee-polli-
parts that are directly involved in producing or receiv- nated flowers have yellow hues, while bird-pollinated
ing pollen, or in the growth of seeds and fruits. But it is flowers are often red or orange. Flowers pollinated by
also true of the other floral parts, such as petals, that moths are usually white to pale yellow, while bat-polli-
function mainly to influence the transfer of pollen. nated flowers are usually white to brown. Butterflies are
Animals visit flowers because they generally con- often attracted to yellow or blue flowers. A note of cau-
tain rewards—mainly nectar, which is rich in sugars and tion is again in order, however: Many flowers reflect
sometimes amino acids, and pollen, which is rich in pro- light in several different wavelengths. Flowers that we
tein. Most people think of floral traits such as shape, perceive as, say, red may be easily distinguished by bees
color, and scent as being important in attracting animal because they also reflect in the blue wavelengths.
pollinators—as advertisements of the rewards the ani- Flower color plays a more complicated role in pol-
mal can expect to acquire at the flower. But the interac- lination than these observations might suggest. Many
tion between plants and their floral visitors often flowers have contrasting colors in the perianth (the
involves a great deal of mutual cheating, false advertis- petals and sepals together). For example, the central por-
ing, and theft. As we will see below, some plants appear tion of the flower may contrast sharply with the rest of
to advertise rewards, but do not deliver them. Some ani- the flower, as in many irises (Iridaceae). In most species
mals visit flowers and remove rewards, but do not pol- that have been studied, these contrasts seem to have one
linate the flowers. Moreover, one species’ attractant may of two functions. In some cases, the contrast itself is the
be another species’ repellant. attractant for pollinators. In other cases, one of the col-
Displays of bright red flowers catch our eye. Al- ors functions as a “nectar guide”—pollinators are first
though most people also enjoy looking at white flowers, attracted to the flower as a whole, and are then attract-
we often do not notice them as easily as we notice red ed by the color contrast to the areas containing nectaries,
ones. Animals vary in their ability to see and discrimi- anthers, and stigma (Figure 8.7).
nate among colors (Kevan 1978). For example, humans Flower colors can vary over time, and numerous stud-
see colors between the wavelengths of about 380 nm ies have shown that these variations make a difference in
(violet) and 780 nm (deep red), and our eyes are most attracting pollinators. In most cases, older flowers fade,
(A) (B)
Figure 8.7
A Caltha palustris (marsh mari-
gold, Ranunculaceae) under
white (left) and ultraviolet
(right) light. The color contrast
that we see only under ultra-
violet light is visible to bees in
ordinary sunlight, and serves
as a nectar guide. (Photographs
courtesy of T. Eisner.)
(A) Determinate inflorescences although there are some, such as those of the
prickly pear and cholla cacti, that grow dark-
er over time. These changes in flower color
seem to function much like nectar guides.
The old flowers add to the attractiveness of
the floral display as a whole, but once an ani-
mal has been drawn to the plant, it tends to
be attracted to young, fresh flowers; these are
the ones containing nectar or pollen.
Other parts of plant bodies have also
fpo been modified to act as attractants. Floral
bracts—specialized leaves found below
many flowers—are often colored and appear
Cyme, opposite branching Umbel (in bud)
to act as part of the plants’ display. A well-
known example is Euphorbia
pulcherrima (poinsettia, Euphor-
Ray florets
biaceae), whose red bracts and
upper leaves are far more color-
ful than its tiny yellow flowers.
Many vase-forming bromeliads
Disk florets
also have leaves that are more
brightly colored than their flow-
ers (see Figure 2.17).
Flowers are often arranged in aggrega-
tions called inflorescences. There are a num-
Cyme, alternate branching Head (capitulum) ber of types of inflorescences, ranging from
the capitulum (head) of composites (Aster-
aceae) to the catkins of beeches and oaks
(Fagaceae) to some rather loosely arranged
(B) Indeterminate inflorescences racemes in plants such as buttercups (Ranun-
culaceae) (Figure 8.8). The terminology
describing inflorescences is extensive and
somewhat obscure; an introductory botany
or plant morphology text can be consulted
for more information. Inflorescence structure
is important in pollination ecology. Having
fpo flowers gathered into an inflorescence may
act to increase the size of the attractive dis-
play without altering the flowers themselves.
Axillary flower
It is likely that the form of an inflorescence
also affects the behavior of pollinators.
Raceme Spike
Attracting Animal Visitors: Floral Odors
and Acoustic Guides
Many flowers have noticeable odors,
which clearly act as attractants. Why have
scent as well as visual attractors? It is gen-
erally thought that scent acts over much
Figure 8.8
Some forms of inflorescences. A principal division is between (A) determinate inflo-
rescences (which have a flower at the end of each flowering branch) and (B) indeter-
Corymb minate inflorescences (which do not). In a determinate inflorescence, the terminal
flower normally blooms first, and the maximum number of flowers in the inflores-
cence is strictly limited.
154 Chapter 8
longer distances than visual attraction, since many ani- right place. Limiting visits to only a small group of pol-
mals are able to detect extremely low concentrations of linators would increase the chance that the next plant
molecules in the air. On the other hand, scents are hard visited will be of the same species. Numerous animals
to locate precisely. Once an animal is in the general area that are not effective pollinators are nevertheless attract-
of a patch of flowers, it is likely to find them much more ed to flowers. From the animal’s viewpoint, pollination
quickly by vision than by scent. is almost always an accidental consequence of this visi-
Much less is known about floral odors than about tation. Thus many plants have undergone selection to
colors, partly because the science of measuring odors is deter certain floral visitors—but, as we all know from
not nearly so well developed as the science of optics. personal experience, it can be hard to get rid of unwant-
Nevertheless, a few general points can be made. Perhaps ed guests!
the most important is that odors can vary independ- Changes in flower shape are the main way in which
ently of colors. Genes that make for yellow or red flow- access to flowers is restricted. The earliest flowers are
ers usually have no effect on flower odor. All of us have thought to have been pollinated by insects unspecialized
been surprised to find no odor, or an unpleasant one, for this role, such as beetles. Not surprisingly, flowers
coming from a visually attractive flower. Pollinators sim- that are typically beetle-pollinated today are often open
ilarly experience floral odors quite separately from the and bowl-shaped. On the other hand, many flowers are
way they experience colors. tubular, and some have long spurs or smaller cones
The second point should, upon reflection, be obvi- behind the main flower tube (see Figure 8.6C) in which
ous: odors that attract one group of animals may repel nectar collects. Only animals with long tongues (or other
others. Casual observation of the responses of humans, long mouthparts), such as birds, butterflies, or moths,
dogs, and flies to various odors should convince you of can reach the nectar and thus pollinate these flowers.
this! Bees are often attracted to what we perceive as The type of reward a plant offers also restricts visi-
sweet odors, and bats to musty odors. A number of tations. Beetles, for example, are generally unable to use
plants, especially the milkweeds in the family Apocy- nectar, whereas birds, bats, and bees use it easily. Some
naceae and the aroids (Araceae), have odors similar to orchid bees (family Euglossidae) collect floral scents (by
animal dung or rotting tissue, which are strongly attrac- scraping the floral cells that secrete odor-producing oils),
tive to the flies and gnats that pollinate these plants. which they use in mating displays. Others use the oils
Some plants have odors that mimic insect mating phero- for nest building. The timing of reward availability is
mones, which cause male insects to visit the flowers and important as well; bat- and moth-pollinated flowers usu-
sometimes attempt to mate with them. ally produce nectar at night.
Possibly the most unusual flower attraction system Mechanisms such as these do not, of course, provide
known is that of the tropical vine Mucuna holtonii any guarantees to the plant against wasted nectar and
(Fabaceae), a bat-pollinated species (von Helversen and pollen. Recent studies have shown that many flowers
von Helversen 1999). The upper petal on the flower is are visited by large numbers of animals that transfer
raised when the flower opens. This petal is concave and little pollen. In a study of Calathea ovadensis (Maran-
acts as a sound mirror—an acoustic guide—reflecting taceae), a tropical Mexican herb, Douglas Schemske and
back bats’ sonar signals and letting them know that the Carol Horvitz (1984, 1988) showed that moths and but-
flower is open. To reach the nectar, a bat has to press its terflies are the most common visitors, but bees account
snout deep into the flower. When it does so, another for the most pollination. Among the bees, the most effi-
petal, which is holding the stamens, bursts open, throw- cient pollinators turned out to be relatively unimportant
ing the pollen onto the bat’s rump. This plant species is because they were uncommon. An extreme case of wast-
the first known to have an acoustic guide, but other bat- ed resources is that of “nectar robbers,” animals that
pollinated species may also have them. insert their mouthparts between petals, or bore or chew
through the side of the flower, and drink the nectar
Limiting Unwanted Visits without ever coming into contact with the pollen. Very
Many plants have adaptations not just for attracting cer- small insects such as ants also sometimes act as nectar
tain visitors, but for repelling others. Why would a plant robbers because they can visit the nectaries without con-
want to limit visitors? Once an animal visitor has tacting stamens or stigma.
removed pollen from a flower, that pollen has to reach
another flower of the same species for it to contribute to Pollination Syndromes
the reproductive success of the plant producing it. If the Color, odor, reward type, timing of rewards, and shape
next flower the animal visits is of another species, then all affect the numbers and types of floral visitors. It is not
the pollen is likely to be deposited in the wrong place. surprising, therefore, that certain combinations of these
From the perspective of the plant, the best pollinator vis- attributes often seem to be associated with particular
its only one species, ensuring that the pollen gets to the types of pollinators. From what we have said, one might
expect that bee-pollinated flowers would be yellowish, in water will find receptive stigmas. Furthermore, tides
sweet-smelling, broad enough for the bee to contact the and currents are likely to make pollen dispersal quite
anthers and stigma, and produce nectar during the day- unpredictable.
time. Bird-pollinated flowers might be expected to be Aquatic plants have generally reduced this unpre-
red or orange, produce copious nectar during the day- dictably by making their pollen dispersal units larger
time (but little scent), and have long tubes or spurs—and (and thus more likely to contact a stigma) and, in many
so forth. Such associations of floral attributes with par- cases, by dispersing them mainly at the water’s surface
ticular types of pollinators (usually taxonomic classes or (so that they travel in a two-dimensional plane, rather
orders) are called pollination syndromes (Faegri and than a three-dimensional volume). A remarkable exam-
van der Pijl 1979). These syndromes are usually thought ple is the freshwater plant genus Vallisneria (Hydrochar-
of as being matched by adaptive suites of pollinator itaceae; Figure 8.9). In this genus, staminate (male) flow-
traits—plant–pollinator interactions are thought of as ers containing pollen are released underwater and float
tightly coevolved mutualisms (Howe and Westley 1997). to the surface. The petals curve downward to keep the
Some ecologists have stood the idea of pollination syn- flower floating on the water, and two sterile anthers act
dromes on its head, treating floral traits as diagnostic as tiny sails. The flowers are thus blown around until
of particular pollinators: if flowers have traits X, Y, and they encounter pistillate (female) flowers containing
Z, the reasoning goes, then the plant’s important polli- ovules, which are held at the surface by long pedicels
nators are in animal taxon A. (stalks). The pistillate flowers create small depressions
There is little doubt that many plants and pollina- in the water surface, and the staminate flowers fall in,
tors are coevolved—that the plants have strongly influ- pollinating the pistillate flowers.
enced the evolution of their pollinators, and vice versa. In most aquatic species, pollen, rather than a whole
There are some well-known cases that offer evidence for flower, is the dispersal unit. In many cases, the pollen
tight coevolution of particular pairs of plants
and pollinators (Box 8A). But there also are
reasons to think that the strength and gen- Staminate
flower
erality of pollination syndromes have been
overstated. In a recent critique, Nickolas
Waser and his collaborators (Waser et al. Petals
fpo Pistillate
flowers
1996) pointed to evidence suggesting that
animals outside these syndromes accom-
plish much pollination—birds at “bee flow-
ers,” bees at “butterfly flowers,” and the like. Young
fruit
A principal criticism of the use of the polli-
nation syndrome idea is that the syndromes Pedicel (stalk)
are tendencies—and often rather weak
ones—rather than laws. The critics have sug-
gested that by concentrating on these ten-
dencies, ecologists may be missing variation
that is important. They have also argued
that, under some fairly broad conditions,
both plants and pollinators can be expected
to evolve to be pollination generalists, rather
than specialists. Thus, focusing on syn-
dromes has caused us to neglect these gen- Pedicel
eralist plant-pollinator interactions.
Staminate
Aquatic Plants and Pollination plant
Pistillate
Some flowering plants, such as waterlilies plant
(Nymphaea or Nuphar, Nymphaeaceae), live
in the water. In most of these, the flowers are
Rhizome
borne above the water’s surface, and insects
or wind pollinate the plants. But in plants of
some 31 genera in 11 families, the flowers are Figure 8.9
borne underwater. As with wind pollination, The aquatic plant Vallisneria (Hydrocharitaceae) has special adaptations for
it is inherently unlikely that pollen grains pollination at the water’s surface.
156 Chapter 8
BOX 8A
Specialized Plants and Pollinators

Plant relations with their pollinators pulpit). The most extreme case is that
range from being highly specialized— of Amorphophallus titanum (voodoo lily),
with the plant depending on visits from a plant native to Sumatra. Not only
only one or a few species of animals— does this plant have the largest inflo-
to quite generalized. There are many rescence in the plant kingdom, but it
interesting cases of specialization and may also have the most powerful odor,
adaptations by plants or their pollina- which is reputed to resemble a combi-
tors. In most cases, the animals are not nation of rotting fish and burnt sugar
as specialized as the plants, as they are (none of the authors has yet had the
pollinating plants for longer periods pleasure). There are reports of people
than any single plant species is in fainting from the odor.
flower. Many African succulents in the dog-
Flowers of Yucca (Agavaceae) are bane family (Apocynaceae, another
pollinated exclusively by Tegeticula family that often has pollen grains
moths. The moths pollinate the yucca clumped together in pollinia) are fly-
flowers while laying their eggs in an pollinated. These plants attract flies by
ovary, and the larvae eat a fraction of having flowers that smell much like
the seeds in that ovary. If too many eggs decaying meat or, in some cases, animal
are laid in a particular ovary, the plant dung.
will abort it. Some yuccas abort a size- Wild figs (Ficus, Moraceae) are main-
able number of pollinated flowers. ly pollinated by specialized wasps. The
In the orchid family (Orchidaceae), flowers in this family are borne in a
pollen grains are clumped together in unique way: they are on the inside of
structures called pollinia. The size of the inflorescence, so that a wasp can
the pollinia makes it relatively difficult Amorphophallus titanum (voodoo lily, Araceae), reach them only by boring a hole in it.
to move them from flower to flower. In a plant native to Sumatra, has the largest single The wasp then lays eggs in the inflores-
some tropical epiphytic orchids, the inflorescence of any plant. The man in the pho- cence. When the brood hatches, they
flowers are shaped so that an insect— tograph is Hugo de Vries, an important early mate (usually, this is a brother-sister
geneticist and one of those responsible for
usually a bee—must crawl through a mating), and the wingless males then
bringing Gregor Mendel’s work to the attention
particular opening to get to the nectar of biologists. (Photograph by C. G. G. J. van die (so every wild fig contains dead
or oil rewards, and in doing so, cannot Steenis, Hugo de Vries archive, Faculty of Sci- wasps). As the young females leave the
avoid hooking the pollinia in a way that ence, University of Amsterdam.) inflorescence, they are dusted with
facilitates their transport to the stigma pollen, and when they visit another
of the next flower visited. Perhaps the inflorescence to lay their eggs, they pol-
most famous case is that of Ophrys (first Many plants in the family Araceae linate it.
studied by Charles Darwin [1877]), have foul-smelling flowers that attract Fascinating as these examples are, it
which has an odor similar to the mating gnats and other small flies. Two com- is easy to get the wrong impression: as
pheromones of female bees. Males visit mon North American examples are noted in the text, most plants have
these flowers and “pseudocopulate” Symplocarpus foetidus (skunk cabbage) much less specialized mechanisms of
with them, thus transferring pollen. and Arisaema triphyllum (jack-in-the- pollinator attraction.
is sticky and tends to form rafts that float on the sur- remarkable case of elongation is found in the seagrass
face until they contact pistillate flowers. Pollen release genus Zostera (Zosteraceae). The pollen germinates
in some subtidal marine species occurs only near very before release, and the pollen tube greatly increases the
low (spring) tides, which also increases the chances of length of the pollen and enhances its chance of becom-
contact between pollen and stigmas. In species with ing caught in a pistillate flower. Pistillate flowers of
entirely underwater pollination, such as the turtle grass aquatic plants also show a number of features, such as
Thalassia (Hydrocharitaceae), pollen tends to be elon- changes in stigma shape, that act to increase the proba-
gated or embedded in strands of mucilage—features that bility of contact with pollen.
also increase the chances of contacting stigmas. A
of variation among individuals can occur simultane-
Who Mates with Whom? ously within a single population.
Plants have remarkably complex sex lives. In this sec- Having a particular set of floral organs, however,
tion we look at plant mating systems—biological fac- does not mean that a plant population of hermaphro-
tors that govern who can mate with whom. Some species dites necessarily uses them. Within a population of her-
are obligately self-fertilizing (individuals can only pol- maphrodites, for example, some individuals may main-
linate themselves), other species are obligate outcrossers ly contribute pollen and set few seeds, acting mainly as
(individuals must be pollinated by others), while some males. The pollen on what appear to be hermaphrodites
species may both self and outcross. This multiplicity of may be inviable or cannot germinate, and those indi-
mating patterns is accomplished by a variety of mecha- viduals are functionally strictly female. When individu-
nisms. The conditions under which one mating system als that appear to us to be hermaphrodites actually are
or another should be favored by selection constitute an functionally unisexual, we speak of cryptic dioecy. The
area of active research, both theoretical (for example, functional gender of a plant may be very different from
Charlesworth and Charlesworth 1978; Barrett and Hard- its apparent physical gender; in other words, the ratio of
er 1996) and experimental (for example, Carr and an individual’s male reproductive output to its female
Dudash 1996; Husband and Schemske 1996; Vogler and reproductive output may be quite different from 1.
Kalisz 2001). Since this kind of variation can greatly affect the
genetic composition of populations, it should be no sur-
Plant Gender prise that plant gender have been the subject of much
Gender expression in plants is complex and varied. study among evolutionary ecologists, conservation biol-
Some individuals are cosexual—they function as both ogists, and population geneticists. Eric Charnov (1982)
males and females simultaneously. The most common developed the basic theory that has motivated many
form of cosexuality is hermaphroditism. Hermaphro- studies of the evolution of gender. The central idea is that
ditic individuals have perfect flowers containing both of fitness maximization: in different settings, fitness may
functional stamens and functional stigmas. Under be maximized by increasing the allocation of resources
monoecy, plants have some flowers with functional sta- to both male and female function, or by increasing one
mens only (staminate flowers) and some with func- and decreasing the other. The genetic basis of variation
tional stigmas only (pistillate flowers); sometimes per- in gender is understood in some plant species; there are
fect flowers also occur on the same individual. By well-known cases in which the controls are quite simple.
contrast, under dioecy, at least some plants in a popu- Discussion of plant gender obviously involves a lot
lation have only pistillate or only staminate flowers. of terminology. To make matters worse, different bio-
These three systems of gender expression can exist logical disciplines use these terms to mean somewhat
in combination with one another. Under gynomonoecy, different things. Animal biologists usually use “her-
both pistillate and perfect flowers occur on the same maphroditic” to mean that both male and female sex
individuals. By contrast, under andromonoecy, both sta- organs are produced by the same individual (since there
minate and perfect flowers occur on the same individ- is no animal analogue to a flower). Geneticists use
uals. Andromonoecy is much more common than “monoecious” to mean the same thing. This terminolo-
gynomonoecy. Under gynodioecy, some plants in a pop- gy is not imprecise. Scientists all mean something quite
ulation have only pistillate flowers, while others have specific when using these terms, but each subdiscipline
either perfect flowers or a mixture of staminate and pis- means something a bit different. In practice, this is not
tillate flowers. By contrast, under androdioecy, some usually a problem: one just needs to be aware of these
plants in a population have only staminate flowers, differences and use the context to determine the mean-
while others have either perfect flowers or a mixture of ing an author intends.
both staminate and pistillate flowers. Gynodioecy is
much more common than androdioecy. Competition for Pollinators and among Pollen Grains
Still another form of gender expression is sequen- As Oscar Wilde remarked, “Nothing succeeds like
tial hermaphroditism: beginning life as one gender excess.” Many plants appear to have followed this
(almost always male) and then switching to the other. In advice. Almost all plants start out with many ovules that
some sequential hermaphrodites the switch is not instan- never turn into viable seeds. These excess ovules may
taneous, and individuals may function as hermaphro- exist for three reasons: because pollen is a limiting
dites for a time. resource, because the flowers have another function, or
Some cosexuals are capable of self-fertilization because the plant is hedging its bets (producing more
(sometimes within the same flower and sometimes only ovules than can usually be matured, because under
between flowers), while others are not. What is even unusually favorable conditions these ovules can be
more remarkable is that some of these different patterns matured). We consider the first two reasons here; bet
158 Chapter 8
BOX 8B
Pollination Experiments
Many questions about plant reproduc- plants, and hand-pollinate plants. The flowers, hand pollination with pollen
tive ecology can be answered with very results of these treatments are then from the same or different individuals
simple experiments. Often we have two compared with results from unmanip- tells us whether the species is self-
questions about a particular species: ulated plants, allowing us to estimate incompatible or simply requires ani-
Are the plants self-incompatible? Are the typical rate of seed set. Bagged mals for pollination. Doing both bag-
the plants pollen-limited? To answer flowers indicate how many seeds in the ging and emasculating tells us whether
these questions, we can do various unmanipulated plants came from out- the species is capable of producing
combinations of three types of manip- crossed pollination. Emasculated plants seeds by agamospermy. If hand-polli-
ulations: cover flowers or inflorescences show how many seeds in the unma- nated plants produce more seeds than
with mesh bags to exclude pollinators, nipulated plants came from self-polli- naturally pollinated plants, the plants
remove stamens from (emasculate) nation. If we get no seeds from bagged are pollen-limited.
hedging is discussed more fully in Chapter 9. tion only to attract pollinators. These flowers are, thus,
For animal-pollinated plants, pollen may be limit- functionally male; they reproduce only by having their
ing because of a lack of pollinators (Box 8B). If there are pollen fertilize the ovules of flowers on other plants, and
more flowers than pollinators to visit them, or if the pol- not by being pollinated themselves. Such cryptic dioecy
linators are inefficient at transferring pollen, the plants is particularly common in legumes (Fabaceae) and in the
may be competing for pollinators. This competition can Bignoniaceae. Competition among pollen grains after
occur both among plants of the same species and among they land on the stigma can also result in sexual selec-
species in a community. These two types of competition tion. There is a race for the ovules as pollen tubes grow
can have different consequences for flowering patterns, down through the style (Figure 8.10). Experimental stud-
as we will see in Chapter 9. ies have shown that in some species, the maternal
Competition for pollinators among individuals of sporophyte “chooses” among its pollinators through bio-
the same species is analogous to competition among chemical interactions that affect pollen germination and
males (or females) for mates in animals. When this con- the rate of pollen tube growth, resulting in more fit off-
cept was first developed in the 1980’s—notably by Mary spring (Marshall and Folsom 1991). Maternal sporo-
Willson (Willson 1983; Burley and Willson 1983)—it was phytes are also known to selectively abort certain seeds
met with some skepticism because the analogy between before they mature (such as Cucurbita lepo [zucchini,
pollen grains on a stigma and male birds competing for Cucurbitaceae]; Stephenson and Winsor 1986).
the attention of females seemed strained to some
botanists. However, there is now a large body of evi- Pollen Dispersal and Its Consequences
dence that pollen donors and pollen grains do indeed Some degree of inbreeding (mating among relatives; the
compete. Some individuals in a population may end up opposite is outcrossing) is common in plant populations.
with many matings (lots of pollen), while others may get The reason is simple: The largest number of matings
very few. This variation results in sexual selection— occur among neighboring plants, which tend to be rel-
selection by means of differential success at mating. Most atives because most seeds do not travel far.
evolutionary ecologists consider sexual selection as a Most pollen ends up near the sporophyte on which
subset of natural selection, it originated. In animal-pollinated species there are sev-
although common usage some- eral reasons for this pattern. First, most plant species
times treats this area of selection have clumped spatial distributions. Moreover, some
theory separately. patches of plants are typically more rewarding to polli-
There are two broad cate- nators than others, and the pollinators usually concen-
gories of sexual selection: male- trate on the more rewarding patches, reducing the aver-
male competition, and female age distance traveled between plants. Finally, most
choice. In plants, male-male com- pollen is transferred to the next flower visited (Richards
petition is often thought to result 1986). In Lupinus texensis (bluebonnet, Fabaceae), for
in extremes in floral displays, example, most pollinator movements between plants are
such as large inflorescences in less than 0.6 m, and over 90% of movements are less than
Mary Willson which most of the flowers func- 1.2 m (Figure 8.11). Gene flow does not precisely match
pollinator movement (again suggesting that pollinator
visitation is not equivalent to pollination), but gene
movement also declines rapidly with distance from the
Pollen
source.
grains
Pollen movement by wind also usually results in
most pollen being distributed near its source, but for dif-
ferent reasons. Wind usually blows in short gusts, most
of which have low velocity, so most pollen movements
will be over short distances. In trees, pollen is released
at varying heights, and the pollen released at the great-
est height usually travels the farthest (Richards 1986).
However, most trees are shaped so that little pollen is
released from the treetops; thus, most pollen travels
Stigmatic short distances. Since most wind-pollinated plants (such
surface as grasses) occur in large stands, most pollen tends to
land on neighboring plants. In some populations, plants
have variable characteristics that can affect the distance
that pollen travels. In Plantago lanceolata (Plantagi-
naceae), Stephen Tonsor (1985) found that individuals
differed in the extent to which they produced pollen as
single grains or as clumps of two, three, or up to fifty
Figure 8.10 or more grains. A wind-tunnel experiment showed that
When pollen grains germinate, the pollen single grains travel much farther than larger clumps
tubes grow, penetrate the stigmatic surface, (Figure 8.12). Thus, some plants are more likely to mate
and grow down the style towards the
ovules. Shown here are two germinated
pollen grains of Erythronium grandiflorum
(glacier lily, Liliaceae). (Photograph cour-
tesy of A. Snow.) (A)
Pollinator movement (%)
Gene flow
Foreign alleles (%)

0.60 0.28
Pollinator
0.45 0.21 movement
0.30 0.14
0.15 0.07
0.6 1.2 1.8 2.4 3.0 3.6 4.2 4.8

Distance (m)
(B)
100
75
fpo
Marker gene (%)
Figure 8.11
(A) The distance traveled by pollinators
(open bars) and the appearance of marker 50
genes (solid bars) in seedlings of Lupinus tex-
ensis (bluebonnet, Fabaceae). While the gener-
al pattern of distribution is the same, using
pollinator movement alone to estimate gene 25
flow would underestimate it, because pollen
from one flower may be deposited on the sec-
ond, or third (and so forth) flower that is vis-
ited. (After Schaal 1980.) (B) The appearance
of marker genes in seedlings of Cucumis 0 2 4 6 8
sativus (Cucurbitaceae) (After Handel 1983.) Distance (m)
160 Chapter 8
Solitary grains
300 Paired grains
3–6 grains
250 7–15 grains
fpo 16 or more grains
Number of clusters
200
150
100
50
0 20 40 60 80 100 120 20 40 60 80 100 120 20 40 50 60 100 120

Distance (cm)
Figure 8.12
Variation in the distance traveled by Plantago lanceolata (Plan-
taginaceae) pollen clumps in a wind tunnel as a function of
size in three trials. In each trial, a plant was placed at one end mating systems are, it seems that only a few major fac-
of the wind tunnel, and rows of Vaseline-coated microscope tors have been important in their evolution. These fac-
slides were placed at 20 cm intervals along the floor of the tors include the fitness consequences of inbreeding ver-
tunnel. Each line shows the number of solitary grains or sus outcrossing, the relative success of male and female
clumps that traveled each distance. (After Tonsor 1985.)
floral functions, the ability to adjust resource allocation
to male and female functions, and the availability of pol-
linators.
with distant individuals, and some are more likely to There is almost always some inbreeding among
mate with close neighbors. plants because the individuals in a population are like-
ly to share common ancestors. As in a small human vil-
Assortative Mating lage, everyone is a cousin of some sort. But strong
Plants with similar phenotypes—for example, flower inbreeding occurs when plants pollinate themselves or
color or flowering time—are sometimes more likely to mate mainly with closely related neighbors. Reproduc-
mate with one another than they would by chance alone. tive traits can promote or limit such inbreeding.
This phenomenon is called assortative mating. In the If inbreeding decreases fitness (inbreeding depres-
case of flowering time, this is especially obvious: early sion), it is advantageous to limit matings with relatives.
flowerers are more likely to mate with other early flow- Self-incompatibility (SI) systems can accomplish this.
erers than with late flowerers. Assortative mating for There are two general types of SI systems. Typically, the
flower color often occurs because many pollinators have self-incompatibility loci are referred to as S loci and indi-
preferences (sometimes on an individual basis) for col- vidual alleles are numbered S1, S2, and so forth. In game-
ors. A bee, for example, may travel mainly from yellow tophytic SI, the pollen grain’s haploid genotype at the
flower to yellow flower or red flower to red flower. S locus prohibits it from growing down the stigma and
When individuals tend to mate with others with dis- style of a sporophyte sharing the same allele By contrast,
similar phenotypes, we refer to negative assortative in sporophytic SI, the diploid genotype of the pollen
mating or disassortative mating. Two common forms parent prohibits it from mating with any sporophyte
of this phenomenon in plants are self-incompatibility sharing either of its S alleles. The two systems act dif-
(meaning that matings can occur only between indi- ferently because of differences in the time the S alleles
viduals with certain genetic differences) and heterosty- are expressed. In gametophytic SI, many matings are
ly (flowers have styles and stigmas of different lengths, “semi-compatible”: if pollen from an S1S2 plant is
and matings within morphs occur only rarely; see Fig- deposited on the stigma of an S1S3 plant, the S2 pollen
ure 6.9). These types of negative assortative mating cause will grow and can fertilize both S1 and S3 ovules. In
obligate outcrossing. sporophytic SI, pollen from an S1S2 plant cannot fertil-
ize any ovules on an S1S3 plant or an S2S3 plant. Thus,
Understanding Plant Mating Systems while self-fertilization cannot occur under either system,
Under what environmental circumstances would these sporophytic SI limits mating with more types of relatives
different mating systems be favored? Complex as plant than does gametophytic SI.
properties. A fruit is just a mature ovary, so even plants
Table 8.1 Ratio of the number of pollen grains produced
per flower to the number of ovules produced such as grasses have fruits, called grains, although these
per flower in species with different levels of fruits contain little maternal tissue beyond the seed itself.
outcrossing Fruits have two major functions: they protect the devel-
oping seeds, and they can affect the dispersal of the
Number Mean pollen:
of species ovule ratio seeds. These functions often conflict with each other, and
Mating system examined (standard error) variation in their relative importance leads to much of
the variation in fruit characteristics.
Cleistogamous 6 4.7 (0.7)
Many fruit characteristics are best understood as
Obligate selfer 7 27.7 (3.1) traits that affect the behavior of animals feeding on fruits
Facultative selfer 20 165.5 (22.1) and seeds. On the one hand, most seeds are quite nutri-
Facultative outcrosser 38 796.6 (87.7) tious, and seeds are therefore important animal foods.
Most human caloric intake, for example, comes from the
Obligate outcrosser 25 5859.2 (936.5)
grains of grasses (for example, rice, wheat, and corn).
Source: Cruden 1977.
There is, therefore, a selective advantage to plants in pro-
tecting seeds from being eaten before they are mature—
otherwise those offspring are simply lost. On the other
Self-pollination and nondispersing seeds often co- hand, after seed maturity, there is often a selective
occur in plants. If a particular genotype is successful in advantage in attracting animal consumers to disperse
a particular spot, it may be advantageous to produce the seeds. Plants have various ways of balancing these
some nondispersing offspring of the same genotype. two functions. Some seeds can pass through animals
This may be part of the explanation for why many plants without being destroyed; for example, while humans
have some cleistogamous flowers (which never open consume vast quantities of wheat and rice seeds, toma-
and always self-pollinate), but evidence for this is equiv- to seeds pass through our digestive tracts without any
ocal. In some species, cleistogamous flowers are located loss of viability.
near ground level; in annual grasses of the genus Amphi- Many fruits attract frugivores (fruit eaters) after
carpum (Poaceae), they are actually underground. More seed maturity. Unripe fruits typically have little in the
familiar open flowers are chasmogamous. Cleistoga- way of attractive odors, flavors, or colors, and sometimes
mous and chasmogamous flowers may be produced these may even be repellent. Moreover, in many taxa,
simultaneously, as in many species of the temperate- unripe fruits contain compounds that act as defenses;
zone herbs Viola (Violaceae), Oxalis (Oxalidaceae), and animals eating them tend to become sick, especially
Impatiens (Balsaminaceae). In some species, cleistogamy when they are eaten in large quantities (Herrera 1982a,b).
appears to be environmentally determined. Flowers of You may have experienced this phenomenon yourself if
the cactus genus Frailea (Cactaceae) are cleistogamous you have ever eaten unripe fruit. As they ripen, how-
except in extreme heat, and so are considered to be fac- ever, most fruits change color, begin to give off attrac-
ultative selfers. In several British plants, including Lami- tive odors, and become palatable through the conver-
um amplexicaule (Lamiaceae) and Hesperolinon (Linaceae), sion of starches to sugars and other chemical and
flowers produced in the dry season are cleistogamous physical transformations.
(Richards 1986). If seed dispersal is to occur, animals must not spend
Plant mating systems often strongly affect the num- too much time feeding at any one plant. If the plant and
ber of pollen grains that are produced by an individual its fruits are too attractive to the animals, the animals will
or a population relative to the number of ovules (the not move away, and the seeds will not be dispersed. How
pollen:ovule ratio). As Table 8.1 shows, self-fertilizing do plants encourage frugivores to move along? First, the
species tend to have much lower pollen:ovule ratios than fruits on an individual plant may mature asynchro-
outcrossing species, and the greater the assurance of self- nously. For example, flocks of birds or troops of monkeys
ing—as in cleistogamy—the lower the ratio. often travel from tree to tree, eating only the ripe fruits.
Second, if ripe fruit remains very mildly toxic, an animal
The Ecology of Fruits and Seeds will eat only a few fruits of a given species at once.
Since most plants are rooted in place, they disperse as Seed Dispersal Patterns
fruits, seeds, or related structures. These dispersal units Seeds can be dispersed by animals, by wind, or, less
are sometimes referred to collectively as diaspores. All commonly, by water. Animals can disperse seeds inter-
flowering plants have fruits, and fruits have a tremen- nally or externally. Seeds can be dispersed singly or in
dous range in size, shape, and number of seeds, char- groups. Furthermore, seeds can be dispersed in time as
acteristics of other tissues, color, odor, and chemical well as space. All of these factors have different effects
162 Chapter 8
on the pattern of seed dispersal. The pattern of seed dis- common among plants that colonize disturbed habitats.
persal is important because it determines the demo- Many of them are annuals that survive by being very
graphic and genetic structure of a plant population, the good dispersers, but poor competitors. We discuss such
amount of competition within and among species, and trade-offs in more detail in the next chapter.
species migration patterns. Maples (Acer, Sapindaceae) are a good example of
There are two modes of dispersal by wind: Some plants that produce fruits that flutter to the ground (Fig-
diaspores float on the breeze, and some flutter to the ure 8.13B). Clearly most maple fruits are unlikely to trav-
ground. Dandelion seeds (Taraxacum, Asteraceae), which el nearly as far as most dandelion seeds. Most maple
children enjoy blowing around, are the archetype of the seeds end up falling very close to the parent tree,
former sort (Figure 8.13A). This form of dispersal is very although a few may be blown a substantial distance.
Carol Augspurger and Susan
Franson (1987) showed that the
(A) Dandelion seed (Taraxacum) (B) Maple fruit (Acer) shapes of diaspores can have a
Plume large effect on how far they trav-
el (Figure 8.14). Even after initial
Seeds
dispersal, significant diaspore
movement can occur. For exam-
fpo ple, seeds of Daucus carota (wild
carrot or Queen Anne’s lace, Api-
aceae) disperse very late in
autumn and can then be blown
Carol Augspurger
across the snow in winter.
Dispersal by animals involves more varied mecha-
nisms and leads to more varied patterns. In species that
rely on animal dispersal through frugivory, the seeds
Seed
have thick, hard seed coats and are either regurgitated
by the animal or passed through the digestive tract in
(C) Cocklebur fruit (D) Mistletoe berries their feces. For the seeds, the latter may be an ideal loca-
(Xanthium strumarium) (Phoradendron) tion because they get dispersed to a location that is pro-
vided with fertilizer. Some fruits contain a mild laxative
to speed their passage through the gut, thereby pre-
venting their digestion and speeding their dispersal.
Some seeds require passage through the gut of an ani-
Berries mal to facilitate germination. In other cases, the fruit may
be eaten and the seeds spit out (as in watermelon: Cit-
rullus lanatus, Cucurbitaceae).
An interesting problem in frugivory was discovered
in the 1970s when scientists noted that Sideroxylon gran-
diflorum (tambalacoque, Sapotaceae), a native of Mau-
ritius, was near extinction. There were only a handful of
individuals left, and these all appeared to be centuries
old. The seeds of this plant have very thick coats, requir-
Figure 8.13 ing scarification (abrasion) to germinate. An early
Seeds often have structures that facilitate particular dispersal hypothesis was that much of this scarification was pro-
mechanisms. (A) Some wind-dispersed seeds, such as those vided by dodos (Raphus cucullatus, large flightless
of Taraxacum (dandelions, Asteraceae), float on the breeze. pigeons), which were driven extinct in the seventeenth
(B) Other wind-dispersed diaspores, such as those of Acer century by hunting by European sailors. As a result of
(maples, Sapindaceae), flutter to the ground. (C) Some ani-
mal-dispersed seeds, such as those of cocklebur (Xanthium, the dodo extinction, the hypothesis went, the seeds were
Asteraceae), are carried externally, stuck to the animal’s fur unable to germinate. Subsequent research has suggest-
or feathers. (D) Other animal-dispersed seeds, such as those ed an alternative culprit: introduced monkeys readily
of Phoradendron (mistletoe, Santalaceae), pass through the eat the fruits before they are ripe, thus eliminating most
gut of a vertebrate before being dispersed. Phoradendron recruitment. Whatever the cause, the species remains
seeds are coated with a sticky mucilage, readily adhering to
the branches of potential host trees; frugivorous birds often near extinction.
rub their cloacas on the branches to remove the seeds from Granivores (animals that eat seeds or grains) some-
their bodies, thus effectively planting the seeds. times act as important seed dispersers. Animals intend-
Artificial fruits
Morphology of
Mass Surface area Shape natural fruit
20 20
30
20
10 (Least) 10
(Lightest) 20
10
10 20 40 60 80
20
20
10 10
30
20
FPO
20 fpo 20
Percentage of fruits dispersed
10
10 10
20 20 40 60 80
20
10
Figure 8.14
10 Carol Augspurger and Susan Franson (1987)
studied the effects of mass and area on the
dispersal of the fruit of the tropical tree Tachi-
20 galia versicolor (Fabaceae). They created 3000
artificial fruits, 200 each of 15 different types,
40 from cardboard of various colors. Each artifi-
10 cial fruit was dropped from the top of a 40-
30 meter tower located in the forest on Barro Col-
(Heaviest) orado Island, a large island in the Panama
Canal. The landing point of each fruit was
20
determined, and then the experiment was
repeated four more times. The figures show,
10 10 for each type of fruit, the proportion traveling
(Greatest) a given distance. Those artificial fruits that
traveled farther were lighter (column 1), had a
20 40 60 80 20 40 60 80 greater surface area (column 2), and were
Distance dispersed (m) shaped more like the real seeds (column 3).
ing to eat seeds often drop them accidentally. Animals nutcrackers have excellent memories, they occasionally
that bury seeds for later use also act as dispersers if they forget where some of their caches are, or they may be
fail to return. Both birds and mammals cache seeds in this killed or driven away, leaving their uneaten seeds to ger-
fashion. For example, Pinus albicaulis (whitebark pine, minate and grow into trees (Lanner 1996).
Pinaceae) in the mountains of the northwestern United Sometimes the mode of animal dispersal can be sub-
States and southwestern Canada is mainly dispersed tle. Many species of ants collect seeds for food. In some
by granivorous birds, especially Nucifraga columbiana, plant species, however, the seeds are attached to a lipid
Clark’s nutcracker. The seeds of this tree are not released body called an elaiosome. Instead of eating these seeds,
from the cones when they are mature, and the tree ants carry the seeds to their nest, remove and eat the
depends on birds to remove and disperse them. The nut- elaiosomes, and toss the seeds outside the nest. Thus
crackers pry the seeds loose from the cones, remove seed dispersal by animals may be highly clumped, with
them, and cache them, storing from 1 to 15 seeds togeth- many seeds ending up in a refuse heap near an ant nest
er in each of many caches just under the soil surface. They or in the dung pile below the feeding tree of a monkey
later retrieve the seeds and eat them. Although Clark’s troop. Other seeds disperse by getting stuck in the hair
164 Chapter 8
or feathers of an animal (Figure 8.13C), which can be a growing in the same spot. Many species growing above
very effective mode of long-distance dispersal. ground have few seeds in the soil. Other species may
While most seeds end up relatively close to the par- exist only in the seed bank at a given time. Species that
ent plant, a small percentage of the seeds may be dis- specialize in colonizing disturbed places by wind dis-
persed long distances, and this long-distance dispersal persal, for example, may exist above ground for only a
can have important demographic consequences. Note, few years following a disturbance.
however, that a long distance for one species could be Seed banks also occur above ground in populations
a short distance for another. When we speak of “long- that have serotinous fruits or cones. Serotinous fruits or
distance dispersal,” we mean that individuals are mov- cone scales are bonded shut with resins and open only
ing far enough from their parent population that they when temperatures are sufficiently high (see Figure 13.4).
can no longer interact with individuals in that popula- Normally this occurs only in response to fire. The seeds
tion. It also helps to have a rough idea of the actual dis- are thus dispersed only under conditions suitable for ger-
tances involved; for many plants, dispersing more than mination (bare mineral soil and increased nutrient avail-
a few hundred meters can move them out of their par- ability) and are protected from granivorous animals at
ent population. other times (Groom and Lamont 1997; Hanley and Lam-
In Chapter 21, we discuss the migration of trees ont 2000; Lamont and Enright 2000). Serotiny is especially
northward in North America following the retreat of the important in many conifers and in a number of species
glaciers. Some tree species are estimated to have migrat- of Australian Proteaceae such as Banksia and Hakea.
ed as fast as 400 meters per year. Since it takes several We actually know very little about how long the
years for a tree to become large enough to produce seeds of many species can remain dormant because the
seeds, migration at this rate must have depended on time scales necessary for such experiments are so long.
very long distance dispersal events. The tail of the dis- In 1879, Edward Beal began a famous experiment at the
tribution of dispersal distributions is therefore quite University of Wisconsin. He took a set of 20 glass jars
important, but estimating the shape of the tail pres- and in each put a mixture of soil and 50 seeds of each
ents some important statistical problems (Portnoy and of 19 common weed species and one crop species. He
Willson 1993). Interestingly, wind-dispersed species then buried the jars uncapped, but upside down. The
migrated faster than animal-dispersed species (see Table idea was to allow moisture, small arthropods, microbes,
21.2); one might expect the reverse, since animal-dis- and so forth access to the seeds, while not allowing new
persed species often appear to have more long-distance seeds to get in. Beal and his successors dug up one jar
dispersal events. every 5 years for the first 40 years, and at 10-year inter-
Colonization of newly disturbed habitats or new vals after that time, and tested the seeds to see whether
islands must also rely on long-distance dispersal. Once they were still viable (Priestley 1986). For 16 species, all
a plant becomes established in such a habitat, short-dis- seeds that germinated did so within 50 years, while the
tance dispersal is responsible for local population seeds of 4 species continued to germinate after that time.
growth. For example, the seeds of Cocos nucifera For two of these species—Malva pusilla (round-leaved
(coconut, Arecaceae) disperse from island to island by mallow, Malvaceae) and Verbascum spp. (Scrophulari-
floating in seawater—one of the few species whose seeds aceae; common mullein, V. thapsus, and moth mullein,
can survive this experience. On a given island, howev- V. blattaria, were not distinguished)—some germination
er, most seeds travel only very short distances, as you occurred among 100-year-old seeds. There were also
might imagine would be so for such large seeds interesting patterns of germination and dormancy. For
(although they travel from tree to ground very fast!). example, seeds of M. pusilla germinated after 5, 20, and
100 years of burial, but not at other time intervals. As the
Seed Banks experiment was designed to look for maximum longevi-
Dispersal in time, as well as in space, can be impor- ty rather than variation in germinability, we cannot esti-
tant for some plant species. Once a seed is buried in the mate the latter. However, half the species showed gaps
ground, it can remain there for years or decades. The in germination like that seen in M. pusilla, albeit not so
seed bank is the collection of seeds in the soil; this term dramatic, suggesting that complex patterns of dorman-
is sometimes used to refer to the seeds of a single species cy may exist in many species.
and sometimes to the seeds of an entire community. The existence of a seed bank can have important
There is a tendency for short-lived plants (such as annu- demographic consequences for a population. It can also
als) to have long-lived seeds, while long-lived plants help to buffer a population against year-to-year varia-
(such as trees) and plants with large seeds tend to have tion in demographic rates (bet hedging; see Chapter 9).
short-lived seeds, although this is not a hard-and-fast Seed banks can also affect evolution in two ways. First,
rule. As a result, the abundances of species in the seed they slow the response to selection by maintaining geno-
bank may have little relationship to those of the plants types produced in previous years. Second, mutation
rates in seeds can actually be substantial, and seed banks adaptations affect pollen transfer, including changes
can thus act as a source of genetic novelty (Levin 1990). in the sizes, shapes, colors, and odors of flowers as well
as in many pollen characteristics. These changes have
led to some remarkable floral adaptations and cases of
Summary
coevolution with specific pollinators, but most polli-
Growth and reproduction are key ways in which plants nation in most species is probably by generalized ani-
interact with the animals and the physical environment mal taxa.
around them through their uptake of nutrients, use of Plants have complex mating systems. The main
space, pollination relationships, and dispersal of fruits causes of this complexity appear to be that inbreeding
and seeds. Plants grow by adding repeated modules sometimes has important fitness consequences, that
such as leaves, flowers, and branches. This mode of male and female floral functions may have different rel-
growth leads to plasticity, allowing plants to vary their ative success rates, and that it is often beneficial to be
growth pattern in response to light availability or other able to adjust the relative allocation of resources to male
environmental factors. There are a variety of ways of and female functions.
using this capacity, and clonal plants use it to explore The dispersal of fruits and seeds is affected by a large
their environments in varied ways—sometimes by fol- number of traits. Again, interactions with animals can
lowing concentrations of resources, and sometimes by be an important mode of dispersal, although plants must
more active “foraging.” accomplish seed dispersal while avoiding seed con-
Both sexual and asexual reproduction are common sumption. Most seed dispersal is local, but rare long-dis-
in plants. Some asexual reproduction is vegetative, but tance dispersal events can be demographically important.
some is from seeds. Most seeds are produced sexually, The seeds of some species remain dormant in the soil for
however, and the evolution of pollen and seeds laid the many years. The movements of pollen, seeds, and fruits
basis for much of the diversification of plants. Many are key causes of spatial patterns in plant populations.
Additional Readings
Classic References Waser, N. M., L. Chittka, M. V. Price, N. M. Williams and J. Ollerton. 1996.
Generalization in pollination systems, and why it matters. Ecology
Darwin, C. 1877. The Various Contrivances by Which Orchids are Fertil-
77:1043–1060.
ized. John Murray, London.
Horn, H. S. 1971. The Adaptive Geometry of Trees. Princeton University
Press, Princeton, NJ. Additional Resources
Charnov, E. 1982. The Theory of Sex Allocation. Princeton University Press,
Princeton, NJ.
Lanner, R. M. 1996. Made for Each Other: A Symbiosis of Birds and Pines.
Barrett, S. C. H. and L. D. Harder. 1996. Ecology and evolution of plant
Oxford University Press, Oxford and New York.
mating. Trends Ecol. Evol. 11:73–79.
Willson, M. F. 1983. Plant Reproductive Ecology. Wiley, Chichester, UK.
Schemske, D. W. and C. Horvitz. 1988. Plant-animal interactions and fruit
production in a Neotropical herb: A path analysis. Ecology 69:1128–1137. Willson, M. F. and N. Burley. 1983. Mate Choice in Plants: Tactics, Mech-
anisms, and Consequences. Princeton University Press, Princeton, NJ.
C H A P T E R
9 Plant Life Histories
W
hy do plant species have such a wide variety of life cycle patterns?
Annual species, such as most weeds and field crops, begin repro-
ducing within a relatively short time after germination, produce many
seeds in one or a few bouts of reproduction, and then die within the year in
which they germinated. Other plants—the giant Lobelia telekii (Campanulaceae;
see Figure 6.12) of Mt. Kenya, Argyroxiphium (Hawaiian silverswords, Aster-
aceae; see Figure 7.6), and Digitalis (foxgloves, Scrophulariaceae; Figure 9.1),
wait for years before they reach a sufficient size for reproducing, then pour
their resources into reproduction, producing seeds in a single bout before they
die. Still other plants, woody perennials such as most trees and shrubs and
herbaceous perennials such as spring-flowering garden plants, must reach a
certain minimum size before reproducing (which can take months to years),
and produce relatively few seeds each time they flower, but survive for a
long time and reproduce many times over their life span.
Organisms that reproduce in a single bout are called semelparous (the
botanical term is monocarpic, because the plant produces flowers once), while
those that reproduce repeatedly are iteroparous (polycarpic). A plant’s sched-
ule of birth, mortality, and growth is called its life history. Variation in plant
life histories has evolved over time, and this variation has important conse-
quences for population dynamics.
In this chapter we discuss schedules of birth, mortality, and growth over
a plant’s life span and the causes and consequences of variation in those sched-
ules—for example, the implications of being perennial versus being annual.
We also discuss the timing, or phenology, of growth and reproduction with-
in a year.
Size and Number of Seeds

The idea of trade-offs caused by limited resources (that increasing one thing
necessarily means decreasing something else) is central to most thinking about
selection on life histories. To see why this is so, ask yourself what would be the
most favorable plant life history if no trade-offs were necessary. Such a plant
would produce unlimited quantities of seeds, each of which was large enough
to maximize its chance of becoming established and growing quickly. New
cohorts of seeds would be produced continually. The plant itself would grow
rapidly and live indefinitely. Clearly, this combination of traits is not possi-
tives over time, why don’t all plants produce huge num-
bers of seeds?
Most plant ecologists studying this question have
focused on the idea of a trade-off between the number
of seeds a plant can make and the sizes of those seeds.
Larger seeds generally contain more endosperm (or
other nutritional stores), and thus generally have a
greater chance of establishment success, than seeds pro-
vided with fewer resources. Large amounts of endo-
sperm in the seeds of some arid-zone perennials, for
example, allow them to sink deep roots quickly, increas-
ing their chances of survival.
Thus, in fitness gained through maturing seeds
(often called a plant’s female fitness), there is a trade-
off in the use of limited resources. Producing more (but
smaller) seeds probably reduces the chances of success
for each seed because smaller seeds typically grow more
slowly. On the other hand, there are likely to be dimin-
ishing returns in provisioning only one seed. The joint
chance of success of, say, two medium-sized seeds can
be greater than the chance of success of a single large
seed. Thus, natural selection on maternal plants favors
those that can divide resources among seeds so as to
optimize the number of descendants left behind. Is there
a single best seed size, or is it best to produce seeds in a
range of sizes? What is the best size or range of sizes?
John Harper (1977) suggested that variation in seed
size is quite limited in most plant species, as was sug-
gested by the data available at the time. However, stud-
ies in the 1980s and 1990s revealed much variation in
Figure 9.1
Digitalis purpurea (foxglove, Schrophulariaceae) is a “bienni- seed size within populations, even when natural selec-
al”—a short-lived monocarpic perennial. These plants take tion was acting to constrain seed size because a partic-
two or more years to reach maturity, then reproduce in a sin- ular size was optimal for a species (Mazer and Wolfe
gle bout in which an enormous fraction of their available 1992; Platenkamp and Shaw 1993; Mojonnier 1998). In
resources is devoted to reproduction. Relatively few plant fact, individual plants often produce seeds of quite var-
species are strict biennials; in most so-called biennials, some
plants reproduce in two years, and some in more. (Photo- ied size (Venable 1985; Winn 1991; see Figure 5.1)
graph courtesy of M. Rees.) Using a graphical model, Christopher Smith and
Stephen Fretwell (1974) showed that there is a single
optimal size for seeds (Figure 9.2) if environmental con-
ditions are predictable. McGinley et al. (1987) showed
ble. Even if we substituted definite numbers for words that Smith and Fretwell’s conclusion holds under fairly
such as “unlimited” and “indefinitely,” it would not be broad circumstances. However, their analysis found that
possible for a plant to maximize all of these components variable seed size is sometimes favored when the envi-
of fitness simultaneously; trade-offs would be necessary. ronment varies over time and the geometric mean of fit-
A good way to see this is to consider the size and num- ness (see Chapter 7) is used as the criterion for fitness.
ber of seeds. The geometric mean gives the average rate of popula-
The number of seeds per reproductive bout can vary tion growth over a series of variable years, and is there-
tremendously. There are some plants that mature only a fore the appropriate measure of fitness.
few dozen or so seeds per flowering episode, such as Individual plants actually produce seeds of variable
Cocos nucifera (coconut palms, Arecaceae). A mature cot- size, as noted above, so this theoretical result seems to
tonwood tree (Populus fremontii, Salicaceae), on the other differ from what we actually see in nature. McGinley et
hand, may produce tens of thousands of seeds per al. (1987) suggested that this difference might be a con-
episode; some orchids may produce hundreds of thou- sequence of evolutionary constraints that limit the abili-
sands to millions of seeds. If natural selection favors ties of plants to reduce variation in seed size. For exam-
those genotypes that leave behind the most representa- ple, many species in the sunflower family (Asteraceae)
PAGE PROOF: 2ND PASS Plant Life Histories 169
1 ral selection is the dominant evolutionary force, it may

not be possible to eliminate variation in seed size with-
in or between plants, because this variation is also deter-
mined by environmental and developmental factors.
In determining seed size, there is an inherent con-
Seed fitness
flict between the fitness of the maternal sporophyte and

Optimal the fitness of the individual seeds. Natural selection on
seed size maternal plants favors those that leave the largest num-
ber of descendants, but it acts differently on seeds: the
favored seeds are those that have the best individual
size—often, the largest. An analogy may be helpful: if a
pair of parents have a large sum of money to leave their
children, they may best assure the success of the largest
0
Small Large number of descendents by dividing the money evenly
Seed size among the children—but an individual child may best
assure his or her success by getting as large an individ-
Figure 9.2 ual inheritance as possible. This sort of competition
Model for finding the seed site that maximizes fitness of the among offspring—and conflict between parents and off-
parental plant. The curve gives the fitness of seeds of a par-
ticular size under particular environmental conditions.
spring—also occurs in plants.
Increasing seed size increases fitness, but above a certain Individual seeds can vary in their efficiency at
size, there are diminishing returns in producing large seeds. sequestering a share of the maternal sporophyte’s
Below a minimum size, the seeds have no fitness. The opti- resources. Genetic differences, positional effects, differ-
mal seed size is the value of the curve tangent to the 45- ences in the timing of fertilization, or environmental
degree line. (After Smith and Fretwell 1974.)
variation can lead to differences in seed size. Selection
on seeds favors increased ability to garner maternal
resources whenever the fitness gained by doing so
have flowering heads with two kinds of flowers, disk flo- exceeds the inclusive fitness lost by depriving siblings
rets and ray florets, which can be constrained to produce of those resources. An individual’s inclusive fitness is
different-sized seeds by their physical shape (Figure 9.3; a measure of how well that individual passes on its
Venable 1985). Seeds within a legume pod can differ in genes as well as how well its relatives pass on theirs,
size because those more proximal (closer) to the mater- weighted by how closely they are related. For example,
nal plant are able to garner more resources than more dis- if every seed in a fruit has the same pollen parent, then,
tal (farther) seeds (Silvertown 1987). It is important to selection favors an individual’s increasing its seed size
realize that other factors besides natural selection deter- if doing so increases its fitness by more than the fitness
mine the outcome of evolution (see Chapter 5). Even lost by two siblings, because, on average, the seeds share
when selection favors a single seed size and when natu- half their genes with one another.
Disk florets
Ray florets
Figure 9.3
Helianthus annuus (sunflower, Asteraceae) is a
widespread North American annual plant that is
widely cultivated for its seeds, but also has many
wild populations. Sunflower inflorescences (also
called heads) are composed of two types of flow-
ers (florets), each of which can produce a single
seed. Disk florets compose the center of the head,
while ray florets—which have a single very large
petal and four very small petals—occur around
the outside of the head. In some species in this
family, seeds produced by different kinds of flo-
rets can differ in size, shape, and dispersal and
dormancy properties. (Photograph by S. Scheiner.)
Classic studies of seed size were conducted by is clear that both the timing of reproduction and the
Edward Salisbury in England (1942) and Herbert Baker amount of resources allocated to reproduction are often
in California (1972). Both of them focused on the eco- genetically variable and have evolved over time. This
logical correlates of seed size. Baker, for example, con- realization leads to the more general problem of the evo-
cluded that plants with larger seeds were characteristic lution of life histories.
of more open habitats, such as deserts and coastal dunes.
These classic studies have stimulated much research, Life History Strategies
partly because this kind of study makes it difficult to dis-
tinguish the effects of selection from the effects of phy- We now consider the more general problem: how does
logeny. For example, seed size may have different eco- selection act on the schedule of reproduction and sur-
logical consequences in different environments, as Baker vival throughout a plant’s life? Martin Cody (1966) sug-
concluded. But it may also be true that open habitats gested that throughout their lives, organisms must
have many large-seeded plants simply because the apportion resources among competing demographic
plants that colonized those habitats are from taxa with functions—survival, growth, and reproduction. How an
large seeds. In an extensive study of the flora of the Indi- organism does this determines its fitness. Many stud-
ana Dunes at the southern end of Lake Michigan (where ies have been conducted using the assumption that these
Cowles conducted his pioneering studies of succession; three demographic functions really are competing. It has
see Chapter 13), Susan Mazer (1989) was able to con- become clear, however, that this is not always the case,
trol for these different effects statistically. She found that because the functions of plant body parts do not corre-
habitat accounted for only 4% of the variance in seed size spond in any simple way to the three demographic cat-
by itself. In the Indiana Dunes, the relationship between egories (see “Reproductive allocation” below). Howev-
seed size and habitat is thus largely caused by the his- er, there is no doubt that life history traits are subject to
tory of the taxa that occur in each habitat. natural selection.
Another factor that can affect selection on seed size Arabidopsis thaliana (Brassicaceae; Figure 9.4) is a
and number is granivory, or seed predation. If a sub- weedy European plant that is now widely naturalized
stantial fraction of seeds are eaten by granivores, there throughout the temperate world. It is perhaps best
can be selective advantages to reducing seed size and known as the “fruit fly” of plant molecular biology—
increasing seed number. While tiny seeds may have
reduced chances of establishment, they may also be too
small to be worth the attention of granivores. There is a
large body of theory concerning animal behavior, called
“optimal foraging theory,” concerned with how animals
should be expected to search for, select, and handle food
items. This theory predicts that granivores should not
pay attention to seeds that are too small, and this pre-
diction is supported by much empirical evidence. In
addition, small seeds, scattered widely, are far more dif-
ficult for vertebrate seed predators such as rodents or
jays to find than are a few large seeds (this is not true for
fungal seed pathogens, however).
Seed size may also be dictated by dispersal strategy.
If seeds are dispersed by wind (see Figure 8.14A), then
there may be an optimal size to maximize the distance
traveled. It is not just the resources invested in the seed
that matter, but also the resources invested in the dis-
persal structure (see Figure 8.15). Variation in seed size
may ensure that seeds are dispersed different distances
Figure 9.4
away from the parent plant, resulting in less competi- Arabidopsis thaliana (mouse-ear cress, Brassicaceae) is a small
tion among siblings. annual mustard. Native to central Europe, it has become nat-
In this discussion of the ecology and evolution of uralized in much of the temperate world. A. thaliana has one
seed size and number, we have assumed that there is a of the smallest genome known in any angiosperm, and
fixed total amount of resources allocated to reproduc- therefore is widely used in genetic studies. Because laborato-
ry strains have been selected for extremely short life cycles—
tion at any time. When this is true, the problem reduces as little as a few weeks—the species is widely used in a
to the trade-off between number and size of seeds. This number of other studies, including a growing number of
assumption is useful in gaining an initial insight. But it ecological studies. (Photograph courtesy of K. Scheiner.)
that is, as an important model organism—because it has and Lawrence Venable (2000) found only a slight corre-
a small genome and reproduces quickly. Under some cir- lation between rain at germination time and subsequent
cumstances, some Arabidopsis genotypes can begin flow- rain during the growing season. This correlation was
ering three weeks after germinating. There is also greatest at mesic sites (that is, favorable microhabitats).
remarkable variation in Arabidopsis life spans. Individu- However, populations at xeric sites (unfavorable micro-
als from some wild European populations live as long habitats) would be expected to experience the strongest
as a year. Arabidopsis from other populations can have selection for being able to predict when adequate rain-
two generations within a year—one of the few plants fall would be available. Clauss and Venable pointed out
that do so. In some places with mild winters, such as the that predictive germination might still be occurring in
mid-latitude U.S. states of Kentucky and North Caroli- these plants, but if so, it is a complicated phenomenon.
na, this plant is a winter annual—germinating in the fall, For example, heavy rain at germination time does pre-
growing slowly through the winter, and flowering in the dict subsequent rain in xeric (but not mesic) sites in El
spring. Where the winters are colder, as in the north- Niño years (see Chapter 18).
eastern United States, it is a spring annual—germinat- Predictive germination may involve factors other
ing in the early spring, growing quickly, and flowering than climate. Annuals in fire-prone shrublands in Cali-
by late spring or early summer. There is a genetic basis fornia are stimulated to germinate by compounds in
for much of this variation within Arabidopsis (Kuittinen smoke (Keeley and Fotheringham 1997). Fire reduces
et al. 1997). Those genotypes with the shortest life his- aboveground biomass and the density of competitors,
tories have been the focus of most research because of and minerals in the ash add nutrients to the soil; there-
their convenience, but evolutionary ecologists are now fore, conditions following fire are more favorable for
exploring the molecular basis of life history variation growth of annuals. Many species classified as weedy
in this species. (ruderal) use cues of recent disturbance as signals for
germination.
Seed Germination
A number of different environmental factors help to trig- Life Span
ger germination. In many plants, temperature plays a key Annuals, as we saw above, are plants in which the veg-
role in regulating germination. For example, winter etative life cycle is completed in less than a year. Indi-
annuals in the Sonoran Desert do not germinate during vidual plants are “born” as seeds; since most annual
the summer rainy season, and the reverse is true of sum- plants have seed banks, this means that most annuals
mer annuals. At least in some plants, some cell mem- that you see growing are actually more than a year old.
branes in seeds undergo temperature-dependant phase In temperate-zone environments, many annuals can be
transitions that render cells either more or less permeable categorized as either winter annuals (like Arabidopsis in
to water, thus allowing (or preventing) imbibition the southern United States) or summer annuals. These
(uptake of water by seeds) depending on the season terms can be confusing. Winter annuals germinate in the
(Bewley and Black 1985). Light also plays a key role in autumn, overwinter as vegetative plants, and flower and
many species. For example, many weedy species require die in the spring. Summer annuals complete their above-
light to germinate; this is why many species germinate ground life cycle in the warm months. Research with
following soil disturbance. There are also species such as annuals in the Sonoran Desert (Mulroy and Rundel 1977)
Eschscholzia californica (California poppy, Papaveraceae) has showed that winter and summer annuals differ con-
in which germination is inhibited by light. siderably in their developmental and physiological
When should a seed germinate? This might seem to traits, and this is probably true of winter and summer
be obvious: it should germinate when conditions are annuals in other environments as well.
favorable. But can plants use environmental conditions Annuals are especially common in certain environ-
at germination time to predict subsequent conditions— ments, including deserts, many dunes, and recently dis-
that is, can plants evolve predictive germination? The turbed sites. Most common herbaceous weeds are annu-
answer depends critically on the predictability of the als. Annuals in warm deserts have been the subject of
environment. In a maritime climate (see Figure 18.15C), much ecological investigation. In years favorable to plant
for example, summers are often warm and wet. Germi- growth, these plants produce many seeds. Subsequent
nation in late spring usually provides a fairly good years may well have poor rainfall, leading to poor ger-
chance of surviving to maturity. In contrast, rainfall in mination, survival, and growth for these plants. It is only
desert environments can be highly unpredictable (see in “good” years that population growth may actually be
Figure 18.20), and many plants that germinate do not positive (see Figure 7.13).
survive to maturity (Fox 1989; Venable and Pake 1999). Semelparity includes plants with a range of life
In a study of the annual Plantago insularis (Plantag- spans. Biennials (Figure 9.5; see also Figure 9.1) are
inaceae) growing in the Sonoran Desert, Maria Clauss semelparous plants that flower after two or more years.
In discussing semelparity, it is important to be clear

as to whether it is the genet ot the ramet that is semel-
parous. Monocarpic ramets may be favored by selection
when production of inflorescences is very costly and it
takes plants long periods to acquire enough resources to
flower. This can be the case when plants are pollen-lim-
ited (see Chapter 8) and large inflorescences receive dis-
proportionately many pollinator visits (Schaffer and
Schaffer 1979). It is important not to confuse monocarpic
ramets with true semelparity: a genet with monocarpic
rosettes (such as an agave (Figure 9.6) or a bromeliad)
can reproduce many times in its life, while a semel-
parous plant (such as a Hawaiian silversword) has a
much riskier life history in that it can reproduce only
once.
In most habitats, it is the perennial plants that give
the landscape its characteristic appearance. The catego-
ry “perennial” includes a great diversity of plants. Some
live for only a few years, while others live for centuries
Figure 9.5
Cirsium canescens (Platte thistle, Asteraceae) is a “biennial.”
(Photograph courtesy of S. Louda.)
Flowering
stalk
The term itself is something of a misnomer, as it takes

most biennial species longer than two years to reach
flowering size, and individuals in most biennial popu-
lations vary in this respect. Semelparous perennials are
plants that live for a number of years before flowering.
There is no firm boundary between biennials and semel-
parous perennials (Young and Augspurger 1991),
although generally biennials die back to ground level in
winter (or the unfavorable season), persisting only
underground, while semelparous perennials have sub-
stantial aboveground structures year-round.
Both theoretical and empirical studies point to two
factors that might select for semelparous life histories: a Nonflowering
critical size for reproduction and diminishing returns ramet
from retaining resources for subsequent bouts of repro-
duction. Biennials and short-lived semelparous plants
often occur in early successional habitats, where canopy
closure after several years causes a decrease in the qual-
ity of the environment. Long-lived semelparous plants Figure 9.6
are characteristic of relatively unproductive but persist- Agaves (century plant, Agavaceae) have semelparous rosettes
ent habitats, such as deserts, many grasslands, and that flower only once, mobilizing all the energy stored during
a lifespan of a few decades to produce a large inflorescence
alpine regions. Factors such as competition for pollina- with hundreds of flowers. The genet itself is iteroparous,
tors may be more important in selecting for semelpari- because vegetative reproduction precedes flowering.
ty in these settings. (Photograph © Frans Lanting/Photo Researchers, Inc.)
or even longer. While most woody plants live for at least mathematically straightforward to show that selection
a few years, the reverse is not necessarily true. Some under logistic growth always favors increases in both
herbaceous plants, such as many tundra plants, live for parameters (Boyce 1984; Emlen 1984).
centuries. Among forest trees there are species that live Using the assumption that high population densi-
for decades and others that live for centuries, even in the ties lead to reduced survival of juveniles, Eric Pianka
same type of habitat. In pine flatwoods in the southeast- (1970) developed detailed predictions for animal popu-
ern United States, for example, Pinus palustris (longleaf lations, which were later extended to plants: K-selection
pine, Pinaceae) can live for 300 or more years, while Cer- should favor delayed reproduction and a reduced num-
sis canadensis (eastern redbud, Fabaceae) lives for only a ber of offspring, while r-selection should favor early
few decades. Typically, long-lived perennials have short- reproduction and an increased number of offspring. This
lived seeds and a small or nonexistent seed bank. Most prediction is what most people mean when they refer to
perennials, however, have the ability to become dormant r- and K-selection: weedy plants, for example, are taken
as vegetative plants during winter or the dry season. to be r-selected, while forest trees are taken to be K-
Perennials have more ability to withstand variable and selected. As Boyce (1984) put it, references to “r strate-
harsh environmental conditions in their aboveground gists” generally mean small organisms that reproduce
phase than do annuals; this ability increases with the quickly and are poor competitors, and thus should be
longevity of the species (long-lived temperate trees, for favored in disturbed habitats, while references to “K
example, are more likely to withstand a hard freeze or a strategists” generally imply that organisms have the
drought than a short-lived herbaceous perennial would opposite characteristics, although these characterizations
be). were not derived from the original model.
How do we explain all of these different life histo- Perhaps the greatest problem with r- and K-selec-
ries, as well as the variation within each? What are their tion theory is that its predictions are qualitative and rel-
consequences for population growth? Early efforts at ative, rather than quantitative, and so are difficult to test.
such explanations focused on the theory of r- and K- Plants that are r-selected, for example, are predicted to
selection, while recent work has centered on demo- evolve an earlier age at first reproduction and greater
graphic approaches. seed numbers than those that are K-selected. But what
can one measure to test this? Plants within one com-
r- and K-selection munity might be classified as r-strategists or K-strate-
Robert MacArthur and his students and colleagues gists based on which reproduced at an earlier age, but
developed their ideas about life history evolution, this classification might be different in another com-
focused on the concept of r- and K-selection, during the munity, or in a comparison of one of these species with
1960s. This work stimulated much of the early research another species. What is more, such a classification says
on the evolution of life histories. The concept of r- and nothing about whether the differences among the
K-selection plays little role in current research on life his- species are actually a result of selection on age of first
tories. Nevertheless, it continues to be influential in pop- reproduction, and certainly not whether the differences
ular writing on ecology, in textbooks, and in some are part of an entire life history strategy. Finally, the
aspects of community ecology, so we explain it briefly emphasis on the distinction between plants that live
here. mainly in disturbed habitats (r-strategists) and those that
Using the logistic model of population growth, live in stable habitats (K-strategists) may have been mis-
placed. Many “stable” habitats, such as mature forests,
dN = rN  K – N 
dt  K  have been shown to depend on disturbance for their
maintenance (see Chapter 13). Previous notions of cli-
where r is the intrinsic rate of population growth, K is max communities and stable equilibria have changed
the carrying capacity, and N is the population size, dramatically in recent years, reducing the utility of this
MacArthur and Wilson (1967; MacArthur 1972) sug- equilibrium-based theory.
gested that at low population densities, selection will be
strongest on traits that increase the intrinsic rate of pop- Grime’s Triangular Model
ulation growth, r. At high densities, they suggested, Responding to some of the limitations of r- and K-selec-
selection will be strongest on traits that increase popu- tion theory, plant ecologist Philip Grime (1977, 1979) pro-
lation size, or the carrying capacity, K. Loosely, they posed an extension of that theory, classifying life histo-
argued that there is stronger selection for productivity ries as being influenced primarily by selection for one of
than for efficiency at low densities, and vice versa at high three traits: colonizing ability (which he called R selec-
densities. Selection for increasing r and K was widely tion, for ruderal plants), competitive ability (C selection),
interpreted as counterposed. There is, however, no rea- or stress tolerance (S selection). Grime’s theory was not
son why there must be a trade-off between r and K. It is derived from any particular model of population growth
or evolution. He assumed that there are necessary trade- amount for the annuals to have a greater lifetime fit-
offs between these syndromes. In fact, Grime (1979) ness—thus resolving the paradox.
pointed out that plants could have any combination of
traits associated with R, S, and C selection. Like r- and K- Reproductive Allocation
selection theory, then, Grime’s theory made predictions If it matters when a plant reproduces, and how much it
that do not lend themselves to critical tests. While C-S- reproduces at each life stage or age, the obvious ques-
R theory made assumptions and predictions about life tion becomes, how does selection act on the allocation
histories, it had a particular emphasis on the relative of resources at each stage? This question, especially with
importance of competition in different habitats (see the added assumption of a trade-off between reproduc-
Chapter 10). Consequently, this theory has received a fair tion and survival, has provided the focus for much
amount of attention and generated considerable contro- research in life history evolution.
versy among plant community ecologists, but has not Key to addressing this problem is the concept of
had much influence in the study of life history evolution. reproductive value (v; see Chapter 7). William Schaffer
and Michael Rosenzweig (1977) showed that fitness is
Demographic Life History Theory maximized at every age if the sum of present reproduc-
The demographic approach to life history theory centers tion plus future reproduction weighted by relative repro-
on studying how variation in survival and fecundity as ductive value,
individuals vary in size, age, or other life history stages pv
affects population growth. Its utility can be seen in its Fi + i i + 1
vo
solution of “Cole’s paradox.” In studying the conse-
quences of schedules of reproduction, Lamont Cole is maximized at every age. The fraction of resources allo-
(1954) developed a model that compared the fitnesses of cated to reproduction at age or stage i is the reproduc-
a perennial phenotype with an annual phenotype in the tive effort at that age or stage. Schaffer and Rosenzweig
same species. If a fraction p of the plants survive to analyzed selection on life histories graphically by plot-
reproduce each year, and produce F seedlings if they do ting current and future reproduction against reproduc-
reproduce, then the perennial’s fitness is λP = p(F + 1), tive effort.
while the annual’s fitness is λA = pF. Cole concluded that Figure 9.7 shows two simple alternatives. In Figure
annuals need only increase their reproductive output by 9.7A, the curves for both present reproduction (Fi) and
one new seedling per year to do as well as perennials. future reproduction [(pivi+1)/v0] are convex, so their sum
He pointed out that this finding was paradoxical is maximized when reproductive effort is either 0 or 1 at
because most organisms are perennial, not annual, sug- any given age or stage. These curves can differ for differ-
gesting that there must be other advantages to being ent ages or stages. Since a reproductive effort of 1 means
perennial. putting all resources into reproduction and none into sur-
Using a demographic approach, Eric Charnov and vival, this graph describes selection for a semelparous life
William Schaffer (1973) pointed out a flaw in Cole’s rea- history. In Figure 9.7B, both curves are concave, so selec-
soning: his result depends on the assumption that rates tion favors intermediate reproductive effort at every age
of survival and fecundity do not depend on age. If his after the onset of reproduction—an iteroparous life histo-
model is modified slightly to give first-year plants a dif- ry. Combinations of curves of different shapes can thus
ferent chance of surviving than older perennials, the par- describe a wide variety of life histories.
adox disappears. Charnov and Schaffer demonstrated Testing this kind of theory has proved difficult. One
this as follows: If the fraction of first-year plants sur- problem, pointed out by Schaffer and Rosenzweig
viving is c, and the surviving annuals produce FA seeds (1977), is that in many cases more than one life history
while the surviving perennials produce FP seeds, then is optimal. But there is a deeper problem in actually
λP = cFP + p and λA = cFA. Thus the population growth measuring reproductive effort: How does one assign the
rates of both phenotypes are given by the chance of indi- fraction of resources used for reproduction? Little is
viduals surviving to reproduce times their fecundity if known about the underlying developmental and bio-
they do survive; for the perennials, this rate is increased chemical processes of reproduction. An early study
by the chance of surviving from year to year. Therefore, (Hickman and Pitelka 1975) found that the dry weight
the annuals have greater fitness only when λP < λA, that of plant parts, except for some unusual tissues, is high-
is, when FA > FP + p/c, or when their fecundity exceeds ly correlated with their energy content. This result led to
the perennials’ fecundity plus the survival chances of hundreds of studies comparing the dry weights of plant
adults relative to those of juveniles. This means that parts under varied circumstances.
where survival of adults is much greater than survival But how does one assign a plant part to “current”
of juveniles, the annuals’ reproductive output would versus “future” reproduction? For flowers and fruits, the
need to exceed that of the perennials by a considerable choice is clear, but this is not so for organs such as roots.
(A) Semelparity sophisticated mathematics, such as optimal control the-

ory, which can handle the technical problem of opti-
mizing reproductive allocation throughout the life span.
Present or future reproduction
pivi + 1 Additional work along these lines has been slow because
Fi + vo
meaningful predictions require that the models be much
more biologically realistic. Further work in develop-
mental biology and physiology is thus needed to make
pivi+1 Fi much more progress in this approach to life history the-
ory.
Matrix models (see Chapter 7) play an important
role in both theoretical and empirical studies of life his-
0 1 tories. In the simple case of the Charnov-Schaffer model
Reproductive effort at age i of perennial life histories, the original model stated that
the numbers of juveniles and adults (J and A, respec-
(B) Iteroparity fpo tively) next year will be Jt+1 = cFPJt + pFPA and At+1 =
cJt + pAt. Rewritten in matrix form, this is
pivi + 1  J t + 1  cFp pBp   J i 

A  =  × 
Present or future reproduction
Fi + vo (9.3)
 t + 1   c p   Ai 
pivi+1
vo Fi Analysis of this model tells us that the population will
eventually grow at the rate λP = cFP + p, the dominant
eigenvalue of the matrix. At that time, the population
will be composed of FP times as many juveniles as
adults. In this simple case, it is easy to see the conse-
quences of varying the survival and fecundity rates. For
example, increasing juvenile survival c increases λP at
0 1
the rate FP. With more realistic models, it is not so easy
Reproductive effort at age i
to do this sort of thing visually. However, the methods
described in Chapter 7 make it possible to completely
Figure 9.7 analyze the effects of varying any survival or fecundity
Selection for semelparity versus iteroparity. Fitness is great- rate on population growth or composition.
est when the sum of present reproduction plus future repro-
duction (green curves) is maximized at every age. (A) Semel- Bet Hedging in a Variable Environment
parity is favored when the curves for both present and
future reproduction are convex—fitness is greatest when Year-to-year variation in survival and fecundity also
reproductive effort is always either 0 or 1. (B) Iteroparity is affects fitness. How can fitness variation among years
favored when both curves are concave—fitness is greatest at be reduced when some of that variation is inherently
intermediate reproductive effort once maturity is reached. unpredictable? Many plants appear to achieve this by
(After Schaffer and Rosenzweig 1977.)
“hedging their bets” (spreading the risk) among years.
To see why variation among years changes fitness, com-
pare a hypothetical annual species that always produces
One attempt at handling this problem was a study by an average of 1.5 seeds every year (that is, half of the
Edward Reekie and Fakhri Bazzaz (1987). These inves- plants in the population produce 1 seed, and half pro-
tigators measured the amount of root mass present dur- duce 2) with another annual that produces 0.5, 1, and 3
ing flowering of a grass and the amount present at other seeds in dry, average, and wet years. If these types of
times, and found no difference. They concluded that years occur with equal frequency (1/3), the average
none of the root mass counted toward reproduction. number of descendants of the variable species will be
The problem with testing theory of this kind is that 0.51/3 × 11/3 × 31/3 ≈ 1.14, or 0.36 fewer than the number
the parts of a plant do not relate in any simple way to produced by the invariant species. In general, increas-
their demographic consequences. Moreover, reproducing the variance in fitness among years decreases its
tive effort is a ratio. As such, it does not uniquely long-term geometric mean. This implies that there is a
describe the life history (Schaffer 1983) because there are trade-off between mean and variance in fitness: gaining
many different life histories that could yield a meas- added fitness in “good” years can come at the expense
urement of, say, 75% allocation to reproduction at a par- of reducing long-term average fitness. For matrix mod-
ticular time. This observation led to models using more els, the problem is somewhat more complex, because the
order of multiplication matters. Analyses of both struc- ing aspen tends to be an early successional species (see
tured and unstructured population models suggest that Chapter 13). We cannot tell whether the growth rate ver-
selection often acts to increase fitness by reducing sus wood density trade-off is inherent in the biochem-
among-year variation in fitness. istry of wood, or if the right mutation has simply not
Spreading reproduction among more years or more come along, or if selection always favors just these two
evenly among years, increasing the area over which combinations of traits.
seeds are dispersed, and increasing dispersal through Thus, we need to know more about variation among
time by means of seed banks are all examples of bet individuals in resource availability, as well as about the
hedging. Seed banks are perhaps the best studied of mechanisms determining resource allocation to differ-
these mechanisms. Most annual plants have seed banks. ent functions. Using a genetic model, David Houle
Without environmental variation, among-year seed dor- (1991) showed that even when trade-offs do exist, genet-
mancy is opposed by selection to germinate immedi- ic correlations between life history traits could evolve to
ately, since seeds that remain dormant in the soil cannot be either negative or positive, depending on the genet-
reproduce, but those that germinate can reproduce. A ic details. This finding points to the importance of
number of studies (e.g., Brown and Venable 1991; Kalisz increasing our understanding of the biological basis of
and McPeek 1993) suggest that seed banks play an life history traits.
important role in buffering populations from environ-
mental variation. Phenology: Within-Year Schedules
Difficulties in Measuring Trade-Offs of Growth and Reproduction
There are an enormous number of studies comparing Phenology—the timing of growth and reproductive
species or higher taxa that show a negative correlation activity within a year—can vary greatly among species,
between traits such as longevity and fecundity. Never- populations, and even individuals. In temperate cli-
theless, a substantial number of studies at the popula- mates, there are usually some plants flowering at all
tion level—where selection must be acting on these times between the first and last frost. Some woody
trade-offs—not only fail to show a negative correlation, species are evergreen, while others are deciduous. Even
but often show the opposite. This kind of result led a few within groups that we usually think of as all of one
scientists in the 1980s to argue that there are really no type—such as conifers, which we think of as evergreen,
trade-offs. The problem was clarified when Arie van there may be exceptions—such as the deciduous conifer
Noordwijk and Gerdien de Jong (1986) pointed out that species Larix laricina (the larch or tamarack, Pinaceae;
a positive correlation between, say, reproductive alloca- Figure 9.8). In general, plant phenologies are constrained
tion and survival-related traits could actually be expect- by seasonality, mainly by temperature or moisture avail-
ed if the total amount of resources available varied ability. Phenologies have a more complex relationship
among individuals. They suggested a useful analogy: with animal pollinators, seed dispersers, and herbivores
for people operating under a fixed income, the amount because the plants and animals can be agents of selec-
of money available to spend on housing is negatively tion on each other.
correlated with the amount available to spend on cars.
Nevertheless, if you looked at the data, you would find Vegetative Phenology
that people who spend more on housing also spend In temperate deciduous forests, many herbaceous plants
more on cars—because they have greater total wealth. on the forest floor expand their leaves and flower before
The trade-offs are real, but they operate in the context of the canopy trees begin leaf expansion. The consequence
variable access to resources. is that these plants do most of their growing and repro-
A critical distinction must be made between those ducing in temperatures that are far colder than those
traits that must be traded off because the physical rules experienced by the canopy trees during the same life
of the universe demand it, and traits that are traded off stages. Many studies have shown that this timing is actu-
because selection has molded species that way. An exam- ally advantageous to the forest-floor herbs. Once the
ple of the first category is resource allocation to leaves canopy closes, very little sunlight—sometimes as little
versus flowers: limiting resources (carbohydrates, nitro- as 1%—gets through to the forest floor. Thus, the for-
gen, or water) used to produce and maintain leaves can- est-floor herbs’ growth and reproduction would be even
not also be used to produce flowers. A possible example more limited by light availability than they are by cool
of the second category is an apparent trade-off between temperatures.
speed of wood growth and wood toughness. Fast-grow- In most temperate species, it appears that tempera-
ing trees such as quaking aspens (Populus tremuloides, ture and photoperiod (day length) are the main factors
Salicaceae) have weak wood, and individual trees gen- determining vegetative phenologies. It is important to
erally live only 50 or 60 years. Consequently, the quak- realize that temperature plays this role in a particular
As the example of temperate forest-floor herbs sug-

gests, there are at least two main types of selective fac-
tors that act on vegetative phenologies. Abiotic factors
limiting growth—such as the timing of killing frosts or
seasonal droughts—often determine the beginning or
end of growth episodes (or both). Biotic factors—in par-
ticular, competition for light or water—are quite impor-
tant for many species.
This observation raises an interesting question:
Why do the canopy trees in temperate forests wait so
long before expanding their leaves? Why don’t they use
the early spring to add to their growth, as the forest-
floor herbs do? There are a number of selective forces
affecting the timing of leaf expansion. First, the canopy
is elevated, and temperatures in the treetops can be con-
siderably colder than those at ground level (see Chap-
ter 3). Thus, leaf expansion for trees occurs later on the
Figure 9.8 calendar, but may not really be much later in terms of
Larix (the larch, Pinaceae) is a deciduous conifer found in degree-days. Second, late frosts are a common occur-
taiga around the world. Left: L. kaempferi (ochudo) at 3500 m
elevation on Mt. Fuji, Japan. Right: A branch of L. laricina
rence. Leaves on trees are much more vulnerable to frost
(tamarack) in a bog at 200 m elevation in northern Michigan. damage than those on forest-floor herbs because the lat-
(Photographs by S. Scheiner.) ter are partly sheltered by the trees and because the
ground-level temperatures are higher. Third, all meta-
bolic enzymes have defined temperature ranges over
which they can operate and are most efficient at par-
way. It is generally the number of degree-days—the sum ticular temperatures. Enzymes adapted for peak func-
of temperatures experienced over some period of time— tioning at warm temperatures are unlikely to be efficient
and not the temperature on a particular day that deter- in the early spring; it is possible that earlier leaf expan-
mines the timing of leaf expansion. In cold years, then, sion might reduce the total annual growth of the trees,
leaf expansion (sometimes called “leaf flushing”) is not increase it. Finally, many temperate-zone trees are
delayed. Plants, like many other organisms, often use wind-pollinated, and pollination in most of these
photoperiod (see Chapter 2) as a reliable predictor of the species occurs while the trees are leafless. The presence
average temperature. If they relied exclusively on tem- of leaves earlier in the season would be likely to limit
perature, a warm spell in midwinter would typically pollen transfer.
cause many plants to expand their leaves. By using tem- The phenologies of temperate forest-floor herbs are
perature as a cue, plants can respond to an early spring actually more complex than we have implied so far.
by flushing their leaves early, but because they also use Most of these plants in northeastern North America have
photoperiod as a cue, this response is limited. These phenologies like those we have described, but there are
plants, in other words, respond to environmental fluc- others that are able to capture and use light at other
tuations, but they do so cautiously. times. For example, some species can use light flecks on
the forest floor to grow during the warmer months (see 10,000
Figure 2.26). Still others use the additional light avail-
able in autumn, when some canopy species have begun
Relative growth rate per week

shedding their leaves (Mahall and Bormann 1978).
In warmer habitats, it is usually the availability of 1000
moisture that has the strongest effect on vegetative phe-
(mg g–1)
nologies. In tropical dry forests and tropical rainforests,
as well as temperate and tropical deserts, rainfall is sea-
sonal. Where it is strongly so—as in deserts and tropical
100
dry forests—many species are drought-deciduous,
shedding their leaves during dry periods. Is the timing
of leaf expansion and leaf drop a plastic response to the
environment or an evolved strategy based on the aver-
age environmental conditions? To a great degree, the 10
1 10 100
answer is “both.” There are some species that retain their
Leaf life span (months)
leaves if given additional water, and others that shed
them at the usual time of the onset of the dry season, Figure 9.9
regardless of current water availability. There are species The relationship between the relative growth rate (RGR) of
that expand their leaves weeks before the rains begin young plants and the life span of leaves is negative. Species
with long-lived leaves tend to have low RGR. (After Reich
and others that do so only after abundant rainfall. All of et al. 1992.)
these strategies are evolved traits. Perhaps the most
extreme case of a drought-deciduous phenology is that
of Fouquieria splendens (ocotillo, Fouquieriaceae; see Fig-
ure 3.5). This desert plant of the southwestern United tions. A number of pine and juniper species are char-
States and Mexico is drought-deciduous. However, it acteristic of the margins of North American deserts.
produces new sets of leaves whenever water is available, Similarly, evergreen oaks (Quercus, Fagaceae) in North
as long as temperatures are high enough. Thus ocotillos America are found where soils are very poor in nutri-
can have four or five sets of leaves over a year. ents, such as the coastal plain of the Gulf of Mexico, and
on the edges of deserts and some grasslands. Oaks from
Evergreen versus Deciduous Habits higher elevations and more northern latitudes are typ-
Why do plants shed their leaves at all? At first glance, ically deciduous.
it might seem more advantageous to be evergreen, so The leaves or needles of an evergreen may live from
that the plant could photosynthesize whenever light was one year to over forty years (Aerts 1995), although more
available. In actuality, there are major trade-offs involved commonly they last between one and four years. A neg-
in being evergreen versus being deciduous. The leaf ative relationship has been demonstrated between leaf
traits of evergreens include being thick, tough, waxy, and life span and the growth rate of young plants (Figure
generally reduced in area. Although these traits reduce 9.9). Leaf life span is also negatively correlated with pho-
the rate at which the plant loses water and nutrients tosynthetic rate, leaf nitrogen concentration (on a weight
(Aerts 1995), they also reduce the plant’s rate of net pho- basis), and specific leaf area (leaf area per unit leaf
tosynthesis (see Chapter 2). Thus, they are advantageous weight). It has been hypothesized that the costs of con-
under only some conditions. structing evergreen leaves are higher than those for
Evergreens are typically found in nutrient-poor deciduous leaves. This hypothesis has been shown not
soils, as well as in habitats that are subject to drought to be true; rather, different evergreen species have dif-
over much of the year. For example, pines and other ferent leaf construction costs due to different ratios of
conifers in North America are most common in acidic biochemical constituents in their leaves (Figure 9.10).
soils, which have low nutrient availability, and at high Evergreen leaves tend to have relatively high invest-
elevations and latitudes, which are subject to tempera- ments in materials that protect the leaf and contribute to
ture extremes. They are also found in sandy or shallow, its long life span, such as lignin and fibers, and lower
rocky soils (which also tend to be acidic and low in investments in materials that contribute to high photo-
nutrients). They frequently experience drought during synthetic and growth rates. Deciduous leaves tend to
winter because of the limited water-holding capacity of have high investments in the latter materials, particu-
such soils or, in northern and alpine habitats, because larly photosynthetic enzymes and other proteins, and
most of the water is in the form of snow. They also lower investments in lignin and fibers. Overall, the car-
undergo summer drought in many such habitats, either bon costs of producing leaves are roughly comparable
because of limited summer rains or due to soil condi- in evergreen and deciduous plants.
(A) Figure 9.10

Carbon costs of the chemical components of leaves in 13
100 species of tundra plants. Costs vary substantially among
species and growth forms; for example, fiber and lignins
Concentration (%)
are relatively expensive for mosses, and lipids are rela-

tively expensive for evergreen shrubs. Deciduous leaves
50 contain somewhat more protein, most of which is RuBP-
carboxylase (rubisco), which is essential to photosynthe-
sis. There is no evidence in these data that evergreen
leaves are more costly to produce. Species shown are
Salix pulchra (Sp), Betula nana (Bn), Vaccinium uliginosum
0 (Vu), Arctostaphylos alpina (Aa), Rubus chamaemorus (Rc),
Pedicularis capitata (Pc), Ledum decumbens 1st and 2nd
year leaves (L1 and L2, respectively), Vaccinium vitisidaea
(B)
800
fpo (Vv), Dryas octopetala (Do), Eriophorum vaginatum (Ev),
Carex bigelowii (Cb), Aulocomnium turgidum (At), and
Carbon cost (mg C g–1)
700 Hylocomium splendens (Hs). (After Chapin 1989.)

600
500
400
300 Protein
200 Lignin
100 Tannin
0 Lipid
Sp Bn Vu Aa Rc Pc LdVvDo Ev Cb At Hs Carbohydrate
Deciduous Forbs Evergreen Graminoids Mosses Fiber
shrubs shrubs
As a consequence of retaining their leaves for long begins flowering while still covered with snow (Galen
periods of time and having low tissue nutrient concen- and Stanton 1991). The North American desert annual
trations, evergreen plants tend to have low nutrient loss Eriogonum abertianum (Polygonaceae, Figure 9.12) has a
rates—that is, they excel at retaining the nutrients they remarkably variable reproductive phenology: individu-
obtain from the soil. This ability appears to confer sub- als may flower in either or both of two rainy seasons sep-
stantial advantages in low-nutrient soils, allowing these arated by as much as six months of drought (Fox 1989).
plants to achieve annual dry matter production at least In both of these plants, flowering time is affected by
equal to that of deciduous species under nutrient-limited physical factors: temperature and light availability in
growth conditions. In fertile soils, evergreen species can- Ranunculus and temperature and moisture availability
not accumulate as much total growth as deciduous species. in Eriogonum. Both plants are unable to reproduce in
Litter decomposition in both nutrient-poor and arid winter. Ranunculus is covered by deep snow then, while
soils is quite slow. Because conifer needles and oak Eriogonum germinates only during the cool desert win-
leaves decompose very slowly and tend to acidify soils ter, so it is usually much too small to flower in winter. In
under moist conditions, the presence of large numbers both cases, there is apparently strong selection to flower
of these trees tends to favor plants with leaf traits that at what appear to be unusual times.
reduce nutrient loss. Thus, on nutrient-poor soils, the Eriogonum may flower during either the spring or
evergreen habit tends to be self-reinforcing (Aerts 1995). the summer rainy season, or both. Like most plants, Eri-
Indeed, the dominance of conifers in boreal forests plays ogonum produces more flowers and fruits the larger it
a key role in the acidification and nutrient depletion of gets—and it gets very large only after the summer rains.
these soils (see Chapter 4). This phenomenon is unlike- The fact that many or most individuals, depending on
ly to be important in arid soils. site and year, die before summer does not necessarily
provide sufficient selection for earlier flowering. For
Reproductive Phenology: Abiotic Factors example, at one site, well under 1% of the plants alive in
While we often think of flowers as something we see in the spring managed to survive and flower in the sum-
the spring, plants can flower under an amazing range of mer, but they accounted for about 25% of the seeds pro-
conditions. Ranunculus adoneus (snow buttercup, Ranun- duced (Fox 1989).
culaceae) is a plant of alpine snowfields in the Rocky In both species, there is substantial variation in flow-
Mountains of Colorado (Figure 9.11). This perennial herb ering time within a population. Much of this variation
1988, r = 0.97, P < 0.001

1989, r = 0.92, P < 0.001
90
Distance from edge of snow bowl (m)
84
72
60
48
36
24
12
ture lead to big differences in reproductive timing. Plants
0 that reach sufficient size and have access to sufficient
1 5 9 13 17 21 25 29
Date of flowering (July) moisture begin flowering in the spring; these plants have
a greater chance of surviving to the summer rainy sea-
Figure 9.11 son than plants that do not flower in the spring (Figure
Ranunculus adoneus (snow buttercup, Ranunculaceae) is 9.12). Despite this plasticity, it is clear that differences
found in alpine snowfields in the Rocky Mountains; it begins
flowering while still covered by snow. Flowering time between populations also reflect local adaptation. Pop-
depends on the plant’s position in a snowfield. Plants on the ulations that experience a greater risk of dying before
edge of the snowfield tend to flower earlier. (Data after Galen summer have more early-flowering plants than those at
and Stanton 1991; photograph courtesy of M. Stanton) a lesser risk, (Table 9.1), and there is much genetic vari-
ation for life history traits between these populations
(Fox 1990a).
It is widely believed that moisture limitation—and
appears to be due to phenotypic plasticity rather than perhaps other types of physiological stress—tends to
genetic variation within populations (Fox 1990a; Stan- induce flowering in many plants. There is no evidence
ton and Galen 1997). This is especially clear in Ranun- for this effect in annual plants (Fox 1990b). In perenni-
culus. Plants on the edge of the snowfield flower much als, the problem is more complicated, because flower
earlier, mainly because the snow melts earlier there, so buds are often formed one or more years before they
that they receive more light earlier in the year (see Fig- open. The effect of drought or other stress therefore
ure 9.11). With Eriogonum, small differences in soil mois- probably depends on its timing. Drought that occurs
Table 9.1 Frequencies and sizes of spring-flowering classes of Eriogonum abertianum in several
populations
Plants alive in
spring that flowered Mean (standard error) height to highest node
Site and year in spring (%) Spring flowerers Non-spring flowerers n
Chihuahuan Desert
Portal 1985 2.6 4.4 (0.58) 2.2 (0.05) 494
Portal 1986 2.6 4.8 (0.43) 1.7 (0.03) 1404
Sierra Vista 1986 0.0 — 1.3 (0.04) 200
Sonoran Desert
Tucson Mts. 1986 56.7 2.8 (0.10) 1.7 (0.06) 201
Waterman Mts.1986 31.0 4.2 (0.43) 2.0 (0.05) 200
Organ Pipe 1985 21.3 2.5 (1.9) 1.5 (0.06) 150
Organ Pipe 1986 23.1 3.1 (0.7) 1.7 (0.02) 1382
Source: Data from Fox 1992.
Note: More plants flower in spring in the Sonoran Desert (which receives less rain, and less predictable rain, in the summer) than in
the Chihuahuan Desert. Spring-flowering plants are larger on average, and have improved chances of surviving to summer (see Fig-
ure 9.11), even when compared with same-sized plants (Fox 1989). Moisture availability, as well as size, strongly influences pheno-
logical variation within populations; the differences between the Portal and Organ Pipe populations, at least, have a strong genetic
basis (Fox 1990a).
Spring-flowering plants
1
Cumulative survivorship
0.1 fpo
Plants not flowering
0.01 in spring
0.001
0 50 100 150 200
Days from marking
Figure 9.12
Eriogonum abertianum (Polygonaceae) is an annual plant of the Sonoran and Chi-
huahuan Deserts of the southwestern United States and Mexico. Its flowering time
depends on plant size and moisture availability. Therefore spring-flowering plants
are more likely to survive to the summer rainy season, when most seed is set.
These data are for a population at Organ Pipe Cactus National Monument, Ari-
zona, in 1985. (Data from Fox 1989; Photograph courtesy of K. Whitley.)
when buds are forming may have a different effect than Poole and Beverly Rathcke (1979) showed that many of
drought that occurs just as the buds are opening. the claims that observed phenologies showed strong
How tightly is reproductive phenology coupled correlations among plant species were based on inap-
with vegetative phenology? Evidence for temperate for- propriate data analyses. For example, many of the
est trees suggests that their coupling is generally weak effects that had been claimed disappeared when Poole
(Lechowicz 1995). The main reason is that plants can and Rathcke took into account the fact that the seasons
store resources such as photosynthate and remobilize limit when plants can flower. If one analyzes temper-
them at other times. This mechanism is what allows so ate-zone data without accounting for winter, or tropi-
many temperate trees to flower when they are leafless. cal data without accounting for dry seasons, it is easy
to wrongly conclude that many species’ flowering times
Reproductive Phenology: Biotic Factors are clustered more than one would expect by random
Biotic factors may also influence reproductive phenol- chance.
ogy. Because of the inherent difficulties in studying mul- Frugivores (fruit-eaters) and granivores (seed
tispecies interactions, many studies of these influences eaters) may also exert considerable selection on phe-
have relied heavily on circumstantial evidence—but the nologies. In most cases, however, it appears that the ani-
circumstantial evidence in some cases is quite strong. mals’ phenologies are more strongly dependent on plant
If pollinators are only briefly available, there may be phenologies than the reverse. Animals may also act as
strong selection for flowering while these animals are important causes of phenotypic plasticity in reproduc-
present and active. Although this seems clear in princi- tive phenology. For example, herbivory (like many other
ple, it is important to realize that such selection will be stresses, including crowding and drought) may cause an
strong only if the plants require animal pollination and individual’s flowering time to be delayed (Lyons and
seed set is limited by the availability of pollen. Some Mully 1992).
plants that are mostly self-fertilizing still require animals Finally, other plants may influence reproductive
for pollination. phenology, either directly or indirectly. Crowding by
A number of authors in the 1970s inferred that flow- neighbors, for example, may act to delay flowering
ering schedules of entire communities were structured (Lyons and Mully 1992). Such delays, in turn, may result
by competition for pollinators. They concluded that this in selection for faster growth or earlier flowering.
competition was a crucial factor in shaping phenologies,
causing the evolution of separation in phenologies Summary
among overlapping species. Other authors suggested
that many plant species act as mutualists by having sim- Plant life histories vary considerably. Life history traits
ilar phenologies and thus jointly attracting more polli- have strong effects on plant fitness and have conse-
nators than any would by flowering alone. quently evolved over time. Studies of the evolution of
Clearly both phenomena are possible, but there is these traits—both singly and as syndromes of associat-
remarkably little evidence for either. Analyses by Robert ed traits—are an important part of plant ecology.
Trade-offs–the necessity of decreasing one thing if In confronting environmental uncertainty, plants are
something else is increased—are thought to be impor- often subjected to a different kind of trade-off, because
tant in the evolution of life history traits. They occur variation in fitness among years acts to reduce long-term
because of some basic constraints. If the amount of fitness. Many plants appear to reduce among-year fit-
resources allocated to reproduction were unlimited, a ness variation by bet hedging—spreading risk by evolv-
plant could produce an infinite number of seeds and ing traits such as iteroparity and seed banks.
make them as large as possible. Since resources are lim- Phenology, the within-year timing of growth and
iting, however, there is a trade-off between seed size and reproductive events, is also affected by various trade-
number of seeds. An important group of models offs. Evergreen plants are able to photosynthesize under
attempts to understand the direction of natural selection a wider variety of conditions than deciduous plants, but
by asking what combination of traits (such as seed size to be evergreen they must have a number of character-
and number) confers the greatest fitness on plants. There istics—such as smaller leaf surface areas—that act to
have been several important attempts to develop broad reduce the rate at which they can fix carbon under favor-
theories to describe how trade-offs shape syndromes of able conditions. Plants living on forest floors frequent-
life history traits. The theory of r- and K-selection, and ly must either experience most of their growth when the
Grime’s extension of that theory, do not lend themselves canopy trees are bare or evolve the ability to effectively
to critical tests because their predictions are relative. use light flecks that travel across their leaves when the
Models of optimal reproductive effort have also encoun- canopy is in leaf. The timing of flowering involves trade-
tered difficulties with testability. One important conclu- offs between the availability of abiotic resources such as
sion is that we need more information about the underwater and light and biotic factors such as pollinators and
lying biology to make much progress in understanding frugivores.
the constraints governing life history evolution.
Additional Readings
Classic References Mazer, S. J. 1989. Ecological, taxonomic, and life history correlates of seed
mass among Indiana Dune angiosperms. Ecol. Monogr. 59: 153–175.
Baker, H. G. 1972. Seed mass in relation to environmental conditions in
California. Ecology 53: 997–1010. Stanton, M. L. and C. Galen. 1997. Life on the edge: adaptation versus
environmentally mediated gene flow in the snow buttercup, Ranunculus
Cole, L. C. 1954. The population consequences of life-history phenome-
adoneus. Am. Nat. 150: 143–178.
na. Q. Rev. Biol. 29: 103–137.
Monk, C. D. 1966 An ecological significance of evergreeness. Ecology 47:
504–505. Additional Resources
Salisbury, E. J. 1942. The Reproductive Capacity of Plants. Bell, London. Aerts, R. 1995. The advantages of being evergreen. Trends Ecol. Evol. 10:
402–407.
Roff, D. A. 2002. Life History Evolution. Sinauer Associates, Sunderland,
Contemporary Research MA.
Brown, J. S. and D. L. Venable. 1991. Life history evolution of seed-bank
Stearns, S. C. 1976. Life-history tactics: A review of the ideas. Q. Rev. Biol.
annuals in response to seed predation. Evol. Ecol. 5: 12–29.
51: 3–47.
Kalisz, S. and M. A. McPeek. 1993. Extinction dynamics, population
growth and seed banks. Oecologia 95: 314–320.
PART
III From Populations
to Communities
10 Competition 185
11 Herbivory and Plant-Pathogen
Interactions 213
12 Community Properties 235
13 Disturbance and Succession 253
14 Local Abundance, Diversity,
and Rarity 275
10
C H A P T E R Competition
C
ompetition has been studied intensively by plant ecologists, and even
more intensely debated. Key questions have included: What are the
effects of competition on individuals, populations, and communities?
What determines the outcome of competition among different individuals?
What are the resources for which plants compete, and how does the type of
resource competition affect the nature or outcome of competitive interactions?
Does the intensity of competition vary with productivity? How important is
competition relative to herbivory and other factors, and how do these factors
modify competitive interactions?
Competition can be defined as a reduction in fitness due to shared use of
a resource in limited supply. Other definitions emphasize different aspects of
competition, such as the mechanism of resource exploitation. More importantly,
there are different ways of measuring competition, each of which reveals some-
thing different about the nature of competitive interactions.
Ecologists who study animal populations have argued a great deal about
the importance of competition in determining population structure and abun-
dance. Plant ecologists, on the other hand, have generally accepted that the
effects of competition are obvious and pervasive. Anyone who has ever grown
a garden knows how much weeds can affect the plants one is trying to grow,
and the millions of dollars spent every year on herbicides by farmers argue
that competition has major effects on plant productivity. However, because its
effects are so complex, plant ecologists have debated just about everything else
regarding plant competition, from how to define and measure it to when and
where it is important.
A fair amount is known about the effects of competition on individual
plants. Competitors can reduce a plant’s biomass and growth rate and decrease
its ability to survive and reproduce. Among plants of the same species in the
same environment, the number of seeds produced is highly correlated with
maternal plant mass, so successful competitors that accumulate more mass will
have more resources to put into reproduction.
We begin this chapter by examining what is known regarding the effects
of competition on seedling survival and growth, primarily from greenhouse
experiments, but also from studying single-species (monospecific) plantings
of trees for timber. Next we look at the effects of competition on populations,
population distributions, and community composition. It is these aspects of
competition that have been the subject of the greatest controversy. We exam-
(A)
Dry weight (g)
1000 10
Average individual plant weight (g)

182 days
0
0 2.5 12.5 25
Thousands of seeds/m2
1
131 days
(B)
80 High
Dry weight (g/pot)
60 0.1
40
Med
20 61 days
Low
0 0.01
0 3 6 12 25 50 0 days
Density (plants/pot)
Figure 10.1 1 10 100 1000
Relationships between yield of dry matter (g) and plant den- Plant density (plants/pot, log10 scale)
sity for two pasture plants. (A) Trifolium subterraneum
(Fabaceae), measured after flowering (density is expressed Figure 10.2
as thousands of seeds sown/m2). (B) Bromus unioloides Average individual plant weight (g) for Trifolium subterra-
(Poaceae) at low, medium, and high levels of fertilization neum (Fabaceae) planted over a range of densities (expressed
with nitrogen (density is expressed as plants/pot). (After as number of plants/pot) and harvested at 0, 61, 131, and
Donald 1951.) 182 days after seeds were planted. Note that both axes are
on a log10 scale. The arrows show the densities at which
plants began to reduce one another’s growth at different
ages. At early harvests, only very dense plantings showed
ine some of these debates, and take a look at the evi- reductions in average plant weight, but by the last harvest,
dence ecologists have gathered on the role and impor- all but the lowest-density planting demonstrated weight
reductions related to density. Decline in weight with density
tance of competition. is linear on a log-log scale. (Data from Donald 1951 and Kira
et al. 1953, after Harper 1977.)
Competition at the Level of Individuals

Size and Density showed that as plant density increased, the mean weight
Plant growth is highly plastic, and the mass, height, of individual plants decreased in a linear fashion when
number of leaves, and reproductive output of an indi- both were expressed on a log scale (Figure 10.2).
vidual plant can vary over orders of magnitude depend- Mean plant size can be a misleading measurement,
ing on growth conditions. When plants are grown with- however. Size relationships among individuals in even-
out close neighbors, they are generally much larger than aged, dense single-species plantings have been well
similar individuals surrounded closely by others, and studied in greenhouse and in a few field studies. Indi-
often have a very different morphology, or form (see Fig- vidual plant sizes are generally extremely uneven in
ure 6.6). such stands. Typically, a few large individuals dominate
In a classic experiment, C. M. Donald (1951) showed the available area, while most individuals remain very
that British annual pasture plants sown over a wide small. These highly unequal size distributions are called
range of densities had a remarkably consistent total final size hierarchies.
dry weight in a given area (Figure 10.1). Whether seeds It has been hypothesized that size hierarchies are
were planted sparsely or very densely (above a certain caused by asymmetric competition for light (Weiner 1990;
minimal density), the total aboveground dry matter at Schwinning and Fox 1995), in which the largest individ-
final harvest was constant. The total yield was increased uals have disproportionate negative effects on their small-
when more resources were supplied, but the same rela- er neighbors. It has been suggested that small initial dif-
tionship held. While the average plant size at low den- ferences in access to light are responsible for progressively
sities was relatively large, the average plant size became greater inequality in size over time, as long as density
smaller as density increased. In a re-analysis of these remains constant (Figure 10.3). Among a group of
data, Tatuo Kira and his colleagues (Kira et al. 1953) seedlings germinating together, a small head start may
PAGE PROOF: 2ND PASS Competition 187
2 4 6 8 Age (weeks)
60 400 60 395 60 250 60 103 Number surviving

(initial = 400)
50 50 50 50
Frequency (%)
40 40 40 40
30 30 30 30
20 20 20 20
10 10 10 10
Smallest Largest Smallest Largest Smallest Largest Smallest Largest

Individual plant dry weights
Figure 10.3
Frequencies of dry weights of individual seedlings of Tagetes patula (marigold, Aster-
aceae), an annual plant, grown in a greenhouse experiment at 2, 4, 6, and 8 weeks.
The number of surviving plants is shown above each graph. At 2 weeks, the distribu-
tion is close to a normal (bell) curve, but the distribution of dry weights becomes
increasingly unequal (hierarchical) as the population becomes older, with many
small plants and a small number of very large individuals. Over time, death removes
the smallest individuals from the population (self-thinning), so that by 8 weeks the
population is somewhat less unequal than at 6 weeks. (After Ford 1975.)
3000 confer a large advantage. Ross and Harper (1972) showed

a strong relationship between the order of emergence and
1000 plant size for densely planted Dactylis glomerata (orchard
Population A grass, Poaceae; Figure 10.4). In natural populations, this
Population B
300 Population C
advantage can be offset by variation in hazards (such as
Seedling dry weight (mg)
frost) and resource availability in early spring.

100 While size hierarchies have been found in many
greenhouse studies and in some tree plantations, much
less is known about them in natural populations. Wilson
30
and Gurevitch (1995) examined the spatial relationship
of plant sizes in a dense natural stand of Myosotis micran-
10
tha (forget-me-not, Boraginaceae), a small winter annu-
al. These plants had extremely unequal sizes (Figure
3 10.5). The researchers hypothesized that if asymmetric
competition was the cause, then large individuals should
1 be surrounded by small, suppressed neighbors.
They found instead that the opposite was true.
0 10 20 30 40 50 60 70
Large plants had large immediate neighbors, and small
First Last
plants were associated with small neighbors. Individual
Order of emergence (percentile ranking) plant mass was also highly correlated with the combined
mass of neighbors, so that the population formed a
Figure 10.4 mosaic of patches of large plants and patches of small
The effect of relative order of emergence (percentile ranking)
on seedling dry weight (mg). Each line shows the relation- plants. Plants without close neighbors were much larg-
ship (the regression) for one of three different populations of er than plants with neighbors, however, so competition
Dactylis glomerata (orchard grass, Poaceae) in a greenhouse was probably important in affecting plant size. The
experiment. The consequences of emerging sooner or later researchers concluded that asymmetric competition was
than one’s neighbors can be enormous: seedlings with the unlikely to have caused the extreme size hierarchy found
lowest percentile ranking—those that germinated and
appeared aboveground first—were more than 1000 times in this natural population. Rather, the size hierarchy was
larger than those emerging last. Seedling dry weight (mg) is probably caused by variation in plant density or patchy
graphed on a log scale. (After Ross and Harper 1972.) resource distribution.
300 and density is determined by area (a quadratic meas-

ure), Yoda and colleagues (1963) proposed what they
called the –3/2 thinning law:
250
w = cN –3/2
where w is the mean dry weight per plant, c is a constant
that differs among species, and N is density. On a log-
200
log scale, the relationship between mean dry weight and
Number of plants
density is predicted to have a slope of –1.5 (Figure 10.6).

Although the –3/2 thinning law was widely accept-
150
ed until the 1980s, it has become a matter of consider-
able debate since then. There is no doubt that mortality
occurs in crowded stands; what is in dispute is whether
100 the process is so regular that it can be described by a sin-
gle numerical relationship for all populations. Studies
of single populations have been plagued by serious sta-
50 tistical problems (Weller 1987), as have review papers
attempting to evaluate the generality of the relationship
(Lonsdale 1990; Weller 1991). On the other hand, Silver-
town and Lovett Doust (1993) make a case for the rela-
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 tionship holding well as an upper limit.
Individual plant mass (g)
Mechanisms of Competition
Figure 10.5
Distribution (number of plants) of individual plant dry Plants compete for light, for water and mineral nutrients
weights (g) in a natural population of the annual Myosotis from the soil, for space to grow and to acquire resources,
micrantha (forget-me-not, Boraginaceae), harvested at flower- and for access to mates. Competition for animal vectors
ing. There were large numbers of very small plants, with a
steep drop in numbers of plants in the larger categories, and
very few of the largest individuals. (After Wilson and Gure-
vitch 1995.)
Slope = –1
Mortality, Size, and Density F
F
If seeds of herbaceous plants (usually annuals) are plant- t=3 3
Mean dry weight/individual
ed in dense monospecific stands and their survival and F Slope = – 3

2
weights are followed through early life history stages,
the seedlings grow until they begin to crowd one anoth- 3
t=2
er. As crowding becomes severe, some individuals even- 2 F
tually begin to die; usually the smaller and weaker 3
2
plants succumb most readily. More densely planted Final
stands begin to experience mortality sooner and at t=1 2
dry weight
smaller individual plant sizes than more sparsely plant- 1
ed stands. Factors such as greater soil fertility or greater 1
initial seed size that result in larger individuals also 1
increase mortality because larger plants cannot maintain
as high a density as smaller plants.
This process of density-dependent mortality is I I I I
Sparsely planted Densely planted
known as self-thinning. In a crowded monospecific,
even-aged stand of plants, a log-log plot of density Log density (m–2)
against total plant mass will have a negative slope— Figure 10.6
plant growth can occur only as some individuals die. It Effects of planting at different densities on the mean dry
is important to realize that the term “self-thinning” does weights of individuals as seedlings age. Each of the green
not imply a voluntary, altruistic self-sacrifice of the lines represents a different initial planting density. The low-
ermost point on each line is the initial dry weight at germi-
weaker individuals for the general good (which is cer- nation (I), and the uppermost point is the final dry weight
tainly not what is happening!). Since the weight of a (F), with weights shown at time intervals t = 1, 2, and 3.
plant is directly related to its volume (a cubic measure) (After Kays and Harper 1974.)
(A) (B)
Figure 10.7
(A) A strangler fig (Ficus sp., Moraceae), a tropical plant that
for pollen and seed dispersal is a special kind of com- begins life as a vine, killing the tree that it grows on and sub-
petitive interaction because these mobile biotic resources sequently becoming a tree itself. (Photograph courtesy of J.
can respond to plants in ways abiotic resources do not. Thomson.) (B) Celastrus orbiculata (Oriental bittersweet,
Light is in some ways the most peculiar resource for Celastraceae), a vine native to eastern Asia that is now inva-
sive in the eastern United States, is shown here completely
which plants compete (see Chapter 2). In one sense, cloaking a tree. (Photograph by J. Gurevitch.)
plants do not really compete for light, because no mat-
ter how much light plants take up, an essentially infinite
supply is always available directly above their canopies.
Yet plant canopies can very effectively mop up photons, Tropics, strangler figs (Ficus spp., Moraceae) grow on
reducing light at ground level to less than 1% of the inci- trees, encircling and eventually killing their hosts (Fig-
dent sunlight. It can be very dark in many communities, ure 10.7A). In temperate forests in the eastern United
not only for the sapling struggling to grow in a dense States, trees along forest edges or gaps are sometimes
forest understory, but also for the seedling emerging in overgrown by either native or invasive exotic vines (Fig-
a grassy meadow. One might argue that plants are com- ure 10.7B). It is not always clear whether vines cloak and
peting for access to light, or for sunlit space, rather than weaken otherwise healthy trees or overwhelm only trees
for light itself. After all, light is available in unlimited previously weakened by other factors, such as insects or
supply, but space is not. Perhaps it is most useful to think disease. Once covered by vines, trees become much
of plants as competing for light, but in a manner differ- more likely to be toppled in windstorms and killed. This
ent from that in which they compete for soil resources. vulnerability probably results both from the weakening
Overgrowing or overtopping neighbors is an obvi- of the root system as a consequence of loss of photo-
ous means of competing for light. Overtopping is prob- synthate as the vines increasingly shade the leaves and
ably a common mechanism leading to successional from the tremendous extra weight of the vines.
change from old fields to forests, in which shorter herba- Most of the competitive interactions experienced by
ceous plants are replaced by shrubs and eventually by a plant occur on a very local scale. The shade of larger
trees (see Chapter 13). Another dramatic example of neighbors reduces the plant’s ability to photosynthe-
overtopping is the overgrowth of trees by vines. In the size, while the roots of the plants immediately sur-
Figure 10.8
A transect from a natural population of Myosotis micrantha (forget-me-not, Boragi-
naceae). Each circle represents an individual plant; the size of the circle shows the rela-
tive plant size. Plants in the same population may experience relatively low densities of
neighbors, as in the part of the transect shown at 1, or high densities of neighbors, as in
the part of the transect shown at 2. The x- and y-axes are not on the same scale: the dis-
tance along the vertical dimension from region 1 to region 2 is about 35 centimeters; the 1
width of the transect is 20 centimeters. (Data from Wilson and Gurevitch 1995.)
rounding it may absorb some of the water and nitrogen

2
that it needs to function. Plants even a short distance
away, however, may experience no effects at all from
those particular plants. For this reason, the density
experienced by an individual plant is almost always
strictly the density of the plants in the immediate patch
surrounding it. The average density of plants in the
population is irrelevant to the degree of crowding it
experiences (Figure 10.8). This pattern is quite different
from the effects of density on mobile animals, which
may be competing with many other animals in the same
general area for food. One exception to the very local However, plants of extremely different sizes or ages
nature of plant competition is competition for animal may not be competing at all, or they may be competing
vectors (pollinators and dispersers). Because these ani- in different wys than more similar-sized or similar-aged
mals may be highly mobile, competition for pollinators individuals. A forest tree may not compete directly with
may occur between plants some distance apart, rather an herbaceous understory plant growing beneath it:
than strictly between immediate neighbors. understory plants commonly grow in the shade, and the
Resources such as light, water, nitrogen, and phos- roots of trees and herbs are often found at very different
phorus can be very patchy. One of the ways plants depths in the soil, drawing on different supplies of soil
respond to this patchiness by preferential growth in the resources.
areas where these resources are available. Forest trees, for The assumption that plants of very different growth
example, respond to newly created gaps in the canopy forms do not compete can be deceptive, however. Such
by the growth of branches into the gap. Young saplings competition may be occurring, although it may be one-
in a newly created gap shoot up rapidly in height. Some sided. A small sapling growing beneath a canopy tree
individuals and some species are much better at respond- may have no measurable competitive effects on the
ing to gaps and other patches of high resource availabil- mature tree, but the shade of the tree may completely
ity than are others, and this ability may make them much prevent the sapling from growing larger. Cut down the
better competitors under some conditions. tree, and the sapling is likely to rapidly increase in height
Dense masses of roots can develop in nutrient-rich and biomass. But things are not always so simple:
patches of soil (Robinson et al. 1999). Root proliferation foresters have found that suppression of understory
may, under certain conditions, allow a plant to monop- plants with herbicides can lead to great increases in the
olize the nitrogen in the soil. Little is known about how growth rates of mature lumber trees, presumably
species differ in their abilities to monopolize soil because the understory plants intercept water or reduce
resources through root proliferation, but such differences soil nutrients available to the trees.
could lead to differences in competitive ability. The roots Being larger is not inevitably an advantage, howev-
of different species are known to differ in their nutrient er, and in some cases, larger plants become susceptible
uptake rates, at least under laboratory conditions, and to other factors that reduce their effectiveness in com-
these rates may also affect their competitive ability. petition. A plant with a greater amount of leaf surface
area, for example, loses much more water by transpira-
Is Bigger Better? tion than a similar plant with less leaf area. Consequent-
Size generally confers an advantage in competion—big- ly, the larger plant may be affected more by drought than
ger is usually better. Larger plants can affect smaller its smaller neighbors, increasing its likelihood of dying
neighbors without being much affected themselves. or reducing its future effectiveness in competition.
Among young plants growing side-by-side in an aban-
doned field, for example, the larger neighbor will pro- Apparent Competition
duce disproportionately more leaves and roots, have One type of animal-mediated interaction among plants
more flowers, and set more seed. is apparent competition—density-dependent negative
interactions between species that might at first appear to based on two-species replacement series experiments
be due to competition for resources, but are actually due (described below), which he argued are biased. Biased
to a shared predator or herbivore (Holt 1977). Apparent or not, extrapolation of the results of these greenhouse
competition might occur if, for example, as the combined experiments to natural plant communities is necessari-
density of two plant species increased, they increasingly limited. We need to have more information, particu-
ly attracted the attention of herbivores. Thus, both species larly from field experiments, before reaching a conclu-
would suffer from the other’s presence and abundance, sion as to the general existence of competitive hierarchies
but because of herbivory rather than competition. in nature. Unfortunately, while there have been hun-
Apparent competition among plants has been dem- dreds of studies of plant competition in the field, few
onstrated only rarely. A recent study showed that a have addressed this issue, and thus the data on this
native New Zealand fern, Botrychium australe (adder’s question are still too limited to allow any reliable con-
tongue, Ophioglossaceae) was declining in abundance clusions to be drawn (Goldberg and Barton 1992).
due to apparent competition with an introduced grass,
Agrostis capillaris (Sessions and Kelly 2002). The grass Quantifying Competition
may serve as a refuge for an introduced slug. The native How one measures the intensity of competition will
fern survives well after fire. However, fire also increas- affect the interpretation of the outcome of plant compe-
es populations of the introduced grass, which leads to tition studies. Some of the disagreements about the
an increase in slug density. Increased slug populations nature of competition among plants can be traced to dif-
following a severe fire led to severe defoliation and mor- ferences in the way competition has been studied and
tality of the fern due to the positive effects of the greater measured (Grace 1991, 1995). Goldberg and Werner
grass cover on the slugs. (1983) distinguished between the competitive effect of
a plant on its neighbors and the competitive response
of a plant to its neighbors. Recognition of these two dis-
Populations and Competition tinct components of plant competitive ability can help
us to understand the ways in which plants interact. The
Competitive Hierarchies rankings of competitive effect and competitive response
Plant ecologists differ strongly on the question of among the plants in a community are not necessarily
whether consistent competitive hierarchies exist among correlated (Figure 10.9; Goldberg 1990).
plant species. Are some species always competitively The interpretation of the results of a competition
superior, and is the rank order of subordinates relative- experiment depends on the units in which the outcome
ly fixed? Some ecologists argue that consistent compet- is expressed. In many experiments, the results are cal-
itive hierarchies exist. Others argue that fluctuating culated on the basis of area (such as biomass per unit
dominance in competitive networks is the general rule area). Alternatively, one can examine the effects of com-
for plant interactions. petition per gram of biomass of competitors, or on a per-
What are the implications of these two opposing individual basis.
views? The issue of competitive hierarchies is important Not only does each of these measures mean some-
because it touches on questions regarding the basic thing different, but their usefulness depends on the
structure of plant communities. If competitive hierar- purpose of the measurement. For example, some mod-
chies were the general rule, ecologists could predict the els of competition require results that are expressed per
competitive abilities of plant species based on their traits, individual, while others make the most sense per unit
such as growth form and size (Herben and Krahulec area. Furthermore, while assessing competitive results
1990; Shipley and Keddy 1994). Explanations of species per individual might be straightforward for many trees
diversity would require an understanding of the factors or even annuals, it is likely to be impossible (and not
that prevent competitive exclusion, such as nonequilib- even very interesting) for clonal perennials. These
rium processes. issues are clearly a consequence of the great plasticity
In contrast, if communities consisted of competitive of plants.
networks with frequent reversals of competitive domi- One argument for measuring competition on a per-
nance, an understanding of community structure would gram basis is that this practice eliminates the effects of
require studies of niche partitioning and species pack- size. But is eliminating size always desirable? If two indi-
ing and of how life history or other trade-offs permit viduals begin at more or less the same size, and one
stable coexistence. In that case, species diversity in eco- acquires resources more quickly and efficiently than the
logical communities would be maintained primarily by other and becomes the dominant competitor, is the
competitive interactions, implying that communities are greater size of that individual a bias that we want to
generally deterministic and equilibrial in character. get rid of, or is it the essence of the competitive interac-
Connolly (1997) pointed out that most of the data tion? Of course, if per-gram effects are calculated at the
that support the existence of competitive hierarchies are beginning of such an experiment, the faster-growing
(A) (B) index of performance is the absolute competition index

(ACI), which is simply the difference:
Target performance
ACI = Pmonoculture – Pmixture
N3 > N2 > N1 T1 > T2 > T3 Using these two different indices as measures of the out-
come of an experiment may lead to very different con-
clusions.
Despite its popularity for measuring competition
N3 N2 N1 T3 T2 T1 intensity, the RCI is subject to a number of limitations.
Neighbor density
Ratios expressed on an arithmetic scale have poor sta-
tistical properties and are asymmetric (changing the
Figure 10.9 numerator affects the ratio differently than changing the
Hypothetical examples of different strengths of competitive
effects and competitive responses for different target plant
denominator by the same amount). The ACI is also sub-
species (T1, T2, and T3) with different neighbor species (N1, ject to both conceptual and statistical problems. One
N2, and N3). In this example, neighbors are grown at a range alternative is to use the log response ratio (LRR):
of densities. (A) As neighbor density becomes higher, target
plant performance (across all species) declines; the rate of Pmixture
LRR = ln
decline is steepest when species N3 is the neighbor, less steep Pmonoculture
for N2, and least for N1. Plants of neighbor species N3, there-
fore, have the greatest competitive effect on the target plants. This index has the advantage of expressing performance
(B) Increasing neighbor densities have the greatest effect on relative to potential performance, as RCI does, but has
target species T3 and the least effect on T1— plants of species better statistical properties, including symmetry (Hedges
T1 decline less in performance as each individual of the et al. 1999). LRR is very similar to the difference in rela-
neighbor species is added and neighbor density increases
(the slope is shallower). Target species T1 has the best com-
tive growth rates between monoculture and mixture if
petitive response because it is least sensitive to competition. the initial sizes of the plants are similar and small in
(After Goldberg and Landa 1991.) comparison with the final sizes.
The debate regarding the superiority of one index
of competition over the others rests on the assumption
that an index can accurately represent the essential fea-
plant in this scenario will have greater per-gram effects, tures of a competitive interaction. Any index simplifies
but if measured in terms of final size (as is often done), reality, however, and thus has certain drawbacks as a
its per-gram effects will be artificially low. tool for understanding competitive interactions (as do
Goldberg and Landa (1991) compared the compet- indices expressing any kind of relationship, from species
itive effects and competitive responses of seven species diversity to stock market performance). Any attempt to
during the early stages of competition among seedlings compress complex data into a single index suffers from
in a greenhouse experiment. They used regressions of a loss of information. Other limitations to competitive
target plant biomass on neighbor biomass to determine indices include that they unrealistically assume linear
the per-unit-biomass effects of neighbors, and regres- responses to neighbor density; that many different out-
sions of target plant biomass on neighbor density to comes can result in the same index; and that competi-
determine the per-individual-plant responses to neigh- tion indices tend to be strongly influenced by initial
bors. They found consistent competitive hierarchies, but plant size (Grace et al. 1992.)
the ranks differed for effect and response, and for per- In making comparisons of competition intensity
plant and per-gram measures of competitive ability. across environments or species, one alternative is to
There are several approaches to quantifying the examine interaction diagrams and test for statistical
intensity of competition. One of the most common is to interactions, using performance directly without con-
use an index, usually a ratio that standardizes respons- verting to indices (Box 10A). Better still would be infor-
es across species and environments so that they can be mation over time on the performance of the competitors,
compared on the same scale. One of the most common following their trajectories to determine the progress of
indices is the relative competition index or relative com- the interaction.
petition intensity, RCI:
Pmonoculture – Pmixture
RCI =
Pmonoculture Experimental Methods
for Studying Competition
where Pmonoculture and Pmixture are plant performance in
monoculture (single-species stands) or in mixture Greenhouse and Garden Experiments
(stands with two or more species), and performance is The majority of plant competition experiments have
usually measured as dry mass or growth rate. Another been conducted in greenhouses. Greenhouse experi-
ments offer the advantages of relatively controlled con- specific competition by altering the frequencies of two
ditions and precise manipulation of the factors of inter- hypothesized competitors while keeping the total den-
est. Limitations are the uncertainty in extrapolating sity constant (Harper 1977). One of the most common
results to natural communities, as well as various arti- forms of this type of experiment is the de Wit replace-
facts of greenhouse conditions, including very differ- ment series (de Wit 1960), in which the relative percent-
ent wind and humidity conditions. An important poten- ages of two competing plants are varied from pure
tial artifact is that plant water relations and root structure monocultures to 50/50 mixtures at a constant density,
are drastically affected by growth in pots. often with intermediate mixtures as well (Figure 10.10).
Three basic experimental designs are commonly These experiments were once the most important tool
used: substitutive (replacement), additive, and additive for studying plant competition, but have been criticized
series designs (Gibson et al. 1999). Substitutive designs on a range of issues. They are subject to theoretical and
test the relative strength of intraspecific versus inter- statistical limitations, suffer from restrictive assump-
(A) (B) (C)

Model I Model IIa Model IIb
Performance (x dry wt., etc.)
I+J I+J I+J
I I I
J J
J
Species J High (J) 0 (J) High (J) 0 (J) High (J) 0 (J)
Species I 0 (I) High (I) 0 (I) High (I) 0 (I) High (I)
Relative density
(D) (E)
Model III Model IV
Performance (x dry wt., etc.)
I+J
I
I
J
Species J High (J) 0 (J) High (J) 0 (J)

Species I 0 (I) High (I) 0 (I) High (I)
Relative density
Figure 10.10
Different possible outcomes in a de Wit replacement series competitor in (B) and J is the superior competitor in (C); in
competition experiment. In each experiment, plants of two each case, the superior competitor affects members of the
species (I and J) are grown in trays with different relative other species more than members of its own species, and is
mixtures of the two, while total plant density is kept con- affected less by the other species than by its own. Combined
stant. (A) In an experiment with a model I outcome, the performance is again the sum of the performances of the two
effects of individuals of species I and J are the same on them- species. (D, E) In model III and IV outcomes, the performance
selves (intraspecific competition) as on each other (interspe- in mixture is not additive. In (D) the combined performance
cific competition). The combined performance in mixture is less, and in (E) it is more, than the sum of the performances
(top line) is the sum of the performances of the two species in of the two species when grown alone. (D) Here each species
the mixture; neither species is a superior competitor. (B,C) In affects the other species more than it affects its own, and both
a model II outcome, plants of one of the species (the superior suffer more in mixture than in monoculture; neither is com-
competitor) have a greater interspecific than intraspecific petitively superior. (E) Here both species have a greater effect
effect, while the other species (the inferior competitor) has a on their own species than on the other species (perhaps due
lower interspecific than intraspecific effect. I is the better to having somewhat different niches).
BOX 10A
The Implications of How Competition is Measured
Imagine that an ecologist is studying seedlings are planted without neigh- weight at the start of the time interval
two co-occurring species; we will call bors to assess what their performance (Hunt 1990). (We could also have cho-
them Species A and Species B. Species would be without competition. The sen to use an absolute growth rate as a
A contributes most of the biomass in results are: measure of performance.) In this case,
the area being studied, while species B (A) RCI for the intensity of competition
contributes only a small proportion of 101.0 g experienced by species A, based on
the biomass. Plots are cleared, and RGR as the measure of performance,
seedlings of the two species are plant- is
ed together in experimental plots. One 1.0 g RCIA = (4.62 – 4.61)/4.62 = 0.002
seedling of each species is planted per
plot, each seedling weighing on aver- The intensity of competition expe-
age 1 gram. After some period of time, rienced by species B (RCIB) is 0.28,
the plants are harvested and weighed, reflecting the large effect of species A on
with the following results (the numbers species B and the very small effect of
(B)
shown are plot means): 25.0 g species B on species A.
Alternatively, an interaction dia-
Species A 1.0 g
Harvested biomass gram using RGR would look like this:
100.0 g
5
Seedling biomass
A
1.0 g To proceed, we next calculate the 4
RGR
natural logs of all biomass values (ln 1.0
= 0, ln 10.0 = 2.30, ln 25.0 = 3.22, ln 100.0
= 4.61, ln 101.0 = 4.62). How should 3 B
these competitive effects be assessed?
First, we calculate the relative growth
Species B 2
rate (RGR) over some time interval for Competitor Competitor
10.0 g
1.0 g
each species as present absent
RGR = ln W2 – ln W1
RGRA,mono = 4.62 – 0 = 4.62
The graph likewise reflects the very
RGRA,mix = 4.61 – 0 = 4.61
Now imagine that the ecologist small effect of competition on species A
wishes to test the hypothesized effects where W2 is the dry weight at the end and the much larger effect on species B.
of each species on the other. Single of the time interval and W1 is the dry A statistically significant interaction
tions, and offer limited ability to extrapolate their results tions and for too often basing their conclusions on a sin-
(Connolly 1986), although they may be useful in partic- gle measure of final yield. They offer various advan-
ular circumstances (Firbank and Watkinson 1990; Ship- tages, however, for addressing a number of questions
ley and Keddy 1994). Perhaps their most important lim- (Gibson et al. 1999). For example, the experiment by
itation is that they consider only a fixed density of plants, Goldberg and Landa (1991) discussed above used an
making it impossible to determine the circumstances additive design to compare competitive effects and
under which each population could be expected to grow responses on a per-individual and per-unit-biomass
or decline (Inouye and Schaffer 1981). basis.
Simple additive designs manipulate the total den- Several authors have discussed some of the advan-
sity of neighbors, usually over a range of densities, while tages of additive series and “complete additive” designs
keeping the density of the target species constant (usu- (also called response surface experiments), in which
ally a single individual). These experiments have been both densities and frequencies are varied (e.g., Firbank
faulted for confounding density with species propor- and Watkinson 1985; Silvertown and Lovett Doust 1993).
would confirm that interspecific com- showing less of a difference between tion is greater for species Y. In such
petition has a greater effect on species the two species in response to the pres- cases, it is particularly important to be
B than on species A. ence of competitors than when aware that additive effects can be dif-
Connolly (1987) suggested the use of expressed on the basis of RGR. One ferent from proportional effects, and
the difference between the relative would not want to express RCI on a per that ACI will give the former and RCI
growth rates of the two species in mix- unit biomass basis; if we divided by the the latter.
ture to predict the dynamics of the mix- mass of the competitor at the end of the An interaction diagram graphed on
ture. This index, the relative efficiency experiment, it would appear that the the basis of the (untransformed) final
index (REI), in this case is competitive effects of the two species biomass will emphasize additive com-
REI = RGRA – RGRB = 4.61 – 2.30 = 2.31 on each other were very similar. Mea- parisons and give similar results to
sured as log response ratios (LRR), the ACI. An interaction diagram based on
Grace (1995) suggests that comparing values are relative growth rates will emphasize
REI in mixture and in monoculture proportional differences (assuming that
LRRA = ln (101/100) = 0.01
allows one to assess competitive inter- the initial sizes were similar and small
actions and predict the “winner” in LRRB = ln (25/10) = 0.92
relative to the final sizes for both
mixture. Here, REImono = 4.62 – 3.22 = A problem arises when the compar- species) and will therefore give similar
1.40; the greater value of REI in mixture ison of the absolute and the relative dif- qualitative results to RCI, perhaps with
indicates that species A is competitive- ferences differ (Grace 1995). Grace gives somewhat more information. LRR will
ly dominant and is predicted to replace the example of species X reaching 30 also emphasize proportional differ-
B in mixture. grams in competition and 40 grams ences, like RCI, but again, has better
If, as is commonly done, RCI were alone, and species Y growing to 2.5 statistical properties than RCI. It may
calculated on the basis of final biomass grams in competition and 10 grams be informative to carry out these calcu-
rather than RGR, the results would be alone. The absolute difference (biomass lations and draw the graphs for your-
RCIA = 101 – 100 / 101 = 0.0099 alone – biomass in competition) for self; you may wish to use Grace’s exam-
and species X is greater than for species Y, ple and assume various initial sizes for
RCIB = 25 – 10 / 25 = 0.60 but the proportional effect of competi- the two species.
These designs offer more information than simple addi- A few other experimental approaches have been
tive designs, but are large and complex, and they have used on occasion in greenhouse competition studies.
not commonly been used, mostly for that reason. The Experiments in which root and shoot competition are
information gained may not always be worth the addi- decoupled (Figure 10.11) can be useful in determining
tional effort and cost involved. Multispecies competition the ways in which competitors affect one another more
experiments, likewise, are complex and are done only precisely. Fan designs were created to efficiently study
occasionally. the effects of density and frequency simultaneously (Fig-
(A) (B) (C)

Figure 10.11
One type of experiment designed to
separate root and shoot competition
while keeping total soil volume con-
stant. (A) Shoot competition alone
(note barrier to root interaction). (B)
Root competition alone (note barrier
to shoot interaction). (C) Both roots
and shoots are allowed to compete.
One could also combine root and
shoot barriers to eliminate both
aboveground and belowground
interaction.
(A) (C)
(B)
5 6 4 3 0 1 2
4 6 5 3 2 1 0
6 5 4 1 0 2 3
6 2 0 1 3 4 5 Figure 10.12
Fan designs for varying plant density and frequency. (A)
Hexagonal fan design. Each dot represents an individual
plant. (B) Two-species variation on fan design. Shaded and
solid circles represent individual plants of two different
5 6 4 2 1 0 3 species; numbers indicate the number of shaded-circle
4 6 5 0 1 2 3 neighbors of plants of either the shaded-circle or solid-circle
6 3 1 0 2 4 5 plants. The number of solid-circle neighbors is equal to 6
5 4 3 2 0 1 6 minus the number of shaded-circle neighbors. (After
Antonovics and Fowler 1985.) (C) Two other types of fan
designs. (After Mead 1988.)
ure 10.12). Hexagonal fans consist of a series of contigu- natural communities. There is no strict demarcation
ous hexagons of increasingly greater sizes, in which each between greenhouse, garden, and field experiments.
plant is surrounded by between zero and six neighbors Recent technical advances have extended the ability to
at different distances. In addition to efficiency of design, manipulate many environmental factors in field exper-
these experiments provide information on density and iments, from soil temperatures to atmospheric CO2.
spacing effects as well as on the effects of the proportions
of inter- and intraspecific competition. However, they Field Experiments
have some statistical drawbacks, including lack of inde- Competition has been extensively studied in field exper-
pendence, certain kinds of bias, and other limitations iments (reviewed by Goldberg and Barton 1992). The
(Gibson et al. 1999). majority of plant competition experiments in natural
Using yet a different approach, Gurevitch et al. communities have been concerned with simple ques-
(1990) compared the performances of individual plants tions regarding whether neighbors affect biomass,
grown alone (no competition), with an intraspecific growth, or some other fitness component in a single loca-
neighbor, and with an interspecific neighbor (as well as tion at one point in time, but a few studies have ad-
at higher densities of neighbors). They emphasized the dressed more complex questions (see the section below
importance of comparing intraspecific and interspecific on competition along environmental gradients).
competition with no competition, rather than weighing The most common experimental design in natural
interspecific competition against intraspecific effects, as communities involves the removal of all or some neigh-
is commonly done. They also grew all species combi- bors of a target individual (Goldberg and Scheiner 2001).
nations in a range of pot sizes to separate the effects of Removals of different species or different functional
sharing space with a competitor from those of having groups (e.g., woody versus nonwoody neighbors) may
less absolute space available. be contrasted. While there is no reason that neighbor den-
Many of the same kinds of experiments carried out sities could not be increased rather than reduced, this has
in greenhouses can also be carried out in gardens, or in rarely been done in plant competition experiments. This
pots set in the ground or above ground in gardens or approach would almost certainly yield interesting results.
Other experiments manipulate target species abun- scales, or that competitive equilibrium is rarely reached,
dance over natural gradients of productivity or neighbor with herbivory, physical disturbance, or other process-
density, usually by transplanting individuals into native es preventing the superior competitor from displacing
vegetation. Generally, only the growth of mature plants other species. The R* rule has been subjected to only a
is measured in such experiments, although a few stud- few experimental tests (reviewed by Grover 1997), and
ies have examined population responses. Both seed addi- carrying out good long-term tests presents some prac-
tions and adult removals were used by Norma Fowler tical challenges. Nevertheless, more extensive testing of
(1995), for example, to examine the density-dependence this hypothesis would certainly be valuable.
of demographic responses in two perennial grasses, Another means by which plants can compete is by
Bouteloua rigidiseta and Aristida longiseta, in Texas. She “getting there first”—arriving at a newly available locale
found that the density dependence of demographic before other propagules (seeds, vegetative reproductive
parameters for both Bouteloua and Aristida was weak— units, or other dispersal units) and being able to hold
competition was much less important than other fac- that space against later arrivals. It is clear that this is one
tors for population regulation in these plants. In other of the means by which some species dominate some
experimental approaches, target species may be trans- habitats, to the exclusion of other species. Peter Grubb
planted into communities that differ in species compo- (1977) hypothesized that differences in the conditions
sition rather than in productivity or density. and circumstances required for germination and estab-
lishment among species, which he called the species’
regeneration niche, might be an important factor in
Mechanisms of Interspecific Competition
species coexistence. We discuss disturbance, coloniza-
What happens when plants compete with one another? tion, and succession further in Chapter 13.
What are the superior competitors doing that makes
them superior competitors, and what is happening to Trade-offs and Strategies
the losers? We have already touched upon some of the Do certain traits generally make for better competitors,
answers to these questions above. These factors begin to or are there trade-offs that make some traits advanta-
account for possible mechanisms of competition, geous in certain kinds of competitive situations but a
although we do not yet have a full understanding of the handicap in others? This is yet another area in which the
various mechanisms by which plants compete. predictions made by different ecologists sharply diverge.
Seventy years ago, Weaver and Clements (1929) ob-
Resource-Based Competition served that in prairies, taller grasses had a competitive
One general approach to thinking about mechanisms of advantage, and at least some later researchers in herba-
plant competition is to consider the effects plants have ceous perennial communities also found that plant
on resources. The extent of resource depletion can be height was associated with competitive dominance
used as a measure of the competitive effect of one species (Keddy 2001). But traits that confer an advantage to indi-
on others. Suppose we call the level of a limiting viduals or species competing for light may not be advan-
resource at equilibrium R*. That is, R* is the amount of tageous in competing for soil nutrients.
the resource left when a population of a single species David Tilman and other ecologists have assumed
grown alone reaches its equilibrium density. It has been that particular traits have associated costs that put
proposed that the outcome of competition should be organisms possessing them at a competitive advantage
determined by the R* rule: Over the long term, in a con- only under certain conditions or in certain environments.
stant environment, the species with the lowest R* when In any environment, the best competitors under the con-
grown alone will competitively displace all other species ditions present will be dominant.
(Armstrong and McGehee 1980; Tilman 1982; Grover In contrast, J. Philip Grime (1977) and others have
1997). This is predicted to happen regardless of the rel- argued that certain traits, particularly those that confer
ative initial numbers of the competing species. The rapid growth rates under favorable conditions, are
resources typically suggested to be involved are light always associated with competitive superiority. In favor-
and soil nitrogen. able environments, such species will always dominate,
The R* rule implies that a relatively uniform habitat whereas in unfavorable environments, characteristics
should be dominated by a single species, and that dif- other than competitive superiority will determine
ferent species should dominate in different environ- species’ dominance and persistence.
ments. Since monospecific stands of superior competi- To examine the relationship between plant traits and
tors are uncommon, it must be that the R* rule is competitive effect and response, Paul Keddy and col-
inadequate, that every microhabitat is sufficiently dif- leagues (Keddy et al. 1998) carried out a large garden
ferent from the others around it that each can be com- experiment with 48 wetland species. They found that
petitively dominated by a different species at very small competitive effect was strongly related to relative growth
(A) Figure 10.13

Relationships between (A) relative competitive response and
100
relative growth rate, (B) relative competitive response and
relative competitive effect, and (C) relative competitive effect
80 and relative growth rate for 48 wetland species from
Relative competitive response (%)
Ontario, Canada. Relative growth rate and relative competi-

60 tive effect have little influence on relative competitive
response. Relative growth rate has a strong and fairly tight
relationship to relative competitive effect. (After Keddy et al.
40 1998.)
20
0 rate. Competitive response, however, was not related to

either relative growth rate or competitive effect (Figure
–20 10.13). Also, species’ competitive responses were simi-
lar across a range of very different communities; that
– 40 is, the response of a species was consistent even when
0 0.05 0.10 0.15 0.20 0.25 0.30 0.35
its neighbors were very different. These results suggest
Relative growth rate
that fast-growing, tall species generally are better com-
(B) petitors, and that in natural communities, a few domi-
100 nant plant species should account for most of the bio-
mass. However, there may be few communities in which
80 this is true (herbaceous temperate systems are one of the
Relative competitive response (%)
few examples), and there is little evidence as to whether

60 the most abundant species in a community are general-
ly competitively superior.
40 Relative allocations to roots, photosynthetic organs,
stems, and reproduction are traits that can affect plant
20 competitive abilities. Shoot architecture has a large effect
on the shading of neighbors. Root structure and distri-
0 bution in the soil (including lateral extent, depth, and
the degree to which roots fill space) affect water and
–20 nutrient uptake relative to neighbors’ abilities to access
these resources. Relative allocations to nonwoody ver-
– 40 sus woody structures and to perennating versus storage
0 10 20 30 40 50 60 70 80 90 100
organs also affect competitive interactions. It is difficult
Relative competitive effect (%)
to make generalizations about how different allocation
(C) strategies determine competitive interactions, however,
100 particularly because their consequences differ in differ-
90 ent environments.
For many years, plant ecologists have tried to iden-
80 tify plant strategies: characteristic suites of traits that
Relative competitive effect (%)
70 should be most successful in certain environments or

under particular circumstances. Defining generally use-
60
ful strategies has been difficult, but the basic idea of
50 strategies continues to come up in various forms (see
40
Chapter 9), and it has intuitive appeal (reviewed by
Grover 1997). Mark Westoby and colleagues (Westoby
30 1998; Westoby et al. 2002) have attempted to overcome
20 objections to previous strategy schemes by proposing
objective, widely applicable schemes that can be used
10
to make worldwide comparisons of plant species and
0 communities. Westoby’s strategy schemes are based on
–10
the assumption that there are trade-offs along a small
0 0.1 0.2 0.3 0.4 number of axes that define major, easily quantified traits
Relative growth rate such as specific leaf area, leaf life span, leaf size, twig size,
canopy height, seed mass, and seed number. His goal in Inside the exclosures, a lush growth of herbaceous plants
developing these schemes is similar to that of people appeared. Apparently the animals spent most of their
seeking to define plant functional groups (see Box 12A): time foraging under the cover of the shrubs, seeking pro-
reducing the vast diversity of terrestrial plants to con- tection from predators (particularly birds). Bartholomew
ceptual categories so that plants can be grouped in ways concluded that the bare zones were created by herbivory,
that enable one to pose testable hypotheses and make not by allelopathy.
predictions. If the ecology of every plant species must be Another blow to allelopathy came from Lawrence
described uniquely, only very limited progress will ever Stowe (1979; Stowe and Wade 1979), who showed that
be made in understanding the ecology of any multi- allelopathic effects were more likely to be found in high-
species units, from plant communities and ecosystems to ly artificial conditions (such as in a petri dish) than in
biomes. It is as yet unclear whether this goal is attainable. more natural conditions (such as in a pot with soil), that
these effects were unrelated to previous claims of
Allelopathy allelopathy in the species tested, and that laboratory and
Do plants engage in “chemical warfare” with other greenhouse results had no relationship to the likelihood
plants, releasing toxins to poison neighbors? If this pur- that two species grew together in an old field. Other
ported phenomenon, called allelopathy, could be problems with allelopathy include the question of how
demonstrated, it would be a potential mechanism by allelopathy could evolve in the face of autotoxicity
which plants could secure uncontested access to (Newman 1978; Harper 1977). Ecologists have become
resources. However, allelopathy has been notoriously largely skeptical of the literature on allelopathy. This
difficult to demonstrate unambiguously in the field. The does not imply that allelopathy cannot exist, only that it
major problem is in separating the effects of allelopath- has yet to be convincingly demonstrated to be operat-
ic chemicals from the effects of other factors affecting ing in real communities.
neighbors’ performances and distributions, particular- What appears to be allelopathy may sometimes
ly resource competition and herbivory. Another diffi- reflect more ordinary mechanisms of competition. Cen-
culty is detecting the toxic compounds in the field and taurea diffusa (diffuse knapweed, Asteraceae) is an impor-
determining whether they are released naturally in suf- tant invasive weed of arid grasslands—especially range-
ficient quantities to harm neighboring plants. To further lands—in western North America; it presently infests
complicate the situation, the compounds released by the about 1.2 million hectares in the United States. Ragan
plant may not be harmful in themselves; rather, their Callaway and Erik Aschehoug (2000) grew C. diffusa in
degradation products may be the effective agents. There combination with a number of species from its native
is no doubt that many plants contain toxic substances Eurasian range and its range in North America. They
that can harm other plants when extracted and applied found that C. diffusa had stronger negative effects on the
to those plants, but these compounds may be serving growth of the North American plants; moreover, the
strictly or primarily antiherbivore, antibacterial, or anti- Eurasian plants, but not the North American plants,
fungal functions. reduced the size of C. diffusa when grown in common
A good bit of the early research on allelopathy was pots. While their study pointed to competition for phos-
motivated by the observation that in some systems, bare phate between C. diffusa and the other species, Callaway
patches of ground often surrounded plants of certain and Aschehoug argued that allelopathy was also
species. These plants often seemed to be ones with pun- involved. In one treatment, they added activated carbon
gent aromatic scents, indicating the release of volatile to the soil to absorb any hypothesized allelopathic chem-
chemicals into the environment. In the early 1960s, Muller icals. This treatment improved the relative competitive
and his colleagues and students claimed to have found ability of the North American plants, but not their abil-
evidence for allelopathy among shrubs of the California ity to absorb phosphate. Besides absorbing chemicals,
chaparral (Muller 1969). These aromatic woody plants are however, activated carbon has other effects on soil and
often striking in the lack of any herbaceous understory, soil microbes, as the researchers noted. However,
whereas herbaceous plants are abundant just outside of Katherine Lejeune and Timothy Seastedt (2001) have
the bare halos surrounding the shrubs. Another area of suggested that C. diffusa and four congeners (all impor-
allelopathy research was based on the suggestion that tant invasives in western North America) may achieve
dominant grassland species could inhibit succession to competitive dominance by altering the balance of soil
woody vegetation for many decades by allelopathic inter- nutrients, rather than through allelopathic interactions.
actions with soil microorganisms (Rice 1974).
In a very clever experiment testing the claims of Modeling Competition
Muller, Bartholomew (1970) placed small wire exclo-
sures in the halos under chaparral shrubs purported to Models of plant competition seek to explain the vast
be allelopathic to keep out herbivorous small mammals. diversity of plant life by identifying the circumstances
under which competitors can coexist. Models of plant effects on species A. Raising this term to the power b
competition may also try to explain the mechanisms by allows density to have nonlinear effects, and the entire
which competitors interact and the characteristics of quantity is multiplied by aA, which measures the effect
winners and losers. Many basic textbooks in ecology of these standardized neighbors on mean weight. The
introduce a single model of competition, the Lotka- equation for numbers of individuals is similar.
Volterra model of two-species competition, or at least Because mortality changes in response to density,
use this model as a departure point for their discussion the number of individuals of species A at the end of the
of the way species compete. We do not do so in this experiment will be equal to the number at the start,
book, as this model is not very useful for describing com- divided by one plus the standardized initial number of
petition among plants. plants of both species (by using β, the per-individual
Unfortunately, no comparably simple or generally effect of species A on species B), together multiplied by
accepted model exists for competition among plants. We the effect of neighbors on the survival of species A, mA.
discuss several widely cited models. The first two of Law and Watkinson (1987) extended this model by
these are equilibrium-based models, in that they relaxing the requirement for a constant value for the
depend on the populations reaching a stable end point competition coefficients and allowing those coefficients
at which the outcome of competition is determined. The to vary with the frequency and density of the two
other models do not make this assumption, and in gen- species. Their model describes rather than predicts com-
eral depend on chance events in one way or another in petitive outcomes because the outcomes are determined
determining competitive outcomes. by fitting parameters empirically—the values used in
the equations are determined by the results obtained in
Two Equilibrium Approaches the experiment.
One model that describes competition in additive design In an attempt to model resource competition mech-
experiments was developed by Watkinson and his col- anistically, Tilman (1982) introduced the resource ratio
leagues (Firbank and Watkinson 1985; Law and Watkin- model for plant competition (Figure 10.14). This model
son 1987) to generalize the self-thinning equation for predicts that stable coexistence among competing plants
two-species mixtures. Recall that the self-thinning law is possible for plants that are competitively superior at
describes a linear, negative relationship between aver- different ratios of essential resources. However, the
age plant weight and plant density when expressed on resource ratio model has not been well supported by
a log-log scale. In the two-species model, the intensity experimental tests. This may be because plant commu-
of competition is estimated by the competition coeffi- nities are likely rarely at equilibrium, and many other
cients α and β. For species A, mean weight per plant in processes, such as herbivory, contribute to determining
mixture, w, is predicted to be the outcome of competition. The few experimental tests
wmA of one component of the model, the R* rule (discussed
wA =
(1+ aA (NA + α NB ))b
above), offer some support for this one component, par-
ticularly when plants are competing for soil nitrogen
and density-dependent mortality is described by (summarized by Grover 1997).
NiA
N sA = Patch-Based and Nonequilibrium Models
1+ mA (NiA + βNiB )
A number of conceptual and mathematical models of
where the subscripts A and B specify which of a given competition and its role in community structure have
variable represents the two species; wm is the average used nonequilibrium and nondeterministic approaches,
weight per plant in the absence of competition (i.e., wmA in which chance events have a key role in determining
is the mean weight for species A in absence of competi- competitive outcomes. Alexander Watt (1947) introduced
tion); aA and b are fitted parameters; α and β are com- the idea of the plant community
petition coefficients that measure the average effect of as a mosaic of patches. These
an individual of species B on an individual of species A patches, he maintained, are dy-
and vice versa; and Ni and Ns are the initial and final namically related to one another.
densities (Firbank and Watkinson 1990). Analogous There is a tension between pre-
equations are used for species B. dictable order and chance events
This model expresses the mean weight for species that tend to disrupt that order,
A as being equal to the mean weight in the absence of resulting together in the structure
competition, divided by one plus the total competitive of the community. Competitive
effects of neighbors. The (NA + αNB) term gives the stan- interactions played a major role
dardized number of neighbors because α allows us to in Watt’s view of patch dynamics
equate individuals of the two species in terms of their (Figure 10.15). Watt specifically Eddy van der Maarel
(A) Exclusion (B) Exclusion species-rich grasslands on the island of Öland in south-
ern Sweden. Van der Maarel emphasized the small-scale
A B B A mobility of plants both as individuals, as parts of plant
die and shoots colonize adjacent gaps, and through
1 2 3 1 6 5
propagules. The “carousel” is the cycle of species replace-
ZNGIB ment, which may go around quickly or slowly.
ZNGIA A formal model that incorporated competitive inter-
actions and environmental patchiness was developed by
Avi Shmida and Stephen Ellner (1984). This model had
R2 a strong influence on ecologists’ thinking about com-
(C) Exclusion (D) Equilibrium coexistence petition and species coexistence. Shmida and Ellner’s
CB CA lottery model assumed that the outcome of competition
A B A B among juveniles for space is determined by chance (a
1 2 3
CA
1 2 3 “lottery”). This lottery for space can be biased, in that
4 4
CB some juveniles may have better chances than others. The
5 5 model incorporates asymmetric competition (juveniles
6 6
cannot displace adults) for microsites, nonuniform seed
dispersal, and larger-scale spatial and temporal patchi-
ness (Figure 10.16). In contrast to most previous models,
this one suggested mechanisms for species coexistence
R1
without requiring differences in habitat or resource use.
Figure 10.14 Several recent advances have been made in model-
The possible outcomes in David Tilman’s resource ratio ing plant competition and species coexistence, particu-
model of competition in which species A and B are compet-
ing for resources R1 and R2 with resource uptake vectors CA larly the explicit inclusion of the effects of spatial loca-
and CB. Each point on the graph represents the supply tion. Using invasion analysis models, Lavorel and
(availability) of the two resources in a particular environ-
ment. Growth is positive for a species in the region above or
to the right and negative below or to the left of the isocline
(the line where growth is 0) for that species (ZNGIA for Calluna
species A and ZNGIB for species B). The numbers indicate
different resource supply combinations in which neither,
either, or both species are able to exist (alone and/or in com-
petition with the other). (A) Competitive exclusion of species
B by the competitively superior species A, where species A
can reduce resource levels below those needed by species B.
(B) Competitive exclusion of species A by the competitively Arctostaphylos Cladonia
superior species B. (C) Unstable equilibrium occurs when
resources are drawn down to the point indicated by the
open circle, at which there is unstable coexistence of the two
species in region 4 of the graph and competitive displace-
ment of one or the other species in the other regions. (D) Sta-
ble equilibrium, in which there is coexistence of the two
species in region 4 of the graph and competitive displace-
ment of one or the other species in the other regions. Stable Bare soil
equilibrium is reached at the point indicated by the closed
circle where CA and CB cross. The resource uptake (or con- Figure 10.15
sumption) vectors CA and CB define the region where coexis- Watt’s model of vegetation dynamics in the Dwarf Callune-
tence is possible. (After Tilman 1982.) tum. This plant community, at about 800 meters altitude in
the Cairngorm Mountains of Scotland, has a natural pattern
of double strips of Calluna (heather, Ericaceae) and Arc-
tostaphylos (bearberry, Ericaceae) separated by wind-swept
bare soil. Watt explained the dynamics of this pattern as fol-
excluded successional change from this conceptual lows: The young shoots of Calluna exclude lichens, but as
model of the maintenance of community structure. they age, their competitive ability diminishes, and the area is
Contemporary views of plant communities are invaded by Cladonia (a mat-forming lichen). The constant
remarkably similar to Watt’s perspective (Pickett wind prevents the Cladonia mat from persisting, however,
and White 1985). Eddy van der Maarel and coworkers and in time it disintegrates, leaving bare soil. Arctostaphylos
then invades by vegetative spread, eventually achieving
(van der Maarel and Sykes 1993) extended Watt’s per- complete occupation of the patch. Young shoots of Calluna in
spective in the “carousel model” they developed to time spread in from the margins, eventually competitively
describe the dynamics of species on very small scales in replacing the Arctostaphylos. (After Watt 1947.)
Figure 10.16
“Lottery model” of the role of
microsite occupation and dis-
Establishment
Adult
persal in a community. The
Population change = by – mortality diagram pictures what hap-
Disperser juveniles
pool pens within a single microsite
Dispersed over the course of one year.
seeds Adults within the population
and from outside the popula-
tion contribute seeds to the dis-
perser pool. Some of these
Input from
seeds leave the population, but
other sites some proportion of the incom-
Nondispersed
seeds
ing and nondispersed seeds
germinate to produce juve-
niles. Seeds produce successful
juveniles only when an adult
Adult dies Many Single dies, because they cannot sur-
Adult Adult
Year t juveniles juvenile Year t + 1
vive under an adult. A juvenile
can grow to become an adult in
the following year; an adult
may die or survive as an adult
Adult lives in the next year. (After Shmida
and Ellner 1984.)
Chesson (1995) simulated competition between two tions concerning competition among plants, and we do
species with detailed attention to spatial pattern and dis- not yet have good answers for them. Within a topic that
persal strategies under different disturbance regimes. generates a great deal of controversy, these questions
They found the greatest species coexistence at interme- seem to be the most vigorously debated, perhaps
diate levels of habitat suitability and disturbance fre- because they bring together all of the other issues dis-
quency. Spatial patterns interacted with species disper- cussed in this chapter about which there is disagreement.
sal patterns and regeneration strategies in a complex There is insufficient evidence as yet to either accept or
fashion. reject any of the general theories of plant community
Cellular automata models represent another organization or of the role of plant competition in deter-
approach to modeling competition spatially (Balzter et mining community composition.
al. 1998). These dynamic models have been used for a We must first distinguish between the importance,
variety of processes at a variety of scales, from spatial- intensity, and significance of competitive interactions.
ly dependent molecular interactions to the development Competition is important if it plays a major role in deter-
of spiral galaxies, as well as spatially dependent eco- mining community composition. Competition can be
logical processes. Balzter et al. (1998) demonstrated the intense—that is, have major effects on various aspects of
usefulness of this approach for plant communities with individuals’ performances—without having much effect
a preliminary model of the dynamics of three plant at the community or even at the population level. Com-
species—Lolium perenne (ryegrass, Poaceae), Trifolium petition is ecologically significant if it plays a substantial
repens (white clover, Fabaceae), and Glechoma hederacea role in determining species coexistence within a com-
(ground ivy, Lamiaceae)—growing in a lawn over a munity.
three-year period. Eventually this approach may prove In one of the earliest experimental studies of species
valuable in making predictions about species’ abun- interactions, Arthur Tansley sought to determine the role
dances and conditions for coexistence in a spatially of competition in the distribution of two species of Gal-
explicit context. ium (bedstraw, Rubiaceae), both small herbaceous peren-
nials (Tansley 1917). Galium saxatile, the heath bedstraw,
Effects of Competition is found primarily on sandy soils, while Galium sylvestre
(probably the species now called G. sterneri), the lime-
on Community Composition stone bedstraw, is found primarily on calcareous, lime-
Is competition important in most environments? What stone-derived soils.
is its role in determining community composition, and Tansley grew plants of both species alone and
how important is it relative to other processes such as together in each of the two soil types in large wooden
herbivory? These are some of the most important ques- boxes outdoors. When grown alone in calcareous soil,
M M Limestone Transect A
M site
m c
M Deep Sand
M
MM site
C C West Plain
C C
m c C C C site
m c
Transect C
East Protea
West Protea Middle Protea site C C
site site C s c
S S S S S S S S C S C S C S C C C
s c s c
M C
Leucadendron meridianum Leucadendron coniferum
S C
Protea susannae Protea compacta Figure 10.17
Diagram of two transects, A and C, in the fynbos of southern
Ericoid shrubs Restioid shrubs Africa, showing the locations of the three experimental
study sites along each transect. Dominant species and plants
Planted seedlings s c m c
of some other growth forms are also shown. (After Richards
(experimental)
et al. 1997.)
limestone bedstraw grew normally, as expected. Heath among communities. What determines the typically sharp
bedstraw survived, but grew slowly and had yellow- discontinuities among communities dominated by dif-
ish leaves, indicating nutrient deficiency. When grown ferent species, and what is the role of competition in deter-
alone in sandy soil, heath bedstraw grew vigorously, mining these boundaries?
while limestone bedstraw survived, but grew poorly. Richards and his colleagues chose three transects,
When the two species were grown together in calcare- each of which crossed a sharp community boundary with
ous soil, limestone bedstraw overtopped heath bedstraw a transition from one distinct soil type to another. Each of
and eliminated it from the mixture. In sandy soil, heath the transects contained a different species pair in which
bedstraw became dominant, but did not completely one dominant species replaced the other (Figure 10.17).
eliminate limestone bedstraw during the course of the In a three-year experiment to compare the influences of
experiment. Tansley concluded that while each species soil type and interspecific competition, seeds of both
appeared to be adapted to the soil in which it lived in species were planted in monoculture and in mixture at
nature, competition also played an important role in three sites along each transect. Interspecific competition
determining the restriction of the two species to differ- consistently decreased growth, but the magnitude of its
ent soil types. effect was small compared with that of soil type. Adap-
A recent study in a different system found quite dif- tation to soil conditions strongly affected both seedling
ferent results. M. B. Richards and his colleagues at the uni- growth and survival at two of the three sites (Figures 10.18
versity of Cape Town in South Africa (Richards et al. 1997) and 10.19) and the researchers suggest that soil type, not
examined the relative importance of competition and competition, may be the critical factor determining species
adaptation to soil type among six shrubs belonging to the distribution.
Proteaceae. These species grow in the fynbos of south- Competitive interactions may change over time, and
ern Africa, one of the most species-rich places on Earth the ultimate outcome of competition may be different
for plants (see Box 20A). The fynbos is dominated by an than initial observations suggest. Lythrum salicaria (pur-
astonishing diversity of shrubs as well as other growth ple loosestrife, Lythraceae) is an invasive species in North
forms. The environment is characterized by nutrient-poor America that appears to be displacing native wetland
soils, summer drought, and rolling, dissected terrain. species (see Chapter 11 and Figure 14.3). Some ecologists
Large numbers of apparently ecologically similar species have argued, however, that evidence that Lythrum is
coexist in the fynbos, and there is great species variation actually displacing native species is weak. Mal and col-
(A) Transect A Figure 10.18

L. meridianum L. coniferum Mean survival (%) and biomass (g dry weight,
100 ± 1 standard error; error bars omitted where
too small to be visible) after three years of
80 experimental seedlings of (A) Leucadendron
meridianum (Proteaceae) and L. coniferum at
Survival (%)
60 transect A and (B) Protea susannae (Proteaceae)

and P. compacta at transect C when grown at
40 sites within their natural distribution (green
bars) or outside of it (gray bars). In transect A,
20 survival and biomass were both far greater
within the natural distributions of both species
than outside them. In transect C, survival and
biomass were lowest outside the natural distri-
18 L. meridianum L. coniferum butions of both species for most sites. (After
Richards et al. 1997.)
16
Biomass (g dry wt.)
14
12
10
8
6
4
2
Limestone Deep Sand West Plain Limestone Deep Sand West Plain
Sites
(B) Transect C
P. susannae P. compacta
100
80
Survival (%)
60
40
20
14 P. susannae P. compacta
12
Biomass (g dry wt)
10
8
6
4
2
West Protea Middle East Protea West Protea Middle East Protea
Sites
leagues (Mal et al. 1997) carried out a four-year field com- inant, displacing Typha. The researchers attribute this
petition experiment between Lythrum and Typha angusti- result to differences in life history strategy between the
folia (cattail, Typhaceae), a dominant native wetland two species. Typha has large rhizomes with substantial
species. Typha was initially competitively superior, but in stored resources, which might give it an initial competi-
the second and third years of the experiment, the species tive advantage, but the high costs of producing new ram-
were relatively evenly matched. By the fourth year of the ets and the strong suppressive effects of Lythrum led to
study, Lythrum, the invader, became competitively dom- the eventual competitive replacement of Typha.
(A) Limestone site Deep Sand site West Plain site

Within natural distribution Outside of natural distribution
100
Interspecific competition
80 Intraspecific competition
Survival (%)
60
40
20
Low Med High Low Med High Low Med High

Density
(B) Limestone site Deep Sand site West Plain site

Within natural distribution Outside of natural distribution
30
Biomass (g dry wt)
25
20
15
10
5
Low Med High Low Med High Low Med High

Density
Figure 10.19
(A) Mean survival (%) and (B) biomass (g dry weight, ± 1 Conceptual Models of Competition in Habitats
standard error) after 3 years for experimental seedlings of L. with Differing Productivities
meridianum planted at three densities, in monoculture and in
mixture, at three sites along transect A. Competition did not Grime (1977, 1979) proposed that competition is unim-
appear to have much effect on biomass or survival, or may portant in unproductive environments, and that success
have had inconsistent effects that were far smaller than the in these environments is dependent largely on ability to
effects of being within or outside the natural distribution. L. tolerate abiotic stress (low nutrients, drought, or cold,
meridianum did not decrease in survival at increased densi-
ties, and decreased in biomass at increased densities only at
for example), rather than on competitive ability. He fur-
the Limestone site. Performance in mixture (green bars) was ther argued that in environments where disturbance fre-
generally better than performance in monoculture (gray quently reduces plant mass, competitive exclusion
bars). Patterns were similar for other species and transects. should be prevented. The dominant plants in such envi-
(After Richards et al. 1997.) ronments should not be competitively superior, but
rather should possess traits that allow them to withstand
disturbance or recolonize rapidly following disturbance.
Most subsequent discussion has focused on unpro-
ductive environments. Newman (1973) disagreed with
Competition along Environmental Grime’s characterization, arguing that competition is
important in low-resource as well as high-resource envi-
Gradients
ronments, but that the resources for which plants com-
One of the most critical and contentious issues concern- pete differ—light in productive environments, nutrients
ing the importance of competition in plant communities and water in unproductive environments. Later work
is whether there are particular kinds of habitats in which by Tilman (1987) reinforced and developed Newman’s
competition is predictably strong, determining commu- ideas. Tilman argued that competition in low-productiv-
nity composition, or predictably weak and unimportant. ity environments would be for belowground resources
Ecologists agree that competition is intense in produc- (primarily nitrogen), whereas competition in high-pro-
tive, nutrient-rich habitats, at least when disturbance and ductivity environments would be primarily for above-
herbivory are low. In these environments, plants are able ground resources (light). Aerts (1999) reiterated Grime’s
to develop large canopies quickly, and competition is argument in part, maintaining that selection in nutrient-
thought to be primarily for light. The relative intensity poor habitats would favor traits that reduce nutrient loss-
and importance of competition in productive and unpro- es rather than those that enhance the ability to compete
ductive habitats, however, remain matters of debate. for nutrients, resulting in slow growth rates.
Keddy (1990) adapted to plant community organi- is freely available for short periods of time, but there are
zation the “centrifugal theory of community organiza- long interpulse intervals in which water is partly or com-
tion” that Rosenzweig and Abramsky (1986) had pro- pletely unavailable. They hypothesized that growth and
posed for desert rodent communities. The centrifugal competition should be limited to the periods of high
model proposes that there is a core habitat type preferred water availability (the pulses) in arid environments.
by all species in a region, presumably with ideal grow- Growth and competition in low-nutrient soils, in con-
ing conditions. Other habitat types, called peripheral trast, should not be limited to pulses of short duration.
habitats, are defined by particular negative conditions
(stresses or disturbance) to which only some of the Experimental Evidence
species are adapted (Figure 10.20). Interspecific compe- The evidence for and against variation in the intensity
tition is most intense, Keddy suggests, in core habitat, of competition along productivity gradients remains
and is more relaxed in the peripheral habitats because confused and ambiguous. Wilson and Keddy (1986)
fewer species are adapted to the particular conditions in compared the competitive abilities of six species that are
each of them. Thus the peripheral habitats serve as dominant at different points along a productivity gra-
refuges, preventing competitive exclusion. In wetlands dient on the shore of Axe Lake, Ontario. The gradient
in Ontario, Canada, for example, all species prefer sites ranged from wave-disturbed, nutrient-poor beaches
with high fertility and low disturbance rates (core habi- with low standing biomass to sheltered, nutrient-rich
tat), while the peripheral habitats are defined by infer- sites with dense vegetation. Plants were grown in com-
tile soils and disturbances such as ice scouring. The core petition in plastic beakers at a protected site in the field,
habitat is dominated by Typha latifolia, while different using substrate from the favorable (sheltered, nutrient-
species dominate as one moves toward more extreme rich) end of the gradient. The researchers found that
conditions in each peripheral habitat type. competitive ability (measured as both competitive
One of the problems with trying to resolve the response and competitive effect) was positively corre-
debate about the relative intensity and importance of lated with mean position along the productivity gradi-
competition in productive and unproductive habitats ent (Figure 10.21). They interpreted these results as evi-
is that environments can be unproductive for very dif- dence that species with high competitive ability
ferent reasons. Many of the hypotheses about competi- occupied nutrient-rich, undisturbed sites, while species
tion in unproductive environments implic-
itly focus on low nutrient levels. But low
productivity may also be due to cold tem- Sandy
peratures, short growing seasons, saline river banks
soils, or toxic materials in the soil that inhib-
it growth, such as heavy metals. Inadequate d Drosera
re s
ou ank
In sw
c Hypericum
water supply is one of the most important
te ale
s b
e
rd s
factors limiting productivity globally. Ic ver is
un
ri ar
Scirpus
Rh
al
c h americanus
yn
Plants are not only likely to have very el o s

ch
E u us
irp an m
os
different adaptations to cold, heavy metals, Scirpus

po
Sc eric niu
Cl
r
od
and drought than to low nutrients, but

a
m g a acutus
iu
a ar a
m
species’ interactions under each of these con- p h

Ca
S p
Ty
re
ditions are also likely also to be very differ- Typha

x
Scirpus americanus
Scirpus fluvlatilis
Eleocharis calva
sand and clay
ent. Goldberg and Novoplansky (1997) pre- Ty

Ice scoured
Polygonum
ph
Beaver
a
Leersia
Leersia
dicted that the effects of competition in

ponds
Bidens
Typha
Typha
Carex
nutrient-poor environments would differ

substantially from those in water-stressed
10
00
habitats. Water availability, they pointed out,

75
is pulsed: even in arid environments, water

0
Cl
50
ad
0
D iu
ro m
Xy ser
25
o Rh ri a
0
Figure 10.20 ex s Cladium

The centrifugal model of plant community organiza- ia
Sa sh
la
tion, applied to the distributions of a number of wet-

ke
g/
nd or
Xyris
m
y es
land species in Ontario, Canada. Typha latifolia (cattail, Drosera

2
Typaceae) occupies core habitat, while other species Sabatia

become more prominent as one moves toward more
extreme conditions in more peripheral habitats. (After Gravel
Keddy 1990.) lake shores
(A) around target Hesperostipa individuals at three sites along

1.4 a gradient from a ridge crest to a moister lower slope.
Competition affected growth of mature plants, flower-
Lysimachia
1.3 Juncus ing, seedling establishment, and survival.
Dulichium When competitors were removed, growth and flow-
Target scores
1.2
ering for the mature Heterostipa plants were greatest on
1.1 Hypericum the lower slope, where its abundance was lowest. Com-
petition had the smallest effect on estimated population
Rhyhchospora
1.0 growth rates on the ridge tops, where Heterostipa was
r = 0.77
Eriocaulon P = 0.03
most abundant, and increasingly greater effects down-
0.9
slope. The largest effects were at the lowest sites where
Competitive ability
0.8 Heterostipa was present (Figure 10.22). These results

0 2 4 6 8 10 12 14 16 18
strongly suggested that competition was a major factor
in determining the distribution of Hesperostipa along this
(B)
gradient of productivity and environmental favorability.
1.4 r = – 0.68 Theodose and Bowman (1997) suggested the exis-
P = 0.07
Eriocaulon
tence of the opposite pattern, in which competition pre-
1.3
vented a species from a more productive area from
Neighbor scores
1.2 Rhyhchospora growing in a resource-poor site. The perennial Deschamp-

Hypericum sia caespitosa (hair grass, Poaceae) is common in moist
1.1 alpine meadows in the tundra of the Front Range of Col-
Lysimachia orado, but is rare in dry meadows. The dry meadows
1.0
are dominated by a sedge, Kobresia myosuroides (Cyper-
Juncus aceae). The authors hypothesized that Deschampsia was
0.9 Dulichium
prevented from growing in the dry environment by
0.8 competition with Kobresia. An earlier study had demon-
0 2 4 6 8 10 12 14 16 18
Exposed shores Sheltered shores strated that Kobresia was kept out of the moist meadows
Sediment organic matter content (%) by deep winter snow.
Figure 10.21 Theodose and Bowman transplanted individuals of
Competitive abilities of six wetland plant species and the each species, as well as two-species pairs of plants to a
locations where they are naturally found along a gradient dry meadow, either clipping the existing vegetation
from exposed, nutrient-poor to sheltered, nutrient-rich (largely Kobresia) at ground level or leaving it intact.
shores, corresponding to a gradient in the percentage of Deschampsia had a greater increase in survival in
organic matter contained in the sediment. (A) Competitive
abilities expressed as target scores, defined as the mean rela- response to vegetation clipping than did Kobresia, and
tive growth (increase in dry mass) of the target species when soil moisture was substantially depressed in plots with
grown in the presence of all neighbor species; this score is intact vegetation compared with those in which vegeta-
similar to competitive response. (B) Competitive ability tion was clipped (Figure 10.23). The researchers con-
expressed as neighbor scores, defined as the mean relative cluded that interspecific competition with Kobresia
growth (increase in dry mass of all neighbor species in the
presence of the target species; this score is similar to compet- excluded Deschampsia from the dry meadows by
itive effect. (After Wilson and Keddy 1986.) depressing soil moisture below the drought tolerance of
Deschampsia. Kobresia, with greater tolerance of drought,
was able to survive during periods of low moisture.
One problem with this conclusion, however, is that
with low competitive ability were displaced to disturbed the mortality of Kobresia (28%) was actually higher than
sites with poor soils, where competitive exclusion was the mortality of Deschampsia (15%) in intact vegetation.
prevented by wave action and low soil nutrients. Furthermore, most of the Deschampsia plants survived
Gurevitch (1986) carried out a field study of com- in the intact vegetation of the dry meadow. Therefore, it
petition along an environmental gradient in southeast- is difficult to argue convincingly that competitive exclu-
ern Arizona. She hypothesized that Hesperostipa neomex- sion was the result of high mortality in Deschampsia. The
icana, a C3 grass, was limited to arid ridgetops by growth of Deschampsia in intact vegetation was also
competitively superior C4 grasses. This is precisely the greater than the growth of Kobresia, and the growth of
opposite of what one would expect if physiology were both species was about equally affected by vegetation
determining species distributions, as C4 species should clipping. Nothing is known about the effects of compe-
be better able to tolerate the unfavorably hot, dry con- tition on reproduction or establishment for these species
ditions on the ridge crests. She removed neighbors from in these environments. So, although this study clearly
Ridge crest Midslope Lower slope Wash

Hes C4’s Hes C4’s Hes C4’s Hes C4’s
High, not
Biomass (g/m2) 140 16 76 111 1 282 0
measured
% Cover 20.5 9.5 8.5 24.5 3.0 37.0 0 ~100%
Hesperostipa: Ridge crest Midslope Lower slope

Control Removal C R C R 0.0
Seedlings/m2 4.55 9.36 0.58 2.04 0.10 0.35 Ridge

–1.0
ψ Soil (MPa)
Seedling survival 0.47 0.79 0 0.56 0 0.50
final size –2.0
Growth initial size 0.77 1.97 0.69 2.58 1.38 4.28
Number of flowers 1.50 13.30 4.10 28.70 35.50 58.90 –3.0 Wash
Population growth 0.93 1.04 0.83 0.92 0.59 0.88
(λ) –4.0
Figure 10.22 10 Feb 08 Mar 29 Apr 18 Aug

Results of a removal experiment at three topographic posi-
Date
tions in a southeastern Arizona grassland (elevation 1400 m).
The table at the top shows means for biomass and cover (N
= 40; confidence limits available in Gurevitch 1986) at the
three experimental sites (ridge crest, midslope, and lower
slope) and at a lower topographic position below them
(wash). The lower table gives experimental results for Con-
trol (neighbors not removed) and Removal (mature neigh-
bors removed) treatments at the three experimental sites.
Means for seedlings/m2 (in 1980 plus 1981), mature plant
growth, and number of flowers produced by mature plants
(N = 20) over the 20 months of the experiment are shown.
Seedling survival was based on the total number of
seedlings surviving, and population growth was estimated Neighbors removed
as described in the original paper. The graph at right shows Intact vegetation
soil water potential at the top (ridge crest) and bottom 40
(wash) of the topographic gradient at the 15–20 cm depth 35
(where the bulk of the roots were located) over a drying and
rewetting cycle of 6 months in 1980. (After Gurevitch 1986.) 30
Mortality (%)
25
20
demonstrated intense and statistically significant effects 15
of competition in this resource-poor habitat, more work
10
needs to be done to conclusively demonstrate that com-
petition leads to the exclusion of Deschampsia in the dry 5
meadows. 0
Deschampsia Kobresia
Evidence from Research Syntheses Figure 10.23
While the results of individual studies are apparently Mortality of Deschampsia caespitosa (Poaceae) and Kobresia
contradictory, there have been several attempts to gain myosuroides (Cyperaceae) in a dry meadow when transplant-
a better overview of competition along environmental ed into intact vegetation (gray bars) and with neighbors
clipped (green bars). Values are mean mortality (± 1 stan-
gradients. In a cross-continental set of field experiments, dard error) based on four plots, each of which had 10 experi-
the intensity of competition was compared for trans- mental plants per species. (After Theodose and Bowman
planted Poa pratensis (bluegrass, Poaceae) individuals at 1997.)
(A) Figure 10.24

Results from a large set of field com-
0.07
9 petition experiments in which
9 perennial bluegrass (Poa pratensis)
was planted in 44 plots across nine
Absolute competition
0.05 sites in locations across the world.

intensity (ACI)
9 Each point represents a plot; num-

8 8 bers indicate particular sites. At
1 6
0.03 1 1 6 each site, competition was studied
5 6
33 6 8 8 over a gradient of neighbor bio-
1 19 2 5 mass, so there is more than one
6 4 3 8 8
4
0.01 1 2 3 32 5 76 experimental outcome shown for
347 each site. Competition was meas-
5 7
7 ured as (A) absolute competition
9 7
–0.01 intensity (ACI) and (B) relative com-
0 100 200 300 400 500 600 700 petition intensity (RCI). (After Read-
er et al. 1994.)
(B)
3
1
Relative competition
2
12 3
intensity (RCI)
1
2 2 9 9
6 9
1 6 6 66 6
1 1 38 5 8
4 34 3 83 57
8 8
9 47 5 88
7 5 7
0
7
9
–1
0 100 200 300 400 500 600 700
Biomass of neighboring plants (g/m2)
12 locations in Europe, North America, and Australia mass as a surrogate for productivity. The meta-analysis
(Reader et al. 1994). The intensity of competition was had surprising results: when the measure of competition
compared across sites over the range of standing bio- intensity used was the log response ratio (LRR), the
mass (a surrogate for productivity) within each site. The researchers found a strong negative relationship—the
authors reported the results using two indices of com- opposite of what was expected—between competition
petition intensity, relative competitive intensity (RCI) intensity and productivity for final plant biomass and
and absolute competitive intensity (ACI). There was survival, but not for growth (Figure 10.25). When com-
some suggestion that ACI increased as neighbor biomass petition intensity was measured using RCI, they found
increased, but RCI showed no clear relationship to no clear relationship between RCI and vegetation bio-
neighbor biomass (Figure 10.24). The authors conclud- mass except, again, a negative relationship for survival.
ed that there was no convincing evidence to support the It is difficult to know what these results mean for our
hypothesis that competition increased along a gradient understanding of the role of competition in environments
of increasing neighbor biomass, when measured across of differing productivity. Clearly, however, the hypoth-
a wide range of sites and productivities. Since there are esis of increasing competition intensity with increasing
flaws in both ACI and RCI, it would be interesting to productivity is not supported as a general pattern, at least
examine the results using other approaches to measur- when competition is measured using these indices.
ing competition intensity.
In another approach to synthesizing results over a Resolution of Differing Results
greater range than is possible in an individual experi- What explains the lack of agreement among these stud-
ment, Goldberg et al. (1999) carried out a meta-analysis, ies, and what can be done to achieve a better under-
or quantitative synthesis, of 14 papers reporting a large standing of the importance of competition over a range
number of outcomes of competition experiments. The of environments? Performing experiments in a relative-
synthesis examined competition and facilitation among ly consistent way across many sites, as Reader and col-
plants along productivity gradients, using vegetation bio- leagues did, is a big step in the right direction, as is using
Relative competition intensity (RCI) Log response ratio (lnR)

Final biomass Final biomass
1 4
0
2
ln R
RCI
–1
0
–2
–3 –2
0 3500 7000 0 3500 1500
Exponential growth Exponential growth

3 4
2
2
1
ln R
RCI
0
0
–1
–2 –2
0 500 1000 1500 0 500 1000 1500
Survival Survival
1 4
0
–1 2
ln R
RCI
–2
–3 0
–4
–5 –2
0 500 1000 1500 0 500 1000 1500
Standing crop (g/m2) Standing crop (g/m2)
Figure 10.25
Values for two measures of response to competitors—relative competition intensity
(RCI) and log response ratio (LRR)—across 14 separate published studies over a
range of natural biomass (standing crop, g/m2). Regression lines are shown only
where there was a statistically significant relationship between the variables. Positive
values of RCI or LRR indicate that competition is occurring, and negative values
indicate that neighbors have a beneficial effect on target plant performance. (After
Goldberg et al. 1999.)
modern quantitative synthesis techniques to bring competition look at individual growth (or other meas-
together large numbers of independent studies, as Gold- ures of the responses of individual plants) in response
berg and her colleagues did. It is clear from these two to competition, rather than estimating population
efforts at synthesis that the conclusions can vary great- responses. Comparisons of RCI, for example, cannot
ly depending on how the experiments are designed, how show whether competition restricts where a species is
long they are maintained, what is measured, and how found; it just does not provide enough information. If
the results are analyzed. the question is what limits the distributions of species
Some of the inconsistencies in the conclusions or determines community composition, studying pop-
among different researchers are almost certainly a result ulation-level responses is the only way to get an accu-
of these artifacts. More profoundly, most studies on plant rate answer.
species distributions, although the experimental evi-

Summary
dence for its existence is weak.
The importance of competition in shaping species dis- Whether and how competition structures commu-
tributions and community composition has been debat- nities may depend on whether dominance hierarchies
ed by ecologists for many years. Competition among are consistent among species and environments. If a
plants is a reduction in fitness due to shared use of a species is dominant in one environment, will it contin-
resource that is in limited supply. The resources for ue to be dominant in other environments? Is the inten-
which plants compete include light, water, mineral nutri- sity of competition the same in all environments, or is
ents, space, pollinators, and seed dispersers. Most com- it weaker in low-resource or stressful environments?
petition occurs among adjacent individuals. Thus, local Resolving these questions is difficult, in part because the
density, not the total density of a population or com- answers depend on which measure of competition is
munity, determines the intensity of competition for an used and whether one measures competitive effect—the
individual. effect of an individual on its neighbors—or competi-
Competition often results in large size differences tive response—the effect of the neighbors on the indi-
among individuals, ultimately translating into differences vidual. Many experimental designs are used to assess
in survival and reproduction. Larger individuals usual- competition, including various ways of manipulating
ly have a competitive advantage over smaller individu- densities and frequencies. These experiments can be car-
als. However, larger size is not always advantageous. ried out in greenhouses, transplant gardens, and natu-
The outcome of competition can be determined by ral populations.
a variety of mechanisms. One attempt to predict com- Models of competition are built on one of two
petitive outcome is the R* rule: over the long term, in a assumptions: (1) that populations reach a stable end
constant environment, the species with the ability to point at which the outcome is determined, or (2) that
deplete resources to the lowest level when grown alone chance events, such as disturbance and herbivory, deter-
will competitively displace all other species. There is lim- mine the outcome of competitive interactions. In the next
ited experimental evidence to support this proposition. chapter we explore the role of herbivory and other types
Allelopathy—the release of toxins to poison competing of species interactions in shaping plant populations and
neighbors—has long been hypothesized to explain some communities.
Additional Readings
Classic References Goldberg, D. and A. Novoplansky. 1997. On the relative importance of
competition in unproductive environments. J. Ecol. 85: 409–418.
Newman, E. I. 1973. Competition and diversity in herbaceous vegetation.
Nature 244: 310. Grace, J. B. 1995. On the measurement of plant competition intensity.
Ecology 76: 305–308.
Tansley, A. G. 1917. On competition between Galium saxatile L. (G. her-
cynicum Weig.) and Galium sylvestre Poll. (G. asperum Schreb.) on dif- Lavorel, S. and P. Chesson. 1995. How species with different regeneration
ferent types of soil. J. Ecol. 5: 173–179. niches coexist in patchy habitats with local disturbances. Oikos 74:
103–114.
Watt, A. S. 1947. Pattern and process in the plant community. J. Ecol. 35:
1–22.
Yoda, K., T. Kira and K. Hozimu. 1963. Self thinning in overcrowded pure Additional References
stands under cultivated and natural conditions. J. Biol., Osaka City Univ. Keddy, P. A. 2001. Competition. 2nd Ed. Kluwer Academic, Dordrect
14: 107–129. (Netherlands) and Boston.
Grace, J. B. and D. Tilman (eds.). 1990. Perspectives on Plant Competition.
Contemporary Research Academic Press, New York.
Aerts, R. 1999. Interspecific competition in natural plant communities: Grover, J. P. 1997. Resource Competition. Chapman and Hall, London.
Mechanisms, trade-offs and plant-soil feedbacks. J. Exp. Bot. 50: 29–37.
Balzter, H., P. W. Braun and W. Kohler. 1998. Cellular automata models
for vegetation dynamics. Ecol. Modelling 107: 113–125.
11
C H A P T E R
Herbivory and Plant-Pathogen
Interactions
G
reen plants are the foundation of almost all terrestrial food webs. All
animals (including humans, of course) ultimately depend on plants for
their existence. Yet casual observation seems to reveal a green world
teeming with (uneaten) plants. Why is the world so green? Conversely, what
are the consequences of herbivory from the plants’ perspective?
Herbivory is the consumption of all or part of a living plant. Some ecolo-
gists use the term “predation” when an herbivore eats and kills an individ-
ual. Seed predators, or granivores, are herbivores that consume seeds or grains,
killing the individual within. Grazers are herbivores that eat grass and other
ground-growing plants, while browsers eat leaves from trees or shrubs. Fru-
givores are herbivores that consume fruits, sometimes without damage to
the seeds.
Plants are consumed by organisms from a variety of kingdoms: animals,
fungi, bacteria, and even other plants. Herbivory can have ecological effects at
the level of the individual plant, the population, the community and landscape
(e.g., patterns of coexistence of plant species), and the ecosystem (e.g., nutri-
ent cycling). Herbivores can also influence the evolution of plants. Ecologi-
cally and evolutionarily, some of the most important herbivores are grazing
mammals and insects. However, other types of herbivores, such as birds, mol-
lusks, and nematodes, can be very important in particular systems.
In this chapter we look at the consequences of herbivory and of plant-her-
bivore interactions for plant population dynamics and for the structure of plant
communities. Can herbivory affect the trajectory of a plant population? Can
the presence of herbivores—or their exclusion—have a decisive role in deter-
mining what plant species are able to coexist in a community? We begin by
examining the effects of herbivory on individuals, populations, and commu-
nities, and briefly touch on its landscape effects. In addition, we look at how
plants defend themselves against herbivores and at how those defenses
evolved. Then we turn to the evolutionary consequences of herbivory and
the responses of plants. Finally, we take a look at what is known about the role
of plant-pathogen interactions in plant communities.
Herbivory at the Level of Individuals

What do herbivores do to individual plants? Herbivores can consume an entire
plant, causing the death of the individual. By eating seeds, granivores—such
as some ants, rodents, and birds—kill individual plants. Alternatively, herbi-
212 Chapter 11
vores can eat (or manipulate) only some of the parts of individuals, it would be called compensation.) The
a plant, damaging, removing, or destroying those parts, researchers suggested that overcompensation was due
but not necessarily immediately killing the plant. Deer, to the coevolution of plants and herbivores, particular-
for example, frequently restrict their consumption to ly grasses and mammalian herbivores. Buffalo saliva
only the newest leaves or parts of shoots. Herbivores can and urine, for example, were thought to contain growth
also live on or within a plant and consume some of the stimulants for grasses (Detling et al. 1980). These ideas
plant’s resources. Insects called aphids, for example, were highly controversial and received a great deal of
extract dissolved sugars and other nutrients directly attention; it seemed difficult for many ecologists to
from the phloem; pathogenic microorganisms can par- believe that being eaten could actually be a good thing
asitize a plant, depleting its resources over time. Some for plants.
plants are parasites on other plants, tapping water, sug- While there was some experimental evidence for
ars, proteins, and other resources for their own use. these ideas, when taken as a whole, they were not well
The effects of herbivory on the plant depend, among supported by the available data (Belsky 1986; Belsky et
other things, on what parts of the plant are consumed. al. 1993). One of the possible explanations for reported
Removal of or damage to roots can reduce or prevent the overcompensation was that researchers had measured
plant’s uptake of water and mineral nutrients and can only aboveground plant parts, while underground
make the plant more vulnerable to being toppled by reserves may have been depleted to stimulate the
wind, flooding, or soil erosion. Consumption of leaves observed aboveground growth. Long-term herbivory
reduces the photosynthetic surface area, while removal might result in significant depletion of these under-
of phloem sap may reduce the energy and materials ground reserves, ultimately harming the ability of the
available for growth and reproduction. Consumption of plant to recover from subsequent bouts of herbivory.
leaves, stems, and twigs may alter the competitive rela- Another explanation was that in dense plant stands, if
tionships among neighboring plants. Removal of meris- herbivores eat only shaded, unproductive understory
tems may alter the growth form of the plant. Consump- leaves, there may indeed be no reduction in the photo-
tion of flowers, fruits, and seeds may reduce the synthetic capacity of the plant, and thus no negative
potential contribution of the plant to the next generation. effects of herbivory. Joy Belsky and her associates (Bel-
Of course, to the new individual in each seed, con- sky et al. 1993) argued that overcompensation seemed
sumption means death. Alternatively, fruits are often to occur mainly in the experimental treatments most
consumed without damage to the seeds, in which case favorable to plant growth, such as the combination of
the frugivore may disperse the seeds to potentially high nutrient availability and reduced competition.
favorable locations. Ecologists now generally believe that herbivory usu-
Also important is the life history stage at which the ally harms individual plants (Hawkes and Sullivan 2001).
plant is attacked or damaged. Seedlings are particularly While there are cases of overcompensation, on average,
vulnerable to herbivores. One mouthful for the herbivore herbivory reduces both growth and reproductive output,
can kill a seedling, but hardly affect a more mature plant. resource addition increases both, and there is no evidence
Grazing on grasses that have just begun to flower can of an interaction between the two effects (Figure 11.1). One
critically affect their ability to produce seeds, whereas interesting observation is that plant responses to her-
heavier grazing after seeds have been shed may have less bivory depend to some degree on their growth forms and
of an effect on population dynamics. phylogenies: for example, monocots with basal meristems
How much do herbivores eat, on average? It has (such as grasses) have more regrowth after herbivory
been estimated that about 10% of the leaves of forest under high-resource conditions, while dicot herbs and
trees are lost every year to herbivores (Coley and Barone woody plants regrow more after herbivory under low-
1996). Herbivory is greatest in dry tropical forests, some- resource conditions. Unfortunately, most studies have
what less in tropical rainforests, and least in temperate used only short-term measures of plant response to her-
forests. Young leaves tend to be eaten more readily than bivory, not measures of lifetime fitness.
mature leaves in the Tropics. As you might expect, there
is tremendous variation among species, locations, and Herbivory and Plant Populations
years in the degree of damage caused by herbivores.
Can herbivory ever actually help plants to grow The extent to which herbivores affect plant population
or reproduce? In the 1980s and early 1990s, a group of dynamics is a highly controversial and unresolved
scientists postulated the existence of overcompensation, question. Two different explanations have been put
in which plants purportedly respond to herbivory by forward for why herbivores should not be important.
growing more (McNaughton 1983). (If a plant’s hypoth- The “top-down” school of thought argues that herbi-
esized extra growth in response to herbivory resulted vores are maintained at such low densities by their own
in no net difference between grazed and ungrazed predators that they rarely exert negative effects on
Herbivory and Plant-Pathogen Interactions 213
Growth the southeastern United States (Figure 11.2; Powers et

Reproduction al. 1999). While most bark beetles inhabit branches and
2 trunks of trees already undergoing stress, such as those
damaged by a lightning strike, some species attack and
– 95% CI)
kill healthy trees. The beetles enter the tree by chewing

1 a hole through the bark into the cambium, the actively
Log response ratio (+
growing layer under the bark, and lay their eggs there.
0
After hatching, the larvae feed on the cambium, destroy-
ing it and the vascular tissue. Coniferous trees (particu-
larly pines) usually respond to boring by oozing sap
–1 (also called pitch) into the wound, thus either suffocat-
ing the beetle or pushing it out of the hole. However, a
massive attack by a large number of beetles appears to
–2
Herbivory Resources Interaction reduce the ability of the tree to “pitch out” the beetles.
Stresses created by drought or injury may have the same
Figure 11.1 effect in making trees more vulnerable.
The effect of herbivory on both growth and reproduction is
generally negative. The graph shows the average effects on
growth in 82 studies and reproduction in 24. The response
ratio is the plant performance when subjected to herbivory
or treated with resources divided by plant performance
without additional resources or herbivory; error bars are
95% confidence intervals. A negative ratio means that the
treatment decreased plant performance, while a positive
value indicates increased performance. Resource addition
improved plant performance, but there is no evidence of an
interaction between resource addition and herbivory.
(After Hawkes and Sullivan 2001.)
plant populations (e.g., Strong et al. 1984). The “bot-

tom-up” school argues that plant populations are lim-
ited by abiotic factors such as water, light, and soil
nutrients, not by herbivores (Hairston et al. 1960;
Slobodkin et al. 1967; Hairston and Hairston 1993).
In contrast, others have argued that herbivores do
play an important role in controlling plant populations.
In all likelihood, herbivory regulates plant population
dynamics in some cases, while “top-down” or “bottom-
up” processes dominate in others. More work is needed
to determine not only under what circumstances each of
these processes is important, but also what factors lead to
the predominance of one kind of regulatory process over
others, and when their interaction becomes important.
One of the most obvious situations in which plant
populations are dramatically affected by herbivores is the
killing of large stands of forest trees by insects. Outbreaks
of lepidopteran larvae can cause massive defoliation, in
some instances resulting in tree mortality. Repeated defo-
liation of oaks in the northeastern and midwestern Unit- Figure 11.2
ed States by gypsy moths has caused die-offs of large A “ghost forest” of Pinus albicaulis (whitebark pine,
numbers of trees (Davidson et al. 1999). Trees in more Pinaceae) killed by Dendroctonus ponderosae (mountain pine
mesic areas appear to be more likely to die from a single beetle, Scolytidae) in an extensive outbreak during the late
episode of gypsy moth defoliation than those in more 1920s. The young trees growing on this site are mainly Abies
lasiocarpa (subalpine fir, Pinaceae) and Picea engelmannii
xeric sites. (Engelmann spruce, Pinaceae). Scolytid beetles cause exten-
Bark beetles (family Scolytidae) are an important sive damage to conifers in western and southeastern North
cause of conifer mortality in western North America and America. (Photograph courtesy of K. Kipfmueller.)
214 Chapter 11
Bark beetles that are responsible for widespread tree imental plants, which led to increases in the total num-
mortality in North America include the Mexican bark ber of seeds set, seedling density, and the number of
beetle, the western pine beetle, the spruce beetle, the flowering adults.
western balsam bark beetle, and the southern pine bee-
tle. As their names suggest, bark beetles are fairly selec- Herbivory and Spatial Distribution of Plants
tive, generally specializing on one or a few species of The spatial distribution of plants can also be affected, or
conifers. The southern pine beetle, for example, prima- even determined, by herbivores. The role of granivorous
rily attacks Pinus echinata (shortleaf pine, Pinaceae), P. rodents in the distribution of a range grass with large,
taeda (loblolly pine), P. palustris (longleaf pine), and P. nutritious seeds, Achnatherum hymenoides (Indian rice-
elliottii (slash pine); the spruce beetle kills spruces. The grass, Poaceae), was studied in the desert of western
beetles function with symbiotic fungal partners that also Nevada. Indian ricegrass is common in sandy soils, but
attack the trees, eliciting specific biochemical responses rare in adjacent rocky habitats (Breck and Jenkins 1997).
from their plant targets. We return to the effects of these The grass was able to survive and grow in both soil
fungal symbionts of bark beetles later in this chapter. types when planted experimentally, although the plants
Chronic herbivory—herbivory that occurs over grew taller on the sandy soils. However, rodents cached
long time periods—can have large effects on plant seeds only in the sandy areas, and this seed dispersal
demography. Pinus edulis (pinyon pine) trees subjected behavior appears to be an important factor in deter-
to chronic herbivory have reduced rates of growth, an mining the distribution of this plant.
altered shape, and produce male cones almost exclu- The distribution of the shrub Haplopappus squarrosus
sively (Whitham and Mopper 1985). In a study of Euca- (yellow squirrel cover, Asteraceae) in California is strong-
lyptus in subalpine habitats in the Snowy Mountains of ly affected by granivory. Louda (1982) showed that H.
Australia, Patrice Morrow and Valmore LaMarche (1978) squarrosus was most abundant in an inland transitional
found that trees treated with insecticide experienced zone between coastal and interior climates, but produced
large increases in growth. Given the growth history of more seeds closer to the ocean (Figure 11.3). This pattern
trees on that site, they concluded that growth was sup- appears to be mainly a result of increased insect
pressed by chronic herbivory. These results imply that granivory on the plants growing closest to the ocean.
chronic herbivory can greatly reduce lifetime fitness. By Donald Strong and his colleagues have unraveled a
spraying insecticides on experimental plants several complex set of interactions that apparently underlie the
times a year, Waloff and Richards (1977) showed that population dynamics of a shrub, Lupinus arboreus
chronic herbivory reduced the seed yield of the British (lupine, Fabaceae), along the central coast of California
shrub Sarothamnus scoparius (broom, Fabaceae) by about (Strong et al. 1995, 1996). Large patches of this woody
75% over 10 years. Similarly, in a study of the semel- perennial periodically die off and are eventually regen-
parous plant Cirsium canescens (Platte thistle, Asteraceae; erated from seed, so that the population fluctuates over
see Figure 9.5), Svata Louda and Martha Potvin (1995) time. The plants are killed primarily by subterranean
excluded inflorescence-feeding herbivores from exper- caterpillars of the ghost moth, Hepialus californicus, which
(A) (B)
100
Flower heads destroyed
1.00
80
by insects (%)
Relative frequency
.75
Observed 60
.50 Expected
40
.25
0
Maritime Coastal Transition Interior Maritime Coastal Transition Interior
Figure 11.3
(A) Observed (bars) versus expected (dashed line) relative each 167 meters in length. The expected frequency is derived
frequency of Haplopappus squarrosus (yellow squirrel cover, by assuming that relative population size should be propor-
Asteraceae) over a climatic gradient in San Diego County, tional to relative seed production. (B) Percentage of flower
California. The observed distribution is based on the pres- heads destroyed by insect granivores in each climate zone;
ence or absence of plants along 15,250 roadside segments, error bars are 95% confidence intervals. (After Louda 1982.)
bore into the roots. However, an insect-killing nematode, and erratic variation among years in the size of the seed
Heterorhabditis hepialus, with its symbiotic bacterium, crop produced, and this variation is generally synchro-
Photorhabdus luminescens, is a highly effective predator nized among most of the plants in a population. This
on the ghost moth caterpillars. Consequently, lupine- phenomenon, called masting, has been widely
dominated areas that are heavily colonized by the nem- explained as an adaptation to granivory. Various ecolo-
atode are protected from attack by the caterpillar, while gists have hypothesized that during masting years, the
those areas without the nematode experience massive large numbers of seeds overwhelm the capacity of grani-
periodic die-offs. Few other studies have reported con- vores to eat them all, allowing at least some seeds to sur-
trol of plant populations by underground herbivores, vive. This explanation has been questioned recently by
not to mention such trophic complexity, but then, few other ecologists, who argue that other factors may be
studies have looked for either. more important in selection for masting. They argue that
masting can more easily be explained as an adaptation
Granivory to wind pollination, which is increasingly efficient at
Granivory can also have important consequences for high pollen densities (Smith et al. 1990; Kelly et al. 2001).
plant populations. In some populations, granivores con- In a large survey of studies on variation in seed pro-
sume a large fraction of the seeds. Tachigali versicolor duction, Carlos Herrera and associates (Herrera et al.
(alazan, Fabaceae) is a monocarpic tree from Central 1998) argued that the distinction between masting
America with unusually large (500–600 mg) seeds. In a species and non-masting species is false because there is
detailed study of the seeds and seedlings produced by a continuum of variability. Most iteroparous woody
two large adult trees, Kaoru Kitajima and Carol plants have variable seed production. Even when phy-
Augspurger (1989) found that 51% to 83% of seeds died logenetic relationships are controlled for, there is slight-
prior to germination, depending on the tree and the dis- ly more variability in wind-pollinated species than in
tance of the seed from the tree. The important granivores animal-pollinated species. However, these researchers
were bruchid beetles (eating 13% to 38% of seeds) and also found more year-to-year variation in species dis-
vertebrates (eating 0% to 59% of seeds). Seedlings had persed by granivores or abiotic mechanisms (such as
somewhat lower mortality rates in their first two months wind or water) than in those dispersed by frugivores.
(24% to 47%, again depending on the tree and distance It seems clear that more studies are needed to assess the
from the tree). Seedling mortality was primarily due to importance of pollination and granivore satiation in
herbivory (6% to 17% of seedlings) and disease (3% to among-year variability in seed production.
25% of seedlings).
The chemical properties of seeds can deter or Biological Control
enhance granivory. Some seeds contain compounds that Biological control is the deliberate use by humans of
strongly deter granivores (we discuss such defensive herbivores or pathogens to control populations of unde-
compounds in more detail below). Seeds of Erythrina sirable plant (or other) species. Many such biological
(coral bean trees, Fabaceae), for example, contain com- control agents are introduced from other continents, and
pounds that are strongly neurotoxic in vertebrates; as a successful instances of biological control of plant pests
result, they are rarely eaten. On the other hand, a study offer examples of herbivores controlling plant popula-
of squirrel granivory on oaks provided some surprises. tions. One well-known example is the introduction of
Quercus rubra (red oak, Fagaceae) has high tannin con- the moth Cactoblastis cactorum to Australia to control the
centrations in its acorns, while Q. alba (white oak) does introduced and invasive prickly pear cacti Opuntia iner-
not. One might expect that gray squirrels (Sciurus caro- mis and O. stricta (Cactaceae). The cacti had spread to
linensis) would prefer the acorns of white oaks. How- cover vast areas in Australia, rendering them useless for
ever, Peter Smallwood and Michael Steele (Smallwood sheep grazing (Figure 11.4). The moth was introduced
et al. 2001; Steele et al. 2001) showed that the interaction in 1925 from Argentina, where its caterpillars were found
is more complicated. Squirrels prefer to eat white oak to be specialist herbivores on prickly pears. By 1935, the
acorns, but prefer to cache red oak acorns because the prickly pear populations had been decimated, and they
tannins protect them from fungi and bacteria, making have remained at low levels since that time.
them less perishable. Moreover, the squirrels often Unfortunately, the success of Cactoblastis in Australia
remove the embryos of the white oak acorns before is colored by recent problems with this species in North
caching them. Thus, more red oak than white oak America. C. cactorum was introduced to the Caribbean
seedlings emerge from squirrel caches, and the squirrels in 1957 to control Opuntia species that had spread due
are more effective as dispersers of red oaks—not because to overgrazing. Subsequently, the moths invaded Flori-
the squirrels avoid the tannin-laden red oak acorns, but da and are now spreading rapidly in the southeastern
because they actively prefer them! United States, where they are affecting several endan-
Many plants—especially woody plants—show large gered Opuntia species (Johnson and Stiling 1998). There
216 Chapter 11
(A) (B)
Figure 11.4
(A) A dense stand of Opuntia in Queensland, Australia, prior
to the release of Cactoblastis. (B) The same stand three years they prefer L. salicaria to related native North American
later. (Photographs by A. P. Dodd and the Commonwealth species. However, these insects feed on native species
Prickly Pear Board.)
when there is little purple loosestrife available, creating a
potential for negative effects on the native species.
An agent introduced to control the severely inva-
is much concern that this invasion will spread to the sive exotic plant Carduus nutans (musk thistle, Aster-
southwestern United States and Mexico. Opuntia species aceae) and its relatives in the midwestern United States
are ecologically important in both countries, and are also has proved to have serious negative effects on rare
economically important in Mexico. native thistles. Rhinocyllus conicus (flowerhead weevil)
A recent example of a similar phenomenon has been was introduced to control Carduus species after stud-
the introduction of several specialist insects to control ies indicated that this weevil prefers Carduus to native
purple loosestrife (Lythrum salicaria), an introduced, thistles of the genera Cirsium, Silybum, and Onopordum.
highly invasive species that now dominates vast areas However, despite this preference for the invasive musk
of wetlands in eastern and central North America (see thistle, studies by Louda and her associates (Louda et
Figure 14.3). Bernd Blossey and his colleagues found that al. 1997) showed that the weevil is also consuming, and
purple loosestrife was attacked by many different insects is negatively affecting, three native thistle species—Cir-
in central Europe, where it is native. These researchers sium canescens, C. centaureae, and C. undulatum. Ob-
imported populations of a root-mining weevil (Hylobius served levels of R. conicus infestation were as great or
transversovittatus) and two leaf-feeding chrysomelid bee- greater on the native thistles than on the exotic this-
tles (Galerucella calmariensis and G. pusilla) and released tles, and were also greater than the levels of native
them in the United States and Canada in 1992 and 1993 insect infestation on the native thistles. As a result,
(Blossey et al. 2001). These and other host-specific insects infested plants set many fewer seeds. In the case of C.
released later appear to be successfully eliminating the canescens, which is already uncommon and restricted in
dense populations of purple loosestrife without attack- habitat, this new herbivore may present a danger to
ing other species, allowing the recolonization of native population persistence, as reduction in seed production
wetland plants in areas where purple loosestrife had is expected to reduce the population growth rate of this
maintained monospecific stands. Purple loosestrife may species (Louda et al. 1997). The best protection for the
be limited by herbivores in its native habitat. In Europe, native C. canescens may be the maintenance of the inva-
it never reaches more than 5% cover and remains a sive Carduus populations in the same habitat to attract
minor component of wetland vegetation, in dramatic the weevils away from the native thistle—thus the orig-
contrast to its spread in North America when it was inal target species may act as sort of a biological control
introduced without specialist herbivores. agent for the weevils!
An important question, however, is the extent to These studies, together with other recent empirical
which these and other biological control agents may affect studies (Henneman and Memmott 2001), have caused
nontarget species. The control agents released for purple many biologists to begin reconsidering the safety of bio-
loosestrife were characterized as “specialists” because logical control in general (Simberloff and Stiling 1996).
While biological control has proved strikingly success- others. The effects of specialists, on the other hand,
ful in cases such as that of Opuntia in Australia, its strong depend on the roles that their preferred food species
point—the fact that the control agents disperse and play in the community. It is likely that a specialist on a
reproduce on their own—is also what makes it poten- potentially dominant species will have a very different
tially risky. effect on the community than a specialist on a less com-
mon species. Thus, herbivory can either increase or
decrease the diversity of the plant community. The out-
Effects of Herbivory come depends not only on patterns of consumption, but
at the Community Level also on interactions between herbivory, plant competi-
tion, and abiotic factors such as soil moisture, nutrients,
Consequences of Herbivore Behavior and light levels. The effects of herbivory can vary over
Herbivores show preferences and exhibit behavior with spatial scales and over the course of time.
respect to what they eat and how and when they eat it. Nancy Huntly (1987) studied the foraging behav-
Their behavior can have profound consequences for ior of pikas and the consequences of their foraging for
species richness and abundance. One important way in the plant community. The pika, a small, territorial rel-
which herbivore behavior can affect plant community ative of rabbits and a strong contender for the world’s
composition is the extent to which animals behave as cutest mammal, lives on high-altitude, rocky talus
generalists versus specialists with regard to what they slopes in western North America. Pikas forage out-
eat. At one extreme, a pure generalist herbivore eats ward from their dens. They are generalists, but prefer
plants in the same proportions at which they are pres- certain plant foods to others. Pikas do not hibernate;
ent in the community. At the other extreme, a pure spe- instead, they collect large haypiles during the summer
cialist herbivore eats only plants (or even only specific for winter use (Figure 11.5). Huntly experimentally
plant parts) belonging to a single species or to a small excluded pikas from vegetation plots at different dis-
group of closely related species. Extreme specialists are tances from their dens into the surrounding meadows.
the most desirable choices for biological control agents She found that the foraging animals had large effects
to minimize effects on nontarget species. on plant community composition close to their dens,
Most herbivores are somewhere between these two where they spent the most time, with decreasing
extremes, preferring certain foods to others, but able effects farther away.
to eat a variety of plants. Thus, it is misleading to strict-
ly contrast specialists and generalists. However, it is Introduced and Domesticated Herbivores
often believed that generalist herbivores tend to pro- One of the classic stories in the ecology of herbivory is
mote or maintain species diversity because they keep that of rabbits, chalk grassland vegetation, and a rabbit
faster-growing, dominant species from outcompeting disease, myxomatosis. Chalk grasslands are highly
Figure 11.5
Pikas (Ochotona princeps), relatives of
rabbits and hares, are common in
high-altitude habitats in western
North America. Pikas feed on almost
all the plant species that grow around
their rockpiles. Because they forage
from a central location—a den—their
effect on the plant community is
strong, but becomes weaker at greater
distances from the den. As shown
here, pikas collect haypiles during the
summer for winter use. (Photograph
courtesy of C. Ray.)
218 Chapter 11
diverse plant communities found on limestone-derived competitive plant species declined in abundance, but did
soils in southern England. These grasslands alternate not disappear altogether. In addition, plants not charac-
with woodlands, and they have been used for centuries teristically found in the chalk grasslands invaded some
to graze sheep and, beginning in the twentieth century, of the ungrazed areas. However, the predicted large-
cattle. Rabbits were brought to England in medieval scale colonization by woody species did not occur, per-
times as a source of food and for sport hunting, where haps because the exclosures were too small or too far
they, well, bred like rabbits. They became widespread, away from sources of tree seeds.
attaining very high numbers after about 1850, undoubt- This experiment was repeated unintentionally on a
edly due to direct and indirect human influences, includ- larger scale some 30 years later, when a viral rabbit dis-
ing predator reduction. Rabbit densities varied greatly ease, myxomatosis, was accidentally introduced into
from one spot to another. Great Britain and nearly wiped out rabbit populations
What effects have the rabbits had on the vegetation in the chalk grasslands and elsewhere. Immediately fol-
of the chalk grasslands? The father of modern experi- lowing the decimation of the rabbit population in 1954,
mental plant ecology, Arthur many rare plant species were found in the chalk grass-
Tansley, observed that where rab- lands that had not been seen in years (Thomas 1960,
bit populations were high, the 1963). These species had been selectively grazed by rab-
turf was typically chewed down bits, never becoming large enough to flower and be
to a height of 1 to 2 centimeters, noticed, or had been grazed as soon as they grew past
while that of sheep-grazed grass- the seedling stage. Various rare orchids and other showy
lands was typically 5 to 10 cen- species, such as Helianthemum chamaecistus (frostweed,
timeters in height. He hypothe- Cistaceae) and Primula veris (primrose, Primulaceae),
sized that if sheep grazing and appeared and flowered in abundance. Some of these
rabbit feeding were prevented, species had been common a hundred years before, prior
the grasslands would revert to to the great increase in rabbits. Other species decreased
Sir Arthur George Tansley forest. Tansley experimentally as the rabbits disappeared, either because they were out-
fenced plots of vegetation to competed or because they had been favored by the nitro-
exclude rabbits and sheep (Tansley and Adamson 1925). gen from the rabbits’ urine. Tall grasses began to become
At first, there was a great increase in the growth of plants more prominent, and woody species began to invade.
inside the exclosures, and many plants flowered abun- When the rabbit populations recovered from the epi-
dantly inside the exclosures that never succeeded in demic of myxomatosis in the early 1960s, the vegetation
reproducing outside the fences. After some time, how- largely reverted to its previous state.
ever, perennial grasses, the preferred food of the rabbits, Striking pictures of the effects of grazing are offered
grew up and shaded shorter dicot species. Total biomass by fence lines where one side is heavily grazed and the
and average vegetation height increased substantially. other ungrazed or lightly grazed (Figure 11.6). Like
The palatable grasses became more dominant, while less pikas, cattle are selective generalists, eating many
Figure 11.6
A fence line in northern Arizona.
The area at the left has been grazed
by cattle; the area at the right is not
grazed. (Photograph by S. Scheiner.)
species, preferring some, and avoiding others. They usu-

ally avoid woody and spiny species, as well as those
with toxic or noxious defensive chemicals. Cattle can
poison themselves, for example, by grazing on species
such as Digitalis (foxglove, Scrophulariaceae; see Figure
9.1) and Astragalus (locoweed, Fabaceae). Cattle can have
dramatic effects on community composition as a result
of grazing, particularly when their densities are high
or when grazing occurs at particularly sensitive times of
the year for plant recovery and regeneration (e.g., when
grass seeds are ripening). Over time, particularly with
heavy grazing, preferred plants such as palatable and
nutritious grasses decline in abundance and are replaced Figure 11.7
by less edible species, drastically changing the compo- The degu (Octodon degus) is a small rodent that is a principal
sition and appearance of the plant community. Heavy herbivore in Chilean deserts. (Photograph courtesy of B.
overgrazing leads to bare patches of ground, weed inva- Lang.)
sion, and severe erosion, especially on slopes. The very
landscape can be changed, with deep ravines and gul-
lies replacing rolling, grass-covered slopes, as a result of (Brown and Munger 1985; Brown et al. 1997). Over time,
long-term damage to the plants that once held the soil removing either rodents or ants caused substantial
in place. These problems tend to be more severe in arid changes in plant species composition. Where rodents
environments, but can occur even in mesic habitats. were removed, large-seeded species increased and
Other grazing animals, including sheep and goats, can small-seeded species decreased. Where ants were
cause similar effects. Problems caused by overgrazing removed, the opposite results were found (Guo and
are widespread in western North America, but also Brown 1996).
occur in Africa, the Mediterranean region, and Australia, Valerie Brown and her colleagues have carried out
among other places. Overgrazing by domestic animals many innovative field experiments to investigate the
has contributed to turning vast areas of grassland into effects of herbivores on plant communities. In one large
shrublands or deserts. study, they used insecticides to kill either aboveground
or belowground insects in an early successional field in
Effects of Native Herbivores Great Britain (Brown and Gange 1992). The purpose of
Seed-eating and vegetation-eating small mammals have the experiment was to see how the effects of root-feed-
been shown to affect plant community structure in a ing insects on the plant community might differ from
number of arid environments. In the semiarid shrubland the effects of foliage feeders. They found that both
of northern Chile, Javier Gutiérrez and his colleagues aboveground and belowground herbivory by insects
used fences and netting to selectively exclude small had major (but different) effects on the timing and direc-
mammals (principally the herbivorous degu, Octodon tion of succession. In this field, the aboveground insects
degus; Figure 11.7) and predatory birds (particularly were largely sap-feeding Hemiptera, which preferred
owls) from large plots (Gutiérrez et al. 1997). The exclu- perennial grasses. Their herbivory suppressed perenni-
sion of degus resulted in an increase in the cover of al grass colonization, slowing succession. The removal
shrubs and perennial grasses, an increase in the diver- of the insects led to a luxuriant growth of the grasses,
sity of perennial species, and a decrease in annual plant which then shaded and replaced the lower-growing
cover. The researchers found some indirect effects of herbaceous dicots, leading to a steep decline in species
predator exclusion on the vegetation (presumably by richness. Underground, chewing insects in the
allowing increases in herbivores), and they also docu- Coleoptera and Diptera fed primarily on the roots of the
mented strong effects of weather as well as interactions dicots. Reducing belowground insect numbers led to the
between weather and herbivore effects. persistence of annual dicots and a great increase in col-
Interestingly, while the arid and semiarid vegetation onization by perennial dicots, and consequently a great
of Chile and Argentina is primarily home to herbivorous increase in species richness. The ordinary presence of
and insectivorous species, with few granivores, seed-eat- these root-feeding insects thus speeds up succession by
ing rodents dominate North American deserts. James reducing the dicots and increasing the colonization of
Brown and his colleagues have conducted a series of the field by grasses.
very long-term experiments to exclude different small Herbivory is not the only way in which plant-eating
mammals (particularly heteromyid rodents) and ants animals affect plant communities. Other kinds of her-
from plots in the Chihuahuan Desert of eastern Arizona bivore behavior can also change the environment and
220 Chapter 11
have strong effects on plant communities. Mammalian 7 Ungrazed

herbivores, in particular, create gaps and patches when Grazed
Species richness or diversity

6
making burrows and trails, and trample vegetation
around available water sources. Domesticated cattle 5
herds can cause severe damage to vegetation along
4
streams and watercourses by repeated trampling. Ele-
phants consume and trample enormous amounts of 3
plant material. Tree canopy cover in Serengeti National
Park in Kenya, for example, was reduced by about 50% 2
by elephants (Pellew 1983). 1
Not all such effects of herbivores are negative. In
tallgrass prairies in the midwestern United States, native 0
Species Species Species Species
bison (see Figure 1.2C) create depressions where they diversity, richness, diversity, richness,
roll in the dust. In the spring these areas contain small (H′), 20-cm 20-cm (H′), 4-m 4-m
temporary pools of water, and in the summer they are scale scale scale scale
inhabited by annual species that otherwise would be Figure 11.8
excluded by perennial grasses (Collins and Uno 1983). Plant species richness and plant species diversity, as meas-
Large herbivores can also affect plant distributions ured by the Shannon-Weiner index (see Chapter 12), are
and species richness through strong effects on soil nutri- greater in grazed than in ungrazed grasslands in Yellow-
stone National Park. The effect was greater for comparisons
ents caused by their urination and defecation. In a study of 20 × 20-centimeter quadrats than for 4 × 4-meter quadrats,
at Yellowstone National Park, David Augustine and implying that the effects were occurring at a very small spa-
Douglas Frank (2001) compared soil characteristics and tial scale. This scale effect was contrary to expectations
community characteristics between ungrazed grasslands because urination and defecation by large mammals is
and grasslands grazed by large herbivores—elk, bison, thought to lead to a very patchy distribution of soil nitrogen
and nitrogen-mineralization rates. Species richness is the
and pronghorn. Species richness and diversity were mean number of species per plot on the 20-centimeter scale,
greater in the grazed grasslands, particularly at very and the mean number of species per plot divided by 10 on
small scales (Figure 11.8). the 4-meter scale. Error bars are ± 1 standard error. (After
Even smaller animals can have dramatic effects, par- Augustine and Frank 2001.)
ticularly when they reach very high numbers. Lesser
snow geese breed on the coastal marshes of the tundra
of Hudson Bay and James Bay in Canada during the
summer, migrating in autumn to the Gulf Coast of there (Abraham and Jefferies 1997). The greater snow
Louisiana and Texas. Over the past 30 years their popu- geese that summer in the eastern Canadian Arctic and
lation has grown tremendously, with midwinter counts migrate to the southeastern United States are poised for
climbing from about 0.8 million in 1969 to about 3 mil- a similar population explosion, with similar effects pre-
lion in 1994. Their winter and summer habitat use has dicted for the vegetation in their breeding habitat.
expanded greatly, both in area and in the variety of habi-
tats they occupy (Abraham and Jefferies 1997). Much of Generality
this population growth appears to be caused by the How general and important are the effects of herbivores
response of the geese to changes in land use by humans. on plant communities? This question has been a con-
Expansion of farmland and the use of crops as food troversial one. In a classic and much-cited conceptual
sources by the geese, both during migration and on their paper, Hairston et al. (1960) suggested that herbivores
wintering grounds, have decreased overwinter adult are kept at low densities by their own predators, limit-
mortality. Changes in predation rates may also be a facing their effects on plants. Widely referred to as the HSS
tor. In summer, ever larger flocks of geese return each hypotheses (after the authors’ initials), this point of
year to the tundra in excellent nutritional condition, capa- view has been controversial ever since it was proposed.
ble of rearing large numbers of young. Snow geese are William Murdoch (1966) argued that the logic of HSS is
grazers; however, they are also capable of killing plants circular. Pointing to its argument that because they do
by pulling them up by their roots or rhizomes. Large not eat all the available plant material, herbivores must
flocks of the birds can have dramatic effects on the tun- not be food-limited, Murdoch observed that one could
dra landscape, not only changing plant species compo- apply the same logic to the next trophic level: because
sition but also leaving entire areas bare of plants. It has they do not eat all the available herbivores, carnivores
been estimated that over a third of the tundra coastal must not be food-limited—exactly the opposite of what
marsh ecosystem has been severely overgrazed, threat- Hairston et al. argued was the case. Murdoch also
ening some 200 other species of water birds that summer argued that the HSS hypotheses were poorly defined
and therefore untestable. Paul Ehrlich and L. Charles generally better defended. There are a great many dif-
Birch (1967) argued that the HSS hypotheses hinge ferent types of plant defenses, just as there are a great
upon the belief that because species persist, they must many different ways that plants can be attacked.
be regulated at or near an equilibrium—a “balance of
nature.” Ehrlich and Birch argued that, to the contrary, Plant Physical Defenses
most populations experience so much random envi- Physical and mechanical defenses include obvious struc-
ronmental variation that their populations undergo tures such as thorns and spines, which probably serve
large fluctuations. best to discourage mammalian browsers and birds, but
Much research and debate has been stimulated by do little to deter insects. Single-celled plant hairs—tri-
these ideas in the ensuing decades; to do justice to this chomes—serve many functions, including protection
body of work would require another book! Confining (Figure 11.9A). Insects are deterred by leaf hairiness and
ourselves to the question with which we began, how- can be impaled by some trichomes. Other trichomes
ever—whether herbivory is important for plant com- secrete noxious compounds that can deter vertebrates or
munities—the answer seems clear. In a recent meta- sticky substances that impede insects. Species of Urtica
analysis of a large number of experimental studies on (stinging nettles, Urticaceae) have brittle, elongate tri-
herbivory, David Bigger and Michelle Marvier (1998) chomes that break off when brushed, leaving a pointed
concluded that on average, herbivory causes a substan- fragment that pierces the skin and injects a painfully irri-
tial reduction in plant biomass in natural plant com- tating fluid (Figure 11.9B). Trichomes of this kind have
munities. Contrary to the assumptions of many ecolo-
gists, invertebrates such as insects have a much greater
effect than vertebrates.
Plant Defenses
Because plants cannot move, most plants have to “sit
there and take it” from herbivores, resulting in natural
selection for plants that are tougher, less palatable, and
(A) (B)
Figure 11.9
(A) A haze of trichomes covers the stem and leaves of this Solidago plant (gold-
enrod, Asteraceae). (Photograph © Jerome Wexler/Visuals Unlimited.) (B) Scan-
ning electron micrograph of trichomes on the leaf of Urtica dioica (stinging net-
tle, Urticaceae). The long trichome in the center is a stinging hair; when its tip is
broken off by some action (such as the touch of a human hand or foot), it easily
penetrates the skin and injects the neurotransmitter acetylcholine along with
histamines that produce allergic reactions. The shorter trichomes do not have
these protective chemicals. (Photograph © Andrew Syred/Science Photo
Library.)
222 Chapter 11
evolved independently in four plant families: Urticaceae, glass, or large amounts of sand). The silica is contained
Euphorbiaceae, Loasaceae, and Boraginaceae. in specialized epidermal cells as well as in other plant
Other physical defenses include materials that make parts. A few other taxa also use silica as a deterrent to
entry into the plant difficult, such as thick bark on trunks herbivory, including horsetails (Equisetaceae) and palms
or roots or the tough coats that protect seeds and some (Arecaceae).
fruits, such as nuts. Various cells and tissues that make Another characteristic that has been suggested to act
up the plant body probably also serve a defensive func- as a defense against herbivores is the nutritional quali-
tion, such as the cap of thick sclerenchyma (a plant tis- ty of leaf tissue. Eating leaves with a low nitrogen and
sue with lignified cell walls) that surrounds the vascu- water content results in poor growth and survival for
lar bundles carrying food and water in young stems, and herbivores. Leaves with a higher nitrogen and water
the waxy cuticle on the surface of leaves, which both content are usually preferred by herbivores over those
reduces water loss and protects against fungal attack. with a lower content (all else being equal). This fact
Leaf toughness presents an important mechanical poses an evolutionary dilemma for plants: The metabolic
barrier to chewing insects, mammalian browsers, and enzymes responsible for growth and photosynthesis
other herbivores, as well as interfering with tissue contain nitrogen. Reducing the concentration of nitro-
digestibility. Toughness is primarily a result of the con- gen-containing compounds in the leaves to discourage
tent, type, and placement of sclerenchyma fibers (elon- herbivory could result in reduced photosynthesis and
gated cells), although short sclereids (sclerenchyma cells growth.
with thick, lignified, much-pitted walls) may be respon- Nitrogen is almost always limiting to plants, but it
sible for the hardness of some structures. Toughness may is even more limiting to herbivores. Animals need much
also depend on other thick-walled cell types, such as more nitrogen to function than plants do, and they main-
xylem and collenchyma (supporting tissue made of liv- tain much higher concentrations of nitrogen in their bod-
ing elongated cells with walls thickened irregularly). ies. Plants growing in soils with low available nitrogen
Newly formed, expanding leaves are considerably less may have higher concentrations of carbon relative to
tough than older ones because the structural tissues nitrogen in their tissues, reducing their nutritive value.
involved in making leaves tough interfere with the leaf’s As global CO2 concentrations increase because of human
ability to grow to full mature size. For this reason, new emissions (see Chapter 22), many herbivore populations
leaves are vulnerable to attack, and must depend on are expected to decline because of the relative decline in
alternative means of protection, such as chemical defens- carbon to nitrogen ratios, and thus the nutritional value,
es. of many plants. Recent studies by Peter Stiling and his
Grasses are unpalatable to most generalist herbi- associates (Stiling et al. 1999) found that several species
vores because they sequester large amounts of silica, of leaf miners consumed more tissue from oak leaves
which makes them difficult to chew and difficult to grown under enriched CO2 conditions, but nevertheless
digest (picture chomping down on ground-up bits of had greater rates of mortality (Figure 11.10).
(A) (B)
Ambient CO2
Elevated CO2
2 15
Figure 11.10
An experiment comparing the per-
Leaf area consumed (cm2)
formance of leaf miners of the genera

Stigmella, Cameraria, and Stilbosis on
Percent mortality
Quercus myrtifolia (myrtle oak, 10

Fagaceae) grown under ambient ver- Quercus myrtilfolia
sus enriched CO2 conditions. (A) The 1
insects removed more tissue from the
plants grown under enriched CO2 5
conditions (error bars are ± 1 standard
error.) (B) Nevertheless, the insects
feeding on those plants had higher
mortality rates from a number of 0 0
causes (error bars are ± 1 standard Stigmella Cameraria Stilbosis Died in mine Preyed on Parasitized
deviation). (After Stiling et al. 1999.) Leaf miner species Fate
Plant Secondary Chemistry contain subunits of more than one type. Additional
Plants can marshal an arsenal of chemical weapons in defensive compounds include toxic proteins and amino
response to herbivores. Chemical ecologists distinguish acids, protease inhibitors, and cyanogenic compounds.
between primary metabolites and secondary chemicals Phenolics include a large variety of chemicals consisting
(or secondary metabolites). Primary metabolites (such of an aromatic ring with an attached hydroxyl group,
as sugars, amino acids, and DNA) are compounds nec- OH (Figure 11.11). Probably the most important defen-
essary for the basic functioning of the plant. Secondary sive phenolic compounds in angiosperms and gym-
chemicals constitute a broad group of compounds that nosperms are the tannins, which reduce the digestibili-
serve a wide variety of purposes, including defense and ty of plant tissues. They are packaged in cell vacuoles
attraction of pollinators, rather than primary functions and are present in high concentrations in the leaves of
such as photosynthesis. Secondary compounds are gen- many woody plants, such as those in the Fabaceae,
erally found only in particular species or groups of Fagaceae, Myrtaceae, and Polygonaceae. Another impor-
species, and often only in specific organs or tissues. The tant group of phenolic compounds are the lignins,
distinction between primary and secondary compounds which impregnate secondary (woody) cell walls, giving
is somewhat arbitrary; some primary metabolites, for them structural strength as well as providing a barrier
example, are also used in defense. The terms are to some to attack by herbivores and pathogens. Other pheno-
extent a holdover from an earlier era when secondary lics include poisonous saponins as well as the flavonoids
chemicals were believed to be waste products by plant and anthocyanins, pigments that also give flowers and
physiologists and biochemists unaware of their ecolog- fruits their colors.
ical functions.
The three major categories of defensive secondary
chemicals are phenolics, alkaloids, and terpenes. These
categories are not exclusive as some large molecules can OH
OH
HO O
OH
OH
OH
HO O
HO OH
OH
HO HO COOH OH
HO O
HO
HO OH
Tannins OH
HOOC OH OH OH
Hydrolysable Condensed
tannin tannin
(Gallotannin) (Catechin)
R1
Flavone R2 Flavan—3—ol
OH
HO O Flavonoids HO O
OH
OH
OH O OH
Catechin
R1 = R2 = OH: Luteolin Basic skeleton
Flavan
B
O
A 3
R1
R1 Anthocyanidin
Flavonol OH
R2 B
HO O—
HO O A R2
R3 Isoflavone
Figure 11.11 OH
Defensive phenolic compounds HO O OH
include tannins and flavonoids. OH O
R1 = R2 = H: Pelargonidin
Examples of specific ones are R1 = R3 = H; R2 = OH: Kaempferol
R1 = OH; R2 = H: Cyanidin
given below the structures. R1 = R2 = OH; R3 = H: Quercetin HO O OH R1 = R2 = OH: Delphinidin
(After Larcher 1995.) Genisteine
224 Chapter 11
The alkaloids are another broad group of com-

pounds and include many that are used as pharmaceu-
ticals (Figure 11.12). Some 10,000 alkaloids have been iso-
lated and their structures analyzed. Alkaloids are
relatively small molecules that contain nitrogen. They (A)
5/14-Photos at
have a bitter taste, and many are toxic to herbivores. HO Sinauer
Alkaloids are generally highly specific to the plant
NCH3
species or group of species in which they are found. O Photo to come
They are often effective in small quantities—as in the
cases of cocaine, nicotine, and caffeine—although in HO
some cases particular plants or plant parts produce high
Morphine
concentrations of alkaloids.
Terpenes are found in all plants, and an individual
may contain many different terpenes (Figure 11.13). Ter-
penes play a wide variety of functional roles in plants. (B)
They are composed of multiple units of the hydrocarbon O
isoprene (C5H8), and they may be large or small,
depending on how many isoprene units they contain. O C
Photo to come
The compound isoprene is emitted in sunlight by the C OCH3
leaves of some plant species (such as Eucalyptus), some- N
O
times in large quantities, and may serve to protect them CH3
from heat damage. The latex (milky sap) found in mem-
bers of the spurge (Euphorbiaceae) and dogbane (Apoc- Cocaine
ynaceae) families contains toxic terpenoid defenses. The
oils that are responsible for the characteristic flavors and
scents of the mints (Lamiaceae) are terpenes; they deter (C)
herbivores and reduce the growth of bacteria and fungi.
O CH3
Cardiac glycosides can cause heart damage in verte- H3C
brates and are poisonous to many insects; these terpenes N
N
are produced by many unrelated plant species, includ- Photo to come
O N N
ing species in the Scrophulariaceae (such as Digitalis, fox-
gloves) and the Apocynaceae (such as Asclepias, milk- CH3
weeds, and Apocynum, dogbanes). Phytoecdysones are Caffeine
terpenes that mimic insect molting hormones and alter
the development of insect larvae; they are produced by
a number of taxa, including ferns, cycads, and some
angiosperms.
Plants in the Brassicaceae (mustard family) contain (D)
characteristic secondary compounds, chiefly glucosino-
lates, which are effective repellents of most generalist N Photo to come
herbivores. These mustard-oil precursors (responsible N
CH3
for the “bite” of the many cultivated species in this fam-
ily) almost completely deter mammals and nonadapted Nicotine
insects. Glucosinolates are not as effective against insect
herbivores that have become specialized on this family,
however, and only high concentrations can offer some
protection. Louda and her colleagues have found that
Figure 11.12
glucosinolates not only can affect the interaction Examples of various alkaloids used by humans and their
between plants and their herbivores, but also can indi- sources. (A) Opium comes from the milky sap of seed pods
rectly control the distribution and abundance of the of Papaver somniferum (opium poppy, Papaveraceae); mor-
plants (Louda and Rodman 1996). Cardamine cordifolia phine is derived from opium. (B) Cocaine is found in the
(bittercress, Brassicaceae) is native to the Rocky Moun- leaves of Erythroxylum coca (coca, Erythroxylaceae). (C) Caf-
feine is found in coffee “beans,” the fruits of Coffea (coffee,
tains of North America, where it grows only in shaded, Rubiaceae), as well as many other plants. (D) Nicotine comes
moist forest edges. Bittercress may be restricted from from the leaves of Nicotiana tabacum (tobacco, Solanaceae)
sunny sites by greater vulnerability to chronic herbivory and other species in this genus. (After Raven et al. 1999.)
H3C CH3 OH Figure 11.13

O Triterpenes play important roles in plant
O HO defenses against herbivores. (A) Hedera helix
H3C COOH (English ivy, Araliaceae) contains the
HC saponin hederagenin in its leaves and fruits.
CH3
CH3 OH (B) Asclepias (milkweeds, Apocynaceae) con-
HO
HOH2C CH3 RO
tains the cardiac glycoside calotropagenin.
(C) Solanum demissum (potato, Solanaceae)
(A) Saponin (Hederagenin) (B) Cardenolide (Calotropagenin) produces the steroid alkaloid demissin in its
leaves. (D) A number of taxa produce
ecdysones, compounds that mimic insect
R OH hormones and interfere with insect meta-
morphosis. (After Larcher 1995.)
OH
N CH3
HO
Gluc H OH
Gluc–Gal–O HO
H
Xyl
(C) Steroid alkaloid (Demissin) (D) Sterol (Ecdysone)
at those sites. In a series of experiments, plants grown in trait (Dirzo and Harper 1982a,b). In the presence of snails,
sunny sites experienced greater water stress, which plants that had the ability to produce cyanogenic glyco-
reduced their glucosinolate concentrations. Conse- sides had higher survival rates than those that did not.
quently, those plants had more insect herbivores and suf- In the absence of snails, however, those plants had
fered greater damage. Thus, herbivory may be control- reduced growth and reproduction compared with
ling the plant’s distribution by causing differential acyanogenic (non-cyanide-producing) plants. As pre-
damage among microhabitats. dicted by evolutionary theory, areas with high snail den-
sities had a higher proportion of cyanogenic individuals,
Constitutive versus Induced Defenses while those with fewer herbivores had a preponderance
Constitutive defenses are those that are present in a of acyanogenic individuals.
plant regardless of herbivore damage. They may be pres- In a study of Lepidium virginianum (pepper-grass,
ent throughout the life of a plant or may change as an Brassicaceae), Anurag Agrawal (2000) showed that
individual grows and matures. Induced responses are induced defenses were effective against a generalist her-
elicited by an attack by herbivores. If these responses bivore, but not against an herbivore that specialized on
serve to protect the plant (whether or not they harm that plant (Figure 11.14). Whether provided with a choice
the herbivore), they are called induced defenses; if they of food items or not, generalist caterpillars did substan-
have a negative effect on the attacking herbivores tially more damage to plants whose defenses had not
(whether or not they defend the plant), they are termed been induced. However, specialist caterpillars produced
induced resistances (Karban and Baldwin 1997). Both the same amount of damage in induced and noninduced
physical and biochemical defenses can be induced. plants.
There has been a great deal of discussion regarding the The production of sap by conifers in response to
relative benefits of constitutive versus induced defens- attack by bark beetles is an example of an induced
es (Karban and Baldwin 1997; Lerdau and Gershenzon response. In addition to the physical defense of the
1997). Historically, most research on plant responses to sticky, suffocating pitch, this response has a biochemical
herbivory has focused on constitutive defenses, but function. Coniferous sap contains substantial amounts
beginning in the 1980s, interest in induced defenses of various terpenes (including monoterpenes and
greatly increased. sesquiterpenes) as well as phenolics and other com-
A classic example of an induced defense is the pro- pounds that are particularly toxic to the symbiotic fungi
duction of cyanogenic glycosides by Trifolium repens that attack with their beetle partners. Raffa and col-
(white clover, Fabaceae) when the leaves are damaged; leagues (1987, 1991) found that wounding a tree mechan-
this response can also be induced by frost damage. ically caused the production of small amounts of
Cyanogenic glycosides are an effective deterrent against monoterpenes, while trees attacked with living fungi
snails, which are voracious herbivores in regions with responded with massive amounts of monoterpenes.
mild, wet winters. The ability to produce these defensive Trees responded most strongly to those particular fungi
compounds is controlled by two genes, and populations associated with the bark beetle species that ordinarily
of white clover in Great Britain are polymorphic for the attack them.
226 Chapter 11
(A) four to ten times, reaching concentrations that can kill

Generalist Control an herbivore after a single meal.
Induced Nicotine is synthesized in the roots of tobacco plants
15
in response to leaf damage and transported via the
Percentage of leaf
xylem to the leaves. The signal indicating that the leaves

consumed
10 have been damaged, cuing the roots to produce nicotine,

is apparently a hormone transported from leaves to roots
5 through the phloem. Greater leaf damage induces more
nicotine production. Because of the long-distance nature
0
of this response and the time needed for biosynthesis, it
Choice No choice takes from ten hours to several days for nicotine to
increase to its maximum levels. What might be the
(B) advantages of belowground synthesis of nicotine, given
Specialist the vulnerability of the plant during this prolonged
wait? If nicotine production occurred largely in the
40
shoots, herbivores consuming the leaves might severe-
Percentage of leaf
ly compromise the plant’s ability to defend itself from

consumed
30
further attack. The site of the synthesis of this defen-
20 sive compound is thus protected belowground from
leaf-eating herbivores. Even if large amounts of the
10 aboveground parts of the plant are damaged or
0
destroyed, the plant can continue to defend itself.
Choice No choice Some plants increase physical defenses in response
to herbivory. Grass plants can increase their silica con-
Figure 11.14
Induced defenses in Lepidium virginianum (pepper-grass, centrations. Opuntia (Cactaceae) can secrete additional
Brassicaceae) were effective against caterpillars of (A) gener- mucilage and fiber, effectively walling off insect herbi-
alist herbivores (noctuid moths), but not against caterpillars vores.
(B) of the specialist butterfly Pieris rapae (Pieridae). The
defenses were induced by allowing a fixed number of larvae
to feed on plants several days before the experimental trials. Evolutionary Consequences
The amount of damage was the same whether or not the of Plant-Herbivore Interactions
caterpillars had a choice of induced or noninduced plants.
Error bars are ± 1 standard error. (After Agrawal 2000.) You may be wondering, after reading about the arsenal
of defenses possessed by plants, why all herbivores do
not simply starve to death. There are several reasons
that they do not. First, fortunately for the rest of the bio-
Many different classes of chemicals have been found logical world, plants are not uniformly well defended.
to act as induced defenses. Ian Baldwin and his associ- Species, populations, individuals, life stages, and plant
ates (Baldwin 1988, 1991; Karban and Baldwin 1997) parts vary in their defenses, and the less well defend-
have done much to increase our understanding of the ed are often sought out and depended on for food. Sec-
process of induction. In some cases chemical defenses ond, some herbivores have evolved various means of
are induced by highly specific cues, such as caterpillar overcoming plant defenses, leading to what has been
saliva. In other cases any mechanical damage is suffi- termed a “coevolutionary arms race” between con-
cient. The toxic alkaloid nicotine, for example, is pro- sumed plants and their would-be consumers. The
duced in response to mechanical damage to the leaves behavior, biochemistry, and morphology of herbivores
of Nicotiana (tobacco, Solanaceae). Nicotine mimics the may contribute to their ability to overcome plant defens-
neurotransmitter acetylcholine, blocking the receptor for es. Some insects, for example, are capable of carefully
that neurotransmitter in both insects and vertebrates. At avoiding the ducts and canals that carry toxic latex in
low doses, nicotine stimulates the nervous system, but certain plants. Even more surprisingly, some
at higher doses it is a depressant, ultimately capable of chrysomelid beetles neatly incise these canals
causing paralysis and death. Smokers become addict- “upstream” before beginning to feed, preventing the
ed to the excitatory effects of low doses of this potent flow of the toxic compounds to the leaf they are con-
nerve toxin; caterpillars or grazing mammals may be suming. Giraffes, with their tough mouths and phe-
paralyzed and killed by the same toxin. Small amounts nomenally long tongues, freely browse Acacia
of nicotine are present in undamaged tobacco plants. (Fabaceae) in the African savanna, undeterred by the
After leaf damage occurs, nicotine levels can increase plants’ abundant and sharp thorns (Figure 11.15). In the
Figure 11.15
Giraffes browse Acacia (Fabaceae) trees, despite the long, sharp thorns that protect
these species. Browsed trees can give the impression of having been “shaped” into a
topiary. (Photograph courtesy of E. Orians.)
northeastern United States, white-tailed deer avidly to depend on them as a primary food source. In most
consume vines of Smilax glauca (catbrier, Liliaceae), at mammals, the teeth cease to grow in adults, but those
least when the stems are young, despite their ferocious that are adapted to eating grass have teeth that grow
hooked spines; it is not clear how the deer avoid being continuously. As the silica in the grass erodes their teeth,
“clawed.” Clearly, one cannot always depend on broad new growth replaces the worn material. Grasses are
generalizations in making predictions about patterns of highly adapted to being grazed as well. Grasses, like all
herbivory. One must understand the biology of the plants, grow from regions of actively dividing, undif-
species involved. ferentiated cells called meristems (see Chapter 8). The
A number of leaf-eating herbivores have symbioses meristems of grasses are generally located at ground
with microbial species that can digest cellulose. Almost level, out of the way of a passing grazer’s teeth. Thus,
no animals can digest cellulose, a major component of the plants are able to rapidly regrow the tissue lost to
plant tissue, but various bacterial species can. Ruminants grazers, at least under ideal conditions. Heavy grazing,
(including deer, cattle, and antelopes) depend on bac- however, may result in destruction of these meristems,
terial fermentation of their food, which occurs in a spe- or in plants being torn up by the roots. Low resource
cialized stomach compartment called the rumen; the ani- levels or repeated heavy grazing can also compromise
mals receive nourishment from the fermented material the ability of the plants to produce new leaf tissues.
and from digesting some of the bacteria. Cellulose- Many specialist herbivores are able to detoxify or
digesting bacteria are also symbiotic with other groups sequester secondary compounds of their preferred host
of mammals and, most notably, with termites. plants that would deter or kill other herbivores. It has
Grasses and grazers provide examples of mutual long been known, for example, that the caterpillars of
adaptations between plant and herbivore. While many monarch butterflies (Danaus) are specialists on milk-
generalist leaf eaters are effectively discouraged from weeds (Asclepias, Apocynaceae) (Figure 11.16). The milky
eating grasses, a large group of mammals has evolved latex of milkweeds contains cardenolides, which are
228 Chapter 11
investigators, allows the plants to kill small monarch lar-

vae, which are sensitive to the cardenolides. But the sub-
sequent decrease reduces the supply of cardenolides to
the bigger, older caterpillars, which have the ability to
sequester the chemical. Other specialist herbivorous
insects also have evolved the ability to sequester plant
secondary compounds and use them for their own pro-
tection, such as cucumber beetles that use cucurbitacin
from cucumbers to deter invertebrate predators.
Plant defenses can have complex effects. Producing
a defensive compound, for example, can affect not only
herbivores, but also pollinators. Pointing out that her-
bivory and pollination can interact in complex ways,
Sharon Strauss (1997) suggested a conceptual model of
the effect of herbivory on plant fitness (Figure 11.17). Her
ideas were tested in a study of the annual plant Castilleja
indivisa (Indian paintbrush, Scrophulariaceae). Castilleja
Figure 11.16 (like many species in its family) is a root hemiparasite (its
A monarch caterpillar on Asclepias syriaca (common milk- roots facultatively tap into the roots of other plants and
weed, Apocynaceae). (Photograph by S. Scheiner.) divert mineral resources and secondary compounds).
Lynn Adler and her associates (Adler et al. 2001) grew
some Castilleja plants on a strain of Lupinus albus
(Fabaceae) that produces a high concentration of alkaloids
highly toxic to most insects and act as cardiac poisons and some on a strain that produces a low concentration.
in vertebrates. While most herbivores avoid these plants, Half of the Castilleja plants from each treatment were then
monarch butterfly larvae are capable of sequestering the sprayed with an insecticide, which both deterred herbi-
toxins, accumulating them in their own bodies to protect vores and limited pollinators. That is, the experiment con-
themselves from the birds that are their own predators. sisted of high and low alkaloid levels combined with or
However, there is more to this classic evolutionary without pesticides. As a result, the researchers were able
ecology story. Malcom and Zalucki (1996) found that to estimate the fitness effects of herbivory (seed produc-
plants of Asclepias syriaca (common milkweed, Apocy- tion), both directly and indirectly through its effects on
naceae) rapidly spike up their levels of cardenolides after pollination. The alkaloids had a positive effect on plant
they are wounded, but drop the levels immediately fitness because they reduced herbivory on buds, leaving
afterward. This specialized type of induction, reason the more flowers and increasing pollinator visitation rates.
Direct effects on Indirect effects Fitness

floral traits (through pollinators) components
Petal size
Flower no. Attractiveness to

pollinators No. seeds
Ovule no. per plant Female
No. flowers fitness
Ovule size
visited per Seed size
Herbivory Pollen production plant
Pollen size/quality
Nectar production Time spent per No. seeds Male

flower sired fitness
Nectar sugar conc.
Floral chemistry/volatiles Figure 11.17

A conceptual model of the effect herbivory exerts on fitness through its effect
on floral characteristics. Single-headed arrows indicate a causal relationship;
double-headed arrows indicate a correlation. (After Strauss 1997.)
Pathogens important pathogens not only of plants, but also of

For reasons that are not fully apparent, the ecology of humans, other mammals, and arthropods. Spiroplas-
plant-herbivore interactions has received much greater mas (helical, motile mollicutes) can be transmitted
attention than the ecology of plant-pathogen interactions. between insects and plants (including major crops) and
Yet plants are clearly subject to diseases, and they defend can cause disease in both.
themselves against disease organisms. Pathogens can Some of the chemical defenses outlined above for
affect plants at all of the ecological levels, and in most herbivory also, or even primarily, serve as defenses
of the ways, discussed previously for herbivores. In one against fungal, bacterial, and viral pathogens. Interest-
of the best-known examples, chestnut blight, caused by ingly, it appears that the biochemical cascade that sig-
the fungus Cryphonectria parasitica, virtually eliminated nals herbivore damage, mediated chiefly by a group of
Castanea dentata (American chestnut, Fagaceae; Figure plant hormones called jasmonates, is quite distinct from
11.18), a forest tree species that once dominated much the series of biochemical events signaling pathogen inva-
of the eastern deciduous forests of North America. By sion, which is generally mediated by salicylic acid.
doing so, the fungus drastically changed the composi-
tion of these communities. Today a number of oak species Responses of Individual Plants to Infection
(Quercus, Fagaceae) dominate these forests. Some diseases rapidly kill large numbers of plants. Phy-
As with herbivory, seedlings are particularly vul- tophthora infestans, the fungal pest that causes potato
nerable to pathogens. A fungal disease called damping blight, can kill a field of potatoes within a few days
off, for example, kills many plants at the seedling stage. under the right weather conditions. Most pathogens
Damping off is an important cause of seedling mortali- have less severe effects, however. The bacterium Xan-
ty in tropical forests (Augspurger 1984). One explana- thomonas axonopodis pathovar citri causes citrus canker;
tion for the high local diversity in tropical rainforests is infected Citrus (Rutaceae) trees drop their fruits and
the effect of such fungal pathogens. If the fungal leaves prematurely. In other species, fungi and bacteria
pathogens are species-specific, Connell (1978) argued, lead to spotting of leaves and fruits, making fruit crops
they are likely to be found at high concentrations around less marketable.
maternal parents, leading to strong selection for dis- Phytoalexins are secondary compounds that act as
persal away from the parent plant. specific defenses against pathogens. Phytoalexins are
Fungi, bacteria, and viruses are the most common produced at the site of an infection to kill microbes. There
infectious agents in plants. An important class of plant are a host of different classes of chemical compounds
pathogens just beginning to be understood is the molli- known to function as phytoalexins, including many of
cutes, tiny wall-less bacteria that includes the somewhat the chemical defenses mentioned above. Through mech-
more familiar mycoplasmas. The mollicutes include anisms that are not yet fully understood, plants can also
Figure 11.18
A Castanea dentata (American chestnut,
Fagaceae) tree in full flower in northern
lower Michigan, far outside its normal
range. This individual was undoubtedly
planted, and only its great distance from
natural populations has allowed it to
avoid infection by the chestnut blight
fungus and reach full size.
(Photograph by S. Scheiner.)
230 Chapter 11
acquire both localized and even systemic resistance to severely reduced the competitive ability of merram grass
pathogens. The systemic resistance observed in plants relative to sand fescue. In the absence of the pathogens,
is analogous to the acquired immunity humans devel- merram grass was not competitively inferior to sand fes-
op to some diseases, although it is very different bio- cue. Thus, the interaction between competition and the
chemically, physiologically, and evolutionarily. effects of pathogens may be responsible for the decline
A physical defense against infection is phloem plug- of the merram grass on stabilized dunes and its replace-
ging, in which the phloem clogs up in response to dam- ment by sand fescue.
age, preventing the spread of the infectious agent Herbivory can also interact with the effects of
through the vascular system of a plant. Similarly, in pathogens. Bowers and Sacchi (1991) studied the effects
some plants, cells in and around the infected tissue die, of herbivore exclusion in an early successional old field
sealing off the area so that the infection will not spread. in Virginia. A dominant plant species, Trifolium pratense
This tissue death reduces the amount of photosynthet- (purple clover, Fabaceae), increased in response to her-
ic surface area and can therefore reduce the plant’s bivore exclusion. However, these high-density, ungrazed
growth rate. clover populations became severely infected with the
fungus Uromyces trifolii, which killed many of the plants.
More Complex Interactions In contrast, grazed areas had much less disease. As a
Insects spread many plant pathogens. Insects visiting result, in the year following the infection, the density of
flowers of Silene (campion, Caryophyllaceae), for exam- clover plants was much higher in the presence of graz-
ple, spread Ustilago violacea (anther smut fungus). The ing than in its absence. Other factors affecting the
infection prevents the anthers from producing pollen; progress of disease in natural populations include plant
instead, they produce fungal spores (Thrall and genotype, microhabitat, and the diversity of the plant
Antonovics 1995). Similarly, infection by the rust fungus community, with the presence of neighbors of other
Uromyces pisi prevents Euphorbia cyparissias (Euphor- species reducing infection rates (Morrison 1996).
biaceae) from flowering. Instead, the plant forms pseu-
doflowers that shed fungal gametes, which are dispersed Summary
by insects (Pfunder and Roy 2000). Some fungal infec-
tions are thus analogous to sexually transmitted diseases Plants are the foundation of nearly all terrestrial food
in animals. webs and, thus, are eaten by a very wide range of ani-
In some cases, however, there is a closer symbiotic mals and other heterotrophic organisms. In addition,
association between an insect and a pathogen. Bark bee- plants can be infected by an array of fungi, bacteria, and
tles are associated with fungal species that harm their viruses, many with pathogenic effects. Herbivory—the
target trees more than the insects do. The fungi may be consumption of plant material—can either kill a plant or
present on the insects’ surfaces or may be carried in spe- reduce its growth and reproduction, depending on what
cialized structures. A tree is infected with the fungi when part of the plant is consumed and the life stage at which
the beetles tunnel into the cambium. The spreading fun- the herbivory occurs. Consumption can sometimes be
gal mycelia cause “blue-stain disease,” which destroys beneficial, as when an animal consumes a fruit and dis-
the tree’s vascular tissue, killing it. The insects and fungi perses rather than destroys the seeds inside. Whether
appear to benefit mutually from their concerted attack other types of herbivory are ever favorable to plants
on the plant. Bark beetles are responsible for the spread through the mechanism of overcompensation is disput-
of Dutch elm disease in North America, caused by the ed among ecologists, but empirical evidence suggests
introduced fungus Ophiostoma ulmi, which has killed mil- that herbivory is generally harmful to individual plants.
lions of Ulmus americana (American elm, Ulmaceae). Both Despite herbivory, the world looks mostly green.
beetles that are native to North America and introduced We do not yet know to what extent herbivory controls
beetles from Europe are involved. The fungus kills the plant population dynamics and community structure
trees by blocking water flow in the xylem and by pro- in general. It may be that most of the time herbivores
duction of a toxin. are kept at low levels by their predators and so have lit-
In a study of the interaction between soil-dwelling tle overall effect on plants. Alternatively, chronic her-
pathogens (fungi and nematodes) and plant succession bivory may change relative abundances and population
in the Netherlands, VanderPutten and Peters (1997) distributions. The effects of herbivory on plant popula-
found that the pathogens facilitated species replacement tions are obvious in some instances, such as during
within the plant community. On coastal sand dunes, insect outbreaks. Changes in herbivore populations can
Ammophila arenaria (merram grass, Poaceae) is replaced also result in changes in community structure, with for-
as the dunes become stabilized by another grass, Festu- merly rare species becoming common. Generalist her-
ca rubra ssp. arenaria (sand fescue). In a series of experi- bivores are thought to promote species diversity.
ments, the researchers found that the soil pathogens Amounts of herbivory on leaves vary among biomes,
with greater herbivory in tropical than in temperate are covered with defensive hairs, or stems covered with
forests. This observation suggests that the importance spines. Other plants have leaves that are low in nutri-
of herbivory may differ among communities. Animals ents or contain substances that make it difficult for her-
can also affect plant community structure indirectly bivores to digest them. To defend themselves against
by manipulating the physical environment through herbivores and pathogens, plants produce toxic sub-
activities such as digging or trampling. Pathogens can stances such as phenolics, alkaloids, terpenes, toxic pro-
have effects similar to those of herbivores on plant pop- teins, protease inhibitors, and cyanogenic compounds.
ulation and community structure. As plants have evolved these defenses, herbivores have
Plants defend themselves in a variety of ways. Some evolved countermeasures, although the extent to which
plants have tough leaves that resist attack, leaves that one-to-one coevolution occurs is unclear.
Additional Readings
Classic References Blossey, B., L. C. Skinner and J. Taylor. 2001. Impact and management of
purple loosestrife (Lythrum salicaria) in North America. Biodiversity and
Tansley, A. G. and R. S. Adamson. 1925. Studies of the vegetation of the
Conservation 10: 1787–1807.
English chalk. J. Ecol. 13: 77–223.
Gutierrez, J. R., P. L. Meserve, S. Herrera, L. C. Contreras and F. M. Jaksic.
Hairston, N. G., F. E. Smith and L. B. Slobodkin. 1960. Community struc-
1997. Effects of small mammals and vertebrate predators on vegetation
ture, population control, and competition. Am. Nat. 44: 21–425.
in the Chilean semiarid zone. Oecologia 109: 398–406.
Rhodes, D. and R. Cates. 1976. Toward a general theory of plant antiher-
bivore chemistry. Rec. Adv. Phytochem. 10: 68–213.
Burdon, J. J. 1987. Diseases and Plant Population Biology. Cambridge Uni-
Contemporary Research versity Press, Cambridge.
Agrawal, A. A. 2000. Benefits and costs of induced plant defense for Lep- Coley, P. D. and J. A. Barone. 1996. Herbivory and plant defenses in trop-
idium virginianum (Brassicaceae). Ecology 81: 804–1813. ical forests. Annu. Rev. Ecol. Syst. 27: 305–335.
Augustine, D. J. and D. A. Frank. 2001. Effects of migratory grazers on Karban, R. and I. T. Baldwin. 1997. Induced Responses to Herbivory. Uni-
spatial heterogeneity of soil nitrogen properties in a grassland ecosystem. versity of Chicago Press, Chicago, IL.
Ecology 82: 3149–3162.
Rosenthal, G. A. and M. R. Berenbaum, eds. 1992. Herbivores: Their Inter-
actions with Secondary Plant Metabolites, 2nd ed. Academic Press, San
Diego.
232 Chapter 11
12
C H A P T E R Community Properties
A
s we hike up a mountain in the central Rockies, we might start in a for-
est dominated by ponderosa pine (Pinus ponderosa), then soon find our-
selves among lodgepole pine (Pinus contorta), then move on to spruce-
fir forest (Picea engelmannii and Abies lasiocarpa) and end up in alpine tundra.
Each area contains a very different collection of species, yet some species are
found in many different areas, and boundaries can be hard to discern. In other
environments, such as grasslands with patches of serpentine soils (soils derived
from serpentine rock that are nutrient-poor and sometimes high in toxic ele-
ments; see Box 15A), differences in species composition between communities
can be abrupt and dramatic. Humans also create boundaries that define com-
munities. A vacant lot is a community, as is a park in the middle of an urban
landscape.
In previous chapters, we looked at how species interact with one another
and their environments; here, we explore the community-scale patterns cre-
ated by those interactions. What determines the boundaries of a community?
Are communities real entities with their own properties, or are they just hap-
penstance collections of individuals and populations? This chapter investigates
what a community is, and discusses methods for describing communities. The
following chapters will cover some additional processes that create commu-
nity patterns: disturbance, succession, dominance, and species invasions.
What Is a Community?
A community is a group of populations that coexist in space and time and inter-
act with one another directly or indirectly. By “interact” we mean that they
affect one another’s population dynamics. This definition of “community”
includes all plants, animals, fungi, bacteria, and other organisms living in an
area. It seems simple enough, but ecologists often use terminology inconsis-
tently, and unfortunately, traditional usage in plant ecology sometimes differs
from that in animal ecology. For example, we speak of “plant communities”
even though plants are only a subset of the community—we are ignoring the
decomposers, herbivores, diseases, pollinators, and many other organisms.
Fauth et al. (1996) have discussed some ways through this terminological
tangle (Box 12A).
Two closely related older terms that were formerly in wider use in plant
ecology have been incorporated recently into some modern vegetation classi-
234 Chapter 12
BOX 12A
Communities, Taxa, Guilds, and Functional Groups
A very confusing array of terms has Geography
Resources
developed to describe communities.
Sometimes the same term is used in dif-
ferent ways by different ecologists,
while in other cases different terms are Local
given to the same concept. Here we Communities guilds
Guilds
describe a recently proposed scheme to
define these terms. Ensembles
This scheme defines groups of
species based on three criteria: geogra-
phy, phylogeny, and resource use, as Assemblages
shown in the accompanying figure.
Geography in this scheme defines com-
munities, sets of organisms living in
the same place at the same time. Phy- Taxa
logeny is the pattern of relationships A scheme for defining sets of species into
among species (or higher taxa) based communities, guilds and ensembles based
on evolutionary ancestry. Phylogeny on the combination of geographical loca-
tion, common ancestry, and shared
defines taxa, sets of organisms that Phylogeny resource use. (After Fauth et al. 1996.)
share a common ancestor. Resource use
defines guilds, sets of organisms that
use biotic or abiotic resources in a sim- ing in the same place. The intersection ed in the plant community. The plant
ilar way. The term “guild” is taken from of geography and resource use defines community could therefore be consid-
animal ecology, but has been used by local guilds. The trees in a forest in ered an assemblage, as it includes
plant ecologists as well. Ontario are an example of a local guild: species that use different resources. The
Plant ecologists often use the term they include distantly related species true community, of course, includes all
functional group to describe a concept such as sugar maple (Acer saccharum), species (e.g., animals, fungi, bacteria),
related to the guild. Functional groups an angiosperm, and eastern hemlock not just plants. Thus, the traditionally
can be defined in a variety of ways, (Tsuga canadensis), a conifer. The inter- defined plant community is actually a
depending on the application, but these section of all three sets defines ensem- subset of the full community and has
definitions are all based on a set of traits bles. The grass species living together properties of ensembles, local guilds,
that identify functionally similar in a prairie are an ensemble. The annu- and assemblages.
species. The traits used to identify func- al Asteraceae in Australia’s Great Sandy
tional groups can be chosen informal- Desert are another ensemble.
ly or based on formal mathematical Plant communities as traditionally
algorithms. Ecologists have used the defined are made up of a combination
concept of functional groups in a vari- of ensembles, local guilds, and assem-
ety of contexts, including studies of the blages. Typically, terrestrial plant com-
relationship between productivity and munities are defined as all the vascu-
diversity (see Chapter 14) and attempts lar plants living in a given space. Most
to reduce the number of types of plants species in this group are primary pro-
that must be taken into account in glob- ducers with similar resource require-
al climate modeling. Extensive recent ments. So, for example, all of the grass
reviews of the concept of plant func- species in a forest understory are one
tional groups and ecological applica- ensemble of that community. The com-
tions of this concept are provided by bination of all understory forbs and
Lavorel and Cramer (1999) and Wood- graminoids would constitute a local
ward and Cramer (1996). guild, as these species would include
Intersections of the sets described by more distantly related species of mono- Monotropa uniflora (Indian pipes, Ericaceae) is
an example of a flowering plant that is a sapro-
the terms above define more narrow cots, dicots, and possibly ferns. Some phyte rather than a primary producer, obtain-
distinctions. The intersection of geog- flowering plants are not primary pro- ing its nourishment not by photosynthesis but
raphy and phylogeny defines assem- ducers, but parasites or saprophytes; from dead or decaying organic matter. (Photo-
blages, groups of related organisms liv- however, these species are also includ- graph by S. Scheiner.)
Community Properties 235
fication schemes (see Table 16.3). An association was a middle ground between these viewpoints, and to a
defined as a particular community type, found in many large extent have moved beyond both of them.
places and with a certain physiognomy and species com- The Clementsian view was the majority view among
position; modern usage is not much different (e.g., Table English-speaking plant ecologists during the first half of
16.3 refers to the juniper-sage woodland association). the twentieth century. Clements saw plant communities
The term formation was originally used to denote a as highly organized entities made up of mutually inter-
regional climax community; modern usage is generally dependent species. In his view (Clements 1916), com-
more specific, referring to a physiognomic subtype. munities are superorganisms—the analogue of indi-
In practice, the boundaries of plant communities are vidual organisms—that are born, develop, grow, and
usually defined operationally, based on changes in the eventually die. Succession, in this
abundance of the dominant, or most common, species. view, is analogous to the process of de-
Sampling is then confined within those boundaries. A velopment and growth; its trajectory
stand is a local area, treated as a unit for the purpose of and end point are highly predictable
describing vegetation. Typically, a number of stands are (Clements 1937). Two of the hallmarks
used to sample the presence and abundances of species of the superorganism concept were the
as well as associated environmental variables. Based on presence of very tight linkages among
data from a number of stands, the community can be species within communities and coop-
characterized. eration among the species in a com-
Only in special cases (e.g., islands, ponds, forest pre- munity for the benefit of the function
serves surrounded by suburban development, vacant of the entire community.
lots) are community boundaries defined easily. Even Even at the height of Clements’s Frederic and
Edith Clements
then, the movement of organisms and the transport of influence, many ecologists held more
matter by wind and water make their boundaries fuzzy. moderate views. This version of Clementsian ecology
Ecologists, therefore, are often of two minds when deal- asserted only that species interactions such as competi-
ing with communities. On the one hand, we recognize tion, mutualism, and predation are important in deter-
that their boundaries are fuzzy; on the other hand, we mining community structure. Clements himself ac-
often need to define discrete entities for convenience of knowledged the effects of abiotic factors such as site
analysis. Typically, plant ecologists define a community history and soils in determining community composi-
based on the relative uniformity of the vegetation and tion. He focused on the idealized nature of communities,
use their knowledge of species biology to decide when and saw them as spatially distinct, with one superor-
they are moving from one community to another. ganism complex giving way to another with a very dif-
Ecologists based in different countries, educated in ferent collection of species. His major focus was on the
different historical traditions, tend to view communities nature and development of the community as a super-
in somewhat different ways. In particular, ecologists in organism, however, rather than on the boundaries
continental Europe were historically influenced by the between communities. Some ecologists of the day accept-
floristic-sociological approach, most extensively devel- ed a more moderate version of this view, which admits
oped by Josias Braun-Blanquet (see Chapter 16). This that communities are not entirely discrete, but still
approach emphasizes the discreteness of communities. divides them into nonarbitrary groups with recogniza-
In contrast, most ecologists in English-speaking coun- ble boundaries.
tries have been more strongly influenced by the history In striking contrast to Clements,
described in the following paragraphs; as a result, they Gleason believed that communities
tend to think of communities as blending continuously are the result of interactions between
into one another. These distinct ways of thinking are individual species and the environ-
becoming less prominent as a result of increased travel ment (biotic and abiotic factors) in
and communication among ecologists worldwide. combination with chance historical
events. Each species has its own envi-
The History of a Controversy ronmental tolerances and thus re-
Within plant ecology, there has been, and still is, a range sponds in its own way to environ-
of views on the nature of communities. The extremes are mental conditions (Gleason 1917,
sometimes labeled the Gleasonian and Clementsian 1926). The implication of this belief Henry A. Gleason
views, named after Henry A. Gleason and Frederic was that along an environmental gra-
Clements, their first major proponents in the English- dient, different species would have their boundaries at dif-
speaking world. These two extreme viewpoints differ in ferent places. Not only were communities not tightly
the importance they ascribe to biotic versus abiotic fac- linked superorganisms, but defining the collection of
tors and predictable versus random processes in shap- species living together in a particular place as a com-
ing community structure. Today, most ecologists hold munity was an arbitrary human construct.
236 Chapter 12
According to the Gleasonian view, within the range Salix nigra
(Black willow) Acer saccharum
of environmental conditions a species can tolerate, 150 (Sugar maple) Quercus rubra
chance events determine whether a species is actually (Red oak)
found in a given location. At the local scale, chance dic- Quercus alba
Importance value
tates whether a seed happens to get to a particular spot. Ulmus americana (White oak)
On larger scales, the chance events of history play a 100 (American elm)
strong role. For example, species in the family Cactaceae
(the cacti) are found in the desert communities of the Ulmus rubra
Americas because the family originated in this region, (Slippery elm)
50
while deserts elsewhere have no cacti (except where they
have been recently introduced by humans). Further-
more, the mix of species changes from place to place as
one moves across the landscape. The Gleasonian view-
0 Wet Wet-mesic Mesic Dry-mesic Dry
point posits gradual changes in community composition
Environmental gradient
as opposed to abrupt boundaries between communities
unless there are abrupt and large environmental bound- Figure 12.1
aries. A more moderate viewpoint was that some iden- Change in the importance of various tree species along a
tifiable community types exist, but that these tend to moisture gradient in Wisconsin. Importance was measured
as the sum of the relative cover, relative density, and relative
blend into other community types. frequency of a species in a community. (After Curtis 1959.)
Clementsian ecologists were not receptive to Glea-
son’s ideas. While Gleason’s work was well known, it
failed to influence many plant ecologists until after
Clements’s death in 1945. Why Clements’s views held
sway for so long is not fully understood. Clements was matic turnover from one community type to another as
said to have an extremely strong personality that could altitude increases. Whittaker realized that if he could
dominate scientific meetings. Given the very small num- demonstrate that even along such a gradient, species
ber of plant ecologists active during the first half of the turnover was gradual, this would provide very power-
twentieth century, perhaps this alone was sufficient. ful evidence in support of Gleason’s ideas and in con-
Gleason, finding little interest in his ideas, abandoned tradiction to Clements’s superorganism model. Whit-
his work in plant ecology by 1930 and spent the rest of taker did just that. He showed that forest communities
his career as a taxonomist. along an elevational gradient in the Great Smoky Moun-
Not until the 1950s did the separate but almost tains of Tennessee changed gradually in species com-
simultaneous work of John T. Curtis and Robert H. position without abrupt boundaries (Figure 12.2). He
Whittaker convince many ecologists that Gleason’s then repeated this work in other areas, including the
views were largely correct. Curtis and his students Siskiyou Mountains of Oregon and the Santa Catalina
mapped the vegetation of Wisconsin and looked at how Mountains in southern Arizona.
species’ optima and ranges were distributed along envi- Another important line of evidence that strongly
ronmental gradients (Curtis 1959). The Clementsian the- affected many ecologists’ views of communities was a
ory predicts that species optima and ranges should show series of studies conducted since the 1970s on the distri-
distinct groupings, whereas the Gleasonian theory pre- butions of plant species during and after the most recent
dicts independence of optima and ranges. The re- glaciation (see Chapter 21). Many of these studies looked
searchers found just such independence; each species at fossil pollen from lake bottoms. Some of the earliest
had a different set of environmental tolerances (Figure and most influential of this research was carried out by
12.1). A key innovation that con- Margaret Davis. She showed that many species that co-
tributed to this study was the devel- occur today did not always do so during glacial periods;
opment of ordination, a set of tech- rather, species were distributed among communities in
niques for describing patterns in the past in very different combinations from those that
complex vegetation; we discuss ordi- are found today (Davis 1981). For example, Pinus strobus
nation in detail in Chapter 16. (eastern white pine, Pinaceae), Tsuga canadensis (eastern
Whittaker (1956) also demon- hemlock, Pinaceae), Castanea dentata (American chestnut,
strated that Gleason was right about Fagaceae), and Acer spp. (maples, Sapindaceae; maple
the nature of boundaries between pollen cannot be identified to species) are often found
communities. One of the most strik- together in the eastern deciduous forest of North Amer-
ing patterns one encounters going up ica today. However, these forest tree species were not
Robert H. Whittaker an elevational gradient is the dra- always associated in the past. White pine and eastern
40 Tsuga canadensis
Liriodendron tulipifera (Canadian hemlock) Halesia monticola
30 (Mountain silverbell)
(Tulip poplar)
20
10
0
40 Acer spicatum
Acer saccharum (Mountain maple,
30 Tilia heterophylla (Sugar maple) moosewood)
(White basswood)
20
Percent of stand
10
0
40 Fraxinus americana
Betula allegheniensis
30 Carpinus caroliniana (Yellow birch) (White ash)
(American hornbeam) Aesculus octandra
20 (Yellow buckeye)
10
0
80
Fagus grandifolia
60 (American beech)
40
20
0
500 700 900 1100 1300 1500 1700
Elevation (m)
Figure 12.2
Changes in plant species frequencies along an altitudinal gradient in the Great
Smoky Mountains of Tennessee. (After Whittaker 1956.)
hemlock survived the Wisconsin glaciation (about 75,000 plowed at some time in the past. Thus, current envi-
to 12,000 years before present) in a region east of the ronmental boundaries do not always match past bound-
Appalachian Mountains, while during the same period, aries. Ecologists still disagree as to the relative impor-
chestnut and maples were found near the mouth of the tance of biotic and abiotic processes and chance events
Mississippi River (Davis 1981), more than a thousand in determining community structure.
kilometers to the southwest. Echoes of the controversy between the Clementsian
Today most plant ecologists take a middle position and Gleasonian views of community among English-
between Clements’s and Gleason’s views, and in many speaking ecologists of North America and the United
ways have diverged from both of their views. There is Kingdom continue to influence ecologists schooled in
wide agreement that species are distributed individual- those traditions. Related issues were debated among
istically, and that community composition typically European and Russian ecologists, but not to the extent
changes gradually along environmental gradients. that they were argued in American and British journals;
Abrupt changes are most likely to be found where there the ecologists of continental Europe and Russia had dif-
are abrupt changes in the environment. However, abi- ferent concerns. Heavily influenced by Braun-Blanquet,
otic boundaries and community boundaries do not they focused primarily on systems of community clas-
always match. Because of processes such as dispersal sification. We will return to their traditions in Chapter
from one habitat into another (see Chapter 17), a popu- 16. The key difference is that the European tradition was
lation may extend partway into an unfavorable envi- more concerned with describing patterns than with ana-
ronment. Abrupt changes may also reflect past events, lyzing processes, mainly sidestepping the arguments
such as the edge of a fire or a part of a forest that was described here and in the next chapter.
238 Chapter 12
A Modern Perspective on the Issues in Contention documenting patterns (as Curtis, Whittaker, and others
The primary issues surrounding the nature of plant com- did), understanding the processes responsible for creat-
munities divide roughly into those of pattern and those ing those patterns, and posing plausible explanations—
of process. Underlying these issues is theory: the effort elucidating the mechanisms causing those patterns and
to explain the patterns and processes, which includes a processes.
search for the mechanisms responsible (see Chapter 1). Whether communities can be considered to have
The issues of pattern focus on how species and com- emergent properties depends in part on the relative
munities are distributed over the landscape. Are bound- importance of biotic and abiotic processes in shaping
aries among communities abrupt or gradual? How pre- community structure, including how strongly species
dictable are the patterns? Questions of pattern are critical interact with one another within communities. (We view
because they set the stage for the rest of the debate. Once the issues in somewhat different terms today than either
patterns are identified, theories can be constructed to Clements or Gleason viewed them, as we will see in
explain them. In this chapter we examine ways of meas- Chapter 13.) Emergent properties are those that come
uring patterns within communities; among-community about through interactions, such as competition, preda-
patterns are dealt with in Chapters 16 and 17. tion, and mutualism, that occur among the populations
The issues of process focus on what processes actu- in a community (Box 12B). If these processes play a
ally function in natural communities and which of these major role in shaping communities, then communities
are most important in determining the observed pat- have at least some emergent properties. But if commu-
terns. We know of many processes that influence com- nities are mainly structured by the tolerances of indi-
munity composition, including the physiological toler- vidual species for abiotic factors (such as minimum tem-
ances of species, competition, herbivory, biogeography, peratures) in the environment, then community
historical contingency, and random factors influencing properties are largely aggregates of the individual
colonization. All can be shown to operate at some times species’ properties. It is important to note that one can
in some places; the question is their relative importance. recognize that emergent properties are important with-
Do some processes predominate in determining pat- out accepting Clements’s superorganism view, by dis-
terns? Is the relative importance of these processes carding the ideas that communities consist of species
always the same, or are some more important some of adapted for one another’s benefit and that all or most
the time, or in certain situations, or in certain types of species in a community are tightly interlinked. Among
communities? One fundamental issue is whether com- modern ecologists, for example, E. P. Odum, H. T.
munities are primarily static, exist at some sort of Odum, R. V. O’Neill, and their co-workers have strong-
dynamic equilibrium, or are always changing. We will ly emphasized the idea that communities and ecosys-
explore this issue in detail in Chapters 13 and 14. Final- tems have emergent properties, but none of them
ly, pattern and process are tied together by theories embrace the superorganism view (E. P. Odum 1971; H.
regarding the nature of communities—such as T. Odum 1983, 1988; O’Neill et al. 1986).
Clements’s superorganism theory—which seek to On the whole, ecologists have shifted their attention
explain the mechanisms responsible for the patterns and to working out the mechanisms behind the patterns
processes that are documented. found within particular communities. This shift from the
An overarching issue is the problem of scale. Recent- study of emergent properties to the study of mechanisms
ly, ecologists have paid increasing attention to the real- is emblematic of the seesawing between reductionist
ity that different patterns and processes may exist and (mechanistic) and holistic (emergent) approaches that
function at different scales. The processes that are impor- has characterized community ecology for the past cen-
tant within communities at the scale of meters may dif- tury. Ecologists in the 1980s and 1990s tended toward
fer from the processes that are important across biomes reductionist approaches. Now there is some movement
at the scale of thousands of kilometers. Throughout this back in the direction of holistic studies under the man-
book, we examine how scale affects both the patterns tle of macroecology (Brown 1995, 1999; Lawton 1999).
and the processes one finds, as well as the answers one
reaches to questions of the relative importance of dif- Are Communities Real?
ferent causes and mechanisms. As we have seen, the extent to which communities are
In Clements’s original superorganism theory, the “real” has been a contentious issue among plant ecolo-
community was an organic entity: a distinct unit with gists for much of the twentieth century. The heart of
strong emergent properties. Species within the com- the debate has been philosophical: What types of enti-
munity were tightly linked and interdependent. In con- ties are real, and what types are just mental constructs?
trast, in Gleason’s view, any community-level proper- It is clear that populations and species are real entities,
ties were simply the sum of the properties of individual but are communities real entities as well, or are they
species. Determining the truth of the matter requires merely convenient but arbitrary human inventions? In
BOX 12B
A Deeper Look at Some Definitions:
Abiotic Factors and Emergent Properties
Ecologists have distinguished between Chapter 22) is affected by living things. property of the biomass of each indi-
the effects of abiotic (nonliving) and We do not have a good substitute to vidual: We can determine community
biotic (living) factors in the environ- suggest for the term “abiotic,” so while biomass by just adding up the biomass
ment for a very long time. Typical biot- we use it here to mean “things like cli- of all of the species in the community.
ic factors include competition and pre- mate and soils,” we recognize that these But we can always measure communi-
dation; typical abiotic factors include may have major biotic components. ty biomass simply by weighing all of
soil nutrients, microclimate, weather, Another term that bears closer ex- the individuals; community biomass is
and general climatic influences. The amination is “emergent properties.” A not an emergent property.
problem with this terminology is that central issue in the argument over the In contrast, consider canopy photo-
the distinction between abiotic and nature of communities is the question synthesis: we cannot measure the pho-
biotic factors may be far less clear than of whether emergent properties exist. tosynthetic rate of an entire forest
we might think. As we emphasize in An emergent property is one that is canopy just by measuring the photo-
Chapters 4 and 15, soils are a product found at a certain level of organization synthetic capabilities of the individual
of organisms and their interactions due to properties, structures, and pro- plants. Canopy photosynthetic rates
with the environment. The nitrogen cesses that are unique to that level of depend on how individuals interact in
available to the roots of a plant, for organization. Emergent properties can several ways, including shading one
instance, depends on the actions of be contrasted with properties that are another, interfering with wind, and tak-
many different kinds of soil organisms merely aggregates of properties at a ing up CO2. Canopy photosynthetic
and the interactions of the plant with lower hierarchical level. For example, rate is an emergent property of the
those organisms. Likewise, the micro- the total biomass of all of the plants in community.
climate—and even global climate (see a community is merely an aggregate
the past, these questions often focused on the problem them if it is not. The debate as to whether communities
of boundaries—of identifying where a community are real entities or imaginary constructs is more than just
begins and ends. Many studies have shown that, except an academic exercise. For example, The Nature Con-
where there are abrupt physical discontinuities, plant servancy (TNC), based in the United States, makes deci-
communities tend not to have discrete boundaries. At sions about land acquisition and restrictions on land use
one extreme, this observation might be interpreted as based on the classification of the communities present.
implying that there are no community-level processes In collaboration with the Natural Heritage network,
worth studying. TNC has devoted substantial effort to describing and
This conclusion would be wrong; in fact, the debate classifying communities under the U.S. Natural Vegeta-
is miscast. Instead of focusing on patterns, let’s shift tion Classification (USNVC) system (see Table 16.3; May-
the focus to processes and rephrase the question by ask- bury 1999). Classifications of communities are even
ing whether community-level processes are important incorporated into law in a number of places. In south-
in structuring the living world. We have already dis- ern California, for example, land development is regu-
cussed several processes responsible for interactions lated quite differently for “coastal sage scrub” than for
among species: competition (see Chapter 10), herbivory “chaparral” communities, although by any definition
(see Chapter 11), and mutualisms (such as those dis- the two include many of the same species, and different
cussed in Chapter 4). These are all community-level scientists may have used one name or the other to cate-
processes that occur among the component parts of com- gorize the same place.
munities (e.g., populations). If such processes are sig-
nificant factors in the structuring of a particular system, Describing Communities
then we can regard that system as a community with
unique properties. Such an outlook eliminates the need Determining which processes are most important in
to worry as much about the existence of clear bound- shaping communities requires that we be able to
aries between communities. We can recognize the exis- describe and compare communities. What kinds of prop-
tence of communities if it is useful to do so, and ignore erties do plant communities have, and how can they be
240 Chapter 12
characterized? One set of community properties encom- (A)
passes the numbers of species present and their rela-
14
tive abundances: species richness and diversity. A sec-
Number of species
ond set encompasses the physical structure of the plant 12
community: its physiognomy. 10
8
Species Richness 6
One way to describe a community is by a list of the 4
species in it. Species richness is the number of species 2
on such a list. Because we often have information about
the biology and ecology of those species, the list can 0 1 1 1 1 1 2
16 8 4 2
point to many other types of information, such as the Quadrat area (m2)
numbers of trees or herbs present, or the numbers of
species in different taxonomic groups. (B)
How would one gather such a list? A simple and 1m
widely used method is to establish the boundaries of the
community and then walk through it, identifying all of
the plants and listing them. Such a survey should be
done several times during the year because some species
may be visible only during a single season. Spring
ephemerals, for example, are perennial plants common
in temperate deciduous forests whose aboveground
parts are present for only one to two months in the
spring; during the rest of the year they exist only under-
ground as dormant corms or rhizomes.
While simply finding and listing all species is useful,
this method has limitations. Most importantly, if we wish
to compare communities, we need comparable sam-
ples—otherwise we might find that two communities are
“different” simply because we sampled one more inten-
sively than the other. The area sampled can have strong Figure 12.3
effects on the number of species found. This sort of sam- (A) A hypothetical species-area curve. (B) A system of nested
pling effect is best dealt with by using plot-based meth- quadrats used to determine a species-area curve.
ods, in which a series of sample plots, or quadrats, are
marked out in a community and a list of species is col-
lected for each. The quadrats may be any shape, such as
square, rectangular, or round. They may be nested, con- curve reaches a plateau. Of course, if the area becomes
tiguous, spaced along a line, placed in a grid, or placed large enough to encompass new environments, the
randomly. These different arrangements can be used to curve begins to rise again.
ask different types of questions or to control for variation The best mathematical model for the shape of the
that occurs on different spatial scales (Krebs 1989). species-area curve has been debated over the years.
By looking at how the total number of species found Three candidates have been proposed: the exponential
increases as quadrats are combined, one can examine the curve,
effects of area on species richness. The species-area
S = Z ln(A) + C
curve (Arrhenius 1921; Gleason 1922; Archibald 1949)
describes the increase in the number of species found as the power curve,
the area sampled increases (Figure 12.3A; see also Fig-
S = CAZ
ure 17.2). This increase comes about for two reasons:
first, as more individuals are sampled, the chance of and the logistic curve,
encountering a new species increases; and second, a larg- B
S=
er area is more environmentally heterogeneous. For a C + A−Z
given community, if it has a relatively uniform envi-
ronment, the number of new species found for each where S is the number of species, A is the area, and B, C,
increment in sampling area decreases. Eventually, very and Z are constants (Figure 12.4). The three equations
few or no new species are found, and the species-area are actually members of the same mathematical family,
such that the exponential function is a special case of the An alternative method of estimating species rich-
power function—that is, they are equivalent when ness is to ask how many species are found in a sample
Z ln( A) of some number of individuals. For some functional
C= groups, such as trees, in which individuals are easy to
AZ − 1
identify, this approach can be useful. Mathematical tech-
and the power function, in turn, is a special case of the niques, called rarefaction, exist for standardizing esti-
logistic function—that is, the power and logistic func- mates among samples that include different numbers of
tions are equivalent when individuals and for estimating the number of very rare
species that were missed in the sample (Krebs 1989).
ln (B − C )/ C 2 

Z=   These techniques were developed first and are used
ln( A) most often in animal ecology, although they are being
used increasingly by plant ecologists. Because animals
Which of the three curves best fits a given data set? This often move, area-based measures are less useful to ani-
depends on the area being sampled and the scale of envi- mal ecologists. Plant ecologists, on the other hand, tend
ronmental variation. For small or intermediate-sized to emphasize area-based techniques because the clonal
areas, a logistic curve often fits bests, while for large nature of many plants can make it difficult to distinguish
areas a power curve is often the best fit. individuals (see Chapter 8). In plant ecology, dry weight
These three models make different assumptions (the weight of a sample after it has lost all its moisture
about the relationship of species number to area. Both and is at a constant weight), or biomass, is often used
the exponential and power functions result in curves instead of number of individuals.
that continue to rise indefinitely. The logistic function, How is the number of species in a community actu-
on the other hand, eventually reaches a plateau. This dis- ally determined? One can use species-area curves to
tinction is important if one wishes to answer the ques- determine the total area required for standardized sam-
tion, “How many species are in the community?” If the pling. By using equal areas in different stands within a
species-area curve rises indefinitely with increasing area, particular community, we avoid differences due to sam-
then the answer depends entirely on the area of the com- pling intensity. We may also wish to compare different
munity. However, the “area” of a community is often community types in a study using the same methods. For
arbitrary, making the question meaningless. Instead, we example, we might sample a beech-maple community
can rephrase the question as, “How many species are and an oak-hickory community using the same
contained in an area of size X?” This value is known as approach. For very different kinds of plant communities,
species density. If the species-area relationship is best however, such as grasslands and forests, we might need
described with a logistic curve, then this question has to use different sampling areas and somewhat different
meaning, because the curve has a definite plateau or techniques to best determine the characteristics of each.
asymptote, calculated as B/C. The issue of the shape of How large an area should be sampled? We want a
the species-area curve will become important when we sample that is large enough to contain most of the species
explore comparisons of diversity among communities and which will minimize differences due to random sam-
in Chapter 16. pling effects, such as where plots happen to be placed.
On the other hand, for practical purposes, we do not
want the area to be bigger than necessary. With respect
to a species-area curve, we want to sample a total area
for a given community that is at least as large as the point
Power
at which the curve begins to level off at a plateau, or, if
there is no plateau, at which the rate of increase with
Number of species
Logistic
increasing area is very small. We will return to the par-
ticular methods used for sampling vegetation shortly.
Diversity, Evenness, and Dominance

Species richness is only one aspect of diversity. Not all
Exponential species exist in equal numbers: some are rare, some are
common but not numerous, and others are very abun-
dant. Imagine two forest stands, both of which contain
ln (Area) a total of 100 individuals (or 100 kg of biomass) belong-
Figure 12.4 ing to 5 different species. In one forest stand, there are
Examples of three differently shaped species-area curves: 20 individuals of each species. In the other, one species
exponential, power, and logistic. has 60 individuals, while each of the other 4 species has
242 Chapter 12
Community A Community B There are various measures and meth-
ods for expressing species diversity (Table
12.1). Which one is most appropriate in a
given situation depends on what aspects of
a community one wishes to highlight; each
has different assumptions, advantages, and
limitations. Two commonly used measures
of species diversity (which combine the
effects of species richness and evenness)
are the Shannon-Weiner index and the
inverse Simpson’s index.
The Shannon-Weiner index is comput-
ed as
s
H′ = ∑ (pi ln pi )
i =l
where s is the number of species, pi is the

proportion of individuals found in the ith
species (pi = ni/N), ni is the number of indi-
viduals of species i in the sample, and N
Figure 12.5 is the total number of individuals sampled.
Samples from two different communities. In the sample on This index assumes that individuals were sampled from
the left all the species have equal numbers of individuals; this
a very large population and that all species are repre-
sample has greater evenness than the sample on the right.
sented in the sample.
Simpson’s index measures the chance that two indi-
viduals chosen at random from the same community
10 individuals. These two samples differ in a property belong to the same species:
called evenness (Figure 12.5). The first, in which the s
∑ pi
species are represented by the same number of indi- 2
L=
viduals (or the same amount of biomass), is more even, i =l
and thus has one of the essential elements of being more
diverse than the second. The species diversity of a com- where the summation is over all the species. Both of these
munity depends on both its richness and its evenness: indices are commonly transformed so that they range
higher species numbers, with the individuals (or bio- from a minimum of 1 to a maximum of S, the total num-
mass) more evenly distributed among them, contribute ber of species in the sample, when all species are equal-
to higher community diversity. A way to think about ly common. For the Shannon-Weiner index, this trans-
evenness as a contributor to diversity is to consider the formation is the exponential (eH′), and for Simpson’s
following thought experiment: Pick two plants at ran- index it is the inverse (1/L = D). The ratio J′ = H′/ln S,
dom from a community. Are they members of the same called the Shannon evenness index, provides a measure
species or different species? In a highly diverse com- of evenness.
munity, it is more likely that they will belong to differ- If L is the probability that two randomly chosen indi-
ent species. viduals are from the same species, then (1 – L) is the prob-
In this example, the species that has by far the great- ability that they are from different species. This number
est number of individuals, or biomass, in the second is also known as the Gini coefficient. While the Gini coef-
community is the dominant species in that communi- ficient has been used as a measure of evenness among
ty. The first community, with greater evenness, does not individuals in a population, it could also be used as a
have any dominant species. The greater the numerical measure of species evenness in a community (Lande
preponderance of one or a few species, the lower the 1996).
diversity of the community tends to be. Of course, a Which of these indices is more useful? This question
community may have one strongly dominant species has been the topic of much discussion, which has led to
and a large total number of species, giving it a high the development of numerous ways of correcting for
diversity value, but this is actually fairly unusual. Other sampling bias (departure of the estimated value for the
ways of characterizing variation among species in abun- sample from the true value due to systematic, non-
dance are considered in Chapter 14. chance causes). Lande (1996) showed that Simpson’s
Table 12.1 Some species diversity and evenness indices
Usual
Index symbol Formula Emphasizes
Species number indices

S
Species density None area sampled
Margalef ’s index DMg S −1

ln N
Menhinick’s index DMn S

N
Proportional abundance indices
Shannon-Weiner index H′ Σ
– [p ln (p )]
i i Rare species
Inverse Simpson’s index D 1/Σpi2 Common species
 
log 2  N ! 
Pielou’s index HP  ∑ n i ! Rare species
N
Brillouin index HB ln(N !) − ∑ ln(n i !) Rare species

N
McIntosh’s U index U ∑ n i2 Rare species
∑ni
2
N−
McIntosh’s D index D Common species
N− N
N max
Berger-Parker index d Common species
N
∑ n i − NS
Cuba’s index DC S +1− Common species
2N
1
2
n R1 + ∑ n r + 21 n R2
Q statistic Q Rare species
R 
log  2 
 R1 
Evenness indices
H'
Shannon evenness J ln(S )
HB
 
ln  n!
Brillouin evenness EB
1

N  [ N ]!S − r {( N + 1)!}r 
 S S 
McIntosh evenness EU
N− ∑ n i2
N− N
S
Definitions of symbols
S = the number of species in the sample
pi = the proportion of individuals in the ith species (pi = ni/N)
ni = the number of individuals of species i in the sample
N = the total number of individuals sampled
Nmax = number of individuals of the most abundant species
nr = the total number of species with abundance R
R1 and R2 are the 25% and 75% quartiles
nR1 = the number of individuals in the class where R1 falls
nR2 = the number of individuals in the class where R2 falls
Note: While these indices were originally defined with regard to numbers of individuals, other measures, such as
cover or biomass, can be used. For more information on diversity indices, see Magurran (1988).
244 Chapter 12
index can be estimated without bias, but that estimating There are other methods besides the use of indices
the Shannon-Weiner index requires actually knowing for quantifying diversity and comparing diversity
the true number of species. The unbiased estimate of D among communities. These methods include graphical
is (1 – N)/(1 – NL). This lack of bias is an advantage for approaches based on comparisons of dominance-diver-
Simpson’s index over the Shannon-Weiner index. Which sity curves (discussed more fully in Chapter 14). Other
index is preferable in a given study depends in part on approaches have been used, but they go beyond the
what you wish to emphasize: the Shannon-Weiner index scope of this book.
is sensitive to changes in the proportions of rare species,
while the inverse Simpson’s index is sensitive to changes Sampling Methods and Parameters for Describing
in the proportions of common species (Table 12.2). If one Community Composition
had hypothesized, for example, that some environmen- Several techniques can be used to sample a communi-
tal factor were having a negative effect on community ty. Which to use in a given case will depend on the type
diversity over time by disproportionately affecting rare of vegetation being sampled and the goal of the survey.
species, reducing their abundance or eliminating them, Plant ecologists influenced by the Zurich-Montepellier
the Shannon-Weiner index might pick up such effects school (the Braun-Blanquet approach; see Chapter 16)
more readily. typically sample and compare multiple communities by
Just documenting that one community has a higher placing a single large sample plot—called a relevé—in
diversity than another community is not enough to con- each stand. The relevé is located subjectively, with an
clude that the two communities are truly different. The attempt to place it within a uniform patch of vegeta-
communities may actually be identical, and the differ- tion that is representative of the community. How large
ences in diversity found might be due to chance sam- should the relevé be? Ecologists determine the appro-
pling events. One more piece of information, a meas- priate relevé size for a given type of community by using
ure of the precision of our diversity parameter, is needed a set of nested relevés in one or a few communities,
before comparisons can be made. Methods for estimat- beginning with a small one and increasing the area sam-
ing precision are described in the Appendix. A formula pled around it in a series of steps (see Figure 12.3B). As
to estimate the variance for Simpson’s index is given in the total area is increased, a species-area curve is con-
Lande (1996), and the variance for the Gini coefficient structed until the curve reaches a plateau. That size
can be estimated with methods in Lande (1996) or Dixon relevé is then used for the remainder of the communi-
(2001). More detailed information on such topics as sam- ties sampled (if they have similar enough characteristics
pling effects, effects of quadrat size and shape, and sen- and structure). Because European botanists carried out
sitivities to various patterns of abundance can be found many of the early vegetation surveys in the Tropics,
in Greig-Smith (1964), Pielou (1975), Magurran (1988), often the only available quantitative descriptions of these
and Krebs (1989). areas have been done using the relevé method.
Ecologists in English-speaking countries
have long used the approach of sampling a
given stand or community using some num-
Table 12.2 A comparison of three species diversity measures as applied to ber of smaller quadrats, or sample plots,
six communities, each containing five species making the size and number of quadrats the
Community same across stands or communities. These
quadrats can be placed either randomly with-
1 2 3 4 5 6
in the community or regularly along a grid.
Species A 20 30 40 50 60 960 The advantage of using a number of smaller
Species B 20 30 30 20 10 10 quadrats, rather than one large relevé, is that
Species C 20 20 10 10 10 10 any local patchiness will even out across the
Species D 20 10 10 10 10 10 entire sample. This method also provides a
Species E 20 10 10 10 10 10 measure of local heterogeneity. The issue of
Sample size 100 100 100 100 100 1000 regular (systematic) versus random place-
Species richness 5 5 5 5 5 5 ment of the quadrats has been heatedly
eH′ 5 2.84 2.07 1.95 1.36 1.04 debated, and involves considerations beyond
D 5 4.17 3.57 3.13 2.50 1.08 the scope of this book (interested readers
J 1 0.65 0.45 0.42 0.19 0.02 should consult the methodological references
at the end of this chapter).
Note: The exponential Shannon-Weiner index and the inverse Simpson’s index differ in how
they change as evenness changes from sample 1 to sample 6. The numbers refer to numbers The size and shape of the quadrats is
of individuals, but could also represent biomass. also an issue. Forests in North America are
often sampled using 0.1-hectare quadrats. The total area Quadrats and transects can be used to estimate sev-
sampled should generally be determined by the species- eral parameters, based on species’ presence or abun-
area curve, as we saw above. As for quadrat shape, dance in the samples. Frequency is the percentage of
square quadrats are typically used, especially for small quadrats in which a species appears. Such a measure has
areas (< 4 m2). Usually the quadrat is marked with a the advantage that it can be done quickly, even by a sin-
rigid sampling frame, formerly made of wood or alu- gle person. In particular, it is not necessary to be able to
minum, now typically made of PVC pipe. Larger determine how many individuals of a given species are
quadrats can also be set up using posts and measuring in a quadrat. The limitation of this method is that it is
tapes in the field, particularly in forests. For the largest inaccurate for very rare species or species with clumped
areas, circular quadrats may be more efficient. In this distributions and uninformative for very common
case, a post is placed at the center of the sampling area species (those that appear in every quadrat).
with a rope or tape measure attached to a swivel point Another quantitative measure is cover: the percent-
at the top. The desired radius is marked on the rope, and age of the ground covered by a given species. Cover may
a circular area is swept out. This method has the advan- be measured as basal cover or basal area—the area
tages of avoiding the problem of precisely determining occupied by the base of a plant, such as a tussock grass
90° corners for a square or rectangular plot and of hav- or a tree with a definable base—or as canopy cover.
ing a lower area-to-circumference ratio. Cover values can tally to more than 100% if there are sev-
Ecological conditions may also dictate the shape of eral layers of vegetation (e.g., in a forest, trees may cover
the quadrat. For example, on a steep hillside, if you 70% of the ground, herbs 50%, and shrubs 20%). Cover
wanted a given sample to be within a particular eleva- can be estimated in several ways.
tional contour, a narrow rectangular plot would be most The most precise, but most laborious, method of
appropriate. A narrow rectangle might also be prefer- estimating cover is to map the vegetation. Efficient map-
able in a patchy habitat if one wanted to include more ping techniques exist for large individuals—shrubs and
patches in the samples. With a narrow rectangle, how- trees. Alternatively, for herbaceous vegetation, if it is not
ever, there is a greater chance of miscounting because too dense and the species are distinctive, a photograph
more plants are near the edge of the quadrat. The shape can be taken from above. Areas of cover are then deter-
and placement of quadrats are known to affect the num- mined using image analysis software. The point-inter-
ber of species found. Depending on the pattern of envi- cept or point-quadrat method for quantifying cover uses
ronmental heterogeneity, rectangular quadrats may con- a sampling frame with a grid of strings or a frame that
tain more species than do square or circular quadrats is used to drop pins at precise locations. The identity of
of the same size. the plant at each point where the strings cross or the pin
A related approach to sampling vegetation is the use is dropped is then determined. The total number of
of transects: very long, narrow sampling areas along “hits” (points occupied) by a species is an estimate of its
which species abundances are determined. Often tran- cover, given that enough points are sampled. In other
sects are designed to run either across or parallel to gra- words, if a species were found at 12% of the points, its
dients of environmental change, depending on the pur- estimated cover would be 12%.
pose of the sample. Transects are particularly useful in Photographic and point-intercept techniques are
sampling very large areas. One kind of transect is just a intensive and time-consuming, and are best for small
line. In the line-intercept method, a line of a set length sampling areas. For larger areas or less intensive efforts,
(e.g., 100 m) is laid through the community, and the pro- visual estimates are typically used. Humans are able to
portion of the length along the line occupied by each discern differences of about 10% by eye, so cover can be
intercepted species is determined. In a forest, a tape estimated to approximately the nearest 10% in this way.
might be suspended at some height, and the cover of Nonequal cover classes can also be used (Table 12.3). The
overstory species above the tape and understory species use of cover classes is an established part of relevé sam-
below the tape recorded along the length of the line. pling.
Another method is the belt transect. As with the line- A further quantitative measure is density, the num-
intercept method, a line of a set length is laid through a ber of individuals of a species per unit area. Density can
community. Now, all individuals lying within a belt of be measured only for species with distinct individuals
some set width on either side of the line (e.g., 0.5 m) are (e.g., trees) or where ramets are treated as individuals.
counted (belt transects can also be used to measure For small plants, quadrats are used and all of the indi-
cover). A belt transect, in other words, is a long, thin viduals are counted. For larger individuals, especially
quadrat. A long belt transect may encompass more topo- trees, plotless methods, such as the point-centered quar-
graphic or soil variation, and thus may include more ter method, can be used. To implement this method, a
species, than a square grid for a given total area sampled. series of random points are located in the community,
246 Chapter 12
Table 12.3 Three different systems for estimating cover by categorizing

estimates into a limited number of classes
Braun-Blanquet Domin-Krajina Daubenmire
Class Cover range Class Cover range Class Cover range

5 75–100 10 100 6 95–100
4 50–75 9 75–99 5 75–95
3 25–50 8 50–75 4 50–75
2 5–25 7 33–50 3 25–50
1 1–5 6 25–33 2 5–25
+ <1 5 10–25 1 0–5
r <<1 4 5–10
3 1–5
2 <1
1 <<1
+ <<<1
Source: Mueller-Dombois and Ellenberg 1974.
often along a transect line. The area around each of the usually determined by harvesting samples of the vege-
points is divided into 90° quarters, usually along com- tation, but can sometimes be estimated nondestructive-
pass headings. Then, in each quarter, the distance of the ly. For trees, a typical measure of biomass is diameter
nearest individual from the central point is measured. at breast height (DBH). DBH is measured at a standard
These distance measures are then converted into densi- height above the ground, often by measuring circum-
ties. Other methods, such as nearest individual, nearest ference using a special tape that converts circumference
neighbor, and random pairs, are variants of this tech- to diameter units. For some tree species, especially those
nique. Greig-Smith (1964) and Krebs (1989) discuss the that are commercially important, the relationship
details of using these methods, none of which are per- between DBH and total biomass is known. The investi-
fect. For example, clumping of individuals can result in gator can also determine this relationship by harvesting
incorrect estimates of density, depending on the details a sample of individuals, if necessary.
of how sampling is carried out.
Because different species may be represented more Physiognomy
heavily by one or another of the above measures of Physiognomy is the form, structure, or appearance of a
abundance, it is sometimes useful to combine them in a plant community. Knowledge of the physiognomy of a
weighted measure called an importance value (IV). Cur- community may tell us about the adaptations of its dom-
tis and McIntosh (1951) first defined the importance inant species to environmental conditions. The Danish
value of a species as the sum of the relative cover (i.e., ecologist Christen Raunkaier developed one widely
cover of that species divided by total cover of all species used system for describing community physiognomy.
present), relative density, and relative frequency of the He classified species based on their growth form and on
species in a community. Importance values can also be the location of their overwintering organs (those that can
defined by combining other measurements of relative survive an unfavorable season), such as meristems and
abundance, depending on what is most relevant in a underground bulbs (see Figure 8.2). According to this
given study. system, a tropical forest is made up almost exclusively
Biomass is generally correlated with cover. Cover, of phanerophytes (Table 12.4). A desert community is
however, is limited to two dimensions, whereas biomass dominated by shrubs—chamaephytes—and annual
is correlated with three-dimensional properties because herbs. A prairie is dominated by grasses and perennial
plant parts fill space. In some ways, it is a better meas- forbs—hemicryptophytes and cryptophytes. We can use
ure of abundance than cover. For many species, biomass this system both to classify general vegetation types
is a more useful measure than number of individuals (Raunkaier’s original intent) and to begin to make infer-
because individuals within a population of the same ences about the environmental factors that shape these
species can often differ in size by several orders of mag- communities. The prairie, for example, is dominated by
nitude (consider a seedling and a full-grown oak tree), species whose meristems are at or below ground level,
and the effect of an individual within its community typ- suggesting that meristems above ground level are vul-
ically depends on its size (see Chapter 10). Biomass is nerable to damage. In this case the critical environmen-
(A) Figure 12.6
40 (A) A bisect of a tropical rainforest in Trinidad,
British West Indies. (After Beard 1946.) (B) A
35 diagrammatic bisect of a temperate woodland.
The letters in the key can be used as “short-
30 hand” to provide information about the life
form, leaf shape, function, and leaf texture of
25 the species. For example, Tmdh(v)zi would
Height (m)
indicate a deciduous tree of medium stature

20
with large, compound membranous leaves and
15 discontinuous coverage. (After Dansereau
1951.)
10
0 10 20 30 40 50 60 70 80
Horizontal distance (m)
tal factors are fire and grazing (see Chapters
13 and 19).
(B) An important feature of plant commu-
nities is vertical structure. A forest commu-
nity, for example, may consist of canopy
trees, understory trees, shrubs, and herbs. A
10
tropical rainforest may have as many as six
different vegetation layers, or strata. Grass-
lands can consist of a mixture of ground-
Meters
hugging forbs and tall grasses. These prop-

erties can be quantified in several ways. The
5 bisect is a visual technique consisting of a
scale drawing of the vegetation along a tran-
sect (Figure 12.6A). A more diagrammatic
2 form of such a drawing is also possible (Fig-
0.5 ure 12.6B). Data from bisects or other sur-
0 veys can be used to determine the vertical
profile of the community. Along with den-
Life form Leaf shape and size sity, cover, or basal area, the height of each
individual is measured. Each species will
T Trees n Needle occupy some range of heights; often seed-
F Shrubs g Graminoid lings or saplings of canopy tree species are
treated separately from adults. By combin-
H Herbs a Small ing species that occupy the same stratum,
M Bryophytes h Large, broad one can produce a graph that summarizes
the distribution of vegetation in different
E Epiphytes v Compound vertical strata (Figure 12.7). These data can
L Lianas q Thalloid also be used to statistically compare verti-
cal structure among sites or community
Function Leaf texture types.
d Deciduous f Filmy
s Semideciduous z Membranous Long-Term Studies
e Evergreen x Sclerophyll
j Evergreen-succulent, leafless k Succulent or fungoid The measurements described above are like
snapshots of a community at a single point
Size Coverage
in time. But ecological processes, such as
t = Tall (T = to 25 m; F = 2–8 m; H = 2 m) b = Barren
m = Medium (T = 10–25 m; F, H = 0.5–2 m) i = Discontinous succession (see Chapter 13) or response to
l = Low (T = 8–10 m; F, H = to 50 cm) p = Tufts, groups global change (see Chapter 22), can take
c = Continuous many years, so these measurements may
be inadequate to describe a community.
Changes in a community may be gradual,
as the environment changes from year to
248 Chapter 12
Table 12.4 Distribution (percentage) of growth forms for various representative communities based on
the Raunkaier systema
Phanerophytes Chamaephytes Hemicryptophytes Cryptophytes Annuals
World average 46 9 26 6 13
Tropical rainforest 96 2 0 2 0
Subtropical forest 65 17 2 5 10
Warm-temperate forest 54 9 24 9 4
Cold-temperate forest 10 17 54 12 7
Tundra 1 22 60 15 2
Mid-temperate 34 8 33 23 2
mesophytic forest
Oak woodland 30 23 36 5 6
Dry grassland 1 12 63 10 14
Semidesert 0 59 14 0 27
Desert 0 4 17 6 73
Source: Whittaker 1975, Table 3.2.
a
These terms are defined in the legend of Figure 8.2.
(A) Eastern deciduous forest

year, or dramatic, as after a hurricane or major fire.
Therefore, a sample taken in a single year may not be
30
very meaningful; in any case, its value is greatly Canopy trees
enhanced by other samples taken at other times. If we
want to study the effects of environmental variation, we
Height (m)
20
need to sample that variation adequately; even if we are
interested only in knowing what happens on average, a
single year’s data may not be “average.” 10 Understory trees
Some of our most important ecological questions
can be answered only with long-term studies (Leigh and Shrubs
Johnston 1994). A good example of such a question is the 0 Forbs and graminoids
effect on global warming. The changes predicted to 25 50 75 100
occur in temperature and precipitation will, in some Cover (%)
cases, take decades to be manifested (see Chapter 22).
From one year to the next, these changes are expected to (B) Boreal coniferous forest
be small relative to normal year-to-year variation. By
monitoring the same communities and individuals over 30
long periods of time, we may be able to most precisely
detect the effects of long-term climate change.
Height (m)
Within long-term study areas, quadrats or individ- 20

uals are permanently marked. Repeated measure- Canopy trees
ments—perhaps of cover, productivity, diversity, phe-
nology, or other population, community, or ecosystem 10
features—can then be taken according to a set schedule,
often yearly but possibly at shorter or longer intervals. Shrubs
In such situations, nondestructive sampling methods are 0 Forbs and graminoids
25 50 75 100
usually preferable to destructive ones. If destructive
Cover (%)
methods must be used, they need to be done in prede-
termined locations so that, for example, the destructive Figure 12.7
sampling does not affect other sampling areas in the Vertical profiles of two different forest types. Each bar repre-
future. sents a different stratum of vegetation. The average height
range of a stratum is given by the width of the bar. The total
Ecology is a relatively young science, so most stud- cover by that stratum is given by the length of the bar. (A)
ies that we call “long-term” are less than a few decades Typical eastern deciduous forest. (B) Typical boreal conifer-
old. However, there are several studies that have been ous forest.
BOX 12C
The Long-Term Ecological Research Network
The Long-Term Ecological Research deciduous forest (Coweeta, Georgia), ing on international networking in long-
(LTER) network is a collaborative effort old fields (Kellogg Biological Station, term ecological research. Representa-
involving more than 1100 scientists and Michigan), and the desert-urban inter- tives of scientific programs and net-
students, with the goal of investigating face (Phoenix, Arizona). Besides pro- works that focus on ecological research
ecological processes that operate at long moting long-term studies, these sites over long temporal and large spatial
time scales. The network promotes syn- also encourage studies that are inte- scales decided to form the Internation-
thesis and comparative research across grated across a range of disciplines: al Long-Term Ecological Research
sites and ecosystems and among relat- physiological ecology, population biol- (ILTER) network. They recommended
ed national and international research ogy, community ecology, and ecosys- developing a worldwide program to
programs. The National Science Foun- tem science. facilitate communication and the shar-
dation established the program in 1980 Global scientific interest in developing of data. As of May 2000, 21 countries
to support research on long-term eco- ing long-term ecological research pro- had formed national LTER programs
logical phenomena in the United States. grams is expanding rapidly, reflecting and joined the ILTER network. Ten
The network now consists of 24 sites the increased appreciation of their more are actively pursuing the estab-
representing diverse ecosystems and importance in assessing and resolving lishment of national networks, and
research emphases, including arctic complex environmental issues. In 1993, many others have expressed interest in
tundra (Toolik Lake, Alaska), southern the U.S. LTER network hosted a meet- the model.
going on for much longer. The oldest experiment in ecol- processes responsible for community structure. One
ogy is the “Park Grass Experiment” of nutrient addition extreme view is that communities are highly predictable
and mowing on pasture communities at the Rothamst- entities in which species are tightly interlinked. An
ed Experimental Station (now the Institute for Crops opposing view is that communities are chance group-
Research-Rothamsted) in Hertfordshire in the United ings of species that are shaped by abiotic factors and
Kingdom, which as been ongoing since the 1850s. Stud- historical contingency. The current consensus lies
ies on the Swedish island of Öland, led by Eddy van der between these views. Deterministic and chance events
Maarel and conducted since the 1960s, have included and biotic and abiotic processes are all considered
both descriptions of vegetation dynamics and experi- important.
mental studies of factors maintaining species diversity. Communities can be described by several measures.
Forest dynamics have been studied at Hubbard Brook The simplest of these is species richness, the number of
Experimental Forest in the White Mountain National species in a community. Other indices of species diver-
Forest, near Woodstock, New Hampshire, since 1963. sity take into account both the number of species and the
Hubbard Brook is now part of the Long-Term Ecologi- relative abundance of each species. Relative abundance
cal Research network that was initiated in 1980 (Box can be measured as numbers of individuals, frequency,
12C). cover, or biomass, using a variety of techniques. Phys-
iognomy, the general form of the vegetation, is another
community property that can be measured. These meas-
Summary ures are used to describe community patterns in time
A community is a group of organisms that coexist in and space. Describing patterns is a first step in deter-
space and time and interact with one another. Ecologists mining which processes are responsible for shaping the
have long debated the relative importance of the plant communities that we see around us.
Additional Readings
Classical References Contemporary Research
Clements, F. E. 1916. Succession. Carnegie Institution of Washington, Hoagland, B. W. and S. L. Collins. 1997. Gradient models, gradient analy-
Washington, DC. sis, and hierarchical structure in plant communities. Oikos 78: 23–30.
Curtis, J. T. 1959. The Vegetation of Wisconsin. University of Wisconsin Lande, R. 1996. Statistics and partitioning of species diversity, and simi-
Press, Madison, WI. larity among multiple communities. Oikos 76: 5–13.
Gleason, H. A. 1926. The individualistic concept of the plant association. Lawton, J. H. 1999. Are there general laws in ecology? Oikos 84: 177–192.
Bull. Torrey Bot. Club 53: 7–26.
Mueller-Dombois, D. and H. Ellenberg. 1974. Aims and Methods of Veg- Additional Resources
etation Ecology. John Wiley & Sons, New York.
Krebs, C. J. 1989. Ecological Methodology. Harper Collins, New York.
Magurran, A. E. 1988. Ecological Diversity and Its Measurement. Prince-
ton University Press, Princeton, NJ.
13
C H A P T E R Disturbance and Succession
T
he living world is in a constant state of flux. In previous chapters, we
looked at processes of change at the level of the individual (Part I) and
the population (Part II). Change also occurs at the level of the commu-
nity. Communities change as the species present change, as the populations
change in number, age structure, or size structure, resulting in changes in phys-
iognomy and ecosystem function.
In Chapter 12 we described a variety of techniques for quantifying com-
munity characteristics. In this chapter we explore patterns of community
change and their causes. We also explore a variety of other aspects of com-
munities: their continuity in time, their predictability, and whether they reach
an equilibrium.
The process of succession has been central to plant ecologists’ study of
plant communities. Succession is directional change in community composi-
tion and structure over time. The term denotes changes over periods longer
than a single season, although very long-term trends, such as those due to cli-
mate change, are not regarded as part of succession. Succession begins when
a disturbance—an event that removes part or all of a community—is followed
by colonization or regrowth of the disturbed site by plants. Consider the change
in the plant community following the abandonment of an agricultural field
in North Carolina (Figure 13.1; Oosting 1942; Keever 1950). In the first year, a
variety of species are likely to colonize the field. The dominant species are typ-
ically short-lived ones—annuals and biennials such as Erigeron canadense (daisy
fleabane, Asteraceae) and Gnaphalium purpureum (purple cudweed, Asteraceae).
Over the next few years these species tend to be replaced by herbaceous peren-
nials such as Andropogon virginicus (broomsedge, Poaceae) and Aster ericoides
(white aster, Asteraceae). By 10 years after abandonment, shrubs such as Rhus
radicans (poison ivy, Anacardiaceae; this species also grows as a vine) and trees
such as Pinus taeda (loblolly pine, Pinaceae) begin to dominate the field. Even-
tually, after 150 to 200 years, the pines are replaced by hardwood tree species—
Quercus rubra (red oak, Fagaceae) and Carya spp. (hickory, Juglandaceae). If no
subsequent large-scale disturbance occurs, these hardwood species will remain
the dominant species.
Discussions about the causes and nature of the successional process are
part of the debate about the nature of communities (see Chapter 12). One
extreme view is that succession is an orderly and predictable process that is a
result of emergent community properties. An alternative view is that succes-
252 Chapter 13
(A) Figure 13.1
Old-field succession in the Piedmont
region of North Carolina. (A) This field,
three years after abandonment, is domi-
nated by perennial species, Andropogon
virginicus (broom-sedge, Poaceae) and Sol-
idago sp. (goldenrod, Asteraceae). (Photo-
graph courtesy of N. Christensen.) (B)
This field, eight years after abandonment,
has been colonized by Pinus taeda (loblolly
pine, Pinaceae). (C) A mature oak-hickory
forest with a canopy of Quercus alba (white
oak, Fagaceae), Q. rubra (red oak), Carya
glabra (pignut hickory, Juglandaceae), and
C. tomentosa (mockernut hickory). Note
the absence of a ground layer, a character-
istic of the infertile soils of this region.
(Photographs courtesy of R. Peet.)
(B)
ory, what we have is a complex set of
interlocking models (Figure 13.2).
These models can be applied to partic-
ular systems, providing descriptions of
the patterns of succession and the
processes responsible.
Theories of the Causes

of Succession
Several issues form the core of discus-
sion about the processes responsible for
succession. One group of questions
flows directly from the sharply diver-
gent views of communities advocated
(C) by Clements and Gleason (see Chap-
ter 12). First, what processes are respon-
sible for successional changes? Do
emergent properties and interactions
among species play a major role? Or is
succession the mere unfolding of the
life history of each species independent
of the others? If species interactions are
important, are they mainly in the form
of mutualisms, or is competition
among species more important?
Second, how predictable is succes-
sion? If two adjacent areas, or the same
area at two different times, were to
undergo succession, would the same
changes occur, leading to the same end
point? A climax is the hypothetical
sion is a rather unpredictable series of events that result deterministic end point of a successional sequence. Do
from interactions between individuals and the abiotic plant communities undergoing succession usually reach
environment. As we will see, succession involves a num- a stable climax state? To the extent that they do not, is
ber of different patterns, mechanisms, and causes. Ecol- this mainly because extrinsic disturbances (such as
ogists have amassed a great deal of information about storms) prevent it, or because of internal dynamics (such
successional processes. Rather than a single, unified the- as population cycles) inherent in the communities?
Disturbance and Succession 253
Figure 13.2 VEGETATION DYNAMICS
A hierarchical framework for a theo-
ry of plant community dynamics.
Three main processes affect commu-
nity dynamics: site characteristics, DIFFERENTIAL SPECIES DIFFERENTIAL SPECIES
SITE AVAILABILITY
availability of species, and species AVAILABILITY PERFORMANCE
characteristics. Each of these process- Course-scale Dispersal Resource availability
es is, in turn, driven by particular disturbance Agents Soil
interactions and conditions. (After Size Landscape Microclimate
Pickett et al. 1987.) Severity
Timing Propagule pool Ecophysiology
Dispersion Decay rate Germination
Land use Assimilation
Growth rate
Life history
Allocation
Third, are most plant communities static and Reproductive time
unchanging, or are they in a state of dynamic equilibri- Reproductive mode
um, or are they always in a fundamental state of flux? A Stress
stable equilibrium implies that they would return to the Climate
same structure and composition after small to moderate Prior occupants
perturbations, remaining in that state over time. Competitors
Clements’s view was that communities are unchanging. Identity
Dynamic equilibrium implies very different processes, Consumers
Disturbance
but in some ways is a modern substitute for Clements’s Resource base
perspective.
The Clements-Gleason dichotomy, however, does Allelopathy
Soil
not account for many of the other questions we consid- Microbes
er relevant to understanding succession today. For exam- Neighbors
ple, it seems obvious that one might ask about the role
Consumers
of competition (see Chapter 10) in successional process- Identity
es. This question, though, does not fit neatly into the Cycles
Clements-Gleason dichotomy, as neither of these scien- Plant defenses
Patchiness
tists ascribed an important role to competition in suc-
cession. This is especially odd, given that Clements
wrote extensively about competition, but (given his
superorganism viewpoint) he thought of it in a rather the 1890s and early 1900s, largely through the work of
different sense than modern ecologists do. Similarly, nei- Cowles (1899, 1911) and Clements (1916). These ideas
ther Clements nor Gleason appears to have thought of were a major force in plant ecology in North America
herbivory as an important factor affecting plant com- (and to some extent in the entire English-speaking
munity structure and function, though they were cer- world) for the next half century or more. The move away
tainly well aware of it. Thus, while it is useful to begin from these ideas took several directions. One influential
with the historical differences between Clements’s and new perspective on community dynamics came from
Gleason’s views on succession, it is a mistake to think Alex S. Watt (1947). In his view, a plant community is
that those differences are all that one must consider in composed of a mosaic of patches, and the patches are
trying to piece together a contemporary view of suc- dynamically related to one another (see Chapter 10).
cessional processes. Community structure results from a balance between
As with much in ecology, the answer to all of the factors that are predictable and those that are unpre-
questions posed above is that a variety of patterns and dictable (what we now might view as factors that drive
processes can be found in different communities and, to the community toward equilibrium and unpredictable
some extent, at different times. We will return to these disturbances that disrupt that tendency). Watt was care-
questions after we examine the factors that lead to suc- ful to distinguish this view—that the community is a
cessional change. Then we will ask, is climax a fixed, dynamic mosaic of patches—from that of a succession-
immutable state—an end point at which the communi- al sequence in which a homogeneous community tends
ty remains for all time, barring a major catastrophic toward a final, determinate end point. He emphasized
change? the ubiquitous patchiness of natural plant communities
The initial ideas about succession as a predictable and stated that the persistence of patches is a funda-
process leading to a static climax were developed in mental characteristic of plant communities.
254 Chapter 13
Several other early ecologists shared this perspec- different kinds of disturbances
tive. R. T. Fisher studied old-growth forests on Pisgah over a range of scales clearly
Mountain in New Hampshire, including both hardwood play a crucial role in community
and coniferous forests, over several decades, beginning ecology.
in 1910. The dynamic and patchlike character of these A disturbance can be de-
ancient forests was cogently summed up by Fisher in a fined as a relatively discrete
preface to a study by his successors (Cline and Spurr event in time that causes abrupt
1942, as cited by Dunwiddie et al. 1996): change in ecosystem, communi-
The primeval forests, then, did not consist of stag- ty, or population structure, and
nant stands of immense trees stretching with little changes resource availability,
change in composition over vast areas. Large trees substrate availability, or the Steward T. A. Pickett
were common, it is true, and limited areas did sup- physical environment (White
port climax stands (or preclimax stands), but the and Pickett 1985, p. 7). Disturbances can result in the
majority of the stands were in a state of flux result- removal of a substantial portion of the existing vegeta-
ing from the dynamic action of wind, fire and other tion in an area. They can range from damage to a single
forces of nature. The various successional stages plant to the destruction of forests or prairies over thou-
brought about, coupled with the effects of elevation, sands of hectares. Sources of disturbance include fire,
aspect, and other factors of site, made the virgin for- storms (windstorms, ice storms, tornadoes, etc.), land-
est highly variable in composition, density and form.
slides, earthquakes, mudslides, volcanic eruptions,
Interestingly, a major hurricane on September 22, 1938, floods, the activities of animals, and disease. Because
uprooted many of the trees in this forest, subsequently of the wide diversity of disturbances and their effects, it
changing the vegetation, and ongoing studies suggest is almost always more useful to discuss specific kinds of
that such infrequent, large-scale disturbances are an disturbance than disturbance as a single phenomenon.
important factor in forest community dynamics (Foster One way of classifying disturbances is by whether
1988). or not they completely remove a community, including
By the 1980s, influenced in particular by work in any organic soil. These two categories of disturbance
tropical forests and by broader trends in ecological think- lead to two categories of succession. Primary succession
ing, many ecologists had begun to view plant commu- occurs when plants colonize ground that was not pre-
nities as mosaics of patches rather than as homogeneous viously vegetated. Examples include the establishment
entities. In other words, they ceased trying to average of plant communities on lava fields, earth laid bare by
over the entire community and refocused on the small- the retreat of a glacier, rock outcrops, sand dunes, and
er scale of gaps and patches (Forman and Godron 1986; newly formed beaches and sandbars in rivers, or, at a
Pickett and White 1985). At the same time, ecologists much larger scale, islands newly emerged from the sea.
abandoned the search for idealized, static types—such Primary succession also occurs where human activity
as climax communities—to embrace concepts of vari- results in massive soil disturbance, as on mine wastes,
ability and dynamic processes as essential to the true roads blasted through rock outcrops, and other sites that
nature of communities. In the 1980s, this changing per- are left with undifferentiated soil parent material (Walk-
spective intersected with a very different tradition from er 1999). A key aspect of primary succession is that the
Europe. There, the floristic-sociological school was part process begins with the development of soil, which
of the development of landscape ecology. The European forms as the plant community develops (see Chapter
floristic-sociological approach focused on measuring 4).We will take a closer look at primary succession later
static patterns. When wed to the thinking of North in this chapter.
American ecologists interested in patch dynamics, the Secondary succession occurs when plants colonize
result was a dynamic outlook that helped put succes- ground previously occupied by a living community. In
sional studies into a wider landscape framework. We this case, soil already exists, and plant propagules, such
discuss these issues in more depth in Chapter 17. as seeds and ramets, are readily available. Forest re-
growth following a catastrophic fire is an example of sec-
Disturbance ondary succession, as is the colonization of an abandoned
agricultural field, called old-field succession. Once soils
Let’s begin by considering the mechanisms that under- are well established, there is no fundamental difference
lie successional change. Complex as the questions list- between primary and secondary succession, as long as
ed above may seem, a loose consensus has emerged sources of colonizing plants are readily available.
among ecologists since the early 1980s on at least two Since disturbances are normal characteristics of all
crucial points. First, most ecologists now think that most ecosystems, ecologists often discuss the disturbance
communities do not tend to stable equilibria. Moreover, regime of an ecosystem—the characteristics of the dis-
(A) Figure 13.3
(A) A surface fire in a forest in northern
lower Michigan, in which only ground-
level vegetation is burnt. (B) A crown
fire in central New Jersey in 1986 in
which entire trees were killed. (C) A
crown fire occurred three years earlier,
killing most of the mature trees in this
spruce-fir forest in the Rocky Moun-
tains. At this time only low-growing
herbs and shrubs are found. (Pho-
tographs A and C by S. Scheiner; B by
D. Burgess.)
(B) (C)
turbances occurring in that ecosystem. We can describe intensity are generally much more frequent than large,
disturbance regimes using three characteristics: intensi- intense disturbances. These factors vary, however,
ty, size, and frequency. The intensity of a disturbance depending on the type of community and disturbance.
is the amount of change that it causes. A forest fire (Fig- The timing of the disturbance can also be important. For
ure 13.3), for example, can be a light surface fire, which example, a fire that occurs in early spring before plants
runs along the soil surface or ground vegetation and are actively growing can have very different effects from
destroys only herbaceous plants or low-growing shrubs, one in the late summer when they are just beginning to
or it can be a severe crown fire that spreads from the set seed (Biondini et al. 1989; Howe 1995).
canopy of one tree to another, killing most of the canopy
trees as well as most of the shrubs and other vegetation. Gaps
The size (spatial extent) of a disturbance is the amount The size of a disturbance is a major factor in the types of
of area affected. A windstorm, for example, might blow species that can colonize a disturbed site. Disturbances
down a single tree, while a hurricane might blow down create gaps in communities that can be filled by colo-
hundreds or even thousands of trees over an entire sec- nizing species. In any given community, gaps of various
tion of a forest. The frequency of a disturbance—also sizes can exist at any given time, so that succession may
called the return interval—quantifies how often, on aver- be occurring on a variety of scales. For example, mounds
age, it occurs in a particular place. These three charac- made by badgers (Taxidea taxus) in North American
teristics are often correlated: small disturbances of low prairies are 0.2–0.3 square meters in size. Each year about
256 Chapter 13
0.01% of the prairie is turned over by the creation of new Fire
mounds (Platt and Weis 1977). The mounds are colo- Fire is a major source of disturbance in many commu-
nized by nearby plants whose seeds have limited dis- nities. Fires can vary tremendously in intensity, size, and
persal abilities. In contrast, a massive forest fire that frequency. One measurement of fire intensity is the
burns thousands of hectares may leave no nearby seed amount of heat transferred per unit area per unit time.
sources. In this case, the initial colonists will be seeds The most intense forest fires can release as much as
that have traveled long distances by wind or animal or 500,000 kJ/m2 in a few minutes, which is enough heat
have remained dormant in the soil. to melt an aluminum-block engine. The Mack Lake fire
A forest can be thought of as a set of patches of wide- (discussed below) released approximately 3 × 1012 kJ of
ly varying sizes that have experienced disturbances of energy, as much as 90 thunderstorms, or 9 times the
different types and intensities. The size of a patch has energy of the atomic bomb dropped on Hiroshima (Pyne
major effects on species composition, the successional et al. 1996). The speed with which a fire spreads is anoth-
trajectory within that patch, and ecosystem processes. er aspect of its intensity. Surface fires can move quickly
In an experimental study on gaps of different sizes in the through an area. In a North American short-grass prairie
southern Appalachian Mountains in North Carolina, on a very windy day, a fire can move as fast as 22
solar irradiance was two to four times higher in large km/hr—as fast as an Olympic sprinter—for at least
than in small gaps, and soil and air temperatures were short periods. Crown fires have been clocked at 12
much higher (Phillips and Shure 1990). Standing crop km/hr.
biomass and aboveground net primary productivity (see Fire frequency is a key determinant of community
Chapter 15) were three to four times greater in the largest structure and composition. In mesic prairies, fires can
than in the smallest gaps. Species richness was also recur as often as every 2 to 3 years. As we will see in
greater in large gaps, and species composition differed Chapter 18, these frequent fires destroy colonizing trees
among gaps of different sizes. We will return to the ques- and so maintain the prairie as grassland. Chaparral com-
tion of how gap formation and gap size influence over- munities in southern California burn approximately
all community species richness in Chapter 14. once every 25 years. Average fire return intervals in dif-
Ecologists’ thinking about disturbance has changed ferent community types vary greatly—from every few
substantially over the past few decades. Early ecologists, years to once a century to once every millennium (Table
for example, thought of disturbances as being unusu- 13.1).
al—or unnatural—occurrences that disrupted the ordi- Many plant species have adaptations to fire. One
nary and orderly processes of a community. In the past such adaptation is the location of the meristems (see
quarter century, however, disturbance has been recog- Chapter 8) in a place protected from fire. Prairies, where
nized as a natural part of many communities. This fires recur frequently, are dominated by grasses and
change in viewpoint has altered management practices, forbs, which have meristems at or below the ground sur-
as we will see in the next section. face. A light, quickly moving surface fire is unlikely to
Table 13.1 Examples of fire regimes in various community types in North America
Fire regime Communities

Natural fires rare or absent Pacific Northwest coastal forests; wetter regions of eastern deciduous
forests; southwestern deserts
Infrequent, low-intensity surface fires with a ±25-year Most eastern deciduous forests; pinyon-juniper woodlands of the Southwest;
return interval some montane meadows in the Rockies and Sierras
Frequent, low-intensity surface fires with a 1- to 25-year Sierra mixed coniferous forests; western montane-zone pine forests; south-
return interval combined with high-intensity surface eastern pine forests; prairies of Nebraska and Oklahoma; sawgrass prairie
fires with a 200- to 1000-year interval in the Florida Everglades
Infrequent, intense surface fires with a +25-year return Pine forests and boreal forests in the Great Lakes region; lower-elevation
interval combined with crown fires with a 100- to forests of the Rockies; California redwood forests
300-year interval
Frequent, intense surface and/or crown fires with a Most boreal forests; higher-elevation western forests; chaparral from
25- to 100-year return interval California to Texas
Infrequent crown fires with a return interval >100 years Northwestern wet coastal montane forests; subalpine forests of the western
mountains; rainforests of Hawaii
Source: Davis and Mutch 1994.
harm these meristems, although the tops of the plants could serve as a “fire ladder,” with the result that fire
may be destroyed. (Having meristems at ground level cannot as easily spread to the canopy.
also protects these plants from damage by grazers, Some species release their seeds from fruits or cones
which may have been the primary selective force acting following a fire. Some pines have a trait called serotiny,
on this adaptation; see Chapter 11.) Several tree and in which they retain their seeds in tightly sealed cones
shrub species have the ability to resprout from roots, rhi- for many years, releasing them only after exposure to
zomes, or buds underneath the bark if the aboveground fire. The cones are sealed by resin, which melts during
portions of the plant are killed or severely damaged by a fire, releasing the seeds. Following a fire, these trees
fire. Examples from the frequently burned pine barrens can release many years of accumulated seed production.
of the eastern United States are Pinus rigida (pitch pine, The sealed cones, therefore, serve as an aboveground
Pinaceae) and Quercus ilicifolia (scrub oak, Fagaceae). seed bank. Cleared mineral soil may be required for ger-
Other species with such adaptations include Eucalyptus mination in many serotinous species. Among the pine
spp. (Myrtaceae) in Australia, Populus tremuloides (quak- species with serotiny are Pinus contorta (lodgepole pine)
ing aspen, Salicaceae) at high latitudes or high altitudes in western North America, P. banksiana (jack pine) in
in North America, and Adenostoma fasciculatum (chamise, north-central North America, and some populations of
Rosaceae) in southern California. Other species, such as P. rigida in the eastern United States (Figure 13.4).
Quercus velutina (black oak) in the eastern United States Serotiny also occurs in other fire-adapted taxa unrelat-
and Pinus ponderosa (ponderosa pine) in the western ed to pines, such as many Australian species in the Pro-
United States, have very thick bark that protects the teaceae. A related adaptation occurs in a number of
tree’s cambium (the living, growing part of the trunk just annual and short-lived perennial herbs that appear after
under the bark) during a surface fire. As they grow, these fires. Some of these plants have a soil seed bank (see
species also tend to shed their lower branches, which Chapter 8) in which seed dormancy is broken by chem-
ical compounds found in ash or smoke.
Pinus palustris (longleaf pine) is a dominant tree in
forests of the southeastern United States, where surface
(A)
fires recur approximately every 3 to 5 years. Young
seedlings are especially vulnerable to these fires. P. palus-
tris has an unusual life history that appears to be an
adaptation to these frequent fires: after the first year or
so, the young plant exists as a small tuft of needles grow-
(C)
(B)
Figure 13.4
Serotiny occurs in some Pinus rigida (pitch pine, Pinaceae)
populations in the pine barrens of Long Island, NY. (A)
Ordinary open cones on a typical, non-serotinous tree. (B)
Closed, serotinous cones on a tree belonging to a frequently
burned population with a high frequency of serotiny, which
is a genetically determined trait. (C) A serotinous cone that
has opened as a result of a forest fire, releasing its seeds.
(Photographs by J. Gurevitch.)
258 Chapter 13
stems, and leaves of the dominant species. In some other
community types, the probability of fire does not increase
with time. In these communities, weather conditions,
rather than fuel availability, limit fire frequency and
intensity. In the boreal forests of North America, for
example, the chance of a fire occurring in a given stand
is roughly independent of the time since the last fire in
that stand. There is usually adequate fuel for a fire, but
weather is only sometimes conducive to forest fires (John-
son 1992). Interestingly, a roughly constant chance of fire
also occurs in stands of California chaparral that have
long been unburned: apparently, once the fuel load reach-
es a critical level, fuel availability no longer strongly
affects the chance of fire (Johnson 1992; Keely et al. 1999).
Some plant species are pyrogenic—that is, their
accumulated leaf or twig litter tends to promote fire more
than one would expect based on the mass of the litter
alone. Good examples include many Eucalyptus, some
chaparral shrubs, and possibly some pines. Oils and
other flammable chemicals produced by these plants are
often of central importance in pyrogenicity. Mutch (1970)
Figure 13.5 proposed that this property might be an evolved adap-
A seedling of Pinus palustris (longleaf pine, Pinaceae) looks tation. This proposal remains quite controversial, how-
very much like a clump of grass—hence the term “grass ever, as the conditions required for such a trait to evolve
stage.” The apical meristem is located out of sight, just as a direct adaptation to fire may be quite restrictive (see
below the ground, where it is protected from frequent fires. Chapter 6; Kerr et al. 1999). Regardless of whether pyro-
(Photograph © G. Grant/Photo Researchers, Inc.)
genicity is an evolutionary adaptation in itself or a by-
product of selection on other traits, pyrogenic species
reestablish themselves following a fire either because
ing at ground level, looking very much like a clump of adults resprout or because the population has a seed
grass (Figure 13.5). The apical meristem sits just below bank (which may be located in the soil or in serotinous
the ground, where it is undamaged if a low-intensity fire cones or fruits). Meanwhile, competing species are often
occurs. During this time, the plant grows a large root sys- killed by the fire. Thus, these pyrogenic plants create an
tem and creates a large nutrient store. Finally, when its environment that enhances their own persistence (or
nutrient store is large enough, it grows very quickly. facilitates their invasion into a new community).
Within a few years, the tree is big enough, and has bark The problem of increasing fuel load with time led to
thick enough, to be quite fire-resistant. Thus, its growth a strenuous debate about how best to manage forests
pattern minimizes the number of years that it is vul- in the United States. For most of the twentieth century,
nerable to fire. the policy of the U.S. government was to suppress fire
If weather conditions are favorable to fire, the prob- as much as possible. As a result, fuel loads and the den-
ability of a fire occurring increases with the fuel load, sity of young trees increased. Therefore, when fires did
the amount of combustible plant material in the com- occur, rather than light surface fires, they often became
munity. In prairies, the grasses die back each winter, and raging crown fires, often killing mature trees and threat-
highly combustible dead grass builds up. This buildup ening human lives and property. In the 1970s, this man-
of material increases the probability of fire with time, as agement practice began to change for two reasons. First,
well as increasing the intensity of fire when it finally the problems and dangers associated with accumulat-
occurs. Likewise, when exotic grasses invade nongrass- ing fuel loads were recognized. Second, because of the
dominated communities, they can contribute to increased understanding gained by ecologists, U.S. gov-
increased fire frequency by creating fuel where it did not ernment agencies started to recognize that disturbance
previously exist, which in turn appears to facilitate fur- and fire are natural parts of ecosystems. In the absence
ther grass invasion (D’Antonio and Vitousek 1992). of the natural disturbance regime, many properties of a
In many Eucalyptus forests in Australia, the pine bar- community can change, including species composition.
rens of the eastern United States, and chaparral in Cali- In the southeastern United States, for example, in the
fornia and other parts of the world, similar processes of absence of fire, hardwood-dominated forests eventual-
fuel buildup occur. Here the buildup is of living twigs, ly replace longleaf pine forests.
The new U.S. government management practices gested by some environmentalists and foresters is to thin
include letting naturally occurring forest fires burn commercially valueless small trees and saplings, which
themselves out (if they are not threatening human lives are the most likely to contribute to catastrophic fires, but
or property) and setting prescribed or controlled burns this proposal has not been enthusiastically embraced by
to reduce accumulated fuel loads (to avert a much big- the logging industry because it is not profitable. The
ger, uncontrolled fire). Both of these practices are con- results of these controversies have been protracted court
troversial, however. The extensive wildfires in Yellow- battles without any clear resolution at present. Even
stone National Park in the summer of 1988, which began without a political, economic, or environmental agen-
naturally and were allowed to burn, were critical in turn- da to consider, there is also the problem of determining
ing public opinion against a policy of allowing fires to what a “natural” fire regime for a forest community
burn. Prescribed burns, too, have become a subject of might be and attempting to get to that state without hav-
controversy. ing a catastrophic wildfire.
Prescribed burns are planned by foresters and set
only after careful evaluation of weather and other con- Wind
ditions. Most have been safe and effective in reducing Wind can be another significant source of disturbance.
fuel load and in facilitating regeneration of fire-adapted At one end of the scale, wind can blow down a single
species such as longleaf pine. They are used routinely to branch or tree. Such windthrows range from blow-
manage forests in many regions of the United States and downs of branches or large parts of trees to losses of sin-
elsewhere. There have been occasional but spectacular gle trees and neighboring groups of trees. Windthrows
mistakes, however, in which controlled burns were set are important in many tropical forests (Figure 13.6;
without the proper precautions and became uncon- Brokaw 1985), where trees can be very tall and are often
trolled wildfires. The Mack Lake fire on May 5, 1980, in
the Huron National Forest of Michigan began as a pre-
scribed burn, set to help manage the jack pine commu-
nity. Unfortunately, the fire got out of control. In 12
hours, the fire had burned 10,000 hectares; tragically, it
also killed a firefighter and destroyed 44 homes (Pyne
et al. 1996). More recently, a prescribed burn in the spring
of 2000 near Los Alamos, New Mexico, escaped control,
burning thousands of hectares and destroying several
dozen homes; fortunately, no lives were lost.
Managing forests and forest fires has become more
complex and difficult as suburbanization and the rise in
ownership of vacation and retirement homes in scenic
areas has led to increasing numbers of homes encroach-
ing into large, previously almost uninhabited forests.
This was the case in Los Alamos, as well as in some of
the areas devastated by the wildfires (started by light-
ning) that burned more than 2.5 million hectares in the
western United States during the very dry La Niña sum-
mer of 2000 (see Chapter 18). Furthermore, incidents
such as the Mack Lake and Los Alamos fires have made
decisions regarding fire management political and con-
tentious. Many of these discussions are underlain by
profound disagreements between environmentalists,
logging corporations, and loggers over the extent to
which the U.S. government should facilitate the logging
of public forests. The logging industry proposes that
they be permitted to thin forests to prevent severe fires
from occurring. Critics of this proposal say that this is
just an excuse to continue extensive logging of public Figure 13.6
lands, as the type of logging that is actually done A large almendro tree (Dipteryx panamensis, Arecaceae) has
created a gap in the tropical forest of La Selva Biological Sta-
removes most of the larger, commercially (and ecologi- tion in Costa Rica. This photograph was taken approximate-
cally) valuable trees, leaving the smaller trees that are ly one year after the tree fell. (Photograph © G.
most vulnerable to fire. An alternative that has been sug- Dimijian/Photo Researchers Inc.)
260 Chapter 13
Figure 13.7
Locations of all tornadoes recorded the
United States from 1981 to 1990. Torna-
does are most frequent in a region
stretching from eastern Texas and
Louisiana northward to Minnesota and
Michigan, an area sometimes called
“Tornado Alley.”
connected by vines. As a result, when one tree falls, it nity dynamics. Soil erosion by wind can also be a major
often brings down others (Putz 1983). In the MPassa for- disturbance factor in more open communities, from arid
est of Gabon, Africa, for example, 51% of gaps were lands to plowed former grasslands (see Chapter 4).
caused by single falling trees and were responsible for
38% of the total gap area (Florence 1981). Trees that fell Water
in a domino-like fashion accounted for 14% of gaps and Water can be an important source of disturbance in both
16% of the total gap area. Once a gap forms, neighbor- its liquid state, as floods and nonflood erosion (includ-
ing trees become more susceptible to being blown down. ing soil erosion), and in its solid state, as snow and ice.
In MPassa, such adjacent treefalls made up 13% of the Floods are most important in riparian habitats (areas
gaps and 36% of the total gap area. In that forest, the adjacent to streams and rivers) and areas near swamps
average time between gap formation at any one spot is and bogs. In many such systems, annual floods are con-
about 60 years, and the average size of a gap is 3 ha. tinually creating and destroying habitat. In Alaska, for
Treefall in such forests is clearly a very important source example, the sandbar willow (Salix exigua, Salicaceae)
of disturbance. grows on sandbars along rivers that are continually
Extremely powerful storms such as hurricanes and being destroyed and re-formed by snowmelt-caused
tornadoes, while rare, are an important source of wind floods. Heavy rainstorms can also lead to landslides.
damage. Hurricanes, typhoons, and cyclones are impor- These types of disturbances can initiate primary suc-
tant in coastal regions. Caribbean hurricanes, for exam- cession because they tend to create new, previously
ple, pass over any given patch of forest, on average, unvegetated ground. Similarly, snow avalanches are
every 15–20 years. These storms can also be important responsible for disturbances in temperate mountain
in temperate regions. In the northeastern United States, regions. In the Canadian Rockies, avalanches remove 1%
hurricanes severe enough to blow down large tracts of of the forests each winter.
forest recur once or twice a century. In 1938 a hurricane Ice storms are important sources of disturbance in
blew down 253,000 ha of forest in central New England many temperate regions. In temperate deciduous forests,
(the same hurricane that destroyed most of the Pisgah such as those of the southeastern United States, they are
forest mentioned above). Tornadoes are another impor- often responsible for many small-scale disturbances,
tant source of wind damage, especially in the midwest- such as the removal of single branches from trees. Such
ern U.S. “Tornado Alley” (Figure 13.7). In the forests of disturbances create new patches in the forest. Ice storms
northern Wisconsin, catastrophic windthrows create, on can also cause large-scale disturbances. In January 1998,
average, 52 patches each year, ranging in size from 1 to a massive ice storm covered large areas of New York and
3785 ha, with a return interval of 1210 years (Canham New England in the United States and Ontario, Quebec,
and Loucks 1984). and New Brunswick in Canada. Approximately 25% of
The gaps and patches created by all of these kinds the trees in the region were damaged moderately to
of wind damage are important factors in forest commu- severely in that single storm.
The combination of disturbance by wind and by 1868, in an ill-advised attempt to establish a silk indus-
water can be important in some communities, such as try in North America by crossing it with the silk moth.
the cove forests of the southern Appalachian Mountains It escaped from cultivation and spread rapidly through-
(Runkle 1985). These highly diverse forests occur in shel- out the eastern deciduous forests of North America;
tered areas, near creeks, at middle elevations. Fire is very today, it can be found as far west as Oregon and as far
rare in these forests; windthrows are the major source of south as Arkansas. During an outbreak, gypsy moth
disturbance. Most gaps are caused by the deaths of sin- caterpillars can defoliate large tracts of forest, under some
gle trees due to ice storms, lightning, or high winds. The circumstances killing many of the trees (Figure 13.8).
average gap is small, about 31 m2, although gaps can be Many other kinds of animal activities can create dis-
as large as 0.15 ha when several trees fall together. On turbances (see Chapter 11). Large animals, such as the
average, a single spot experiences a gap-causing distur- American bison in North America and the African buf-
bance once every 100 years. falo in Africa, graze and also create wallows. In the
Konza Prairie Research Natural Area (see Chapter 1),
Animals bison wallows are important habitats for annual species
For the most part, disturbances caused by animals are found nowhere else on the prairie.
small and often associated with herbivory (see Chapter
11). Sometimes this herbivory can be quite dramatic, Earthquakes and Volcanoes
however, as when a herd of elephants strips the foliage Seismic events such as earthquakes and volcanic erup-
from a stand of trees. A single elephant can consume 225 tions lead to primary succession. Volcanic lava flows can
kg of forage per day! create new islands in the ocean or bury existing lands.
Even small animals such as insects can cause wide- The eruption of Mount St. Helens in Oregon on May 18,
spread disturbance during periods of population out- 1980, created nearly 16 km2 of bare ground from lava
breaks. The gypsy moth (Lymantria dispar), for example, flows (Figure 13.9; we will look at the patterns of suc-
was introduced from Europe into Massachusetts in about cession that followed this eruption below). Seismic
FPO
Figure 13.8
Gypsy moths caterpillars (Lymantria dispar) defoliated this Figure 13.9
Fagus grandifolia (American beech, Fagaceae) in an infesta- The 1980 eruption of Mount St. Helens in Washington State
tion that occurred in Connecticut in 1980. Deciduous trees created nearly 16 km2 of bare ground, leading to primary
do not always recover from such an attack. (Photograph by succession. (Photograph © D. Weintraub/Photo Researchers,
J. Bova/Visuals Unlimited.) Inc.)
262 Chapter 13
events can also have indirect affects by triggering land- tryside around the world (see Chapter 11), in many cases
slides. In Chile, a series of devastating earthquakes in greatly accelerating soil erosion. Even the trampling
1960 set off thousands of debris avalanches, landslides, associated with recreational hiking and other uses of
and mudflows, which covered 25,000 ha, or 2.8%, of the wilderness areas can be substantial in areas with large
land surface of the province of Valdivia (Veblen and Ash- numbers of visitors, or in areas that are especially vul-
ton 1978). nerable, such as deserts and tundra.
Disease
Colonization
The spread of disease can also be a source of distur-
bance. Chestnut blight was introduced to North Amer- All successional processes begin with propagules (seeds
ica in the early twentieth century, leading to the death or other dispersal structures, such as seed clusters). Pri-
of chestnut trees (Castanea dentata, Fagaceae, Figure mary succession differs from secondary succession in
11.17) over most of the eastern deciduous forest biome. the types of propagules involved and in their sources.
Other examples of pathogen-caused disturbances are Generally, in primary succession, all propagules must
described in Chapter 10. be seeds or spores carried to the newly bare ground by
wind or water. Animal dispersal plays a small role in
Humans primary succession because relatively few animals visit
One of the most important causes of disturbance in nat- unvegetated sites. First to appear are species of mosses
ural communities today is human activity (see Chapter and lichens that specialize in colonizing bare ground.
22), including disturbances caused by domestic animals. These species tend to be slow-growing and to possess
Sometimes the disturbances we create are accidental or unique adaptations that enable them to live under the
due to casual mischief. For example, approximately 35% difficult conditions found at primary successional sites.
of the forest fires in the United States, representing 19% In addition to the difficulties posed by the lack of organ-
of the area burned, are due to either human carelessness ic matter, nutrients, and soil structure, other conditions,
or arson. Other human disturbances are more purpose- such as widely fluctuating temperatures and exposure
ful. Commercial logging removes large numbers of trees to high winds, make survival at these sites difficult for
from U.S. forests annually in a combination of clear-cut- most plants.
ting (removal of all trees) and selective logging Colonization by dispersing
(removal of some commercially valuable trees, leaving seeds is also important in sec-
others). After clear-cutting, timber companies are legal- ondary succession, but the colo-
ly required to replant to replace the trees they have nizers usually differ from those
removed. However, because the replanted trees are typ- involved in primary succession.
ically a single economically valuable species that grows Because soil is already present,
quickly, the resulting forests are often dramatically dif- many fast-growing species can
ferent from the original forest communities. Following grow in the newly available
clear-cutting and replanting, most forests become mono- patches. Peter Grubb (1977), in-
cultures (plantations composed of a single tree species), spired by Watt’s work, defined
with all of the trees usually of the same age. Further- the regeneration niche as the set
Peter Grubb
more, logging roads can act as corridors for the move- of environmental requirements
ment of invasive species, while creating barriers to the necessary for a plant species to reproduce. Grubb pro-
movement of other species (see Chapter 17). posed that by differing in their regeneration niches,
In many tropical forests, clear-cut timber harvesting species that were otherwise very similar could coexist.
and clearing of land for farming has been extensive. While newly disturbed areas such as gaps present
Worldwide, rainforest destruction is currently estimat- many challenges to colonizing individuals, they also
ed at 1 ha per second (about equal to two U.S. football offer numerous advantages. Light levels are generally
fields), 87,000 ha per day (an area larger than New York high because of the lack of a dominant plant canopy. Soil
City), and 32 million ha per year (an area larger than nutrients may also be high because of both reduced com-
Poland). But not only the forest that is destroyed suffers petition and the release of nutrients from decaying
disturbance. The remaining areas of forest are also affect- plants or plant parts (e.g., leaves from the canopy trees
ed, as fragmentation of the landscape changes species killed in a large blowdown). The colonizing species in
migration patterns and population sizes (see Chapters secondary succession are often (but not always) very fast
17 and 22). growers with small, wind-dispersed seeds. Taraxacum
Grazing by domestic cattle, sheep, and goats has officionale (dandelion, Asteraceae), for example, has
drastically altered the composition of grasslands and wind-borne seeds. Often these colonizing species are
shrublands in huge tracts of arid and semiarid coun- found at a site for only a year or two following the dis-
(A) Figure 13.10
Disturbance and early succession in a for-
est in northern lower Michigan. (A) The
site was clear-cut and burned in August
1980. (B) Early in June, 1981, the site is
covered by Populus grandidentata (bigtooth
aspen, Salicaceae) about 10 cm tall that
has re-sprouted from roots in response to
the fire. (C) One year following the fire, in
August 1981, the aspen are now about 1
meter tall. (Photographs by S. Scheiner.)
tats have seeds that are able to remain

dormant for many years and germinate
under specific conditions that follow a
(B) disturbance. In particular, the seeds of
many short-lived herbs have dorman-
cy that is broken only by exposure to
sufficient light, which occurs only after
soil disturbance. You may have ob-
served this in a garden: after you turn
the soil over, many weedy species seem
to appear suddenly. These plants have
a soil seed bank and light-induced ger-
mination. Other cues can also be impor-
tant; for example, some seeds germi-
nate in response to the chemicals in ash
or smoke left from a fire.
The composition of the soil seed
bank can be quite different from that of
the plant community growing above it
(C) (Oosting and Humphreys 1940; Liv-
ingston and Allessio 1968). In an
upland oak-pine forest in Michigan
(Figure 13.10), for example, Arenaria ser-
pyllifolia (sandwort, Caryophyllaceae)
was common in forest seed banks, yet
was absent above ground. In contrast,
Erigeron canadensis (fleabane, Aster-
aceae), which has wind-dispersed
seeds, was one of the most common
species following a fire, but was absent
from the soil seed bank (Scheiner 1988).
A gap in this forest might be quickly col-
onized by both species: Arenaria from
the seed bank and Erigeron by dispersal
of wind-borne seeds from other areas.
Vegetative propagules can also be
an important source of regrowth fol-
lowing disturbance. Aspens (Populus
turbance before being outcompeted by other species (see tremuloides and P. grandidentata) in northern temperate
Figure 13.1). forests specialize on growing in early successional habi-
Soil seed banks (see Chapter 8) can also be an impor- tats. These trees can spread by “root suckering”; that is,
tant source of propagules for secondary succession. through the growth of new stems as offshoots from rhi-
Some species that specialize on early successional habi- zomes. Fire stimulates their tendency to produce new
264 Chapter 13
shoots. Following an experimental burn in northern thesized that the dunes were ini-
lower Michigan, there were 24,750 aspen stems per ha, tially colonized and stabilized
and these shoots grew to an average height of over 1 m by plants with extensive root
in a single summer (see Figure 13.10). systems, such as the grasses
Ammophila breviligulata (beach
grass) and Elymus canadensis (wild
Determining the Nature of Succession
rye), and small woody plants
Early ecologists thought of succession as an orderly pro- such as Prunus pumila (sand-cher-
gression of seral (pre-climax) communities leading to a ry, Rosaceae). Once the dune is
predictable end point—a stable climax state. Beginning stabilized by these plants, Cowles
in the 1950s, as we saw above, a number of new views argued, larger shrubs, such as Henry Chandler Cowles
about succession emerged. Ecologists are still trying to Salix glaucophylloides (dune-wil-
sort out the patterns and processes of succession. low, Salicaceae) and Cornus stolonifera (red osier, Cor-
naceae), can colonize it. Eventually, trees such as Tilia
Interaction between Methodology and Understanding americana (basswood, Tiliaceae) enter the community,
The ability of scientists to understand nature is affected converting it from a thicket to a forest. Finally, under
by the methods they use to study it. This is nowhere Cowles’ scenario, the forest becomes dominated by Acer
more apparent than in the study of succession. Changes saccharum (sugar maple, Sapindaceae) and Fagus grandi-
in methods for studying succession have affected our folia (beech, Fagaceae).
understanding of the process; conversely, changes in The chronosequence approach rests upon three
ecologists’ understanding of succession have also fos- important assumptions. First, it assumes that succession-
tered changes in methods. al processes are highly predictable; that is, that similar
A central difficulty in studying succession is that it communities will all undergo the same successional
is often a very slow process—it can take decades or cen- sequence. Second, it assumes that the climate and other
turies. Ideally, one would like to be able to observe the aspects of the environment, as well as the species pool
entire successional process in a single community, but (the species available for colonization), will remain the
this is generally not possible. One traditional way to get same during the entire chronosequence. We know, how-
around this limitation is by studying a chronosequence, ever, that this is not always true. For example, new
a set of communities of different ages since disturbance. species may migrate into a region, species combinations
Users of the chronosequence approach assume that dif- may change during major periods of climate change (see
ferences among these communities represent what Chapter 21), or species may go locally extinct. In recent
would occur within a single community over time. H. C. years, especially, invasive species have changed succes-
Cowles (1899) pioneered and popularized this approach, sional trajectories (see Chapter 14). Finally, the chronose-
which was based on his studies of patterns of primary quence approach assumes that communities in the same
succession in the dunes and forests along the southern general region are similar enough to be considered part
shore of Lake Michigan (Figure 13.11). Cowles hypo- of the same successional sequence. This last assumption
Figure 13.11
A foredune along the shore of Lake
Michigan at Big Point Sable, Luding-
ton, Michigan, taken on September
13, 1916. On the left is Ammophila bre-
viligulata (beach grass, Poaceae), and
on the right is Prunus pumila (sand-
cherry, Rosaceae). The root systems of
these species stabilize the open
dunes. (American Environmental
Photographs Collection, AEP-MIS136,
Department of Special Collections,
University of Chicago Library.)
was called into question during the 1950s undermining
the use of chronosequeces—an example of how concep-
tual shifts drive changes in methods.
Two new research approaches were used to direct-
ly test the first assumption, that the outcome of succes-
sion is highly predictable—an example of how methods
sometimes drive conceptual advances. The first of these
methods was the use of long-term (or longitudinal) stud-
ies, which provided data that were not available to the
earliest ecologists. Beginning in the 1920s and 1930s,
ecologists established permanent study sites that were
monitored for decades or longer. Ecologists have now
directly observed long successional sequences and com-
pared them with those predicted by the chronosequence
method. In a famous example from North Carolina,
Oosting (1942), along with his student Catherine Keev-
er (1950), used the chronosequence method to predict
that pine forests would eventually yield to hardwood-
dominated communities. This hypothesis was tested
using long-term study plots established in the Duke For- Figure 13.12
est in 1934 and continuously monitored ever since. The Annual rings show clearly in this cross section of the trunk
methods of observation included tagging every tree with of a loblolly pine (Pinus taeda). (Phograph © G. Grant/Photo
Researchers, Inc.)
a trunk greater than 1 cm DBH. Based on these studies,
Christensen and Peet (1980) determined that Oosting’s
hypothesized successional sequence was correct. A
drawback to long-term studies is that time is still a lim- establishing plots with different sets of species (see
iting factor. A single researcher can examine succession Chapter 15). Time is again an important limiting factor
over a period of decades, but not centuries. In addition, in these kinds of experiments. They are most useful for
this method can be used only in situations where sites addressing questions over a time span of a few years to
can be established and monitored without fear of human over a decade in unusual circumstances. That is why
disturbance, and where there is a dedicated effort to they have mainly been used in communities such as
maintain the study over a long period as investigators, prairies and old fields, where successional change ini-
funding, and even institutions change. tially proceeds at a relatively fast pace.
The second shift in methodology was an increase in Another method that has been used extensively over
the 1970s in the use of manipulative experiments to the years to reconstruct the history of particular com-
study community processes. Until then, plant commu- munities is dendrochronology, the study of tree rings.
nity ecology was mainly an observational discipline, This method is particularly useful for picking up the
involving the collection and analysis of extensive quan- study of succession where the other two methods leave
titative data. Now ecologists began to test the explana- off: over a period of centuries. Many temperate-zone
tions proposed for successional processes by applying trees produce annual growth rings (Figure 13.12). By tak-
different experimental treatments to communities and ing a core from a tree and counting the rings, the tree’s
studying their responses. Three of the most extensive of age can be determined. Knowing the ages of the trees in
these studies in the United States have been conducted a forest allows one to determine when various species
at sites that are part of the Long-Term Ecological Re- became established at that site. Other characteristics of
search (LTER) network (see Box 12C). In Chapter 1 we tree rings, such as size, appearance, chemical composi-
describe one of these sites: the Konza Prairie Research tion, and isotopic composition, can provide additional
Natural Area in Kansas. A second is Hubbard Brook, information about the climate, soil, and growing con-
New Hampshire, where Frank Bormann and Gene ditions in times past. We discuss other methods of deter-
Likens used whole-watershed manipulations to study mining long-term, nonsuccessional change in plant com-
succession (Bormann and Likens 1979). The third is the munities in Chapter 21.
Cedar Creek Natural History Area in Minnesota, home
to an ongoing study of prairie succession established by Mechanisms Responsible for Successional Change
David Tilman. The experimental manipulations used Using these various methods, plant ecologists have
at these sites include fertilizing plots with different lev- slowly uncovered successional patterns and the process-
els of nitrogen, creating plots of different sizes, and es responsible for them. The first comprehensive suc-
266 Chapter 13
cessional theory was that of Clements (1916); it was part
of the superorganism concept, his overall theory of the
nature of plant communities (see Chapter 12). In
Clements’s view, some species played the role of prepar-
ing a site for the later occupation of other species. Inter-
actions among species were seen as benefiting the func-
tioning of the whole community. Some species existed
in the community for a limited time before bowing out
and allowing others to take over. Clements believed that
these species were somehow designed for this purpose.
Today, we recognize that Clements’s ideas were pro-
foundly non-Darwinian. He developed them at a time
when the Darwinian theory of evolution by natural
selection was still a highly debated issue. Today, no sci-
entist would accept the notion of a species existing only
for the good of the rest of the community. Mutualistic
interactions can be favored by natural selection (Wil-
son 1980). The evolution of mutualisms is complex, and
their role in driving succession is much more limited
than envisaged by Clements.
How, then, do populations of successional species
interact with one another? To develop a set of testable
hypotheses, Connell and Slatyer (1977) proposed three
possible mechanisms by which succession might occur.
First, early successional species might facilitate the
colonization and growth of late successional species.
This hypothesis looks at succession from a viewpoint Figure 13.13
akin to Clements’s, albeit within an evolutionary con- In Arizona’s Sonoran desert, a Carnegia gigantea (saguaro
cactus, Cactaceae) grows in the shade of its nurse plant, a
text and in a more mechanistic fashion. A second Parkinsonia microphyllum (palo verde tree, Fabaceae). Once
hypothesis is that a process of inhibition drives suc- the cactus grows large enough, it no longer requires protec-
cession. An early successional species might inhibit col- tion and will displace the tree as Carnegia outcompetes
onization by later successional species by monopolizing Parkinsonia for water. (Photograph courtesy of T. Craig.)
resources such as light, water, or nutrients. Succession
would eventually occur when later species outcompet-
ed earlier species. Finally, a process of tolerance, in Secondary succession in very harsh environments
which species neither help nor hinder colonization by also provides examples of facilitation. In the Sonoran
other species, might determine succession. Desert, seedlings of Parkinsonia microphyllum trees (palo
There is evidence for all three of these mechanisms. verde, Fabaceae) survive poorly in the open. Instead,
Facilitation is best demonstrated during primary suc- they germinate and grow in the shade of shrubs such as
cession. Some species are adapted to eking out a living Ambrosia deltoidea (Asteraceae). Because of this role in
under difficult conditions. Lichens, for example, can the community, the latter are called nurse plants (see
grow on bare rock. They do so by slowly dissolving Chapter 10). Similarly, Carnegia gigantea (saguaro, Cac-
away the rock to gain nutrients, thus beginning the taceae) germinate and grow under both Parkinsonia and
process of soil formation. Another example of facilita- Ambrosia (Figure 13.13), so Parkinsonia shifts roles to
tion during primary succession occurs along lakeshore become a nurse plant. The eventual replacement of early
dunes, such as those of the Great Lakes in the north-cen- successional species by later ones in these instances is
tral United States. Species such as the grass Elymus not a result of one species voluntarily giving way to
canadensis can colonize newly deposited sand, with its another, as Clements suggested. The later species even-
very low levels of nutrients and poor water-holding tually outcompete the early species—in fact, there is
capacity. In areas of active sand dune formation, the good evidence that saguaros eventually kill off their
fibrous root systems of this grass act to stabilize the sand nurse trees. The early successional species persist in the
(see Figure 13.11). Once the sand is stabilized, other landscape because they are able to colonize and live in
species with smaller root systems are able to colonize the ephemeral habitats.
site; the establishment of these species is thus facilitated Inhibition and tolerance are the easiest succession-
by Elymus. al processes to demonstrate and observe. Inhibition
comes about because of competition for light, water, and before any of the other trees that germinate in that gap.
nutrients (see Chapter 10). One or more of these So pin cherry inhibits—slows down—the colonization
resources is always in limited supply in a given com- of sugar maple, while sugar maple is able to colonize
munity. In a desert, the limiting resource is often water. because of its tolerance of low light.
In very sandy soil, the limiting resource is often nitro- Another theory points to additional processes that
gen (see Chapters 4 and 15). In a forest understory, the are important in succession. In 1954, Frank Egler
limiting resource is usually light. Competition for these described two alternative scenarios for succession (Fig-
limiting resources results in reduced growth and repro- ure 13.14). The first, known as relay floristics, was large-
duction. Thus, competition reduces a species’ ability to ly a Clementsian view, in which integrated communities
colonize and grow in a site. replace one another like runners in a relay race. Egler
Competition is less intense if the competing species proposed a second scenario that he thought was more
are adapted to somewhat different conditions. For exam- likely: the initial floristic composition hypothesis,
ple, some species grow best under high light conditions, which emphasized the process of colonization and dif-
while others do best under low light conditions. During ferences in the life spans of species. Focusing on suc-
forest succession, light levels decrease as
trees grow up and the canopy fills in. In this
(A)
case, species adapted to high light intensi- Field
ties tend to occur early in succession (or in abandoned Perennial
gaps in a mature forest), while species adapt- Annual forbs and
ed to low light usually occur in older com- Crop forbs grasses Shrubs Trees
munities. In the temperate forests of the
northeastern United States, Prunus pensyl-
vanicus (pin cherry, Rosaceae), a species that
grows best in high light conditions, occurs
in early successional forests. Because of its
Relative biomass
high light requirement, its seedlings cannot

grow up underneath its own canopy.
Instead, stands of pin cherry are often colo-
nized by Acer saccharum, which can grow
under much lower light conditions. Growth
under these conditions is actually very slow,
and a sugar maple sapling would never
become a full-sized tree if it were always
growing in the shade of another tree. Rather,
the strategy of the sugar maple is one of tol- Time
erance. It can survive, growing slowly, under
(B)
low light conditions. Then, if a canopy tree Field
happens to fall and create an opening, the abandoned Perennial
sugar maple sapling can grow quickly to fill Annual forbs and
that gap. Because the sugar maple is already Crop forbs grasses Shrubs Trees
at the sapling stage, it can reach the canopy
Figure 13.14
F. E. Egler’s theories of succession (Egler 1954),
Relative biomass
stylistically diagrammed for a hypothetical

abandoned field in North Carolina. Each line
represents a single species of the vegetation
type indicated. The thicker the line, the more
important the species at a given time. (A)
According to relay floristics, groups of species
replace one another like runners in a relay race.
(B) Egler’s modification of relay floristics was
the initial floristic composition theory. Here all
of the species are present at the beginning of
succession, which is simply a process of the
unfolding of their various life histories. Time
268 Chapter 13
cession in abandoned fields in the eastern United States Successional time
(see Figure 13.1), Egler claimed that all species reach a Early Late
site early in the successional process. Early succession-
al species, such as annuals and perennial herbs, domi- A Z
nate the site because they grow quickly. Tree species also
arrive early, but do not dominate until much later
because they grow more slowly. Thus, in Egler’s view, a A
particular successional sequence is a direct consequence Z
B
of the initial composition of the plant community. Put
differently, Egler’s theory was that succession is sim-
C
ply a process of the unfolding of various life histories
that proceed at different rates; species neither facilitate
nor inhibit other species. In this sense, Egler’s theory of A X
initial floristic composition resembles Connell and Slaty-
er’s tolerance model of succession. B Y
Life history differences do cause shifts in dominance
C Z
in some communities. Populus grandidentata (big-toothed
aspen) is an early successional species found in upland
forests with poor soils around the North American Great
X
Lakes. It is replaced by a variety of species, including
Quercus rubra (red oak). This replacement occurs because A Y
aspen grows very quickly. Its quick growth means that
its wood is not very strong and is susceptible to damage Z
by fungal attack and by high winds. Aspen trees rarely
live beyond 60–70 years. In contrast, oaks grow very Figure 13.15
slowly, producing the strong, dense wood prized by fur- Schematic of predictable and unpredictable successional tra-
jectories. (1) Early ideas about succession saw all sites in an
niture makers. Oaks may colonize an aspen stand very area as starting from approximately the same starting point
quickly, but they will not dominate for decades. In other and following a single predictable trajectory to the same end
cases, however, later dominance is clearly the result of a point. (2) Different starting points might still lead to the
species’ inability to colonize an early successional site. same end point. In this scenario, deterministic processes
For example, Tilia americana cannot colonize new sand dominate, and the outcome is predictable. (3) Different start-
ing points might lead to various unpredictable end points.
dunes. It requires other species to stabilize the dunes and (4) Even similar starting points might lead to different
begin the process of soil formation. unpredictable end points.
The Predictability of Succession

One of the most important contributions of the initial
floristic composition theory was that it called attention succession takes to its end point (Figure 13.15). The start-
to the process of colonization. While earlier theories of ing point encompasses the physical conditions follow-
succession saw colonization as an important process, ing a disturbance and the propagules that either survive
they viewed it as a deterministic one—a process with the disturbance or arrive at the site. The early ideas
a fixed outcome. Colonization, however, is a process about succession saw all sites in an area as starting from
fraught with random elements. These elements include approximately the same point and then following a sin-
the timing of seed arrival, the number of propagules that gle trajectory to the same end point. Once we recognize
arrive at a site, whether those propagules land in a safe the possibility of unpredictability, however, three other
site (a favorable spot for their germination and growth), outcomes are possible. The following descriptions
and whether climatic conditions are favorable for their assume that basic conditions (e.g., climate, soil type,
establishment when they arrive. Two identical sites slope, aspect) are similar among sites.
might experience different successional sequences due First, different starting points might still lead to the
to these chance events. Thus, the predictability of suc- same end point. While early successional communities
cession is an open question. (Recall that the use of might differ considerably in composition, and while the
chronosequences to study succession assumes that suc- exact successional sequence might differ, all sites in an
cession is predictable and repeatable.) area might eventually converge on the same type of com-
One way of thinking about the predictability of suc- munity. In this scenario, deterministic processes domi-
cession is by dividing it into two components: the start- nate over random ones in the long run. Second, different
ing point of succession, and the path (or trajectory) that starting points might lead to various end points. Given
different starting points, different successional sequences Probably the best-studied site of primary succession
might unfold. The initial floristic composition hypothe- is at Glacier Bay, Alaska (Cooper 1923). William Cooper
sis would fit under this scenario. Here, random process- set up permanent study plots there in 1916 in areas from
es dominate in setting the stage for a given successional which a glacier had retreated fewer than 40 years earli-
sequence, although deterministic processes take over er (Figure 13.16). In this system, the first colonist of bare
thereafter. Third, even similar starting points might lead ground usually is a cryptogamic crust, a thin layer of
to different end points. In this scenario, random process- mosses, fungi, various unicellular photosynthetic organ-
es dominate during the entire course of succession. isms, lichens, and photosynthetic bacteria (cyano-
Currently we have limited data indicating which of bacteria). The cryptogamic crust limits erosion and helps
these three scenarios occurs most often. The best evi- to retain moisture, dead parts of the crust provide ini-
dence that we have favors the first two scenarios (i.e., tial organic matter for soil formation, and dispersing
that deterministic processes dominate in the long run). seeds may lodge in cracks in the crust. The small shrub
A study of tropical rainforest communities in the Ama- Dryas drummondii (Rosaceae) is typically the next colonist.
zon basin of eastern Peru, for example, found that A plant with nitrogen-fixing symbionts in its roots, Dryas
species composition tended to converge among sites forms dense mats. In some regions of Glacier Bay, Alnus
over time (Terborgh et al. 1996). Similarly, in Oosting’s sinuata (alder, Betulaceae), another plant with nitrogen-
study of old-field succession in North Carolina, fields of fixing symbionts (see Chapter 4), also forms dense thick-
similar age had very similar sets of species across many ets. Not surprisingly, it appears that these early nitro-
different sites. Other factors can complicate the pathway gen fixers facilitate the later colonization of Populus
to convergence, however. In a study of succession on
mudflows resulting from the eruption of Mount St.
Helens, carried out 16 years after the eruption, an area
that was relatively close to an intact forest—a source of
many propagules—showed evidence of convergence to
the composition of that forest, while an area farther away
did not (del Moral 1998). So, dispersal rates and oppor-
tunities do affect the successional trajectory. In this case,
FPO
however, given that the entire successional process takes
several centuries, even the more distant site may con-
verge, given enough time. Ecologists will need many
more examples from many different types of commu-
nities before we can reach a final decision on this issue.
Primary Succession
There are three main reasons why primary succession
can be quite different from secondary succession. First,
in primary succession, there is generally little or no true
soil—the substrate lacks structure and any substantial
organic component. Most species are incapable of per-
sisting under these circumstances and can colonize the
site only after substantial soil development. Thus, a key
issue in many primary successional settings is how the
initially colonizing plants affect soil development. A sec-
ond characteristic of primary succession is that, at least
initially, propagules (mainly seeds and spores) that
arrive at the site have few or no interactions with resi-
dent populations, because there are none. Competition
is absent or of minor importance, and herbivory may be
drastically reduced. Third, depending on the spatial
scale of the disturbance that initiates succession, the
physical environment (e.g., temperature at the ground
Figure 13.16
surface) can be much more variable and much more A rock-strewn area of bare ground at the base of Franz Josef
extreme in primary succession than in secondary suc- Glacier in New Zealand. This area has not yet been colo-
cession. nized by any plants. (Photograph courtesy of L. Walker.)
270 Chapter 13
trichocarpa (cottonwood) and Picea sitchensis (Sitka spruce, of its size and isolation. In 1883, Krakatoa erupted so
Pinaceae), which do not fix nitrogen. Spruce is much forcefully that the explosion was heard 3500 km away.
more successful as a colonizer of alder thickets than at Only about a third of the original island remained above
earlier successional stages, no doubt partly because of the sea level after the eruption, but several new islands were
soil nitrogen, soil organic matter, and presence of myc- formed from the volcanic material ejected. Continued
orrhizae that the alder provides. Some spruce trees are volcanic activity has added to the land area. Early
inhibited by, while others are facilitated by, the presence descriptions of Krakatoa following the main eruption
of cryptogamic crusts, Dryas, and alder, as well as by say that it was essentially scoured of living things and
other spruce trees (Chapin et al. 1994). Thus, both facili- that colonists arrived slowly from distant sources. It is
tation and inhibition occur in this system. Different sites impossible to judge whether this was truly the case or
at Glacier Bay follow different successional trajectories; whether the observers failed to look for the “biological
in particular, alder does not occur in some areas, mainly legacies” that have proved to be important in studies
due to its limited dispersal ability (Chapin et al. 1994). of other regions.
Volcanic activity is important in much of the world, Recent research on the Krakatoa archipelago (Schmitt
and lava flows and ash deposits create new ground on and Whittaker 1996) suggests that limitations on dis-
which primary succession occurs. Though volcanic erup- persal may be quite important there. Shoreline species
tions can be massive disturbances, it is a mistake to think recolonized the islands very rapidly, but these were large-
that they always create large areas devoid of life. Stud- ly water-dispersed species. Species that grow farther
ies of succession following the 1980 eruption of Mount inland colonized the islands much more slowly. Wind-
St. Helens have shown that most of the initial growth dispersed species, especially grasses, composites, and
and spread of plant populations was dominated by “bio- ferns, predominated among the first inland colonists, cre-
logical legacies”—buried roots, small stands of surviv- ating grasslands. Woody plants began colonizing the
ing plants, and seeds from nearby slopes (Frenzen et al. islands somewhat later; forests dominated the islands by
1986; Franklin 1990; Halpern et al. 1990). These foci for the 1920s. The successional trajectories were similar
the spread of populations remain important in the area among the islands, but there were also differences among
today. Around the Kilauea volcano in Hawaii, lava flows islands and among sites within islands. These differences
often leave small islands of vegetation (kipukas), which seemed to be due to chance differences in dispersal and
are important sources of propagules for the colonization environmental differences (Schmitt and Whittaker 1996).
of the lava (Figure 13.17). Continued volcanic activity, resulting in the heavy dep-
The Krakatoa volcanic complex in Indonesia has osition of ash on some islands, has contributed to increas-
provided important data on primary succession because ing differences among the islands over time.
FPO
Figure 13.17
A lava field on the island of
Hawaii showing a young tree
fern Sadleria cyantheoides (amau-
mau fern, Blechnaceae) that has
colonized the bare ground. This
species is an endemic and com-
mon on relatively new lava.
(Photograph courtesy of J.
Thompson.)
Although (by definition) primary succession occurs rium may exist at the scale of the landscape, even though
in unvegetated areas, dispersal distances may not be no single community is at equilibrium. We will return to
great enough to limit the ability of particular species to this issue of hierarchies of scales when we discuss land-
colonize a disturbed area. At Mount St. Helens and scapes in more detail in Chapter 17.
Kilauea, remnants of the prior vegetation are scattered Today, plant ecologists recognize that communities
throughout the disturbed sites, where they serve as and landscapes never attain a constant, unchanging state.
important sources of propagules. Most other primary Rather, communities may attain a dynamic equilibri-
succession, such as that on new beaches and sand bars um, if an equilibrium is reached at all. Long-term climate
in rivers, occurs on smaller disturbed sites, where dis- changes (see Chapter 21), introduction or evolution of
persal distances may be short. new species, and landform changes such as erosion,
mountain building, and continental movements mean
Climax Revisited that the world is always in flux. At best, these changes
occur slowly enough that communities or landscapes are
The concept of the climax—the hypothesized static, at quasi-equilibrium. Thus, one’s view of change and
deterministic end point of succession—once dominated equilibrium depends on the scale considered.
ecologists’ thinking about succession. According to this
idea, once a community has reached the climax state, it Summary
stops changing, unless a disturbance occurs that resets
the community back to an earlier seral stage. There are In this chapter, we have explored the process of succes-
serious difficulties with this concept, which have become sion and the implications of that process for our under-
increasingly apparent over the years. standing of the nature of communities. Succession starts
The notion of a successional climax was devised with disturbance. All communities experience distur-
by Hult (1885) and developed over the next several bances. Those disturbances can be small (a limb falling
decades (Clements 1916). One of its proponents was from a tree, a badger creating a mound) or large (a for-
Frederick Clements. His ideas may have been influenced est fire, a volcano erupting). They can occur every year,
by classical Greek philosophy, a major part of the intel- or once in a millennium. They can result from a multi-
lectual tradition of his day, which emphasized idealized plicity of factors—fire, wind, rain, snow, animals, dis-
types in nature. Ecologists of the time viewed the natu- ease. After a disturbance, new species may colonize the
ral state of the community as being unchanging and dis- site. These species then interact with one another so that
turbance as an unnatural, external process. As we have the community changes over time, resulting in the
seen in this chapter, however, various kinds of distur- process of succession. Finally, if disturbances are rela-
bances are intrinsic to most communities. As plant ecol- tively infrequent, the community may reach a state of
ogists have recognized this fact in recent decades, many dynamic equilibrium. Earlier ecologists thought that a
have called into question the entire concept of the climax static end point to succession, called the climax, was the
as the end point of succession. usual state of affairs in plant communities. There are
The result was a conceptual shift to considering pat- now strong reasons to think that even a state of long-
terns of stasis and change at different spatial scales. Con- term dynamic equilibrium may be exceptional.
sider a landscape dotted with many different communi- In this chapter and the previous ones, we addressed
ties, each containing many different patches. A single a number of questions about the nature of communities.
small patch may always be in a state of flux as popula- We began by asking whether communities are entities
tions change and one species replaces another (as in van in their own right, or merely collections of populations
der Maarel’s carousel model; see Chapter 10). At this scale, that happen to co-occur. We described a number of ways
change may seem directionless and at the constant mercy in which species interact. These interactions can be direct
of unpredictable factors such as climate and herbivory. In or indirect, positive or negative. For example, competi-
contrast, the community as a whole may be undergoing tion for common nutrients is a direct, negative interac-
a slow successional process of gradual replacement of one tion. In contrast, the stabilization of a sand dune by Ely-
set of species by another. At this larger scale, disturbances mus canadensis allowing colonization by other species
may still occur every century or so, resetting the succes- is an indirect, positive interaction. These interactions
sional cycle. Finally, the entire landscape may consist of a help to shape communities. Succession occurs, in part,
mosaic of communities at different points along the suc- because of these interactions. Thus, a community is more
cessional cycle. Although each community is changing, than simply the sum of its constituent species. How
the proportion of the entire landscape at each successional those species interact is also important in shaping com-
stage remains about the same. Thus, a dynamic equilib- munity composition and structure.
272 Chapter 13
Additional Readings
Classic References Keeley, J. E., C. J. Fotheringham, and M. Morais. 1999. Reexamining fire
suppression impacts on brushland fire regimes. Science 284:1824–1832.
Cowles, H. C. 1899. The ecological relations of the vegetation on the sand
dunes of Lake Michigan. Bot. Gaz. 27:95–117, 167–202, 281–308, 361–391. Phillips, D. L. and D. J. Shure. 1990. Patch-size effects on early succes-
sion in southern Appalachian Forests. Ecology 71:204–212.
Cooper, W. S. 1923. The recent ecological history of Glacier Bay, Alaska.
II. The present vegetation cycle. Ecology 4:223–246.
Oosting, H. J. 1942. An ecological analysis of the plant communities of Additional Resources
Piedmont, North Carolina. Am. Midl. Nat. 28:1–126. Johnson, E. A. 1992. Fire and Vegetation Dynamics. Cambridge Universi-
ty Press, Cambridge, UK.
Contemporary Research Pickett, S. T. A., and P. S. E. White. 1985. The Ecology of Natural Distur-
bance and Patch Dynamics. Academic Press, Orlando, FL.
Chapin, F. S., L. R. Walker, C. L. Fastie, and L. C. Sharman. 1994. Mecha-
nisms of primary succession following deglaciation at Glacier Bay, Alas- Pyne, S. J., P. L. Andrews, and R. D. Laven. 1996. Introduction to Wild-
ka. Ecol. Monogr. 64:149–175. land Fire, 2nd ed. Wiley, New York.
Halpern, C. B., P. M. Frenzen, J. E. Means, and J. F. Franklin. 1990. Plant
succession in areas of scorched and blown-down forest after the 1980
eruption of Mount St. Helens, Washington. J. Veg. Sci. 1:181–195.
14
C H A P T E R
Local Abundance, Diversity,
and Rarity
I
In this chapter we address a series of questions that raise fundamental
issues in ecology as well as having important implications for applied
ecology and conservation. What determines the relative abundances of
the species in a community? Are abundant species competitively superior to
less common species, or are they more abundant for other reasons? Why are
some species rare, others common, and still others invasive? Our focus in this
chapter is mainly at the scale of communities; we return to some of the same
issues at larger scales in Chapter 20.
The abundances of individuals within species are only part of the story,
however. G. Evelyn Hutchinson asked a deceptively simple question in the
title of his 1959 paper, “Homage to Santa Rosalia, or why are there so many
kinds of animals?” Ecologists are still probing the implications of this general
question. What determines the number of species that can coexist in a com-
munity? The number of species in a community might be determined largely
by deterministic processes such as competition, or by chance processes, such
as what species are able to reach the community. How do species interactions
affect the ability of a new species to become integrated into the community?
It may be that biotic and abiotic factors make a plant community more sus-
ceptible or more resistant to invasion by exotic species.
Much effort has gone into looking at how processes that determine local
abundances play out within a single community. More recently, there has been
a growing recognition that landscape and regional processes can also be very
important in determining the relative numbers of individuals per species as
well as the number of species in communities. In this chapter we examine the
interaction between local and regional processes from a local perspective. In
Chapter 17 we will look at landscape-level processes that determine popula-
tion dynamics and species distributions, and in Chapter 20 we will consider
abundance and diversity at regional and global scales.
Dominance
Even a cursory walk through any plant community will reveal that not all
species are equally abundant. Typically, at least in temperate communities,
most of the individuals (or biomass) in a community will belong to one or a
few species, while many other species will be rare. The most conspicuous
and abundant species (numerically or in terms of biomass) in a plant com-
munity are called the dominants. Dominant species are in the process of succession and by allowing other
often used to characterize plant communities: a cattail species to increase in abundance much more than they
marsh, a pine barren, an oak-hickory forest, or a saguaro did in nonremoval plots. This finding implies that the
“forest” are all characterized by their most conspicu- goldenrod was not only numerically dominant, but also
ous or abundant plants. competitively superior.
Other researchers have also reported community
Are Dominant Species Competitively Superior? responses to the removal of dominant species. For exam-
The term “dominant” implies competitive superiority. ple, Kutiel et al. (2000) found that removal of dominant
Are numerically dominant species really competitively shrubs on coastal dunes in Israel led to greater abun-
dominant as well? While plant ecologists have general- dance of annual species and therefore higher diversity.
ly assumed that this is the case, few have directly test- Removal of the C3 grass Lolium perenne (English rye
ed that assumption. We have already talked about grass, Poaceae) in a New Zealand grassland resulted in
differing views on the consistency of competitive hier- increased species diversity, but removal of C4 grasses did
archies in Chapter 10. Here we examine the extent to not have that effect, presumably because no other
which numerical abundance is determined by compet- species in the community had a phenology similar to
itive ability. Random processes, such as which species that of the C4 grasses (Wardle et al. 1999).
happen to occupy a site first, can also contribute to the
pattern of species abundances at a site. The current pat- Abundance Curves
tern of abundances in a community may also be tem- One way of describing hierarchies of relative abundance
porary. For example, if a site is undergoing succession is the use of abundance curves (Figure 14.1; also known
following a disturbance (see Chapter 13), one species as dominance-diversity curves). Abundance curves offer
may be more abundant than another because it tends an alternative to diversity indices (see Chapter 12) for
to dominate during early succession, but may be dis- expressing the number and relative abundances of all
placed as time passes. the species in a community (or sample, or stand).
Timothy Howard and Deborah Goldberg (2001) Because they are a graphical presentation rather than an
used field and garden experiments on herbaceous peren- attempt at condensing everything into a single num-
nials in a Michigan old field to test whether rankings of ber, they avoid some of the limitations of diversity
competitive ability were related to observed species indices. Abundance curves can provide an excellent
abundances in the field. Competitive ability at the stage graphical contrast among different kinds of community
of seedling germination and early growth was strongly structures—a curve representing a community with a
related to species abundance. The effects of adult com- high degree of dominance by a single species and low
petitors on germination and seedling growth were small species richness will have a very different shape from
for the most abundant species, Poa compressa (bluegrass, one representing a highly diverse community without
Poaceae), and large for the least abundant species, Achil- strong dominance, with many common species, and
lea millefolium (yarrow, Asteraceae). In contrast, the com- with a high total species number. However, these curves
petitive ability or survivorship of adult plants had only are more useful for descriptive purposes than for quan-
a limited relationship to species abundance. Other titative comparisons.
researchers who have examined the effects of competi- During the 1960s and 1970s it was suggested that
tive ability at single life history stages (mostly early abundance curves could serve as an indicator of the
seedling growth) have generally reported positive rela- processes responsible for community structure, but this
tionships between abundance and competitive ranking. effort proved disappointing. Robert MacArthur (1960)
Another approach to testing whether numerical proposed that the shape of a curve was created by the
dominants are really competitive dominants is to exper- way in which species divided up the available energy in
imentally remove the dominant species from a com- a system. Each species was hypothesized to monopolize
munity and observe the effects on community structure. a random fraction of that energy. This theory is referred
Gurevitch (unpublished data) found that removing the to as the broken-stick model because the resulting pat-
dominant species from an old-field community had dif- tern is the same as would be obtained if one took a stick
ferent effects depending on the species removed. Re- (representing the available energy) and broke it into
moval of one of the dominants, Dactylis glomerata pieces at random (Figure 14.1, curve A).
(orchard grass, Poaceae), had little influence on other Others demonstrated, however, that one could
species, but removal of another, Solidago rugosa (wrin- obtain an identical pattern through a strictly determin-
kle-leaved goldenrod, Asteraceae), had dramatic effects. istic process. Depending on the assumptions used,
Removal of S. rugosa from a plot resulted in an increase curves of any shape can be generated, but such assump-
in species richness by permitting the persistence of tions are extremely difficult to verify. Assume, for exam-
species that had disappeared elsewhere in the old field ple, that the most dominant species can preempt 50% of
PAGE PROOF: 2ND PASS Local Abundance, Diversity, and Rarity 275
given dominance pattern by examining the shape of the

Theoretical distributions
abundance curve for a given community. Instead, ecol-
100 Broken-stick model (A) ogists have shifted their attention to working out the
Geometric series (B) mechanisms causing dominance patterns within differ-
Lognormal distribution (C)
ent communities.
Data plots
10 Subalpine fir forest (D)
Deciduous cove forest (E)
Rarity and Commonness
A Species differ greatly not only in how abundant they are
Relative importance (%)
in any one place, but in the number of places in which

1.0 they are found. Some species have a wide geographic
distribution, while others are endemic (found in only
one place). All species are rare when they first evolve;
some go on to become common, while others remain
0.1 B rare. An understanding of rarity and the processes
D
responsible for it is central to the preservation of endan-
C gered and threatened species. Endangered and threat-
E
ened species are defined as being rare, although not nec-
0.01 essarily globally rare. A species can be rare and listed
as endangered in one state of the United States while
being common in another state. In the United States and
many other countries, the official listing of a species as
0.001 endangered triggers a host of laws and regulations con-
0 10 20 30 40
Most common Least common cerning how the species and its habitat must be pro-
Species rank tected. Thus, the topic of rarity is fraught with legal,
Figure 14.1 political, and economic consequences.
Examples of both theoretical and empirical abundance
curves. The vertical axis is the relative abundance of a The Nature of Rarity
species in a community—that is, the percentage of the com-
munity that it represents. Besides numbers of individuals, While a plant species may be rare, common, or invasive
the relative abundance could be measured in terms of bio- (see below), these categories are not necessarily fixed
mass, cover, frequency, or a combination of measures. As is characteristics of the species. A remarkable example is
often the case, the vertical axis is shown at a log scale. The Pinus radiata (Monterey or radiata pine, Pinaceae) (Lav-
horizontal axis arrays the species of the community in rank ery and Mead 1998). Once a rare endemic limited to a
order, from most common to least common. The most com-
mon species has rank 1 (left side), the next most common narrow strip of the California coast, Monterey pine has
species has rank 2, and so forth. The green curves represent escaped from timber plantations in the Southern Hemi-
three theoretical distributions. The black and gray curves sphere to become a major invasive pest in a number of
represent two sets of empirical data for vascular plant com- countries. In its native habitat, the trees are small-
munities in the Great Smoky Mountains of Tennessee—sub- statured and gnarled and, thus, are not used as timber.
alpine fir forests and deciduous cove forests. (After Whittak-
er 1975.) In plantations in the Southern Hemisphere, however,
Monterey pine grows rapidly into large, straight trees.
Chile, South Africa, New Zealand, and Australia all now
have major naturalized populations of this species, and
the available energy in the system, that the next most its geographic ranges and population sizes in these
dominant species can preempt 50% of the energy that countries vastly exceed those in its native habitat. We
remains, and so forth. This deterministic process would will return to the factors that may make a species inva-
produce a geometric series (Figure 14.1, curve B). On the sive after taking a look at some ecological aspects of rar-
other hand, if dominance were determined by many ity and commonness.
independent factors that affected different species in dif- Deborah Rabinowitz (1981) identified three differ-
ferent ways, the curve would be lognormal (Figure 14.1, ent aspects of species distributions: geographic range
curve C). (wide or narrow), habitat specificity (broad or restrict-
The problem with deciding among these alterna- ed), and local abundance (somewhere large or every-
tives (random division, preemption, or many factors) where small; Table 14.1). These three aspects result in
is that all three models (broken stick, geometric, and log- eight possible combinations of characteristics, and a
normal) could be fit to any given set of data. Thus, it is given species could display any one of those combina-
not possible to decide which process is responsible for a tions. Only one of the eight possible combinations—
Table 14.1 A scheme for describing eight categories of commonness and rarity
based on three traits
Geographic range Wide Narrow
Habitat specificity Broad Restricted Broad Restricted
Somewhere Common Predictable Unlikely Endemic
abundance
large (habitat specialists)

Local
Everywhere Sparse Unlikely Rare on all

small counts
Source: Rabinowitz 1981.
wide range, broad habitat specificity, and large local In a similar fashion, species that can live in a wide
abundance somewhere—is classified as commonness. variety of habitat types tend to have large geographic
The other seven each constitute some form of rarity. ranges because in any given area there is likely to be
Commonness and rarity are the result of processes some suitable habitat. While these relationships are the
that operate at local, regional, and even global scales. typical ones, exceptions exist. Some species are habitat
Local differences in species abundances may or may not specialists with small geographic ranges, yet dominate
be reflected at larger scales, and these patterns may the communities they are found in. An example is the
change through time. Plants that were once common mint Mentha cervina (Lamiaceae), a specialist of montane
may decline in numbers, while rare species may, over wetlands of the eastern Mediterranean. While it is found
time, become common or even dominant, particularly in a very limited range of sites in central Spain, where
in new or changed environments. Local and regional it is found, it exists at high densities.
patterns may be uncoupled or may be
linked in a variety of ways. By chance alone,
a species may be introduced to a new site to 5
which it is well adapted and may become
highly abundant, as the Monterey pine has
done. From there, it may spread and
become abundant everywhere in a broad 4
region. In this case, events at a local level
spread to a regional scale. Conversely, Mentha cervina
processes at a regional scale (such as gene
Mean local abundance
flow among sites; see Chapter 5) may result

3
in a species becoming adapted to varying
conditions across a wide geographic range.
A high average level of adaptation, coupled
with metapopulation processes (see Chap-
ter 17), may result in the species reaching 2
high abundances everywhere it occurs. In
this manner, regional-scale processes can
determine local-scale patterns.
1
Patterns of Rarity and Commonness
While all possible combinations of local and
regional patterns of commonness and rari-
ty are conceivable, what combinations do
0 10 20 30 40 50
we actually find in nature? There is often a
Range
positive relationship between abundance
and geographic range (Figure 14.2): species Figure 14.2
Correlation of mean local abundance and geographic range for vascular
with large population sizes tend to be wide- plants in montane wet meadows in Spain (rS = 0.17, P < 0.01). Mentha
spread, while species with small geograph- cervina is an exception to the general trend that species with the largest
ic ranges tend to have small population geographic ranges also have the highest abundances where they are pres-
sizes. ent. (Data from Rey Benayas et al. 1999.)
Table 14.2 Percentages of species falling into each of eight commonness and rari-
ty categories in three studies
(A) Vascular plants of the British Isles (n = 160)

Local abundance large 36 44 4 9
Local abundance small 1 4 0 2
(B) Vascular plants of montane wet meadows in Spain (n = 220)

(C) Trees of a tropical rainforest in Manú National Park, Peru (n = 381)

Sources: (A) Rabinowitz et al. 1986 ; (B) Rey Benayas et al. 1999 ; (C) Pitman et al. 1999.
In an analysis of the entire native British flora, Rabi- lefs 2000). These results may have important implica-
nowitz and colleagues (Rabinowitz et al. 1986) found tions for the conservation of tropical biodiversity
that 36% of species were common, with wide geographic because if most species are widespread, it will be easier
ranges and large local abundances; almost 60% were to find land to preserve them.
restricted to particular habitats, while 15% had small We can also look at rarity from a taxonomic per-
geographic ranges and 7% had small maximum popu- spective. A study of rarity in the vascular plants of the
lation sizes (Table 14.2A). This study was the first to ana- United States and Canada found that taxonomic groups
lyze these patterns and is often used as a basis for gen- (genus, family, order, and class) with many species also
eralizations about rarity. However, other researchers tended to have higher than average numbers of rare
have found that these patterns vary depending on how species, while smaller groups had fewer rare species
one looks at the flora. A study by José Rey Benayas and than expected (Schwartz and Simberloff 2001). This pat-
colleagues considered the entire set of vascular plant tern would occur if large groups were more likely to pro-
species inhabiting a specialized habitat type, wet mead- duce new species through allopatric speciation (see
ows in the mountains of central Spain (Table 14.2B). In Chapter 6), in which case, on average, the new species
this instance, the most frequent category was species would be isolated, have small range sizes, and hence be
with narrow geographic ranges, restricted habitat speci- more likely to be rare. Conversely, groups with high
ficities, and small local abundances. In another study, rates of extinction might lose their rare species first, so
Nigel Pitman and colleagues (Pitman et al. 1999) exam- that small groups would be left with common species
ined tree species in a highly diverse tropical rainforest only.
in Manú National Park and the surrounding region in
eastern Peru (Table 14.2C). Most of the tree species in Causes of Rarity and Commonness
this study were locally common in at least one plot and Why is a particular species common or rare? The pat-
were found across a wide range of habitats (fewer than terns described above suggest that ecological special-
25% were habitat specialists), and all had broad geo- ization may be an important cause of rarity. Species that
graphic distributions. This pattern contrasts strongly are ecological specialists tend to have low abundances
with the Rabinowitz et al. study of the British flora, in and small geographic ranges. This observation raises
which many more species were habitat specialists (Rick- another question: why are there both ecological special-
ists and ecological generalists? The traditional explana- the most frequent class of species in any given landscape
tion is that there are trade-offs that result in common- was those found at only a single site, especially in mid-
ness or rarity. By specializing, according to this theory, dle- and low-latitude areas.
a species can better exploit a particular niche. The theo- In contrast, Mark Burgman (1989), in a test using the
ry predicts that local abundance should be negatively flora of the mallee vegetation of western Australia, did
correlated with the number of habitats occupied because not confirm the major predictions of Brown’s theory.
a specialist should do very well where it is found, but be This region has a mediterranean climate and highly vari-
found in only a few places. However, this prediction is able soil conditions. The vegetation of the region is char-
contrary to what is typically observed (see Figure 14.2). acterized by high species richness, with many endemic
The “environmental control” or “superior organ- species, and is dominated by low shrubs and herbaceous
ism” theory was proposed and later elaborated by James perennials. Burgman’s analysis strongly refuted Brown’s
Brown (1984, 1995) to account for differences among “environmental control” theory. It found no support for
species in abundance. Brown began with the premise the hypothesis that rare species have smaller funda-
that species vary in their abilities to exploit nature. This mental niches than do widespread, abundant species.
premise is built on Hutchinson’s (1957) concept of the The low abundances of the rare plants in this study were
multidimensional niche. A niche is the set of all of the not caused by limited environmental tolerances, nor was
conditions and resources that determine the ability of the high abundance of common plants attributable to
the individuals of a species to survive and reproduce. A wide environmental tolerances. Burgman explained
fundamental niche is the set of conditions and resources these results by saying that even if a species can tolerate
that the species could use in the absence of competitors a broad range of environmental conditions, if those
(or, by extension, predators, pathogens, and parasites). combinations of conditions and resources are found in
Brown assumed that some species have broad fun- only limited geographic areas or patches, those species
damental niches, while others have narrow ones. The will be limited as well.
center of a species’ geographic range, he proposed, is the The flaw in Brown’s theory, therefore, could be the
place where it can exploit the widest range of resource assumption that the range of variation in environmen-
combinations and, therefore, the widest number of habi- tal conditions is positively related to the geographic
tats. In this region it should be at high abundance. As the extent of the range. Imagine, for example, that a plant
distance from this center increases, the conditions favor- species can tolerate a wide range of soil textures, from
ing the species become less common, because, from its sandy to loamy soils, but that in the environment in
perspective, environmental conditions are deteriorating which it is found, there are no sandy soils, a few patch-
(since environmental variables tend to be correlated with es with loamy soils, and large areas with clay-loam and
one another). Therefore, the species’ local habitat speci- clay soils. Then, even though the species could do well
ficity narrows and its abundance decreases. Eventually, on many different soil types, in that environment it will
the edge of its geographic range is reached. Those be restricted to a small number of patches and will be
species that begin with a broad niche will have a large rare. The rare species in Burgman’s study appeared to
geographic range, while those that begin with narrow be rare because, although they could tolerate a wide
niches—that is, with restricted environmental toler- range of environmental conditions, those conditions
ances—will be rare and have a small geographic range. were found at few sites. In contrast, the conditions tol-
This theory has similarities to the centrifugal theory of erated by the common and widespread species were
community organization (see Chapter 10). found in many places in Western Australia.
A major problem with Brown’s theory is that it does Gaston (1994) and Kunin (1997) list a number of
not account for the persistence of inferior species. hypotheses proposed to explain what makes a species
According to the theory, the superior performers should rare. They suggest that the explanations for the imme-
simply overwhelm the inferior ones, eventually replac- diate causes of rarity fall into three categories: ecologi-
ing them. In all likelihood, both mechanisms (the exis- cal specialization, lack of dispersal, and historical acci-
tence of trade-offs and the superior adaptations of some dent. While various explanations have been proposed
species) play a role in determining patterns of com- for why species become ecologically specialized (Rosen-
monness and rarity. zweig and Lomolino 1997; see also other chapters in
Evidence relating to Brown’s theory is mixed. An Kunin and Gaston 1997), the causes of specialization are
analysis of the abundance and occurrence of plant not well understood, and the evidence connecting spe-
species in 74 landscapes across the globe found support cialization with rarity is limited.
for his theory (Scheiner and Rey Benayas 1994). Two pre- Colleen Kelly (1996; Kelly and Woodward 1996)
dictions of the theory were confirmed: first, there was a searched for ecological correlates of commonness and
positive correlation between the number of patches a rarity in the floras of Great Britain and Crete. She asked
species occupied and its local abundance, and second, whether there are particular attributes of species that give
them wide geographic ranges. She looked at a number Invasiveness is related to, but not exactly the same
of attributes, including growth form, pollination syn- as, being weedy or ruderal. Weeds are generally defined
drome, and dispersal mechanism. Only growth form as uncultivated species that proliferate in agricultural set-
(trees were more common than shrubs) and pollination tings, interfering with crop production, although the term
syndrome (wind-pollinated plants were more common has been used in a broader context on occasion. Weeds
than animal-pollinated plants) were found to explain are often exotic species, but may sometimes be natives
geographic range. that thrive in the conditions of high nutrient and light
availability and frequent soil disturbance typical of agri-
Invasive Species and cultural fields. Ruderal species possess traits that allow
them to do well in temporary habitats, including agri-
Community Susceptibility cultural fields and roadsides, as well as areas subject to
Ecologists and environmentalists have become alarmed frequent disturbances such as rockslides or flash floods.
at the tremendous spread of invasive species in the past The word “ruderal” derives from “rude,” in the sense
few decades because their spread is often associated of “wild” or “rough.” While “weedy” and “ruderal” are
with the displacement of native species. In some cases often used synonymously, the former is typically used in
invasive species have drastically altered community and agriculture and the latter is used in ecology.
ecosystem patterns. Many definitions for invasive Ecologists are concerned with a number of questions
species have been proposed, but the term generally regarding invasiveness. In this section we ask what
refers to species that are rapidly expanding outside of makes species successful invaders and what factors
their native range. Invasive species may be native to the make a community susceptible or resistant to invasion
general region, but usually are exotics that are far from by exotic species. In Chapter 22 we return to this topic
their native habitats, often from other continents. In most to look at the ecological effects of invasive species.
cases they have not dispersed naturally, but have been
brought either purposely or inadvertently to a new habi- Why Do Some Species Become Invasive?
tat by people. A large proportion of the plants that have Ecologists have struggled to predict which plants might
become seriously invasive were deliberately introduced, become invasive for many years. Charles Elton (1958)
planted, and cultivated by people (Figure 14.3). Between and Herbert Baker and Ledyard Stebbins (1965) attacked
2000 and 3000 exotic plant species have been introduced this question four decades ago, tabulating the charac-
into the United States, most within the last 100 years. teristics and attributes of species known to be success-
(A) (B)
Atlantic Ocean
1830s
1900
1940
1985
Isolated
Gulf of Mexico records–1985
Figure 14.3
(A) Lythrum salicaria (purple loosestrife, Lythraceae) is an invasive wetland species
that now dominates many marshes and lakes in the northeastern United States.
(Photograph courtesy of P. O’Neil.) (B) Purple loosestrife spread across North
America over a 150-year period. It may have been introduced into the United States
in the 1700s, but it did not begin spreading for a half-century or more. (After
Thompson et al. 1987.)
ful invaders. But the attempt to successfully predict inva- other plants, allowing invaders to become established
siveness has been a frustrating exercise. Different plant and multiply. “Empty niches” may exist in the new com-
invaders have contrasting strategies: for example, one munities in which an invader may be able to proliferate.
might have a very high seed output, while another An invasive shrub, for example, might succeed in a for-
invests in extensive rhizomatous spread. Most of the est that has no native shrubs. Finally, invaders may alter
data addressing this question are qualitative observa- ecosystem characteristics in ways that favor their own
tions of unrelated organisms; such observations are subsequent increase (see Chapter 22).
difficult to compare or synthesize across studies. Fur- Williamson and Fitter (1996) proposed the “tens
thermore, data on failed invasions are, of course, un- rule” in an attempt to make (and explain) statistical gen-
available, leaving us with only half the picture. eralizations about the success of invaders. The tens rule
Recently, however, there has been progress in under- states that 1 in 10 of the plant and animal species brought
standing what might make some plant species likely to into a region will escape to appear in the wild, 1 in 10
become successful invaders. Rejmánek and Richardson of those escaped species will become naturalized as a
(1996) quantified the traits of 24 species within a single self-sustaining population without cultivation, and of
genus—the pines—12 that are highly invasive and 12 those naturalized species, 1 in 10 will become invasive.
that have been widely planted worldwide but have Rules of this sort are appealing, but we need to know
never become invasive. Because the pines are evolu- why (and how well) they hold and when to expect them
tionarily closely related and are ecologically and physi- to fail. Many invasions seem to follow the tens rule, but
ologically similar in many respects, this is a powerful there are significant exceptions, suggesting that it should
approach for isolating those characteristics that might be applied with caution.
be closely linked to invasiveness. The researchers ana- The process by which a species makes the transi-
lyzed ten life history traits, and found that three were tion from its first introduction to invasive status con-
most closely related to invasiveness: reproduction at sists of several phases (Mack et al. 2000). Many invasive
younger ages, smaller seeds, and large seed crops pro- species have a long lag phase (Figure 14.4) during which
duced at shorter intervals. Rejmánek and Richardson their numbers are low and they are not particularly
then applied this information to 34 other pine species noticeable. This lag phase may be brief or may last well
that were not in the original data set, but whose inva- over a century. Because most introduced species disap-
sive status was known, and accurately predicted which pear during this lag phase (see the tens rule above), it is
of those species were invasive. They also offered a ten- difficult to tell during this period which introduced
tative set of rules for predicting woody plant invasive- species are headed for extinction and which for inva-
ness (Table 14.3). Mack et al. (2000) discuss more gener- sive status.
ally the (usually inadequate) attempts of ecologists to Populations of successful invaders increase during
predict invasiveness. this period even though their total numbers are still low.
Plant invaders may have the ability to grow rapid- Many invasive plant species are introduced not once, but
ly in a new environment for several reasons. They may multiple times, and at multiple sites, during the lag
escape from herbivores and parasites that control them phase. At some point, however, the population begins to
in their native environments (see Chapters 7 and 11). grow rapidly and at an accelerating pace in both num-
Human disturbances may reduce competition from bers and area occupied—at this point, the species has
Table 14.3 Tentative rules for predicting which woody seed plants are likely to
become invasive
Opportunities for vertebrate dispersal
Z* Type Absent Present
>0 Dry fruits and Likely invasive Very likely invasive
seed mass > 2 mg
Dry fruits and Likely invasive in
seed mass < 2 mg wet habitats
Fleshy fruits Unlikely invasive Very likely invasive
<0 All Noninvasive unless Possibly invasive
dispersed by water
Source: Rejmánek and Richardson 1996.
*Z = 19.77 – 0.15√M – 3.14√J – 1.21S, where M = mean seed mass (mg), J = minimum juvenile period
(years), and S = mean interval between large seed crops (years). Z is thus a function that integrates
several life history traits.
8 Figure 14.4
Some outcomes for a spatially explicit simula-
tion model of the spread of an invasive species
in a new environment. In this model, about
Spread ( area)
half the time, the rate of spread is slow at first,

then becomes more rapid as the invading
species begins to more fully occupy the new
habitat. This pattern corresponds to the lag
phase seen in many actual plant invasions that
have been documented, in which the rate of
spread is unnoticeable at first, sometimes for
0 decades to hundreds of years, but at some
Initial Final I F I F I F point becomes very rapid. (Model results
(I) (F) taken from Hastings 1996.)
Time
8
Spread ( area)
0
I F I F I F I F
Time
become invasive. Ultimately the invader reaches new different effects, remain very limited (Lodge 1993;
geographic and ecological limits at which its popula- Rejmánek 1989). In examining the role of disturbance in
tion stabilizes. promoting or inhibiting invasion, elucidating the effects
of specific factors (such as soil disturbance, logging, or
What Makes a Community Susceptible to Invasion? fire) is a much more satisfactory approach than lumping
Various predictions have been made regarding the fac- those factors together into a general category.
tors that make communities vulnerable to invasion. Likewise, evidence as to whether species-poor com-
Some kinds of communities appear to be highly sus- munities are indeed more vulnerable to invasion is lim-
ceptible to invasion by exotic species, while others ited and contradictory. In a recent review, Levine and
appear better able to resist invasion. In his classic study D’Antonio (1999) evaluated data and theory linking
reviewing invasions by plants and animals, Elton (1958) diversity and biological invasions. They found that
hypothesized that invaders were most likely to become while both classic and recent theory predicted that more
established in disturbed and species-poor communities. diverse communities would be more resistant to inva-
While most ecologists continue to believe that dis- sion (e.g., Elton 1958; Tilman 1999), data from experi-
turbance is a critical factor in promoting invasion, there mental and descriptive studies were ambiguous. Some
is conflicting evidence on its role, and its effects may dif- studies supported this predicted relationship and oth-
fer among communities (Lodge 1993). In a native grass- ers refuted it.
land community in Kansas, for example, Smith and Two recent studies in grasslands showed negative
Knapp (1999) showed that increased disturbance (annu- relationships between vulnerability to invasion and both
al burning) strongly decreased invasion by exotic species and functional group diversity (Tilman et al.
species. 1997; Symstad 2000). Other recent studies, however,
Furthermore, disturbance clearly encompasses more have shown the opposite pattern. Thomas Stohlgren and
than a single factor; it may reflect changes in factors as his colleagues (1999) studied the relationship between
diverse as soil surface microclimates, predator-prey rela- plant species richness and invasions in the Colorado
tionships, resource availability, and competitive inter- Rockies and in the Central Great Plains of western North
actions (Orians 1986). Rigorous data appropriate for test- America across a wide range of plant community types
ing the importance of disturbance, or of its many (including forests, meadows, and grasslands) and over
(A) may be particularly important in systems on soils with

20 low soil nutrient resources, such as soils on glacial out-
Central grassland sites
Tallgrass wash (e.g., Long Island, New York) and very old soils
Northern
Exotic species richness
prairie (e.g., Australia). A study of Long Island forest commu-

15 mixed
prairie nities found that soil characteristics are critical in the facil-
itation or inhibition of invasion by exotic species (Howard
10 et al., unpublished). There was a strong positive rela-
Mixed tionship between both nitrogen and calcium levels and
prairie plant invasions, and a weaker but also positive relation-
5
ship between phosphorus levels and invasions. This
Shortgrass steppe study also found a strong positive relationship between
0 native species diversity in forests and the degree to which
30 35 40 45 50 55 60 65
those forests had been invaded by exotic species.
Native species richness
These results echo the pioneering experimental work
(B) of Hobbs (1989) in Australia, which showed that distur-
12 Colorado Rockies
bance alone did not make a plant community suscepti-
11 ble to invasion, but that disturbance coupled with the
Aspen addition of soil nutrients greatly facilitated invasion.
Exotic species richness
10
9 Wet
Clearly, more information is needed before we can defin-
8 meadow itively answer why plant communities differ so much in
7
their likelihood of becoming invaded by exotic species.
Ponderosa
6
pine
5 Dry meadow Abundance and Community Structure
4
Lodgepole pine
As we have just seen, species can be rare or common
3
within a community. They may be long-term or short-
2
60 70 80 90 100 110 120 130 term dominants, or they may be invading the commu-
Native species richness nity. If we shift our focus from individual species to the
entire plant community, we recognize that communities
Figure 14.5
The relationship between native species richness and exotic may differ greatly in the number of species present and
species richness in the central Great Plains grasslands (A) in the degree to which individual species are dominant.
and in the Colorado Rockies (B). The results are based on In other words, there are great differences in species rich-
lists of species in 1000-square-meter plots of each vegetation ness and diversity among communities. Total biomass
type. Results of linear regressions (grasslands: r2 = 0.85, also differs greatly among communities, as does pro-
P = 0.078; Rockies, r2 = 0.81, P = 0.038). (After Stohlgren et al.
1999.) ductivity—the rate of energy flow through a trophic
level (see Chapter 15). What are some of the causes of
these contrasts among communities? Are variations in
these different aspects of communities causally related?
a range of spatial scales. They concluded that exotic Various hypotheses have been proposed to explain
species primarily invaded areas of high species richness the observed relationships between relative species
(Figure 14.5). In a survey of 184 sites distributed glob- abundances and the productivity and diversity of com-
ally, Lonsdale (1999) also found a positive relationship munities, and we briefly discuss some of those hypothe-
(on a landscape scale) between the degree of invasion ses here. We examined some of the causes of species
and native species richness. diversity in Chapters 10, 11, and 13. Hypotheses explain-
The same positive relationship between native ing differences in species diversity at local and regional
species richness and invasion by exotics holds in sys- scales have been reviewed by Rosenzweig (1992), Hus-
tems as diverse as the South African fynbos (Higgins et ton (1994), Ricklefs and Schluter (1993), and Tilman
al. 1999), New Zealand mountain beech forests (Wiser et (1982). We will examine patterns and possible explana-
al. 1998), and Great Plains grasslands (Smith and Knapp tions for differences in species diversity at larger spa-
1999). In a review of plant invasions globally, Levine tial scales in Part V.
(2000) states that, in general, the most diverse plant com-
munities are the most likely to be invaded. Productivity and Diversity
Variation in resource availability has been proposed Productivity can affect community structure in a variety
as an underlying cause for the positive association of ways. If rates of herbivory and decomposition differ
between native and invasive species diversity. This factor little between communities, a community with higher
productivity will have a greater total biomass, or stand- example, rather than a greater number of individuals,
ing crop (the total living biomass in an area): individual there might be no change in the number of species, or it
plants will be larger, or there will be more of them, or might even decrease if the larger individuals crowded
both. In most terrestrial ecosystems, this strong positive out smaller ones. Such declines in individual density and
correlation between standing crop and productivity sets species richness are typically seen after fertilization in old
the stage for many important processes that result in fields and other herbaceous communities (Figure 14.6).
structural differences among communities.
Differences in productivity can have dramatic effects
(A)
on physiognomy and community structure. If higher
productivity, for example, gives rise to larger individual 16
plants, the community may have trees instead of shrubs,
or shrubs instead of herbaceous perennials. We will
examine these differences in more detail in Chapters 18 a
12
Species richness (No./m2)

and 19 when we look at global vegetation patterns. More a
subtly, higher levels of productivity can lead to larger, b
more vigorous individuals within a species, or to a
change in the particular species with a given growth 8
form occupying a site (for example, a large tree such as
eastern white pine, Pinus strobus, instead of a small tree c
such as jack pine, P. banksiana, or even the replacement
4
of pine species with hardwoods).
Changes in productivity can also result in changes
in the number and diversity of species in a community.
The nature and causes of the relationship between pro- 0
ductivity and species diversity have been disputed by 0 8 16 24 32
ecologists for many years (we visit the converse ques- NPK fertilizer (g N/m2)
tion of the effect of increased diversity on productivity (B)
later in this chapter). The nature and shape of the rela-
50
tionship between productivity and diversity in plant
communities depends on spatial scale. Here we look at
the nature of this relationship at local to regional scales.
Stem density (1000 stems/m2)
a a
In Chapter 10 we saw how competition affects this rela- a
tionship, while in Chapter 20 we examine the patterns
observed at larger spatial scales. 10
Preston (1962) argued that when the total area (or,
by extension, the resources) available to support species
in a given region is small, then the numbers of individ- b b
uals of the rarest species will be too low to maintain the c
b, c
existence of that species in that region. Therefore, as
the available area (or resources) increases, the number
of species will increase. 1
Wright and his colleagues (reviewed by Wright et al.
1993) extended Preston’s theory to explicitly predict that 0.5
0 8 16 24 32
an increase in productivity should lead to an increase in
NPK fertilizer (g N/m2)
species diversity. They began by assuming that greater
productivity reflects a greater amount of energy avail- Figure 14.6
able to the system, resulting in a greater number of indi- The results of experimental fertilization of 72 plots in an old
field in northwestern Pennsylvania. A slow-release NPK fer-
viduals at a site. If a site can support more individuals, tilizer was applied in May and June, and all aboveground
then more species can each have the minimum number biomass was harvested and measured in September. (A)
of individuals necessary to prevent their extinction. Species richness declined by about 50% at the highest level
Wright et al. allowed, however, that the shape of the rela- of fertilizer treatment. (B) Stem density declined nearly 80%,
tionship between productivity and diversity could be with the greatest decline in the experimental block with the
highest density at the lowest level of fertilizer treatment.
unimodal (a hump-shaped curve), or even negative, Means (bars are ± standard error) with different letters are
under certain circumstances or at smaller scales. If significantly different (P < 0.05). (After Stevens and Carson
increased productivity resulted in larger individuals, for 1999.)
(A) (B) Figure 14.7

45 36
The relationship between pro-
ductivity and species richness
40 can have different shapes in
32
Number of species
Number of species
35 different communities. All
30 28 relationships are statistically
significant. (A) Aboveground
25
24 yearly productivity versus
20 total number of tree, shrub,
15 20 and herb species in a decidu-
10 ous forest in the Great Smoky
16 Mountains. (Data from Whit-
5
500 1000 1500 2000 2500 100 120 140 160 180 200 220 240 260 taker 1966.) (B) Aboveground
Productivity (g C/m2/yr) Productivity (g C/m2/yr) yearly productivity versus
total number of herb species in
a grassland in western North
(C) (D) Dakota. (Data from Redmann
1975.) (C) Standing crop versus
70 total number of ground-layer
50
60 species in grasslands and
Number of species
Number of species
40 deciduous forests in Estonia.

50
(Data from Zobel and Liira
40 30 1997.) (D) Standing crop ver-
sus total number of herb
30
20 species in a fen in the United
20 Kingdom. (Data from Wheeler
10 and Shaw 1991.)
10
0 100 200 300 400 500 600 700 0 500 1000 1500 2000 2500 3000
Productivity (g C/m2/yr) Productivity (g C/m2/yr)
Many authors have asserted that at local and region- Rosenzweig (1971) called the pattern of decline in
al scales, there is generally a unimodal, or hump-shaped, diversity at high productivity the “paradox of enrich-
relationship between productivity and diversity (e.g., ment.” This pattern has been found for many different
Grime 1973; Whittaker and Niering 1975; Huston 1979; taxa, not just terrestrial plants. Plant ecologists have
Tilman 1982; Rosenzweig 1992; Huston and DeAngelis addressed this paradox experimentally by fertilizing
1994). That is, diversity increases as productivity increas- plots and seeing how that treatment affects diversity. All
es up to a maximum at intermediate productivity, but of these experiments (reviewed by Grime 1973; Huston
then decreases with further increases in productivity. 1979; Tilman 1982) were done in herbaceous plant com-
James Grace (1999) reviewed evidence for this relation- munities because these communities are likely to re-
ship at small spatial scales in herbaceous communities spond to treatment within a year or two. In most cases,
and concluded that the data were consistent with such a decline in diversity was found after fertilization. How-
a unimodal pattern. ever, this decline occurs even when very low-produc-
Gary Mittelbach and colleagues (2001) conducted tivity sites are raised to intermediate values (Gough et
a broad review of the literature on the relationship al. 2000). It may be that such short-term declines in
between productivity and diversity. They concluded that diversity are due to changes in competitive dominance
while unimodal patterns are found about 40% of the among species (see Chapter 10), while an increase in
time, there are also many examples of positive patterns, diversity requires the migration of new species into the
negative patterns, and even U-shaped patterns (Figure plot, a longer-term process.
14.7). They emphasized that the shape of the relation- A number of reasons have been suggested for the
ship changes as a function of scale. Hump-shaped pat- increases in diversity observed as productivity increas-
terns, for example, are more common in studies that look es from very low to moderate levels. As mentioned
across communities than within a single community above, one explanation is that more energy or resources
(Figure 14.8). Nearly all studies of the relationship can support more individuals, leading to a greater num-
between productivity and diversity are carried out at a ber of species (Wright et al. 1993). Another is that very
single scale, and almost no theories explicitly incorpo- few species are adapted to survive the difficult condi-
rate scale. The issue of scale and productivity-diversity tions that underlie low community productivity (such
relationships is further explored in Chapter 20. as low temperatures, low water availability, infertile
(A) Within communities larger than in other environments. Denslow (1980) sug-
40 gested that because habitats of intermediate productiv-
n = 39 ity are far more common than those with very low or
very high productivity, more species should have
Percent of studies
30
evolved to survive in the conditions presented by those
intermediate habitats, and far fewer should be adapted
20
to conditions found in environments of very high or very
low productivity.
10 As you might gather, the question of the relation-
ship between productivity and diversity is currently an
0
Unimodal U-shaped Positive Negative No
active area of research. Because more than one process
correlation may contribute to the observed patterns, with different
Productivity-diversity pattern processes dominating in different circumstances, resolv-
ing their causes is difficult. While it is satisfying to reach
(B) Across communities
a conclusion on such questions, it is the process of work-
ing out complex matters such as these that makes ecol-
50 ogy an exciting and challenging science.
n = 26
Percent of studies
40 Niche Differentiation, Environmental Heterogeneity,

and Diversity
30
Why does diversity exist? Another way of asking this
20 question is, given that a positive relationship exists
between the number of individuals and the number of
10
species in a community, why is it that adding individu-
0 als leads to more species, rather than just more individ-
Unimodal U-shaped Positive Negative No uals of a single species?
correlation
One reason why a single species does not replace all
Productivity-diversity pattern
the others in a community by outcompeting them is that
Figure 14.8 different species are best adapted to different conditions.
Percentage of studies showing various relationships It is not possible to be best at everything simultaneous-
between productivity and plant species richness at two eco- ly, and environments are commonly heterogeneous in
logical scales: (A) within communities and (B) across com-
munities. (Data from Mittelbach et al. 2001.) both space and time. Adaptations to different conditions
combined with environmental variation allow species
to divide up the environment in various ways. Under-
story species, for example, are able to tolerate low light
soils, or short growing seasons; Grime 1973). According conditions, while canopy species are superior perform-
to this hypothesis, as conditions become more favorable ers under conditions of full sunlight. In areas of outcrops
and productivity increases, more species are able to sur- and patchy, rocky soils, some species are able to domi-
vive because the specialized adaptations necessary to nate pockets of deeper soil, while others are found in the
survive in stressful conditions are not required. Yet shallower soils on ridges and among boulders. Some
another explanation is that in nutrient-poor environ- plants are able to tolerate saline soils, or soils with heavy
ments, the best competitor for the limiting soil nutrient metals, while others are restricted to soils free of these
should dominate, while at intermediate levels of fertili- materials.
ty (and productivity), spatial heterogeneity in soil nutri- Temporal heterogeneity may also offer opportuni-
ents should lead to the coexistence of multiple species ties for specialization. In the Sonoran Desert, which has
due to niche differences among those species (Newman both a winter and a summer rainy season (see Chapter
1973; Tilman 1982; Tilman and Pacala 1993). 18), some annuals are specialized to germinate in
There are also many contradictory explanations for response to winter rains in December, January, and Feb-
the often-reported decline in diversity at high produc- ruary and grow in cool spring temperatures. They flower
tivity (see Table 20.2). Grime (1973) proposed that high- in March and are dead by May. Others germinate in
ly competitive species dominate the most favorable envi- response to summer rains in July and August, grow in
ronments and suppress other, less competitive species the hot temperatures of summer, flower in September,
(Grime 1973; see Chapter 10). Tilman (1982; Tilman and and die in October. Because these two sets of species
Pacala 1993) theorized that in the most fertile and pro- have different phenologies, they do not compete with
ductive environments, individual plants should be much each other. This specialization leads to species coexis-
40
tence and, thus, increased diversity in the community.
Explaining the maintenance of species diversity on the 35
n = 197
Percent of studies
basis of environmental heterogeneity and niche differ- 30
entiation also may depend on differences in species’ 25
competitive abilities under different conditions (see 20
Chapter 10). 15
10
Gaps, Disturbance, and Diversity
5
Disturbances can create gaps in mature communities,
0
particularly in forests, as we saw in Chapter 13. Unimodal U-shaped Positive Negative No
Researchers have proposed two different reasons why correlation
these gaps might increase community diversity. The first Disturbance-diversity pattern
is that some species might be specialized to take advan- Figure 14.9
tage of the different conditions afforded by gaps of dif- Percentage of studies showing various relationships
ferent sizes (the niche-partitioning hypothesis). If this between disturbance and species richness. (Data from
were the case, certain species could be maintained in the Mackey and Currie 2001.)
larger community by being superior at colonizing and
dominating small gaps, for example, while others could
dominate intermediate or large gaps, and still others
would largely be found only in intact forest. affect diversity. If disturbances reduce the abundance,
The second reason why gaps might increase total distribution, or height and spread of competitive dom-
forest diversity is based on chance, or stochastic (ran- inants, competitively inferior species may be able to
dom) events. According to this second hypothesis, the coexist with them. But if disturbances are too frequent
plants that dominate any particular gap are those that or severe, few species will be able to survive. This obser-
happen to be there when the gap forms, or those with vation led to the formulation of the intermediate dis-
propagules nearby. Species coexistence in the case of turbance hypothesis (Connell 1978), which states that
niche partitioning would depend on competitive inter- species diversity should be highest at intermediate lev-
actions, with different species being competitively supe- els of disturbance. According to this hypothesis, com-
rior in different gap types. In the case of stochastic petitive exclusion reduces species diversity at low lev-
events, competitive exclusion would be prevented els of disturbance. When disturbances are very frequent,
because many competitively inferior species would be most existing species are destroyed, with insufficient
able to persist by chance, essentially by being in the right time available for full recolonization before the next dis-
place at the right time. turbance. The intermediate disturbance hypothesis has
Brokaw and Busing (2000) reviewed the evidence been tested in many different systems (not only in plant
for the role of gaps in maintaining tropical and temper- communities) and a conceptual model has been devel-
ate forest diversity and evaluated the data in support of oped and formalized (reviewed by Huston 1994). In a
these two hypotheses. They concluded that gaps are survey of 197 studies, Mackey and Currie (2001) found
indeed very important in maintaining species diversity that the most common pattern was no relationship
in both temperate and tropical forests. They did not find between disturbance and diversity, while humped, pos-
any differences in the role of gaps in temperate and in itive, and negative relationships were all frequent (Fig-
tropical regions. The evidence pointed to a major role for ure 14.9). Thus, while it is clear that disturbance pro-
stochastic events in allowing species coexistence in motes diversity in some cases, what we do not know is
forests. There are clearly some species that are better how important it is relative to other factors in deter-
than others at dispersing to and growing quickly in gaps mining diversity.
(pioneer species). However, while there is some evidence
for niche differentiation in different kinds of gaps, this Effects of Increasing Diversity
evidence is much weaker than the evidence for the role
of stochastic processes, and niche partitioning proba- In the previous section we looked at some of the evi-
bly does not play a large role in maintaining diversity in dence for the effects of increased productivity on diver-
forests. Other studies have agreed with these conclu- sity. The direction of causation can also be reversed:
sions for both tropical and temperate forests (e.g., Brown some ecologists believe that productivity, as well as other
and Jennings 1998; Busing and White 1997), although ecosystem properties (see Chapter 15), should be
Schnitzer and Carson (2001) found evidence for niche enhanced by increased diversity. (“Diversity,” in this
partitioning in tropical forest gaps. context, can refer to either species diversity or diversity
Gaps are not the only result of disturbance that may of functional groups.) It is easy to imagine how this
might occur. Imagine, for example, a forest with no on what the main issues were and on the types of data
spring ephemerals; no species are growing actively in needed to settle those issues.
the early spring. If a spring ephemeral species were
introduced into the forest, it would not displace any of Testing the Effects of Diversity on Ecosystems
the current species; rather, it would occupy an empty The effect of diversity on productivity has been tested
niche. It might use up nutrients that another herbaceous experimentally in herbaceous communities (old fields
species growing later in the year would use. But if light and prairies). One large recent experiment was con-
were generally limiting for those species, the effect of the ducted by a group of plant ecologists in Europe (Hec-
spring ephemeral on the growth of the later species tor et al. 1999). They sowed seeds of the same species
might be small relative to the amount of new biomass at two sites each in Germany, Greece, Ireland, Portugal,
it would add to the community. Switzerland, Sweden, and the United Kingdom. Among
This idea has been the subject of heated debate plots within each site, they varied the number of species
among ecologists. Shahid Naeem and his colleagues and the number of functional groups. At the end of the
(Naeem et al. 1996), David Tilman (1996), and others have growing season, they counted the number of species
been strong supporters of the present in each plot and harvested all of the above-
idea that an increase in biodiver- ground biomass to measure productivity. They found
sity increases productivity as well that greater diversity was related to greater productivi-
as affecting other ecosystem ty at all of the sites (Figure 14.10A). Most of this effect
properties. The idea has been could be attributed to differences in the number of func-
strongly criticized by David War- tional groups present in a plot (Figure 14.10B). Other
dle and other ecologists, howev- experimental tests of the effects of species or functional
er (e.g., Wardle et al. 1997, 2000), group diversity on ecosystems have also been carried
for a number of reasons, particu- out (e.g., Tilman 1996; Hooper and Vitousek 1998; see
larly its proponents’ interpreta- Chapter 15), but the results have been subject to very dif-
tions of experiments and data. ferent interpretations (Wardle et al. 2000; Naeem 2000).
David Tilman Wardle and his colleagues have As with the effects of productivity on diversity, it is like-
argued that increased productiv- ly that the effects of diversity on productivity will
ity and the other ecosystem properties attributed to depend in some as yet unknown ways on spatial and
increased biodiversity are mostly a result of traits of temporal scale.
the particular dominant species and functional groups These effects, if substantiated in other systems, could
present, rather than of increased diversity itself. Recent- have important implications for conservation efforts. A
ly, many of the protagonists in this debate published a failure to conserve large tracts of wildlands could lead
joint paper (Loreau et al. 2001), in which they agreed to a decrease in plant species diversity (see Chapter 17).
(A) (B)
1500 800
Aboveground biomass (g/m2)
Aboveground biomass (g/m2)
700
1000
600
Germany 500
500 Ireland
UK 8 species
Switzerland 400 4 species
Portugal 2 species
Sweden 11 species
0 Greece 300
32 16 8 4 1 3 2 1
Decreasing species richness (log2 scale) Decreasing functional group richness
Figure 14.10
The relationship between plant diversity and aboveground productivity in experi-
mental plots at eight sites across Europe. (A) Species richness. The points represent
individual plots; the lines are slopes from a series of regressions. (B) Functional
group richness. The points are averages over all sites. (After Hector et al. 1999.)
If such a diversity loss led to a decrease in primary pro- More recently, several ecologists have studied a
ductivity, the effects could potentially reverberate quite different type of stability: they have attempted
through the food chain, leading to extinctions of herbi- to determine whether more diverse communities have
vores and carnivores. less variation in productivity from one year to the next.
The hypothesis is simple: with more species present, it
Diversity and Stability is more likely that (regardless of environmental fluctu-
Yet another contentious issue of long standing in ecolo- ations) in any given year, some species will have a
gy is whether community diversity is related to the sta- “good” year. In other words, the ability of different
bility of communities through time. The answer depends species to grow well under different conditions should
on which of several senses of “stability” one has in mind. mean that more diverse communities vary less in pro-
In mathematics, dynamic stability means that a system ductivity.
tends to return to its original state after a small pertur- In a study of prairie communities in Minnesota,
bation. Robert May (1973), using a mathematical model, Tilman and Downing (1994; Tilman 1996) found that
predicted that increased diversity would have the oppo- more diverse communities (those with a greater num-
site result: after perturbation, a more diverse communi- ber of species) maintained higher productivity, even dur-
ty should be less likely to return to its original state than ing a period of drought (Figure 14.11). Some species
a less diverse community. Subsequent theoretical stud- were better adapted to low water conditions than oth-
ies have slightly modified this conclusion: the robust the- ers; in a drought year, those species constituted a greater
oretical result is that increasing diversity in ecological percentage of the total biomass than in other years. Their
models is never likely to increase the dynamic stability biomass increased when drought caused other species
of a community, but may not decrease it. to grow poorly. Ecologists have debated whether this
(A) (B)
80 Field A r = – 0.39 90 Field B r = – 0.32
for community biomass
80
Coefficient of variation
70
70
60
60
50
50
40
40
30 30
20 20
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Average species richness Average species richness
(C) (D)
80 Field C r = – 0.09 90 Field D r = – 0.53
70
80
60 70
50 60
40 50
30 40
20 30
10 20
2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16 18 20 22
Average species richness Average species richness
Figure 14.11
Relationship between plant species richness and stability of productivity (variability
in total biomass among years) for 4 × 4 m2 plots in four fields in a prairie in Minneso-
ta from 1984 to 1994. A very severe drought occurred during this period in
1987–1988. (After Tilman 1996.)
finding represents an important emergent property of

Summary
communities (Tilman 1997) or a statistical inevitability In most communities, most species are rare, and only a
of sampling larger groups of species (Doak et al. 1998). few are common. The most conspicuous or numerically
abundant species in a plant community are called the
Regional Processes dominants. The limited data available suggest that
In this chapter we have focused on the role of local numerically dominant species tend to be competitive
processes in determining community structure. How- dominants as well, at least in some communities. Species
ever, local diversity, dominance, and abundances might may be rare in different ways and for different reasons,
be determined not only by the interactions among and a species’ status as rare or common may sometimes
species within a site, but also by regional processes. If a change dramatically over time. Rare plant species can
species is common throughout a region, for example, be limited in terms of geographic distribution, local pop-
then a continual influx of migrants into a site might keep ulation size, or range of habitats occupied. A common
its densities high. We explore this process in more detail explanation for rarity, specialization and narrow envi-
when we look at metapopulation models in Chapter 17. ronmental tolerances, does not appear to be a good gen-
Similarly, the number of species found at a site eral explanation for why species are rare.
might depend, in part, on the regional species pool. If Various explanations have also been proposed for
we compare the number of tree species in deciduous the ability of some plant species to become invasive
forests in the eastern United States, Europe, and China, when they are introduced into new environments. The
for example, we see a curious result. Although we can characteristics of species themselves, and of their situa-
find sites with nearly identical climates and soil condition in a new environment, have been used with vary-
tions, we will not find the same numbers of species. ing success to predict invasiveness. Characteristics of
Species richness will be highest in China, somewhat communities that may make them susceptible to inva-
lower in the United States, and least in Europe. These sion by exotic species have also been proposed and, to a
local differences reflect differences in the regional species limited extent, tested.
pool. Europe has the fewest tree species because of Changes in productivity can alter the richness and
extinctions during the Pleistocene glaciations. China, on diversity of species in a community. One common pat-
the other hand, has the most species because its tem- tern is a unimodal (hump-shaped) relationship, in which
perate zone is adjacent to tropical land areas that have diversity is highest at intermediate productivity and
been a source of new species over geological time (Qian lower at both low and high productivity, but other pat-
and Ricklefs 1999). These regional differences in species terns are also commonly found. Species diversity can
richness are a result of nonequilibrium processes that also be related to the frequency and severity of distur-
operate over very long time scales. Determining how bance. Gaps created by disturbance are important in
local and regional patterns and processes are linked is maintaining species diversity in temperate and tropical
currently one of the grand challenges in ecology (Law- forests. The question of the effects of species diversity
ton 1999). We take up these issues in more depth in on ecosystem properties is still under active investiga-
Chapters 17, 20, and 21. tion and remains highly controversial.
Additional Readings
Classic References Kelly, C. K. and F. I. Woodward. 1996. Ecological correlates of plant range
size: Taxonomies and phylogenies in the study of plant commonness and
Hutchinson, G. E. 1959. Homage to Santa Rosalia or Why are there so
rarity in Great Britain. Phil. Trans. R. Soc. Lond. B 351: 1261–1269.
many kinds of animals? Am. Nat. 93: 145–159.
Smith, M. D. and A. K. Knapp. 1999. Exotic plant species in a C4-domi-
Preston, F. W. 1962. The canonical distribution of commonness and rar-
nated grassland: Invasibility, disturbance, and community structure.
ity. Ecology 43: 185–215, 410–432.
Oecologia 120: 605–612.
Tilman, D. 1982. Resource Competition and Community Structure. Prince-
ton University Press, Princeton, NJ.
Huston, M. A. 1994. Biological Diversity. Cambridge University Press,
Contemporary Research Cambridge.
Howard, T. and D. E. Goldberg. 2001. Competitive response hierarchies
Rosenzweig, M. L. 1995. Species Diversity in Space and Time. Cambridge
for different fitness components of herbaceous perennials. Ecology
University Press, Cambridge.
82:979–990.
PART
IV From Ecosystems
to Landscapes
15 Ecosystem Processes 295

16 Communities in Landscapes 319
17 Landscape Ecology 333
15
C H A P T E R
Ecosystem Processes
T
he focus here is somewhat different from that of the other chapters in
this book. In this chapter we look at plants as critical players in ecosys-
tems, rather than focusing on plants themselves. An ecosystem consists
of all of the organisms in an area and all of the abiotic materials and energy
with which they interact; it is a bounded ecological system. The word ecosys-
tem was devised by Sir Arthur Tansley (1935) to include the entire system of
living organisms in the context of the physical factors on which they depend
and with which they are interconnected.
Tansley also stated that ecosystems could exist over a range of scales—a
very contemporary concept. He further stated that “the [eco]systems we iso-
late mentally are not only included as parts of larger ones, but they also over-
lap, interlock and interact with one another. The isolation is partly artificial,
but is the only possible way in which we can proceed.” Ecosystem ecologists
ask questions about the roles of plants as conduits for energy and materials
in ecosystems and as agents for the transformation of energy and materials
into different forms, and about the effects of the supplies and flows of energy
and materials on plants.
What determines plant productivity, and why is it so different in differ-
ent places on Earth? What are the important controls over the flow of energy
and materials in ecosystems? What are the most important chemical elements
affecting plant growth and decomposition, and how do the relative amounts
of these nutrients affect these processes? How do ecosystem processes affect,
and how are they affected by, characteristics of communities such as species
diversity and plant-animal or plant-microorganism interactions?
We begin with a look at biogeochemical cycles in general, then look at
the water cycle as an example. Then we turn to the cycling of carbon in ecosys-
tems and its roles in biogeochemical processes from decomposition to pro-
ductivity. (We will revisit the global carbon cycle in Chapter 22, when we
consider it in the context of climate change.) We then reexamine the nitrogen
cycle, first introduced in Chapter 4, and go on to the phosphorus cycle, with
a brief look at how some other elements move through the biosphere. Here we
begin by looking at how a young plant grows, widening our focus to the
ecosystem context from there.
296 Chapter 15
Biogeochemical Cycles: tion from accurate measurements of the other terms

Quantifying Pools and Fluxes (Vitousek and Reiners 1975).
Ecosystem ecologists also seek to make predictions
A seed, with its embryo and other associated tissues, about the future sizes of pools and the magnitude and
contains very limited quantities of energy and materi- direction of fluxes. The major fluxes and pools of a sub-
als. Seeds are generally very small—orders of magnitude stance in a system are collectively called a cycle; for
smaller than the plants into which they will grow. The example, we might diagram the global water cycle, or at
spore that will develop into a mature fern is even small- a smaller scale, the nitrogen cycle or phosphorus cycle
er than most seeds. As the new individual grows into a of a forest ecosystem. Most of the nutrients that are
mature plant, it transforms inorganic forms of carbon important to plants are cycled biogeochemically—that
and other materials into complex organic molecules, and is, both biological and chemical reactions are involved.
it stores the energy captured from photons of light in the The material contained in a plant is composed most-
form of chemical bonds in these molecules (see Part I). ly of carbon (C) and oxygen (O) (see Table 4.2). The oxy-
In terrestrial systems, most of the materials and energy gen available to aboveground parts of plants is almost
used by heterotrophs—animals, fungi, bacteria, and pro- never limiting to terrestrial plant growth (oxygen can,
tists—are transformed from unusable inorganic forms however, be limiting below ground, particularly in
into usable organic materials by plants. (Photosynthet- waterlogged soils). Carbon, in the form of CO2, is on
ic bacteria, diatoms, green and red algae, and other average equally available everywhere on Earth (at sea
organisms play a similar role in aquatic systems and level; it is considerably less available at high altitudes).
sometimes in soils.) The story is actually more compli- Yet the production of plant biomass varies enormously
cated than that, however, because some of the materials over Earth’s surface (see Table 19.2). What are the caus-
needed by plants must first be transformed by microor- es of this tremendous variation?
ganisms into forms the plants can use (see Chapter 4). Briefly, primary productivity is determined both by
The quantity of material transformed into organic climatic factors (such as moisture, light, temperature,
forms by an individual plant can range from fractions of and length of the growing season) and by the supply of
grams to metric tons (1 metric ton = 1000 kg), depending nutrients essential for plant growth. We have examined
on the size of the plant and the substance being trans- some of these factors in Part I and in Chapter 14. Here
formed. At the ecosystem scale, this can add up to many we look at the major factors controlling the ability of
metric tons of material. Terrestrial primary productivi- plants to obtain essential nutrients, as well as the roles
ty—the amount of carbon transformed from CO2 into of plants as conduits for the fluxes of those nutrients,
organic carbon (C) by terrestrial plants per unit area per as agents controlling those fluxes, and as reservoirs for
year—is typically on the order of 5 to 10 metric tons per their storage.
hectare (far smaller amounts are also fixed by other Some essential nutrients are needed by plants in rel-
organisms). On the global scale, fluxes of carbon, for atively large quantities,* but their availability to plants
example, are on the order of 1015 grams of carbon per year is limited. Nitrogen (N) and phosphorus (P) are in this
(gigatons). category; both are major constituents of essential organ-
A major focus of ecosystem ecology is understand- ic molecules, but their supply is limited in most soils.
ing what regulates the pools (quantities stored) and Consequently, the availability of N and P controls pri-
fluxes (flows) of materials and energy in the various mary productivity in many ecosystems. In contrast, in
abiotic and biotic components of ecosystems. The pools the case of some nutrients that are available in larger
and fluxes of some major mineral nutrients in a Cali- quantities, such as sulfur (S) and potassium (K), primary
fornia chaparral system are shown as an example in productivity can determine the rate at which they cycle
Table 15.1. in the ecosystem (Schlesinger 1997). In both cases, living
A basic approach that is often used in accounting for organisms have a major effect on the geochemistry—the
the magnitude of pools and fluxes is the mass-balance pools and fluxes—of these major components of living
approach to constructing nutrient budgets. Simply put, things.
the mass-balance approach states that In contrast, the cycling of elements that are not major
constituents of living things, such as sodium (Na) or alu-
inputs – outputs = ∆ storage
minum (Al), is relatively independent of the actions of
In other words, if we can measure everything that goes living organisms (Schlesinger 1997). It is because plants
into a system and everything that comes out, the dif- have a major role in the local, regional, and global cycles
ference between the two quantities must be reflected by of water and some chemical elements that we include
a change, ∆, in the material stored in the system. The
mass-balance approach allows ecosystem ecologists to *You may wish to review Table 4.2, which summarizes the roles of the var-
account for difficult-to-measure quantities by subtrac- ious mineral elements necessary for healthy plant growth and nutrition.
Ecosystem Processes 297
Table 15.1 Nutrient cycling in a 22-year-old stand of the chaparral scrub Ceanothus megacarpus near
Santa Barbara, California
Biomass N P K Ca Mg
2
Atmospheric input (g/m /year)
Deposition — 0.15 — 0.06 0.19 0.10
Nitrogen fixation — 0.11 — — — —
Total input — 0.26 — 0.06 0.19 0.10
Pools (g/m2)
Foliage 553 8.20 0.38 2.07 4.50 0.98
Live wood 5929 32.60 2.43 13.93 28.99 3.20
Reproductive tissues 81 0.92 0.08 0.47 0.32 0.06
Total live 6563 41.72 2.89 16.47 33.81 4.24
Dead wood 1142 6.28 0.46 2.68 5.58 0.61
Surface litter 2027 20.5 0.6 4.7 26.1 6.7
Annual flux (g/m2/year)
Requirement for production
Foliage 553 9.35 0.48 2.81 4.89 1.04
New twigs 120 1.18 0.06 0.62 0.71 0.11
Wood increment 302 1.66 0.12 0.71 1.47 0.16
Reproductive tissues 81 0.92 0.08 1.47 0.32 0.07
Total in production 1056 13.11 0.74 4.61 7.39 1.38
Reabsorption before abscission — 4.15 0.29 0 0 0
Return to soil
Litterfall 727 6.65 0.32 2.10 8.01 1.41
Branch mortality 74 0.22 0.01 0.15 0.44 0.02
Throughflow — 0.19 0 0.94 0.31 0.09
Stemflow — 0.24 0 0.87 0.78 0.25
Total return 801 7.30 0.33 4.06 9.54 1.77
Uptake (=increment – return) — 8.96 0.45 4.77 11.01 1.93
Streamwater loss (g/m2/year) — 0.03 0.01 0.06 0.09 0.06
Comparisons of turnover and flux
Foliage requirement/total requirement (%) — 71.3 64.9 61.0 66.2 75.4
Litterfall/total return (%) — 91.1 97.0 51.7 84.0 79.7
Uptake/total live pool (%) — 21.4 15.6 29.0 32.6 45.5
Return/uptake (%) — 81.4 73.3 85.1 86.6 91.7
Reabsorption/requirement (%) — 31.7 39.0 0 0 0
Surface litter/litterfall (year) 2.8 3.1 1.9 1.2 3.3 4.8
Source: Schlesinger (1997), modified from Gray (1983) and Schlesinger et al. (1982).
this topic in a plant ecology textbook. The large part that growth, tissue composition, leaf longevity, litter pro-
plants—and humans—play in global biogeochemistry duction, and litter decomposition rates. These intercon-
also has major implications for global change, as we will nections determine the flux and pool sizes of C, N, and
see in Chapter 22. P. All three cycles are also simultaneously affected by cli-
Cycles of energy and materials are often illustrated matic and local environmental factors such as tempera-
by diagrams that show the fluxes and pools of a single ture, precipitation, and light availability.
element. While this makes for a convenient presentation, There are large differences in the magnitudes of
it is important to recognize that the cycles of various pools and fluxes among different materials and ecosys-
materials can be highly interconnected, and that they tems. Fluxes of water and carbon are vastly greater than
interact to affect plant growth. The carbon cycle, for fluxes of phosphorus, for example. Pools of many nutri-
example, both depends on and strongly affects the nitro- ents are much larger in living plants in tropical forests
gen and phosphorus cycles (Shaver et al. 1992). These than they are in tundra plants. There are generally huge
cycles are interconnected by their mutual effects on plant pools of stored carbon in bog soils, substantial (but far
298 Chapter 15
less) stored carbon in grassland soils, and very little

The Global Water Cycle
stored carbon in desert soils.
The ultimate sources of the different essential nutri- Terrestrial plants are the only living things (except
ents also differ. The atmosphere is the source of N, O, and humans) to have a substantial effect on the global water
C, although plants obtain N from the air only indirectly. cycle (Figure 15.1). Most of the world’s available water
The atmosphere contains about 78% N and about 21% O. (96.5%) is stored, not surprisingly, in the oceans, with all
The concentration of CO2 in the atmosphere, while rising of the fresh (i.e., non-salty) water in the world’s rivers,
(see Figure 22.2), is surprisingly low—about one-third of lakes, ice, atmosphere, and groundwater constituting the
one percent on average, or about 370 parts per million by remaining few percent. Living things store relatively
volume (ppmv). Nevertheless, because the volume of the minuscule amounts of water, but the amount of water
atmosphere is so large, the total pool of carbon in the that moves through plants by transpiration is large and
atmosphere is very large. Large amounts of carbon are globally important.
also stored as carbonate ions (CO32–) in rock and dis- The flux of water through plants can also be impor-
solved in seawater; additional carbon is contained in liv- tant regionally. Plants can contribute a major share of the
ing biomass and in soils. Liquid water is the source of water in the atmosphere and in regional precipitation,
hydrogen (H), and is the ultimate source of the O released but the effects of plants on local precipitation patterns
to the atmosphere by photosynthesis. Rock weathering vary greatly. Precipitation in California, for instance,
is the major source for most of the other elements need- comes largely from water evaporated over the Pacific
ed by plants (such as Ca, Mg, K, Fe, and P), while S comes Ocean. In contrast, between one-quarter and one-half of
both from atmospheric deposition and rock weathering the rainfall over the Amazon basin comes from evapo-
(Schlesinger 1997). In addition to new inputs from the transpiration (evaporation and transpiration) from the
atmosphere and the weathering of rock, plants rely on Amazon forest itself, with the rest coming from outside
ecosystem recycling of elements (Table 15.2). the region (Salati and Vose 1984; Eltahir and Bras 1994).
Before proceeding with a discussion of individual Cutting large areas of the Amazon forest can therefore
cycles, we must confront the issue of scale (see Chapter result in decreased evapotranspiration, decreased rain-
17), which we purposely blur in this chapter. Ecosystem fall, and increased temperatures at the ground surface
ecologists treat local, regional, and global cycles sepa- (Lean and Warrilow 1989; Shukla et al. 1990).
rately. Our focus here is that point at which biogeo- Plants can also indirectly affect water fluxes by
chemistry most strongly interacts with plants. We con- intercepting precipitation, reducing the impact of rain
sider aspects of the ecosystem ecology of several hitting the ground, and decreasing runoff from the soil.
elements in other chapters where they are particularly If vegetation is removed—for example, by clear-cut-
relevant. Water moves over large scales on Earth, and ting forests—runoff and soil erosion can be greatly
we focus on the pools and fluxes of water at large scales increased, particularly in areas where the terrain is hilly
here. Carbon, nitrogen, and phosphorus move through or mountainous. This can sometimes result in flooding
plants and affect plants at a range of scales, from the rhi- and mudslides, further damaging vegetation and threat-
zosphere to the globe, and we consider these elements ening human lives and homes. Even in semiarid regions,
over a range of scales both in this chapter and else- removal of vegetation can result in reduced precipita-
where. tion, increased soil warming, and the onset of desertifi-
Table 15.2 Annual nutrient requirements supplied by various sources in the Hubbard Brook
Experimental Forest, New Hampshire
N P K Ca Mg
Requirement kg/ha/year 115.4 12.3 66.9 62.2 9.5
Percentage supplied by:
Intersystem inputs
Atmospheric 18 0 1 4 6
Rock weathering 0 1 11 34 37
Intrasystem transfers
Reabsorptions 31 28 4 0 2
Detritus turnover (includes return 69 67 87 85 87
in throughflow and stem flow)
Sources: Reabsorption data are from Ryan and Bormann (1982). Data for N, K, Ca, and Mg are from Likens and Bormann (1995)
and for P from Yanai (1992).
Atmospheric water (vapor) Figure 15.1

10.0 The global water cycle. The
numbers show the pools (in
Vapor transport units of 1000 km3 of water,
40 shown in regular type) held in
Precipitation over land various major components of
111 the global ecosystem, and the
Ice Precipitation
over oceans fluxes among those components
24,000 Evaporation + Transpiration (1000 km3/year, shown in bold-
385
71 faced type). Notice that the
Evaporation amount of water leaving the
Surface Living from oceans
runoff organisms land to return to the oceans
425
1 (40,000 km3/year) is equal to
the amount of water that
Percolation
Lakes
returns from the oceans to the
and rivers Rivers land as water vapor evaporated
40 Oceans
Soil <200 1,338,000 from the oceans. The numbers
Land given are approximate, as esti-
water
<20 mates vary considerably
between different authors.
Aquifer (After Schlesinger 1997 and
Groundwater 23,000 Gleick 1996; data from Lvovitch
1973 and Chahine 1992.)
cation (Schlesinger et al. 1990; Dirmeyer and Shukla and tropical forests generally lose less water by evapo-
1996). transpiration than they receive as precipitation, so con-
In intact vegetation, the balance between precipita- siderable amounts often run off into rivers and streams
tion, evapotranspiration, and runoff varies greatly among and percolate to the groundwater. Over the Earth as a
biomes (Table 15.3). In deserts and grasslands, all or most whole, rivers carry about a third of the precipitation that
of the water that enters the system as precipitation is lost falls over land back to the oceans. While the amount of
by evaporation or transpiration, leaving almost nothing rainfall over the oceans is greater than that over land (see
for runoff and recharge of groundwater. Both temperate Figure 15.1), the evaporation of water from the oceans
Table 15.3 Relative importance of pathways leading to the loss of water

in a variety of terrestrial ecosystems
Runoff and
groundwater
Biome Evaporation (%) Transpiration (%) recharge (%)
Tropical rainforest 25.6 48.5 25.9
Tropical rainforest 10 40 50
Tropical rainforest 11 56 32
Temperate forest 13 32 53
Temperate grassland 35 65 0
Desert 28 72 0
Desert 20 80 —
Desert 73 27 —
Desert 65 35 —
Source: Schlesinger (1997).
300 Chapter 15
is also much greater than the evapotranspiration of water soil processes (Schlesinger 1997). This is not surprising,
over land. The net movement of water vapor from the given the dependence of plant and microbial activity on
oceans to the land through the atmosphere is balanced soil water content and the close functional linkage
by the runoff of water from the land to the oceans in the between carbon assimilation and water loss in plants
form of groundwater and river flow. While only a minus- (see Chapters 2 and 3).
cule amount of water is stored as water vapor in the
atmosphere, the flux of water vapor through the atmos-
phere is tremendous. Carbon in Ecosystems
Potential evapotranspiration (PET) is the maximum
amount of water that would be lost to evapotranspira- Decomposition and Soil Food Webs
tion in a particular place if water were freely available From the shortest-lived microorganism to the longest-
in the soil and plant cover were 100%. PET depends on lived tree, everything that lives eventually dies. If the
the energy available to evaporate water, which depends organic detritus—the dead bodies and waste products of
on the ambient temperature. You may be surprised to living and once-living organisms—were not decomposed,
know that PET can be much greater per unit area than most ecosystem processes would come to a screeching
the amount of water that would be lost by evaporation halt, and needless to say, the piled-up dead bodies and
from an open body of water. This is because the surface wastes would not be a pretty sight. Sooner or later, much
area of the roots taking up water is much greater than of the carbon and other material and some of the energy
the surface area of the ground in which the plants are released by the process of decomposition are taken up by
rooted, and because the surface area of the leaves from living organisms and used again. How are materials that
which water is evaporating is often greater than the sur- were once part of living organisms recycled?
face area of the ground—that is, the leaf area index Decomposition transforms dead plants and animals,
(LAI) is often greater than 1.0. their shed or removed parts (e.g., leaves, bark, branches,
Actual evapotranspiration (AET) is equal to the and roots), and animal feces into soil organic matter and
amount of water that enters the system in precipitation ultimately into inorganic nutrients, CO2, water, and ener-
minus the amount that is lost in runoff and percolation gy. In terrestrial ecosystems, decomposition involves both
to groundwater (the amount of water stored in living physical and chemical alteration of the original materi-
systems is minimal relative to the other terms). In areas als by a complex food web, which exists mostly (but not
with long dry periods, PET is much greater than actual entirely) on or just under the surface of the soil (Figure
evapotranspiration, but in tropical rainforests PET and 15.2).
AET are approximately equal. Actual evapotranspira- Decomposition is largely an aerobic process (one
tion is closely linked to other major ecosystem process- requiring oxygen), and it occurs slowly, if at all, in water-
es, particularly productivity, decomposition, and other logged soils or other anaerobic environments. Warm,
Fungal-
Shoots Plant-feeding feeding Predatory
nematodes mites nematodes Figure 15.2
Roots Diagram of a soil food web.
Autotrophs at the soil surface
and upper layers of the soil
Mycorrhizal Fungal- include plants, with above-
Algae fungi feeding ground photosynthetic parts
nematodes Predatory and roots below ground, pho-
mites tosynthetic single-celled pro-
Photosynthetic
bacteria tists (including diatoms and
Bacterial- green algae), and cyanobacte-
feeding ria (photosynthetic bacteria).
Lichens nematodes
Saprophytic Higher Heterotrophs include many
fungi level different kinds of bacteria, sin-
predators gle-celled animals (amoebae,
Flagellates ciliates, flagellates), mycor-
Organic rhizal and saprophytic (decom-
matter posing) fungi, nematodes, soil
Amoebae microarthropods and macro-
Bacteria arthropods, including mites,
Ciliates and a variety of vertebrates as
Dead well. (After Ingham, unpub-
material lished, at http://
(from all boxes) www.soilfoodweb.com)
100 r 2Pinus = 0.63 moist conditions favor the highest rates of decomposi-
r 2Drypetes = 0.76 tion. Cold, anaerobic (oxygen-free) conditions inhibit
decomposition, resulting in the buildup of organic mate-
60 rial. Decomposition rates also depend on other factors,
Mass remaining after 5 years (%)
especially the physical and chemical characteristics of

the material being decomposed (Figure 15.3).
60 Pinus Scavenger animals begin the process of decomposi-
tion of materials of animal origin by consuming animal
bodies. These scavengers include large animals (hyenas,
40 vultures, and so forth) as well as insects (such as Nicro-
Drypetes
phorus burying beetles and the Dipteran larvae known as
maggots). Dung beetles are important and astonishing-
20 ly effective consumers of mammal feces in tropical grass-
lands.
The major source of material for decomposition is
0 plant litter, which consists of a variety of materials—from
–10 –5 0 5 10 15 20 25 30
dead roots and shed leaves, needles, or bark, to dead tree
Mean annual temperature (°C)
trunks. Litter must be broken up into small pieces of
Figure 15.3 organic matter before it can be effectively decomposed
Decomposition rates of the roots of two species of trees with by microorganisms. This process of litter fragmentation
highly contrasting properties: Drypetes glauca (Euphor-
is largely accomplished by animals. Large animals, from
biaceae), a tropical hardwood, and Pinus elliotii (slash pine,
Pinaceae), a temperate softwood. Mesh bags with root mate- deer and bears to gophers and voles, break litter apart as
rial were buried in 28 sites in North and Central America. they search for food in the soil and on its surface.
The roots of both species decomposed faster at sites with At an intermediate scale, earthworms, where they
higher annual temperatures. Drypetes roots decomposed are present, ingest and process huge volumes of soil,
faster than Pinus roots at all sites. (After Gholz et al. 2000.)
causing dramatic physical and chemical alteration of the
soil and the organic matter contained in it. Termites eat
living and dead plant material, particularly decomposi-
(A)
tion-resistant wood, digesting cellulose by means of
mutualistic protists and bacteria that inhabit their
hindguts. These insects can be major contributors to
tropical ecosystem dynamics.
The smallest multicellular animals in soil food webs
are microscopic insects and arachnids known as soil
microarthropods, or soil microfauna and mesofauna (Fig-
ure 15.4). These organisms are par-
(B)
ticularly important in the decompo-
sition process, especially in forests.
Among them are the Collembola
Figure 15.4 (springtails), which are insects that
Some invertebrates that are important in eat fungi, and mites, which are
soil food webs. (A) Sminthurinus elegans (a arachnids that directly consume lit-
springtail; Collembola). Springtails are very
common tiny soil insects that consume ter as well as eating soil fungi and
fungi. The organism pictured is about 3 mm bacteria. Nematodes (Figure 15.4C)
long. (Photograph courtesy of K. Brockle-
hurst.) (B) A soil mite (in the Cepheidae;
genus and species unknown), from a sub- (C)
tropical rainforest in Queensland, Australia.
These tiny (about 2 mm) arachnids consume
fungal hyphae and spores, as well as algae,
lichens, and bacteria. Soil mites are common
worldwide. (Photograph courtesy of D. E.
Walter.) (C) A nematode, Acrobeloides nanus,
showing its front (anterior end). This nema-
tode species is probably one of the most common invertebrates in the world. It lives on
all continents, including the subantarctic islands, and in favorable sites may have pop-
ulations of over 1,000,000 individuals per square meter. Its entire length is about 0.2
mm. (Photograph courtesy of S. Boström.)
302 Chapter 15
are spectacularly abundant and critical in soil food webs among different species (see Figure 15.3) and different
as herbivores that eat plant roots, carnivores that eat soil plant parts (leaves versus wood, for instance). The exten-
animals, and consumers of soil-dwelling bacteria and sive multicellular mycelia that make up a fungus can
fungi. Unicellular protists, including ciliates and amoe- obtain carbon from one source and nitrogen from a dif-
bas, live in the films of soil water between and sur- ferent source at some distance, further enhancing the
rounding soil particles and prey on soil bacteria. By their ability of fungi to use dead plant materials as a source
voracious consumption of soil bacteria and fungi, soil of nutrients and energy.
microarthropods, nematodes, and protists are responsi- Bacteria play an essential role in soil food webs and
ble for converting large amounts of microbial nitrogen nutrient cycling, although they are generally not very
and phosphorus into forms available to plants, as well important in the initial stages of decomposition in most
as releasing microbial carbon as respired CO2. ecosystems. Bacteria can lyse (rupture) and break down
Saprophytic fungi rely on non-living organic mate- living and dead cells of plants, animals, fungi, and other
rial for their carbon and energy. They are the major de- bacteria. Their presence in the soil is highly variable in
composers of dead leaves and other plant litter. They space and time; they are found in greatest abundance at
account for approximately half of the microbial biomass warm temperatures and when water and nutrients are
in grassland soils and about 90% in temperate forest available. They are found largely in soil macropores
soils. Fungi are heterotrophs that secrete powerful diges- (larger soil pores), in the rhizosphere, and coating aggre-
tive enzymes externally, decomposing their substrate. gates of soil particles (see Chapter 4). They move
The fine hyphae of fungi can penetrate plant organs through the soil passively, carried by water or animals.
and cells, overcoming the defensive cuticle to gain access We will return to the roles of various functional groups
to the interior and speeding decomposition. However, of bacteria below.
both plant structural defenses, such as lignin, and vari-
ous aspects of plant defensive chemistry, including tox- Productivity
ins and pH levels, can inhibit their effectiveness. Such Through photosynthesis, plants and other photosyn-
differences in plant chemistry and physical structure thetic organisms capture carbon and energy, transform-
contribute to large differences in decomposition rates
Figure 15.5 Glaciers, 0 6.0 – 8.0 t/ha/yr

Global map of estimated NPP (total net primary productivi- 0 – 1.0 t/ha/yr 8 – 11 t/ha/yr
ty, above and below ground) for terrestrial plants in metric
tons/ha/year of total dry organic matter. (Unpublished data 1.0 – 2.5 t/ha/yr 11 – 16 t/ha/yr
from N. I. Bazilevich, Global Primary Productivity Database, 2.5 – 4.0 t/ha/yr 16 – 30 t/ha/yr
1993, http://ceos.cnes.fr:8100/cdrom-006/ceos1/casestud/ 4.0 – 6.0 t/ha/yr Varies with elevation
ecoreg/datasets/602/baz.htm)
(A) (B)
10,000 1400 New Hampshire
NPP predicted (g m–2 yr–1)

1200
1000 Wisconsin
Florida
800 Oregon Mass. hardwood
N. Carolina
600 Alaska, spruce Mass. pine
NPP (g m–2 yr–1)
1000
400 Alaska, aspen
200 Wyoming
0 200 400 600 800 1000 1200 1400

100 NPP observed (g m–2 yr–1)
Coniferous forests
Deciduous forests
Grasslands
Deserts NPP varies enormously over Earth’s surface (Figure
10 15.5). The greatest amounts of carbon are fixed in warm,
10 100 1000 10,000
moist ecosystems; much less is fixed in arctic systems,
Leaf biomass (g/m2)
and the least in deserts (see Table 19.2). Across a vari-
Figure 15.6 ety of ecosystems within the North American continent,
(A) Net primary productivity (aboveground, in g/m2/year) aboveground NPP varies over more than two orders of
across a wide range of ecosystems in North America, from
deserts to grasslands to deciduous and coniferous forests. magnitude (Figure 15.6A; Table 19.2). Even if we con-
NPP is closely related to leaf biomass. The scale for both sider only forests within this continent, there can be
axes is logarithmic, so these variables have a very large almost as great a range in aboveground NPP (Figure
range. (After Schlesinger 1997) (B) Predicted versus 15.6B). At a larger scale, across the continental United
observed NPP (aboveground, in g/m2/year) in ten North States, NPP depends on latitude, climate, and elevation.
American forests. Aboveground NPP in these forests ranges
from < 300 g/m2/year to almost 1300 g/m2/year, with a NPP in terrestrial ecosystems is broadly limited by
median of about 750 g/m2/year. Predicted values are closely climate, particularly by inadequate temperatures and
correlated with observed values. (After Waring and Running insufficient moisture for growth, and by the length of
1998.) the growing season (Figure 15.7) In addition, produc-
tivity can be limited either by a lack of essential nutri-
ents or by an excess of certain other materials (e.g., toxic
ing oxidized carbon (CO2) into reduced forms as organ- levels of salt or heavy-metal ions) in the soil. More sub-
ic carbon compounds. Productivity is the rate of energy tle factors, such as differences in the amount of respir-
flux through a trophic level in a particular ecosystem. ing tissue in forests with predominately young versus
Productivity is often measured in terms of the amount old trees, also affect NPP. Climatic or soil factors can
of carbon transferred because the energy we are con- influence properties of the vegetation, such as LAI and
cerned with is captured together with, and stored in, car- total leaf biomass, which in turn influence productivity
bon compounds. Productivity is typically expressed as (Figures 15.6A and 15.8).
a rate per unit area (e.g., g C/m2/year). Productivity can vary greatly even among adjacent
The total energy (or carbon) fixed by producers in sites. Gaius Shaver and F. Stuart Chapin III (1991) com-
an ecosystem is called gross photosynthetic production, pared the relationship between production and biomass
or gross primary production (GPP). Since photosyn- among four contrasting types of tundra vegetation in
thesizing organisms use some of the total energy fixed, northern Alaska at the Toolik Lake LTER site (see Box
net primary production (NPP) is equal to the total ener- 12C). The study sites were chosen to represent extreme
gy captured (GPP) minus the losses to respiration by pri- examples of the wide range of plant growth forms com-
mary producers. The term “primary” indicates that we monly found in arctic tundra, and included tussock tun-
are concerned with the first trophic level in the system— dra, deciduous shrub-dominated riparian tundra, ever-
that is, the capture of solar energy and the fixation of green heath tundra, and wet sedge tundra (see Figure
atmospheric CO2 by autotrophs—in contrast to the pro- 19.23). Shaver and Chapin found that the four sites dif-
ductivity of herbivores, carnivores, or detritivores. In fered greatly in aboveground vascular plant biomass
practice, GPP and respiratory losses by primary pro- (ranging from 217 to 1877 g/m2), in addition to differ-
ducers are very difficult to measure directly on an ing strongly in plant growth form. Aboveground NPP of
ecosystem basis, and most data on productivity are col- vascular plants varied over an even greater range (32–305
lected as NPP. g/m2/year), and the element content and element
304 Chapter 15
(A) Figure 15.7

(A) Net primary production (aboveground, in kg of carbon
per m2/year) across Canada, based upon a simulation model
with input for leaf area index and land cover from satellite
data, plus meteorologial and soil data. (After Liu et al. 1997)
(B,C) Net primary production (aboveground, in g/m2/year)
for sites across several continents, as related to (B) mean
annual temperature and (C) precipitation at or near each
site. NPP is lowest in cold, dry ecosystems (tundra, desert)
and greatest where temperatures are high and moisture is
generally available to plants throughout the year (such as
tropical rainforests). NPP declines at extremely high precipi-
tation rates due to lack of oxygen in the soil and excessive
leaching of soil nutrients. (After Chapin et al. 2002.)
0.01 0.1 0.2 0.3 0.4 0.5

(kg m–2 yr–1)
(B) (C)
5000 5000
4000 4000
3000 3000
2000 2000
1000 1000
–15 –10 –5 0 5 10 15 20 25 30 0 2000 4000 6000 8000

Temperature (ºC) Precipitation (mm/yr)
requirements (such as N and P) also varied greatly ent cycling (e.g., turnover rates of N), the amount of
among the different sites. Among plants with different nutrients required to produce a gram of biomass, and
growth forms there were also dramatic differences in the total biomass turnover. Turnover time, or mean resi-
allocation of carbon and nutrients to the various plant dence time, is a measure of how rapidly materials move
parts, and a wide range in leaf replacement rates and the through a system. It can be used to describe systems at
amount of resources needed to produce each leaf. any scale—from an individual plant to an ecosystem to
Despite these dramatic differences among the four the atmosphere. Turnover time is equal to the total mass
different sites in total biomass, aboveground NPP, and of a component divided by its flux in or out of the sys-
plant nutrient content, Shaver and Chapin found sur- tem. The fraction lost each year, –k, is equal to 1/turnover
prisingly little difference among the sites in the rela- time in years, and the amount remaining after t years as
tionships between nutrients and NPP, including nutri- a fraction of the starting amount is e–kt.
1500 at the site—or are we interested in the actual productiv-

ity—the amount that is actually fixed in a given time peri-
od by the existing vegetation? The actual productivity
will always be less than potential productivity because,
for example, not all plant species are equally efficient at
fixing carbon, and factors such as competition might limit
1000 the growth of some species. In general, we are more con-
cerned with actual productivity, which we focus on here,

although making this distinction is important for sorting
through theories about how productivity determines
species diversity (see Table 20.4).
NPP can be estimated in a variety of ways. In the
500 simplest case, we would start the year with a patch of
bare ground. At the end of the year we would measure
the standing biomass (total aboveground plant mass) by
cutting down all of the plants, drying them, and weigh-
ing them. In practice, in perennial herbaceous ecosys-
tems, standing biomass is harvested from matched plots
at the beginning and at the end of the growing season,
0 5 10 15 20
with the difference in mass providing an estimate of the
Leaf area index (m2/m2)
amount of material produced. Losses to herbivory and
Figure 15.8 other factors must be estimated to obtain an accurate
Relationship between aboveground net primary production assessment of NPP. One way to estimate the magnitude
and leaf area index (defined as one-half the total leaf area of such losses is to exclude herbivores from other
per unit of ground area) for forests in the northeastern Unit- matched plots by fencing, insecticides, or other means.
ed States. (After Schlesinger 1997.)
Harvesting plant material to measure productivity
has two important limitations. First, harvesting stand-
The large differences in leaf replacement and allo- ing biomass is destructive. This method would not be
cation patterns that were so evident among different very useful, for instance, in a site that we wanted to
growth forms disappeared at the whole-ecosystem level, study over many years. One way around this limitation
because sites or growth forms with rapid leaf replace- is to measure productivity in a number of harvested
ment had slow stem or rhizome replacement. Thus, in plots while other measurements are made in adjacent,
these tundra ecosystems, nutrient storage, particularly undisturbed plots.
in stems and rhizomes, compensated for differences in Second, harvesting only aboveground plant parts
leaf replacement and resources used for leaf production. ignores belowground productivity, which is very impor-
The wide variation in leaf replacement and metabolism tant in many ecosystems. Measuring belowground pro-
among the different species and contrasting growth ductivity, however, is difficult. Not only is there a great
forms was buffered at the ecosystem level by stem deal of labor involved in successfully digging up roots,
processes, which dominated because stems were by far but it creates a substantial disturbance in a plant com-
the largest biomass and nutrient pools. munity under study. Furthermore, much of the root
material may be contained in very fine root hairs that are
Methods for Estimating Productivity hard to separate from the soil. For these reasons, below-
Measuring productivity and being able to predict its ground productivity is often ignored.
magnitude has great commercial value (for instance, in The ratio of belowground to aboveground NPP
determining timber production). It is also tremendous- varies greatly among ecosystems. Vogt and colleagues
ly important for understanding local ecosystems and as (1982), for example, compared aboveground and below-
a basis for understanding and predicting components of ground NPP in 23-year-old and 180-year-old Abies ama-
global carbon balance. While the concept of productiv- bilis (silver fir, Pinaceae) forest stands in the Cascade
ity may seem simple, actually measuring it in practice Mountains of Washington. In the young forest, below-
raises various questions and involves solving some sub- ground NPP was 45% of the total; in the old forest, the
stantive problems. belowground component was 75% of total NPP.
To begin with, if we wish to know how productivi- To get around these limitations, ecologists often use
ty affects community structure, do we want to know the indirect measures of productivity. One such approach
potential productivity of a site—the maximal amount depends on developing formulas relating changes in
of carbon that could be fixed given the abiotic conditions plant size to changes in biomass. For many tree species,
306 Chapter 15
especially economically important ones, the statistical Along with information on soil clay content and
relationship between average increases in stem diame- long-term weather information, these data are entered
ter and average annual biomass accumulation in stem into a simulation model. This method has been tested
wood, leaves, branches, and large-diameter roots has successfully on tropical grassland and woodland sites.
been determined by cutting and measuring trees. Such Its limitations include the requirement of a reasonably
allometric relationships are then used to calculate esti- good knowledge of the site’s climate and soil charac-
mates for NPP and for stores of carbon and nutrients in teristics, the assumption that the ecosystem is at a steady
the different tree components at different stages of for- state, as well as the need to account for the amount of
est development. Plant litterfall can be collected to carbon removed from the system each year by respira-
account for that aboveground NPP not retained by the tion, burning, logging, or crop harvest. Substantial soil
plants. Biomass and productivity are sometimes esti- erosion would also invalidate the assumptions of the
mated for shrubs and herbaceous understory plants by method. Under the right circumstances, however, this
similarly relating biomass increases statistically to method may prove to be an important complement to
increases in cover, with the relationships determined ini- other approaches to measuring NPP.
tially for harvested plants. These allometric formulas are Remote sensing methods (see Chapter 16) can be
widely used by forest ecologists and timber companies used to estimate NPP at larger scales than other meth-
because extensive data on such relationships are readi- ods. Images that record the different wavelengths of
ly available, since measuring tree growth rates is critical light reflected by the ground surface can be analyzed
for forestry and timber production. While this approach to provide an estimate of “greenness,” or the amount
works for well-studied forests, it is not practical for of chlorophyll present, because chlorophyll absorbs (and
many other ecosystems, because a formula is needed for reflects) light at specific wavelengths (see Chapter 2 and
each species in the community, and obtaining this infor- Box 16A). Since the amount of chlorophyll is related to
mation is time-consuming and expensive. the amount of live photosynthetic material, this infor-
A very simple, indirect estimate of productivity can mation can be used to estimate live biomass at a site.
be obtained by extrapolating from observed correlations Successive estimates of biomass can be used to obtain
between AET and productivity. AET has been suggest- an estimate of productivity.
ed as a simple, widely available correlate of potential The normalized difference vegetation index
productivity for examining large-scale productivity pat- (NDVI), measures standing crop based on the reflect-
terns (Rosenzweig 1968). While AET has been shown to ance at all the near-infrared (NIR) and visible (VIS)
provide fairly good estimates of productivity over a wavelengths:
broad range of climates, it is a poor predictor of pro-
NDVI = (NIR – VIS)/(NIR + VIS)
ductivity where precipitation and temperatures are both
high. Because it is inadequate as a surrogate for pro- where the reflectance is integrated over all the relevant
ductivity in tropical regions, and because the Tropics wavelengths. Regressions relating NDVI to NPP have
account for a large proportion of Earth’s total primary been worked out by relating the reflectance values to
productivity, AET is not very useful for predicting glob- those from actual harvests of vegetation. NDVI has been
al productivity. widely used, and provides good estimates of relative
A different approach to estimating long-term NPP, NPP in many systems (Figure 15.9; Cook et al. 1989).
based on analyses of soil organic matter, has recently Accuracy can be increased by measuring NDVI at mul-
been suggested (Jenkinson et al. 1999). This method has tiple times of the year. If an absolute measure of NPP is
the advantage of providing an estimate of NPP over a needed, it may be necessary to calibrate the measure-
much longer period of time than is usually possible with ments using on-the-ground harvests. This measure may
harvest data; it also includes both aboveground and be inaccurate in communities with many vegetation lay-
belowground productivity. It estimates NPP by calcu- ers. New methods are being developed to overcome
lating how much organic matter must enter a soil annu- some of the current limitations.
ally to maintain the existing pool of soil organic matter Another approach to the indirect estimation of pro-
at a steady state. This input of organic matter comes ductivity is the use of techniques based on micrometeo-
from sources such as plant material, soil fungi, dead ani- rology that measure concentrations of CO2 in the air.
mals and heterotrophic bacteria, and carbon fixed by soil These techniques have become increasingly important
autotrophs such as cyanobacteria. The three measure- in measuring productivity (and other plant-atmosphere
ments used are soil mass in each soil layer, the organic interactions) since the 1980s. The eddy covariance (or
carbon content of each soil layer, and the radiocarbon eddy correlation) method is the most widely used of
age of each soil layer based on 14C measurements (it is these methods. This technique measures net ecosys-
also possible to use data, if available, on the soil micro- tem exchange (NEE), which is the net transfer of CO2
bial biomass). into or out of an ecosystem .
30°E 40°E 50°E Greenness Figure 15.9

(at height of Map indicating the maximum NDVI,
growing representing greenness during the
season) height of the main growing season, for
20°N 20°
High
the “horn” of east Africa, 1982–1993.
These data can be monitored to identify
problems in the food production that
potentially lead to famine, flood, or
other food-insecure conditions in subsa-
haran Africa. The lowest values are in
the northern part of this region and in
10°N 10° Low the tip of the “horn” and other eastern
portions. (From U.S. Geological Survey
1996.)
0° Equator 0°
instance, it includes respiration by soil microorganisms).

However, NEE is in itself a valuable measure.
10°S 10°
Carbon Storage
30° 40° 50°
Over time, ecosystems typically accumulate and store
carbon. Net ecosystem production (NEP) is the net accu-
mulation of carbon per year by an ecosystem (Odum
1969). Our vast underground stores of coal, for example,
Eddies are parcels of air that move essentially as dis- are the result of geological periods over which NEP was
crete units, and within which gas concentrations are high. NEP is equal to NPP minus total heterotrophic res-
approximately uniform. Eddies carry CO2 into and out piration (primarily due to soil organisms). During the
of the ecosystem, which consists of a complex mix of car- course of the growing season, NEP is positive because
bon sources (such as respiring organisms) and sinks plant photosynthesis is far greater than heterotrophic
(primarily photosynthesizing plants). During the day- respiration. During the season when plants are dormant
time, by far the most important sink for CO2 is its uptake (winter or dry season) and photosynthesis is minimal,
by the photosynthesizing leaves in the ecosystem. As NEP is negative (Figure 15.10).
an eddy comes into contact with a surface, such as a leaf, NEP may also be defined as GPP minus total ecosys-
it slows down because of friction. At that point, CO2 can tem respiration (the sum of all autotrophic and het-
diffuse out of the eddy and into the boundary layer of erotrophic respiration), assuming that no other losses of
the leaf. When conditions are favorable, there will be a carbon occur. (Other losses of carbon besides respiration
net transfer of CO2 into the ecosystem from the sur- can result from leaching, plant volatile emissions, meth-
rounding air due to photosynthetic uptake by the canopy ane flux, fire, and logging; erosion or deposition and ani-
(that is, gross uptake minus losses due to respiration in mal movements can add carbon to or subtract carbon
the ecosystem). from a system as well.) NEP generally depends more
During the night, there will be a net transfer of CO2 on the length of time that an ecosystem has remained
out of the ecosystem into the atmosphere, due primari- undisturbed than on temperature or moisture, and is gen-
ly to respiration by the plants and soil organisms in the erally small: undisturbed ecosystems typically show
ecosystem. Using the eddy covariance method requires small positive accumulations of carbon each year.
intensive measurements of local wind speeds and veloc- Over the course of ecosystem development follow-
ities as well as local gas concentrations. It is highly instru- ing a major disturbance, the biomass of woody tissue in
ment- and computer-intensive; most systems are mount- long-lived plants increases. The sum of all organic matter
ed on either permanent or movable instrument towers in in living vegetation (including the nonliving structural
forests, although some have been mounted on aircraft. material in wood)—the total standing biomass—may
Because NEE includes all sources and sinks for CO2 in become very high, particularly in older forests, where car-
the ecosystem, it is not a direct estimate of NPP (for bon gradually accumulates in long-lived trees over many
308 Chapter 15
GPP ly among ecosystems. Flux rates among the components

Ecosystem
carbon uptake within ecosystems also vary considerably. For example,
NEP in a grassland system, the turnover rate for undecom-
+
posed litter was two orders of magnitude greater than
Carbon exchange
that for humic acids in the lower parts of the soil pro-
0
file, while the amount of carbon stored in the humic acids
was much greater than that in the litter (Figure 15.11).
NPP tends to be closely tied to soil microbial respi-
– RE ration because the carbon and energy fixed by plants is
Ecosystem
carbon loss the substrate for microbial respiration, and because
microbial respiration releases nutrients that plants need
in order to capture carbon by photosynthesis (as well as
Winter Midsummer Winter to survive and grow). In some cases, however there is
Figure 15.10 no strong link between NPP and soil microbial activity,
Net ecosystem production, NEP, is the net accumulation of and NEP rises or falls accordingly. For example, in peat
carbon per year by the entire ecosystem, including living bogs, the anaerobic soil environment severely limits
organisms and soil carbon. NEP increases during the spring microbial activity, but does not strongly limit plant pro-
to a peak in midsummer, because photosynthesis by plants
is greater than total ecosystem respiration. Ecosystem respi-
ductivity, resulting in high carbon accumulation in the
ration is minimal during the winter. It increases in the soils, and thus high NEP. Disturbances such as fire or
spring, causing larger ecosystem carbon losses as soil tem- logging can result in substantial losses of carbon from
peratures warm. The ecosystem reaches a peak of carbon an ecosystem. These disturbances remove living plant
loss in late summer or early autumn. As GPP declines in the biomass and sometimes soil organic matter as well, and
autumn, NEP also declines, even though losses due to respi-
ration are also being reduced. (After Chapin et al. 2002.)
can alter subsequent NPP in complex ways.
Models of Ecosystem Carbon Cycles

Models have been developed to describe and predict
centuries. Eventually, the allocation of carbon to new fluxes and pools of carbon from the ecosystem to the
woody tissue is balanced by losses due to the deaths of global scale. These models typically include the pools of
individuals and plant parts, and the ecosystem no longer carbon in different fractions in the soil and in vegetation
gains additional standing biomass. The amount of carbon as well as the fluxes of carbon from the atmosphere to
in the ecosystem may continue to rise after that time as plants (NPP), from plants to the soil, and from the eco-
soil organic matter continues to accumulate. system (plants and soil) back to the atmosphere. These
The relative proportions and total amounts of car-
bon stored in different ecosystem components vary great-
Soil respiration
0.41
Aboveground 0.004
litterfall 0.005
0.16 0.419
0.26 Undecomposed
Root litter Turnover in
Roots 0.5 kg C/m2 10s of years
turnover
0.80 kg C/m2
0.004
Fulvic acids Turnover in

2.3 kg C/m2 100s of years
0.003 0.003
Figure 15.11
Pools of carbon in various components in a grassland soil (in Humins Humic acids Turnover in
kg C/m2/year, in boxes), and relative rates of turnover (in 3.8 kg C/m2 3.8 kg C/m2 1000s of years
kg C/m2/year, numbers next to lines showing change). Soil
humus (non-cellular organic matter) includes various com-
ponents (humins and humic acids) with very long turnover Permanent accumulations
times. (After Schlesinger 1997; data from Schlesinger 1977.) in the lower profile
models vary from being highly detailed to very coarse quences for NPP and other variables. The modeling of
and general, require different inputs of data, and have carbon fluxes is critical to our understanding of and our
different particular objectives. Waring and Running ability to predict human effects on the global carbon
(1998) discuss different ecosystem modeling approach- cycle (see Chapter 22).
es and compare some of the major models.
One example is the CENTURY model (Parton et al. Nitrogen and the Nitrogen Cycle
1988, 1993), which began as a model of carbon storage
in agricultural grasslands but has since been extended
at Ecosystem and Global Levels
to model carbon cycling in many other biomes (Figure In striking contrast to the water cycle, living organisms
15.12). It is a highly detailed model that requires a great other than plants are important at many junctures in the
deal of information about the system and produces very nitrogen cycle. Water and nitrogen are similar, though,
complex simulations. While there are advantages to this in that they are both commonly limiting to individual
approach, one disadvantage is that, due to the model’s plant growth and to community and ecosystem pro-
complexity, it may be difficult to understand how a par- ductivity (Figure 15.13). Water and nitrogen cycling are
ticular factor acts to produce a particular result. also functionally linked.
The various models in use incorporate different Nitrogen has a fairly rapid flux through living
approaches to predicting how environmental factors organisms coupled to a large global pool with slow
may affect ecosystem processes, as well as the conse- turnover—that is, the N2 in the atmosphere. Nitrogen
enters the biosphere from the atmosphere through the
process of nitrogen fixation. Eventually it is returned to
Structural the atmosphere by the process of denitrification (and
carbon by other means, including fires). The balance between
(3 years) nitrogen fixation and denitrification maintains the con-
centration of N2 in the atmosphere at equilibrium
Plant
A 1–A residue (Schlesinger 1997). Nitrogen is recycled among living
things at a relatively rapid rate, so primary productivi-
L/N ty does not depend solely on nitrogen fixation.
Metabolic
carbon
(0.5 years) Nitrogen Fixation
CO2 CO2 Very few organisms—only a relatively small number
0.45 0.55
SL 0.55
of prokaryotes—have the ability to fix atmospheric N2,
BL
creating NH4+ (ammonium) by breaking the double
Active soil
F(T) + (0.85 – 0.68T)
bond between the two atoms in gaseous N2. These nitro-
carbon
CO2
(1.5 years)
CO2 0.42
CO2 10,000
0.55
0.3 0.03
Slow soil 0.004
carbon 1000
(25 years) 1 – F(T) – 0.004
NPP (g C m–2 y–1)
CO2
100
0.55
Passive soil
SL Surface litter
carbon
(1000 years) BL Soil litter 10
L/N Lignin to
nitrogen ratio
A Lignin fraction
F(T) Soil silt + clay 1
content (fraction) 10–2 10–1 100 101 102 103
–2 –1
Nitrogen input (g N m y )
Figure 15.12
The CENTURY model of carbon flow through the biosphere. Figure 15.13
The fractional numbers indicate the proportion of carbon Net primary productivity versus annual nitrogen input for a
moving along each pathway. The boxes represent carbon wide range of terrestrial, freshwater, and marine ecosystems.
reservoirs, and turnover times are given for each reservoir. NPP reaches an asymptote at about 10 g N/m2/year. (After
(After Parton et al. 1988.) Schlesinger 1997.)
310 Chapter 15
Figure 15.14
Traditional rice paddy culture in Mai Chau, Vietnam.
Rice is the primary crop and primary export of Viet-
nam. An important component of the nitrogen econ-
omy of paddy agriculture is the symbiosis between
the freshwater fern Azolla and the nitrogen-fixing
cyanobacterium Anabaena (see Figure 4.12). (Photo-
graph courtesy of M. Harms.)
gen-fixing microorganisms exist both in free-living forms expensive energetically, plants that are capable of sup-
and in symbiotic relationships with the roots of a few porting nitrogen-fixing bacteria cease to do so when soil
plant species, as described in Chapter 4. The most com- nitrogen is abundant, instead obtaining nitrogen from
mon nitrogen-fixing symbioses are those between bac- the soil. This is the case, for example, in agricultural
teria in the genus Rhizobium and members of the Fa- fields fertilized with nitrogen.
baceae (legumes; see Figure 4.13) and that between an
actinomycete, Frankia, and alder (Alnus spp.), Ceanoth- Other Sources of Nitrogen Input to Living Organisms
us, and a few other tree and shrub species. Free-living In addition to the atmospheric N2 that is fixed by prokary-
nitrogen fixers are found in soil, in aquatic environ- otes, nitrogen that is in suspension in the atmosphere in
ments, and in anoxic environments such as mud. more available forms can enter terrestrial ecosystems in
A particularly important group of free-living nitro- several different ways. The major forms of N deposition
gen fixers is the cyanobacteria (the microorganisms for- are wet deposition of nitrogen dissolved in rainwater, dry
merly known as blue-green algae). These photosynthet- deposition of nitrogen in particles and dust, and fog or
ic bacteria are abundant as free-living organisms in the cloud deposition of nitrogen dissolved in water droplets
oceans, in fresh water, and on soil surfaces. Cyanobac- onto plant sufaces (Figure 15.15). Nitrogen in a wide vari-
teria are also important partners in various nitrogen-fix- ety of forms, including nitrate (NO3–), ammonium (NH4+),
ing symbioses. Lichens are symbiotic associations organic N, and gaseous forms (NH3, HNO3, or NO2), can
between a photosynthetic species and a fungal species. be transferred to plants by deposition.
The photosynthetic species can be cyanobacteria or Much of this atmospheric nitrogen deposition
members of the Chlorophyta (green algae); lichens con- results from human activities, and it has increased enor-
taining cyanobacteria can fix nitrogen. mously in recent decades, with potentially severe envi-
Another nitrogen-fixing symbiosis is that between ronmental consequences (Vitousek et al. 1997). In the
Azolla, a tiny aquatic fern, and Anabaena (or other cyano- eastern United States, for example, atmospheric depo-
bacteria) in rice paddies and other freshwater tropical sition of N is now 10 times greater than it was before
systems (Figure 15.14). This Azolla-cyanobacterium sym- industrialization, while in northern and central Europe
biosis can be a major source of nitrogen in the ecosystems it is 20 times greater. The nitrogen enters the atmosphere
where it occurs, including traditional rice production largely from the burning of fossil fuels, particularly from
(where it is deliberately maintained by farmers). automobile exhausts and industrial sources. Humans
Nitrogen fixation uses a great deal of energy. The also add about 80 million metric tons of nitrogen per
symbiotic nitrogen fixers obtain this energy in the form year to the biosphere by fixing atmospheric N for nitro-
of carbohydrates from their host plants, supplying nitrogen fertilizer, which is a large source of the N found in
gen to the plant in return. Free-living nitrogen fixers runoff and groundwater (Figure 15.16).
(other than cyanobacteria) must obtain carbon and ener- Volatilization of methane and ammonium by domes-
gy from other sources, such as soil organic matter. It ticated cattle (from flatulence and decomposing manure)
takes a great deal more energy for a plant to support is a substantial source of anthropogenic N particularly
nitrogen-fixing bacteria than to obtain NH4+ from the in some ecosystems, but also globally. Nitrogen deposi-
soil, as long as NH4+ is readily available. Because it is less tion by humans is spatially highly variable; it is usually
3.9 × 1021
Atmospheric
nitrogen
20 Nitrogen
Industrial 50
fixation Human Fires compounds Denitrification
80 activites in rain + dry 110
3.5 × 1015 deposition
60 Fixation
by
Fertilizer lightning
factory
Biological 3
≤ 140 fixation
140
Excreta and
Denitrifiers dead bodies Biological
(bacteria) Uptake Plant Leaching + fixation
1200 residues Symbiotic runoff 15
fixation 36 Oceans
Immobilization Soil
Nitrates N carried
by bacteria organic matter Free-living N2 by rivers
and fungi fixers (bacteria)
~120 × 1015
Mineralization 8000 Internal
Mineralization (ammonification cycling
Leaching to by bacteria)
groundwater Nitrites
11 Ammonium Long-term
burial 570 × 1015
salts
Figure 15.15
The global nitrogen cycle, showing some of the major fluxes (in 1012 g N/year,
shown in regular type) and pools (shown in boxes; units as indicated). The atmo-
sphere is by far the largest pool of nitrogen. (Data from Schlesinger 1997 and Taiz
and Zeiger 1998.)
much greater near industrialized urban centers and where

prevailing winds deposit nitrogen from those areas. Not
surprisingly, dumping huge amounts of nitrogen on nat-
ural systems can have striking effects on ecosystem func-
100 tion. High rates of nitrogen deposition have been shown
to have disruptive effects on forests, streams, and coastal
waters and estuaries (Aber 1992; Aber et al. 1998) and can
lead to general forest decline and dramatic changes in
Percent of 1988 value
stream ecology (Schlesinger 1997).

75
Nitrogen Mineralization
Most of the nitrogen used by plants comes from the recy-
cling of N from decaying biomass (see Figure 15.15). The
50 Forest initial stages of decomposition transform these materi-
destruction als into soil organic matter. Further decomposition
Global releases nutrients via a set of processes collectively called
Industrial
human
N fertilizer
mineralization, in which microorganisms release car-
population bon as CO2 and nutrients in inorganic form. The nutri-
25
1700 1800 1900 1975 1990 ents may then be taken up by plants directly, or they may
Year be taken up after further conversions occur. The process
Figure 15.16 of N mineralization begins with ammonification, by
Global changes in forest destruction, human population size, which nitrogen in organic compounds, such as proteins
and industrial fertilizer production from 1700 to the late and nucleic acids, is released in the form of ammonium
1980s. The chart shows the year in which each variable ions (NH4+). The NH4+ ions may be directly taken up by
reached 25%, 50%, and 75% of its value at the end of this time
period. Industrial production of nitrogen fertilizer has grown some plants, or may undergo nitrification to nitrite
exponentially since the late 1940s. (After Vitousek et al. 1997; (NO2–), and nitrate (NO3–) (Figure 15.17). Ammonifica-
data from Kates et al. 1990.) tion can be carried out by a variety of soil-dwelling
312 Chapter 15
Figure 15.17
Atmosphere N2 Atmosphere Nitrogen cycle within a soil ecosys-
tem, including the soil surface, indi-
N2 fixation cating many of the major components
Nitrifi
cati of nitrogen and types of transforma-
on
tions between components. (After
on
Waring and Running 1998, with data
ti
ica
Plant N from Davidson et al. 1992, Schlesinger

rif
e
iza 1997, and Drury et al. 1991.)
Nit
rh
cor
my
via
Pl
on
e
ati c.
an
tak
liz , et
tu
ra ss
ine f lo up
pt
a
M , le
ak
nt
ath
e
Pla
De
Dentrification
Non-living Nitrification NO2–
NH4+ N2
organic N Mineralization NO3–
Dentrification
Immobilization Immobilization
N2O
Death,
exudation, etc.
Microbial
biomass N
microorganisms, but nitrification is carried out only by require oxygen (i.e., they are obligate aerobic organisms).
specialized nitrifying bacteria (or nitrifiers) and a few The two groups are generally found together in soils,
other organisms. with the result that high levels of NO2– do not often accu-
In ammonification, heterotrophic microorganisms mulate. There are also a few other organisms that can
release N from organic matter by breaking the chemi- carry out nitrification. They belong to a variety of het-
cal bonds in carbon compounds to release fully reduced erotrophic groups, including bacteria and fungi. They
ammonium ions (NH4+) in a series of enzymatically con- utilize NH4+ to produce NO3– or NO2–, obtaining energy
trolled steps. Various things can happen to the nitrogen from the breakdown of carbon compounds in soil organ-
released by these microorganisms, including being used ic matter. These organisms may be particularly impor-
by the same or other microorganisms or by plants. tant in soils that are low in nitrogen.
Ammonium ions that are “immobilized” by being taken The most important immediate factor limiting nitri-
up by microorganisms (and thus converted back into fication rates is the amount of available NH4+. The fac-
organic form) can be recycled one or more times through tors limiting nitrification therefore include anything that
microorganisms before being ultimately released to be affects the production of ammonium as well as its immo-
taken up by plants, bound to soil minerals or soil organ- bilization, uptake by plants, and capture on clay parti-
ic matter, volatilized to ammonia gas (NH3) and lost, or cles and organic matter. Nitrification rates are also affect-
oxidized. ed by factors that control the population size and
Ammonium ions are relatively immobile in the soil physiological activity of nitrifying bacteria, particularly
and are likely to be taken up only by roots in close soil water content, but to some extent temperature, pH,
proximity. The rate at which ammonification occurs is and characteristics of the soil and the plant materials.
determined by a variety of environmental factors, partic- Soil water content is important because very dry soils
ularly temperature and moisture and the chemical char- limit the growth and activity of the bacterial populations,
acteristics of the soil (including pH and other factors). while overly wet and waterlogged soils limit these aer-
In many soils, much of the NH4+ produced under- obic microorganisms by excluding oxygen.
goes further transformation by being oxidized to NO2–
and NO3– through the process of nitrification. Bacteria in Denitrification and Leaching of Nitrogen
the genera Nitrosomonas (which convert NH4+ to NO2–) Unlike positively charged ammonium ions, negatively
and Nitrobacter (which oxidize NO2– to NO3–) are the two charged nitrate ions do not readily bind to soil miner-
most important nitrifiers. These bacteria are chemoau- als and organic matter, and are therefore highly mobile
totrophs that obtain energy from the oxidation of NH4+ in the soil solution. Nitrate ions in the soil solution can
or NO2– and fix carbon from atmospheric CO2. They be transported some distance to soil roots both by dif-
fusion and by mass flow. Because they are so mobile, those from deciduous plants and plants from fertile sites.
they can also be easily leached from the soil into ground- The leaves of many oak species, which are high in mate-
water and runoff, and thus are readily lost. (In some rials that physically and chemically resist decomposition,
tropical soils, however, anions can be adsorbed onto pos- are incorporated into soil organic matter much more
itively charged aluminum or iron oxides and thus pre- slowly than, for instance, maple leaves from the same
vented from leaching.) Leaching of nitrate from fertil- site. Leaves of nitrogen-fixing species decompose more
izer and animal waste can be an important source of quickly than those of co-occurring nonfixers, with conse-
water pollution. quently less immobilization of soil nutrients. Decompos-
Nitrate is also lost from ecosystems by the process ing animal tissues have much lower C:N ratios, and tend
of microbial denitrification, which is the reduction of to decompose much more rapidly, than plant tissues.
NO2– and NO3– to various gaseous nitrogen compounds,
including N2 and N2O. The latter is an important green- Plant Uptake of Nitrogen
house gas that can contribute to global warming (see The balance between NO3– and NH4+ in soils differs
Chapter 22). The composition of the denitrifying bacte- among habitats, largely as a result of environmental
rial community can differ strikingly between disturbed effects on nitrification. While most plant species are bet-
and undisturbed soils, and these differences can be ter able to take up NO3–, plants adapted to environments
reflected in very different responses to environmental where soil N is largely in the form of NH4+ (i.e., where
conditions and in different rates of denitrification and nitrification is inhibited, as in cold, waterlogged soils)
production of N2O (Cavigelli and Robertson 2000). may show a preference for uptake of nitrogen in that
Fire can be a major cause of nitrogen loss in some form. In most plants, both forms of nitrogen are con-
ecosystems. Most of the N released into the atmosphere verted in the roots to amino groups (NH2–), which are
during a fire comes from the vegetation itself, but sub- then attached to various organic compounds (NH4+ is
stantial amounts can come from the burning of accu- toxic if it accumulates in plant tissues).
mulated litter, and very hot fires can burn the soil organ- The conversion of NO3– to amino groups is ener-
ic matter in the uppermost part of the profile, thus getically expensive. The conversion of NH4+ to NH2–
volatilizing nitrogen from the soil. requires much less energy. However, NO3– is transport-
ed to plant roots readily by mass flow in the soil solu-
Decomposition Rates and Nitrogen Immobilization tion, but the lower mobility of NH4+ in the soil may
As soil organic matter is decomposed by microorgan- require greater energy expenditure in the form of root
isms, much of the N and P they contain is incorporated proliferation (Schlesinger 1997).
into the living microbial biomass (see Figure 15.15) The Nitrogen uptake can be energetically costly for
ratio of C to N in plant tissues ranges from about 25:1 to plants for a number of reasons. The uptake of nitrogen
150:1, but the ratios are much lower in fungi (between as either NO3– or NH4+ is aided by enzymes that expend
4:1 and 15:1) and decomposer bacteria (3:1–5:1). As energy to transport the ions across cell membranes in the
decomposition occurs, soil C: N (as well as C: P) ratios roots. Active root growth and root proliferation into
decline as the biomass in decomposing plant material is areas of the soil where N is more available also requires
replaced by biomass in the fungi and bacteria growing energy. In environments where N is less available, plants
on it. When more of the soil N and other nutrients (such tend to allocate more biomass to roots, and root:shoot
as P) become incorporated into living soil microorgan- ratios are generally high, compromising the plant’s
isms as organic molecules, these nutrients become less capacity to capture carbon.
available for uptake by plants. This phenomenon is Uptake of positively charged ions, such as NH4+ and
called biological immobilization of the nutrients. many other nutrients, such as K+, from the soil could
Eventually the microorganisms die, and some of the potentially lead to a charge imbalance in plant tissues. To
nutrients again become available for uptake by plants. counteract the positive charges, plant roots release hydro-
However, nitrogen immobilization by microorganisms gen ions (H+) into the soil solution, acidifying the soil.
may be a very important short-term limitation on plant When plants take up NO3–, they release negatively
nutrient uptake and growth. This effect can be particu- charged ions such as organic acids and HCO3– to main-
larly strong when the plant tissues that are decomposing tain their charge balance. The release of these ions in turn
have relatively high C: N ratios or are rich in plant mate- affects soil chemistry and the availability of nutrients to
rials (such as lignin) that resist decomposition. High C: plant roots and soil microorganisms. Some ions (e.g., Ca2+
N ratios and the presence of lignin and similar materi- and Na+) are actively excluded by the roots of some
als result in slower decomposition rates and a longer plants in environments where their availability is too
period of N and P immobilization. For instance, ever- high.
green leaves and leaves from plants native to low-nutri- In some unusual cases, plants obtain nitrogen from
ent environments typically decompose more slowly than sources other than NO3– and NH4+ in the soil solution.
314 Chapter 15
Figure 15.18
Sarracenia purpurea (purple pitcher plant,
Sarraceniaceae) in a bog in northern New
York. Bogs are low in nutrients. The
pitcher-plant obtains substantial
amounts of nitrogen by absorption from
detritus captured in its pitcher-shaped
leaves. The detritus comes from insects
that fall into the leaves, but cannot crawl
out because of downward-pointing tri-
chomes (hairs), and drown. Decomposi-
tion is carried out by a community of
bacteria and invertebrates living in the
water that collects in the “pitcher.” (Pho-
tograph by S. Scheiner.)
“Carnivorous” plants native to extremely nitrogen-poor Chapter 4 for a discussion of how mycorrhizal fungi and
bogs may obtain nitrogen from decomposing insects other organisms can access this phosphorus).
(Figure 15.18). Other species may take up nitrogen in the The original source of most of the phosphorus that
form of amino acids directly from the soil. enters the biosphere is weathering of rock containing
apatite minerals. The minerals in the apatite group are
the only minerals common in the lithosphere that con-
Phosphorus in Terrestrial Ecosystems
Although phosphorus is needed for plant growth and
function in smaller amounts than other macronutrients
Living
(see Table 4.2), it is fairly common for plant growth to be organisms
limited by P availability. We considered P availability in
the soil and its uptake by plants in Chapter 4. Here we
discuss P at the ecosystem and global levels.
Most of the phosphorus that cycles through ecosys-
Soil
tems is found in microorganisms, in decomposing plant organic P
material, and in soil organic matter—that is, in organic
material (Figure 15.19). This is curious, because most of
the P in terrestrial systems occurs as minerals in rocks.
Stable
Inorganic P is often complexed with other soil minerals organic P
and is relatively unavailable to plants directly (but see Soluble P
Labile P
in soil water
Primary P
Secondary P
minerals
minerals
(from rock)
(inorganic)
Figure 15.19
Phosphorus from minerals in rock (primary P) dissolves in the soil
solution, from which it can be absorbed by plants and other living Leaching loss
organisms. Some soluble soil phosphorus is transformed into vari- to groundwater Occluded P
ous inorganic forms or is leached; however, the vast majority of the (inorganic)
phosphorus taken up by organisms in an ecosystem has been recy-
cled through other living organisms. Biological transformations
are shown in green, geochemical transformations in white. (Modi-
fied from Schlesinger 1997, after Smeck 1985.)
tain a substantial amount of phosphorus. As rock weath- Ecosystem Nutrient Cycling

ers and soils form, and as the soils develop and age,
and Plant Diversity
much of the phosphorus becomes bound in the interior
of iron and aluminum oxide crystals, from which liv- Is there any relationship between the characteristics of
ing organisms cannot remove it. a plant community, such as diversity, and ecosystem
In old soils that have become dominated by iron and processes such as nitrogen and phosphorus cycling?
aluminum oxides (such as those that are common in Recently considerable effort has been made to determine
Africa, Australia, and other environments, particularly and understand the potential connections between these
in the Tropics), the lack of available phosphorus is one important ecosystem and community properties (Lore-
of the most important environmental factors affecting au et al. 2001; Tilman et al. 1997; see Chapter 14).
plants. Inorganic P may also be held in more available David Hooper and Peter Vitousek (1998) set out to
forms on the surface of other soil minerals by anion answer this question experimentally in a California
adsorption and other reactions. The complicated soil grassland. The grassland was located on serpentine soil
chemistry involved in these reactions depends on soil (Box 15A). The plants were categorized into four func-
pH and other factors. It is therefore generally difficult tional groups that differed in phenology (seasonal
for scientists to accurately measure the amount of phos- growth pattern), root:shoot ratio, and other traits rele-
phorus available to plants. vant to the use and cycling of N and P. The four func-
Unlike the global cycles of carbon, nitrogen, and tional groups were early-season annual forbs (herba-
water, the global cycle of phosphorus does not have a ceous plants), late-season annual forbs, nitrogen fixers,
major atmospheric component (Figure 15.20). Most of and perennial bunchgrasses. Experimental plots con-
the phosphorus that is used by organisms has been recy- tained plants of either a single functional group or com-
cled in organic form through other living organisms, binations of functional groups. The researchers meas-
although its ultimate source is rock weathering. Organ- ured pools and fluxes of nitrogen and phosphorus, as
ically bound phosphorus is released in the form of PO43– well as other ecosystem variables, on the experimental
(orthophosphate) by plant roots, fungi (including those plots.
in mycorrhizae as well as free-living fungi), bacteria, and Hooper and Vitousek found that plant functional
algae. All of these organisms mineralize phosphorus group diversity increased the total use of nutrients, as
through the action of extracellular phosphatases they had hypothesized, because of seasonal differences
(enzymes that cleave ester bonds). This process is expen- in plant growth and resource use among the functional
sive both energetically and in terms of the nitrogen groups. They also found, however, that functional group
invested in the phosphatase enzymes. Microorganisms richness did not decrease nitrogen leaching, although
as well as plants can take up the PO43– that is released, much more nitrogen was lost when plants were totally
but if it is not immediately taken up, the orthophosphate absent than when plants of any of the functional groups
does not remain available for long in the soil. Rather, it were present. Ecosystem nutrient retention was deter-
readily binds to organic particles or to soil minerals,
becoming partially or completely unavailable to plants.
Figure 15.20
The global phosphorus cycle, showing major fluxes (in 1012 g
P per year, shown in bold) and pools (in 1012 g P). Marine
1.0 sediments, at approximately 4 × 10 21, are by far the largest
3000 Dust transport
pool, and cycling by land plants, at 3000 × 1012 g P per year
Land plants
is the largest flux. (After Schlesinger 1997, with data from
Jahnke 1992.)
Internal 60
cycling 12
River flow
Mining 21
Mineable Reactive Ocean

Soils
rock 90,000
200,000 2
10,000 Bound Burial
Internal
1000 (sedimentation)
19 cycling
2
Sediments
4 × 109
316 Chapter 15
BOX 15A
Serpentine Soils
Important ecological research has been species, surrounded by grasslands. Ser- es. In addition, Kruckeberg (1954)
conducted over the years on various pentine vegetation is often very high in showed that the serpentine species
aspects of the ecology of vegetation on species diversity and may have many appear to be excluded from the sur-
serpentine soils (Kruckeberg 1954; Whit- endemic species. The species from surrounding regions by competition (echo-
taker 1954; Hooper and Vitousek 1998). rounding areas are unable to tolerate ing Bradshaw’s findings for plants
Serpentine soils are derived from ser- the toxic, low-nutrient soils, and this growing on mine tailings, discussed in
pentine rock, which is a magnesium sil- excludes them from serpentine patch- Chapter 6).
icate rock of metamorphic origin. These
soils are poor in many nutrients—par-
ticularly calcium, nitrogen, and phos-
phorus—and they may have either very
low or very high pH. Toxic elements,
particularly heavy metals, are also char-
acteristic of serpentine soils.
Areas with serpentine soils occur in
many places in the world. Serpentine
soils often occur in patches of limited
extent, or “islands,” surrounded by
more ordinary nonserpentine soils.
These serpentine islands (sometimes
called serpentine barrens) typically sup-
port distinctive plant communities that
contrast strongly with the vegetation of
the surrounding region. For instance,
serpentine vegetation in California
often contains many showy dicot
Serpentine grassland in the state of California.

(Photograph courtesy of D. Hooper)
mined at least as much by the indirect effects of plants Sulfur

on microbial activity and subsequent nutrient retention Sulfur is necessary primarily for building the amino
by microorganisms as by the indirect effects of plants acids cysteine and methionine, important constituents
on microbial activity. The researchers concluded that the of various proteins. Like nitrogen, phosphorus, and car-
identity of the particular plants present explained more bon, sulfur is cycled biogeochemically in ecosystems.
about ecosystem nutrient cycling than could be explained The weathering of rocks with minerals containing sul-
by considering the number of functional groups present. fur (such as pyrite and gypsum) is one major source of
the sulfur that ultimately becomes available to plants.
Ecosystem Processes Atmospheric deposition is also an important source of
sulfur. A large proportion of the sulfur pool in soils exists
for Some Other Elements as a part of soil organic matter. Inorganic sulfate ions
We have examined the major ecosystem processes of pri- (SO42–) are adsorbed onto soil mineral particles, in equi-
mary importance for plants: those involving water, car- librium with SO42– dissolved in the soil solution. Bacte-
bon, nitrogen, and phosphorus. Other nutrients, of course, rial mineralization and immobilization of sulfur are
are necessary for plant growth (see Table 4.2). Here we analogous to the same processes for nitrogen and phos-
briefly discuss the cycling of two other important ele- phorus.
ments, sulfur and calcium, both as general examples and Sulfur is lost from ecosystems in a variety of forms.
because of the influence of human activities on their In addition to the loss of sulfate ions from ecosystems
cycles. by leaching, many plants release volatile organic sulfur
(particularly H2S), and waterlogged, anaerobic soils can sequent litter decomposition, as well as throughflow
also produce large quantities of sulfur gases by bacteri- (rainwater) that passes through the canopy and travels
al reduction. down the surface of tree trunks, carrying dissolved ions
On a global scale, large pools of sulfur are dissolved leached from the plant surfaces. Some calcium is carried
in seawater. In the atmosphere, sulfur-containing gases in particulate form in the atmosphere and deposited in
are short-lived and are found at low concentrations. rainfall or by dry deposition (Figure 15.21).
However, dust storms, sea spray, and volcanic eruptions The requirement of plants for calcium is generally
can contribute large amounts of particulate sulfur (as very high, second only to that for nitrogen (see Table 4.2).
well as some sulfur in the form of gases and aerosols) to Calcium is unusual in that it plays both chemical and
the atmosphere. Human activities are the largest current structural roles in plant tissues, including regulation of
source of atmospheric sulfur gas, although recent air pol- growth and response to stress, stomatal function, cell
lution controls have decreased the yearly input of sulfur division and cell wall synthesis, and structural support
from industrial sources considerably. The SO2 that enters in both leaves and wood (McLaughlin and Wimmer
the atmosphere reacts with water to form sulfuric acid, 1999). Decreasing calcium availability may lead to de-
which, along with nitric acid (from anthropogenic NO creased growth and higher mortality in forest trees, mak-
and NO2), is the chief cause of acid precipitation (see ing them more susceptible to pathogens, for example.
Chapter 22). Some recent evidence suggests that calcium deple-
tion is occurring in many forests, with negative conse-
Calcium quences for forest vitality (Likens et al. 1998), although
Calcium is abundant in the Earth’s crust. It is an impor- other researchers have disputed these findings (Yanai et
tant component of limestone and other common rocks al. 1999). The potential for calcium depletion is a result
such as gypsum (CaSO4 •H2O) and calcium carbonate
(CaCO3). Weathering of these rocks releases calcium ions
(Ca2+), which are carried by rivers to the oceans; large
pools of calcium ions are found in seawater. Marine
invertebrates use this calcium to build calcium carbon-
ate shells; when these organisms die, their remains form
thick sediments. Through burial, metamorphosis, and
crustal movement, these marine sediments may ulti-
mately be transformed into calcium-bearing terrestrial 34
rock.
Calcium ions readily dissolve in the soil solution and Precipitation
11
are taken up by plants passively in the transpiration
stream. Calcium concentrations can vary dramatically
among soils, from being apparently limiting to plant 23
growth to being present in excessive amounts. In humid
Leaf
regions with low-pH soils, calcium ions can be heavily litter
leached from the soil. Huntington (2000) studied fluxes Branch
17
of calcium in southeastern U.S. forests, and found that litter, etc,
losses of calcium caused by logging and soil leaching 122 12
often were greater than inputs from atmospheric depo- Leaching
sition and rock weathering, leading to severe calcium 12 14 19 Leaching
Herb
depletion in the forest soils. His results suggest that cal- litter
5
cium depletion could result in widespread problems for 93 Litter 141
southeastern forest trees over the next few decades.
Decomposition
In contrast, in many deserts, calcium can accumu- 43
late in the soil. Rainwater dissolves the calcium ions and
then evaporates, precipitating and thus concentrating Uptake Soil 437
41
the calcium into either a soft or a cement-hard layer Groundwater loss
called a calcic horizon (also called a “pan” or “caliche”). 11
The excessive amounts of calcium in these calcareous,
high-pH soils can be actively excluded by the roots of Figure 15.21
Local calcium cycle within a forest ecosystem in the United
some plant species. Kingdom. Pool amounts are given in kg/ha, and annual
Within an ecosystem, the calcium contained in plant fluxes (arrows) are shown in kg/ha/year. (After Schlesinger
biomass is returned to the soil through litterfall and sub- 1997 and Whittaker 1970.)
318 Chapter 15
of several factors, including decreasing amounts of cal- Net primary productivity (NPP) the total energy
cium in rainfall and removal of calcium from the ecosys- captured minus respiratory losses by primary produc-
tem in biomass (by logging and other human activities). ers, is a critical component of many ecosystem process-
Most important are the direct negative effects of anthro- es. Recent technological developments have made it pos-
pogenic nitrogen deposition (particularly in the form sible to obtain far better measurements of productivity
of HNO3), which results in displacement of calcium ions than was possible in the past. Nevertheless, we are still
in the soil, causing them to be leached into stream water not able to model the carbon cycle with full confidence,
and lost. although doing so is important for predicting the con-
sequences of deforestation and CO2 emissions from fos-
sil fuels. The carbon that is captured in NPP accumulates
Summary
in an ecosystem primarily in living biomass, plant litter,
Ecosystem ecologists study the flow of energy and mate- and soil organic matter, and is lost from the ecosystem
rials through ecosystems. In terrestrial ecosystems, primarily through soil microbial respiration.
plants play the essential role of primary producers: near- The nitrogen cycle depends strongly on microbial
ly all production of biomass depends on plant photo- activity. Biologically available nitrogen originates in soils
synthesis. Fluxes of water and carbon, as well as nitro- mainly by microbial fixation, and the decomposition of
gen and most other nutrients, also depend critically on biomass recycles available nitrogen in the soil. Nitrogen
the plants in an ecosystem, since plants constitute a large is often limiting to plants; increases in the amount of
proportion of the living biomass, and other organisms available nitrogen can have major effects on plant
acquire carbon and nutrients from food webs that are growth rates. Use of synthetic fertilizer by humans has
based mainly on plants. These materials cycle through roughly doubled the amount of biologically available
the biosphere at very different rates. Thus, changes in nitrogen in the world, with serious consequences for
the amount of plant cover or in the efficiency of uptake both terrestrial and aquatic ecosystems.
by plants have differing effects on each biogeochemical The global phosphorus cycle does not have a major
cycle. Similarly, changes in the different biogeochemical atmospheric component. Phosphorus is limiting for
cycles have very different effects on plants. Plants and plant growth in many ecosystems, particularly those on
the water cycle affect one another quickly, as changes in old soils. Sulfur and calcium are necessary for plant
regional plant cover can affect humidity and rainfall growth; their availability varies widely among different
within years to decades, and changes in moisture can soils. Human activities may be depleting calcium in
affect plants within days to weeks. Changes in the car- some ecosystems. Improving our understanding of bio-
bon cycle occur on a larger spatial and longer time scale, geochemical cycles is critical both for understanding
because CO2 is (on average) evenly distributed in the how plants and plant communities change over time
atmosphere, and much carbon is stored as biomass and and for understanding the environmental challenges
dissolved in the oceans. presented by anthropogenic change.
Additional Readings
Classic References Shaver, G. R. and F. S. Chapin III. 1991. Production-biomass relationships
and element cycling in contrasting Arctic vegetation types. Ecol. Mono-
Tansley, A. G. 1935. The use and abuse of vegetational concepts and terms.
gr. 61: 1–31.
Ecology 16: 284–307.
Vogt, K. A., C. C. Grier, C. E. Meier and R. L. Edmonds. 1982. Mycorrhizal
Lindeman, R. L. 1942. The trophic-dynamic aspect of ecology. Ecology 23:
role in net primary production and nutrient cycling in Abies amabilis
399–418.
ecosystems in western Washington. Ecology 63: 370–380.
Odum, E. P. 1960. Organic production and turnover in old field succes-
sion. Ecology 41: 34–49.
Chapin, F. S., P. Matson and H. A. Mooney. 2002. Principles of Terrestrial
Contemporary Research Ecosystem Ecology. Springer-Verlag, New York.
Gohlz, H. L., D. A. Wedin, S. M. Smitherman, M. E. Harmon and W. J.
Schlesinger, W. H. 1997. Biogeochemistry: An Analysis of Global Change.
Parton. 2000. Long-term dynamics of pine and hardwood litter in con-
2nd ed. Academic Press, New York.
trasting environments: toward a global model of decomposition. Global
Change Biol. 6:751-765. Waring, R. H. and S. W. Running. 1998. Forest Ecosystems: Analysis at Mul-
tiple Scales. 2nd ed. Academic Press, New York.
16
C H A P T E R
Communities in Landscapes
C
ommunities exist as components of a larger landscape. The populations
that make up a community are linked to populations of the same species
in other communities by migration (see Chapter 5). Communities are
also linked by the movement of air, water, and nutrients (see Chapter 15). Land-
scape ecology is the study of these larger-scale relationships among commu-
nities. In this and the following chapter we will examine ecological patterns
and processes that occur at the level of landscapes. In this chapter we explore
methods for examining patterns of variation among communities within a
landscape.
How can we quantify differences in community composition and struc-
ture? How are species distributed among communities? Can we relate these
differences to variation in the environment? These types of questions and analy-
ses are some of the oldest in plant ecology. Comparing communities across a
landscape and looking for environmental causes of differences and similari-
ties goes back to the origins of plant ecology at the end of the nineteenth cen-
tury. While technological advances have enhanced ecologists’ ability to collect
and analyze complex data to use in making these comparisons, many of the
basic underlying questions and approaches have remained the same for the
past 100 years.
Comparing Communities
The first question we might want to ask about the communities in a landscape
is how different they are from one another. In Chapter 12 we described sever-
al ways of characterizing communities. Those characteristics can be compared
among two or more communities. The techniques used for such comparisons
depend on the nature of the variables being compared. If each community is
described by a single characteristic (such as total biomass or mean canopy
height), then univariate (single dependent variable) statistical procedures are
used. In contrast, if each community is described by multiple parameters (such
as a species list), then multivariate (multiple dependent variable) procedures
are needed. The latter methods can be complex and have been the subject of
much debate among ecologists.
Table 16.1 Using the Jaccard index to measure the similarity of sites based on presence/absence data
(C) Matrix of Jaccard similarity values

(A) Typical presence/absence dataa (B) Reordered matrixb for the data in (A) or (B)
Sites Sites Sites

Species A B C D E Species A B E D C Site A B C D E
1 1 1 0 0 1 8 1 1 0 0 0 A 1.00 0.57 0.33 0.13 0.13
2 1 1 1 0 0 1 1 1 1 0 0 B 0.57 1.00 0.30 0.00 0.25
3 0 0 1 1 1 9 1 1 0 0 1 C 0.33 0.30 1.00 0.57 0.22
4 0 0 1 1 0 2 1 1 0 0 1 D 0.13 0.00 0.57 1.00 0.33
5 1 0 1 1 0 6 0 1 1 0 0 E 0.13 0.25 0.22 0.33 1.00
6 0 1 0 0 1 10 0 1 0 0 1
7 0 0 1 1 1 5 1 0 0 1 1
8 1 1 0 0 0 3 0 0 1 1 1
9 1 1 1 0 0 7 0 0 1 1 1
10 0 1 1 0 0 4 0 0 0 1 1
a
The presence of a species in a site is indicated by a 1.
b
A reordered matrix attempts to group sites that share species and species that share sites.
Non-numerical Techniques nities across landscapes. Dissatisfaction was expressed

While the methods used in plant community ecology with this method, however, because it is partially sub-
have changed considerably in response to the avail- jective. For example, site C shares more species with site
ability of powerful computers, the questions asked have A than site E shares with site A. So one might rearrange
not. One such basic question is, which communities in the columns in the table to place sites A and C next to
a landscape are most similar to each other? Later in this each other. Also, we decided that there were two groups
chapter we will show how we can quantify the answer of sites, (A, B) and (C, D), even though there is consid-
to this question. However, we can also address this erable overlap in the species shared by sites in different
question in a more qualitative way, using a method groups. To avoid this subjectivity, ecologists developed
favored by European ecologists during the mid-twen- a variety of quantitative techniques.
tieth century.
We might begin, for instance, with a survey of five Univariate Techniques
sites that contain a total of ten species (Table 16.1A). The Univariate techniques are used whenever a single type
order of the sites and species in Table 16.1A is arbitrary. of measurement is made, such as biomass per unit area.
(Because this is a hypothetical example, we use numbers We might determine biomass in ten quadrats within
for the species, rather than names.) We can then try to each of six communities. A typical question would be,
change the order of the sites and species so that we “Do these communities differ in their average biomass
group the sites that share the most species and group the per square meter?” There are many standard statistical
species that are most often found together. This reorder- techniques for analyzing such data, such as analysis of
ing is shown in Table 16.1B. The reordering has pro- variance (ANOVA). A number of statistics textbooks
duced some patterns, with a group of species that are describe these methods in detail (e.g., Sokal and Rohlf
shared by sites A and B appearing in the upper left-hand 1995; Zar 1999; Scheiner and Gurevitch 2001). Most
corner and a group of species that are shared by C and often, though, we compare communities by measuring
D in the lower right-hand corner, while site E shares more than one variable.
species with both groups. Species 9, 2, 10, and 5 are
shared by both groups, while some of the species might Multivariate Techniques
be considered indicators of each group (8 and 4) since All multivariate techniques rely on the same basic prin-
they each appear in just one group. With a much larger ciples and approach. Each is concerned with comparing
table showing many more sites and species, such pat- how different the members of a group of objects (such
terns usually become even more apparent. as communities) are, based on the values obtained for
Using this sort of reordering method, plant com- a set of characteristics measured for all of the objects. The
munity ecologists tried to discern patterns of commu- mathematical techniques then arrange those objects in
PAGE PROOF: 2ND PASS Communities in Landscapes 319
one or more dimensions, based on their differences in 6.0

the entire set of characteristics taken together. C
Let’s begin with a simple univariate (one-dimen- 5.5
sional) example. Imagine that we wish to compare three
Mean foliage height (m)

plant communities (A, B and C) with mean biomasses 5.0
of 500 g/m2, 725 g/m2 and 625 g/m2. We can put them
dBC
in order based on their mean biomass (A C B). Another dAC
4.5
way of accomplishing the same goal is to first determine
the difference between each pair of means (dAB = 225,
4.0
dAC = 125, dBC = 100). Using this information, we deter-
mine that A and B are most different, or that the order is dAB B
(A C B). Note that dAC + dBC = dAB because there is only 3.5
one way to put these numbers in order from smallest A
to largest—they fall on a line. While the second method, 3.0
450 500 550 600 650 700 750
using differences, seems more complicated than the first,
Mean biomass (g/m2)
it is actually what you do when you use ANOVA to
determine whether groups differ from one another. Figure 16.1
With two descriptive variables, the process becomes Three plant communities (A, B, and C) measured for bio-
a bit more complicated, but the basic principles are the mass and foliage height. Euclidean distances between the
points are given by dAB, dAC, and dBC.
same. Let’s assume that we also measured mean foliage
height for the three communities and found that the val-
ues are A = 3.2 m, B = 3.5 m, and C = 5.6 m. Instead of a
single number, or scalar, giving the difference between
two communities, we now need an ordered list of num- The first step is to devise a distance measure. If the
bers, or vector (see Chapter 7), to give the difference. In descriptive variables are related monotonically (mean-
this case, we can use that old standby, the Pythagorean ing that the abundances of species 1 and species 2 always
theorem (x2 + y2 = z2), where x and y are the lengths of the increase together, for instance, and when species 1
sides of a right triangle and z is the length of the increases, species 3 always decreases), and also increase
hypotenuse. The measure of the difference between each or decrease together linearly (when graphed against one
pair of communities is referred to as the Euclidean dis- another, the points fall on a straight line), then we can
tance. If we let x represent mean biomass and y represent use the multivariate equivalent of the Euclidean dis-
mean foliage height, then for communities A and B, x = tance. This measure is the Pearson product-moment cor-
725 – 500, y = 3.5 – 3.2, and z = 225.0002 = dAB (Figure relation, which is the familiar correlation coefficient (see
16.1). Similarly, dAC = 125.0230 and dBC = 100.0220. Now Appendix A). However, species abundances are usual-
the distances do not sum because the points no longer lie ly not related in such a simple fashion.
on a straight line. Typically units of measure are not used Consider two species with different soil moisture
because the distances are a complex combination of the requirements: species 1 does best in wet soils, while
units of measure for each descriptive variable. species 2 does best in partly moist soils. Going from very
With only two dimensions, it is easy to graph and dry soils to partly moist soils, both species would
visualize our data (Figure 16.1). But what if we have increase in abundance. But going from partly moist soils
many descriptive variables? Information on species pres- to wet soils, species 2 would continue to increase in
ences and abundances constitutes just such a large set of abundance, while species 1 would decline in abundance
descriptive variables. Think of the two axes in Figure (Figure 16.2). The descriptive variables would no longer
16.1 as representing the abundances of species 1 and be related in a monotonic fashion. In addition, many of
species 2. Now, imagine that we add a third axis pro- the descriptive variables used are not continuous. Data
jecting out of the page, on which the abundance of on the presence and absence of species, for example—
species 3 is graphed. We might not be so readily able to the information contained in a species list—are dichoto-
imagine a fourth axis for the abundance of species 4. mous, consisting of just 1’s and 0’s (see Table 16.1A).
Typical data include, however, not just three or four, but For these types of data, ecologists have devised a
tens to hundreds of species. Clearly, we cannot graph number of distance metrics called similarity measures.
such a multidimensional object. Instead, we resort to Consider first those that are used for presence/absence
various techniques that reduce the problem from n data. For a pair of sites, we can define four types of
dimensions (n = the number of species) to the two or species: those that are present in both sites (a), those that
three dimensions that we can visualize. are present in the first site but absent in the second (b),
those that are present in the second site and absent in the Trientalis latifolia
first (c), and those that are absent in both (d). This pat- 50 Xerophyllum tenax
Galium ambiguum
tern can be represented as follows: Iris bracteata
40
Viola lobata
First site
Frequency (%)
Eriophyllum lanatum
Second site Present Absent
30
Present a b
Absent c d 20
For sites A and B in Table 16.1, a = 4, b = 1, c = 2, and d = 3. 10

The simplest, and oldest, measure of similarity,
invented by the French ecologist P. Jaccard, is the Jaccard 0
index (Jaccard 1901). It is the percentage of species con- Moist Dry
tained in two sites that are shared by those sites: Figure 16.2
Patterns of species abundance along a moisture gradient in
Sj = a the Siskiyou Mountains, Oregon. As seen in this case,
a+b+c changes in the abundances of species along an environmen-
tal gradient are unlikely to be related monotonically. Species
abundance was measured as the frequency of quadrats with-
For sites A and B, SJ = 4/(4 + 1 + 2) = 0.57. in a site in which a species appeared. (Data from Whittaker
Over the years a number of similarity measures 1960; after Brown 1984.)
have been invented for both presence/absence and
abundance data (Table 16.2). While various indices work
best in different situations and with different types of Landscape Patterns
data, for presence/absence data the Jaccard index con-
sistently performs the best, or close to the best, in the Ordination: Describing Patterns
widest number of situations. Using the Jaccard index, Ordination is the process of taking information such
we can now proceed to the second step, measuring the as that contained in Table 16.1C—points arrayed in an
distance between each of our sites (Table 16.1C). n-dimensional space—and reducing it to fewer dimen-
Table 16.2 Some similarity measures used by plant ecologists
Index Formula Index Formula
Presence/absence indices Abundance indices
Jaccard index SJ = a Percentage of similarity SPS = ∑| p1i − p2 i |

a +b+ c
S SD = 2a Asymmetrical percentage S APS =

∑ (n 1i − n 2 i ) , for n ≠ 0
Sørensen-Dice index
2a + b + c of similarity ∑ n 1i + ∑ n 2 i 1i
Simple matching S SM = a+d Minimum similarity S MS = ∑ min ( p1i , p2 i )

a +b+ c + d
Ochioi index SO = a
Bray-Curtis index S BC =
∑ min(n 1i , n 2 i )
(a + b) + (a + c ) ∑ n 1i + ∑ n 2 i
b
Asymmetrical similarity S AS =
2a + b Morisita’s index SM =
2∑ (n1i n 2 i )
,λ =
∑ n ji (n ji − 1)
(λ1 + λ 2 )N 1N 2 j N j (N j − 1)
Definitions of symbols:
a = number of species in both sites p1i = the proportion of individuals in the ith species in sample 1 (p1i = n1i/N)
b = number of species in second site only n1i = the number of individuals of species i in sample 1
c = number of species in first site only N = the total number of individuals sampled
d = number of species in neither site min(x,y) = the smaller of the two values
Note: These indices differ in the factors they emphasize (e.g., common versus rare species) and their robustness to deviations
in assumptions (Whittaker 1972; Janson and Vegelius 1981; Wolda 1981; Austin and Belbin 1982; McCulloch 1985).
Figure 16.3
The principle of dimension reduction. The flashlight causes the balls,
arrayed in three-dimensional space, to cast a two-dimensional shadow on
the back wall.
Axis 3
is 2
Ax
Axis 1
sions; in other words, ordination is just a quantitative re- randomly throughout the model. Rather, they are prob-
description of the data. Often the results are shown as ably arrayed in some more compact shape, such as an
a two- or three-dimensional graph. There are many ordi- elongated ellipse. If we shine the flashlight so that it is
nation techniques, each of which has had its advocates. at right angles to the long axis of the ellipse, then points
They are usually known by acronyms: PO (polar ordi- that were the farthest apart in the original model are still
nation), PCA (principal components analysis), PCoA far apart in the shadow.
(principal coordinates analysis), RA (reciprocal averag- Of course, with actual data, the points are arrayed
ing), DCA (detrended correspondence analysis), NMDS in a much more complicated n-dimensional space, and
(nonmetric multidimensional scaling). Each technique the “flashlight” is a set of mathematical steps that deter-
uses somewhat different mathematical approaches and mine the coordinates of those points in the reduced two-
has different assumptions, limitations, and advantages. or three-dimensional space. Figure 16.4 shows two ordi-
While the techniques differ in their mathematical details, nations of the data from Table 16.1C using two different
they all rely on the same basic principles, and they all techniques; notice how both techniques produce the
provide broadly similar information. The distinctions same overall picture of the relationship between the
between and details of these methods are beyond the sites. Since ecologists are usually interested in broad pat-
scope of this book, but those who wish to explore them terns, rather than the exact numerical relationships
further will find the topic covered in depth by Legendre between sites, the choice of distance metric and ordina-
and Legendre (1998). tion technique is often a matter of convenience.
The process of dimension reduction—taking high- Multivariate statistical methods started to permeate
ly multivariate data and collapsing them into a small plant community ecology in the mid-1950s, sometimes
number of dimensions—can be visualized as follows. by importation of techniques being used in other fields
Imagine a physical representation of a three-dimensional and sometimes by independent invention. Perhaps the
data set like that shown in Figure 16.3, in which each axis most influential work was the development of polar
shows the abundance of a species and each ball repre- ordination by J. Roger Bray in the mid-1950s while he
sents a different site. Now, we shine a flashlight through was a student of John Curtis (Bray 1955; Bray and Cur-
the model so that a shadow falls on one wall. Each ball tis 1957). Independently, David Goodall (1954) showed
(site) is now represented by a point on that wall. We have how factor analysis (now referred to as principal com-
reduced the data from three dimensions to two. In the ponents analysis) could be applied to community sur-
process we have lost some information. Two points that veys.
were actually far apart in the three-dimensional model Robert Whittaker developed a general approach
may now appear to be close to each other because of the called direct gradient analysis, in which the ecologist
way their shadows fall. We can minimize this problem, chooses environmental axes and orders vegetation sam-
though, by carefully choosing the angle at which we ples (stands) along those axes, examining the resulting
shine the flashlight. Most likely the balls are not arrayed patterns. Bray and Curtis adopted a different approach,
(A) (B) Figure 16.4

0.8 0.8 Ordinations of the data in Table
16.1C using two different ordination
Second principal component
C
techniques: (A) principal coordinates
0.4 C 0.4 A analysis (PCoA), and (B) nonmetric
D
A D
multidimensional scaling (NMDS).
Dimension 2
0.0 Notice that the two techniques pro-
0.0 duce the same broad pattern of rela-
B B
tionship between the sites.
–0.4
–0.4
–0.8
–0.8 E E
–0.8 –0.4 0.0 0.4 0.8 –1.2 –0.8 –0.4 0.0 0.4 0.8 1.2
First principal component Dimension 1
called indirect gradient analysis, in which the stands primary causes of differences in species composition are
are ordered by their similarity in species composition climatic, topographic, and edaphic factors—the physi-
and the environmental factors responsible for the result- cal factors that determine plant growing conditions (see
ing patterns are inferred. The latter approach (a form Part I). A number of techniques exist for determining
of ordination) has the advantage of being more objec- which of these factors are most important in a given set
tive, but it cannot be used to examine patterns of vege- of communities. All of these techniques use basically the
tation along particular environmental axes that are same strategy, looking for correlations between species
known in advance to be of interest. distributions and environmental variables.
Ordination and related methods were further refined Consider a simple problem, that of determining the
in the 1960s, 1970s, and 1980s by a number of ecologists, causes of variation in the abundance of a particular
most notably Whittaker’s students and collaborators in species across a landscape. One strategy for attacking
the United States (e.g., Gauch and Whittaker 1972), this problem is to sample a number of sites across the
Michael Austin in Australia (e.g., Austin 1977), and Cajo landscape. At each site we measure the abundance of the
ter Braak in the Netherlands (e.g., ter Braak 1986). Over species as well as a number of environmental variables,
the same period, the development of multivariate tech- such as temperature, precipitation, elevation, slope,
niques in systematics (under the heading of numerical aspect, soil texture, and various soil nutrients. (Ideally,
taxonomy; Sneath and Sokal 1973) further influenced the climatic data would be based on long-term aver-
their use in ecology. Advances in computer hardware and ages.) Then we calculate the correlation of each envi-
software at this same time were particularly important ronmental variable with species abundance. This pro-
in the spread of these techniques. During the 1960s and cedure contains two unstated assumptions: (1) that the
1970s, using these methods required writing complex pattern of species distribution is based on a long-term
programs, making stacks of punch cards, and gaining equilibrium with the environment, and that the current
access to a mainframe computer. pattern is close to that equilibrium, and (2) that we have
Today, these techniques are still commonly used, measured the correct environmental variables.
especially for pattern detection and hypothesis genera- Of course, correlation does not demonstrate causa-
tion. A series of communities might be sampled to doc- tion, but a strong correlation provides one piece of sup-
ument relationships involving a particular ecological porting evidence for a possible cause of the observed
phenomenon—for example, nitrogen deposition, inva- pattern. The method just described has a problem, how-
sion by exotic species, or damage by a pathogen—with ever: the environmental variables are themselves corre-
the object of relating the phenomenon to species com- lated with one another. Sites at higher elevations, for
positions and environmental variables across commu- example, are also likely to have lower temperatures. To
nities. The patterns that are revealed can form the basis account for these correlations, we use a technique called
of experiments to test hypotheses about the mechanisms multiple regression. This technique determines the
causing the patterns. regression (Appendix A) of the abundance of our species
with each environmental variable while correcting for
Determining Causes of Patterns correlations among the environmental variables them-
Ordination can describe patterns of species distribution selves. The limitation of this technique is that it cannot
among communities in a landscape. But what are the correct for very large correlations. For example, if the
causes of those patterns? For plant communities, the abundance of our species decreased with increasing ele-
(A) Figure 16.5
FPO (A) Stands of Pinus contorta (lodge-

pole pine, Pinaceae) dominate the
landscape in the middle elevations
of the Canadian Rockies. The cones
open in response to fire, and the
result is stands like this one near
Allison pass. (Photograph © John
Worrall.) (B) Direct gradient analysis
of P. contorta stands in the Canadian
Rockies. The stands are arrayed
along two gradients: moisture and
elevation. Green numbers indicate
stands in Banff National Park. Black
numbers indicate stands in Jasper
National Park. (After La Roi and
Hnatiuk 1980.)
(B)
vation and temperature also decreased with increasing
1800
elevation, then we could not determine whether the
cause of decreasing abundance was elevation or tem- 61
perature. However, our analysis would have narrowed
60
the possible factors responsible for decreasing abun-
dance, although the possibility would remain that the 1600 62
63
actual cause was some unmeasured third variable that
is also correlated with elevation and temperature.
Elevation (m)
We could also perform manipulative experiments 52 54

(e.g., growing the plant in growth chambers at differ- 51 47
1400
ent temperatures) to further narrow the possible caus- 46
57 59 50
49
es and to test specific hypotheses. While such experi- 29 48
36
ments are an important source of information, they also 40
have limitations. Because they are often done in highly 23 45 58

artificial environments such as a growth chamber or 27 42 38352841
1200 16 53 31212244
18 33 34
greenhouse, they cannot account for the myriad of abi- 1930 56 20
39 37
otic and biotic factors and their interactions in nature. 4 24 25 7 8 9 102617
6
43
32
3 5 13 14 11 55
Competitive interactions, for example, might differ at 1 12 15
high and low temperatures. Manipulative experiments 2
1000
done in a field setting are technically more challenging,
but can provide a more realistic test of the importance 2.0 2.5 3.0 3.5
of an environmental factor. Drier Wetter
Moisture index
We can do the same sort of analysis for whole com-
munities as was just described for a single species. Now,
instead of using species abundance, we use position
along an ordination axis as our dependent variable. We understory species, soil chemistry, soil temperature, soil
can see how this works with data from a study of Pinus moisture, elevation, slope, and aspect.
contorta (lodgepole pine, Pinaceae) forests in the Cana- Figure 16.5B shows data from these stands arrayed
dian Rockies. Lodgepole pine is the most widely dis- along two environmental axes, elevation and moisture.
tributed tree in western North America (Figure 16.5A). Because each axis consists of a single environmental
In Alberta, Canada, where this study took place, it is the variable, this graph is a form of direct gradient analysis.
dominant species following fire in mid-elevation forests If each axis consisted of combinations of multiple envi-
(1000–2000 m). For this study, 63 stands were sampled ronmental variables, we would refer to this method as
in Banff and Jasper National Parks; measurements canonical correspondence analysis (CCA; ter Braak
included the number and sizes of all trees, cover of all 1986). In effect, canonical analysis is a hybrid between
direct and indirect gradient analysis. Two ordinations

are carried out, one on the species data and one on the Chloride
environmental data, and then the correlations between ions
the axes of the two ordination sets are determined. If we
are lucky, the first one or two axes of each set will be
highly correlated with each other. An analysis of macro-
Peat
phytes growing on dikes in the Netherlands provides an
example of this type of analysis (Figure 16.6). As with
ordination, various distance metrics, dimension reduc- Electrical
conductivity
tion techniques, and correlation procedures can be used
(Ludwig and Reynolds 1985). Phosphate
Clay concentration
Types of Data
Sand
We have described analyses that can use either pres-
ence/absence data or abundance data. Which type of
data is most appropriate? The answer depends on the
questions being addressed relative to the scale of varia-
tion. Different kinds of variation are expected to occur
at different scales. By “scale” we do not mean geo-
graphic extent; rather, we mean ecological scale—how
different the communities are ecologically. A set of neigh-
Figure 16.6
boring old fields all abandoned within the past five years Canonical correspondence analysis of macrophytes growing
would represent small-scale ecological differences. All on dikes in the Netherlands. The points show the average
of the fields would have similar soil conditions and position of each species along the vegetation axes. The first
species composition. Differences among the fields would vegetation axis is highly correlated with electrical conductiv-
most likely show up as differences in species abun- ity (r = 0.83) and phosphate concentration (r = 0.86), while
the second vegetation axis is correlated with chloride con-
dances rather than differences in species presences. In centration (r = 0.86) and to a lesser extent with the amount of
contrast, if those fields were included in an analysis with peat (r = 0.49) and sand (r = –0.40) in the soil. These correla-
surrounding forest patches, or with fields abandoned tions are indicated by how closely the environmental vari-
decades ago, then we would have a much larger eco- ables (arrows) line up with the vegetation axes. (After ter
logical scale. Now species presences might capture the Braak 1986.)
differences among communities. Geographic extent is
not synonymous with ecological scale, but the larger the
geographic extent, the more likely that a wider range any ordination, the number of species used should be at
of ecological conditions will be captured in the analysis. least three or four times the number of sites.
At larger scales, presence/absence data may be more
informative than abundance data because of the sig- Classification
nal-to-noise problem. At large scales, we are usually try- One use of landscape analysis is the classification of com-
ing to find the one or few causes of differences among munities into larger groupings (Table 16.3). This classi-
very different types of communities. But abundances can fication can then be used for landscape management pur-
vary for a number of reasons having to do with random poses. For example, if we wished to preserve particular
demographic factors (see Chapters 7 and 14). While this types of communities, our classification scheme could
variation may be of interest in relation to other questions, help us choose which places to preserve to meet this goal.
in this instance it acts as “noise” (random variation) in Classification techniques are complementary to those of
the analysis, obscuring the main pattern. By using pres- ordination. In effect, what we are doing is taking the pat-
ences and absences instead of abundances, we smooth tern revealed by the ordination and drawing circles
out this noise. However, if we use presence/absence around subsets of the points. For example, in Figure 16.4,
data, it is important that we include as many species as we might put sites A and B into one group, sites C and D
possible, especially rare species. The distribution of each into another group, and site E into a third group. We
species contains information about the environment and could even construct larger groups from these smaller
its effects on the plant community. The more species ones. Just because we have created groupings does not
we include in the analysis, the more information we mean that the communities form discrete types; our dif-
have. There is a redundancy to this information, so that ferent groups might gradually blend into one another.
the more species we use, the more confident we can be Yet, for planning purposes, defining discrete communi-
about our conclusions. A good rule of thumb is that in ty types would still be useful.
Table 16.3 An example of the classification of a North American plant community
Physiognomic categories
Class . . . . . . . . . . Woodlands
Subclass . . . . . . . Mainly evergreen woodlands
Group . . . . . . . . . Evergreen needle-leaved woodlands
Subgroup . . . . . . Natural/seminatural
Formation . . . . . Evergreen coniferous woodlands with rounded crowns
Floristic categories
Alliance . . . . . . . . . . .Juniperus occidentalis
Association . . . . . . . .Juniperus occidentalis/Artemisia tridentata
Note: This classification follows the National Vegetation Classification system proposed by the Ecological
Society of America. The classification uses a dual system in which higher categories are based on physiog-
nomic criteria and finer-level categories are based on floristic criteria.
A variety of methods can be used to create groups. A polythetic-agglomerative analysis works in the
The two major types of classification techniques are mono- opposite direction. Using some distance metric (such as
thetic-divisive and polythetic-agglomerative. A mono- the Jaccard index), the two most similar sites are
thetic-divisive analysis begins by considering all of the grouped togther. Then the next two most similar sites
sites and dividing them into two groups depending on the are grouped together, and so on. The grouping may
presence or absence of a single key species. Each of the involve two individual sites, a single site and a previ-
new groups is further divided based on new key species ously created group, or two groups. The process stops
contained within each. The process is continued until we when all sites have been grouped together. Thus, “poly-
have reached the desired number of groups, a set of thetic” refers to the use of multiple species in the dis-
groups that seems natural and useful, or until division is tance metric, and “agglomerative” refers to the joining
no longer possible. Thus, “monothetic” refers to the use of process. Various distance metrics and joining algorithms
a single species as the criterion at each step, and “divisive” can be used. The result is a tree diagram (Figure 16.7).
refers to the dividing process. As with the monothetic-divisive approach, this diagram
11
6
Distance between cluster centroids
4 CLUSTER 1 2 3 4
1 Figure 16.7
A tree diagram developed by
a cluster analysis using a
0 polythetic-agglomerative
1
6
26
2
3
4
9
53
60
24
25
10
13
15
12
14
5
8
11
30
7
29
18
27
22
23
17
41
43
55
38
39
42
16
34
19
21
31
20
40
56
28
44
36
59
58
47
51
61
62
49
52
50
33
45
46
35
37
54
57
32
63
48
method for the Pinus contorta

Pinus contorta forests of Banff and Jasper stands in Figure 16.5. (After
Polythetic-agglomerative cluster analysis La Roi and Hnatiuk 1980.)
1800 approach and have been used primarily by Russian

plant ecologists. The third approach relies on physiog-
61
nomy rather than species identity. In this approach, all
broad-leaved deciduous forests would be classified
Ledum/Vaccinium together, even if they shared no species in common. It is
1600 63
also possible to combine these approaches. The current
National Vegetation Classification scheme for classifi-
Elevation (m)
cation of North American vegetation (Table 16.3) com-

52 54 bines aspects of the second and third approaches.
1400 Menziesia/Vaccinium 47 Which approach is most appropriate depends on the
49 scale of classification and its purpose. At very broad con-
48
40
tinental or global scales, physiognomy-based classifi-
cations are the most useful because we are often looking
Shepherdia/Elymus 58 for commonalities among communities that transcend
42 41
1200 16
18 21
39
species identities. Our discussion of global biomes in
Alnus/Linnaea
9 17
Chapter 19 uses this type of system. In contrast, at local
8
5 scales, we often are interested in how closely-related
2 Shepherdia/Arctostaphylos species sort themselves into communities, so either of
1000 the other approaches would be useful. We may, on the
other hand, wish to emphasize continuity among sam-
2.0 2.5 3.0 3.5 ples. If that is the case, ordination is most appropriate.
Drier Wetter
Moisture index Remote sensing is the process of collecting data
Figure 16.8
about an entity of interest without being in direct con-
Classification of the Pinus contorta stands in Jasper National tact with that entity. It is especially useful for gathering
Park graphed in Figure 16.5, using a combination of tech- and analyzing ecological information on a large scale, as
niques, including the cluster analysis shown in Figure 16.7 in classifying vegetation over large areas. Data collection
and indicator species analysis. The clusters are named by is done using airplanes or satellites, as in the case of the
the dominate species. The number of clusters differs from
that in Figure 16.7 based on the additional analyses. (After
popular Landsat Thematic Mapper. A landscape is cap-
La Roi and Hnatiuk 1980.) tured in an image that is divided into a grid. The most
common grid size is 30 × 30 m2, although with newer
technology 1 × 1 m2 grids are now available. Each grid
is used to devise some natural and useful classification square is called a pixel, like the pixels of a computer
scheme (Figure 16.8). image. The data consist of the spectral (light wave) prop-
A useful adjunct to classification is the designation erties of each pixel (Box 16A). Different satellite- or air-
of indicator species. An ideal indicator species is found plane-borne cameras record different numbers of wave-
in all communities of a given type and not in any other lengths, from three or four to hundreds.
community type. As with other procedures, there are Classification using remote sensing proceeds by
various methods for picking indicator species (Dufrêne grouping pixels with similar spectral properties into a
and Legendre 1997). The use of indicator species makes single category. There are two basic approaches. Unsu-
classification procedures much easier. An analysis of a pervised classification separates pixels into a user-
landscape might proceed along the following lines. First, defined number of classes based on statistical similari-
one would pick a number of representative sites in the ties between the pixels. This method is similar to the
landscape and do a complete species survey of each. Sec- classification methods described previously. In this case,
ond, one would classify the sites and identify indicator however, instead of each pixel (site) being characterized
species. Third, one would survey the rest of the land- by the presence or abundance of a set of species, each
scape, looking just for the indicator species. By just look- pixel is characterized by the presence or intensity of a
ing for a much smaller number of species, many more set of light waves. In contrast, supervised classification
sites could be surveyed in the same amount of time and requires the groups to be defined explicitly by the user.
effort. This variety of methods leads to three different Specific locations in the image (training regions) are
approaches to classifying landscapes. The first approach characterized using ground-based knowledge. For
relies on using all species. Polythetic agglomerative instance, one begins by surveying a series of sites, which
methods are typical of this approach and are a hallmark are then classified into groups using the techniques
of the floristic-sociological school of central Europe. The described above. Those sites are also mapped using a
second approach relies on using dominant or indicator Global Positioning System (GPS) and located on the
species. Monothetic-divisive methods are typical of this remotely sensed image. These sites then serve as the
BOX 16A
Differentiating Vegetation Based on Spectral Quality

Remote sensing has been widely used tion, which is useful for assessing plant content of their leaves, photosynthetic
to discriminate and map vegetation vigor. The 630 to 690 nm (red) band activity, standing dead biomass, and
cover and community types since the matches a chlorophyll absorption band senescent woody biomass. The radar
1960s. Discrimination is based on the that is important for distinguishing wavelengths (0.1 to 70 cm) provide
different responses of different plant vegetation types. The 1550 to 1750 nm information on vegetation density,
species to the various regions of the (reflected infrared) band indicates the architecture, and flooding. For exam-
electromagnetic spectrum (microwave, moisture content of soil and vegetation ple, Pope et al. (1994) distinguished six
visible, infrared, and radar) and to dif- and also provides a good contrast marsh types with different species
ferent bands within those regions. between vegetation types. For example, composition, standing biomass, and
Thus, the 520 to 600 nm (green) band Rey Benayas and Pope (1995) were able flooding status using radar bands of 68,
within the visible light region matches to distinguish six different tropical for- 24, and 5.7 cm.
the green reflectance peak of vegeta- est types by differences in the moisture
training regions. It is best to have multiple training Today science is becoming increasingly global in
regions per group in order to account for the inevitable scope and outlook, and it is becoming increasingly dif-
variation between sites. Following either type of classi- ficult to assign particular viewpoints to particular geo-
fication, accuracy is judged using ground-truthing oper- graphic areas. For example, as exemplified by European
ations in which randomly selected sites are surveyed journals such as Oikos, Journal of Vegetation Science, and
to see if their actual species composition matches that Plant Ecology, one can find articles representing all tra-
predicted by the remote sensing classification. ditions being published by plant ecologists from all over
the world.
Views on Continuous versus Discrete Landscapes
Do communities tend to blend into one another, or do
they form discrete types? One’s answer to this question Landscape Diversity
can be influenced by how one analyzes landscape data— Imagine yourself walking through a forest. You stop and
by ordination or by classification. Ordination assumes count the number of species around you. You walk a bit
that landscapes are continuous, while classification farther and stop again. Again you count the number of
assumes that they are not. The dominance of ordination species you find. You are measuring species diversity
as an analytic technique among North American aca- (see Chapter 12). But now you ask an additional ques-
demic ecologists is a reflection of the current consensus tion: How different is the forest from one place to the
among those ecologists that landscapes are mostly con- next? Did you mainly find the same species as you
tinuous and are likely to be dominated by one or two walked along, or different ones? Are communities com-
very strong gradients. Applied ecologists in North posed of equal proportions of common and rare species,
America often use classification methods to categorize or do some communities contain many more rare species
vegetation for practical reasons, however. than others? Answering these questions both informs us
Most classification systems and techniques were orig- about the processes responsible for structuring land-
inally devised by European and Australian ecologists scapes and provides critical information for deciding
(Mueller-Dombois and Ellenberg 1974). The Zurich-Mont- how many and what kinds of communities we need to
pellier or floristic-sociological approach was developed preserve.
by the Swiss ecologist Josias Braun-Blanquet (1932). While
these ecologists recognized some continuity, this approach Differentiation Diversity
emphasizes discontinuity. Northern Europeans, primari- Any ecological unit—be it a single quadrat, a commu-
ly in Sweden, took an even stronger position toward dis- nity, a landscape, or a biome—has several measurable
continuity in the early part of the twentieth century. More properties (Table 16.4). Inventory diversity is the species
recently they have blended their methods with the floris- diversity found within the unit. Differentiation diver-
tic-sociological approach. Russian ecologists participated sity is the way in which the species are grouped into sub-
in many of these same debates. They were most strongly units. Differentiation diversity is most commonly stud-
influenced by ecologists in northern Europe, and had sim- ied at the landscape level and is often simply referred to
ilar views on discontinuity. as β diversity.
A second measure of differentiation diversity is the

Table 16.4 Definitions of some diversity concepts percentage of total species richness:
 S 
Inventory diversity  
 Si • 
Measures: Species density; Shannon-Weiner index;
Simpson’s index where S is the total number of species found and Si. is
Spatial scale: the mean species richness of each sample.
1. Point diversity: The diversity of a single small or microhabitat A third measure is the turnover of species along a
sample within a community that is regarded as homogeneous gradient—the average number of species that appear
2. Alpha (α) diversity: The diversity of a sample representing and disappear as you pass from one community to the
a single community next along a gradient. Turnover can be estimated with
3. Gamma (γ) diversity: The diversity of a landscape or a set of abundance data or presence/absence data. To do so, it is
samples that includes more than one community
essential that the samples can be arrayed along a single
4. Epsilon (ε) diversity: The diversity of a broader geographic
area including different landscapes gradient. Turnover is often calculated in conjunction
with an ordination of the samples.
Differentiation diversity
Measures: Mean similarity; percentage of species richness; Pattern Diversity
turnover A third aspect of diversity is the arrangement of subunits
Spatial scale: within an ecological unit, referred to as pattern diver-
1. Beta (β) diversity: The difference in community composition sity. These patterns can be of three types: spatial, tem-
between communities along an environmental gradient or
among communities in a landscape
poral, and compositional. Spatial pattern diversity is
2. Delta (δ) diversity: The difference in community composition
the arrangement of subunits in physical space (Turner
between communities between geographic regions 1989). The measurement of spatial patterns often begins
with a map created from an aerial photograph or by
Pattern diversity
satellite remote sensing. From this map we can calculate
Measures: Mosaic diversity; nestedness metrics such as patch size, nearest neighbor probability
Spatial scale: No explicit terminology (the probability that two types of habitat will be adja-
Sources: Whittaker 1977 and Scheiner 1992. cent), and fractal dimension (a measure of the complex-
ity of a spatial pattern). These aspects of spatial pattern
are of particular importance for habitat preservation and
the design of nature reserves. We explore these meas-
In a study in Spain, landscapes that were found in ures in more detail in Chapter 17.
transition zones were more species-rich—had a greater Temporal pattern diversity is the arrangement of
inventory diversity—than landscapes outside those subunits in time. The same techniques that are used to
zones (see Figure 20.5). In other words, these transition measure spatial pattern diversity and compositional
zones contained a mingling of species from two differ- pattern diversity can be used, with the restriction that
ent biogeographic floras. This mingling could potentially all subunits must be a single spatially defined area
be of two types: either a patchwork of different types measured at different times. Compositional pattern
of communities from each zone, or a mixture of species diversity is the arrangement of subunits in the mathe-
from both zones within single communities. Measures matical space defined by the site-species composition
of differentiation diversity tell us which one of these is matrix. Two metrics have been developed for quantify-
the actual pattern. ing compositional pattern diversity: mosaic diversity
The differentiation diversity of a set of sites can be (Istock and Scheiner 1987; Scheiner 1992) and nested-
measured in several ways. One measure is mean simi- ness (Patterson and Atmar 1986). Mosaic diversity is a
larity—for example, the mean of the values in Table measure of landscape complexity due to variation in
16.1C. This measure indicates whether the communities species richness among communities and variation in
in the landscape all tend to contain the same species (a commonness and rarity among species within the land-
high mean similarity) or whether they tend to contain scape. A high value for mosaic diversity indicates a
different species (a low mean similarity). In Spain, the landscape with many environmental gradients and
landscapes in each transition zone had the same mean strong differentiation among communities, from those
similarity as landscapes outside the zone, indicating that that are rich in common species to those that are rich
species from the two biogeographic floras were mingling in rare species. In contrast, a simple landscape is one
within single communities (Rey Benayas and Scheiner, dominated by a few species and controlled by one or a
in press). few environmental gradients.
Table 16.5 Two examples of nested landscapes
(1) Site (2) Site

Species A B C D E F G H I Species A B C D E F G H I
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0
2 1 1 1 1 1 1 1 1 0 2 1 1 1 1 0 0 0 0 0
3 1 1 1 1 1 1 0 0 0 3 1 1 1 1 0 0 0 0 0
4 1 1 1 1 1 0 0 0 0 4 1 1 1 0 0 0 0 0 0
5 1 1 1 0 0 0 0 0 0 5 1 1 0 0 0 1 1 1 1
6 1 1 0 0 0 0 0 0 0 6 1 1 0 0 0 1 1 1 0
7 1 0 0 0 0 0 0 0 0 7 1 1 0 0 0 1 1 0 0
8 1 0 0 0 0 0 0 0 0 8 1 0 0 0 0 0 0 0 0
China is more species-rich than the United States replace one another along gradients. Plant communities
because eastern Asia is rich in older lineages (Latham that contain all the species in a landscape, like site A in
and Ricklefs 1993; Qian and Ricklefs 1999; see Chapter our example, rarely exist except when the species are
20). Is the greater species richness in eastern Asia due to drawn from a very small spatial and ecological scale.
a few communities that are especially rich in rare
species? Perhaps the species from the old lineages tend Summary
to concentrate in just a few unusual community types.
However, a comparison of alpine tundras in eastern Asia A plant community exists as part of a set of communities
and western North America found that mosaic diversi- that make up a landscape. To study landscapes, we need
ty was the same in the two areas, indicating that rare methods and measures for comparing communities and
species were distributed in similar ways in both regions arraying them within this larger context. Communities
(Qian et al. 1999). can be compared by a variety of measures. Properties of
Nestedness is the tendency of communities to be communities such as biomass or productivity can be
subsets of other communities. To put it another way, compared using standard univariate and multivariate
when communities are highly nested, a rare species will statistical techniques, such as analysis of variance. Using
be found only in communities that also contain all of the information on species presence or abundance, various
more widespread species. The pattern of nestedness is measures of similarity are possible.
critical in deciding how many communities are neces- This information can be further utilized to exam-
sary to protect in order to preserve the most species (see ine patterns among sets of communities. Ordination is
Chapter 17). a method for arranging these sets in continuous space.
Two examples of nestedness are given in Table 16.5. These arrangements can then be used to look for corre-
In the first example (Table 16.5A), site A contains all eight lations between vegetation patterns and environmen-
species, site B contains six of the eight species contained tal variables. Several classification methods can be used
in site A, and so forth, to site I, which contains just one to sort these sets into groups, which is useful for man-
species, which is found in all sites. In the second exam- agement purposes.
ple (Table 16.5B), site A again contains all eight species Diversity is another property of landscapes. Most
and site B contains seven of those eight. But now sites diversity studies are concerned with inventory diversi-
C–E form one nested subset of the landscape, while sites ty: the number of species present and their relative abun-
F–I form another. dances. Variation among communities can also be meas-
Nestedness is a frequent condition of animal com- ured by differentiation diversity and pattern diversity.
munities, especially communities on islands or in island- All of these measures of communities within landscapes
like habitats, such as mountaintops (Wright et al. 1998). provide additional ways of looking at the world. Using
However, plant communities within a landscape are these measures of pattern, we can study the processes
generally not nested. Instead, plant species tend to that shape the world’s vegetation.
Additional Readings
Classic References Wartenberg, D., S. Ferson and F. J. Rohlf. 1987. Putting things in order: a
critique of detrended correspondence analysis. Am. Nat. 129: 434–448.
Bray, J. R. and J. T. Curtis. 1957. An ordination of the upland forest com-
munities of southern Wisconsin. Ecol. Monogr. 27: 325–349.
Greig-Smith, P. 1980. The development of numerical classification and Additional Resources
ordination. Vegetatio 42: 1–9. Legendre, L. and P. Legendre. 1998. Numerical Ecology. 2nd English ed.
Whittaker, R. H. 1972. Evolution and measurement of species diversity. Developments in Environmental Modelling, Vol. 20. Elsevier, Amsterdam.
Taxon 21: 213–251. Ludwig, J. A. and J. F. Reynolds. 1988. Statistical Ecology. John Wiley &
Sons, New York.
Contemporary Research McCune, B. and M. J. Mefford. PC-ORD: Multivariate Analysis of Eco-
logical Data. MjM Software Design, Gleneden Beach, OR.
Jackson, D. A. and K. M. Somers. 1991. Putting things in order: the ups
and downs of detrended correspondence analysis. Am. Nat. 137: 704–712.
ter Braak, C. T. F. 1986. Canonical correspondence analysis: a new eigen-
vector technique for multivariate direct gradient analysis. Ecology
1167–1179.
17
C H A P T E R
Landscape Ecology
I
n the last chapter, we looked at communities embedded within landscapes,
as well as patterns of species composition and diversity among commu-
nities. Here we shift our perspective to consider the landscape as a whole.
Science involves the search for patterns and for the processes that cause those
patterns (see Chapter 1). What might seem to be a third activity—prediction—
depends on having identified the processes that explain the observed patterns.
These two aspects of science are especially apparent in landscape ecology.
Landscape ecology is the study of the spatial distributions of individuals,
populations, and communities and the causes and consequences of those
spatial patterns. Landscape ecologists have a particular interest in spatial scale,
and have brought this topic to the forefront of contemporary ecological think-
ing.
While the roots of landscape ecology go back to the origins of ecology, it
gained its current identity only in the 1980s. The term “landscape ecology” was
coined by Carl Troll, a German geographer, in 1939. Until 1980, the field was
mostly confined to Europe, where it was most closely allied with the floristic-
sociological approach of Braun-Blanquet and emphasized static patterns
(Naveh and Lieberman 1994). In the early 1980s, Richard T. T. Forman (1995)
and others brought landscape ecology to North America, modifying the Euro-
pean tradition by adding dynamics and scale. The theoretical roots of some
of these efforts go back to the 1960s, to island biogeography theory and
metapopulation theory. The coalescence of landscape ecology into its modern
form has also relied on contemporary developments in the field of geography.
Other threads that have contributed to the emergence of landscape ecology are
the development of spatially explicit statistics and modeling and technologi-
cal advances in Geographic Information Systems (GIS) and remote sensing.
Because of the relative youth of landscape ecology, the field is still very
much in flux. New perspectives, methodology, and conceptual advances con-
tinue to develop. As a result, links between theory and data are still limited;
questions have been posed that do not have answers, and relationships have
been hypothesized but not yet tested. While such ambiguities can be frustrat-
ing for students, the study of landscape ecology provides an opportunity to
grapple with a scientific field at a very dynamic stage in its development.
Many ecological phenomena are best studied at the landscape level. For
example, low-intensity fires can be quite patchy on a local scale: severely dam-
aged trees may be found near unscathed trees. As burn intensities increase,
however, this local patchiness is greatly reduced (neigh- scape elements, others are most interested in how eco-
boring trees tend to experience similar burn damage), logical phenomena are affected by changes in scale, or
but a new spatial pattern emerges because intense fires in the effects of spatial patterns on ecological proper-
tend to “spot”: flying embers ignite material at a distance ties regardless of spatial scale.
from the initial fire. The spatial arrangements of indi-
vidual trees, groups of trees, lakes and rivers, and topog- Spatial Patterns
raphy all combine to determine fire spread and intensi-
ty. Thus, we can gain unique insights into fire ecology There are two general ways to study patterns on a land-
and the resulting responses of plant populations by scape level. One, a spatially explicit approach, depends
studying how the elements in a landscape are arranged. on determining the particular spatial arrangement of
Landscape ecology is concerned with questions such species and landscape elements (e.g., habitat patches,
as, How are individuals, species, and communities dis- farms, roads). The other, often called the mean field
tributed within a landscape? How many types of patch- approach, focuses on describing average parameter
es are there in a landscape? What are the sizes, shapes, values.
and distributions of those patches? How do these pat- To see this distinction, consider a species-area curve
terns affect the movement of individuals, materials, and (see Chapter 12). Imagine that we do the following: On
energy among patches and across the landscape? What the landscape shown in Figure 17.1, we overlay a grid
processes are responsible for which patterns? Do the of 10 × 10 m squares, and we compile a list of every vas-
processes occur at the level of the landscape (e.g., migra- cular plant species growing in each square. We can now
tion) or within communities (e.g., competition)? How do build a species-area curve in two ways (Figure 17.2).
the patterns we find, and the interpretation of those pat- Using the first, spatially explicit approach, we start in
terns, differ as scale changes? The answers to these ques- one spot. Imagine that this spot is the bottom left corner
tions have important implications for conser-
vation and landscape management. The
design of nature reserves, for example, is
sometimes explicitly based on landscape ecol-
ogy theory. Realistically, political and eco- Shrubland
nomic realities often intrude on scientific con-
siderations in the establishment and Agricultural
management of nature reserves; nevertheless, field
it is important to know what one is aiming for Agricultural
field
in making management decisions. We return Forest B
to issues of conservation and management at
the end of this chapter.
The name “landscape ecology” can be
misleading—landscape ecology does not nec-
essarily concern large (to humans) spatial Meadow Shrubland
scales, but rather scales that are large rela-
tive to the species or phenomena of interest.
Road
Road
For example, a study in landscape ecology Grid

concerned the movements of beetles over a
10-m2 patch (Johnson and Milne 1992). Many
studies of landscape ecology are carried out
at the scale of a watershed—an area that Forest A
includes a stream or river and all of the land Agricultural
that drains into it. While the focus of inter- field
est for many landscape ecologists may be at
Agricultural
the scale of watersheds or other whole land- field
Figure 17.1
Diagram of a landscape showing a patchwork
of meadows, forests, and shrublands. For sam-
pling purposes we can lay down a 100 × 100 m
grid that arbitrarily divides the area into small-
er plots. 100 m
PAGE PROOF: 2ND PASS Landscape Ecology 333
Edge of Edge of would not want to do this by hand, it can be done in a

meadow forest A few seconds on a typical desktop computer.) We con-
tinue this process with all sets of three squares, four
Meadow Forest A squares, and so forth. The result is a smoothly rising
Totalof number of species sampled
curve. Any discontinuities in the environment that

would have produced a stairstep effect with the spatially
explicit approach are, instead, averaged out.
Forest B
Which of these approaches is preferable? The
answer depends on the questions one wishes to address.
The pattern produced by each of these species-area
Spatially explicit approach curves can provide important information about both
landscape patterns and the processes that account for
them. Next, we explore these patterns in more detail.
Defining Patches
The basic unit we use to examine spatial patterns is the
Mean field approach patch. A patch is any specified area, either an area
(nonspatial curve) defined arbitrarily or, often, an area that is in some way
Area relatively homogeneous or internally consistent. A patch
can be defined by naturally occurring edges (e.g., the
Figure 17.2
Two species-area curves. One curve is from a spatially boundary between a meadow and a forest), but need not
explicit tally, which forms a stairstep pattern. Within a com- be. For example, each of our 10 × 10 m squares in the
munity, the number of species found as area increases will exercise above could be a patch. However, if we start
level off. When a community boundary is crossed, however, with such artificially defined patches, we might want to
the number of species found will once again rise rapidly. The aggregate them into “natural” units. Sometimes this is
other curve, from a mass field analysis that is not spatially
explicit, forms a smoothly rising pattern. done based on compositional similarity—the extent to
which adjacent patches share a similar set of species at
similar frequencies (see Chapter 16). Of course, adjacent
patches usually share at least some species, so the cutoff
of the meadow. Our first data point is the number of point for when patches should be joined or not is some-
species in the 10 × 10 m square in that corner. Next, we what arbitrary. Often the decision is aimed at ending up
expand our square to 20 × 20 m and again count the total with a manageable and interpretable number of patch
number of species. This number will include all of the types.
species in the original square plus any new species in Remote sensing images have become a popular
the larger square. We repeat this operation for a 30 × 30 method for sampling large areas quickly. In this case, the
m square. Initially, the number of species will rise rap- minimum patch size is determined by the resolution of
idly. But as more and more of the meadow is captured the remote sensing equipment. Typically this is 30 × 30
in our ever-growing plot, the number of new species will m, although newly available technology is pushing res-
drop off. At some point, it is likely that we will even stop olution down to 10 × 10 m, and even 1 × 1 m. Our aggre-
finding new species, and the curve will level off. If we gation of patches is again based on their similarity, but
keep going, however, our plot will eventually run into now similarity is determined by spectral quality rather
the forest abutting the meadow. Now, suddenly, we will than by shared species (see Box 16A). Because spectral
find a whole new suite of species, such as trees and for- quality is an indirect measure of community composi-
est-floor herbs. The species-area curve will again rise tion, it limits our view of landscape structure.
rapidly with increasing area until, once again, it levels Aggregating patches and deciding on the number
off as most of the forest gets sampled. This stairstep pat- of patch types is, therefore, as much an art as a science.
tern will be repeated each time we cross a new com- Often it depends on the ability of measurement instru-
munity boundary. ments to distinguish between different patch types. With
In contrast, we can build our species-area curve remote sensing, for example, the indirect measures we
using the mean field approach, which ignores the spa- are using need to be related to the actual vegetation
tial arrangement of the squares. First, we calculate the through a process called ground-truthing: one goes out
average number of species in all of the 10 × 10 m squares. to the field and determines empirically what the remote
Next, we take all possible combinations of two squares, sensing devices are recording. Two communities with
determine the total number of species in each pair of very different species compositions may be indistin-
squares, and again calculate the average. (While we guishable in spectral quality. Thus, if remote sensing
were being used to classify a large region, one would be ecology as well.
forced to lump those two communities into one patch Bruce Milne (1992) connected the concepts of frac-
type. tal dimension and scaling explicitly by quantifying the
Because landscape ecology is built on the study of degree of habitat fragmentation in a landscape. Milne
patches, it is, inevitably, built on a categorical view of the quantified the fractal dimension of a New Mexico grass-
world. Categorical versus continuous views of nature land and showed that the habitat was patchy. Beyond
have been a regular source of controversy in plant ecol- that simple result, he showed that the patchiness was
ogy. In Chapter 12 we explore the history of this con- scale-dependent. Scale dependence implies that species
troversy, beginning with the disagreements of Clements might perceive the landscape differently. A beetle that
and Gleason. It is somewhat ironic that nearly 50 years lived in grass clumps might perceive a very disconnect-
after the Gleasonian continuum was declared tried landscape, while a grazing pronghorn might perceive
umphant, at least among English-speaking ecologists, a that same landscape as one large patch. Some areas of
categorical framework has once more emerged to dom- the landscape showed high connectivity at three differ-
inate landscape-level studies. ent scales, so here the beetle and the pronghorn might
perceive similar landscapes. Other areas consisted of iso-
Quantifying Patch Characteristics lated patches that might attract species that disperse
and Interrelationships readily, but then settle within a single patch (e.g., small
Once we have defined our patches, we can measure rodents).
some of their attributes. For individual patches, we can As researchers become concerned with going
measure sizes and shapes. A typical measurement of beyond the individual patch, they can examine the
shape is the perimeter-to-area ratio, which indicates how arrangement of patches in the landscape. Using a dif-
much edge a patch has relative to its interior. Edges are ferent approach than the one taken by Milne, researchers
important ecologically for many reasons. In particular, may ask, are patches mostly aggregated into like types
they are the focal areas for processes such as the move- or not? A checkerboard pattern, for example, represents
ment of individuals or materials between the outside the maximal interdispersion of two patch types in a
and inside of a patch of forest. landscape (Figure 17.3A). The maximal aggregation of
One approach to quantifying the shapes used in two patch types would have all the patches of each type
landscape ecology is to calculate the fractal dimension grouped into two large clumps (Figure 17.3B). A meas-
of the landscape elements. A fractal dimension is a frac- ure of this aggregation is connectedness (see below). If
tional dimension—for example, something in between there are more than two patch types, then other meas-
a one-dimensional line and a two-dimensional surface, ures of context come into play. For example, for a land-
or in between a two-dimensional surface and a three- scape with three patch types, a patch of type A could
dimensional area. For example, as a line (which has one be surrounded entirely by patches of type B, or entirely
dimension) becomes increasingly convoluted, it becomes by patches of type C, or by half of each. Clearly, such sec-
space-filling, finally turning into a surface with two ond-order patterns can get very complicated with a large
dimensions. The fractal dimension is thus a measure of number of patch types. We can quantify these patterns
the complexity or degree of convolution of a line (such as the nearest neighbor probability, the chance that two
as the perimeter of a patch or the coastline of an island) patch types will be adjacent.
or a surface. Fractal geometry can also be used to quan-
tify patterns of branching and fragmentation.
Imagine that you have a very large map of a coast- Scale
line showing very fine details. You have to measure the
length of the coastline using a ruler with no markings. Definitions and Concepts
Thus, the smallest distance that you can measure is the When we study an ecological pattern, we must establish
length of the ruler. A small ruler will pick up more of the the scale at which we are conducting our analysis. Before
ins and outs of a coastline than a large ruler, so that using we examine the implications of this seemingly straight-
the smaller ruler will result in a longer measurement. As forward statement, however, we must point out a poten-
the coastline or the perimeter of a patch becomes increas- tially confusing difference in how people in different
ingly convoluted, the effect of ruler size becomes greater. fields discuss scale. Ecologists generally equate scale
The magnitude of the ruler size effect is the fractal with the physical dimensions of the phenomenon being
dimension. Pennycuick and Kline (1986) used just such studied: a process that occurs over 10 km is at a larger
an approach to quantifying fractal dimension to com- scale than one that occurs over 10 cm. We follow this
pare bald eagle nest density on two Alaskan islands that practice in this textbook. Some landscape ecologists refer,
differed in the complexity of their coastlines. This instead, to “coarse scale” and “fine scale” for larger and
approach might be useful for various problems in plant smaller physical areas, respectively (Forman and
(A) (B) (C) Figure 17.3

Three different patterns of patch
aggregation in a landscape with two
patch types. (A) Maximal dispersion
of patches. (B) Minimal dispersion
of patches. (C) Random dispersion
of patches.
Godron 1986). However, cartographers and remote sens- in fundamentally different patterns (Figure 17.4; see also
ing researchers use exactly the opposite terminology, in Figure 20.11).
which scale is the relative size at which an object of a Consideration of scale is central to understanding
standard size appears in a map or photo. Thus, a world ecological processes. The same process may function dif-
map has a much smaller scale than a regional map (that ferently at different scales. For example, the ecological
is, a meter is much smaller on a world map). It is there- process of herbivory is generally a very local phenome-
fore important to be explicit as to how one is using the non. However, the effects of an elephant herd might be
term “scale” (e.g., Turner and Gardner 1991), although manifested at the landscape level, and those of a locust
some ecologists have advocated abandoning the term swarm at the regional level. Population dynamics can
entirely, replacing it with its components (Csillag et al. be a local phenomenon, but if metapopulation dynam-
2000). ics is involved, then it becomes a regional phenomenon
Scale is composed of several components, two of as well. The evolutionary process of extinction can occur
which are grain and extent. Grain is the size of the pri- at a local, regional, or global scale; its implications at
mary unit used in a study—for example, a 10 × 10 m these different scales can be very different.
square or a 10 km2 area in which species are counted. Jonathan Levine (2000) found that at different scales,
While the grain is usually the smallest distinguishable different processes explained the success of plant inva-
unit, or the finest level of spatial (or temporal) detail that sions. In northern California, Carex nudata (Cyperaceae),
can be resolved in a given data set, for some types of a tussock-forming sedge, forms clumps along river
data it is possible to extrapolate to a smaller grain; in edges. A tussock (clump) creates a microhabitat that can
contrast, data aggregation will result in a larger grain. shelter up to 20 species of angiosperms and bryophytes
Extent is the total range over which a pattern is exam- during winter flooding. Across a 7-km stretch of river,
ined. The extent of a study might be across a local region Levine found that the higher the species richness of a
of 100 km2 or across the entire globe. For example, a tussock, the more likely it was to be invaded by the non-
research project might have a grain of 1-ha squares and native species Agrostis stolonifera (creeping bent grass,
an extent of 10,000 ha in central Africa, which together Poaceae), Plantago major (common plantain, Plantagi-
define the scale of the analysis. naceae), and Cirsium arvense (Canada thistle, Asteraceae).
Some researchers recognize a third component of This pattern was the result of processes working in
scale, focus or resolution. These terms refer to the sam- opposite directions at smaller and larger scales. Very
pling level or the area represented by each data point. small-scale processes—those occurring within a single
Consider an analysis of diversity in a 1-km2 area that is tussock—were investigated by an experimental manip-
divided into 1-ha squares. Those 1-ha squares become ulation of the number of native species on individual
the focus of our sampling. In each square we randomly tussocks and by the introduction of seeds of the invaders
place 10 1-m2 quadrats and record the species contained into the experimental tussocks. At this small scale, as
in each quadrat. These data can be analyzed in two species richness increased, the germination, survival,
ways: using the average species richness of the 10 and growth of the invaders declined. In contrast, at the
quadrats in each 1-ha square (in which case the grain larger scale of the 7-km stretch of the river, Levine found
equals 1 m2), or using the total number of species in the that seeds of all species (both natives and invaders)
10 quadrats in each 1-ha square (in which case the grain moved with the flow of the river, and so both native
equals 10 m2). In both cases, though, the focus is 1 ha, diversity and frequency of invasion increased down-
because the 1-ha square is the basic unit of analysis. stream. Thus, although native diversity contributed to
Changing the grain, extent, or focus of a study can result community resistance to invasion at very small scales,
(A) (C)
Focus = Plot Focus = Community
1
Extent = Region Extent = Landscape
2
Species richness
Species richness
3
Productivity Productivity
(B) (D)
1 Focus = Plot Focus = Landscape
2 Extent = Landscape Extent = Region
Species richness
Species richness
Productivity Productivity
Figure 17.4
An illustration of how changing the focus and extent of a landscape is analyzed separately, so the extent is the land-
study can alter the results—in this case, the relationship scape. Analyses of the relationship between species richness
between species richness and productivity. Each diagram and productivity for each landscape result in a negative
represents a region. In this region there are three landscapes, slope. (C) Each point represents the average of the plots in a
with three communities in each landscape and five plots in community, so the focus is the community. Each landscape is
each community. In all cases, the grain of the study remains analyzed separately, so the extent is the landscape. Analyses
the plot. Species richness is measured as the total number of of the relationship between species richness and productivi-
species in each plot. (A) Each point represents one plot, so ty for each landscape result in a negative slope. (D) Each
the focus is the plot, and the extent is the entire region. An point represents the average of the plots in a landscape, so
analysis of the relationship between species richness and the focus is the landscape. The extent is the entire region,
productivity results in a slope of zero, or no relationship. (B) and the relationship between species richness and produc-
Each point represents one plot, so the focus is the plot. Each tivity is positive. (From Scheiner et al. 2000.)
at larger scales, other factors that changed with diversi- across the Pacific (see Chapter 18). How the temporal
ty (here, seed supply of invaders) were more important, variation in stream flow affected invasions is not known.
creating a positive relationship between invader and However, such landscape-scale variations in flooding,
native diversity. across temporal changes in weather happening at the
Interestingly, there was even a suggestion of the scale of thousands of kilometers, could potentially influ-
potential importance of factors operating at much larg- ence regional invasion success by creating greater oppor-
er scales. At the beginning of the study in 1998, the river tunities for invasion as tussocks become disturbed by
chosen for the experiment was one of the few whose flooding followed by drought. Thus, the process of inva-
banks were not flooded by high-water conditions, and sion could be influenced in this system by processes
thus had one of the few stretches of Carex tussocks avail- occurring over a range of spatial and temporal scales,
able for study. In the following year, water levels sometimes acting in opposition to one another.
dropped severely, threatening the survival of the Carex Hypotheses that seek to explain large-scale patterns
tussocks on this stretch of river, while they recovered may be based on processes that operate at smaller scales
elsewhere in the region. This variation in local flooding than, or at the same scale as, the phenomena they seek
is related to a 3- to 7-year climatic cycle, the El Niño to explain. If the underlying process on which the expla-
Southern Oscillation, that affects weather conditions nation rests operates at a smaller scale, then the hypoth-
esis must either include an explanation for how local the soil or the amount of disturbance due to wave action
processes scale up to larger areas or assume that large- (Wilson and Keddy 1986; see Figure 10.26). In arid mon-
scale patterns are simply the sum of local-scale patterns. tane regions, by moving a distance of just a few kilo-
Scaling up and down from one set of patterns to anoth- meters up the side of a mountain, one can travel from
er is a major current focus of research in many different desert to temperate forest to boreal forest. Conversely,
areas of plant ecology, from the modeling of carbon flux- traveling across the taiga of Russia, one finds very sim-
es (leaf-canopy-ecosystem-global) to determining the ilar forest communities stretching across many hundreds
causes of diversity patterns. of kilometers.
Spatial and Ecological Scale Quantifying Aspects of Spatial Pattern and Scale
How do we specify (and think about) scale? It is com- Given the complexity of the mosaic of environmental
mon for people to define scale in an exclusively spatial factors and biotic interactions that determine spatial pat-
manner, but we can contrast this approach with a con- terns at different scales, how can we hope to describe or
sideration of scale in an ecological sense. This latter quantify these patterns? Ecologists have developed both
meaning of scale is related to the concept of hierarchy in spatially explicit and nonspatial methods of doing this.
ecology. Using a strictly spatial definition, ecologists con- (By “spatially explicit,” we mean approaches that take
ventionally speak of local, landscape/regional, and con- the particular location of each data point into account.)
tinental/global scales, and they use area to distinguish Methods that are not spatially explicit—ordination and
these scales—for instance, local scales are often thought some statistical approaches to quantifying spatial asso-
of as being up to 102 km2, landscape/regional scales as ciation—are older, and we discuss some of them in
being between about 102 km2 to 108 km2, and continen- Chapter 16.
tal/global scales as being 108 km2 and larger. It is usu- Recently, a spatially explicit graphical technique,
ally important, however, to define scale in terms of the Geographic Information Systems (GIS), has become
particular ecological processes involved. For example, central to landscape ecology. GIS is a method for depict-
102 km2 may be about the scale at which local pollen dis- ing the relationships between different kinds of infor-
persal occurs for pines or some other plants with wind- mation (“layers” in GIS terminology) at particular loca-
dispersed pollen, but it is much too large a scale for tions. For example, we might have several types of
pollen dispersal by animals such as ants, beetles, or Aus- information that relate to each location in the landscape
tralian possums. (Figure 17.5). One set of information might be the species
Ecologically, we often recognize three basic scales: composition of each patch. For each patch, we might also
within a community, across communities, and across have information on soil type, nutrient levels, water
biomes. It is easiest to appreciate the distinctions among table depth, slope, and the last time the patch was
these scales by starting with biomes, the largest scale. logged or burned. Our information could also include
Biomes are defined by distinct differences in physiog- socioeconomic information such as land ownership,
nomy over large regions (e.g., grasslands, shrublands, taxes on that piece of land, the number of people resid-
forests; see Chapters 18 and 19). Patterns that cross ing there, and the legal classification determining how
biome boundaries involve variation in the physical sizes, the land can be used. GIS allows us to map these vari-
morphologies, and ecological niches of the dominant ables singly and in any combination to get a visual sense
species—for example, shrubs versus trees. Within bio- of potential patterns of correlation among them, so that
mes, moving across communities may result in more we may pose hypotheses regarding causality. GIS by
subtle, yet equally important changes—for example, itself does not allow one to quantify these relationships
from a forest dominated by fast-growing hickory (Carya or statistically test hypotheses about them.
spp.) trees to one dominated by slower-growing oaks Spatially based relationships can be quantified using
(Quercus spp.). Even within a single community, such as spatial statistics. Many advances in the field of spatial
the meadow in Figure 17.1, there may be variation in fac- statistics have been made since the late 1980s (Legendre
tors such as soil conditions, land use history, microhab- and Fortin 1989). These advances include the use of geo-
itat, and herbivore activity. statistics, taken from the field of mining geology, and
These two different approaches to thinking of techniques for studying spatial autocorrelation, taken
scale—spatial and ecological—are not entirely inde- from the field of geography (Liebhold et al. 1993; Lieb-
pendent. The farther we move in space, the more likely hold and Gurevitch 2002). Older statistical techniques in
it is that we will cross community or biome boundaries. ecology were developed to analyze and model spatial
But we may also cross community boundaries over very variation in ecological data (Pielou 1977; Greig-Smith
short distances if the environment changes rapidly. Mov- 1983). These methods were useful in distinguishing cer-
ing away from the edge of a stream or lake, for example, tain kinds of spatial patterns, but did not take the actu-
can result in abrupt changes in the amount of water in al spatial location of the data into account, and had
this approach, a patch is seen as existing in a hierarchy

of successively larger patches. Each level of this hierar-
chy defines a scale. Ecological processes occur within
patches, while links among patches are created by the
movement of materials and energy. The pattern of link-
ages defines the next layer in the hierarchy and deter-
Trees mines how processes propagate up the hierarchy. For
example, gas exchange by a single leaf can be combined
with that of the other leaves on a plant to determine the
dynamics of gases around the entire canopy of a tree,
which in turn can be used to model an entire forest stand.
We have discussed a number of the issues involved
in describing and studying patterns in nature. As you
have doubtless concluded, there is no single way to
study or even describe such patterns (Levin 1992). View-
Topography
ing similar questions from different vantage points (such
as the spatial versus ecological scales discussed above)
or at different scales can lead to very different answers.
This may be frustrating if you are expecting simple
truths, but as we have seen, considering information at
more than one scale, while difficult, can also lead to new
insights and understanding of the underlying biology.
We now turn to some approaches to studying the
Streams processes responsible for some of the types of patterns
we have just examined.
Toward a Theoretical Basis for Landscape

Patterns: Island Biogeography Theory
Our exploration of landscape processes starts with arch-
ipelagos, chains of islands. Robert MacArthur and
Edward O. Wilson’s theory of
island biogeography (1967) was
developed to explain patterns of
Sum
species presence and absence on
islands. That theory was impor-
tant in the development of much
of today’s landscape ecology.
Figure 17.5
GIS is used to examine the relationships among multiple
Although the initial descrip-
types of information, which are arranged in layers. tion of the theory applied it to
oceanic islands, many types of
communities have island-like
properties. An inclusive defini-
Edward O. Wilson
numerous shortcomings. The newer approaches have tion of an island is any area that
contributed substantially to our current understanding is suitable for the survival and
of the spatial aspects of organisms’ responses as indi- reproduction of a species and is surrounded by unsuit-
viduals, populations, and metapopulations (summa- able habitat. A landscape made up of isolated woodlots
rized by Dale 2000; Legendre and Legendre 1998). Curi- consists of a set of islands from the point of view of an
ously, spatial statistics have rarely been used by herb that lives only in woodland understories. Moun-
ecologists in conjunction with GIS. Just as we combine taintops are islands for alpine species. Plants can be spe-
graphical and statistical analyses with other types of cialized for life on particular types of soil or terrain, such
data, together GIS and spatial statistics constitute a as serpentine barrens (see Box 15A) or rock outcrops,
remarkably powerful set of tools. which also function as islands. The theory of island bio-
Hierarchical patch dynamics is a new approach to geography therefore has applications far wider than
examining the effects of scale (Wu and Loucks 1995). In oceanic islands.
The basic idea of the theory is quite simple: The Moreover, islands closer to the mainland have higher
number of species on an island is determined by the rate immigration rates because it is more likely that indi-
of immigration of new species to the island and the rate viduals will arrive there from the mainland. Using their
of local extinction on the island (Figure 17.6). The rate of assumption that the islands had reached an equilibrium
arrival of new individuals from the mainland deter- number of species, MacArthur and Wilson predicted
mines the immigration rate—the rate at which new that the greatest number of species would be found on
species arrive. (This immigration rate differs from that large islands near mainlands (the source of immigrant
used in other contexts where it refers to the rate at which species), and that the smallest number would be found
individuals arrive.) As the number of species on the on small islands far from mainlands (Figure 17.6C,D).
island increases, the immigration rate declines, because Real archipelagos, of course, are more complex:
an arriving individual is increasingly likely to be a mem- among other things, immigrants can arrive not just from
ber of a species already on the island. The relationship the mainland, but also from other islands. A more impor-
between species number and extinction rate is a bit more tant limitation of island biogeography theory is that it
complicated. Assume that the island can hold a fixed treats all species as if they were interchangeable. Clear-
total number of individuals. As more species take up res- ly, this is not the case. Even very similar species have dis-
idence on the island, the potential population size of tinctive ecologies. MacArthur and Wilson acknowledged
each gets smaller. The chance that a species will go local- this limitation, although many ecologists subsequently
ly extinct—disappear from the island—increases as its trying to apply their theory did not. The equilibrium
population size gets smaller. Thus, the rate of extinction assumptions needed to predict species number are
increases as the number of species rises. If we make the unlikely to hold, and empirical studies have provided
further assumption (as MacArthur and Wilson did) that little support for the theory’s predictions. Despite these
these processes are roughly constant over long periods, limitations, the theory is useful. It provides a starting
then the number of species on the island eventually point for examining patterns and gives us a way of
reaches a steady state, given by the value of the hori- thinking about the movement of individuals in a land-
zontal axis where the immigration and extinction curves scape.
intersect.
A key insight of MacArthur and Wilson was that Metapopulation Theory
larger islands can hold more individuals and, conse- The theory of metapopulations, originally developed
quently, extinction rates on larger islands are lower. by Richard Levins (1969) and subsequently expanded
(A)
A2 Figure 17.6
A3 (A) Mainland and archipelago. Islands A1 and A2 are the same distance
Mainland from the mainland, while A1 and A3 are the same size. (B) The equilibrium
number of species on island A1 is determined by the balance between the
immigration and extinction rates. (C) The larger island (A2) will have a
A1
higher immigration rate and a lower extinction rate, resulting in a greater
number of species at equilibrium. (D) The nearer island (A3) will have a
higher immigration rate but the same extinction rate, resulting in a greater
number of species at equilibrium.
(B) (D)
(C)
A3
)
A2
Extinction rate (
A1 A1 A1
Immigration rate (
A1 A1
A1 A3
A2
)
s1 s1 s2 s1 s3
Number of species (s)
BOX 17A
Metapopulation Models
Metapopulation models have the gen- The extinction rate might also depend of patches with populations that can go
eral form dp/dt = immigration rate – on p for the same reason: as the number extinct, p, times one minus the fraction
extinction rate where p is the percent- of seeds being dispersed increases of sites that can provide immigrants,
age of patches occupied by a given (with increasing p), there could be a thereby reducing the probability of
species, 0 ≤ p ≤ 1, and dp/dt indicates reduced chance of local extinction. This local extinction, 1 – p.
the rate of change in occupation over phenomenon is called the rescue effect If both extinction and immigration
time. Typically we are interested in (Brown and Kodric-Brown 1977) be- rates are independent of the fraction of
solving this equation for p, which pre- cause new immigrants “rescue” the occupied patches, the model reduces to
dicts the frequency of occupied patch- local population from extinction. dp/dt = i(1 – p) – ep
es in a landscape. Obviously, a species’ If both the immigration rate and the
biology sets its immigration and extinc- extinction rate are functions of p, the Gotelli (1991) refers to occupancy-
tion rates. model takes the form independent immigration as the
It is important to realize that in these propagule rain because immigration
dp/dt = ip(1 – p) – ep(1 – p) provides a steady “rain” of new indi-
models, immigration and extinction
rates may not be independent of each That is, the immigration rate is a viduals to each population. All four
other, and either or both may depend function of the probability of immigra- possible combinations of occupancy-
on the frequency of occupied patches, tion, i, times the fraction of available dependent and occupancy-independ-
p. It is easy to see how the immigration sources of immigrants, p, times the frac- ent immigration and extinction rates
rate might depend on p: the number of tion of available patches, 1 – p. The are conceivable and represent the
seeds arriving in a patch could depend extinction rate is a function of the prob- extremes of a more general metapopu-
on the number of occupied patches. ability of extinction, e, times the fraction lation model.
upon by Illka Hanski (1982, 1999), was another impor- equivalent migration and extinction probabilities, the
tant stimulus for landscape ecology. In some respects, theory also predicts the distribution of population sizes
metapopulation theory is similar to the theory of island among species within a community and the frequency
biogeography, but applied to populations rather than of occurrence of species among communities. It is pos-
sets of species. Begin by thinking of an archipelago, but sible to create more complex versions of the theory by
one far from a continent. The population size of a species relaxing the assumptions of species and patch equality.
on any given island is a result of local population Susan Harrison recently coined the term “metacommu-
dynamics plus immigration minus emigration (this is nity dynamics” for the extension of metapopulation the-
the same as Equation 7.1). A population on a given ory to explain community patterns (Harrison 1999).
island may go extinct, but new immigrants from other One important conclusion that comes from extend-
islands may repopulate that island. The equilibrium ing population dynamics to a landscape is that not all
between immigration and extinction determines the total local populations need be self-sustaining. A sink popu-
size of the metapopulation—the number of surviving lation may be continually on the road to extinction, but
populations in the archipelago and their average size. may be maintained by constant immigration from a
Like the theory of island biogeography, metapopulation source population. The idea of source and sink popu-
theory applies, of course, to many situations other than lations is important because it emphasizes that local pop-
oceanic islands. The ideas of Levins were undoubtedly ulations may often not be at equilibrium. Thus, testing
influenced by those of the population geneticist Sewall theories based on equilibrium assumptions may be mis-
Wright, who envisaged a very similar process as part leading. Island biogeography theory, for example, makes
of his shifting balance theory of evolution (Wright 1931, predictions based on equilibrium assumptions. During
1968). We discuss the details of metapopulation mod- the 1960s, especially in response to island biogeography
els in Box 17A. theory, equilibrium was often the dominant assumption
Metapopulation theory can be used to look at both among ecologists. Since the 1980s, the pendulum has
population and community patterns. The theory pre- swung in the opposite direction, and nonequilibrium is
dicts the distribution of population sizes for a species generally assumed, at least at a local level. A metapop-
across a set of islands or patches, and how it changes ulation perspective moves the equilibrium assumption
over time. If one is willing to assume that species have up one level to the entire landscape. While individual
Table 17.1 Predictions about the correlation between species frequency and abundance and the shape of the frequency
distribution from four different versions of the metapopulation model
Model Predictions
Immigration Extinction Correlation Distribution Source
Dependent Dependent Positive Bimodal Hanski 1982
Dependent Independent Zero Single interior mode Levins 1969
Independent Dependent Positive Single interior mode Gotelli 1991
Independent Independent Zero Single interior mode Gotelli 1991
Note: The models are classified by the dependence of immigration and extinction rates on the fraction of sites that are occupied (see text for details).
populations are not at equilibrium, the entire metapop- occupied patches, then patch occupancy by a species
ulation is often assumed to be at equilibrium, though should have a bimodal distribution (Figure 17.7A). In
this assumption has not been tested. other words, species should tend to be either very com-
mon or very rare, although they may switch over time.
Metapopulation Patterns Hanski (1982) coined the terms “core species” for those
How well does metapopulation theory work at predict- species that are common in all communities in a land-
ing landscape patterns for plant communities? The scape and “satellite species” for the rare ones. This ver-
results are mixed. The key issue is whether plant popu- sion of the theory is often called the core-satellite hypoth-
lations have migration and extinction rates high enough esis. Some ecologists occasionally confuse the
for metapopulation processes to be a significant factor core-satellite hypothesis with the idea of source and sink
in determining species distributions and population populations; they are quite distinct from each other in
sizes. The various versions of metapopulation models that the former idea concerns characteristics of entire
make explicit predictions about two aspects of species communities. The bimodal patch occupancy distribu-
distributions (Table 17.1). First, if local extinction rates tion of the core-satellite hypothesis becomes unimodal
are dependent on the frequency of occupied patches, if either the immigration or extinction rate is occupan-
then those species occurring in a large number of patch- cy-independent. The mode is located somewhere in the
es will also have large populations when present in a middle of the distribution (Figure 17.7B)—that is, typi-
patch—a positive correlation between occupancy and cal species will be somewhat common, with the exact
abundance. If local extinction rates are independent of values depending on immigration and extinction rates.
the frequency of occupied patches, the correlation will One series of studies of tallgrass prairie communi-
be zero. Evidence from many studies shows that in most ties in Kansas (Gotelli and Simberloff 1987; Collins and
plant species, occupancy and abundance are positively Glenn 1990, 1991) found evidence of bimodal patch
correlated (see Figure 14.5; Brown 1984; Scheiner and occupancy distributions for surveys done at a scale of 1
Rey Benayas 1994). Thus, at least some versions of the km2 or less (Figure 17.8). However, a survey of 74 land-
metapopulation models are consistent with available scapes ranging in scale from 0.5 to 740,000 km2 (mean
data. 755 km2) found no evidence of bimodal distributions
The second prediction of metapopulation theory is (Scheiner and Rey Benayas 1994). Thus, at medium to
not so well borne out in plant studies. If both immigra- large scales, the data for terrestrial plants are not con-
tion and extinction rates depend on the frequency of sistent with the hypothesis that metapopulation process-
(A) (B)
25
30
Number of species
Number of species
20
20 15 Figure 17.7
Two possible distributions of species
10 frequencies. (A) Both immigration
10
and extinction rates are dependent on
5
patch occupancy rates. (B) Only one
of the rates, or neither, is dependent
0 5 10 0 5 10 on patch occupancy rates. (After
Number of patches occupied Number of patches occupied Collins and Glenn 1991.)
1981 30 1983
n = 123 n = 137
Number of species
Number of species
30
20
20
10
10
0 2 4 6 8 0 4 8 12
Number of plots occupied Number of plots occupied
1984 1985
n = 144 n = 127
Number of species
Number of species
30
20
20
10
10
0 3 6 9 12 15 18 0 3 6 9 12 15 18
1986 1987
n = 152 n = 148
Number of species
Number of species
30 30
20 20
10 10
0 3 6 9 12 15 18 0 3 6 9 12 15 18
1988
n = 137
Number of species
30
Figure 17.8
20 Species occupancy distributions over eight years in the
Konza Prairie Research Natural Area, Kansas. Eight sites
10 were sampled in 1981, 14 in 1983, and 19 from 1984 to 1988.
N is the total number of species among all sites. Each site
consisted of 20 10-m2 quadrats, a grain of 200 m2, spread
0 3 6 9 12 15 18 across an extent of 3487 hectares. (After Collins and Glenn
Number of plots occupied 1991.)
es are important, but they may be important at smaller culty is easy to overcome, though, because spatial struc-
scales of approximately 1 km2 or less. ture can be incorporated into computer simulation mod-
Why is it that metapopulation theory does not do a els (e.g., Holt 1997).
good job of predicting plant species distributions at large The biological reason is more critical (and you may
spatial scales? There are two reasons, one mathemati- have already realized what it is): the models work only
cal and one biological. The mathematical reason is sim- if immigration and extinction rates are high enough to
ple: to make the mathematics analytically tractable, sim- influence local population dynamics. Immigration rates
ple models need to assume that the landscape has no must be large to generate a positive correlation between
spatial structure—that is, that all patches are equally occupancy and abundance. Consider a set of popula-
close to one another. If dispersal rates are very high, this tions, each of which varies most of the time between 40
is a reasonable approximation, but few organisms, espe- and 200 individuals. If 5 to 6 individuals enter each pop-
cially plants, have such high dispersal rates. This diffi- ulation in any generation—a low immigration rate—
those individuals will be too few to affect the dynamics a new reserve is proposed, there is not sufficient time
of the populations or change the probability of an extinc- and money to do a detailed analysis of the population
tion event. In contrast, if 50 to 60 individuals enter each ecology of many species, or even the most vulnerable
population in any generation—a high immigration species. Instead, reserve designers have looked to land-
rate—those migrating individuals will constitute a large scape ecology theory to provide simple guidelines that
portion of each population and will have a large influ- can be applied in the hope of saving the largest num-
ence on population dynamics and extinction probabili- ber of species.
ties. This goal led (in the 1980s) to a dispute among ecol-
For most plants, immigration and extinction rates ogists called the SLOSS (“Single Large Or Several
are probably too low for metapopulation models to hold. Small”) debate. The debate centered on whether (given
Long-lived perennials dominate most plant communi- a fixed total area) it is better to create one very large
ties. For many of these species, the establishment of new reserve or many small reserves. You may ask, “Why not
individuals is episodic, perhaps occurring just once a always create very large reserves?” One answer is that
decade or less when the right set of conditions comes random environmental variation, or a major catastro-
along. Metapopulation theory has been successful in pre- phe, could potentially eliminate a single population of
dicting abundance patterns for insects and small mam- a species. Island biogeography theory, which favors
mals that have generation times of a year or less and that large reserves, does not consider environmental varia-
can move substantial distances. For plants, the theory is tion. Thus, there were reasonable arguments on both
probably most relevant to populations of early succes- sides in the debate. Interestingly, the SLOSS debate was
sional plants, weedy plants, and annuals. never resolved in an intellectual sense; rather, it was
Lenore Fahrig (1998) has recently developed anoth- ended by the realization that in real life, the alternatives
er type of metapopulation model, the “patchy popula- of single large versus several small almost never pres-
tion model.” This model is an individual-based ent themselves—the point being argued was moot. The
model—it consists of simulations in which the proper- political and economic reality is that assembling a sin-
ties of individual organisms are explicitly modeled. By gle large reserve can be expensive and very difficult, if
comparison, other metapopulation models are mean not impossible. Buying up many small, disconnected
field models—all individuals in the population are parcels of land is sometimes much easier. The consen-
assumed to have the same properties, so only mean pop- sus reached was one you might guess: preserve as much
ulation parameters are simulated. In Fahrig’s model, as you can. In any case, the SLOSS debate brought issues
though, individuals give birth, move, and die based on of landscape ecology to the fore in the planning of nature
individually defined probabilities. Both conventional reserves.
metapopulation models and individual-based models
are spatially structured as was the case in this model. Fragmentation
This model allows population dynamics to occur with- One of the greatest threats to many natural habitats, out-
in each patch, with immigration and emigration rates side of sheer destruction, is the fragmenting of entire sys-
determined by the sum of individual actions, rather than tems into smaller bits and pieces (see Figure 22.14). Even
being predetermined. large reserves may have roads and other intrusions run-
ning deep into their interiors, or even bisecting them.
Our understanding of the consequences of fragmenta-
Landscape Ecology and Conservation tion does not simply depend on theory: both descriptive
and experimental data exist to document its effects.
Reserve Design One of the most important sources of such infor-
Landscape ecology is very important to conservation mation is an ambitious long-term experiment, the Bio-
efforts. For example, the idea of using corridors to con- logical Dynamics of Forest Fragments Project (BDFFP),
nect nature reserves comes directly from landscape ecol- that was begun in the early 1980s and is still under way
ogy. While the goals of conservation efforts are quite var- in Brazil. This collaborative study—involving the U.S.’s
ied (e.g., some efforts are aimed at protecting a single Smithsonian Institution and Brazil’s Instituto Nacional
species while others are aimed at preserving communi- de Pesquisas da Amazonia (National Institute for
ties), landscape considerations are important to all of Research in the Amazon)—aims to examine the conse-
them. Many new nature reserves try to save the largest quences for species diversity of the fragmentation of con-
possible number of species (or communities) while jug- tinuous forest due to the widespread replacement of
gling the realities of private land ownership, develop- tropical forest by cropland and pastureland in Brazil
mental pressures, and public use of lands. Older nature (Bierregaard et al. 1992). The study site consists of a
reserves, such as many national parks around the world, series of forest fragments left standing amid cleared land
were developed mainly to protect scenery. Often when (Figure 17.9). There are four plots of 1 ha, three plots of
Manaus Figure 17.9

Forest fragments and control areas Amazon The Biological Dynamics of
Cattle pasture, regrowth forest River Forest Fragments Project near
Rainforest Manaus, Brazil, in the Amazon
basin. Forest fragments and
control areas are shown in
black. Gray areas are cattle pas-
tures or regrowth forest. The
rest of the area is rainforest.
The solid lines are roads. (After
Laurance et al. 1998.)
Roads
5 km
10 ha, and two plots of 100 ha, thus providing replicat-

ed samples. In addition, the nearby intact forest serves
as a control. The plots are monitored for changes in the
composition and abundance of plant and animal species. (A)
The BDFFP addresses many issues in addition to
those involved in the SLOSS debate. The simplest ver-
)
Extinction rate (
sion of island biogeography theory (in which immigra-
Immigration rate (
tion and extinction rates are a linear function of area)

predicts that the total number of species contained in the
entire collection of fragments does not depend on the
size of individual fragments as long as the total area is
the same. But this simple theory ignores some important
)
biology. Immigration and extinction rates depend on the
identities of the species involved. Some species are bet- s
ter dispersers and colonizers than others. Extinction
rates may be higher in more crowded communities (B)
because interactions among species intensify competi-
tion. Both of these factors result in nonlinear immigra-
)
tion and extinction curves (Figure 17.10). These nonlin-

Extinction rate (
earities mean that the same total area will not result in
Immigration rate (
the same total number of species. The precise sizes of the

fragments become critically important. And it is impos-
sible to predict exactly what set of fragment sizes will
maximize the number of species, because that depends
on the exact shapes of the curves. Attempting to deter-
)
mine those shapes puts us back in the position of need-

s
ing to determine the ecological requirements of the
species involved. In the BDFFP, for example, mortality Number of species (s)
rates of trees declined with fragment size, but not in a Figure 17.10
linear fashion. The rates for 10-ha fragments were near- How changing the dependency of immigration and extinc-
ly identical to those for 100-ha fragments (Figure 17.11A). tion rates on species density affects the equilibrium species
Eventually, the greater mortality rates in the smallest number on an island. (A) Linear relationships. (B) Nonlinear
fragments will result in higher extinction rates. relationships. The shape of the relationships depend on the
biology of the species, such as how the strength of competi-
Fragmentation can sometimes affect communities tion depends on individual density. Depending on the exact
in unexpected ways. Mary Cadenasso and Steward Pick- shapes of the immigration and extinction rate curves, the
ett (2001) quantified the movement of seeds from out- equilibrium species number will go up or down.
(A) unable to evolve in response to changes in the environ-

10 ment. Because the process of fragmentation may itself
Mortality cause environmental change, this problem becomes even
8 Damage more important. For instance, in small fragments of the
Annual rate (%)
Turnover BDFFP, individuals of the Brazilian canopy tree Cory-

6
thophora alta (Lecythidaceae) are all very similar geneti-
4 cally, in contrast to a lower level of genetic relatedness
in large fragments (Hamilton 1999). Because seed and
2 pollen movement are limited, inbreeding is likely to
increase, leading to further loss of genetic diversity.
1 10 100 Controls
Edges, Connectivity, and Nestedness
Fragment area (ha)
Edge effects—systematic differences between areas
(B) inside the edges of habitat patches and the interiors of
those patches—can also be important to plants. Wind
5
speed, for example, is greater at the edge of a forest than
4 in the interior, and the plants there are exposed to more
Annual rate (%)
desiccation. If winds are strong enough, trees on the

3 edges may have more limbs broken off or may have a
greater chance of being knocked down. The effect of
2
wind has been particularly important in the BDFFP
1 plots. Near fragment edges, the rates of tree damage and
mortality were three times higher than inside intact
forests (Figure 17.11B). Other edge effects include dif-
<60 60–100 101–500 >500
ferences in available light (and therefore in the magni-
Distance from edge (m)
tude of diurnal temperature changes) and humidity
Figure 17.11 between edge and interior, and in some cases, differences
Annualized rates (mean ± 1 standard error) of forest dynam- in CO2 concentrations. There may also be differences in
ic parameters of all trees in each fragment for the Biological the presence or activity of herbivores, pollinators, and
Dynamics of Forest Fragments Project. Turnover is the rate seed dispersers between edge and interior. Thus, patch
at which trees are being replaced. (A) Effects of fragment
area (size of fragment). (B) Effects of distance from the edge shape—especially the ratio of perimeter to total area—
of the fragment. (After Laurance et al. 1998.) is important to consider.
In recent years, ecologists have begun to turn their
attention to several other aspects of landscape geome-
try, such as the distances between patches and the degree
side of several forests into the forests’ interiors. Wind- to which patches are connected to one another. Con-
borne seeds, including those of invasive non-native nectivity is the degree to which a landscape facilitates
species, entered the forests in substantial numbers, an or impedes the movement of organisms (Tischendorf
important but not unexpected finding. Long-undis- and Fahrig 2000). It is a function both of the physical
turbed edges with denser, “branchier” vegetation, how- arrangement of patches and of the behaviors of organ-
ever, formed far better seed barriers than recently cut isms in response to that pattern. Different species may
edges. Old edges not only blocked far more seeds from judge the connectivity of a landscape very differently.
entering the forest, but also kept those seeds that made Much attention has been devoted to the idea of con-
it into the forest much closer to the edge, protecting the necting habitat patches with “corridors,” mainly in the
interior better. These results imply that recent increases context of animal conservation. Dispersal among habi-
in habitat fragmentation, creating many new edges, will tat patches, even if it occurs only occasionally, can be an
exacerbate the spread of invasive species into the inte- important means by which populations are preserved
rior of these habitats. in the long run, as isolated populations are more likely
Habitat fragmentation can change not only the sizes, to go extinct.
but also the genetic diversity of local populations. Small Much less attention has been given to connectivity
populations—those in small habitat fragments—are like- in plant ecology, probably because most attention in con-
ly to lose genetic diversity through the process of genet- servation as a whole has been paid to animals, especial-
ic drift (see Chapter 5). In turn, reduced genetic diver- ly vertebrates. An exception is the work of Janez Pirnat
sity (as well as small population size itself) may increase (2000), who used GIS and vegetation sampling to study
the chance of extinction because the population will be connectivity between forest fragments in Ljubljana,
Slovenia. Here, forest patches existed as urban forests and

parks, as suburban forests around the fringe of the met- Summary
ropolitan region, and as connecting forest corridors and Landscape ecology is the study of the spatial distribu-
fragments. Road building and development disrupted tion of individuals, populations, and communities and
these connections, threatening the dispersal of forest the causes and consequences of those spatial patterns.
plant species, particularly endangering the maintenance There are several ways to describe spatial patterns. Most
of diversity in the urban forests that are population sinks such measures deal with the sizes, shapes, and arrange-
for various tree species. Similarly, in many landscapes ments of patches. Other measures, such as some types
affected by natural or anthropogenic fragmentation, from of species-area curves, are not spatially explicit. These
large lava flows or fires to suburban development, the measures highlight the importance of scale in ecological
availability of foci of dispersal—patches of plants that processes. Scale has several components—grain, extent,
can serve as sources of seeds for dispersal into the sink and focus—and can have both spatial and ecological
regions—can be of critical importance in the maintenance hierarchies. How we perceive a pattern depends on the
and recovery of plant populations. scale at which we analyze that pattern. Our ability to
The nestedness of a set of communities—the degree measure patterns and examine scale effects has dramat-
to which communities tend to be subsets of other com- ically increased in the past two decades with advances
munities (see Table 16.4)—is an important landscape- in remote sensing methods and the development of new
level issue in conservation biology. In a perfectly non- analysis techniques such as GIS and spatial statistics.
nested set of communities, each community is composed More significantly, processes can operate different-
of a unique set of species; for example, the sets of species ly as scale changes. Movement—whether of individuals
growing in a forest and in a neighboring lake do not or of materials such as nutrients—is a key ecological
overlap. By contrast, in a perfectly nested set of com- process in landscapes. The movement of individuals
munities, the smallest patches have only species con- affects population dynamics and community structure.
tained in the larger patches. In a perfectly nested system, The dispersal of individuals is a critical component of
one could thus preserve all the species diversity by pre- two important theories—island biogeography theory
serving just the largest patch, while in a perfectly non- and metapopulation theory—that form the basis of land-
nested system one would need to preserve every patch. scape ecology. From these theories have come other the-
A simple application of island biogeography theo- oretical advances, such as hierarchical patch dynamics.
ry predicts that the small patches will contain random At present, however, these theories have had only lim-
samples of the entire set of species, with different species ited success in explaining the landscape patterns of plant
appearing in different patches. But fragmented land- communities.
scapes have a tendency to become more nested over An understanding of landscape dynamics and their
time: small patches often contain the same few species consequences is important in conservation biology.
because those species are the hardiest or the best dis- Nature reserves can be designed to maximize the num-
persers. Animal communities tend to be highly nested ber of species preserved. Island biogeography theory
(Wright et al. 1998), although this is less true of plant was at the center of a debate about whether it is better
communities. The reason is that local environmental to create one large reserve or many small reserves. Land-
effects play a major role in determining plant commu- scape fragmentation leads to decreases in connectivity
nity composition, and patches tend to differ in impor- among patches and increases in the amount of habitat
tant environmental parameters. In the BDFFP, immi- edges. These changes in landscape structure can have
gration of new individuals into the fragments consisted important, and often negative, effects on community
mainly of species found in disturbed areas, rather than structure and composition, including the loss of species
the forest-interior species previously found there. Thus, and decreases in the genetic diversity of populations.
there was a profound shift in the species composition of
the fragments, which was not simply a random sam-
pling of the previous species.
Additional Readings
Classic References Tischendorf, L. and L. Fahrig. 2000. On the usage and measurement of
landscape connectivity. Oikos 90: 7–19.
Hanski, I. 1982. Dynamics of regional distribution: The core and satellite
hypothesis. Oikos 38: 210–221.
Levins, R. 1969. Some demographic and genetic consequences of envi- Additional Resources
ronmental heterogeneity for biological control. Bull. Entomol. Soc. Am. Gotelli, N. J. 2001. A Primer of Ecology, 3rd ed. Sinauer Associates, Sun-
15: 237–240. derland, MA.
MacArthur, R. H. and E. O. Wilson. 1967. The Theory of Island Biogeog- Hanski, I. 1999. Metapopulation Ecology. Oxford University Press, New
raphy. Princeton University Press, Princeton, NJ. York.
Simberloff, D. S. and E. O. Wilson. 1970. Experimental zoogeography of Naveh, Z. and A. S. Lieberman. 1994. Landscape Ecology: Theory and Appli-
islands: A two-year record of colonization. Ecology 51: 934–937. cation, 2nd ed. Springer-Verlag, New York.
Dale, M. R. T. 2000. Spatial Pattern Analysis in Plant Ecology. Cambridge
Current Research University Press, Cambridge.
Harrison, S. 1999. Local and regional diversity in a patchy landscape:
Native, alien, and endemic herbs on serpentine. Ecology 80: 70–80.
Johnson, A. R. and B. N. Milne. 1992. Diffusion in fractal landscapes: Sim-
ulations and experimental studies of tenebrionid beetle movements. Ecol-
ogy 73: 1968–1983.
PART
V Global Patterns
and Processes
18 Climate and Physiognomy 351

19 Biomes 379
20 Regional and Global Diversity 403
21 Paleoecology 419
22 Global Change: Humans and Plants 273
18
C H A P T E R
Climate and Physiognomy
I
f you were to travel any considerable distance, either across a continent
or across the globe, you would see strikingly different plant communi-
ties. In some places you would see tall trees with broad leaves. Other places
would have no trees and dense grass cover, while still others would be home
mainly to shrubs. Some places would have a great diversity of plants, while
others would have few species. These patterns of variation, both in overall
plant diversity and in physiognomy—the general form of the vegetation—are
the subject of Part V of this book. In this chapter, we consider the climatic fac-
tors that control vegetation patterns and how they lead to variation in vege-
tation form. Subsequent chapters will look at large-scale patterns of vegetation
form and species diversity across the globe. Finally (in Chapters 21 and 22),
we will examine patterns through geological time and look at some predic-
tions about what may happen in the future as the global climate changes.
Climate and Weather

Any attempt to determine the causes of large-scale vegetation patterns must
start with climate. Climate refers to the long-term distribution (such as means
and variability; see the Appendix) of the weather in a given area (e.g., London
generally has cool temperatures and a lot of rain), while weather refers to the
immediate or short-term conditions (e.g., this week or this month it might be
unusually warm and dry in London). Although weather has profound effects
on the function, growth, and survival of plants, it is climate that determines
the general type of vegetation in an area (e.g., broad-leaved deciduous forest
versus desert) and influences large-scale patterns of diversity. In this chapter
we describe the patterns of climatic variation around the world (with special
attention to North America), explain the mechanisms responsible for those pat-
terns, and look at how climatic variation translates into variation in plant form.
Our understanding of the connection between climate and vegetation was
forged during the nineteenth century, spurred by the research of the German
naturalist Alexander von Humboldt. He was the first to systematize informa-
tion on the effects of altitude and air pressure on patterns of temperature and
precipitation. He also was the first to codify our understanding of how coastal
climates differ from those inland. The first map of mean monthly world tem-
peratures was published in 1848. Two decades later, in 1866, the first world
352 Chapter 18
vegetation map was produced. Over the next several

decades, a number of naturalists developed classifica-
Temperature
tions of plant communities, noting the relationships If you sit outside on a sunny day, you can feel the sun’s
between different types of communities (such as forests radiant energy on your skin. Radiant energy from the
and grasslands) and the climate at different latitudes and sun, and variation in how much radiant energy reaches
altitudes. different parts of Earth’s surface, is the primary cause of
Our understanding of weather has depended heav- variation in temperature, the first of the two major cli-
ily on the ability of researchers to gather large numbers matic factors. Heat is a measure of the total kinetic ener-
of measurements over a wide geographic area. Prior to gy—the energy of motion—of the molecules in a sub-
the mid-eighteenth century, no one had realized that stance, while temperature is a measure of the average
weather moves in predictable ways across the globe— kinetic energy of those molecules; see Chapter 3. Radi-
news traveled far more slowly than weather (McIlveen ant energy from the sun heats objects (such as plants or
1992). On October 21, 1743, Benjamin Franklin attempt- the ground) directly. This energy is primarily in the
ed to observe a lunar eclipse in Philadelphia, but was shortwave part of the electromagnetic spectrum (wave-
prevented from seeing it by a storm. Later, he was sur- lengths < 700 nm) and includes visible light. As objects
prised to learn that the eclipse was visible in Boston and are heated by sunlight, they emit longwave radiant ener-
that the storm arrived there the following day. By con- gy (wavelengths >700 nm, including infrared radiation)
tacting people living between the two cities, he was able in proportion to the fourth power of temperature (see
to reconstruct the movement of the storm. It was the Chapter 3), becoming a secondary source of heating for
development of communication technology, however, nearby objects. The warming of physical objects by sun-
that really changed the sciences of meteorology and cli- light, and their slow emittance (i.e., the storage and later
matology. The invention of the telegraph in 1835 (with release) of that heat energy over time as longwave radi-
the first message transmitted in 1844) made it possible ation, creates lags in temperature change, which we will
to organize large numbers of people to observe and fore- explore in more detail below. The transport of that heat
cast the movement of storms. The advent of airplanes energy from one place to another by atmospheric and
and the military technology developed in World War II ocean currents—a process called convection—is also a
further spurred the development of weather station net- key element in global climates.
works and made it possible to sample weather condi- The temperature at any spot on Earth’s surface is
tions high in the atmosphere. Today, we gather infor- determined primarily by the amount of radiant energy
mation about the weather with a combination of it receives from the sun. That amount is determined, in
satellites, ground-based stations, and upper-atmosphere turn, largely by the angle of Earth’s surface in that spot
probes. The ready accessibility of powerful computers relative to the sun’s rays, as well as by atmospheric con-
makes it possible to analyze these enormous data sets, ditions. Consider a spot at sea level on the equator at
and both climatologists and meteorologists are actively noon during the vernal or autumnal equinox, when the
developing more sophisticated models of climate and sun is directly overhead. At the top of the atmosphere at
weather (see Box 22A). These efforts are especially that spot, Earth receives approximately 340 W/m2/year,
important in our attempts to predict changes in weath- or 2 langleys (ly) of energy per minute (1 ly = 1 cal/cm2)
er and climate due to the effects of global warming and as shortwave radiation. If we consider the amount of
to understand how different factors influence those energy striking the top of Earth’s atmosphere to be 100%,
changes. we can trace what happens to that energy (Figure 18.1).
The two primary components of a region’s climate Almost a third of it returns to space as shortwave radi-
are temperature and moisture. We first turn to the long- ation, reflected by Earth’s surface (6%) and by clouds
term mean (average) temperature and precipitation and (17%) and scattered by the atmosphere (8%). Almost half
the effects of these average conditions on vegetation pat- (46%) is absorbed by Earth’s surface. The rest of the ener-
terns. Variation around these means is also important, gy entering the atmosphere is absorbed by clouds (4%)
and we look at two aspects of this variation: predictable and by the “greenhouse gases” (19%) in the atmosphere.
changes (e.g., it is usually warmer in the summer than (We will return to the absorption of energy by green-
in the winter) and departure from average conditions house gases shortly and will discuss it at greater length
(e.g., in a particular year it may be unusually dry in the in Chapter 22.)
Amazon basin). The time scale of variation is also impor- Earth’s surface emits energy in the longwave part of
tant, with temperature and precipitation varying on the spectrum, reradiating 15% of the initial total energy
scales from a single day, a week, a season, a year, a that reaches the upper atmosphere and transferring an
decade, up to cycles that take hundreds of thousands additional 24% as latent heat loss (see Chapter 3) and 7%
of years to complete. Here we investigate these patterns as convection to the atmosphere and clouds. The green-
of variation and how they shape plant communities. house effect results when of greenhouse gases in the
PAGE PROOF: 2ND PASS Climate and Physiognomy 353
Space
Outgoing Outgoing
shortwave longwave
Incoming
radiation radiation
(shortwave)
solar radiation
(340 w/m2 100%)
Backscattered 20
17 by atmosphere Absorbed Emission
Reflected by clouds by clouds 40
Atmosphere 8 9
by 4
clouds
6 Emission by
H2O, CO2,
etc.
19
Reflected Absorbed by
by H2O, CO2, etc. 6
surface Absorbed
by surface
46
Latent Convection Longwave

Oceans and land
heat 7 re-radiation
loss from the
24 Earth s surface
15
Figure 18.1 heat exchange also transfer heat energy from the surface to
The radiant energy balance of the Earth, expressed as the the atmosphere. The total amount of energy absorbed by the
percentage of the energy reaching the top of the atmosphere surface, atmosphere, and clouds (as shortwave radiation)
(340 W/m2/year, or 100%). This energy is reflected by the and the total amount that leaves the planet (as longwave
surface and by clouds and scattered by the atmosphere, so radiant energy, latent heat loss, and convection) are in bal-
that about 31% returns to space. Of the remainder, 46% is ance with each other. The total amount of energy entering
absorbed by the surface and 23% by the atmosphere, warm- the top of the atmosphere (as shortwave radiation) is exactly
ing both. Longwave radiation (infrared, shown by boldfaced equal to the sum of the outgoing shortwave radiation (from
numbers) is emitted by the surface, warming the atmosphere reflection and scattering) plus the outgoing longwave radia-
further as well as being lost to space. Convection and latent tion, which together sum to 100%. (After Schlesinger 1997.)
atmosphere reabsorb longwave radiation emitted by the shortwave radiant energy absorbed by the atmosphere
Earth’s surface (see Figure 22.3). Secondary heating of and Earth’s surface (46% + 23%) is balanced by the long-
the Earth’s surface then occurs by reradiation from the wave energy (69%) emitted.
atmosphere.
This reabsorption and reradiation is critical for life Short-Term Variation in Radiation and Temperature
on Earth. Without the greenhouse effect Earth would be The combination of shortwave and longwave radiation
bitterly cold everywhere, all of the time. Secondary heat- determine the ambient temperature at Earth’s surface. As
ing by longwave energy from the atmosphere accounts you travel away from the conditions described above, of
for two-thirds of the total radiant energy received at the maximal solar energy input at the equator at noon, sev-
surface. The total outgoing longwave radiation from eral factors act to determine the amount of incoming
Earth back to space is about 69% of the energy received radiation and the subsequent temperature. Most impor-
initially, coming from Earth’s surface, the atmosphere, tant is the angle of the sun, which determines the amount
and clouds. of incoming shortwave radiation. Over the course of the
Incoming energy must balance outgoing energy. day, Earth rotates on its axis, and the sun’s rays hit the
Temporary imbalances will result in Earth warming or surface at different angles. The solar angle of incidence
cooling, but eventually a new equilibrium will be is the angle that a ray of sunlight makes with a line per-
reached as Earth’s emission of longwave radiation pendicular to the surface. The smaller the angle of inci-
changes according to its new temperature. Thus, the dence, the closer to 0º from the horizontal, the larger the
354 Chapter 18
Sun’s rays 1500
Energy imput to horizontal surface

Less solar energy Maximum amount of solar energy
is intercepted by 1000
is intercepted by horizontal surface
(W/m2)
horizontal surface parallel to surface of Earth (midday)
parallel to surface
of Earth
Subsolar
point 500
(90º from
horizontal)
0º from
horizontal
Shadowed hemisphere 0
(night) 90º 60º 30º 0º
Sun’s angle from horizontal
Figure 18.2
Differences in angle and amount of solar radiation reaching nearly constant solar energy inputs throughout the year
the Earth’s surface. At the equinox, when neither hemi-
sphere points toward or away from the sun, the diagrams
(Figure 18.3B). These effects combine to produce distinct
shows the amount of solar radiation received at a single spot patterns of temperature variation across the globe (Fig-
over the course of the day. The same differences could apply ure 18.4).
to the angles and amounts of solar radiation reaching differ- Atmospheric, ground, and oceanic temperatures
ent latitudes at different times of the year (see Figure 18.3B). have a feature called lag—a delay in effect—that deter-
mines some of the daily and yearly patterns of temper-
ature variation. It is warmer in the late afternoon, even
area over which a given amount of solar energy is spread. when the sun is low in the sky, than when the sun is at
Consequently, when the incoming sunlight is at a steep a comparable angle in the morning. The coldest month
angle, any given square meter of surface receives less of the winter in the Northern Hemisphere is not Decem-
energy than it would if the sun were at a more direct ber (when the sun is lowest and the days are shortest),
angle overhead (Figure 18.2), Also, at steeper angles, but January; conversely, the hottest summer months are
the incoming solar energy must travel a longer distance July and August, not June (when the days are longest
through the atmosphere, and so more energy is absorbed and the sun most directly overhead). The cause of lag
by the atmosphere and less reaches Earth’s surface. is the storage of heat by the ground, the ocean and other
Similarly, as one moves north or south of the equa- large bodies of water, and the atmosphere itself.
tor, the angle of the sun, even at noon, decreases. The Imagine a place somewhere at a temperate latitude
actual angle depends on the time of year and how far at the winter solstice. Incoming solar radiation is at a
one is from the equator. Earth’s axis is tilted relative to minimum. As Earth revolves around the sun and the
the plane in which it revolves around the sun. Seasons year progresses, day length increases and the sun is more
are created by the differences in solar angle, and in the directly overhead. The ground absorbs solar energy and
consequent amount of incoming solar energy, over the begins to warm. The ground has a certain specific
year-long progression of Earth around the sun (Figure heat—the amount of heat energy that is required to
18.3A). The Northern and Southern Hemispheres are tilt- warm 1 gram of it by 1°C. Water has a higher specific
ed toward or away from the sun at the opposite extremes heat than the ground, so the ocean takes longer to warm
of Earth’s trajectory around the sun, at the solstices in up than nearby land. Air has a far lower specific heat
December and June. At the equinoxes, in September and than either land or water, and warms up more readily
March, Earth’s axis is parallel to its plane of revolution, than they do. Even as our imaginary spot on Earth pass-
neither pole is tilted toward or away from the sun, and es the summer solstice and incoming radiation begins to
day and night length are equal everywhere on Earth. decline, the longwave output from the now warmed
When the sun is more directly overhead in the Northern ground and bodies of water nearby continue to warm
or Southern Hemisphere, it is summer in that hemi- the air, keeping air temperatures high.
sphere. In the other hemisphere, steep solar angles result Eventually, as solar input declines, the reverse pat-
in decreased radiant energy input, and it is winter. This tern occurs: air, ground, and finally water become cool-
change in solar angle with the seasons is greatest at the er. A high specific heat results in slow cooling, just as it
poles and least at the equator. As a result, over the course causes slow warming. Coastal environments, influenced
of a year at extreme latitudes, there are large changes by the adjacent bodies of water, consequently cool and
in radiation input. In contrast, equatorial regions have warm more slowly than those in the interior of conti-
(A) March
equinox North
Arctic pole
Circle Angle of sunlight
reaching surface
Tropic of
Cancer 23.5º
Equator 66.5º
90º
Tropic of 66.5º
Capricorn
South
Summer pole Winter
23.5º
47º 0º
Sun
90º 43º
June December
solstice 66.5º 66.5º solstice
43º 90º
23.5º
Winter 0º
23.5º Summer
Angle of
axis tilt 66.5º
= 23º
(constant)
90º
Figure 18.3
(A) The axial tilt of Earth relative to the sun as 66.5º
Earth revolves around the sun over the course
of a year. (B) The graph shows the average daily
total solar radiation received at Earth’s surface
at different northern latitudes as a function of September
the time of year. (After Gates 1962.) equinox (B)
J F M A M J J A S O N D
nents, and thus experience more moderate temperatures. 1000
Solar radiation (cal/cm 2/day)
The water ameliorates very hot temperatures as cooler 0¡ Equator

offshore breezes convectively reduce heat on land. In 0¡
autumn, the water releases heat gradually, slowing tem- 10¡
10¡
perature declines on land by providing warming breezes. 20¡
20¡
These breezes may feel cold to a person, but they are 30¡
500 30¡
warm relative to the ground and the air over it. One sub- 40¡
tle implication of these lags for plants is that times of 40¡
50¡
maximum or minimum solar radiation are not the times 60¡ 50¡
60¡
of highest or lowest temperatures, either daily or sea- 70¡ 70¡
80¡ 80¡
sonally. 0
J F M A M J J A S O N D
The magnitude of daily variation in temperature—
the range from nighttime to daytime temperatures—
depends in part on local atmospheric conditions, par-
ticularly the amount of moisture in the air. At night, there typically humid air, usually has nighttime temperatures
is no incoming shortwave radiation. We do not imme- that are only a little cooler than daytime temperatures.
diately freeze after sunset, however, because the sur- In contrast, Tucson, Arizona, with very dry air, may have
rounding air acts somewhat like a blanket, holding heat dramatic swings, with nighttime temperatures 20°C or
and releasing it as longwave radiation all night. Moist more cooler than those during the day. As altitude
air can hold more heat than dry air, as we will see below. increases, the density of the air decreases—the blanket
A place like Tokyo, adjacent to the Pacific Ocean with is thinner. As a result, areas at high elevations, such as
356 Chapter 18
(A)
–30 –20
–10 –10 –20
–30 0 –30
–40 –20 10
–60 20
–50 –10 –20
0 10 30
20 30 40
40 50
50 60
60
70 70
80
80
90 80
70 70
60 60
50 50
(B)
40
40
50
50
60
60 70
90
70
90
80
80
80
70 70
60 60
50
50
40 40
Figure 18.4
Mean temperatures (Farenheit) across the globe in (A) January and (B) July.
(After Rumney 1968.)
mountaintops, also tend to have large swings in tem- Africa. These spots can experience daily temperature
perature from day to night. Plants must be adapted to swings approaching those experienced between sum-
tolerate the temperature ranges, as well as the average mer and winter at mid-latitudes, although obviously
conditions, where they live. Elevation and latitude joint- they also receive much more radiation during the day.
ly affect climate in interesting ways. As indicated above,
equatorial regions receive nearly constant amounts of Long-Term Cycles
radiation throughout the year. But there are high-ele- Besides the daily and yearly variations in solar radiation,
vation regions at the equator, such as the Ecuadorian there are also longer-term cycles. One of these is the 11-
Andes in South America and the Kenyan highlands in year sunspot cycle. Over this cycle, the amount of solar
radiation reaching Earth varies up to 0.1%. A 22–23-year was first studied by the astronomer Andrew E. Douglass,
double sunspot cycle is also known, and is related to who used the long-term record of good and bad years for
reversals of the solar magnetic field. These cycles prob- plant growth recorded in tree rings. In doing so, he
ably do not directly affect plant growth or local temper- invented the field of dendrochronology (see Chapter 13).
atures. Rather, they affect atmospheric circulation and At very long time scales, there are several different
precipitation patterns, as we will see below, resulting in kinds of changes in Earth’s orbit around the sun that ^
indirect effects on plants due to changes in weather pat- influence climate, collectively called Croll-Milankovic
terns. The linkage between the sunspot cycle and climate effects (Figure 18.5). The longest of these orbital cycles
(A) Ellipticity of orbit Most

Cyclic period = 100,000 years circular
orbit Most
Aphelion elliptical
orbit
Perihelion
(B) Degree of axial tilt
Cyclic period = 41,000 years
24.5° 23.5° 22.1° N
N N
Maximum tilt Present Minimum tilt
(C) Direction of axial tilt

Cyclic period = 22,000 years
Vega (North Star) Polaris (North Star)
N N
Figure 18.5 ^
Croll-Milankovic cycles are periodic changes

in (A) the ellipticity of Earth’s orbit, (B) the
degree of axial tilt, and (C) the direction of
axial tilt. Each affects the amount of solar
radiation received at different points on
Earth’s surface at different times of the year.
These cycles interact to determine long-term
climatic patterns. (After Gates 1993.) 11,000 years ago Present
358 Chapter 18
0.05 Past Future Figure 18.6 ^

Ellipticity
Past and predicted Croll-Milankovic cycles.

of orbit
0.03 Sometimes the cycles can reinforce each other;

for example, when both ellipticity and the
0.01 degree of axial tilt are large. At other times they
can cancel each other. The precentage of axial
tilt is the amount of wobble. (After Gates 1993.)
24.5
24.0
of axial tilt
Degree
23.5
23.0
22.5
22.0
+3
of axial tilt
Direction
–3
Present
250 200 150 100 50 0 –50 –100

Thousand years ago
is a 100,000-year cycle of change in the shape of Earth’s Northern Hemisphere winters currently are milder than
orbit around the sun, from nearly circular to a more ellip- Southern Hemisphere winters.
tical trajectory. Currently the orbit is almost circular, so All three of these cycles thus modify seasonality, and
that the difference between Earth’s nearest point to the they can work together or in opposition to magnify or
sun and its farthest is only 3.5%. When the orbit becomes minimize seasonal variation (Figure 18.6). While the
more elliptical, this difference between the shortest and effects may be small, they are sufficient to cause large-
longest distance of Earth from the sun may be as great scale and long-term changes in Earth’s climate and may
as 30%. Clearly, such changes in orbital shape will affect be responsible for the cycles of ice ages that Earth has
the range of seasonal variation at temperate and polar experienced over the last million years (see Figure 21.8).
latitudes, because the amount of incoming solar radia- We look at the most recent of these ice ages and its after-
tion will change. math in Chapter 21.
The next longest pattern of variation is a 41,000-year Recently there has been extensive debate about glob-
cycle in the degree of tilt of Earth’s axis from 22.1° to al warming, an overall rise in global temperatures. Part
24.5°; currently, the tilt is 23.5°. These changes in the axial of the debate stems from a genuine scientific question:
tilt will greatly affect the range of seasonal variation, is the recent warming of the climate likely to be caused
because changes in the angle of the sun will change the by human addition of greenhouse gases to the atmos-
amount of solar energy input at any given spot in tem- phere, or does it reflect natural cyclic changes and per-
perate and polar latitudes. haps some random variation in the weather? Without
Finally, there is a 22,000-year wobble of Earth’s axis. considering the actual data, either hypothesis (or both)
This wobble is similar to what you would see on a top could be plausible. Are the warmer temperatures we are
that is beginning to slow down, so that its stem no longer experiencing part of a long-term trend, or are they mere-
points straight up but is beginning to circle around. The ly a short-term blip that will soon be reversed? If this
result of this wobble is to change whether the Northern warming is part of a trend, to what extent is it due to
or the Southern Hemisphere is pointing toward the sun human actions? Certainly, it is very difficult to separate
during the time of Earth’s nearest approach to the sun. natural variation in weather and climate from anthro-
Earth is now a little bit closer to the sun at the winter sol- pogenic changes. In any case, one may well wonder
stice in the Northern Hemisphere (the summer solstice what the answers to these questions mean for the future:
in the Southern Hemisphere) than at the opposite point Is the observed warming a harbinger of an increasingly
in its orbit. Changes in the direction of the axial tilt will warm climate? We investigate this issue in depth in
affect the intensity of seasonal variation; for example, Chapter 22.
urated with moisture, which falls as rain. Equatorial

Precipitation regions consequently tend to have high rainfall.
Besides temperature, the other major aspect of climate Above those clouds, the conveyor belt of air con-
is moisture: precipitation, the amount and pattern of tinues to move. Moisture continues to be wrung out of
rainfall and snowfall, and relative humidity, the water the air as this rising, cooling, and condensation process
vapor in the air as a percentage of the amount in fully continues. Rising air cools at the adiabatic lapse rate
saturated air at the same temperature. Precipitation when no heat is exchanged with surrounding parcels of
clearly has a major determining effect on the type and air. Air saturated with moisture cools at 5.4°C/1000 m,
amount of vegetation found in a region. Humidity lev- the saturated (or moist) adiabatic lapse rate, while dry
els are also important ecologically because fog may be a air cools more rapidly, at the dry adiabatic lapse rate of
critical source of moisture for plants in certain habitats 9.8°C/1000 m. The reason these rates differ is that as
(fog occurs only at saturating humidity) and, more gen- air saturated with moisture rises and cools, the water
erally, because humid air holds more heat than dry air. that condenses out of the air releases heat to the air—the
Water molecules in the atmosphere are very effective at opposite of what happens when a surface is cooled by
absorbing longwave radiation and have a high specific evaporating water (latent heat loss; see Chapter 3).
heat. Therefore, humid air acts as a reservoir of stored Eventually, air can no longer rise any higher in the
heat. Clouds can also absorb and reflect incoming solar atmosphere. But the continual addition of new air from
energy. A number of factors determine how precipita- below means that this air has to go somewhere. The ris-
tion and water vapor are distributed over the globe and ing mass of air at the equator is thus forced to move
during the year. away from the equator northward and southward
toward the poles. As this air moves poleward, it slowly
Global Patterns cools further because it is receiving less radiant energy.
Warm air is less dense than cold air, and therefore rises, By the time it has reached about 30° latitude, it has
like a piece of wood floating on water. That is why hot cooled sufficiently to become denser and begin sinking.
air balloons go up, and why it is warmer near the ceil- The falling air becomes warmer, through the reverse of
ing of a room than at the floor. Warm air also has a the process that occurred at the equator. Already dry, this
greater capacity for holding moisture than does cooler rewarmed air has a large capacity to hold moisture, but
air. You can see this directly by observing moisture con- contains little water vapor. These falling air masses
densing on a cold beverage glass on a warm, humid day. absorb any available moisture, and release little of it, cre-
As the air near the surface of the glass cools, its capaci- ating intensely arid regions where they meet Earth’s sur-
ty to hold moisture is reduced, so moisture condenses face. This explains why the great deserts of the world
as liquid on the cool surface of the glass. are largely found in two bands, at 30° N latitude (the
As explained previously, solar radiation is greatest Sahara, Arabian, and Sonoran) and 30° S latitude (the
at the equator, and the air there is warmed. This warm- Atacama of Chile, the Kalahari of southern Africa, and
ing creates a pattern of rising air in a belt around the the Great Sandy Desert and Great Victoria Desert of Aus-
equator. Warm tropical air masses pick up moisture from tralia).
the land surface, both from transpiration by plants and Finally, once the air nears the ground, it must move
by evaporation from the ground, and in large amounts either toward the poles or toward the equator. The Cori-
from the oceans by evaporation. These warm, moist olis effect (Box 18A) causes air to veer to the right (rela-
masses of rising air establish a gigantic three-dimen- tive to the direction in which it is moving) in the North-
sional conveyor belt for moisture and energy. ern Hemisphere and to the left in the Southern
This pattern of air movement has two outcomes: Hemisphere. For the air moving from 30° N and 30° S
large amounts of precipitation in equatorial regions, and toward the equator, this effect creates the trade winds in
air flowing toward the poles at high altitudes. The air both hemispheres, which move from northeast to south-
rising over the equator begins to cool as it reaches high- west in the Northern Hemisphere and from southeast to
er altitudes. Why? Air pressure decreases with altitude northwest in the Southern Hemisphere. Air masses mov-
because as a given packet of air rises, there is less atmos- ing poleward from 30° N also veer to the right, while air
phere above it pushing down on it. The rising warm moving poleward from 30° S veers to the left. This pat-
air therefore expands, because gases at lower pressures tern creates the prevailing westerlies (winds moving
expand to fill up a greater volume. Expanding air cools from west to east ) in both hemispheres. Early travelers
as energy is used to push the air molecules farther apart. in sailing ships going between Europe and North Amer-
The cooler air now cannot hold as much moisture as it ica took advantage of these patterns, allowing the trade
did, and some of the water vapor it contained condens- winds to carry their ships southwestward across the
es. Clouds form, and eventually the clouds become sat- Atlantic on their journey out and returning to Europe
360 Chapter 18
BOX 18A
The Coriolis Effect
You may have heard the persistent left, and if they are in the Northern is found in the Southern Hemisphere,
myth that water goes down toilets and Hemisphere, they tend to be deflected but there objects are deflected to the left.
drains in a clockwise direction in the to their right. Another way to think of the Coriolis
Southern Hemisphere, but spins coun- Why does this strange phenomenon effect is to ask whether objects are
terclockwise in the Northern Hemi- occur? Odd though it sounds at first, speeded up or slowed down relative to
sphere. This myth is false (as we know, the surface of Earth does not rotate at Earth’s surface, considering the direc-
having personally performed the exper- the same speed everywhere. If you tion in which Earth is rotating. Earth
iment), but the basis of the claim—the stand on the equator, in one day your rotates counterclockwise if you are
Coriolis effect—is true. In fact, the gen- body will have traveled much farther looking down at it from the North Pole,
eral pattern of circulation in the oceans in space than if you were standing on or clockwise if you are looking up at it
is clockwise in the Northern Hemi- the Arctic Circle—this must be so if you from the South Pole. Objects moving
sphere but counterclockwise in the are to arrive back at the same place in toward the equator in either hemi-
Southern Hemisphere. Understanding 24 hours, because there is a lot farther sphere veer in a direction opposite to
the Coriolis effect is fundamental to to go at the equator than at the Arctic that of Earth’s rotation because they are
understanding many patterns in glob- Circle. An object, whether it be a mis- relatively slower than the ground at
al climate and weather, including the sile or an air mass, moving toward the their destination, while those moving
reasons for the general patterns of equator from 30° N initially has the toward the poles are deflected in the
atmospheric and oceanic circulation. same rotational speed as the ground same direction as Earth’s rotation
One way of summarizing the Corio- where it is “launched;” as Earth rotates because they are speeded up relative to
lis effect is this: Objects moving toward on its axis, the object moves along with the ground surface.
or away from the equator in the North- Earth at the same speed as the ground This means that air masses moving
ern Hemisphere are deflected toward surface. As this object is shot or forced from 30° latitude toward the equator
their right, and in the Southern Hemi- southward toward the equator, it are deflected toward the southwest in
sphere toward their left. If you were in retains its movement in the direction of the Northern Hemisphere and toward
Paris and you fired a missile due north, Earth’s rotation, but as it gets closer to the northwest in the Southern Hemi-
it would land somewhat to the north- the equator it is moving at a speed slow- sphere—these are the trade winds. Air
east. If you fired the missile due south, er than Earth’s surface beneath it. Rela- masses moving from 30° latitude
it would land somewhat to the south- tive to the surface, it will be deflected to toward the poles are deflected toward
west. The reverse would be true in the its right (toward the southwest). An air the northeast in the Northern Hemi-
Southern Hemisphere: a missile fired mass originating at the equator and sphere and toward the southeast in the
due north from Sydney would land to moving in the opposite direction, Southern Hemisphere—these are the
the northwest, and one fired due south toward the North Pole, will have a rota- westerlies. In turn, these wind patterns
would land in the ocean, to the south- tional speed faster than that of the sur- are the major reason that ocean currents
east. Moving air masses do the same face of Earth farther north, so it, too, move clockwise in the Northern Hemi-
thing: if they are in the Southern Hemi- will veer to the right, being deflecting sphere and counterclockwise in the
sphere, they tend to be deflected to their to the northeast. The analogous pattern Southern Hemisphere.
with the westerlies farther north. These large-scale pat- moves at high speeds from west to east. The Northern
terns of moving air masses have critical influences on Hemisphere jet stream, especially its precise position at
world climates because they convey moisture and ener- any given time, is one of the primary factors determin-
gy over great distances, influencing conditions in the ing short-term weather patterns at mid-latitudes. The
lands they cross. overall patterns of global air circulation are summarized
The three-dimensional conveyor belts of air that rise in Figure 18.7B.
at the equator, traverse a distance, fall, and then move The movement of air masses at the Earth’s surface is
back again along Earth’s surface are known as Hadley a major factor in determining surface oceanic currents
cells (Figure 18.7A). A similar (but weaker) process oper- (Figure 18.8), which in turn play a major role in terres-
ates in polar regions, with air rising at about 60° latitude trial climates. Because of the high specific heat of water,
and sinking at the poles; polar regions also tend to be these currents are even more effective conveyors of heat
very dry. Between these two sets of Hadley cells in the energy than the moving air masses. They bring water
upper atmosphere is the jet stream, a river of air that warmed or cooled far away to the coasts of all of the con-
North pole
Earth’s Speed at pole

We close by returning to the story of rotation 0 km/24 h
toilets and drains: Why don’t they do
what is so widely claimed? It’s a matter
of scale: water moving poleward across
your sink cannot possibly be deflected
very far by Earth’s rotation—unless you Speed at 45ºN:
have a truly enormous sink! Russia ~10,500 km/24 h =
440 km/h
Initial speed
faster than
earth’s surface
Africa
at destination
China
Initial speed
India
slower than
earth’s surface
at destination
Equ
ator
Initial speed
faster than
earth’s surface Speed at Equator
at destination 40,000 km/24 h =
Initial speed 1700 km/h
slower than Australia
earth’s surface
at destination Speed at 45º S:
~10,500 km/24 h =
440 km/h
Speed at pole
0 km/24 h
Earth’s Destination
rotation
South pole Actual Initial
trajectory trajectory
The Coriolis effect causes moving objects in the Northern Hemisphere to veer to their right, and
those in the Southern Hemisphere to veer to their left. The speed of Earth’s rotation at the equator, Origin
at 45° latitude, and at the poles is shown.
tinents north of Antarctica and to most islands, adding surface, so when they are blocked by continents, they
or removing heat energy wherever they travel, and thus are forced to go either north or south (Figure 18.8). This
have a major influence on local temperature and mois- effect creates gyres, enormous circles of water, that move
ture regimes. clockwise in the Northern Hemisphere and counter-
The prevailing winds described above push masses clockwise in the Southern Hemisphere. Major gyres exist
of water, causing currents that can act like rivers of water in the North and South Atlantic, North and South Pacif-
with their own distinct temperatures, salinities, and ic, and Indian Oceans. The gyres carry heat energy to
nutrient concentrations. These oceanic currents move and away from the areas they traverse. The Gulf Stream
through the surrounding ocean waters. Like the winds in the North Atlantic, for example, brings tropical waters
pushing them, these currents are deflected to the right northeastward, warming the island of Bermuda and
in the Northern Hemisphere and to the left in the South- resulting in a far warmer climate in Ireland, Great
ern Hemisphere by the Coriolis effect. Britain, and other parts of northern and western Europe
Unlike air masses, currents cannot rise above the than they would have based on their latitudes alone. The
362 Chapter 18
(A) Figure 18.7

(A) The pattern of rising and falling air in a Hadley cell in the
Northern Hemisphere. Warm air rises near the equator, flows
Warm air Air radiates heat toward the poles, cools, sinks, and flows back toward the equator.
flows poleward to space and ground (B) The general pattern of air circulation at Earth’s surface and for
some of the major upper atmosphere air currents. The Hadley cells
that match the diagram in (A) are shown in green.
Hot air rises Cold air sinks
and is cooled and is heated
by expansion by compression
Cool air flows (B)

Ground warms air toward equator
Polar
Hadley
Earth’s surface cells
Equator 30°N
Polar easterlies
Jet stream Jet stream

L L
Westerlies Tropical
Hadley
H H cells
STH 30°
Northeast trades
Intertropical convergence
0°
Southeast
trades
STH
30°
H H
Westerlies
L
Jet stream Jet stream
Polar easterlies
Polar
Hadley STH Subtropical high
cells pressure cells
L Low pressure cells
Peru Current brings frigid Antarctic waters north along H High pressure cells
the western coast of South America, causing a cooling
and drying effect there. We will consider some of the
effects of these currents when we look at the El Niño viously, cools and loses moisture. The windward sides
Southern Oscillation and related phenomena later in this of mountains thus tend to be areas of high precipitation.
chapter. Eventually the air is pushed over the tops of the moun-
tains and falls down the other side. The falling air warms
Continental-Scale Patterns again, and (like the descending air at 30° latitude) tends
Precipitation patterns at continental and local scales to pick up and hold any remaining moisture. Thus, the
are strongly affected by two factors in addition to those leeward sides of mountains tend to be areas of low pre-
just described: the distribution of mountain ranges and cipitation. These areas are known as rain shadows (Fig-
proximity to large bodies of water. When air moving ure 18.9). This process of pushing air over mountains
over Earth’s surface meets a mountain range, it has also explains why mountaintops tend to be windy
nowhere to go but up. This rising air, as explained pre- places.
Oyashio Labrador
Current Current North
Atlantic
Current
Kuroshio Canary
California Gulf Current
Current Stream
Current
North Equatorial Current North Equatorial Current
Equatorial
Equatorial Countercurrent Countercurrent
South Equatorial Current South

Equatorial
Current
Peru
Current Benguela
Brazil
Current
Current Agulhas
Current
Falkland
Current
West Wind Drift West Wind Drift
Figure 18.8
The world’s major oceanic currents.
Even a single mountain can cast a rain shadow. For

Rain shadows occur in many parts of the world and example, Mauna Kea on the island of Hawaii rises 4200
at various scales. One of the driest areas of the world, meters above sea level. The eastern face of the mountain
the Gobi Desert of central Asia, is in the rain shadow of receives as much as 750 cm of rain a year. Just 50 km
the Himalayas. The Pampas grasslands of Argentina, away on the western face, less than 50 cm of rain falls in
at mid-latitudes in South America, are in the rain shad- a year (Figure 18.10). This pattern is repeated on islands
ow east of the Andes caused by the prevailing wester- in the Caribbean with sufficiently high mountains—for
lies. Farther north in South America, the air masses mov- example, the island of Puerto Rico has a rainforest on
ing to the northwest across the Atlantic, laden with one side of its mountains and semi-arid conditions on
moisture, cross the eastern Amazon basin and are forced the other side.
to rise when they reach the Andes at their eastern edge. Land masses can be affected by their proximity to
This pattern results in very high rainfall on the eastern oceans or other large bodies of water in several ways.
side of the Andes at latitudes north of 30° S, and extreme Air passing over the water can pick up moisture and
aridity on the western side of the Andes. deposit it on land as rain or snow, or the land can tend
to lose moisture to the body of water, depending on rel-
ative air and water temperatures. One example of the
first phenomenon is the “lake effect” around the Great
Lakes of North America: winter storms tend to deposit
much more snow as one moves eastward through this
Windward
Leeward
(rain shadow)
Figure 18.9
Rain shadows are caused when warm, moist
air is forced upward and over a mountain or
mountain range, depositing high levels of pre-
Water cipitation on the windward side. The rain
shadow is created where dry air descends on
the leeward side of the mountain.
364 Chapter 18
102 Figure 18.10

76 127 Precipitation patterns on the island of Hawaii. Contours
191 254 show centimeters of rain per year. The windward side is on
51 the west. Mauna Kea and Mauna Loa are the two largest
318
25 318 mountain peaks on the island. (After Carlquist 1980.)
381 254
Kohala 191
mountains 445 191
254
191
Kawaihae 318
381
51 508
25 Mauna Kea
51 38
Hilo
762
38
635
191 381 318
51 254 381
76 76 102 127
318 254
191
102 191 508
Kona 127
coast 127
102
38 Mauna Loa
102 191
Kilauea 191
51
76
127
N 102 318 102
318
127
Pahala
191 254
127
Hawaii
102
76 76
51
51
although just offshore there are cool, foggy air masses.

Coastal southern California also has limited rainfall
region because they gain moisture from the lakes. That (though its vegetation is chaparral rather than desert)
is why Buffalo, New York (at the eastern edge of Lake because the cold California Current prevents oceanic air
Erie) is locally famous for its winter blizzards, while masses from delivering rain during much of the year. In
nearby Toronto, Ontario (on the north shore of Lake winter, the land cools enough to allow rainfall there
Ontario, and thus out of the way of most storms off the some of the time. Farther north, however, the land is
lakes) receives few blizzards. much cooler, and the cool, moist prevailing westerlies
At a larger scale, when air masses cooled by cold dump large amounts of precipitation, creating temper-
oceanic currents travel to lower latitudes and meet warm ate rainforests in Oregon, Washington, and British
land masses, the air warms as it passes over the land, Columbia. Even there, much of the rain falls in winter
absorbing moisture and preventing precipitation. If this when the ground is coldest.
air veers offshore, it can even tend to lose moisture to Air circulation patterns and geographic features
the ocean, compounding the drying effect on the land. interact to produce continental-scale patterns of precip-
This phenomenon occurs to some degree on the west- itation (Figure 18.11). In North America, for example, the
ern coasts of all continents, all of which have cool cur- major mountain ranges (Appalachians, Rockies, and Sier-
rents offshore. The coastal Atacama Desert in Chile, for ra Nevada) run in a north-south direction. The primary
example, is also affected by the cold Peru Current, and direction of air flow at mid-latitudes, as we have just
as a consequence it is exceptionally arid. Some areas in seen, is west to east. These factors combine to produce a
that desert have been rain-free for stretches of ten years, distinct pattern of precipitation across the continent.
Arctic circle
Tropic of
Cancer
Equator
Tropic of
Capricorn
Under 25 cm Figure 18.11

Global patterns of mean annual
25 — 50 cm precipitation. (After Rumney
51 — 100 cm 1968.)
101 — 150 cm
151 cm and over
Annual precipitation (cm)
150
100
fpo Figure
Number 18.11
Author:
GUREVITCH
Project Title:
Ecology of Plants
0OK Correx Art. Ed.

50
Author's review Editor's review
Along the West Sinauer

Coast, from northern California ISBN #: to
0-87893-291-7
Identifier GUR1811.1811
Great Salt Lake Basin
OK Cx.
Sacramento Valley
southern Alaska, there is a region of high precipitation

Final Size (w x h ):
Lahontan Basin
Coastal ranges
Sierra Nevada
Wasatch Mts.
Pacific Ocean
Image: New Dev. Ed.

Green River
Color: 2-C 42p X 32p6 Rocky Mts.

Uinta Mts.
where moisture-laden air coming in from the Pacific OK Cx.

Date (Y/M/D):
Great Basin
runs into mountain ranges (and the ground is cool

Proof 1st 2nd 3rd final
02/04/10
Initials
Bio. Ed.
enough to allow precipitation, at least in theImagineering
cooler Artist:
sm,jc,sm Date
OK Cx.
months). To the east of these mountains it is dry. The

Elevation (m)
Intermountain Desert of Nevada, for example, is in the

3000
rain shadow of California’s Sierra Nevada (Figure 18.12). 1500
A second area of relatively high precipitation occurs on 0
the western side of the Rockies. To the east of the Rock- 200 km
ies there is another rain shadow. Precipitation levels
gradually increase toward the Atlantic seaboard. The Figure 18.12
Transects showing the patterns of elevation and precipita-
increased moisture in the Midwest and east comes pri- tion across western North America at 40° latitude (from the
marily from moist air moving up from the Gulf of Mex- coast of northern California, through Nevada and Utah, and
ico (Figure 18.13). The Appalachian Mountains are not ending near Boulder, Colorado). (After U.S.D.A. Yearbook
high enough to create a rain shadow. 1941.)
366 Chapter 18
Continental Figure 18.13

Arctic Principal air masses and their direction of
(winter only) movement at different seasons in North Amer-
ica. (After Rumney 1968.)
Maritime
polar
Continental
polar Maritime
(all year) polar
Winter
only
Winter only
Maritime
tropical
Summer
only Maritime
tropical
a nc er
Tropic of C
Maritime
tropical
Seasonal Variation in Precipitation mate, as it is typical of areas around the Mediterranean

Variation in precipitation, as in temperature, occurs on Sea. In North America, mediterranean climates are
several different time scales. On a scale of days, precip- found in coastal southern and central California; other
itation varies as weather systems move across the globe. regions that exhibit this pattern include western Aus-
These weather systems are caused by the movement of tralia, southern Chile, and the southwestern tip of Africa.
high and low pressure areas. Precipitation usually occurs Continental climates are found in the interior of large
at the leading edge of high pressure systems. On aver- continents at mid-latitudes, as in the North American
age, these systems take several days to move through an Midwest and central Russia. Precipitation here is also
area, creating alternating wet/dry cycles. Weekly and highly seasonal, with the greatest amount in the sum-
seasonal variation in larger-scale phenomena, such as mer, although there is generally some precipitation year
the position of the jet stream, causes variability on a dif- round (Figure 18.14B). Maritime climates are often
ferent spatial and temporal scale. found along the eastern edges of continents. Here pre-
The most important scales of variation in precipi- cipitation is more uniform throughout the year, although
tation are seasonal and yearly. Annual variation in tem- it comes as rain in the summer and snow in the winter
perature helps to drive variation in precipitation. Sev- (Figure 18.14C). In equatorial regions, temperatures are
eral different kinds of annual patterns of precipitation high throughout the year. Precipitation in the Tropics
exist. Along western continental margins there are dis- may show one of three patterns: relatively constant pre-
tinct seasonal patterns, with rainy winters and dry sum- cipitation throughout the year, a distinct wet season and
mers (Figure 18.14A). Where temperatures are warm, dry season, or two wet/dry cycles in a year (Figures
this pattern is often referred to as a mediterranean cli- 18.14D–F; see also Figure 19.6).
(A) (B) (C)
Eureka, CA Lincoln, NE New York, NY

Lat. 40° 48' N Alt. 43 ft Lat. 40° 49' N Alt. 1166 ft Lat. 40° 43' N Alt. 132 ft
Mean annual Mean annual Mean annual
precipitation 36.2 in. precipitation 25.7 in. precipitation 43.4 in.
100 20
90 18
Precipitation (in)
80 16
Temperature (°F)
70 14
60 12
50 10
40 8
30 6
20 4
10 2
0 0
J F M AM J J A S O N D J F M AM J J A S O N D J F M AM J J A S O N D
(D) (E) (F)
Andagoya, Colombia Madras, India Colombo, Sri Lanka

Lat. 5° N Alt. 197 ft Lat. 13° 04' N Alt. 51 ft Lat. 6° 54' N Alt. 24 ft
Mean annual Mean annual Mean annual
precipitation 264.3 in. precipitation 50.0 in. precipitation 93.1 in.
120 24
110 22
100 20
90 18
Temperature (°F)
Precipitation (in)
80 16
70 14
60 12
50 10
40 8
30 6
20 4
10 2
0 0
J F M AM J J A S O N D J F M AM J J A S O N D J F M AM J J A S O N D
Figure 18.14 Atlantic Ocean, has a maritime climate. (D–F) Three tropical
Average monthly temperature and precipitation for various locations with different patterns of precipitation. (D)
locations across the globe. (A–C) Patterns for three locations Andagoya, Colombia, in South America, has constant pre-
at approximately 40° latitude in North America. (A) Eureka, cipitation year-round. (E) Madras, India, in southern Asia,
California, near the Pacific Coast, has a mediterranean cli- has a rainy and a dry season. (F) Colombo, Sri Lanka, has
mate. (B) Lincoln, Nebraska, at midcontinent in the Great two distinct rainy seasons, separated by drier periods. (After
Plains, has a continental climate. (C) New York City, on the Rumney 1968.)
368 Chapter 18
Figure 18.15 (A) Normal year

Pacific Ocean events (A) during
Rising air Upper atmospheric
a normal year and (B) during an
air currents
El Niño year. Events just above
and below the ocean surface, in Falling air
the deeper waters below the
surface, and in the upper Rain
atmosphere, as well as the con- Trade winds
sequences for rainfall patterns, Cool surface water Dry
are shown.
Ocean currents
Australia Warm water South

Thermocline America
Upwelling of
Cold water cold water
Pacific Ocean
(B) El Niño year
Upper atmospheric
air currents
Falling air Rising air Rising air
Dry Rain
(Trade winds slacken) Warm surface water
Weak ocean currents
No upwelling
Australia Warm water South
Thermocline America
Cold water
Pacific Ocean
The El Niño Southern Oscillation ENSO effects occur over a very large geographic
Besides these weekly and yearly cycles in precipitation, scale, influencing temperatures and rainfall from Africa
other cycles also exist. Of these, the most important is a to Australia, across the Pacific, and all the way across
3–7 year cycle called the El Niño Southern Oscillation North and South America, affecting by some estimates
(ENSO). Narrowly defined, the ENSO determines about 75% of the globe (among the few places that are
weather patterns in a band from 20° N to 20° S across the not much affected are northern Europe and northern
Pacific Ocean. However, ENSO is closely interconnect- Asia, although parts of southern Asia are strongly affect-
ed with other global atmospheric and oceanic effects. ed). Remarkably, climatologists began to fully recognize
Together, ENSO events and their associated effects drive and study this phenomenon only during the record El
some of the most dramatic global fluctuations in weath- Niño year of 1982–1983. The study of ENSO and the role
er, although the strength of the events, and the magni- of the Pacific Ocean in this cycle has transformed our
tude of their effects, can vary substantially. understanding of world weather patterns, and clima-
tologists are still refining their understanding of all of its As an example, the very severe El Niño year of
effects. These effects may also be changing due to glob- 1997–1998 (Figure 18.16) resulted in extreme drought
al warming (see Chapter 22). in Australia, Indonesia, and southern Africa.
El Niño (literally “the boy”) is named for the Span- Warm, moist air now rises above the eastern Pacif-
ish term for the infant Jesus, because its effects were ini- ic, causing heavy rainfall in ordinarily dry regions such
tially noticed off the coast of Peru beginning around as southern California, northern Chile, and northward
Christmas. El Niño is only one half of the ENSO; it alter- to the coasts of Peru and Ecuador. This rain can result in
nates with La Niña (“the girl”) years, during which con- severe flooding and massive landslides. Finally, a com-
ditions are reversed. plex change in high-altitude winds 12 km over the Pacif-
The heart El Niño is a cycle of tropical Pacific oceanic forces a jet of air to move eastward, past Central Amer-
ic and atmospheric conditions. During an ordinary year, ica, and then across the Atlantic to Africa, depressing
the trade winds in both hemispheres blow west across hurricane activity in the Atlantic and causing other dis-
the Pacific, pushing strong ocean currents along in the turbances in normal weather patterns. These changes in
same direction (Figure 18.15A). The water being pushed upper-level atmospheric winds in the southern Pacific
away from the western coasts of North and South Amer- also force changes in the jet stream in northern mid-lat-
ica is replaced with very cold, nutrient-rich water itudes, spreading El Niño’s effects on weather to a far
dragged up from the ocean bottom. This upwelling of greater portion of the globe.
cold bottom water in the eastern Pacific supports Eventually, these conditions reverse. During La
extremely productive marine food webs as well as affect- Niña years, at the opposite part of the ENSO, strong
ing global climates. Strong trade winds push warm sur- trade winds push warm surface waters into the far west-
face waters into the far western Pacific north of Aus- ern Pacific north of Australia. Evaporation is intensified
tralia. Warm water has increased evaporation, bringing in the eastern Pacific, further strengthening the trade
rainfall to the islands of the western Pacific and to east- winds. Meanwhile, the movement of surface waters east-
ern Australia and Asia. ward causes a strong upwelling of cool, deep waters in
When an El Niño year commences, the trade winds the western Pacific along the coast of South America.
slacken (Figure 18.15B). As a result, upwelling in the The Equatorial Countercurrent, which flows counter to
eastern Pacific ceases, with huge negative repercussions the prevailing winds, is relatively weak. The effects on
for fish populations, marine mammals, seabirds, and weather patterns are essentially the opposite of those
other organisms that are ultimately dependent on the in an El Niño year. At some point, the positive feed-
nutrient-rich bottom water. Because this cold water back that maintains this system reverses, and El Niño
never makes it to the surface, warm surface water now conditions return again.
spreads across the tropical Pacific from east to west. Sur- ENSO is a succession of alternating events. We do
face temperatures in the central Pacific Ocean become not completely understand what triggers the changes
as much as 5°C warmer than normal, leading to drastic between El Niño and La Niña conditions, but the ENSO
decreases in rainfall in the western and central Pacific. clearly depends on a complex interplay between ocean
3
Standardized ENSO index
2
El Niño
1
–1
La Niña
–2
1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000
Figure 18.16
Historical record of intensity of El Niño (above the midline of the graph) and La
Niña events (below the midline) from 1950 to 2001, based on a standardized multi-
variate index that includes ocean surface temperatures, surface air temperatures,
wind characteristics, and other variables. (After NOAA-CIRES Climate Diagnostic
Center, http://www.cdc.noaa.gov)
370 Chapter 18
currents and winds. Ocean waters are always exchang- coast died or abandoned their nests; the same fate befell
ing energy (as heat) with the atmosphere, although they almost all of the 17 million seabirds living on tiny Christ-
hold far more heat, and move heat around the globe mas Island in the Pacific.
more slowly, than do atmospheric currents. This The El Niño of 1997–1998 was even more extreme,
exchange of heat energy with the ocean drives the with the resulting drought exacerbating forest fires in
movement of winds, and the winds drive ocean currents. Indonesia and Malaysia and harming the already threat-
How do these complex changes in atmospheric and ened rainforests there; the rainforests of the Amazon
oceanic interactions affect weather and organisms in dif- basin also suffered from extreme drought. In many
ferent places? In North America, El Niño events may be places, the unusual severity of this particular El Niño,
linked to milder winters along the U.S.-Canada border, coupled with other problems facing these tropical forests
increased winter storms along the California coast, (such as logging and fragmentation), combined to cause
floods in the southeastern United States, increased snow- much greater damage than any of these factors would
fall in the southwestern mountains, and greatly have by themselves (Wuethrich 2000). Other areas affect-
decreased hurricane activity in the Atlantic. La Niña also ed by the drought included Australia and southern
can have strong effects on weather patterns in North Africa, while severe flooding occurred in southern Cal-
America, including increases in reported tornado ifornia and in the coastal deserts of northern Chile and
strength and frequency in the Midwest, stronger and Ecuador.
more frequent hurricanes in the Atlantic, and drought These cycles also affect plant population dynam-
and forest fires in the Southwest. ics. Because the U.S. Southwest normally has low
Globally, the El Niño of 1982–1983 caused droughts amounts of precipitation, this region provides a sensi-
in Africa, Australia, and South America. Coral reefs off tive context for studying these effects. The ENSO effects
Costa Rica, Panama, Colombia, and the Galápagos both precipitation and temperature, as can be seen by
Islands suffered drastic losses (50%–97% die-offs) looking at the climate record for the past century at Las
because of the warmer water and loss of nutrients. In Cruces, New Mexico (Figure 18.17). Periods of dry win-
that same event, 85% of the seabirds on the Peruvian ters combined with wet summers in this area are asso-
(A) (B)
er
er
r
r
g
g
m
m
te
te
rin
rin
m
m
in
in
ll
ll
Su
Su
Sp
Sp
Fa
Fa
W
1980 1980
1970 1970
Cooling trend
1960
1950
1940 1940 Figure 18.17

Patterns of temperature and precipitation for
the past century at Las Cruces, New Mexico.
1930
(A) Contour map of average monthly tempera-
Warming trend
ture (contour interval 5°C). Colder winters are

1920 indicated by larger areas within the 5° contour,
such as the period from 1928 to 1934. Warmer
summers are indicated by larger areas within
1910 the 25° contour, such as the period from 1927 to
1937. Arrows followed by solid lines indicate
years with large numbers of seedlings of
1900 1900 Bouteloua eriopoda (black grama grass, Poaceae).
(B) Contour map of average monthly precipita-
10 5 5 10 20 25 20 10 tion (contour interval 10 mm). Wet periods are
1890 1890 indicated by the dark triangular-shaped areas,
O N D J F M A M J J A S O N D O N D J F M A M J J A S O N D such as the period from 1965 to 1975. (After
Month Month Neilson 1986.)
106
3
Standardized ENSO index

ENSO index
Area burned (ha)
105 2
1
104 0
–1
103
Area burned –2
102 –3
1910 1920 1930 1940 1950 1960 1970 1980
Year
Figure 18.18
Annual area burned in Arizona and New Mexico (black line) and an index of the
intensity of the El Niño Southern Oscillation (ENSO) (green line) from 1905 to 1985.
(After Swetnam and Betancourt 1990.)
ciated with range increases of the native Bouteloua eri- of the Rocky Mountains during the Eocene 34 million
opoda (black grama grass, Poaceae; Neilson 1986). Peri- years ago. As continents drift across the face of the Earth,
ods with dry spring weather are associated with large- their climates change. For the past 100 million years,
scale forest fires (Figure 18.18). Australia has been moving northward; as a result, its cli-
Short-term oscillations such as ENSO are embedded mate has changed from mainly temperate and polar to
in larger, long-term phenomena. Over the past 5000 mainly tropical and subtropical. Changes in atmospheric
years there may have been periods of more frequent El CO2 concentrations, both natural and anthropogenic,
Niño events from 4800 to 3600 years before present (B.P.), also have had large effects on global climate in the past
around 1000 B.P., and after 500 B.P. These periods also as well as currently. We discuss these long-term patterns
appear to be associated with increased flooding in the in detail in Chapters 21 and 22.
southwestern United States (Ely et al. 1993).
Predictability and Long-Term Change

In this chapter we have described patterns of variation
0.3
in temperature and precipitation. These patterns are gen-
erally regular, with some predictability. Random effects Creosote
act to introduce unpredictability into these patterns. The
Among-year annual variablility
degree of unpredictability is one determinant of the

types of plant communities found at different locations 0.2
(Figure 18.19). Uncertainty is also an important deter- Desert scrub
minant of some kinds of adaptations. For example, pre-
dictable variation is more likely to favor adaptation by Grassland
phenotypic plasticity (see Chapter 5), while unpre- Pine forest
dictable variation is more likely to favor adaptation by 0.1
a jack-of-all-trades strategy.
The distinction between random and predictable pat-
terns of change depends on the life span of the plant rel-
ative to the rate at which the climate changes. The ENSO,
for example, varies enough in its details (e.g., intensity 0 250 500 750 1000
and length) that for an annual plant it is effectively ran- Annual mean precipitation (mm)
dom, while for a Sequoia sempervirens (redwood, Cupres- Figure 18.19
saceae) that lives for centuries, it forms a predictable Association between the variation and predictability of rain-
cyclic pattern. fall and the presence of different types of plant communities
At the longest time scale—tens of millions of years— in Arizona. Desert areas dominated by creosote bush (lower
continental drift and tectonic activity play a major role Sonoran Desert) or desert scrub (upper Sonoran Desert)
have low amounts of precipitation and high year-to-year
in changing climate. Mountain building, for example, variation. As precipitation amounts increase and variation
creates rain shadows. The spread of grasslands across decreases, plant communities become dominated by grasses
the middle of North America is associated with the rise and then pine forest. (After Davidowitz 2002.)
372 Chapter 18
and subtropical forests can be dominated by either

Plant Physiognomy across the Globe
deciduous or evergreen angiosperms.
A continent is a complex mosaic of climates and vege- The critical climatic factor determining which one
tation forms. Across a continent, this variation falls into occurs in a given area is seasonality of precipitation.
regular patterns. Here we introduce several of these Areas that receive rain throughout the year are ever-
broad patterns. The first of these is a north-south gra- green, while areas with pronounced dry periods are
dient in vegetation form due to temperature. A second deciduous. As always, there are exceptions to this gen-
major pattern is variation from west to east in response eral pattern. For example, many forests in the south-
to changes in average precipitation. Such patterns exist eastern United States are dominated by a particular
on all continents because of the latitudinal differences in genus of needle-leaved evergreen gymnosperms: Pinus
solar radiation and the movement of weather systems (pines). These pine forests are typically found on soils
described earlier in this chapter. While we concentrate that are very low in nutrients and are subject to fre-
here on North America, and while each continent is quent fires. (The effects of soil nutrients on leaf life span
unique, the principles behind these patterns can be are considered in the context of plant strategies in
applied to the other continents as well. This variation Chapter 9.)
will be addressed in greater detail when we describe the Moving north, the transition from broad-leaved
world’s biomes in the next chapter. evergreen to broad-leaved deciduous forests depends
on the likelihood of freezing temperatures. Broad-leaved
Forests evergreen trees are vulnerable to frost through two of its
Forests are communities dominated by trees whose effects. First, leaf tissues will die if kept frozen. Second,
leaves touch each other, resulting in a closed canopy. If the weight of ice or snow on leaves will cause them to
you were to travel southward in eastern North America be torn off or, worse, will break branches and even tree
from northern Canada to the Carolinas, Georgia, and trunks. A very unusual October snowstorm in 1997 in
northern Florida, you would pass through a variety of the Midwest and Great Plains of the United States,
different-looking forests. Forest composition is affected which occurred before the leaves had fallen, caused
by both climate and soil; to focus on the effects of cli- major damage to deciduous trees as branches and even
mate, we confine our discussion to the relatively rich trunks snapped under the weight of the snow-burdened
soils found inland (away from the nutrient-poor coastal canopies. Tall, narrow conifers such as spruces (Picea)
plain). and firs (Abies) shed snow far more effectively. So it is no
Immediately to the south of the tree line you would surprise, then, that the boundary of broad-leaved ever-
find taiga or boreal forest, which is dominated by conif- green trees corresponds with that of freezing tempera-
erous, evergreen needle-leaved trees. Next you would tures (Figure 18.20).
enter a region dominated by broad-leaved trees—decid- Farther north, the transition between broad-leaved
uous angiosperms that drop their leaves in winter— deciduous and needle-leaved evergreen forests is relat-
although conifers would still be present. Finally, you ed to growing season length. An evergreen tree can take
would come upon broad-leaved evergreen forests dom- advantage of short periods of favorable conditions as
inated by angiosperms that retain their leaves all year. long as there is enough unfrozen moisture in the soil to
In summary, there is a transition from evergreen forests support transpiration (a major limitation for these trees).
to deciduous forests and back to evergreen forests, Conifers are able to avoid leaf damage due to freezing,
although the type of evergreen tree changes from nee- by increasing the osmotic concentration of intracellular
dle-leaved to broad-leaved, and from gymnosperms to water (thereby lowering its freezing point). Ice crystals
angiosperms. These changes in tree type with latitude form only between the cells (which does little damage
are largely driven by differences in seasonal tempera- to membranes); this acts to draw water out of cells,
tures. Of course, the description of this imaginary trip is which further increases the osmotic concentration inside
very general; for the moment, we are ignoring many the cells.
other details to concentrate on temperature effects. Sim- In contrast, a deciduous tree requires time to pro-
ilar patterns are found, for example, in eastern Asia as duce new leaves in the spring before it can begin pho-
one travels from Korea and northeastern Russia to the tosynthesizing. When the new leaves are emerging, they
south of China and beyond to Southeast Asia. are particularly vulnerable to injury from freezing. Thus,
To understand why these changes occur, let’s re- leaf emergence does not occur until the chance of a late
verse our trip. Beginning in southeastern North Ameri- frost is very small. Following leaf emergence, the grow-
ca, growing seasons are longest in the broad-leaved ever- ing season must be long enough for the tree to accu-
green forests found along the Atlantic and Gulf coasts. mulate enough carbon for its maintenance, growth, and
These areas are classified as subtropical because they reproduction. Thus, deciduous trees are able to survive
rarely experience temperatures below freezing. Tropical only in regions with a relatively long growing season.
Arctic
Circle
Tropic of
Cancer
Equator
Tropic of
Capricorn
Antarctic
circle
Frost-free
Episodic or periodic
frost to –10ºC
Frost to –20ºC possible Figure 18.20
Winter minima lower
The occurrence of subfreezing temperatures across the globe. Frost-free areas and areas
than –20ºC possible with only episodic frost are generally tropical or subtropical. Taiga, dominated by nee-
dle-leaved evergreen trees, and tundra are found in areas where fewer than three
Fewer than 3 months
frost-free
months are frost-free. The other areas with frost or winter are generally temperate.
(After Larcher and Bauer 1981.)
The slow leafing out of deciduous trees in the spring below 0°C, plant tissues can supercool, meaning that
has another effect on the form of deciduous forests: they can cool without ice formation. It is the formation
These communities are much more complex in form and of ice crystals in plant and animal tissues that is the pri-
higher in species diversity than boreal forests. Because mary source of damage and injury from freezing.
the canopy is deciduous, there are times of year when
large amounts of light penetrate to the forest floor. Con- Tree Line
sequently, there can be up to four layers of vegetation in Temperature has a direct effect on the types of vegeta-
deciduous forests: herbaceous species along the ground, tion that can grow in a region. One of the most strik-
small shrubs, small trees and large shrubs, and canopy ing examples of this effect is the alpine tree line, an
trees. In the spring, the forests turn green from the often abrupt boundary where trees are replaced by low-
ground up, with each layer leafing out slightly later. This growing vegetation (Figure 18.21). Tree line is most
ground-up effect occurs because the air is warmer near obvious in high mountain ranges, such as the Sierra
the ground, especially early in the season before the Nevada and Rockies in North America, the Alps in
canopy leafs out and reduces convective cooling. The Europe, the Andes in South America, or the Himalayas
diversity of species and form in deciduous forests, there- in Asia, all of which have forests that stop below the
fore, depends on seasonal variation in temperature. summit. If you were to hike up such mountains, you
The evergreen needle-leaved trees that dominate the would reach a point where the forest would thin and
northern coniferous forests pay a cost for their ability to the trees would get shorter. At some point, the trees
withstand freezing temperatures: they grow much more would disappear entirely. A similar sight would greet
slowly than broad-leaved deciduous trees. Once the you at the Arctic or Antarctic tree line on a hike toward
growing season is sufficiently long, deciduous trees out- the North Pole in North America or Eurasia, or toward
compete needle-leaved evergreen trees. The border the southern tip of South America.
between the deciduous forest and the needle-leaved Tree lines are caused by a complex of several factors,
evergreen forest corresponds to a temperature line north including the limits to supercooling mentioned above
of which the minimum temperature during the winter and others detailed in Chapter 19. In addition, wood will
falls below –40°C (Arris and Eagleson 1989; compare form only when temperatures are above 10°C. Microen-
Figures 18.20 and 19.1). Above this temperature and vironmental differences in temperature can lead to inter-
374 Chapter 18
By the time you have reached Iowa, you are in tall-

grass prairie. In addition to taller vegetation, these com-
munities have greater biomass; they are also more pro-
ductive and more diverse. Besides the obvious grasses,
there are many dicot forbs whose stature matches that
of the grasses. This shift in plant stature and diversity
closely follows a pattern of increasing rainfall.
As you continue eastward, more and more trees are
evident. In the areas dominated by grasslands, trees are
clustered along streams. Farther east, they begin to dot
the countryside, becoming more dense. These areas are
woodlands, dominated by trees, but without a closed
canopy. Eventually, at about the Illinois/Indiana border,
you reach forests.
This transition from grassland to woodland to for-
est is a function of changes in both the amount and sea-
sonality of precipitation. Annual rainfall, for example, is
about 80 cm both in the middle of Kansas and to the
northeast in the middle of Minnesota, yet the former is
Figure 18.21 in the center of the tallgrass prairie, while the latter is
Tree line at about 1200 m in Rocky Mountain National Park, forested. The difference is that in Kansas, the rainfall is
Colorado. (Photograph by S. Scheiner.)
much more seasonal, with more of it occurring in spring
and fall, the temperature is warmer, and the chance of
drought is greater.
esting effects, such as the formation of krummholz (Fig- This seasonality of precipitation and propensity for
ure 18.22), a peculiar tree form found at tree line on drought leads to another key factor: fire (see Chapter 13).
mountains in different parts of the world. Under natural circumstances, prairies experience fires
Above tree line on Mount Washington, New Hamp-
shire—the highest peak in eastern North America—you
might notice an odd-looking plant. It looks like a hori-
zontal woody vine growing along the ground (Figure
18.23). If you examine it closely, however, you will find
that it is Picea mariana (black spruce, Pinaceae), the dom-
inant tree species in the nearby forest. Seeds taken from
these plants, if planted farther down the mountain, will
grow into tall trees! The krummholz form of the plant
comes about because the temperature right at ground
level is just a little bit warmer than the air above it—just
enough above 10°C for wood formation (Teeri 1969). The
warmer temperatures near the ground are caused by
reduced convective cooling and longwave radiation by
the ground of the energy it absorbs from sunlight.
Grasslands and Woodlands

Traveling from west to east, starting in the rain shadow
on the eastern face of the Rocky Mountains and tra-
versing the center of the United States, you will see
another gradual transition in the physiognomy of the
vegetation. Your journey begins in short-grass prairie,
in the midst of low-growing clumps of grasses, with bare
patches interspersed between the clumps. As you trav-
el eastward, the vegetation becomes taller and thicker as Figure 18.22
you reach the midgrass prairie. Here you may notice Krummholz vegetation at treeline in Rocky Mountain
fewer grasses growing in clumps and more rhizomatous National Park, Colorado. Note the twisted trunk of the tree.
grasses, as well as many dicots. (Photograph by S. Scheiner.)
Figure 18.23
Picea mariana (black spruce, Pinaceae) growing in a “vine” form above tree line on
Mount Washington, New Hampshire, the highest peak in eastern North America.
(Photograph courtesy of J. Teeri.)
every 3–5 years, depending on climatic and landform depends on disturbance. Fire often favors shrubs, which
conditions. Each year dry grasses and other plant mate- can resprout readily after being burned, over trees.
rial accumulate until there is enough fuel to sustain a Annuals are also very common in shrublands and are
conflagration, and lightning (or human activity) ignites especially prevalent in the years immediately following
a fire. Prairie plants are adapted to these frequent fires. a fire. Other disturbances can also favor shrubs over
Their vulnerable meristems are located at or below trees. Today grazing by domesticated animals is the most
ground level. A fast-moving fire will burn off the tops of important factor spreading and maintaining shrublands
the plants, but they quickly resprout. Trees, however, in subtropical regions of the world.
have their meristems at the tips of their branches. The other major climatic regions with extensive
Seedlings and young saplings are particularly vulnera- shrublands are deserts, which are regions where poten-
ble to fire. Grasslands begin to give way to woodlands tial evapotranspiration exceeds actual evapotranspira-
where the frequency of fire drops low enough that some tion. The amount of precipitation and its seasonality are
trees have sufficient time between fires to grow tall among the most important factors in determining what
enough to become resistant to fire. Even so, these wood- kinds of plants thrive in very dry environments (Fig-
lands are dominated by species that have thick bark and ure 18.24). Summer and winter precipitation represent
are tolerant of fire. The woodlands become closed- very different sources of water for plants. At summer
canopy forests where precipitation levels are high temperatures, rain only wets the soil surface, often for
enough that fire frequency becomes low. only a few days. At winter temperatures, precipitation
can infiltrate deeper soil layers and remain available
Shrublands and Deserts there throughout the year.
Shrublands are another widespread physiognomic form Wet winters with dry summers favor deep-rooted,
dominated by low-growing woody plants with multi- long-lived plants with large leaf areas, such as trees.
ple stems. The locations of these communities are deter- These plants use almost no summer rainwater, even
mined by a combination of temperature and precipita- when it is available. As the amount of winter water
tion effects. They are primarily found in dry to very dry declines, plants with greater shoot biomass relative to
regions. Areas with mediterranean climates, for exam- the amount of root biomass, such as shrubs, are favored.
ple, contain a mixture of shrublands and woodlands. As winter water further decreases, summer water
Which type of vegetation predominates in an area becomes increasingly important, favoring plants with
376 Chapter 18
shallower root systems and even less root material, such many shrubs, the Mojave Desert is dominated by low-
as herbaceous annuals and perennials. These plants are growing shrubs (many of which are deciduous) and has
capable of extracting surface water quickly, before it is a very large number of annual plant species, the Chi-
lost to evaporation. huahuan Desert is dominated by evergreen shrubs and
When there is no winter water stored deep in the grasses, and the Sonoran Desert has a mixture of many
soil, only succulent plants such as cacti, which store growth forms. In all of these deserts, however, several
water internally and lose it slowly, can survive between growth forms coexist; no one growth form can preempt
summer rain events. The more often it rains in sum- all of the available water.
mer, the less need to store water, and the community
becomes increasingly more herbaceous. If wet summers Summary
are accompanied by moderately dry winters, herba-
ceous, shallow-rooted plants, such as grasses or summer The climate of an area is determined by the mean, vari-
annuals, are favored. ability, and seasonality of temperature and precipitation.
These effects of the amount and seasonality of pre- Mean temperature is determined by the amount of
cipitation in deserts explain the differences in the phys- incoming solar radiation, which varies daily, seasonally,
iognomy of the hot deserts of western North America: yearly, and over the course of centuries. At the equator
the Mojave, the Sonoran, and the Chihuahuan. The solar radiation inputs are generally high and even
Mojave Desert (mainly in California) gets almost exclu- throughout the year; in contrast, solar radiation inputs at
sively winter precipitation. The Chihuahuan Desert to high latitudes vary greatly over the year. At much longer
the southeast (mainly in the Mexican states of Chi- time scales—tens to hundreds of thousands of years—
huahua and Coahuila) gets almost exclusively summer changes in Earth’s orbit around the sun and in the tilt of
precipitation, whereas the Sonoran Desert between the Earth’s axis also create variation in solar radiation.
two in Arizona and Sonora gets both summer and win- Patterns of precipitation are caused by differential
ter precipitation. As a result, while all three regions have heating and cooling of the atmosphere at different lati-
Succulents High
(e.g., cacti)
Shrubs
Summer precipitation
Herbaceous
perennials
Herbaceous
annuals
Trees
Low High
Winter precipitation
Figure 18.24
Plant types favored by various combinations of winter and summer precipitation.
(After Schwinning and Ehleringer 2001.)
tudes. At the equator, warm rising air creates a region of It is the combination of the mean climate of a region,
high precipitation. That air tends to descend at 30° north the variability of that climate, and the predictability of
and south of the equator, creating regions of low pre- that variation that determines the types of plants that
cipitation there. A second area of low precipitation is will grow there. Trees can grow only where temperatures
found near the poles. In addition, the prevailing winds exceed 10°C for sufficient periods of time. Warm tem-
push masses of water, creating ocean currents that carry peratures and long growing seasons favor broad-leaved
heat energy to and away from the areas they traverse. deciduous trees over needle-leaved evergreen trees.
A second pattern is caused by the flow of air across Trees require high amounts of rainfall, however. Areas
continents and its interaction with mountain ranges. of lower rainfall are dominated by shrubs or grasses.
Areas on the upwind (western) sides of mountain ranges Succulents are favored in areas with unpredictable sum-
tend to be wet, while the downwind (eastern) sides of mer rainfall. The border between woodlands, shrub-
mountain ranges tend to be dry. The north-south–run- lands, and grasslands is often controlled by rates of dis-
ning mountain ranges of North America create a distinct turbance, especially the frequency of fire. Fire frequency,
pattern of alternating wet and dry areas across the con- in turn, is controlled by patterns of temperature and pre-
tinent. Variation in precipitation occurs on time scales of cipitation. Thus, an understanding of climatic patterns
days, seasons, years, and decades. This variation also across the globe provides the basis for understanding
contains an element of unpredictability that is especial- patterns of vegetation. We explore those patterns in more
ly important in extreme climates. detail in the next chapter.
Additional Readings
Classic References Neilson, R. P. 1986. High-resolution climatic analysis and southwest
biogeography. Science 232: 27–34.
Beard, J. S. 1955. The classification of tropical American vegetation-types.
Ecology 36: 89–100. Schwinning, S. and J. R. Ehleringer. 2001. Water-use tradeoffs and opti-
mal adaptations to pulse-driven arid ecosystems. J. Ecol. 89: 464–480.
Gates, D. 1962. Energy Exchange in the Biosphere. Harper & Row, New
York.
Holdridge, L. R. 1947. Determination of world plant formations from Additional Resources
simple climatic data. Science 105: 367–368. Heinrich, W. 1973. Vegetation of the Earth in Relation to Climate and the
Mather, L. R. and G. A. Yoshioka. 1968. The role of climate in the distri- Eco-Physiological Conditions. Springer-Verlag, New York.
bution of vegetation. Ann. Assoc. Am. Geog. 58: 29–41. Rumney, G. R. 1968. Climatology and the World’s Climates. Macmillan,
New York.
Contemporary References
Ji, J. 1995. A climate-vegetation interaction model: simulating physical
and biological processes at the surface. J. Biogeog. 22: 445–451.
19
C H A P T E R
Biomes
O
ur planet has a complex pattern of climates, which in turn has a major
role in creating the complex pattern of vegetation and community types
we find on Earth. In the previous chapter, we looked at climate and
how it affects plant physiognomy. Here we look at the resulting vegetation pat-
terns in more detail. Ecologists divide these large-scale patterns into units called
biomes: major biogeographic regions that differ from one another in the struc-
ture of their vegetation and in their dominant plant species (Clements 1916).
Biomes represent the largest scale at which ecologists classify vegetation. In
this chapter, we examine these broad-scale patterns, look briefly at the world’s
major biomes, and briefly consider some of the ways in which human activi-
ties affect them.
Categorizing Vegetation
Classification of the world’s vegetation into biomes (Figure 19.1; Table 19.1)
provides useful categories that describe major aspects of vegetation, such as
its structure, function, and adaptations. Not only do biomes tell us a great deal
about what kinds of plants might be found growing in a given area, they also
indicate something about the types of animals and other organisms likely to
be living there, and about the major environmental constraints on living things.
Similar climates in different parts of the world contain similar biomes (Figure
19.2), although their details differ. For example, temperate evergreen forests
are similar in being dominated by trees that stay green all year, but differ in
being dominated by needle-leaved conifers in North America, by broad-leaved
beeches in South America, and by broad-leaved eucalypts in Australia.
Biomes are defined by the physiognomy of the dominant or most obvious
plants. Thus, we recognize different types of forests (e.g., evergreen or decid-
uous, broad-leaved or needle-leaved), shrublands, and grasslands. However,
we must not lose sight of the trees for the forest: there may be a tremendous
amount of variation within a given biome. In her study of communities with-
in a single biome, the temperate deciduous forest of eastern North America, E.
Lucy Braun recognized 12 different major sub-biomes and a number of minor
variants (Braun 1950). Within a biome, there may be patches of vegetation that
appear “not to belong,” such as riparian forests found along streams in grass-
lands or deserts. The occurrence of these patches reminds us that climate is
only one factor in determining vegetation. Local variation in soils and topog-
380 Chapter 19
30ºN
Equator
30ºS
Tropical rainforest High mountains (taiga and alpine tundra) Figure 19.1
Major biomes of the world.
Tropical deciduous forest Temperate evergreen forest
Thorn forest Temperate deciduous forest
Tropical savanna Taiga
Hot desert Arctic tundra
Temperate shrubland Temperate grassland
Cold desert Ice
raphy, especially as the latter fit easily into any given biome. Furthermore, different
affects microclimate, can influ- scientists might include a place in different biomes. As
ence the type of vegetation found the definitions of biomes are somewhat arbitrary, they
in an area. In turn, the animals also differ in detail among scientists. Some use fewer,
living in a given biome are deter- broader groupings, whereas others use a more detailed
mined by the vegetation there, as breakdown of vegetation types. The definitions of bio-
well as by climate and other fac- mes given here may not exactly match those given in
tors. other texts. However, there is a broad consensus on the
Biomes are useful, if some- rough categories and approximate definitions of biomes.
what arbitrary, descriptive classi- Our use of “temperate rainforest,” for example, may
E. Lucy Braun fications, rather than quantitative not be identical to someone else’s use of that term, but
or objective categories. Bound- it will be close. If you were to search the scientific liter-
aries between biomes may represent the range limits of ature, you would find many different descriptions of
the dominant species, whereas other species may have places called rainforests. There is, however, no single,
ranges that span those boundaries. While we draw sharp rigorous definition of the term—it is, in fact, used to refer
boundaries on a map, in reality the boundaries of bio- to forests occurring in places where rain is abundant and
mes are often fuzzy. Similarly, the categories that we there is no extended dry season. Tropical rainforests
define ignore the fact that a particular place might not occur where there is substantial year-round rain; tem-
Table 19.1 Major biomes of the world: dominant growth forms and general climatic conditionsa
Angiosperms or
gymnosperms
Biome Dominant growth form dominant or common Temperature Moisture
Tropical rainforest Broad-leaved evergreen trees Angiosperms Hot Wet
Tropical montane forest Broad-leaved evergreen trees Angiosperms Mild Wet
Tropical deciduous forest Broad-leaved deciduous and Angiosperms Hot Seasonally dry
semi-evergreen trees
Thorn forest Broad-leaved deciduous trees Angiosperms Hot Dry
Tropical woodland Broad-leaved semi-evergreen Angiosperms Hot Moderate
trees and grasses
Temperate deciduous forest Broad-leaved deciduous trees Angiosperms Seasonally cold Moderate
Temperate rainforest Needle-leaved evergreen trees Gymnosperms Seasonally cool Wet
Temperate evergreen forest Needle-leaved or broad-leaved Gymnosperms or Various Various
evergreen trees angiosperms
Temperate woodland Needle-leaved evergreen or Both Mild Moderate
broad-leaved deciduous trees
and grasses
Taiga Needle-leaved evergreen trees Gymnosperms Cold Moderate
Temperate shrubland Evergreen shrubs, annual forbs Angiosperms Mild Moderate
Temperate grassland Perennial grasses Angiosperms Seasonal Moderate
Tropical savanna Perennial grasses Angiosperms Hot Moderate
Hot desert Shrubs, succulents, annual and Angiosperms Hot Dry
perennial grasses, annual forbs
Cold desert Shrubs Angiosperms Mild Dry
Alpine shrubland Deciduous shrubs Angiosperms Cold Moderate
Alpine grassland Perennial grasses Angiosperms Cold Moderate
Tundra Perennial grasses, sedges, Angiosperms Cold Moderate
shrubs, forbs
a
Other factors, especially seasonality, are also important.
—15
—10 Tundra
—5 Alpine shrubland
Alpine
Cold temperate
grassland Cold temperate
Mean annual temperature (…C)
0 Tropical
Taiga montane forest
Te odl
Cold desert
wo
mp an Temperate
5 Temperate
Temperate
era d
deciduous Temperate
evergreen
te
forest rainforest
forest
shrubland
10
Temperate
15 Temperate evergreen
Figure 19.2 grassland forest Warm temperate
The distribution of biomes is deter- Tropical
mined by climate, especially annual 20 savanna
temperature and precipitation. In Tropical
woodlan
regions within the dashed lines, other Tropical Tropical

Tropical
Thorn deciduous rainforest

factors—such as fire, grazing, and sea- 25 forest forest
sonality of precipitation—strongly affect
Hot
d
which biome is present. Climate can desert

also interact with factors such as soil 30
type to determine biome distributions. 0 50 100 150 200 250 300 350 400 450
(After Whittaker 1975.) Mean annual precipitation (cm)
382 Chapter 19
Table 19.2 Primary productivity of selected major biome types
Net primary
Area productivity per Global net primary Biomass per Total global
(× 106 km2) unit area (g/m2/yr) productivity (1012 g/yr) unit area (kg/m) biomass (1012 g)
% of total % of total % of total
terrestrial Normal terrestrial Normal terrestrial
Biome Amount area range Mean Amount productivity range Mean Amount biomass
Tropical rainforest 17.0 11.4 1000–3500 2200 37.4 32.5 6–80 45 765 41.7
Tropical 7.5 5.0 1000–2500 1600 12.0 10.4 6–60 35 260 14.2
deciduous forest
Temperate 7.0 4.7 600–2500 1200 8.4 7.3 6–60 30 210 11.4
deciduous forest
Temperate 5.0 3.4 600–2500 1300 6.5 5.6 6–200 35 175 9.5
evergreen forest
Woodland and 8.5 5.7 250–1200 700 6.0 5.2 2–20 6 50 2.7
shrubland
Taiga 12.0 8.1 400–2000 800 9.6 8.3 6–40 20 240 13.1
Temperate 9.0 6.0 200–1500 600 5.4 4.7 0.2–5 1.6 14 0.8
grassland
Tropical savanna 15.0 10.1 200–2000 900 13.5 11.7 0.2–15 4 60 3.3
Desert 18.0 12.1 10–250 90 1.6 1.4 0.1–4 0.7 13 0.7
Tundra 8.0 5.4 10–400 140 1.1 1.0 0.1–3 0.6 5 0.3
Extreme desert, 24.0 16.1 0–10 3 0.07 0.1 0–0.2 0.02 0.5 0.03
rock, sand,
and ice
Cultivated land 14.0 9.4 100–3500 650 9.1 7.9 0.4–12 1 14 0.8
Swamp and marsh 2.0 1.3 800–1500 2000 4.0 3.5 3–50 15 30 1.6
Lake and stream 2.0 1.3 100–1500 250 0.5 0.4 0–0.1 0.02 0.05 0.003
Total terrestrial 149.0 773 115 12.3 1837
Total world 510.0 333 170 3.6 1841
Source: Whittaker and Likens 1975.
perate rainforests tend to occur where there is heavy In Chapter 12, we discussed a controversy that cen-
winter rain and lighter summer rain but considerable tered on the contrasting views of Frederic Clements and
summer fog. The term “rainforest” is thus useful as a Henry Gleason about the nature of communities.
broad description, but not as a rigorously defined cate- Clements saw communities as highly predictable enti-
gory or unique entity. ties primarily controlled by large-scale climatic patterns.
While we have emphasized that, scientifically, the In contrast, Gleason saw communities as unpredictable
boundaries between biomes (or communities) are arbi- and variable and subject to the vagaries of dispersal,
trary, this is not necessarily true under the law. In Aus- microclimatic conditions, and individualistic species
tralia, for example, areas designated “temperate rainfor- distributions. One way to resolve this conflict is to rec-
est” are legally afforded a greater measure of protection ognize that each scientist emphasized a different spa-
than other areas. Some eucalypt forests in very rainy tial scale. Clements was focusing on patterns at the level
places exist above an understory that is floristically a of biomes. At that scale, we can see regular patterns and
rainforest and would replace the eucalyptus canopy in draw boundaries on our maps. Gleason was most con-
about 300–500 years in the absence of fire. However, cerned with patterns at local levels. At that scale, the
because these forests are fire-dependent, some states do boundaries blur, and we mostly see local variability.
not legally considered them rainforests. If the legal Aus- Both perspectives are valid, but each tells us different
tralian definition were applied in North America, the red- things about the world. In the following sections, we
wood forests of the Pacific coast would not be classified will look at vegetation patterns from a large-scale per-
as temperate rainforests—a categorization with which spective as we survey the major biomes and see what
some North American scientists have agreed. distinguishes them from one another.
Biomes 383
Moist Tropical Forests of which retain their leaves year-round and others that
lose their leaves seasonally.
Tropical Rainforest Tropical rainforests are characteristically multilay-
Tropical rainforests are among the most diverse and pro- ered, with two, three, or even four layers of understory
ductive biomes on the planet (Figure 19.3). Although trees. Tallest are the emergent trees, which are widely
they make up only 11% of Earth’s land area, they spaced and often more than 40 m in height, with umbrel-
account for over 30% of terrestrial net primary produc- la-shaped canopies that extend above the general canopy
tion (Table 19.2) and contain perhaps half of all living of the forest. Below this is a closed canopy of trees, typ-
species. Tree species diversity can be as high as 300 ically 30–40 m high. Light is readily available at the top
species per 0.1 ha. Animal diversity is also high, espe- of this layer, but greatly reduced below it. Under the
cially that of insects and other invertebrates. closed canopy, there may be another layer of trees less
Tropical rainforests are found within a band extend- than 20 m in height. Under these trees, there is often a
ing from 10°N to 10°S latitude at elevations below 1 km, shrub/sapling layer, and finally a sparse ground layer.
in areas that get more than 250 cm of rainfall each year, The floors of tropical rainforests can be very dark,
with some rain falling in all months. The most extensive with less than 1% of above-canopy light levels. While
rainforests are found in the Amazon basin of South Hollywood movies may portray tropical rainforests as
America, the equatorial regions of western Africa, and jungles with a tangle of vines and understory plants, the
parts of southeastern Asia. As precipitation becomes understory is generally sparsely vegetated with highly
more seasonal, this biome grades into tropical semi-ever- specialized plants capable of tolerating low light levels
green forests dominated by broad-leaved trees, some (see Chapter 2). The interior of a rainforest was com-
(A) (B)
Figure 19.3
(A) The interior of a tropical rainforest in La Selva, Costa
Rica. (Photograph by E. Orians.) (B) Epiphytes are symbiotic
commensals (see Box 4.1) that grow on the surfaces of other
plants and are common in most tropical rainforests. Numer-
ous bromeliads (Bromeliaceae) can be seen on the trunks of
several of the trees in this forest in Puerto Rico. (Photograph
courtesy of J. Thomson.)
384 Chapter 19
pared by early European explorers to a dimly lit, soar- throughout the year. As for the mineral nutrients need-
ing green cathedral with an expansive space reaching to ed for growth, there are large quantities of nutrients in
a high, vaulted roof. “Jungles” are the consequences of these forests, but they are mainly found in the living bio-
disturbed, high-light environments on the margins of mass rather than in the soil. Because decomposition is
the forest, along roads, rivers, and in gaps, or where log- extremely rapid in these environments, nutrients cycle
ging has thinned the canopy. very rapidly. Like plants in many areas where nutri-
Because of this vertical light distribution within the ents are limiting, those in tropical forests are very good
rainforest, much of its productivity is concentrated high at acquiring and retaining nutrients. Particularly impor-
in the canopy. Plant life forms that are not present in tant in this regard are mycorrhizal associations (see
most temperate forests greatly contribute to the diversi- Chapter 4).
ty of tropical forests. These plants include many species Not only is a large proportion of the world’s biodi-
of lianas (woody vines rooted in the ground) and epi- versity located in the Tropics; a large proportion of its
phytes (plants that grow on other plants but are not root- human population is located there as well. The greatest
ed in the ground; see Figure 2.16), including many growth in human populations is also found in the Trop-
species of ferns, orchids, and bromeliads. Lianas and epi- ics, along with poverty and diseases associated with
phytes are common in tropical forests because of the poverty and poor living conditions. All of these factors
very low light levels near the ground. An alternative can drive the rapid obliteration of natural communities
strategy to being a tree is to grow high up on a tree to by people. Although until recently much of the human
gain access to light. population in the Tropics was rural, it is becoming
Most epiphytes and lianas have commensal rela- increasingly urban. Lagos, Nigeria, for example, with
tionships (the host neither benefits nor is harmed) with about 13 million people, is one of the fastest-growing
the trees on which they grow unless the burden of epi- cities in the world. Unfortunately, this urbanization does
phytes and lianas becomes too large, weakening the tree not mean that there is less pressure on natural habitats;
and possibly pulling it down. Some species are not so in many cases, the opposite is true. Economic, popula-
benign, however. Strangler figs (Ficus spp., Moraceae) tion, and political pressures have contributed to the
begin as a number of separate individual plants that destruction of a large proportion of the world’s rain-
send roots downward to the forest floor, develop thick forests (Table 19.3; see Chapter 22). Exploitation comes
trunks and canopies, and gradually envelop and kill the not only from within the tropical countries themselves,
host tree (see Figure 10.7). The separate individual plants but also in great measure from individuals and corpo-
comprising the strangler ultimately merge to become a rations based in the United States, Europe, and Japan.
functional individual tree. Thus, this plant has a com- Today all tropical rainforests are threatened by logging,
bined strategy of beginning life high in the canopy as an and even more by the clearing of land for a range of pur-
epiphyte, becoming rooted in the ground, which allows poses, including subsistence slash-and-burn agriculture
more growth, and eventually becoming a tree and out- and corporate-owned cattle ranching (see Chapter 22).
competing other species for light. Logging and the clearing of land for cattle grazing
Moist tropical forests typically have highly infer- or human habitation can have numerous effects on rain-
tile soils. In tropical rainforests, warmth and abundant forests. Because most of the nutrients are tied up in bio-
moisture contribute to an extraordinary amount of res- mass, logging, burning, and agriculture remove sub-
piration in the soil—not only by roots, but also by many stantial amounts of nutrients from the system. The
other kinds of soil dwellers, particularly decomposing remaining soils are often very nutrient-poor, and they
organisms. When CO2 is dissolved in water, it produces can quickly achieve a cementlike texture or become
carbonic acid (that is why natural rainwater is always severely eroded. These changes can make agriculture
slightly acidic). The huge amount of soil respiration in impossible in a very short time—in other words, the pur-
tropical rainforests releases large amounts of CO2, much poses for which the forest was removed are not sus-
of which goes into solution in the soil. The net effect is
that nutrient ions, which bind to soil (especially clay)
particles less tightly than hydrogen ions bind to soil par-
ticles, tend to enter the soil solution. As a result, they are
Table 19.3 Global tropical rainforest loss
either taken up almost instantaneously by plants or
washed away by the high levels of rainfall (see Chap- Area (× 106 km2)
ter 4). Region Original Remaining % lost
If the soils are so poor, how is it that tropical forests
Central/South America 8.0 5.8 28
are so productive? Factors contributing to their high
Asia/Pacific Islands 4.4 2.2 39
productivity are year-round water availability, warm
Africa 3.6 1.2 66
temperatures, and duration of daylight varying little
Biomes 385
tainable. Removal of rainforest from large enough areas Overall diversity is lower than in rainforests, however,
can drastically affect regional rainfall levels. When the because there are fewer lianas. These forests are found
forest is present, rainfall is quickly taken up and tran- at middle elevations in the mountains of Africa, South
spired back into the atmosphere. When the forest trees America, Central America, and New Guinea.
are no longer there, the rainfall runs off, which leads to
increasing erosion, particularly for forests on slopes. Seasonal Tropical Forests and Woodlands
Even when the destruction is not so dramatic, as in selec-
tively logged forests, non-native plant species can invade When people think about the ecology of tropical regions,
disturbed forests, threatening the regeneration of native they typically think about rainforests, especially when
species. they think about species diversity and conservation.
While habitat destruction is the key factor threaten- However, many tropical regions are considerably drier
ing tropical biodiversity, other human activities, from than rainforests, and many of these forests are also rich
the illicit trade in rare species such as orchids and par- in species. The seasonal tropical forest biome, with its
rots brought into the United States and Europe, to the associated species, is being lost very rapidly. In fact,
hunting of chimpanzees and other primates for the del- these forests are disappearing—or have already disap-
icacy called “bush meat” in African cities, also endanger peared—even more rapidly than tropical rainforests, as
animal and plant species within these forests. These they are often on soils that are much better for farming.
forms of forest destruction are not particularly new; This biome includes a range of types that grade from
wealthy nineteenth-century orchid collectors in England deciduous forest to thorn forest and woodlands.
and Germany destroyed entire tropical forests to assure Several geographic factors determine the locations
that their competitors could not collect orchids from of dry tropical forests and woodlands. First, they tend
them. However, the scale of recent damage dwarfs that to occur relatively close to the Tropics of Cancer and
of the past, and, of course, the destruction is cumulative. Capricorn, rather then near the equator. This pattern is
a result of the shifting throughout the year of the
Tropical Montane Forest intertropical convergence—the latitude at which the
Tropical montane forests—also called elfin forests or trade winds from the Northern and Southern Hemi-
cloud forests at the highest elevations—are higher-ele- spheres tend to converge (see Figure 18.7), causing a
vation neighbors of rainforests (Figure 19.4). They are large amount of rainfall. The intertropical convergence
called cloud forests because they are often swathed in is usually near each of these latitudes only once a year,
clouds, and much of the moisture available to plants whereas it passes over the equator twice a year. As a
comes from deposition of condensation. They are cool- result, rainfall is highly seasonal, whereas closer to the
er than tropical rain forests because they are at higher equator, rainfall is more evenly spread throughout the
elevations. The trees are typically shorter, and there are year (see Figure 18.15). Second, dry forests and wood-
many more epiphytes festooning the branches of trees. lands tend to occur in the rain shadows of mountain
Figure 19.4
A tropical montane forest in
the Santa Elena Cloud Forest
Reserve, Costa Rica. (Photo-
graph © G. Dimijian/Photo
Researchers, Inc.)
386 Chapter 19
Figure 19.5
The top view of a tropical deciduous
forest in Palo Verde National Park,
Costa Rica was taken during the rainy
season; the bottom photograph is the
same view shown in the dry season,
when many of the trees are leafless.
(Photograph by Donald L. Stone.)
ranges (see Figure 18.9) and on the

western slopes of continents. For
example, the eastern (Caribbean)
side of the mountain range in Costa
Rica is tropical rainforest, whereas
the western (Pacific Ocean) side is
dry tropical forest.
Tropical Deciduous Forest

Tropical deciduous forests, also
called tropical seasonal forests,
include a range of community
types, from wholly deciduous to
semi-evergreen (Figure 19.5). The
trees are drought-deciduous, mean-
ing that they lose their leaves dur-
ing the dry season. These commu-
nities are found in tropical areas that
have pronounced wet and dry peri-
ods, and were once especially exten-
sive in India and Southeast Asia, but
also occurred in Central and South
America and elsewhere. Depending
on the latitude, there might be a single wet and a single
23°N
dry period, or two of each (Figure 19.6). Although most
photosynthetic activity and plant growth occurs during Dry season Single rainy Dry season
season
the wet season, the dry season is often when flowering 15
takes place. There may be advantages to flowering in the
dry season for both animal-pollinated and wind-polli- 10
nated plants. The lack of leaves makes the flowers more Short
conspicuous to pollinators, and the pollen and nectar 5
dry
rewards may be one of the few food sources available to season
them. Similarly, wind-dispersed pollen can travel with
0 Equator
fewer obstructions.
Tropical deciduous forests are similar in overall form
g
and structure to rainforests, with a species richness lower 5 Two rainy

seasons
than but approaching that of tropical rainforests in some
10
Figure 19.6 15
The length and number of rainy seasons depends on lati- Long dry
tude. Near the equator there are two wet and dry seasons season
each year, while farther north and south there is a single
23°S
rainy season in the summer and a dry season in the winter.
The timing of the seasons is reversed north and south of the J F M A M J J A S O N D
equator. Month
Biomes 387
Figure 19.7
A thorn forest on the island of Mada-
gascar, off the east coast of Africa.
This photo was taken after heavy
rains in December. From August
through November—the dry sea-
son—the trees are parched and
brown. (Photograph by N. Garbutt/
Indri Images.)
Central America, and in Asia

(Burma, India, and Thailand), con-
tinental Africa, and Madagascar.
They have very pronounced wet
and dry seasons, and the total year-
ly rainfall may be half that of trop-
ical deciduous forests. Species of
Acacia and other woody members
of this part of the pea family
(Fabaceae) often dominate these
communities. These plants have
leaves that are divided into many
areas. Localities with lower amounts of rainfall gener- small leaflets, which helps to minimize water loss. Diver-
ally have shorter canopies and fewer species. Human sity is considerably lower than in tropical rainforests and
populations are often very high in these biomes, and deciduous forests, but can still be quite high compared
they face many of the same issues and problems as trop- with that in temperate systems. As in tropical seasonal
ical rainforests. forests, the trees are typically drought-deciduous. Suc-
culents are also common, especially in drier areas. As the
Thorn Forest climate becomes even drier, these communities grade
Tropical areas that are even drier support thorn forests into either desert scrublands or tropical savannas.
(also called thornwoods), named for the menacing arma-
ment of many of the plant species there (Figure 19.7). Tropical Woodland
They are common from South America to Mexico, Forests are dominated by trees whose canopies touch
including those on limestone soils in the West Indies and each other. In contrast, the trees in woodlands are spaced
apart, although the density of
those trees can vary from closely
spaced to sparse. Tropical wood-
lands are found in areas with
strongly seasonal climates, like
tropical deciduous forests, but
which are drier or have less favor-
able soils. They occupy areas of
Brazil (Figure 19.8) and interior
southern Africa (miombo), and
also occur in northern South Amer-
ica, the West Indies, northern Aus-
tralia, and Burma and other parts
of Southeast Asia.
Figure 19.8
The vegetation on Brazil’s Sete
Cidades National Park is an example
of tropical woodland. (Photograph ©
J. Jangoux/Photo Researchers, Inc.)
388 Chapter 19
The trees in tropical woodlands range in height from Tilia (basswood or linden), Juglans (walnut), and Liq-
3 to 7 m. Most trees and shrubs have large, tough, semi- uidambar (sweet gum). Different species of these genera
evergreen leaves and thick, fire-adapted bark. Many of occur on each continent. The diversity of local forest
the trees are members of the Fabaceae, as in thorn forests. types on all continents may be very great, depending on
Palms are also common, whereas succulents and spiny the details of topography, climate, soils, and disturbance
plants are rare. African woodlands—as well as savan- regimes. In North America, for instance, the eastern
nas—are rich in large mammal species, including herds deciduous forest consists of many different forest sub-
of elephants, giraffes, wildebeest, and their predators, types (Braun 1950; Greller 1988).
such as lions and hyenas. At higher latitudes and elevations, temperate decid-
uous forest sometimes grades into mixed deciduous-
evergreen forest. An example is the hemlock-white pine-
Temperate Deciduous Forest
northern hardwood forest in the northern United States
Temperate deciduous forests are found in three disjunct (especially around the Great Lakes) and extending south
areas of the Northern Hemisphere: eastern North Amer- in the upper elevations of the Appalachian Mountains.
ica, eastern Asia, and Europe (Figure 19.9). These forests The dominant trees in this forest are a mixture of ever-
grow on young soils formed since the most recent glacia- green needle-leaved and deciduous broad-leaved
tion; the lack of such forests in the Southern Hemisphere species. Elsewhere, patches of pine-dominated commu-
is probably a result of the absence of such soils there. nities can be found where the soil is especially shallow
Temperate deciduous forests are found in areas with or infertile or there is a high frequency of fire. If such
continental or maritime climates with warm, moist sum- areas are large enough, they may be separately classified
mers and cold, sometimes snowy winters, with a range as temperate evergreen forest.
of 50–250 cm rainfall annually, distributed through the Tree species diversity in temperate deciduous forests
year. Diversity and productivity are generally moderate is generally moderate, with 5–30 dominant tree species
in these forests, but span a fairly wide range. In North in any one place. However, the temperate deciduous
America and Asia, this biome is known for the brilliant forests of the southeastern United States are a center of
red, orange, and gold colors of its leaves in autumn, plant diversity, with the highest tree species diversity in
before they are shed for the winter. The dominant trees North America as well as a very high diversity of ferns
are 18–30 m tall. Growing seasons range from very short and deciduous and shrubby angiosperms. For example,
in the northernmost and high-altitude parts of this there are over 120 species of trees and over 3000 plant
biome to very long in its southern portions. The genera species in the southern Appalachian region. These
of the dominant trees common to temperate deciduous forests escaped glaciation, and some areas have been
forests on all three continents include Quercus (oak), Acer continuously vegetated for as long as 200 million years.
(maple), Fagus (beech), Castanea (chestnut), Ulmus (elm), The dominant genus here is Quercus (oak, Fagaceae).
Other common canopy trees in-
clude numerous species of Car-
ya (hickory, Juglandaceae), Acer
(maple, Sapindaceae), Aesculus
(buckeye or horse chestnut, Sapin-
daceae), and Magnolia (Magno-
liaceae), as well as Tilia americana
(basswood, Tiliaceae), Liriodendron
tulipifera (tulip poplar, Magnoli-
aceae), and Betula lenta (black birch,
Betulaceae).
Common in the understory
are trees such as Cornus florida
Figure 19.9
A temperate deciduous forest in
Michigan, here dominated by Amer-
ican beech (Fagus grandifolia,
Fagaceae) and sugar maple (Acer
Saccharum, Sapindaceae). (Photo-
graph by S. Scheiner.)
Biomes 389
Figure 19.10
An old-growth temperate rainforest
in Olympic National Park, Washing-
ton state. (Photograph © G. Ranalli/
Photo Researchers, Inc.)
Other Temperate Forests

and Woodlands
Temperate Rainforest
Substantial temperate rainforests
are located in southern Chile and
along the northwest coast of
North America (Figure 19.10).
Smaller ones occur on the western
coast of New Zealand, the south-
eastern and southwestern coasts
of mainland Australian, and the
island of Tasmania. Fragments
remain in Norway; in other parts
of Europe, such as Ireland, these
forests were completely destroyed
several centuries ago. These
(flowering dogwood, Cornaceae) and large shrubs such regions are all similar climatically: under maritime influ-
as Kalmia latifolia (mountain laurel, Ericaceae) and Rhodo- ences, temperatures stay cool, and there is abundant
dendron spp. (evergreen rhododendrons and deciduous winter rainfall and summer fog. Productivity and
azaleas, Ericaceae). In wetter areas such as coves or decomposition are extremely high in some of these
stream banks, Tsuga canadensis (eastern hemlock, forests, due largely to the mild temperatures and extend-
Pinaceae) is found, while pines are common on drier ed rainy period, which result in a very long growing sea-
ridges. The formerly dominant Castanea dentata (chest- son. In contrast to tropical rainforests, however, species
nut, Fagaceae; see Figure 11.18) was devastated by chest- diversity here can be quite low. In North America, these
nut blight during the early part of the twentieth centu- forests are dominated by needle-leaved conifers. In con-
ry and has largely been replaced by various oak species trast, temperate rainforests are dominated by broad-
in its former habitats. leaved trees, such as Nothofagus (Fagaceae), in Chile,
Deciduous forests are threatened by a variety of fac- New Zealand, and Australia, although conifers such as
tors. Some of the world’s largest cities (such as Lon- those in the family Podocarpaceae also play an impor-
don and New York) occur on land that was once tem- tant role in some Southern Hemisphere forests.
perate deciduous forest, although large areas of this Temperate rainforests contain some of the tallest
biome still exist. In Europe, air pollution, including the trees in the world, including the giant redwoods (Sequoia
effects of acid precipitation and nitrogen deposition, has sempervirens, Cupressaceae) of California and some of
been a major cause of physiological stress in forest trees the eucalypts of Australia, with heights exceeding 100
and general forest decline (see Chapter 22). Worldwide, m. Not all forests of this type contain trees of such enor-
logging has been a major cause of forest destruction for mous stature, however; the coastal Sitka spruce forests
hundreds of years, as has development for farmland or of Alaska have trees similar in height to those of tem-
residential use. In some places the forests have been perate deciduous forests. Some of these forests (includ-
destroyed, while in others, such as the eastern United ing the redwood forests) depend on occasional fires for
States, secondary forests have regenerated over large regeneration.
regions. In some areas of the world (such as parts of Temperate rainforests are threatened more by their
Japan and Mexico), the only forests to escape cutting economic value than by human population growth
occur in places that are inaccessible or very steep. Road because most of them are not near centers of human pop-
building and expanding suburbanization are renewing ulation. Because of their enormous standing crop of
threats to many secondary forests in North America. wood, many of these forests on the western coast of
Invasion by exotic plant species is a threat to this and Canada and the United States have been clear-cut for
many other biomes worldwide as well (see Chapter 22). lumber, resulting in high rates of erosion and losses of
390 Chapter 19
Figure 19.11
This privately owned temperate
rainforest near Port Angeles in
Washington state is being “clear
cut.” All the trees will be felled for
commercial lumber. (Photograph ©
C. Larsen/Photo Researchers, Inc.)
Temperate Evergreen Forest

Temperate evergreen forest is a
catch-all category that includes
forests of very different types. All
of them are located in areas that
have pronounced seasonal cli-
mates and are drier than areas
with deciduous forests, due to
either low rainfall or soils that do
not retain moisture well (Figure
19.12). These forests generally
experience relatively mild temper-
atures but have strong wet/dry
species diversity. Because of the high potential growth cycles (e.g., in the wetter parts of the Mediterranean
rates for trees, some of these areas have been replanted region, southern California, and southern Australia), or
with tree monocultures for lumbering, with the result have greater temperature ranges and drier continental
being low-diversity tree plantations in place of natural climates (e.g., in the mountains of the western United
old-growth forest. The remaining stands of old-growth States). Temperate evergreen forests often grade, at their
forest have been the focus of extended and bitter fights drier margins, into temperate woodlands or shrub-dom-
between environmentalists and logging interests. Old- inated systems such as chaparral. The trees can be broad-
growth temperate rainforests in North America are dis- leaved or needle-leaved, but their leaves are retained
appearing more rapidly than tropical rainforests, almost throughout the year. Trees with sclerophyllous (tough,
entirely as a result of commercial lumbering operations evergreen, relatively small, roundish and thick) leaves
(Figure 19.11). are common, especially in areas with milder tempera-
tures. In cooler regions, needle-
leaved forests often predominate,
including the pine and fir forests
at mid-altitudes in the Sierra
Nevada of California and the
mixed needle-leaved and broad-
leaved forests of conifer and beech
in Chile and New Zealand. The
latter forests, depending on the rel-
ative proportions of needle-leaved
Figure 19.12
In their natural state, temperate
evergreen forests are one of the most
endangered habitats in the world.
Shown here is Hartwick Pines State
Park in central Michigan. The last
remaining stand of old-growth white
pine forest in Michigan, it is home to
the largest white pine tree (Pinus
strobus, Pinaceae) in the world. (Pho-
tograph by S. Scheiner.)
Biomes 391
evergreen trees and broad-leaved deciduous trees, might stantially from the original forests. Today, because var-
be classified with temperate deciduous forests. ious pine species can be grown quickly for pulp, many
Needle-leaved forests are found in areas of the east- pine forests are maintained in this state by logging and
ern United States and western Europe where drought forestry practices, with none of the natural diversity or
due to sandy soil, low soil fertility, or shallow soil on structure remaining. Periodic fires in unmanaged areas
rocky outcrops favors pines over deciduous trees. Exam- may also help to maintain pine dominance. The remain-
ples are the “pine barrens” and “piney woods” of the ing natural pinelands, especially those dominated by
eastern and southern coastal plains of the United States. Pinus palustris are among the most endangered ecosys-
These forests are dominated by pitch pine (Pinus rigida) tems in the United States. Many of these areas are now
in parts of the eastern coastal plain and on rock outcrops being urbanized.
along the East Coast, and by longleaf pine (P. palustris)
and slash pine (P. elliottii) from Virginia to East Texas, Temperate Woodland
along with various tree and shrub oaks (Quercus spp.) Temperate woodlands, like their tropical counterparts,
and low-growing ericaceous shrubs. All of these forests occur on the drier margins of forests and often form tran-
experience frequent fires, and to varying extents depend sition zones between forests and either grasslands or
on fire for their maintenance and regeneration. Many of shrublands. Diversity and productivity are lower in
the species in these forests have numerous adaptations woodlands than in deciduous forests. Temperate wood-
to fire (see Chapter 13). Productivity and decomposition lands range from nearly continuous tree canopies to very
rates are typically low in these forests. Diversity can vary open, with scattered trees (Figure 19.13). The tree species
extensively; while many have low diversity, longleaf can range from needle-leaved evergreens to broad-leaved
pine forests include some of the most species-rich sys- sclerophylls to broad-leaved deciduous trees with heights
tems in North America. from 3 to 7 m. In North America, coniferous woodlands
In the southeastern United States, pine-dominated are extensive in the area between the Rocky Mountains
forests are currently more extensive than they were sev- and the Sierra Nevada, where they are dominated by sev-
eral hundred years ago. Areas that were originally eral pinyon pine (Pinus) and juniper (Juniperus) species.
broad-leaved forests were cleared for farming, which Oak (Quercus) and oak-pine woodlands occur from
depleted the soil of nutrients. When the farms were Washington to the Central Valley of California and across
abandoned during episodic convergences of bad eco- areas of the southwestern United States and northern
nomics and bad weather, from the nineteenth century to Mexico, east to Texas. They are also common around the
the Great Depression of the 1930s, the forests returned, Mediterranean Sea (where they are again oak-dominat-
but they are now dominated by pines and differ sub- ed) and in South America, Africa, and Australia.
Figure 19.13
A pinyon-juniper woodland in northern Arizona. (Photograph by S. Scheiner.)
392 Chapter 19
Taiga short, warm summers (with only 50 to 100 frost-free
The taiga is a broad swath of primarily coniferous for- days). Precipitation is low (38–50 cm/year), but the cli-
est that covers an enormous region of the circumpolar mate is humid due to very low evaporation rates. Large
subarctic. Similar vegetation is found in many subalpine parts of this region were glaciated during the last Ice
regions globally (Figure 19.14). The name “taiga” comes Age. Recurring disturbances are an important feature of
from the Russian word for boreal forest, which is anoth- the landscape and include forest fires, treefalls due to
er name for this biome. The taiga is one of the largest of storms, and large areas defoliated by insect attack. Pro-
the world’s biomes. It encompasses approximately 12 ductivity and decomposition rates tend to be very low,
million km2, occupying 8% of the Earth’s land surface, and soils are generally very acidic.
makes up about a third of the world’s forested area, and Species diversity and productivity in boreal forests
stretches from Canada and Alaska to Scandinavia, Rus- are low, and these forests are generally dominated by
sia, China, Mongolia, and parts of Japan and the Kore- just a few species of coniferous trees. Four tree genera
an peninsula. In North America, boreal forests can also dominate much of the circumpolar taiga: Picea (spruce),
be found at high elevations in the Appalachian Moun- Abies (fir), Pinus (pine), and Larix (larch or tamarack).
tains, the Rocky Mountains, and the Sierra Nevada; they (Larches are one of the few deciduous conifer species.)
are found in similar mountainous regions across Eura- The understory of a boreal forest is typically shady, with
sia. At their northern or upper elevation margins, these few species. Broad-leaved deciduous trees and shrubs
forests open up into woodlands, muskeg (northern are common dominants in both primary and second-
bogs), or wind-swept shrublands. ary succession, with Alnus (alder), Betula (birch), and
Taiga climates are largely continental, with long, bit- Populus (aspen) particularly common. In much of Cana-
terly cold winters (up to six months below freezing) and da and Alaska, for example, the dominant species are
Picea glauca (white spruce) and Picea mariana (black
spruce), with lesser numbers of Abies balsamea (balsam
fir), Larix laricina (larch), Pinus banksiana (jack pine), Pinus
contorta (lodgepole pine), Populus tremuloides (quaking
aspen), Populus balsamifera (balsam popular), and Betu-
la papyrifera (paper birch).
Because low decomposition rates lead to the accu-
mulation of peat (see Chapter 15), the taiga is a “net car-
bon sink,” playing a critical role in the global carbon bal-
ance by absorbing about 700 million tons of carbon per
year, about equal to the annual emissions of China.
Large areas of the taiga remain as old growth, or at least
are intact and relatively undisturbed. The taiga of Rus-
sia and Canada constitutes more than half the world’s
remaining old-growth forest. Because human popula-
tions in the taiga are low, the primary threat to these
forests is commercial lumbering for timber, pulp, and
paper. In Sweden, for example, large-scale commercial
logging for over a century and replanting with mono-
cultures of timber species has transformed many bore-
al forests into low-diversity timber plantations. Climate
change, mining, and oil and gas drilling also threaten
many taiga ecosystems.
Temperate Shrubland
Temperate shrublands are found in regions with
mediterranean climates: cool, wet winters and warm to
hot, dry summers. They occur from 30° to 40° latitude
on the western coasts of continents, adjacent to cold
ocean currents offshore. These regions are highly isolat-
Figure 19.14 ed from one another on the different continents, result-
Taiga along the southern coast of Alaska. (Photograph cour- ing in many endemic species. Names for this biome dif-
tesy of L. Walker.) fer around the world: garrigue and maquis around the
Biomes 393
Figure 19.15
Temperate shrubland (chaparral)
in California. Chaparral vegetation
such as this is subject to periodic
fires. (Photograph courtesy of J.
Thomson.)
Mediterranean Sea, fynbos in the Cape region of south- ish settlers (Woodward 1997). There is wide variation
ern Africa, matorral in central Chile, kwongan in south- among continents in the current human population size
western Australia, and chaparral on the southern coast in and the nature of the human effects on shrublands.
of California. Species diversity ranges from low to excep- Human populations are not large or rapidly growing
tionally high, most notably in the fynbos (see Box 20A). in most shrubland regions, although in California, sub-
All of these different types of temperate shrubland are urbanization and development coupled with rapid pop-
dominated by sclerophyllous evergreen shrubs (Figures ulation increases are major threats to chaparral. The
19.15 and Figure 19.16). Many members of the shrub shrubland of Australia has a low human population, but
flora are highly aromatic, including many members of in some places has been cleared for agriculture. The fyn-
the mint (Lamiaceae) family (e.g., sage, rosemary, thyme, bos region is particularly threatened by invasive plant
and oregano). Many contain highly flammable oils. and animal species.
Total annual precipitation in temperate shrublands
is 40–100 cm/year, but precipitation is highly concen-
trated in a few months of the cool season. Temperatures Grasslands
are cool to hot, moderated by proximity to the ocean and
by fog caused by the cold ocean currents. There is a short, Temperate Grassland
predictable growing season when soil moisture and tem- Temperate grasslands (Figure 19.17) are known by many
peratures are sufficient for growth; most plant growth names around the world: prairie (North America),
and flowering occurs from late winter to spring. Many steppe (vast parts of north-central Eurasia), veldt (South
of the plants are adapted to drought. Fires are frequent, Africa), pampas (South America), and puszta (Hungary).
as the residents of California suburbs encroaching into They are generally found in moderately dry continental
this biome have been dismayed to discover. In addition climates with cold, dry winters and warm to hot sum-
to the dominant shrubs, annuals are very common, espe- mers with higher rainfall. Grasslands are the largest of
cially in the years immediately following a fire. the North American biomes; at one time, they made up
Temperate shrublands have been affected by 266 million ha (35%) of the United States. The grasses
humans for millennia by both livestock grazing and the can be short, especially in drier areas, or tall. Andropogon
use of fire. The Mediterranean region itself, as we know gerardii (big bluestem), a dominant species of the tall-
from classical Greek literature, was formerly forested grass prairie of central North America, can grow as tall
with evergreen oaks, pines, cedars, wild carob, and wild as 2.5 m. The tallest grasses and greatest productivity are
olive. Many scientists believe that the shrublands of Cal- in the eastern most part of the North American biome,
ifornia, likewise, became much more extensive as a with stature and productivity declining to the west.
result of deliberate burning by Native Americans as a The invention of the steel plow by John Deere in
management tool, as well as livestock grazing by Span- 1837 resulted in the conversion of almost all of the tall-
394 Chapter 19
California Chile Sardinia South Africa Figure 19.16
Examples of leaf morphology in four
widely separated areas with temper-
ate shrubland biomes. The leaves
illustrated are from southern Califor-
nia in North America (chaparral),
Rhus ovata Kageneckia oblonga Arbutus unedo Protea arborea northern Chile in South America
(matorral), the island of Sardinia in
the Mediterranean Sea, and South
Africa (fynbos). Leaves in temperate
shrublands tend to be sclerophyllous:
Brabejum tough, evergreen, relatively small,
Heteromeles arbutifolia Lithraea caustica Quercus ilex stellatifolium rounded, and thick. (After Cody and
Mooney 1978.)
Phillyrea Metrosideros angustifolia

Quercus agrifolia Cryptocarya alba
angustifolia
Arctostaphylos glauca Quillaja saponaria Phillyrea media Psorales obliqua
Leucadendron
Ceanothus leucodermis Colliguaya odorifera Rhamnus alaternus salignum
Ceanothus greggii Trevoa trinervis Cistus salvifolius Ursinia pinnata
Quercus dumosa Satureja Juniperus oxycedrus Agathosma ciliata

gilliesii
Adenostoma fasciculatum Erica arborea Stoebe plumosa
grass prairie, and much of the mixed prairie to its west, In many grasslands, fire is the key factor in the exclusion
to cropland. Today only a few prairie remnants are pre- of trees and shrubs. All of the U.S. prairie regions east of
served, and the only common tallgrass species that you the Mississippi River are fire-maintained; evidence sug-
are likely to see is maize (Zea mays, which does not occur gests that Native Americans used fire to maintain these
at all in nature, but was created by human selection and grasslands in areas that would otherwise be temperate
breeding in pre-Spanish Mesoamerica). In Illinois, for deciduous forest.
example, less than 1% of the original prairie still exists. Human effects on temperate grasslands are extensive.
The grasslands that have been converted to croplands Fire exclusion has led to the invasion of many grasslands
for growing grain worldwide are often described as “the by woody plants. Human hunting of large native grazers,
breadbasket” of their particular country. such as bison in North America, has also favored woody
Rainfall is strongly seasonal in most temperate plants such as shrubs that are vulnerable to grazing. High
grasslands, leading to seasonal drought. In some cases, densities of domestic cattle have also had major effects.
the total precipitation would be enough to support trees One major difference between bison and cattle is that
if it were more evenly distributed throughout the year. while bison were always on the move, allowing the grass-
Biomes 395
Figure 19.17
Tallgrass prairie in Nebraska, in the center of the United
States. It is easy to see why this biome is sometimes poetical- to their original condition is costly, labor- and time-inten-
ly referred to as “a sea of grass.” (Photograph by R. and J. sive, and depends on a range of conditions; it is there-
Pollack/Biological Photo Service.) fore unlikely at the present time to provide workable
solutions for anything other than very limited areas. On
the other hand, the return of ecosystem structure and
es to recover, cattle are kept permanently in one place function, albeit to a very different community, is much
for long periods. The introduction and spread of invasive easier to accomplish.
non-native plant species in North America has also result- Anthropogenic disturbances in grasslands can have
ed in dramatic changes in grasslands at the community, interactive effects, as seen in the disappearance of the val-
ecosystem, and landscape levels. ley grasslands of California. The Central Valley of Cali-
Enormous regions of temperate grassland have been fornia was once a vast perennial grassland, with exten-
converted to irrigated cropland. Such intensive agri- sive wetland areas and fertile riverbeds. Spanish explorers
culture is often not sustainable over the long term. For wrote that, in wet years, it was impossible to ride a horse
example, much of the North American Great Plains is across the Sacramento Valley due to the density and
irrigated with water from the Oglala aquifer that under- height of the vegetation, as well as the widespread wet-
lies the southern part of the region. This water will not lands. This vegetation is now entirely gone, having been
be available forever; the aquifer is being used up much replaced by short Mediterranean annual grasses (and
faster than it is being recharged. Over many years, irri- some exotic forbs). Conversion to irrigated cropland,
gation leads to salinization of soil because the water while important, was not the main reason for this change.
leaves behind concentrated mineral deposits when it Interactions between drought, high densities of non-
evaporates. Moreover, the costs of pumping the water native domestic cattle and sheep, and the introduction
increase over time as the water table drops. Irrigated of non-native plants all contributed to the demise of the
agriculture has already become untenable in parts of the perennial grasslands. The growth form of the native grass-
Great Plains, and the problems being experienced there es in clumps provides open spaces that allow the estab-
will expand to other regions in the future. Meanwhile, lishment of invading plants. The native grasses also do
the native grassland is gone and cannot replace the irri- poorly under heavy or close grazing. When the United
gated cropland—and the change has occurred within States took possession of California from Mexico, and
just two or three generations. The efforts of restoration grazers were introduced in the 1850s and 1860s, the bal-
ecologists to restore grassland ecosystems have met with ance began to tip toward the successful incursion of non-
a range of results, from what optimistically appear to be native plant species. The fact that the area was then
long-term successful restorations to complete failures. undergoing a prolonged drought also played a role in giv-
Successful restoration of prairies and other grasslands ing the invasive annuals an advantage.
396 Chapter 19
occurs on nutrient-poor soils. The
FPO llanos of the Orinoco basin of

Venezuela and Columbia are grass
savannas maintained by the annu-
al flooding of the Orinoco River,
which results in long periods of
standing water that prevent the
establishment of trees. The cerrado
of Brazil has an open canopy of
short trees with a distinctive twist-
ed form; its acidic, heavily leached
soils are high in toxic aluminum
ions and low in plant nutrients. This
region has many endemic species,
and the overall plant species diver-
sity approaches that of tropical rain-
forest. The pine savannas of Central
America occur on dry, low-nutrient
quartz sands. The savannas of
Figure 19.18 Southeast Asia are generally con-
The Masai Mara Reserve of Kenya is part of the great sidered to be created and maintained by humans through
African savanna known as the Serengeti. (Photograph © frequent fires.
A. Jones/Photo Researchers, Inc.) Grazing can combine with fire to establish or main-
tain savanna vegetation. In East Africa, elephants debark
trees and, more dramatically, knock them over, opening
Tropical Savanna woodlands to invasion by grasses. The presence of grass-
Tropical savannas are perennial grasslands that may have es increases the attractiveness of the area to other graz-
an open tree canopy or scattered shrubs (Figure 19.18). ing animals, such as antelopes and zebras, which eat and
“Savanna” is a Hispanicized variation of a Native Amer- trample tree seedlings and prevent woodland regener-
ican word for “plains.” Like temperate grasslands, savan- ation. Only thorny trees and shrubs can become estab-
nas occur in areas with distinct dry and wet seasons that lished in this environment, leading to the creation of an
would be seasonal forests or woodlands if soil conditions acacia savanna.
or disturbances did not limit the establishment of canopy The world’s greatest diversity of ungulates (hoofed
trees. Rainfall is 75–125 cm/year, with a dry winter sea- mammals) is found on the savannas of Africa, including
son that lasts at least five months. Disturbances are buffalo, giraffes, eland, impalas, gazelles, oryx, gerenuk,
caused largely by grazing or frequent fires, or occasion- wildebeest, plains zebra, rhinoceroses, elephants, and
ally by other factors such as soil saturation (due to imper- warthogs. The vast herds of herbivores support carni-
meable soils). The most extensive tropical savannas are vores that include lions, leopards, cheetahs, jackals, and
found in Africa and Brazil (cerrado); others are found in hyenas. In contrast, few Neotropical (Central and South
Central America, Australia, and Southeast Asia. American) mammals or birds are restricted to savannas,
Savannas are dominated by tall perennial grasses although capybaras (a very large, semiaquatic rodent)
1–2 m high, typically interspersed with drought- and are typical of llanos, and anteaters are common in South
fire-resistant trees or shrubs. Palm, eucalyptus, and aca- American savannas and are endemic to the Neotropics.
cia savannas are examples of this vegetation type, each Termites are especially notable residents of all trop-
named for its scattered overstory trees. Savannas have ical savannas; these detritivores are important in soil for-
a wide range of characteristics. Most are fairly high in mation. The termite-eating aardvarks and pangolins that
species diversity. Decomposition rates are lower than live in the savannas are among the oddest mammals
those in tropical forests, but are still generally very high, on Earth.
and soils are quite variable. The acacia savannas of east- Like many other biomes, tropical savannas are
ern and southern Africa are a mosaic of local commu- threatened by human population growth. In Africa the
nities created and maintained by fire and grazing, and greatest challenge of preserving the large herbivores is
are preserved in Kenya, Tanzania, Zimbabwe, Botswana, to find ways in which they can coexist with humans.
Namibia, and South Africa. The famous Serengeti Plains Beyond the problems of poaching, elephants, for exam-
of Tanzania are acacia savanna growing on nutrient-rich ple, are at risk when they wander off the nature reserves
volcanic sands, although most savanna vegetation and humans shoot them to prevent them from destroy-
Biomes 397
ing vital food crops. Conservation plans that include eco- nounced seasonal wet/dry cycles, although some of the
nomic incentives and participation in decision making driest deserts may experience periods of many years
by the people living in the region, and not just by out- without any measurable rainfall.
side conservation experts, are likely to be the most suc- Precipitation in the desert typically comes at one
cessful strategies for preserving these and other natu- time period during the year. In North America, the
ral systems. Mojave Desert receives almost all of its rain in winter,
while the Chihuahuan Desert receives mainly summer
rains. The Sonoran Desert is unique among the deserts
Deserts of the world in that it has two periods of rainfall, a win-
ter and a summer rainy season, with very dry autumn
Hot Desert and spring seasons separating them. As a result, the
Deserts and semi-deserts occupy large areas of western Sonoran is the greenest of the world’s deserts, with the
North America, Australia, Asia, and Africa. Deserts highest standing biomass and productivity, and is one
occur where potential evapotranspiration exceeds actu- of the most floristically diverse.
al evapotranspiration—in other words, where water can As a rule, the lower the average annual rainfall, the
be lost to the atmosphere at a greater rate than it is nor- greater the variance in both the amount and timing of
mally made available to plants. Typically this means that annual rainfall. The “rainy season” in most deserts is a
desert regions receive less than 25 cm of precipitation time when rain is more likely, although it is not uncom-
per year, but such numbers can be deceptive; for exam- mon to have very little rain in any given rainy season.
ple, the coastal regions of southern California receive less Consequently, desert organisms must be adapted not
rainfall than this, but they are not deserts because cold only to limited moisture and high temperatures, but also
offshore currents act to lower temperatures and increase to the lack of predictability in rainfall (see Chapter 3). At
relative humidity. Thus, San Diego, which has a mediter- one extreme of adaptation, annual plants spend most of
ranean climate, typically receives less annual rainfall their lives as desiccation-resistant seeds, living only
than Tucson, in the Sonoran Desert. briefly as photosynthetic organisms when conditions are
Hot deserts (also called “warm deserts”) are found most favorable. At the other extreme, succulents and
in two belts centered at 30° N and S latitudes. The hot evergreen shrubs have photosynthetic tissue that is
desert biome includes the Mojave, Sonoran, and Chi- always exposed to potential desiccation. Drought-decid-
huahuan deserts of North America, the Sahara, Kalahari, uous shrubs are intermediate, with little moisture-los-
and Namib deserts of Africa, the Arabian Desert of Asia, ing tissue exposed during the long dry periods. In part
the Atacama Desert of South America, and the deserts because of the unpredictability of rainfall, succulents are
of Australia (Figure 19.19). These areas all have pro- less abundant or absent in the driest deserts; they are
characteristic of deserts with
relatively greater and more
predictable rainfall. In Death
Valley, California, which is
extremely hot and dry, annu-
al plants are the dominant
growth form. The other com-
mon growth form in Death
Valley is drought-deciduous
shrubs. In contrast, succulent
cacti, agaves, and yuccas are
abundant and diverse in the
Sonoran Desert.
Because annual species
predominate in hot deserts,
their seeds are usually rela-
Figure 19.19
The Sonoran Desert outside
Phoenix, Arizona. (Photo-
graph by S. Scheiner.)
398 Chapter 19
tively abundant in the soil, and consequently there are turbs the soil, increasing evapotranspiration rates by
many granivorous (seed-eating) animals, particularly reducing cover. Use of recreational off-road vehicles,
rodents, ants, and birds. Southern Arizona, for instance, common across the American desert landscape (as well
has the highest diversity of rodents in the world, and as in other regions), is highly destructive to the fragile
almost the highest diversity of seed-eating ants. vegetation. Because the vegetation is responsible for
Although the seeds of desert annuals are abundant, they holding the soil in place, the loss of vegetation through
do not usually disperse well. Many desert plant species grazing and other human activities leads to soil loss and
have adaptations that limit their ability to disperse. severe erosion.
Areas from which plants have been cleared often are not
recolonized for decades or even centuries. Cold Desert
Deserts are important areas for agriculture, grazing, Cold deserts (also called “cool deserts”) are found in rain
and mineral exploitation, and they are becoming increas- shadow areas in the centers of continents, the largest
ingly important for recreation as well. For almost all of being the Kazakho-Dzungarian and Gobi deserts of cen-
human history and prehistory, human populations in tral Asia and the Intermountain or Great Basin desert of
deserts have been very low. In the last few decades, how- North America (Figure 19.20). Unlike hot deserts, cold
ever, there have been rapid increases in human popu- deserts experience freezing temperatures for part of the
lations in some desert areas. The explosive growth of year. All deserts can be quite hot in the summer, how-
desert cities, such as Phoenix, Arizona, has meant that ever. Precipitation is generally concentrated in the win-
human activities are having massive effects on natural ter. Cold deserts are dominated by shrubs and grasses
systems in such areas. Underground aquifers are being and have few of the succulents found in some hot
depleted by urban dwellers and by agricultural use, and deserts, as most succulents are vulnerable to damage by
surface waters (such as the Colorado River in the south- ice crystal formation. Cold deserts also have few annu-
western United States) face very heavy demands. Irri- al plants because moisture is available mainly as
gated agriculture can lead to soil salinization in deserts, snowmelt, precisely when temperatures are too cool for
just as it does in grasslands. This process is thought to rapid growth. There may also be historical reasons for
have caused the collapse of several ancient civilizations the dearth of annuals. In the Great Basin of North Amer-
in the American Southwest. Environmentally damaging ica, for example, there are few native annual species, but
mining and oil extraction are additional sources of the successful invasion of the Old World annual grass
severe disturbance to biotic systems in many deserts. Bromus tectorum is evidence that annuals can do well
Overgrazing, by both livestock and feral ungulates, has there. Most of the anthropogenic threats to hot deserts
widespread effects in many deserts. Heavy grazing dis- also threaten cold deserts.
Figure 19.20
Sagebrush (Artemesia, Asteraceae) dominates this cold desert landscape near Mono
Lake, California, part of the North American Great Basin. (Photograph by E. Ely/
Biological Photo Service.)
Biomes 399
Figure 19.21
Alpine shrubland on Mt. Wash-
ington in the state of New
Hampshire. (Photograph ©
M. P. Gadomski/Photo
Researchers, Inc.)
Alpine and Arctic Vegetation dominated by low-growing ericaceous shrubs, and in

some parts of the world alpine shrublands consist to a
Grassland and Shrubland large extent of heath vegetation. In the northern tem-
Alpine shrublands extend upward from tree line on high perate zone, the shrubs are typically Salix (willow) and
mountains, and are often interspersed with alpine grass- Betula (birch).
lands (Figure 19.21). The vegetation in these areas can Above tree line in mountainous areas or in areas
include both what we typically think of as shrubs— with cold air drainage, alpine grasslands are the major
multi-stemmed woody plants of short stature—and community type (Figure 19.22). These grasslands are
bizarre-looking plants that have a rosette form that can often dominated by sedges, plants in the family Cyper-
grow 3–5 m tall (see Figure 6.12). Alpine shrublands aceae that are closely related to and similar in growth
are found on all of the continents except Australia and form to grasses. These communities can also contain
Antarctica. In South America, they constitute a large part scattered shrubs, dwarf trees, and small, herbaceous
of the high-altitude vegetation type called paramo. perennial plants that are shaped like little cushions, thus
Heath is the name for a general vegetation type that is grading into alpine shrublands and tundra.
Figure 19.22
Alpine grassland in Rocky
Mountain National Park,
Colorado. (Photograph by
S. Scheiner.)
400 Chapter 19
Temperate and tropical alpine areas differ in some iting tree growth is that emergent vegetation above the
important ways. Temperate alpine regions have long, snow cover in winter would experience not only extreme
cold winters, during which the ground is frozen and cold, but also high winds, especially in alpine areas. Tun-
deep snow covers much of the vegetation for long peri- dra soils are generally young and poorly differentiated
ods. Snowmelt generally marks the start of spring (often inceptisols; see Table 4.1) because most arctic tun-
growth. In contrast, tropical alpine regions, especially in dra sites were glaciated in the recent past. Litter decom-
very high mountains, have a strong daily temperature position is generally slow due to the cold temperatures
cycle, as if it were winter every night and summer every and consequently low microbial activity and, secondar-
afternoon. Growth continues year-round, and the soil ily, the anaerobic conditions in many places.
does not freeze (at least not deeply or for long periods Alpine and arctic tundra have similarities and dif-
of time). All alpine regions have very high solar energy ferences. Both have short growing seasons, and alpine
input because the atmosphere is thinner than at sea level, tundra (at least in temperate areas, although not in the
so there is less to filter (i.e., absorb and scatter) incom- Tropics) is subject to severe cold in winter. In both types
ing sunlight. The sunlight is particularly intense in trop- of tundra, most precipitation is in the form of snow, so
ical alpine regions because the angle of incoming solar that there are long periods during which plants are
radiation is most direct over the Tropics. The resulting unable to take up moisture; in many settings (such as the
high levels of ultraviolet radiation can damage bio- alpine tundra at mid-latitudes in western North Amer-
chemical processes in leaves (and can cause severe sun- ica), drought is common during the summer. Arctic and
burn in humans). alpine tundra differ greatly in other factors, including
The total area occupied by alpine vegetation of all day length, atmospheric pressures, partial pressures of
types is low. Precipitation is generally limited in these gasses such as CO2, exposure to ultraviolet radiation,
areas, and productivity and species diversity are also low. and soil conditions.
The harsh climate, steep slopes, and low productivity Many genera and species found in Northern Hemi-
of alpine regions have combined to keep human popu- sphere alpine tundra are also found in the arctic tundra.
lation sizes low from prehistory to the present, although Oxyria digyna (mountain sorrel, Polygonaceae) is a plant
several distinct cultures, such as those of the Incan and found both in the Arctic and in alpine areas of the west-
Tibetan peoples, have developed and thrived in high ern and eastern United States and Europe. A compara-
mountain regions. Even small human populations, how- tive study of arctic and alpine populations of this plant
ever, can still have significant effects on alpine habitats. concluded that they are each adapted to specific local
The cold temperatures mean than many processes, such conditions. Compared with the arctic plants, the alpine
as plant growth, soil formation, and decomposition, are plants had a lower leaf chlorophyll content, a higher res-
very slow. As a result, human activities such as mining piratory rate, and a lower light saturation point. Alpine
and grazing (the primary form of agriculture in these tundra species are often evolved from local lowland
regions) can have very long-lasting effects. species. In the high Andes, for example, there are many
species of high-altitude bromeliads, a group we nor-
Tundra mally associate with tropical forests.
Tundra is found in areas where temperatures are too The growth forms of tundra plants include low
cold for tree growth. There are vast areas of arctic tun- shrubs, taller shrubs alongside rivers, herbaceous peren-
dra in northern North America and Eurasia. Because nials, grasses and sedges, mosses, and lichens. Different
there is little land in the Southern Hemisphere at high plant communities exist within the tundra, depending
latitudes, there is little tundra there—only a few areas on topography, soils, and other factors. Tussock tundra
on the Antarctic Peninsula and neighboring islands are is dominated by grasses and sedges (notably by the
referred to as tundra. Conditions at the tops of high sedge Eriophorum vaginatum), deciduous shrubs, and
mountains, including wind and extreme cold, create evergreen shrubs in roughly equal abundance. Riparian
alpine tundra. bottomlands are dominated by relatively tall (1–4 m) wil-
The arctic tundra has long, dark, and often bitterly lows and other shrubs. Wet sedge tundra, dominated by
cold winters and short, cool summers with up to 24 hours low-statured (< 20–30 cm) rhizomatous sedges, covers
of sunlight. The soil thaws each summer to a depth of flat areas with restricted drainage, forming a broad
only 25–30 cm, resulting in permafrost, a permanently expanse on the Arctic coastal plain of Alaska as well as
frozen layer of soil, which severely limits root growth. in other areas. Soils typically thaw to only 25–30 cm here,
Trees are excluded from the arctic tundra by a combina- and the freeze-thaw cycle often creates a characteristic
tion of factors, including permafrost and extreme cold. ice wedge-shaped pattern in the soil. The soils of wet
Temperatures below 40°C kill almost all living above- sedge tundra have very high organic matter content and
ground plant tissue, preventing long-lived plants such may be covered with water during the summer because
as trees from becoming established. Another factor lim- the frozen soil prevents drainage.
Biomes 401
Plant species diversity in the tundra is very low. that are notable components of the arctic fauna, and
Many genera (such as Salix) and even some species (e.g., marine species such as walrus and various seal and
Dryas octopetala, Eriophorum vaginatum, and Oxyria dig- whale species are native to, or rely heavily on, arctic
yna) are common across the circumpolar Arctic (Figure habitats. The arctic tundra serves as critical nesting area
19.23), and similar communities are found over wide for many bird species, particularly those associated with
ranges. Animal population densities and species rich- wetlands.
ness are also low. However, the Arctic provides critical Human population sizes and densities in the tundra
habitat both for species that are characteristic of this are extremely low (under 1 person/km2), although
region, such as lemmings, and for others, such as humans have lived in, traveled through, and exploited
wolves, that have been reduced to low numbers else- arctic environments since prehistoric times. In spite of
where. Brown bears, grizzly bears, polar bears, snow- these low population sizes, humans have managed to
shoe hare, lynx, moose, and caribou (called reindeer in pollute arctic environments in disproportion to our num-
northern Europe) are some of the other large mammals bers, and the cold temperatures mean that the pollution
persists for a long time. Small soil disturbances can last
a very long time. In the arctic tundra, disturbance may
result in the melting of permafrost, which leads to sub-
sidence and soil compaction. Perhaps the most threat-
ening of all ecological effects is human-caused global
warming, which is predicted to be greatest in polar
regions (see Chapter 22).
Summary
Biomes are large regions that differ from one another
in the structure of their vegetation. They are strongly
determined by climate—temperature, rainfall, and sea-
sonality—and, to a lesser extent, by soil properties.
While the boundaries of biomes are defined arbitrarily,
the concept of the biome remains a useful one so long as
it is clear that it is a descriptive categorization. There is
a broad consensus among scientists on the classification
of biomes.
Biomes are classified based on whether they are
dominated by trees (forests and woodlands), shrubs, or
grasses and other herbaceous species. Forests are fur-
ther classified by whether the trees are evergreen or
deciduous, broad-leaved or needle-leaved. Where con-
ditions become too dry for trees, grasses predominate;
grasslands can be tropical, temperate, alpine, or arctic.
Shrubs dominate under a variety of conditions, includ-
ing those that are too hot or dry for grasses, those with
very seasonal precipitation, in some very cold regions,
or as a secondary result of overgrazing or fire suppres-
sion.
All biomes are heavily affected by humans. Some,
such as temperate deciduous forests, can recover rela-
tively rapidly from human disturbance because many
of their species disperse readily and grow rapidly. Oth-
ers, such as hot deserts, recover poorly because their
plants disperse poorly and grow slowly. Full recovery
in still others, such as tropical forests, is difficult because
of soil destruction and species extinctions in these high-
Figure 19.23
The vegetation of this Arctic tundra along Glacier Bay in ly diverse communities. Recovery of all damaged or
southern Alaska is dominated by species of Dryas degraded biomes is limited by continued encroachment,
(Rosaceae). (Photograph courtesy of L. Walker.) habitat destruction, and global climate change.
402 Chapter 19
Additional Readings
Classic References Pastor, J., Y. Cohen and R. Moen. 1999. Generation of spatial patterns in
boreal forest landscapes. Ecosystems 2: 439–450.
Braun, E. L. 1950. Deciduous Forests of Eastern North America. Blakiston,
Philadelphia, PA. Pitman, N. C. A., J. Terborgh, M. R. Silman and P. Nuñez V. 1999. Tree
species distribution in an upper Amazonian forest. Ecology 80: 1651–1661.
Rübel, E. 1936. Plant communities of the world. In Essays in Geobotany
in Honor of William Albert Setchel, T. H. Goodspeed (ed.), 263–290. Uni-
versity of California Press, Berkeley, CA. Additional Resources
Schimper, A. F. W. 1903. Plant-Geography upon a Physiological Basis. Bowden, C. 1993. The Secret Forest. University of New Mexico Press, Albu-
Clarendon Press, Oxford. querque.
Tansley, A. G. 1939. The British Islands and Their Vegetation. Cambridge Brown, J. H. and M. V. Lomolino. 1998. Biogeography. 2nd ed. Sinauer
University Press, Cambridge. Associates, Sunderland, MA.
Davis, M. B. 1996. Eastern Old-Growth Forests. Island Press, Washington,
Contemporary Research DC.
Brais, S., C. Camire, Y. Bergeron and D. Pare. 1995. Changes in nutrient Reisner, M. 1993. Cadillac Desert: The American West and Its Disappear-
availability and forest floor characteristics in relation to stand age and for- ing Water. Penguin Books, New York.
est composition in the southern part of the boreal forest of northwestern
Quebec. Forest Ecol. Mgmt. 76: 181–189.
20
C H A P T E R
Regional and Global Diversity
A
bout a quarter of a million plant species are currently known world-
wide, and there may yet be many undescribed species. Strikingly, some
parts of the world have a great many more species than others. Some
tropical forests can have as many as 1500 species of flowering plants (includ-
ing 750 species of trees) in 1000 hectares—in Indonesian Borneo, researchers
recorded more than 400 tree species in a 0.75 km2 plot! In contrast, the north-
ern regions of Canada and Russia may have only a few dozen species spread
across hundreds of square kilometers. Why do some regions of the world have
so many more plant species than others?
In this chapter, we describe regional and global patterns of diversity and
discuss progress in understanding their causes. Local-scale patterns of diver-
sity were discussed in Chapter 14. Here we look at patterns at two other scales:
at the largest scale (across continents and the entire globe), and at intermedi-
ate scales (across communities within a region). At each scale, we examine some
of the hypotheses that have been proposed to explain the patterns observed.
Some of these hypotheses have now been ruled out, others explain the observed
patterns at least in part, and still others are being vigorously debated by ecolo-
gists. No one explanation can account for all of the observed patterns of species
diversity. Rather, ecologists seek to assess the relative importance of the vari-
ous factors contributing to biodiversity in different places and over time. In
Chapters 18 and 19 we saw how one of those factors—climate—determines pat-
terns of community physiognomy and the distribution of biomes. Here we see
how climate, among other factors, influences biological diversity.
Besides being an intrinsically interesting subject, understanding patterns
of diversity is important for conservation. An ever greater proportion of the
planet is occupied by human-dominated environments such as farms, homes,
and industrial areas. Our domination of the landscape leaves wildlands more
diminished every year, and alters those that remain. Old-growth forests and
other relatively untouched plant communities are shrinking. With these
changes in land use and in the characteristics of the remaining natural habi-
tats comes an accelerating rate of species extinction (see Chapter 22).
If we wish to preserve species and community diversity, we need to under-
stand current patterns of diversity and the processes that drive these patterns.
Because plants are at the base of almost all terrestrial food webs (the only excep-
406 Chapter 20
tions being anaerobic food webs), the diversity of ani- 60
mals, fungi, and microbes is often linked to plant diver-
sity. Thus, the issues discussed in this chapter have prac- 50
tical as well as fundamental implications.
Number of species
40
Large-Scale Patterns of Diversity
30
Scientists first became aware of large-scale patterns in
species diversity as a result of the voyages of exploration
and colonization undertaken by Europeans in the eigh- 20
teenth and nineteenth centuries. Their ships often car-
ried botanists and zoologists as part of the crew, the 10
most famous example being the voyage of the Beagle
(1831–1836), on which Charles Darwin served as ship’s
0 500 1000 1500 2000 2500 3000
naturalist in. These naturalists made records of the
Elevation (m)
plants and animals they found, brought back specimens,
and added them to the growing catalog of described Figure 20.1
species. In the eighteenth century, Carl Linnaeus named Plant species richness along an elevational gradient in the
over 9000 terrestrial plant species; today there are Santa Catalina Mountains in southern Arizona. Both the low-
and high-elevation communities—the Sonoran Desert and
approximately 250,000 named plant species. boreal forest, respectively—are in extreme environments,
From these records, it soon became apparent that resulting in a peak in diversity at intermediate elevations.
tropical regions were typically very rich in species, polar (After Brown 1988; data from Whittaker and Niering 1975.)
regions were very species-poor, and temperate regions
had intermediate numbers of species. For example,
Brazil has over 56,000 named plant species, the United
States has about 18,000, while Canada has about 4200. ern Asia than in Europe, while North America has an
Species richness in tropical regions is probably under- intermediate number. In high mountain ranges, there are
estimated because many of these areas have still not typically more species at mid-elevations than elsewhere
been thoroughly explored . Even in North America, (Figure 20.1). Within continents, there are longitudinal
which has been studied by botanists for hundreds of gradients of diversity; in the boreal forest of North Amer-
years, new plant species are being named at a rate of ica, species richness within sites is higher in central Cana-
about 60 per year. These plants are occasionally found da than it is in either western or eastern regions (Figure
right under one’s feet—for instance, in 1997, a new 20.2). In the temperate zone, between approximately 35o
species, the mustard Lesquerella vicina (Brassicaceae), was and 60o latitude, there is a peak of within-community
discovered by the botanist James Reveal growing in his species diversity (Figure 20.3). In this chapter we explore
neighbor’s yard in Montrose, Colorado. some of the hypotheses that have proposed to explain
An explanation for the general latitudinal gradient these patterns of terrestrial plant species diversity.
in species richness may be the oldest major ecological
hypothesis (Hawkins 2001). Between 1799 and 1804, Levels of Explanation
Baron Alexander von Humboldt traveled through Mex-
ico, Central America, and northwestern South America. How have ecologists attempted to explain the causes
He subsequently published a series of essays under the of variation in species diversity along geographic or
title Ansichten der Natur (“Views of Nature”), in which environmental gradients? The processes that we use to
he described a global gradient in species diversity. He explain a pattern change with the temporal and spatial
postulated that this pattern was due to differences in cli- scale of that pattern. At short to medium time scales—
mate, specifically winter temperatures and the effects of up to centuries or millennia—ecological processes dom-
freezing. inate. The physiological and ecological properties of each
Besides the latitudinal gradient, many other patterns species result in species sorting among communities
of diversity have been observed at regional and global within a region. Over longer time periods, these prop-
scales. The regional diversity of a particular taxonomic erties, together with the history of each species, help
group may be very similar to or extremely different from determine the distributions of species at regional and
the diversity of that group in similar habitats in other continental scales. At even longer time scales, ecological
parts of the globe (Schluter and Ricklefs 1993). Compar- and historical factors affect the distributions of species
isons among Northern Hemisphere regions with similar across the globe. At the longest time scales—from mil-
conditions show that there are more plant species in east- lennia to hundreds of thousands or millions of years—
Regional and Global Diversity 407
allows it to occupy a new habitat. At regional scales, these
White spruce
Ba Black spruce
processes of adaptation can lead to the formation of new
80
species. If the new species is different enough, it may coex-
Aa Aa ist in the same habitat or landscape with the species from
Ab
which it evolved, increasing species richness in that land-
Number of species
60 Ab
scape. At continental to global scales, as new species
Ab
appear, they may migrate to new regions, so that species
40 richness increases over an entire set of regions.
Some proposed explanations for patterns of species
diversity assume that species distributions are in equi-
20 librium with current ecological conditions, while other
explanations do not make that assumption. Whether a
given explanation assumes that diversityis at equilibri-
0 um or not depends, in part, on scale. Within a single
Western Central Eastern
Canada community, for example, disturbance can increase diver-
sity (see Chapter 14); this is a nonequilibrium explana-
Figure 20.2
Number of species (mean ± 1 standard error) in 30 × 30 m2
tion for diversity. But at a larger scale, across commu-
plots in boreal forests of Canada. Species richness is higher nities in a landscape or region, disturbance in particular
in the central region, especially in white spruce communi- communities balanced by migration among communi-
ties. Bars with different capital letters indicate statistically ties may determine diversity (see Chapter 17). This is an
significant differences among regions within community equilibrium explanation for diversity, although the equi-
types. Bars with different small letters indicate statistically
significant differences within regions among community
librium is a very dynamic one.
types. (Data from Qian et al. 1998.)
evolutionary processes (adaptation, isolation, speciation, 80

Africa
and extinction) predominate. Some processes, including Eurasia
range expansions and contractions and extinctions of 70 North America
populations and species, may happen over long periods South America
of time or very quickly. Humans and their effects on the 60 Australia
environment have greatly accelerated these processes
Number of species per site
for some species (see Chapter 22).

50
The mix of processes governing diversity will also
differ depending on spatial scale. At local scales—with-
in communities—interactions among species are of pri- 40
mary importance in determining species richness. These
interactions include competition, herbivory, and mutu- 30
alisms (see Chapters 10 and 11). As our scale expands to
the level of the landscape or sets of communities, 20
metapopulation processes may become increasingly
important. These processes include migration among
10
communities and extinction within communities (see
Chapter 17). Finally, at continental to global scales,
migration and extinction in response to climate change 0 10 20 30 40 50 60 70 80 90
become important. We discuss one example, migration Latitude
following the most recent glacial retreat, below and in
Figure 20.3
Chapter 21. A latitudinal pattern in vascular plant species richness. Each
Evolutionary processes also operate at all of these point represents the mean number of species per site aver-
scales, although over longer time spans. At small spatial aged across sites in a particular landscape. Species richness
scales, local adaptation can occur (see Chapter 6). In of tropical landscapes is underestimated in this analysis. Lat-
response to competition with other species, a population itude is given in absolute degrees so that landscapes in the
Northern and Southern Hemispheres are superimposed. The
may evolve to occupy a different niche, thereby reducing latitudinal gradient is very weak because many factors affect
competition and increasing local diversity. Similarly, a pop- large-scale patterns of diversity (see text). (After Scheiner
ulation may evolve a defense against an herbivore that and Rey-Benayas 1994.)
408 Chapter 20
where. Random placement of species on a uniform globe
Explanations for Latitudinal Gradients
would result in many ranges overlapping along the
A natural tendency in the face of the strong patterns of equator, thus creating a peak in diversity in tropical
latitudinal diversity is to seek simple, unifying explana- regions and decreases in diversity toward the poles.
tions for those patterns. If, for example, we can account Because the bounded ranges model includes no eco-
for the temperate-tropical diversity gradient on a num- logical processes—only geometry—it serves as a useful
ber of continents with a single explanation, we are like- baseline against which to compare observed patterns.
ly to have a deeper understanding than if we try to For example, the model was found to account for some,
explain the patterns on different continents separately by but not all, of the latitudinal pattern of diversity in New
invoking details of the floras and histories of each conti- World bats (Lyons and Willig 1999). Actual species diver-
nent. Several theories have been put forward to explain sity was higher in the Tropics and lower at higher lati-
latitudinal diversity patterns (Table 20.1), some of which tudes than predicted by the model. Other biological
we delve into here. processes, therefore, are needed to explain the entire pat-
A useful starting point for evaluating hypotheses tern. Thus, a null model provides a useful starting point
is to build a null model. Used throughout the sciences, for discovering where interesting patterns lie and what
null models describe the patterns that would be about those patterns requires explanation. No analysis
observed if only random processes were operating. One of this kind has yet been done with plants.
such null model is the bounded ranges hypothesis (Col- A related idea depends on the observation that
well and Hurtt 1994; Willig and Lyons 1998). Developed species’ range sizes tend to increase with latitude and
as a possible explanation for the latitudinal gradient of elevation (Rapoport 1982; Stevens 1992) . This pattern is
animal diversity, this hypothesis uses a simple geomet- known as Rapoport’s rule. Because temperate-zone
ric argument. Species ranges are bounded by the North species have larger ranges than tropical species, tem-
and South Poles; otherwise, species can appear any- perate species’ ranges are more likely to extend into the
Table 20.1 Theories explaining the latitudinal gradient in species richness
Theory Mechanism References
Ecological and evolutionary theory

Available energy Increased energy supply leads to greater resource availability and thus larger 1, 2
population sizes. With more resources available, more species can coexist.
Larger populations have higher speciation rates and lower probabilities
of extinction.
Evolutionary theories
Speciation rate and temperature The Tropics have higher rates of speciation due to warmer temperatures, which 3
decrease generation times, increase mutation rates, and increase the speed
of selection.
Area and latitude There is more land area in the Tropics than in the temperate zones. Larger areas 4, 5
support more species due to higher rates of speciation and lower rates of
extinction.
Time There are more species in the Tropics because the communities there are older. 6
Climatic stability The Tropics have not been subjected to the large climatic fluctuations experienced 7
by temperate and polar regions during recent glaciations, so more species have
evolved or survived in the Tropics.
Ecological theories
Bounded ranges Peaks in species richness near the equator arise from a random association 8, 9
between the size and placement of species ranges.
Rapoport’s rule and species spillover Relatively low climatic variability at low latitudes leads to narrower species ranges. 10
Species at higher latitudes are thus more likely to spill over into the Tropics than
vice versa, inflating species richness in the Tropics.
Spatial heterogeneity The Tropics contain more different kinds of habitats than other regions.
Predation Higher levels of herbivory in the Tropics prevent competitive exclusion.
Disturbance Diversity is highest in regions with frequent disturbances, which is true of the Tropics.
References: 1, Wright 1983; 2, Wright et al. 1993; 3, Rhodes 1992; 4, Rosenzweig 1995; 5,Terborgh 1973; 6, Fischer 1960; 7, Sanders 1968; 8, Willig and Lyons
1998; 9, Colwell and Hurtt 1994; 10, Stevens 1989.
Tropics than tropical species’ ranges are to extend into which migrational lag cannot be the explanation, weak-
the temperate zones. The result will be more species in ens this hypothesis.
the Tropics—species that live only in the Tropics plus In contrast to the notion that climatic stability in the
those that have overlapping ranges from neighboring Tropics has led to increased diversity, some ecologists
temperate regions. Fine (2001) tested this hypothesis have proposed the opposite argument, suggesting that
using data for North American trees. He found that few speciation and species coexistence are most likely to
tropical species are also found in extratropical areas, and occur with more frequent disturbances. One reason to
concluded that Rapoport’s rule cannot explain the lati- believe that such a relationship might exist is that dis-
tudinal diversity gradient of plants. turbance may reduce the chance of competitive exclu-
Another simple explanation for latitudinal diversi- sion. However, there is no evidence that tropical com-
ty patterns, especially for the most extreme environ- munities differ from temperate ones in frequency of
ments, is differences in opportunities for growth. At very disturbance. In addition, as originally formulated, this
high elevations and latitudes, conditions are so unfa- hypothesis was unclear as to the scale of the disturbance
vorable that few species are physiologically capable of effects involved. That is, it might explain differences in
persisting there. At the summit of the Mauna Kea vol- local diversity, but not the large difference in species
cano in Hawaii (4700 m elevation), there is one vascular diversity between Brazil and the United States.
plant species present, a grass, with only a few individ- Disturbance might also increase diversity through
uals in an entire hectare. The climate, with very cold air its evolutionary effects. For example, speciation rates
and little precipitation, means that the soil is poorly might have been greater in the Tropics because of the
developed. Not surprisingly, there are few organisms of repeated contractions and expansions of the tropical
any kind in this environment, while plant species rich- rainforest during repeated glaciations. Geographic iso-
ness is very high on the lower slopes of the mountain, lation, which would have been heightened by these
especially on the wet side of the island (see Figure 18.11). shifts, is known to promote speciation (see Chapter 6).
An analogous argument explains why species However, both temperate and tropical species should
diversity is so low in cold, arid Patagonia, Argentina, have been subjected to similar kinds of isolation dur-
in southernmost South America, and so much greater in ing recurrent periods of glaciation and warming. The
the Amazon basin near the equator, and why diversity distribution of species in northern boreal and polar
is lower in the Canadian tundra than in the rainforests regions show evidence for this pattern of speciation (Fig-
of Central America. Unfortunately, this simple explana- ure 20.4). In Chapter 21, we will look at general patterns
tion does little to explain the rest of the latitudinal (or of speciation and diversity as affected by long-term cli-
elevational) diversity gradient. More complex explana- mate changes.
tions are needed. An old and influential equilibrium explanation for
One major nonequilibrium hypothesis for the the latitudinal diversity gradient is the available energy,
increase in diversity toward the equator is that com- or productivity, hypothesis (Preston 1962). The notion is
munities in the Tropics are older than those closer to the simple: increasing productivity means that more ener-
poles. If communities in the Tropics have remained gy is available for growth and reproduction and that
intact for a longer period of time, then the species in more individuals can therefore live in the same amount
those communities should have had more opportunity of area (see Chapter 14). Because a sustainable popula-
to diversify into many species. Given enough time, it has tion requires some minimum number of individuals (see
been proposed, there might be more speciation in tem- Chapter 7), more individuals in an area means that they
perate and polar communities, or more migration of can be divided up among more species, resulting in
species from low to high latitudes, and the latitudinal higher species richness.
gradient might slowly disappear. A related idea is that On an evolutionary time scale, larger populations
tropical communities have experienced fewer extinc- reduce the chance of extinction, again resulting in high-
tions because they have experienced fewer climatic fluc- er species richness. Analysis of the latitudinal pattern
tuations, and in particular have never been glaciated. shown in Figure 20.3 revealed that productivity differ-
Explanations such as these that invoke time and cli- ences were associated with about 25% of the total vari-
matic stability have a number of serious weaknesses. For ation in species richness and accounted for most of the
example, there is much evidence that there has been suf- latitudinal pattern (Scheiner and Rey Benayas 1994).
ficient time for plant species to migrate into temperate However, while productivity is an important explana-
and polar regions (see Chapter 21). In addition, a great tion—indeed, it may be the most important single fac-
deal of evidence makes it clear that tropical communi- tor in the latitudinal gradient—it is certainly not the only
ties were strongly affected by recent glaciations, even factor causing the gradient. The effects of productivity
though they were not covered by glaciers themselves. on diversity also vary with scale, as we will see later in
Similarly, the existence of elevational gradients, for this chapter.
410 Chapter 20
Figure 20.4
Lathyrus maritimus complex
Distribution of the Lathyrus
maritimus (Fabaceae) species Essentially glabrous specimens
complex around the Northern (leaves are hairless)
Hemisphere polar region. Pubescent specimens
Populations of this plant in (leaf hairs present)
different regions differ mor- Pubescence unknown
phologically, particularly in
whether and where leaves
have hairs. The different mor-
phological types may be dif-
ferent species or incipient
species. The circumpolar dis-
tribution of this species com-
plex is typical of many polar
and boreal species and genera.
A simple and currently controversial equilibrium and Sandlin 1997; Rohde 1998). While the two ecologists
theory is often called the area hypothesis. It proposes agree that area has some effect, they have been unable
that the Tropics are more species-rich because the total to agree on how important the area effect is relative to
tropical land area is greater than the total temperate land other factors such as productivity. While it is relatively
area (Rosenzweig 1995). This hypothesis contains both easy to identify which factors are important in patterns
ecological and evolutionary components. Ecologically, of diversity, assigning exact levels of importance is much
one can reason that as area increases, spatial hetero- more difficult.
geneity also increases. Because the Tropics encompass-
es more area, it contains more habitat types, and there- Continental Differences
fore more species. Furthermore, over evolutionary time,
a larger area affords greater opportunities for geographic If we compare similar regions in eastern Asia, North
isolation, and therefore promotes greater rates of America, and Europe, we find that eastern Asia has the
allopatric speciation. Michael Rosenzweig and Klaus most plant species, while Europe has the fewest. This pat-
Rohde have conducted a vigorous debate over this issue tern holds over a range of taxa. For instance, if we look
(Rohde 1992; Rosenzweig 1995; Rohde 1997; Rosenzweig at the number of tree species across all plant families,
eastern Asia has the greatest number of species, North of the mountains to escape from the cold, and went
America has an intermediate number, and Europe has extinct. As the glaciers retreated, the species in North
the fewest. These patterns are also robust across higher America and Asia that had retreated southward migrat-
taxonomic levels: the same relationships hold for orders, ed northward again. But in Europe, those species had
families, and genera of trees (Latham and Ricklefs 1993). disappeared and were not available to recolonize the
What can explain these patterns? The first explana- newly ice-free temperate areas. The end result of these
tion relies on differences in species extinctions. A key dif- events, repeated during each of the Pleistocene glacia-
ference between Europe and the other two regions has tion and warming periods, is the greatly reduced species
to do with an accident of topography. The Alps, the main diversity of Europe (Table 20.2). North-temperate and
mountain range in Europe, runs in an east-west direc- boreal communities in North America contain about
tion, while most or all of the principal ranges in eastern 50% more vascular plant species than equivalent Euro-
Asia and North America run north-south. During peri- pean ones (approximately 18,000 versus 12,000: Niemelä
ods of glaciation, climate zones are pushed southward. and Mattson 1996). Fossil evidence supports this expla-
At the height of the most recent glaciation in North nation: European forests before the Pleistocene had
America, for example, the edge of the ice sheet was in much greater species diversity than they do now (Lath-
Wisconsin and boreal forests extended as far south as am and Ricklefs 1993). This explanation depends on his-
the Gulf Coast (see Chapter 21). As climate zones moved torical contingency: the pattern of diversity is the result
southward, the plants followed them, and the southern of a particular sequence of historical events in a partic-
portions of the United States provided refugia for many ular place, not of a general or fundamental process that
species. explains all such patterns.
In Europe, however, the Alps created a barrier to this The difference between the floras of North America
southward migration. Species that were adapted to (particularly the eastern part of the continent) and east-
warmer conditions were unable to migrate over the tops ern Asia is also thought to be the result of a particular
history, but one operating on a longer time scale and
depending on rates of evolution—the rise of orders, fam-
ilies, and species—rather than on extinctions (Latham
Table 20.2 The number of woody species of various and Ricklefs 1993; Qian and Ricklefs 1999). China and
genera that went extinct in Europe during the the United States are approximately the same size and
Pleistocene but are still found in contain a similar range of ecological conditions. Yet
North America China contains 60% more species of vascular plants
Genera Number of species (29,188 vs. 17,997). Much of this difference is caused by
the fact that plant families in China tend to have more
Gymnosperms
species than the same families in the United States.
Chamaecyparis 3
Hong Qian and Robert Ricklefs (1999) hypothesize
Sequoia 1
that these differences in diversity are largely a result of
Taxodium 2
continental-scale geographic relationships. The south-
Thuja 2
ern United States is separated from tropical regions by
Torreya 2
water to the east and south and by very dry conditions
Tsuga 4
to the west. In contrast, southern China is directly con-
Angiosperms nected with tropical and subtropical regions. The tem-
Asimina 3 perate regions, Qian and Ricklefs hypothesize, were
Carya 11 repeatedly invaded by species from adjacent subtropi-
Diospyros 2 cal and tropical areas. These species then underwent
Lindera 2 extensive adaptive radiation in the temperate regions.
Liquidambar 1 What makes this argument especially compelling is the
Liriodendron 1 finding that the greatest differences in diversity between
Magnolia 8 the two continents are found in older taxonomic groups
Morus 2 with strong tropical affinities. Europe also suffers from
Nyssa 3 the same limitation as North America in this regard
Persea 1 because the Mediterranean Sea and the arid expanse of
Robinia 4 North Africa create a barrier between it and tropical and
Sabal 3 subtropical floras. Thus, a happenstance of geography
Sapindus 2 appears to have been largely responsible for differences
Sassafras 1 in species evolution over very long time periods as well
Source: Niemelä and Mattson 1996. as extinction rates over a shorter ecological time scale.
412 Chapter 20
Other Geographic Patterns al continental climates. This seasonality allows for the
coexistence of species with different life history patterns.
Species Diversity and Patterns of Overlap In many temperate deciduous forests, for example, there
There is often an increase in diversity in transition zones, is a suite of plants that are active only in early spring.
areas of contact between communities, regions, or bio- During this time, it is warm enough for plants to grow
mes. This increase in diversity occurs because the tran- close to the ground, but too cool for growth at greater
sition zone contains species from the areas on either side heights. Consequently, herbaceous plants in the under-
(Shmida and Wilson 1985). For example, the mid-latitude story can use this period for growth before the trees leaf
peak in terrestrial plant species richness in North Amer- out and cut off their access to light. Seasonality creates
ica (see Figure 20.1) is associated with the area around this temporal niche. Note that this argument is analogous
the Great Lakes in the biogeographic region called the to the argument that tropical diversity is high due to spa-
hemlock-white pine-northern hardwoods. This region is tial heterogeneity; in this case, however, the heterogene-
a transition zone between the boreal forest to the north ity creates opportunities in time, rather than across space.
and the eastern deciduous forest to
the south.
A similar pattern is found in the
Iberian Peninsula, where there are two (A)
transition zones (Figure 20.5A). One
transition zone is found along the
Atlantic coast and the Pyrenees moun-
tain range, where the Eurosiberian 6°W 0°
flora abuts that of the Mediterranean
region. The other is found in the
southeast, in the region adjacent to
Africa.These two areas have more
species than others in Iberia because
species ranges tend to overlap there
(Figure 20.5B).
Within the boreal forest of North
American there is also a longitudinal 40°N 40°N
pattern of species richness, as men-
tioned above (Qian et al. 1998) (see
Figure 20.3). The differences in species
composition found along this longi-
tudinal transect are apparently relat- Deciduous forest
ed to migration following the most Coniferous forest
recent glaciation (see Chapter 21). The Evergreen woodland
higher species richness in the middle and chaparral
of the continent may be due to an Tall shrubland
overlap of species ranges, caused by Dwarf shrubland
the different pathways from the east 6°W 0° Perennial grassland
and west taken by plants expanding Annual grassland
their ranges northward from refuges (B)
in the southeast and southwest after
250
the glacial retreat. This explanation is
a longitudinal version of the bound- Figure 20.5
200 (A) Map of Iberia, showing
ed range hypothesis.
Number of species
location of the two transition

While the transition zone effect zones (areas along green
150
provides part of an explanation for lines). Each point is the cen-
the mid-latitude peak in species rich- ter of a large sampled land-
100
ness in North America, an additional scape. (B) Landscapes in the
explanation can be found by looking transition zones contain more
50 species (mean ± 1 standard
at patterns of seasonality. The mid-lat- error) than those elsewhere
itude landscapes with the greatest 0 in Iberia. (Data from Rey
species richness have highly season- Transition zone Nontransition zone Benayas and Scheiner 2003.)
Figure 20.6 tures—and west—because of decreasing pre-
Tree species richness as measured in 2.5° × 2.5° quadrats across North cipitation (Figure 20.6). This pattern contrasts
America. The contour lines connect points with the same approximate
number of species per quadrat. (After Currie and Paquin 1987.)
with the pattern for the entire flora, which
shows a peak around the Great Lakes (see Fig-
ure 20.1). The mid-latitude peak does not exist
10
for trees, presumably because seasonal niche
partitioning occurs among species with dif-
ferent growth forms, and so would not be
expected to hold for trees, which belong more
or less to cies richness can differ between the
entire plant community and a single function-
20
0 al group.
30 0 Endemism, Centers of Diversification,

and Isolation
10
All of the patterns we have discussed so far
40 involve gradual changes in species diversity
40 over large spatial scales. Diversity can also
60
vary in patchy patterns, in which some regions
30 60 have many species compared with adjacent
80
regions. Often these differences are a result of
30 80 evolutionary processes. For example, high
100 species diversity in a region may occur be-
120 140 cause a particular lineage had a large evolu-
tionary radiation in that region. Cacti, for
20 160 example, are especially diverse in central and
southern Mexico because many species originated there.
180
Geographic isolation promotes speciation, as we saw
in Chapter 6. Islands are an obvious case of such isola-
tion. Madagascar is a very large island off the south-
eastern coast of Africa that became separated from that
continent 90 million years ago. During that time, many
species evolved in isolation there. Madagascar is now one
of the most floristically diverse places on Earth, with
approximately 10,000 native plant species, about 80% of
As with global patterns, different continental pat- which are endemic (found nowhere else). Similarly, the
terns of diversity exist for different subsets of plants. For Pacific islands of Hawaii, New Zealand, and New Cale-
trees in North America, the highest number of species is donia have all been isolated for millions of years, and all
found in the southeastern United States, with steady have very high species richness and very high propor-
decreases to the north—because of decreasing tempera- tions of endemic plants (Table 20.3).
Table 20.3 Diversity of plant genera and species and percentage of endemism in the native flora on some isolated islands
and island groups
Island Area (km2) Total genera Endemic genera Total species Endemic species Endemism (%)
Cuba 114,914 1308 62 5900 2700 46
Hispaniola 77,914 1281 35 5000 1800 36
Jamaica 10,991 1150 4 3247 735 23
Puerto Rico 8,897 885 2 2809 332 12
Galápagos 7,900 250 7 701 175 25
Hawaii 16,600 253 31 970 883 91
New Zealand 268,000 393 39 1996 1618 81
New Caledonia 17,000 787 108 3256 2474 76
Source: Gentry 1986.
414 Chapter 20
Even continental areas may have endemic species. (A)
Rare endemic plants may be restricted to unusual habi-
tats, such as serpentine barrens (see Box 15A). In some
cases a species may be endemic to an area because it is
the product of a recent speciation event and has not yet
Local species richness

spread. Another possibility is that species may be pre-
vented from spreading from their place of origin. If spe-
ciation rates are high in a region and species cannot
spread, that region will become species-rich and have a
large number of endemic species
The general problem of explaining variation in spe-
ciation rates is still one of the greatest challenges in evo-
lutionary biology. Other than geographic isolation, there
is little agreement about the mechanisms of speciation.
Factors that have been hypothesized to cause high lev-
Regional species richness
els of speciation or endemism in plants include protec-
tion from herbivores, low productivity, the existence of
(B)
many local topographic barriers, habitat diversity, a low
frequency of major disturbances, and isolation. The case 150
of the astounding levels of plant diversity found in the
fynbos of the Cape region of South Africa presents many
of these issues and the difficulties in explaining them
Local species richness
(Box 20A). Such centers of diversification are also impor- 100

tant in conservation, as we will see in Chapter 22.
Relationships between Regional

and Local Diversity 50
Most studies of the causes of diversity have focused

either on local patterns of diversity (see Chapter 14) or
on regional or global patterns, with few studies seek-
ing to connect the two scales. Recently ecologists have 0 100 200 300 400
started to make this connection, although the roles of Regional species richness
large-scale and small-scale processes and how they inter-
Figure 20.7
act to determine local and regional diversity are still not (A) Two possible relationships between the size of the
well understood. regional species pool and local species richness. If local rich-
One question we can ask is whether local diversity ness depends only on the size of the regional pool, then local
is set by the number of species available in the regional richness should increase with regional richness (black line).
If local richness depends in addition on local processes, then
species pool, or whether it is determined by processes
at some point, adding species to the regional pool should not
acting locally. We can test these two hypotheses by sam- result in any more species locally (green line). (B) The rela-
pling a series of regions, measuring both the total num- tionship between regional and local species richness for 14
ber of species in each region and the number of species plant communities in Estonia, including grasslands, shrub-
in a small subset of the region, and then plotting local lands, and forests. (After Pärtel et al. 1996.)
species richness as a function of regional species rich-
ness (Figure 20.7A). If local species richness is deter-
mined by the size of the regional species pool, then we
should see a straight-line relationship, in which a larger no leveling off of the relationship at the highest region-
regional species pool leads to more species locally. On al diversity (Figure 20.7B). As of yet, few other studies
the other hand, if local processes, such as competition, have quantified this relationship, and a general answer
limit the local number of species, the relationship should to this question awaits more research.
be concave; at some point, adding more species to the Some ecologists are attempting to examine the
regional pool should have little or no effect on local mechanisms that link local and regional diversity.
species richness. In an analysis of plant communities in Jonathan Shurin and Emily Allen (2001) modeled the
Estonia, Meelis Pärtel and colleagues (1996) found that conditions under which local species interactions (com-
local diversity was a function of regional diversity, with petition and predation) contrast with regional patterns
in determining species richness at both local and region- dependent on the regional context and on the abilities
al scales. Their model included both deterministic, equi- of species to disperse among sites (see Chapter 17). Thus,
librium processes at the local scale and nonequilibrium, there is a reciprocal relationship between local and
stochastic processes at the regional scale. They found regional scales: feedbacks between processes at the two
that predation expands the range of environments in scales result in interactions such as predation leading to
which competitors can co-exist at a regional level, pro- increased diversity at both scales. As with any mecha-
moting higher levels of regional diversity by preventing nism shown to be plausible through modeling, the cur-
competitive exclusion in some locations (see Chapter 11). rent challenge is to determine whether this mechanism
The effect of these interactions on local diversity was is important in nature. Currently, little consensus exists
Table 20.4 Theories explaining the hump-shaped productivity-diversity relationship observed at regional scales
Hypothesis Mechanism References
Evolutionary theories
Available habitat More species have evolved to tolerate more common habitats, and the most common 1, 2
habitats are of intermediate productivity.
Speciation rates and productivity The difference between rates of speciation and extinction is maximal at intermediate 3
productivities.
Rate of evolution The rate of evolution is maximal at intermediate productivities. 3
Adaptive trade-offs More species can evolve tolerance to intermediate conditions. 3
Ecological theories
Competition for resources and Species compete for resources, and more species coexist where resource heterogeneity 4,5
resource heterogeneity is greater. Effective resource heterogeneity first increases with productivity and then
decreases due to light limitation.
Transport-limited competition Consumption of nutrients by plants and sufficiently high nutrient supply: nutrient 6
diffusion ratios generate heterogeneity in soil resources and allow the coexistence of
competing species. Increased soil fertility allows for the coexistence of more species
until light limitation reduces species richness.
Predation (herbivory) and Species compete for resources, and there is a trade-off between competitive ability and 7
resource competition vulnerability to predators. Predator density increases with productivity, resulting in
a shift to more predator-resistant taxa at higher productivities. Spatial or temporal
variation in resources allows coexistence of multiple species at low to moderate pro-
ductivities, but at high productivities only the most predator-resistant species persist.
Intertaxonomic competition A given taxon can exploit only a portion of the productivity range and is then replaced 2, 8, 9
by a competing taxon. Species richness within a taxon increases and then decreases
with productivity.
Dynamical instability Increased productivity reduces stability in multitrophic-level interactions. 10, 11
Disturbance and competitive Population growth rates are positively correlated with productivity. Disturbance acts 12, 13
exclusion to reduce population density and slow the time to competitive exclusion. Therefore, at
low productivities, species diversity is low because most species’ growth rates are
not high enough for them to tolerate disturbance. At high productivities, species
diversity is low because of competitive exclusion. Species richness is thus maximal at
intermediate productivities.
“Hump-backed model,” Plant species diversity is low in low-productivity habitats due to environmental stress. 14
including environmental In high-productivity habitats, dominant species competitively exclude other species,
stress, competitive exclusion, and species diversity is low. At moderate productivities, species density is high because
and disturbance moderate levels of stress and disturbance suppress competitive dominants.
Predator (herbivore)-victim As productivity increases, predators absorb much more than a proportional share and 8, 15
ratios reduce the diversity of consumers.
Temporal covariance Temporal variation in resources allows the coexistence of competing species. 8, 16
If temporal variation correlates with productivity, then a hump-shaped productivity-
diversity relationship may result.
Note: General discussions of these theories can be found in Tilman and Pacala 1993; Huston 1994; Abrams 1995; Rosenzweig 1995; and Leibold 1998.
References: 1, Denslow 1980; 2, Rosenzweig and Abramsky 1993; 3, VanderMeulen et al. 2001; 4, Tilman 1982, 1988; 5, Abrams 1988; 6, Huston and DeAngelis
1994; 7, Leibold 1999; 8, Rosenzweig 1995; 9, Tilman and Pacala 1993; 10, Rosenzweig 1971; 11, Wollkind 1976; 12, Huston and Smith 1987; 13, Huston 1979,
1994; 14, Grime 1973, 1979; 15, Oksanen et al. 1981; 16, Chesson and Huntley 1988.
416 Chapter 20
BOX 20A
The Fynbos and the Cape Region of Africa
You may never have heard of the fyn-
bos, but it deserves to be far better
known. The fynbos is a small region in
the southernmost part of Africa that has
some of the highest levels of plant
diversity on Earth. Why should anyone
care about this tiny area with its weird
plants and its animals with odd names
such as grysbok, klipspringer, and Cape
dainty frog? We should care because,
among other reasons, it is a priceless
resource for advancing our under-
standing of biodiversity, evolution, and
ecology.
The vegetation of the fynbos is so
distinctive that a few scientists have ele-
vated it to the status of a unique biome.
Its distinctiveness was recognized in
the early twentieth century by the Ger-
man botanist Adolph Engler, who clas-
sified it (including the surrounding A lowland fynbos landscape. The fynbos, a region of about 46,000 km2 (about the size of the states
of Maryland and Massachusetts together), boasts a mediterranean climate and landscapes that
plant communities in the Cape region range from the heathlike lands shown above to craggy montane regions. The vegetation is highly
of Africa; see Figure 22.14) as the small- endemic, with many species found only in this tiny region. (Photograph courtesy of E. Orians.)
est of the world’s six great floristic divi-
sions.* The fynbos is home to about
8000 species, some 5000 of which are phy is diverse, with sandstone moun- and the surrounding Cape region, with
endemic. It has an extremely high contains, steep, dissected valleys, limestone wonderful names such as the Succulent
centration of plant species, with 1300 outcrops, shale flats, and coastal plains. Karoo and Nama Karoo, the Mountain
species per square kilometer and up to Fynbos is defined by the presence of Fynbos, and Renosterveld (small leafed
121 species per square meter. The latter plants in the Restionaceae (Cape reeds). shrublands). Animals are far less
statistic points to a truly remarkable Restios are somewhat grasslike in form diverse in the Cape region than are
feature of the fynbos: its high diversity and replace grasses in the fynbos. The plants, although there are rare and
at small to medium spatial scales. Ericaceae (the heaths—generally low- endemic species in a number of animal
Derived from the Dutch words fijn growing, shrubby plants typically taxa, including fishes, reptiles, mam-
bosch (“fine bush”), fynbos (pronounced found on acid soils) is another impor- mals, birds, and many insect groups.
roughly “fane boss”) is named for the tant fynbos plant family; the genus Most dramatic of all may be the highly
narrow-leaved, grasslike plants and Erica reaches a peak of diversity here, specialized animal-plant interactions on
sclerophyllous evergreen shrubs with with about 600 species, whereas there which many fynbos plant species
fine, small or rolled leaves that are are fewer than 30 species in the entire depend for reproduction and seed dis-
dominant there (see Figure 19.16). It has rest of the world. Other characteristic persal. Protea acaulis and many other
a mediterranean climate, with cool, wet plants of the fynbos are members of the proteas are totally dependent on mice
winters and hot, dry summers. Like Proteaceae and the Asteraceae; Aster- for pollination. Their flowers are held
other mediterranean vegetation, the aceae is represented by over 1000 fyn- near the ground and produce a strong
fynbos is adapted to frequent fires bos species, more than half of them odor that mimics yeast. Mice drink the
(which occur predominantly in late endemic. There are a large number of stinky nectar, carrying pollen on their
summer). The soils are infertile, acidic, plants in the Iridaceae with particular- noses to the next plant. Ants are very
sandy, and very ancient. The topogra- ly beautiful flowers, including many important seed dispersers for many
that are grown horticulturally. Unlike fynbos plants, such as those in the
*The six floral kingdoms categorized by Engler are vegetation in most other places, the fyn- genus Callistemon (bottlebrush, Myr-
the Cape, Antarctic, Australian, Neotropical (West-
bos typically has no one dominant taceae). The seeds of these plants are
ern Hemisphere tropical regions), Paleotropical
(African and Asian tropical regions), and Holarctic species or group of species. attached to special structures called
(the entire Northern Hemisphere north of the Various distinctive local plant com- elaiosomes, which offer a food reward
Tropics). munities occur as part of the fynbos to the ants for carrying the seeds off.
(A)
The great diversity of the fynbos is a

geologically recent phenomenon. Until
some 5 million years ago, the Cape
region was covered with subtropical
forests. As the climate became cooler
and drier, mediterranean shrublands
spread into the Cape from the north.
Much of the diversification of the flora
of the fynbos occurred during the large
climatic fluctuations of the past 2 mil-
lion years, as glacial periods alternat-
ed with warm periods.
Why is this tiny region home to so
many plant species? How did they
evolve, and how do they coexist? Why
is the fynbos so rich in species when
most other mediterranean shrublands
are low in plant diversity? The poor,
acidic soils are hypothesized to be an
important factor in the evolution of this
astounding diversity—but the poor,
acidic soils of the pine barrens of North
America are supposed to explain the
very low plant species diversity there! (A) Endemic fynbos species include the (B)
The soils of the fynbos are very old, mutualistic species Protea magnifica
in contrast to the very young soils of the (queen protea, Proteaceae) and its pollina-
pine barrens, but soil age is not reliably tor, Promerops cafer, the Cape sugarbird.
(B) This tiny Protea is one of the ground-
correlated with diversity. Another growing species dependent upon mice for
hypothesis is that the highly dissected pollination. (Photographs courtesy of C.
and varied terrain of the Cape region Oertel.)
has created many ecological opportu-
nities for specialization. Evolution of
ecological specialization can occur over thropogenic changes in fire regimes
short distances if selection is sufficient- threaten species throughout the fynbos.
ly strong, even if some gene flow still Most threatening to the fynbos flora,
occurs. Whether this phenomenon however, is the invasion of exotic
accounts for part of the diversity of the species. The silky hakea, Hakea sericea, is tralia. An invasive, highly aggressive
fynbos flora is unknown. Future study a fire-adapted member of the Proteaceae ant species from South America,
of the fynbos may hold the key to many from southeastern Australia that was Iridomyrmex humilis (the Argentine ant),
mysteries critical to our understanding brought to South Africa as a hedge plant is eliminating the native ant species on
of plant evolution. and for firewood around 1830. It soon which many fynbos plants depend for
Tragically, the fynbos faces serious escaped cultivation and has become a seed dispersal, threatening the regener-
threats. Some 1700 plant species are major threat to fynbos vegetation. ation of many plant species.
threatened with extinction in the Cape Clumps of this shrub or small tree Finally, global climate change is
region. Many fynbos species are highly develop into dense, impenetrable thick- especially threatening to the survival of
localized, and there are also many very ets that eliminate native species, alter highly restricted ecosystems such as
local centers of species diversification fire regimes, and deplete soil water those of the Cape region and to species
and endemism. The city of Cape Town reserves. Other devastating invaders with very limited distributions. The
is located on two such centers of include Pinus radiata (Monterey or radi- fynbos is located at the very southern
endemism, and its expansion threatens ata pine, Pinaceae; see Chapter 14) from edge of the African continent; if the cli-
several hundred species. Agriculture coastal California, other pine species mate warms and species are driven
and the spread of fertilized commercial from Europe, and Acacia cyclops poleward, there is nowhere for them to
crops threatens lowland fynbos. An- (Fabaceae) and Acacia saligna from Aus- go but the salty deep.
418 Chapter 20
80 Local (<20 km) 80 Landscape (20–200 km) Figure 20.8
Percentage of studies showing vari-
n = 11 n = 11
60 60 ous relationships (humped, U-
shaped, negative, positive) between
40 40 productivity and vascular plant
species richness at different geo-
20 20 graphic scales. Geographic scale is
Percent of studies
based on the farthest distance

0 0 between sites in a data set. Sample
sizes (n) refer to the number of data
80 Regional (200–4000 km) 80 Continental to global (>4000 km) sets. (After Mittelbach et al. 2001.)
n = 27 n = 10
60 60
40 40
20 20
0 0
Humped
U-shape
Positive
Negative
No correlation
Humped
U-shape
Positive
Negative
No correlation
Productivity-diversity pattern Productivity-diversity pattern
on the relative importance of various mechanisms in dif-

ferent systems, or the ways in which they may commonly
interact (Caswell and Cohen 1993; Lawton 1999).
Northern provinces
Conifer-hardwood forest
Productivity and Scale
Pine savanna
We can also look at patterns of diversity in relationship Boreal forest
to environmental factors. Ecologists agree that produc- Southern provinces
tivity is an important determinant of species richness. Southern hardwood forest
Beyond that agreement, however, lies a much murkier Oak savanna
realm, consisting of a myriad of processes operating at Prairie
a variety of scales. The relationship between species rich-
ness and productivity is complex; it can be positive, neg-
ative, hump-shaped, or U-shaped (see Figure 14.7). To
make matters even more complicated, the shape of the
relationship often varies with scale (Figure 20.8). Many
theories have been proposed to explain these patterns
(Table 20.4).
The effects of scale on productivity-diversity rela-
tionships were recently explored
by Scheiner and Jones (unpub-
lished), using data on the vege-
tation of Wisconsin originally col-
lected by John Curtis and his
students in the 1950s (Curtis 1959;
Figure 20.9). For the state as a
whole, the relationship was U-
shaped (Figure 20.10), with the
lowest diversity at intermediate Figure 20.9
productivities—a rather unusual The state of Wisconsin, showing the ecological divisions and
John Curtis pattern (see Figures 14.7 and provinces defined by Curtis. (After Curtis 1959).
(A)
140
120 0.0
100
Number of species
–10.0
Regression curvature
80
–20.0
60
– 30.0
40
– 40.0
20
– 50.0
0
400 500 600 700 800 900 7.5 20 30 40 60 80 120 160 240 Entire
Productivity (g C/m2/yr) state
Extent (km)
Figure 20.10
The relationship between terrestrial vascular plant species
richness and net primary productivity for 901 plots across (B)
the state of Wisconsin. (Data from Scheiner and Jones, un-
published.) 1.0
Regression curvature 0.0
20.8). The relationship is statistically significant despite –1.0

the obviously large amount of variation around the
regression line. The variation is due to a combination of –2.0
error in estimating productivity as well as other, unac-
counted for factors that also affect diversity. Such varia- – 3.0
tion from multiple causes is common with ecological data.
One way to examine the effects of scale is by con- – 4.0
sidering how the relationship between productivity and
1 2 3 4 8 16 24 32 48 64
diversity changes when different components of scale
Aggregated number of sites
are varied (see Chapter 17 for a discussion of scale and
its components). First, Scheiner and Jones grouped their Figure 20.11
data points so as to vary geographic extent (the total area Effects of changing (A) the extent and (B) the grain of the
examined). If very small areas (7.5 × 7.5 km2) were sam- data on the relationship between plant species richness and
productivity across the state of Wisconsin. The graph shows
pled, the relationship between productivity and diver- the average amount of regression curvature for 1000 ran-
sity was hump-shaped—that is, diversity was highest at domly chosen samples. Negative curvature indicates a
intermediate levels of productivity (Figure 20.11A). hump-shaped relationship; positive curvature indicates a U-
Increasing the area sampled changed this relationship. shaped relationship. Error bars show the 95% confidence
At intermediate areas, there was a negative relationship interval. (Data from Scheiner and Jones, unpublished.)
between productivity and diversity, while at the largest
area—the entire state—the relationship was U-shaped.
A second analysis examined the effects of chang-
ing grain size (the size of the sampling unit represent- depends on the scale of the analysis, with different com-
ed by each data point) (Figure 20.11B). Grain size was ponents of scale having different effects.
manipulated by combining data from adjacent sites, A critical factor responsible for the effects of scale on
beginning with two sites, then three sites, and so forth. diversity-productivity patterns in Wisconsin is that the
At the smallest grain sizes the relationship between pro- state sits at the intersection of three major biomes, with
ductivity and diversity was U-shaped; at intermediate eastern deciduous forest to the southeast, hemlock-white
grain sizes the relationship became negative, while at pine-northern hardwood forest to the east, and tallgrass
the largest grain size the relationship was hump-shaped prairie to the west. Nonlinear (hump-shaped or U-
(Figure 20.11B). What these analyses show is that even shaped) patterns are more often associated with scales
for a single set of sites there is no unique relationship that span more than one community type or biome (see
between productivity and diversity. The relationship Figure 14.8), which is what was found for these data.
420 Chapter 20
It is clear, however, that it is not the sole explanation.
Summary
Many different hypotheses have been proposed to
The explanation of large-scale patterns of plant diversi- explain these patterns. Current research is attempting to
ty—which regions have many species and which have penetrate this thicket of ideas. While we have been able
few—has been an ongoing effort of ecologists since to discard some hypotheses, others remain.
before ecology developed as a distinct discipline. Today One likely explanation for the current lack of reso-
these issues have taken on a new urgency as global lution on this problem is that no single process accounts
change and habitat destruction have increased the need for all of the observed patterns. Rather, different process-
to preserve areas of high diversity. We need to know es are important in different circumstances and at dif-
why species are where they are if we are to predict the ferent scales. Other important processes include the eco-
effects of global change and preserve as many species as logical and evolutionary histories of particular regions,
possible. and interactions between local and regional processes
One of the oldest known and best established pat- that determine diversity.
tern is a latitudinal gradient in diversity: highest in the Our job now is to try to find rules defining those cir-
Tropics, intermediate in temperate zones, and low in cumstances and to determine when and how two or
polar regions. Similar gradients are found from low to more processes may function together. Investigations of
high elevations. There are also longitudinal gradients, how patterns of diversity change as a function of scale
such as the peak in diversity in the middle of Canada. is one important piece of this puzzle. The rigorous eval-
The effects of productivity—greater productivity uation of the hypotheses presented in this chapter is an
allows the coexistance of more species—is likely the pri- important first step in all of these efforts.
mary cause for large-scale patterns of species diversity.
Additional Readings
Classic References Mittelbach, G. G., C. F. Steiner, S. M. Scheiner, K. L. Gross, H. L. Reynolds,
R. B. Waide, M. R. Willig, S. I. Dodson and L. Gough 2001. What is the
Curtis, J. T. 1959. The Vegetation of Wisconsin. University of Wisconsin
observed relationship between species richness and productivity? Ecolo-
Press, Madison.
gy 82: 2381–2396.
MacArthur, R. H. 1965. Patterns of species diversity. Biol. Rev. 40: 510–533.
Pianka, E. R. 1966. Latitudinal gradients in species diversity: A review of Additional Resources
concepts. Am. Nat. 100: 65–75.
Huston, M.A. 1994. Biological Diversity: The Coexistence of Species in
Changing Landscapes. Cambridge University Press, Cambridge.
Contemporary Research Rosenzweig, M. L. 1995. Species Diversity in Space and Time. Cambridge
Colwell, R. K. and G. C. Hurtt. 1994. Nonbiological gradients in species University Press, Cambridge.
richness and a spurious Rapoport effect. Am. Nat. 144: 570–595.
Ricklefs, R. E. and D. Schluter (eds.). 1993. Species Diversity in Ecological
Currie, D. J. and V. Paquin. 1987. Large-scale biogeographical patterns Communities: Historical and Geographical Perspectives. University of
of species richness of trees. Nature 329: 326–327. Chicago Press, Chicago, IL.
21
C H A P T E R Paleoecology
I
n previous chapters we have looked at global patterns of vegetation and
diversity and at how those patterns have resulted from ecological and evo-
lutionary processes. But the world is constantly changing, and the picture
presented in those chapters is just a “snapshot” of today’s world. The world
was a very different place at various times in the past. Today, if you were to
stand at a spot in rural northern Illinois, you would see cornfields all around
you. Just 150 years ago, the scene would have been one of rolling prairie. Ear-
lier—20,000 years ago—you would have been standing in arctic tundra, with
sunlight glinting off the massive glaciers to your north. Go even farther back,
and you might have been standing in the middle of a dense forest, or in a
swamp dominated by tree ferns.
In this chapter, we consider some of these long-term changes in plant com-
munities, the scientific methods used for understanding them, and some of
their implications. Knowledge of the past both informs our understanding of
the present and provides critical data for the management of today’s ecosys-
tems (Swetnam et al. 1999). We cannot, of course, cover the entire history of
Earth, or even of terrestrial plants, in one chapter. The ecology of many earli-
er periods is still poorly understood. The field of paleoecology—the study of
historical ecology—depends heavily on inferences and analogies with current
species and communities. Our understanding of the very distant past is espe-
cially tenuous because of a dearth of fossils from those times, and because many
ancient species have no modern analogues. Our knowledge of the past, how-
ever, is increasing rapidly. The development of new methods, such as the use
of carbon isotope signatures of C3 and C4 plants and the creation of extensive
computerized databases of the fossil record, have made paleoecology an active
and lively field of research.
In this chapter we first survey several geological eras and periods (Figure
21.1), especially those that represent significant times of change for plant com-
munities. Then we look in detail at the past 20,000 years. Because of the amount
of detail involved, we focus on North America. The history of this recent peri-
od is now well known and provides special insight into current plant com-
munities and biomes. And what of the future? That is the subject of the next
and final chapter of this book.
Figure 21.1 Eon Era Period Traditional Epoch New
The geological time scale (numbers period period
Cenozoic
are in millions of years). Although designation designation
the Cenozoic era was traditionally 24
.01
split into the Tertiary and Quater- Holocene(recent)
Quaternary
nary periods, a newer system now 65 Pleistocene
1.8
divides it into two periods of more Pliocene
Cretaceous 5 Neogene
equal duration, the Paleogene and
Mesozoic
the Neogene. (After Stanley 1987.) 145 Miocene
Jurassic 24 24
213 Tertiary Oligocene
Triassic 34
248
Phanerozoic
Permian
Paleogene
286 Eocene
Carbon-
Pennsylvanian
iferous
325
Missisipian 56
Paleocene
360 65
Devonian
Paleozoic
410
Silurian
440
Ordovician
505
Cambrian
544
Precambrian
pressure, that their gametophytes produced both male

The Paleozoic Era and female reproductive organs like those of modern
Fossil evidence tells us that plants first invaded land in bryophytes and ferns, and that they had, at most, only
the Paleozoic era, during the late Ordovician through rudimentary roots (Bateman et al. 1998). The forms of
Silurian periods. In Chapter 3 we considered some of the their sporophytes were simple, but at least one species,
evolutionary challenges they faced in accomplishing this
major transition. Land plants are descended from the
Charophyta, a group of branched, filamentous green
algae. Today many green algae grow along seashores,
where they are subjected to daily drying as the tides rise
and fall. Undoubtedly these were the conditions under
which land plants evolved.
The invasion of plants drastically changed terrestrial
environments, setting the stage for two major (and sep-
arate) animal invasions (by arthropods and by verte-
brates) as well as for all subsequent plant evolution.
During the Silurian and Devonian periods, the ter-
restrial landscape was completely transformed. This
period of time saw a steady increase in the morpholog-
ical and ecological complexity of both plants and ani-
mals and the consequent dramatic radiation of major
phyla. At the beginning of this period, the only vegeta-
tion consisted of ground-hugging mats in very damp Figure 21.2
Marchantia is a liverwort, a nonvascular plant. The 3-centimeter
areas, dominated by plants related to liverworts (Figure tall structures in the center are reproductive organs of the female
21.2). Based on fossil evidence, we can infer that these (left) and male (right) gametophytes. Early terrestrial plants
early plants were mechanically supported by turgor may have had a similar form. (Photograph by S. Scheiner.)
PAGE PROOF: 2ND PASS Paleoecology 421
ity probably triggered an “arms

race.” There is generally little
advantage to a plant in becoming
very tall unless it is likely to be
shaded by its neighbors (see Fig-
ure 6.6). If one species in a com-
munity is taller than all of the
others, it will outcompete its
neighbors for light. Natural selec-
tion will then favor individuals of
those other species that can grow
even taller. This process contin-
ues until other constraints bal-
ance selection for height. Selec-
tion for increased height, as well
as other selective forces, result-
ed in the evolution of wood,
which in turn enabled the great
diversity of forms of trees and
shrubs to evolve. The evolution
Figure 21.3 of roots, in contrast, was proba-
A forest in what is now Colorado as it might have looked 300 bly driven by a combination of abiotic factors as well as
million years ago. A shallow sea lies to the east, and rivers competition for nutrients and water. As a result of these
choked with gravel flow from the mountains to the sea. The adaptations, plants were able to spread from wet areas,
banks of the rivers are colonized by forests of scale trees (Sig- eventually spreading over nearly the entire surface of
illaria, Lycopodophyta) and giant horsetails. The trees form a
very minimal canopy with little shade. They have green Earth that is not covered by water or ice.
trunks, and even their surface roots are green; the whole tree As plants diversified, biotic interactions probably
can photosynthesize. Forests of primitive conifers grow in the came to play an increasingly important role in plant evo-
higher terrain of the foothills and mountains. (Painting by Jan lution. Many modern ecological interactions evolved
Vriesen, courtesy Denver Museum of Nature and Science.)
Rhynia gwynne-vaughanii, had already evolved the abil-

ity to spread by clonal growth.
By the end of the Devonian, soils had developed in
many habitats, and tall forests rose above primordial
swamps dominated by giant lycopods up to 40 meters
tall (Figure 21.3). As root systems evolved, land plants
diversified and spread to mesic and drier uplands. By the
end of the period, many major vascular plant phyla—
Psilotophyta (whiskferns), Lycopodophyta (clubmosses;
Figure 21.4), Sphenophyta (horsetails) and Pteridophyta
(ferns)—had all evolved, as had the seed plants (Bateman
et al. 1998). Today the first three of these groups grow
only as small understory plants. Many ferns also grow
as small plants, but tree ferns are large plants that grow
in a number of tropical and subtropical habitats. A con-
sistent trend during this period was the increasing dom-
inance of the sporophyte (diploid) generation. As the
sporophyte became larger and longer-lived, the gameto-
phyte became smaller and shorter-lived.
Figure 21.4
We can speculate about the selective processes that Lycopodium obscurum (ground pine, Lycopodiaceae) is a club-
created this diversification. The evolution of vascular moss, a vascular plant approximately 10 centimeters tall. It has
tissue allowed plants to increase in height, and this abil- a stem, but not true roots or leaves. (Photograph by S. Scheiner.)
(A) Late Permian
PA
NA PA
Tethys Sea
PA
NT ET
I AU
AN Major continental plates
PA = Palearctic
Laurasia
(B) Late Cretaceous NA = Nearctic
PA
NA
NA
Tethyan
ET Seaway
NT
I
AU Major continental plates
NT = Neotropical
AN ET = Ethiopian
I = Indian
Gondwanaland
(C) Middle Eocene AN = Antarctic
AU = Australian
M = Madagascar
NA PA
PA
ET
I
NT
M
AU
AN Mountain ranges
Lowlands
Figure 21.5 (D) Modern world Continental shelf
The positions of the continents (under water)
over geological time: (A) Late Ocean
Permian (255 million years ago),
NA PA
when the land masses were
joined together into one super-
continent, Pangaea. (B) Middle I
Cretaceous (94 million years ET
ago), when the northern conti-
nent, Laurasia, and the southern NT
continent, Gondwana, had M AU
begun to break apart. (C) Mid-
dle Eocene (50 million years
ago), when the continents were
taking on their current configu-
rations. (D) Today. AN
very early, including mycorrhizal symbioses and her- ferns are an extinct group that superficially resembled
bivory during the Devonian and animal pollination and ferns, but reproduced through seeds rather than free liv-
seed dispersal during the Carboniferous and Permian ing gametophytes.
periods. With increasing aridity and continentality—influence
The Carboniferous period is so named because of of land masses on weather, in this case due to the forma-
the enormous amounts of fossil carbon—oil and coal— tion of Pangaea—during the first half of the Mesozoic, the
that were deposited that time. Coal deposits were gen- flora became dominated by conifers, cycadophytes, and
erated mainly from the remains of wetland plants, other gymnosperms, and seed ferns gradually went
including ferns and other spore-producing plants, as extinct. Nonflowering families, genera, and species from
well as early gymnosperms (DiMichele et al. 2001). The the Mesozoic which persist today include: Osmunda (a
organic sources of oil are less well understood; they fern genus), Ginkgo biloba (ginkgo, Ginkgoaceae), Sequoia
appear to be primarily marine autotrophic and het- (a conifer genus that today includes the giant sequoia of
erotrophic plankton, but may also include wetland California), and Araucaria (a conifer genus of the South-
plants. As climates changed in the late Carboniferous— ern Hemisphere that today includes several important
mainly by becoming drier—seed plants began to dom- forest trees, such as A. excelsa, the Norfolk Island pine of
inate the landscape. This change in the vegetation was the South Pacific, and A. araucana, the monkey puzzle tree
rapid, resulting in a nearly complete change in the of the southern Andes of South America). The angio-
species present at any given location, with upland plants sperms came to dominate the terrestrial fauna during the
rapidly replacing those of wet lowlands. Cretaceous.
The early Cretaceous was a time of warm tempera-
tures and very little difference among the seasons. CO2
The Mesozoic Era levels were somewhere between 3 and 4 times higher than
they are today (Figure 21.6), contributing to warmer tem-
The Dominance of Gymnosperms peratures through the greenhouse effect (see Chapter 22).
At the beginning of the Mesozoic era, all of the conti- The warm climate and ready availability of CO2 made the
nents were joined together into one supercontinent, Pan- early Cretaceous a period generally favorable to plant
gaea. They have been carried apart and reshuffled by growth. By the late Cretaceous, however, these favor-
plate tectonics since that time (Figure 21.5). The coales- able conditions had gradually declined, with climates
cence of Pangaea, along with increases in atmospheric becoming much more seasonal and larger differences in
CO2 and temperature, led to widespread continental cli- temperature existing between the equator and the poles.
mates, greater seasonality, and widespread aridity. Any The plant communities of the Mesozoic were simi-
productivity benefits of the high CO2 levels may have lar to those of today in some ways. There were distinct
been more than counterbalanced by the combination of biomes dominated by trees, shrubs, or herbs. However,
high temperatures (from the greenhouse effect) and low the trees were predominately conifers; today, tropical
or seasonal rainfall. forests and many temperate forests are dominated by
By the late Jurassic period, 150 million years ago, angiosperms, with coniferous forests mostly found in
Pangaea had separated into two large continents, north- temperate or boreal regions. Savannas are a particular-
ern Laurasia and southern Gondwana. These continents, ly arresting example of how ancient biomes differ from
in turn, would break up into our current continents by those of today. Today’s savannas are dominated by
the end of the Cretaceous period. The Atlantic Ocean grasses along with scattered angiosperm trees. In the
was coming into being by the early Cretaceous, first with Mesozoic, there were no grasses; instead, ferns provid-
the separation of North America and Eurasia about 130 ed most of the ground cover, while the trees were large-
million years ago, and then with the separation of South ly gymnosperms.
America and Africa about 90 million years ago. As a If we consider the modern distribution of ferns, we
result of these continental movements, areas that were might think of them as a largely tropical group, as they
once inland became coastal. Coastal areas are both wet- are most abundant and diverse in the Tropics. Interest-
ter and less seasonal than inland regions (see Chapter ingly, most of today’s fern families originated in moist
18). Thus, overall conditions for plant growth tended temperate regions (above 30° N and S latitude) during
to improve throughout the Jurassic and Cretaceous. the Mesozoic (Skog 2001). During this time, ferns were
We can consider the history of land plants as the noticeably absent from the Tropics. Long-distant dis-
successive dominance of three major groups. The pteri- persal of spores allowed fern families to spread widely
dophytes (ferns) and pteridosperms (seed ferns) domi- in temperate regions, so that during much of the Meso-
nated the flora during the late Paleozoic, when climates zoic, many families were found in both the Northern and
were warm and wet. Despite the similarity in their Southern hemispheres. Only later did they spread into
names, ferns and seed ferns are not closely related; seed the Tropics.
(A) (B)
300
1800
CO2 concentration (ppm)
CO2 concentration (ppm)

1440
fpo 250
1080
200
720
360 150
500 400 300 200 100 Present 450 300 150 Present
Million years ago Thousand years ago
Figure 21.6
(A) Estimated concentrations of atmospheric CO2 over the
past 500 million years. The shaded areas above and below
the solid line are upper and lower limits of the estimate. The relationships between plants and pollinators. The pol-
black bars indicate periods when Earth's climate was rela- lination of earlier vascular plant taxa depended mainly
tively cool; white areas are periods of relative warmth. CO2
on the wind (as in conifers) or on beetles (as in cycads),
levels have mostly decreased for the past 175 million years.
(After Rothman 2002.) (B) Estimated concentrations of at- which mainly consumed pollen and only occasionally
mospheric CO2 over the past 450,000 years, measured from transferred some of it to other plants. With the evolu-
trapped air in ice cores from the Antarctic. Over this period tionary diversification of flowers came nectar rewards,
CO2 levels have varied, but there has not been a trend either and diversification in flower shapes, colors, and scents
up or down. (After Barnola et al. 1999.)
increased the reliability of insects in moving pollen to
receptive stigmas on other flowers of the same species.
Today some of the best-known examples of mutualisms
The Breakup of Pangaea are those of flowers and their pollinators. The evolution
and the Rise of the Angiosperms of fruits similarly opened new possibilities for interac-
The rise of the angiosperms was a critical event in the his- tions between plants and their animal seed dispersers
tory of plants. A few tantalizing fossils suggest an origin (and seed predators).
as early as the Triassic for this group, but that interpreta- Another important difference between angiosperms
tion is controversial and not well supported by current and the groups that preceded them is that angiosperms
evidence. What is certain is that by the middle to late Cre- include many herbaceous and small woody species that
taceous, angiosperms were spreading throughout the often have faster growth and life histories. In conifers,
world. This group now accounts for approximately 75% seed maturation can take from six months to three years,
of all land plant species and a level of plant diversity not while almost all angiosperms can mature seeds in just
seen previously on Earth. The separation of the continents weeks to months. More generally, angiosperms encom-
during the Cretaceous not only contributed to climate pass an extremely wide diversity of plant architectures,
change, but also helped to accelerate speciation rates in life histories, and reproductive modes.
both plants and animals through geographic isolation (see At the end of the Cretaceous, tropical forests were
Chapter 6). In this world of newly divided continents the found in what is now southeastern North America. These
angiosperms spread, proliferated, and diversified. Their forests had open canopies with relatively tall conifers
rise created a profound change in the structure of com- similar to Sequoia and Metasequoia (Taxodiaceae), with an
munities. First, the nature of animal communities went understory consisting of a variety of angiosperms,
through a dramatic transformation. Insects began to including rosette palms. At this time, the middle of North
diversify; today their species numbers vastly overwhelm America was covered by a vast shallow sea. The areas
those of any other group of animals. In some cases, such west of the sea were covered by a broad-leaved evergreen
as the diversification of lepidopterans (butterflies and community with a variety of species like those in the
moths), the timing of insect radiations can be linked southeast. Temperate forests, dominated by broad-leaved
directly to the evolutionary radiation of angiosperms. evergreen conifers, cycads, and ginkgos, covered what is
The evolution of the flower—a key feature of today the northwestern United States and southwest-
angiosperms—opened possibilities for more specialized ern Canada. In those forests, the angiosperms were gen-
erally deciduous species growing along streams and in the Rocky Mountains and Sierra Nevada continued ris-
other disturbed habitats. The far northwestern areas of ing, creating a mid-continent rain shadow (see Chapter
North America—the area above 60° N latitude—were 18). Along with continued mountain building in the
covered by broad-leaved deciduous forests with a mix- Himalayas, these geological changes led to a cooling in
ture of conifers (such as Taxodiaceae and Gingko) and the global climate. The cooling was strongest in the South-
angiosperms, including species related to birches (Betu- ern Hemisphere: the South Pole was covered by ice dur-
laceae) and elms (Ulmaceae). The milder climates of the ing the Oligocene epoch, but the North Pole was not. This
Cretaceous allowed these broad-leaved communities to cooling may have been reinforced by another effect. Mas-
exist where now there is taiga and tundra. sive mountain building during this period resulted in the
weathering of large amounts of rock: chemical reactions
The Cretaceous-Tertiary (K-T) Boundary during weathering remove CO2 from the atmosphere
(Ruddiman and Kutzback 1989). Temperatures remained
The Mesozoic era came to an end 65 million years ago. high enough, though, that throughout the Miocene epoch
There is strong evidence that a large asteroid (10 ± 4 kilo- the vegetation was tropical even at high latitudes.
meters in diameter) slammed into what is now the These changes in climate had important ecological
Caribbean Sea near the current Yucatán Peninsula. The consequences. By the Oligocene epoch, broad-leaved
asteroid struck from the southeast, throwing vast amounts evergreen forests in the middle of North America had
of debris into the atmosphere to the north. This debris thinned out to tropical woodlands and savannas. Grass-
eventually settled to the ground, forming a distinct geo- es, which originated in the Paleocene, began to spread
logical layer marking the transition between the Creta- widely during the Miocene. Grass eaters evolved from
ceous and Tertiary periods, known as the K-T boundary. ancestors that ate leaves from trees or shrubs. Most
How this asteroid impact led to massive global notably, horses evolved in North America during this
change—including the extinction of the dinosaurs—is time. (Eventually horses migrated to Eurasia over the
still an area of active research and controversy. Under land bridge between Alaska and Siberia, and went extinct
one hypothesis, the debris thrown up by the impact may in North America 10,000 years ago. Today’s wild herds
have been sufficiently dense to darken the skies for long in North America are descendants of horses brought by
enough that photosynthesis was largely shut off. Other the Spanish conquistadores just 500 years ago.)
evidence suggests that the impact radiated enough heat The drop in CO2 levels may have had a second
to start massive and widespread fires that lasted for sev- important effect on plant ecology beyond the change in
eral years. The debris remaining in the atmosphere from temperature. Studies of the isotopic signatures of fossil
the impact and the fires is likely to have led to a drastic plants show that C4 angiosperms first evolved from C3
cooling in the world’s temperatures for a period of years ancestors during the Oligocene. C4 plants are more effi-
to decades. It has also been hypothesized that the dis- cient than C3 plants at CO2 uptake (see Chapter 2), espe-
ruption in the global carbon cycle caused by the impact cially at low CO2 concentrations. During the late
led to major oscillations in global climates for the next Miocene, tropical and subtropical grasslands dominat-
million years. It is now generally accepted that this event ed by C4 grasses spread widely, especially between 7 and
is tied to the extinction of the dinosaurs, along with 70% 8 million years ago. Their spread has often been
of the marine species that existed at the time. explained as a response to the drop in CO2 concentra-
In North America, the flora of the Paleocene epoch tions. However, the drop in CO2 had already become
contained many fewer species than the flora of the late substantial by the early Miocene, some 16 million years
Cretaceous. In the southern part of the continent, almost earlier. So, the delay in the spread of C4 grasslands is
80% of existing plant species became extinct, while polar unexplained. Researchers have suggested that climate
forests had a 25% extinction rate. Even as far away from change, rather than changes in global CO2 concentration,
the impact as New Zealand, a recent study of fossil pollen might be the most important factor in the rise of C4
and spores found that a diverse flora consisting of gym- grasslands (Pagani et al. 1999).
nosperms, angiosperms, and ferns was abruptly replaced We can look at changes in the distribution of C3 and
with one consisting of a small number of fern species C4 grasses in response to much more recent climate
(Vajda et al. 2001). Recovery occurred only gradually over changes in order to understand what might have hap-
more than a million years, with the eventual evolution pened during the Miocene. Studies of fossils from 10,000
of many new species, particularly angiosperms. years ago in two lake beds in Mexico and Guatemala sup-
port the hypothesis that climatic drying was responsible
for the local spread of C4 grasses (Huang et al. 2001).
The Cenozoic Era
Atmospheric CO2 concentrations were the same at both
In the early Eocene epoch, global temperatures began to sites, and thus can be ruled out as a cause of differences
fall, following falling global CO2 levels. In North America, in the vegetation. At the Mexican site, C3 grasses spread
as the local climate become rainier at the end of the most because most temperate-zone trees are wind-pollinat-
recent glaciation; at the Guatemalan site, C4 grasses ed—and wind-pollinated herbaceous groups, especially
spread as the local climate become drier. Studies of grass- grasses and sedges.
lands in southern Africa also suggest that the relative pro- The level to which pollen can be identified varies
portion of C3 to C4 plants varies with climate (Scott 2002). among taxa. Pollen from some trees can be identified to
How will the Earth’s vegetation change in response species. In many genera, however, only subgenera can be
to the current rise in global CO2 levels? Will it favor the distinguished. For example, among pines, the pollen of
spread of C3 plants over C4 plants (see Chapter 22)? It Pinus resinosa (red pine, Pinaceae) and P. banksiana (jack
seems likely that the answer will depend on how changes pine) cannot be told apart, although they can be differ-
in the abundance of CO2 affect regional climates. entiated from that of P. strobus (white pine). In some cases,
such as the grasses, pollen can be identified only to the
Paleoecology Methods family. Despite these limitations, palynology has been key
to reconstructing communities and migration patterns in
How do we know so much about the history of plant temperate regions.
communities? Studies of the past use a variety of tech-
niques. Of primary importance is the study of fossils.
From these remains of leaves, stems, flowers, and other
plant parts, we can reconstruct past communities and
Pinus contorta
climates (Prentice et al. 1991).
(lodgepole pine,
The sources of fossils depend on the habitat where Pinaceae)
they were formed. Because most organic material decom-
poses quickly, anoxic—oxygen-free—conditions are ideal
for its preservation. In nonmarine habitats, these condi-
tions are most often found at the bottoms of lakes and
bogs. Sediments formed in lakes and bogs are the primary Acer rubrum
source of macrofossils, such as leaves, flowers, stems, and (red maple,
seeds. For information from hundreds of thousands or Sapindaceae)
millions of years ago, we rely on hardrock fossils, in which
the organic material has been replaced by nonorganic
material. Material from more recent periods may still be
organic. Bogs, which have highly acidic waters that slows
decay, are especially useful in this regard. Quercus garryana
Fossils can be retrieved from lakes and bogs by (garry oak,
removing a core from the sediments at the bottom. A long Fagaceae)
tube or pipe, open on the bottom, is pushed down into
the sediments, and the core is then carefully drawn up.
Thin slices of the core are examined; fossils, if present, are
identified, and the slices are aged using isotopic dating.
In addition to macrofossils, these cores yield pollen
grains (Figure 21.7) and other microfossils. Palynology Chenopodium album
is the study of spores and pollen. Wind-borne pollen (lamb’s quarters,
Amaranthaceae)
often lands in a bog or lake and settles to the bottom.
Because pollen can travel long distances, cores from lake
or bog sediments provide a profile of the community
over a wide area around the body of water. In contrast,
macrofossils usually provide a picture only of the plants
growing immediately adjacent to the area studied.
Palynology has strengths and weaknesses as a Ipomoea wolcottiana
method of studying the past. One strength is that pollen (bindweed,
is readily preserved, and large amounts of it can usual- Convolvulaceae)
ly be found. A weakness is that not all species are pres- 10 µm
ent or equally represented, or even identifiable. Only the
Figure 21.7
pollen of wind-pollinated species makes it into lakes and Pollen grains preserved in sediments can be used to identify
bogs in quantity; animal-borne pollen is largely missing. plant species present at some time in the past. (After Pielou
Thus, sediment core samples are biased toward trees— 1991.)
Another way in which plant material can be pre-

served is by drying. This means of preservation has pro- (A)
vided macrofossils in arid regions of North America Present
Global mean temperature (¡C)

17 interglacial
where packrats (Neotoma) live. As their name implies,
these rodents accumulate large amounts of plant mate- Present mean temperature
15 15¡C
rial, which they store in their dens. They urinate on the
plant material, caking it with mineral salts. Urination
provides substantial evaporative cooling for the den and, 13
incidentally, provides excellent conditions for preserva-
tion of the plant material. The accumulations of pack- 11
rats are referred to as middens. A given den site may be
Wisconsin maximum
used for centuries by successive generations of packrats, 9
800 600 400 200 Present
with its middens growing to several meters across. A fos-
silized midden may be preserved for thousands of years Thousand years ago
in a dry climate. Samples from packrat middens can be (B)
radiocarbon-dated and the plant material identified,
2 Sangamonian
allowing researchers to reconstruct changes in the plant interglacial
community surrounding the den site (Cole 1985; Betan-
court et al. 1990). In Arabia and Africa, the rock hyrax 0 15¡C
creates a similar structure, called a hyraceum.
—2
The Recent Past
—4
We now leave the distant past and jump ahead to a point Wisconsin
just 20,000 years before the present, when much of the maximum
—6
Northern Hemisphere was covered by glaciers. Over the 140 120 100 80 60 40 20 0
Global temperature difference from present mean (¡C)
previous 2 million years, during the Pleistocene epoch, Thousand years ago
Earth had experienced a period of major glaciations. Start-
(C)
ing about 850,000 years ago, there were gradual periods
of glacial buildup, followed by abrupt switches to warmer Hypsithermal
weather and short periods of glacial retreat (called inter- 1
glacials) (Figure 21.8). Each of these periods of alternat-
ing cold and warm weather lasted about 100,000 years.
These cycles were caused mainly by changes in the tilt 0 15¡C
and precession of Earth’s orbit (see Chapter 18). The var-
ious glacial advances and retreats are known by various
names. In North America, the most recent glacial advance
is called the Wisconsin glaciation, named for one of the —1
locations of its southernmost extent (Figure 21.9). Because
10 8 6 4 2 0
each glacial advance destroyed most of the traces of pre-
Thousand years ago
vious ones, we focus here on the Wisconsin glaciation, for
which the best data exist. (D)
The last warming period began about 12,000 years
Middle Ages Little Ice Age
ago. Since that time, the average global temperature rose, warm epoch
1
peaked about 7000 years ago, and then declined (Figure
21.8B,C). If not for human activities bringing on a new,
rapid period of global warming, we would likely be see-
ing the growth of glaciers in the next few thousand years. 0 15¡C
—1
Figure 21.8
Mean global temperatures over geological time for various
periods of time: (A) the past 850,000 years; (B) the past 140,000 1000 800 600 400 200 0
years; (C) the past 10,000 years; (D) the past 1,000 years. (After (1000 C.E.) (1200 C.E.) (1400 C.E.) (1600 C.E.) (1800 C.E.) (Present)
Gates 1993.) Years ago
What has all of this climate change meant for plants? glaciers were found in mountainous regions such as the
As glaciers advanced and retreated, plants in northern Rocky Mountains in North America, the Himalayas in
latitudes migrated across continents, following mov- Asia, and the Alps in Europe. Few glaciers formed in the
ing climate zones. In the Tropics, these climate changes Southern Hemisphere, excepting Antarctica, because
resulted in the contraction of forests and the expansion there is little continental land area at high latitudes,
of grasslands. Our study of these changes in plant dis- although some glaciers existed in the Andes mountain
tributions has led to new understanding of the causes of range of South America. Regional climatic conditions
diversity patterns across the globe, as well as new maintained some glacier-free regions even at high lati-
insights into the nature of plant communities. tudes. For example, some coastal regions remained free
of glaciers in Canada and Alaska.
At the Glacial Maximum With so much water tied up in glaciers, sea levels
At the maximum of the Wisconsin glaciation, large con- were lower than today’s by 85 to 130 meters, resulting
tinental glaciers extended as far south as 40° N latitude in coastal areas that were substantially broader than at
in North America and 50° N in Eurasia, while smaller present. Of particular importance to plants were the
(A)
Alaska Baffin
I.
Ne
wf
ou
nd
Hudson lan
d
Bay N
500 km
Nova
Scotia
Maine
Wisconsin
Long Island
New Jersey
(B)
Taymyr
500 km
Peninsula
Novaya
Zemlya
Svalbard
Barents Kara
Sea Sea
Arc
t i c Ci
rcle
North y Yamal
Norwa Peninsula
Sea
en
Finland
ed
UK
Sw
Figure 21.9
The greatest extent of continental Russia
glaciers during the most recent Denmark
glaciation in (A) North America St. Petersburg
and (B) Eurasia. (After Mayewski France
1981 and Siegert et al. 2002.)
southeastern coast of North America, which served as by glaciers. Most tropical studies have focused on
a refuge for species pushed southward by the glaciers, events in the Amazon basin. While the exact details of
and the Bering land bridge connecting Alaska with what happened there are still under active study,
Siberia, which was an important migration route for researchers agree that the climate in the Amazon basin
plants and animals between North America and Eura- at the glacial maximum was cooler than at present.
sia. (It was by this route that humans probably first Until recently, it was generally thought that the climate
reached North America.) was markedly drier as well—in fact, precipitation was
South of the glaciers, and in other ice-free areas thought to be the most important factor in determin-
such as eastern Siberia, were vast stretches of tundra ing tropical plant distributions during that time (Gra-
and muskeg—a grassy bog habitat with scattered, ham 1999). However, a recent review suggests that the
stunted conifers. The tree species of the taiga and tem- evidence for decreased rainfall at the glacial maximum
perate forests that today cover these areas were then is equivocal (Burnham and Graham 1999).
found far to the south. In southeastern North America, Previously, almost all paleoecologists believed that
along the coastal plain, there were forests of deciduous cycles of glaciation and warming created, alternative-
trees such as Quercus (oak), Carya (hickory), Fagus ly, highly connected forests and a mixture of forest frag-
(beech), and Acer (maple). Along the southern reaches ments separated by savannas and grasslands. Some
of the Mississippi River Valley were coniferous forests ecologists have hypothesized that these changes in for-
dominated by species such as Picea (spruce), Thuja est coverage in the Tropics contributed to the high
(cedar), Abies (fir), and Larix (larch). species diversity in tropical regions (see Chapter 20)
The discovery of these distributions addresses an Current data do not support this hypothesis. Compar-
important question in plant community ecology: isons between taxonomic groups that should have been
whether communities are highly coordinated units or highly affected by this fragmentation (such as terres-
simply collections of coexisting species (see Chapter trial mammals) and groups less likely to be affected
12). The assemblages of species that existed 20,000 (such as birds) have failed to find differences in the
years ago, at the most recent glacial maximum, have no numbers of species that evolved. Some recent studies
modern equivalent. For example, Tsuga canadensis (east- suggest that forests in the Amazon basin may not have
ern hemlock, Pinaceae) and Acer saccharum (sugar become so severely fragmented after all. Studies of
maple, Sapindaceae) are codominants in forests around pollen deposited at the mouth of the Amazon suggest
the Great Lakes today. Yet during the glacial maximum, that throughout the last 50,000 years, forests have
the former tree species was found mainly along the always dominated the region (Haberle 1999; Colinvaux
Mississippi River, while the latter was found several and De Oliveira 2001). A recent study suggests an alter-
hundred kilometers to the west along the Gulf Coast native mechanism that might have fragmented the
and parts of inland Texas. In the late 1970s and early Amazon basin, however: increases in sea level during
1980s, palynological studies of past species distribu- the interglacials may have broken the area into two
tions and the mapping of northern migrations in major islands and a number of smaller archipelagos
response to climatic warming demonstrated that the (Nores 1999).
species compositions of plant communities have been An important event for North American plant com-
reshuffled many times. These data demonstrate that munities was the arrival of humans, possibly as early
communities are not highly coordinated units, as well as 30,000 years ago, but certainly by 12,000 to 15,000
as showing that communities rarely reach an equilib- years ago. This migration is linked to a major extinc-
rium of either species composition or population struction about 10,000 years ago of most of the large mam-
ture (Davis 1981). mal species in the New World (Martin and Klein 1984).
Studies of the refuges where plant species survived Similar mass extinctions of large animals followed the
during the glacial maximum also point to important arrival of humans in Australia and Madagascar. Such
causes for contemporary differences between North extinctions can create pitfalls for present-day plant ecol-
American and European plant communities. As men- ogists. Today, if you travel the western deserts of the
tioned in Chapter 20, north-temperate and boreal com- United States, you find many shrubs with sharp thorns
munities in North America contain approximately 50% and spines. Yet, except for introduced species such as
more vascular plant species than equivalent European cattle, horses, and donkeys, there are almost no large
ones. This difference is due to the many species in browsers, except for bighorn sheep, pronghorn, and
Europe that went extinct during the Pleistocene (see mule deer. Only after studying the fossil record would
Table 20.2), but which survived in North America you realize that these plants evolved during a time
because they were able to reach refuges. when North America had many large native browsers,
These global climate changes also affected the such as horses, camels, elephants, and giant ground
Tropics, even though those regions were not covered sloths, all of which are now extinct.
(A) (B)
6 4
6 4
6 7 5 5 6
10 8 8 6 6
9 >8 4
6 11 6
4 5
10 10 >10 10 9
8
11 10 4 8
11 9 8
12 10 8
10 10 9 11 10 10 7 7
11 11 7 8 7
9 10 9 6 6
11 10 12 12 6
12 5
8 10
6
10 10
10
10 13
15
15 11
17 13 12 12
12
14
Ulmus spp. Fagus
>13 14 grandifolia
14
Elm
Beech
400 km 400 km
(C) (D)
5 6
8
7 10
8 8 10
7 6 7 99 9
7 9 10 8 111111 10
2 10
8 8 9 12 10 11
8 10 10 8? 12
5 9 14 12
8 10 11
6 12 12 12
0 10 13 11
10 10 11 12 12 12
0 12 12 12 12
11
10? 10 10 11 14 15 14 12
10 15 12
14
0 >13 21 14
12 >13
>11
23 12
0
25
10 22—16 14
0 0
20—13 17
Pinus strobus Picea spp.

>14
0
White Pine Spruce
400 km 400 km
Figure 21.10
Examples of migrations patterns for trees in North America pollen to species, and in others only to genus. (A) Ulmus
based on pollen records. Counters indicate the time of spp. (elm, Ulmaceae). (B) Fagus grandifolia (beech, Fagaceae).
arrival (thousands of years ago). The green area shows the (C) Pinus strobus (white pine, Pinaceae). (D) Picea spp.
current distribution. In some cases we can identify the (spruce, Pinaceae). (After Davis 1983.)
Glacial Retreat
(e.g., Fagus and Castanea) were
Beginning about 20,000 years ago, the climate warmed actually the slowest to migrate
and the glaciers retreated, slowly at first, then more rap- northward.
idly starting about 14,000 years ago. For several thou- The migration of a species is
sand years, as the ice melted, large lakes (much bigger not a steady progression. For
than the current Great Lakes) dominated the interior of example, the migration of Fagus
North America. Areas that had been under ice became grandifolia (American beech,
tundra and then forest or grassland. Fagaceae) over the last 7000
During this warming period, trees migrated north- years has included both periods
ward. However, each tree species was unique in its of rapid expansion and periods
migration pattern (Figure 21.10). Many of these patterns of relative stability (Woods and Margaret Davis
were deciphered by Margaret Davis, using pollen sam- Davis 1989). During this period,
ples gathered from lakes and bogs throughout eastern beech migrated northward and westward from the
North America (Figure 21.11). Among the fastest-mov- Lower Peninsula of Michigan into the Upper Peninsula
ing species was Pinus strobus (white pine, Pinaceae), and Wisconsin. Some of the delays are attributed to the
which migrated at an average rate of 400 meters per year. barrier presented by Lake Michigan; long-distance dis-
In contrast, Ulmus (elm, Ulmaceae) moved at a rate of persal by blue jays or passenger pigeons may have been
250 meters per year, while slowpoke Castanea dentata responsible for breaching this barrier. Other starts and
(American chestnut, Fagaceae) moved a mere 100 meters stops may have been responses to climatic fluctuations.
per year (Table 21.1). Beech arrived in Wisconsin on the western shore of Lake
One might imagine that wind-dispersed and ani- Michigan approximately 6000 years ago, but did not
mal-dispersed species would have migrated at about spread farther west and north for another 1500–2000
equal rates, and that among animals, birds would have years, probably due to cold winter temperatures.
moved seeds faster than mammals. Yet Table 21.1 clear- Migration depends on the successful establishment
ly shows that wind-dispersed seeds generally traveled of seedlings. When we look closely at the frontier of a
much faster than animal-dispersed seeds. The exception, migration, we find a patchwork of populations, with
Quercus (oaks), is dispersed mainly by small mammals establishment and spread highly dependent on local
in eastern North America. The bird-dispersed species conditions. Davis et al. (1998) studied the spread of east-
e
ea
ac
ae
th
ce
Am isia
us
an
e
us
ra
ea
s
in
la
tem
ar
erc
s
mu
pe
cea
ac
nu
ax
tu
Qu
Cy
Po
Ar
Ul
Be
Fr
Pi
Pi
3
Depth from marsh surface (m)
6
Figure 21.11
7 Example of a pollen diagram for Kirchner
Marsh, Minnesota. The deepest samples are the
8 oldest. About 20,000 years ago, this area was
dominated by Picea (spruce) and Fraxinus (ash).
9 These were followed by a succession of species:
Betula (birch), Pinus (pine), Ulmus (elm), and
10 Quercus (oak). The subsequent decrease in oaks
and increase in grasses (Poaceae) indicates the
11 existence of a woodland vegetation. Currently
the area is again dominated by pines mixed
20 40 60 10 20 20 20 20 40 20 10 10 20 40 with oaks, but now with a sedge (Cyperaceae)
Percent of pollen grains understory. (After Grimm 1988.)
on nearby, taller mountains. The initial hypothesis was

Table 21.1 Average rates of northward range expansion that the soil on these mountains was not suitable for
for trees in eastern North America following coniferous species. But no differences from nearby
the most recent glacial retreat
mountains could be found, and coniferous trees plant-
Species Rate (m/yr) Dispersal agent ed on the balds grew well.
Pinus banksiana/resinosa 400 Wind The mystery was solved once the effects of climate
Pinus strobus 300–350 Wind change and migration patterns were considered. As the
Quercus spp. 350 Animals climate warms and cools, climate zones move up and
Picea spp. 250 Wind down mountainsides, and plant species follow this move-
Larix laricina 250 Wind ment. During the most recent glaciation, the balds were
Ulmus spp. 250 Wind covered with coniferous forests. As the climate warmed,
Tsuga canadensis 200–250 Wind the lower edge of the ranges of the conifers moved
Carya spp. 200–250 Animals upward until it was higher than the tops of the moun-
Abies balsamifera 200 Wind tains. At this point, the mountains were entirely covered
Acer spp. 200 Wind with deciduous forests. Then the climate cooled again.
Fagus grandifolia 200 Animals The upper edge of the species ranges of the deciduous
Castanea dentata 100 Animals trees moved downward, below the tops of the mountains.
Now, however, there were no coniferous trees left on the
Source: Davis 1981.
mountains to replace them, leaving the tops bald—dom-
inated by grasses and forbs. On higher mountains the
ern hemlock into an area in the western Upper Penin- coniferous species were still present, so they simply
sula of Michigan. Four communities dominated by migrated back down. Given enough time, long-distance
white pine were invaded by eastern hemlock about 3000 migration might return conifers to these mountaintops.
years ago over a period of 800 years. Over the next 2000 This pattern of vertical migration was even more
years, climate changes caused the water table to rise and important in the American Southwest. This region has
the soil to become wetter. Eastern hemlock became the many high mountains, plateaus, and other vertical fea-
dominant species, and white pine disappeared from all tures, so that during the Pleistocene much of the plant
but one site. In contrast, four other nearby sites that were migration consisted of elevational changes. During peri-
initially dominated by sugar maple and oak were never ods of maximum glaciation, the climate was cooler and
invaded by eastern hemlock, possibly because the wetter, and pine forests and pinyon-juniper woodlands
greater availability of soil nitrogen and higher light lev- spread out over the landscape. Large lakes dotted the
els in the spring under the deciduous canopy allowed landscape in what is now the Great Basin desert. Their
sugar maple seedlings to outcompete hemlock. Thus, the remnants can be seen today as the Great Salt Lake and the
exact details of migration patterns may be highly con- Bonneville Salt Flats. During interglacial periods (like the
tingent on local conditions and on which species hap- present), the low-elevation areas became hotter and drier,
pen to arrive first. and desert shrubland became dominant in these areas.
Studies of these patterns in the Grand Canyon in
Climate Fluctuations in the Recent Past Arizona have disclosed an important phenomenon,
Superimposed on the long-term glacial-interglacial cli- called vegetation lag. Such lags occur when changes in
mate swings are shorter climate fluctuations (see Figure the vegetation happen more slowly than the rate at
21.8D). The period from 1500 to 1850, known as the “Lit- which the climate changes. Using evidence from pack-
tle Ice Age,” was a time of low solar activity, cold tem- rat middens, Kenneth Cole (1985) found that as the cli-
peratures, and glacial advances (interrupted by rela- mate warmed, species disappeared locally as the lower
tively warm interludes in 1540–1590 and 1770–1800). edges of their ranges moved upward. However, these
These shorter-term fluctuations also leave their imprint species were not immediately replaced; it took centuries
on plant communities and often help explain puzzling for new species to become established (Figure 21.12).
aspects of current species distributions. The studies of migration in the Great Lakes region
Grassy balds—treeless mountain peaks—in the described previously also found evidence of vegetation
southern Appalachian Mountains are odd in that they lag, with species taking 1000 to 2000 years to reach and
do not occur where you might expect them, on the then dominate areas after the climate had become suit-
tallest mountains. Rather, they are found on mountains able for them. The studies of the Grand Canyon vege-
of intermediate height, but not on all such mountains. tation also reinforce the notion that plant species
Closer inspection shows that the tree line on these migrate individualistically, with plant communities
mountains occurs at about 1500 meters, the elevation changing through time as species combinations
where deciduous forest gives way to coniferous forest changed (Cole 1990).
Plant species may also undergo evolutionary adap- ation with a population (see Chapter 5). Migration is
tation in response to climate change. Rapid climate limited both by dispersal biology (for example, wind
change—of the kind now occuring in response to ris- can move pine seeds only so far) and by anthropogenic
ing concentrations of atmospheric CO2—imposes very changes in land use that present barriers to species
strong natural selection on populations. Studies of tree movement. Together, these factors suggest that rates of
species point to the conclusion that adaptation and extinction are likely to increase as a consequence of cur-
migration occur simultaneously (Davis and Shaw rent global climate change.
2001); they are not alternative responses to climate
change. Will plant populations be able to meet the chal- Summary
lenge of today’s changing climates through some com-
bination of adaptation and migration? Evolutionary The plant communities that we see today are not stat-
responses are limited by the amount of heritable vari- ic entities; they are the current state of a dynamic sys-
(A)
(B) Figure 21.12

An area near the southern rim of the
Grand Canyon in Arizona at 1500 m
elevation. (A) Today, this area is a
desert shrubland of Ephedra viridis
(Mormon tea, Ephedraceae), Coleogy-
ne ramosissima (blackbrush, Rosaceae),
and Opuntia engelmannii (prickly pear
cactus, Cactaceae), with a few juniper
(Juniperus sp., Cupressaceae) along
the cliff tops. (B) Reconstructed image
of the same site as it looked 20,000
years ago, based on plant fossils from
packrat middens. The bottom slopes
support a woodland of juniper,
Artemisia tridentata (sagebrush, Aster-
aceae), and Atriplex canescens (shad-
scale, Amaranthaceae). In the shady
alcoves of the cliffs grow Pseudotsuga
menziesii (Douglas-fir, Pinaceae) and
Abies concolor (white fir, Pinaceae),
with a few Pinus flexilis (limber pine,
Pinaceae) along the cliff tops. (Cole
1985; photograph and reconstruction
courtesy of K. Cole.)
tem that changes in a complex fashion. These changes communities appeared. These changes were driven by
occur on both ecological and evolutionary time scales. environmental change, caused primarily by continen-
Unless we recognize this, we can be seriously misled tal drift, mountain building, and variations in Earth’s
about the causes of ecological patterns. Among plant orbit. Equally important were evolving biotic interac-
ecologists, appreciation for this dynamic has grown tions such as competition, herbivory, and pollination.
during the past century. Today we have a much better The evolution of these interactions created feedback
understanding of the origins of modern plant commu- effects, spurring further change.
nities, although much still remains to be discovered, Today’s temperate plant communities were shaped
especially about the Paleozoic and Mesozoic eras. mainly by patterns of glaciation during the past 2 million
The Paleozoic era witnessed the establishment of years. The last glacial retreat began about 12,000 years ago,
plants on land and the origins of many of the basic pat- with temperatures first rising, peaking at about 7000 years
terns of plant and community diversity. The early Meso- ago, and then dropping. Along with these climate changes
zoic era saw the dominance of the first seed plants, the have come migrations of species, resulting in shifting
seed ferns and the gymnosperms. An explosion of new mosaics of species and communities across the landscape.
forms and patterns occurred in the Cretaceous with the Today these changes are happening at an even more
diversification of the angiosperms. dizzying pace, but they are now driven primarily by
During the Cenozoic, angiosperms continued to human-caused environmental changes—the topic of the
evolve, and more recognizably modern species and next chapter.
Additional Readings
Classic References Swetnam, T. W., C. D. Allen and J. L. Betancourt. 1999. Applied histori-
cal ecology: Using the past to manage for the future. Ecol. Appl. 9:
Davis, M. B. 1981. Quaternary history and the stability of forest com-
1189–1206.
munities. In Forest Succession Concepts and Applications, D. C. West, H.
H. Shugart and D. B. Botkin (eds.), 132–153. Springer-Verlag, New
York.Contemporary Research Additional Resources
Krasilov, V.A. 1975. Paleoecology of Terrestrial Plants: Basic Principles and Graham, A. 1999. Late Cretaceous and Cenozoic History of North Ameri-
Techniques. Wiley, New York. can Vegetation. Oxford University Press, New York.
Niklas, K. 1997. The Evolutionary Biology of Plants. University of Chica-
Contemporary References go Press, Chicago.
Davis, M. B., R. R. Calcote, S. Sugita and H. Takahara. 1998. Patchy inva- Pielou, E. C. 1991. After the Ice Age. University of Chicago Press, Chica-
sion and the origin of a hemlock-hardwood forest mosaic. Ecology 79: go.
2641–2659.
Betancourt, J. L., T. R. V. Devender and P. S. Martin. (eds.). 1990. Packrat
Jacobs, B. F., J. D. Kingston and L. L. Jacobs. 1999. The origin of grass- Middens: The Last 40,000 Years of Biotic Change. University of Arizona
dominated ecosystems. Ann. Missouri Bot. Gard. 86: 590–643. Press, Tucson.
Skog, J. E. 2001. Biogeography of Mesozoic leptosporangiate ferns relat-
ed to extant ferns. Brittonia 53: 236–269.
22
C H A P T E R Global Change:
Humans and Plants
E
arth is going through one of the most rapid periods of change in its his-
tory, and the primary cause of the ongoing changes is humans. The glob-
al carbon cycle is central to these changes.
One of the major consequences of alterations in the carbon cycle has been
a global rise in temperatures. Some of the other likely changes in weather
include shifts in rainfall patterns, so that some areas will become wetter and
others drier; changes in the seasonality of rainfall in many locations; and
increases in rainfall intensity during storms. These changes will affect most liv-
ing organisms.
The global carbon cycle is intimately intertwined with plant ecology. Plants
both drive and respond to the carbon cycle. As conditions for plant growth
change over the face of the globe, dramatic changes—some predictable, but
others unpredictable—will affect agricultural systems and natural ecosystems
in all biomes. Plant and animal distributions will shift, natural communities
will have different species compositions than they have today, some species
will become extinct, and some natural communities will decline or vanish alto-
gether. Humans and human institutions (such as national economies) will
almost certainly be affected by these changes as well, directly by changing
weather patterns and rising sea levels and indirectly by alterations in agricul-
tural productivity (for better or worse, depending on the location).
While shifting weather patterns are the best known of its effects, anthro-
pogenic global change encompasses far more than warmer temperatures. Other
global changes caused by humans over the past 150 years include greatly
altered patterns of land use and a large and continuing decline in the diversi-
ty of plants and animals living on Earth. To adequately address the role of
plants, and consequences for plants, in this global drama, we must digress in
places from our focus on plants to look more closely at the actions of people.
Carbon and Plant-Atmosphere Interactions

The Global Carbon Cycle
The uptake of carbon by plants was introduced in Chapter 2, and we exam-
ined in detail how carbon moves through ecosystems in Chapter 15. To recap
briefly, plants take up carbon in the form of CO2 from the atmosphere and
incorporate it into organic compounds using energy from sunlight. Carbon
moves through ecosystem food webs via consumption living plants contain the two other major active pools of
of plants and plant parts and by the actions of decom- CO2, with somewhat more carbon stored in the atmos-
posers. Organic carbon accumulates in ecosystems for a phere than in plants. (The organic matter stored in liv-
period of time both in living organisms and as soil ing animals is minuscule because the total biomass of
organic matter. Carbon returns to the atmosphere pri- animals is small relative to the global carbon budget.)
marily through the respiration of soil organisms. We Gross primary production—the total flux of CO2
now widen our focus to a global scale, examining the into living terrestrial plants—removes the greatest
pools and fluxes of carbon in terrestrial environments, amount of carbon from the atmosphere yearly. About
the atmosphere, and the oceans (Figure 22.1). half of this carbon is returned to the atmosphere through
By far the largest active pool of carbon on or near the respiration of terrestrial plants, so that terrestrial
the Earth’s surface is that dissolved in the oceans. (This global NPP (net primary production: see Chapter 15) is
does not include the substantial pools of carbon con- about 60 × 1015 g C/year. Marine phytoplankton—par-
tained in rock, oil, and natural gas, which are naturally ticularly cyanobacteria—exchange CO2 directly with
largely inactive.) Most of the active marine carbon is that dissolved in surface ocean water and only indi-
found in deep ocean waters, and the exchange of this rectly with the atmosphere. NPP in the oceans is esti-
carbon with the atmosphere occurs slowly, with a mated to be about 35–50 × 1015 g C/year, or somewhat
turnover time of about 350 years. Surface waters ex- less than that of terrestrial plants. Most of the carbon
change CO2 with the atmosphere much more quickly. fixed by marine phytoplankton is taken up by het-
It takes about 11 years, on average, for a molecule of CO2 erotrophic bacteria and zooplankton in surface waters.
in the atmosphere to be dissolved at the ocean surface Much of the CO2 taken up by the bacteria is quickly
and then released again. Slightly more CO2 is dissolved transformed to inorganic forms again and released into
every year than is released, so the oceans currently act the water, while the zooplankton form the basis for
as a net sink for carbon. marine food webs. Only a tiny proportion of the NPP
Another substantial pool of carbon is found in soil eventually sinks to the ocean bottom. The total pool of
organic matter. A major flux of CO2 into the atmosphere carbon in marine organisms at any one time is very
comes from the respiration of soil organisms, primarily small, although the turnover rate of carbon in this pool
microorganisms (Schlesinger 1997). The atmosphere and is very rapid.
6 GPP
120
Land Atmospheric pool
plants RP
750
Anthropogenic 560 60 Annual change: +3.2
sources
60
0.9
Rivers 92 90
0.4 DOC
Net destruction
0.4 DIC 38,000
of vegetation
Soils
Ocean
1500
0.1
Burial
Figure 22.1
The global carbon cycle. The numbers show the pools (in phere. Earth contains about 1023 g carbon; most of this car-
units of 1015 g C) of carbon in various major compartments bon is buried in sedimentary rocks. Only about 1018 g carbon
and the fluxes (in units of 1015 g C/year, shown in boldfaced is contained in active pools, as shown here, with the largest
type) of carbon moving between portions of the land, of these as marine dissolved inorganic carbon. DOC, dis-
oceans, and atmosphere. Notice that land plants take up solved organic carbon; DIC, dissolved inorganic carbon;
about twice as much carbon annually as they release, and GPP, gross primary production; RP, plant respiration. (After
that land plants store almost as much carbon as the atmos- Schimel et al. 1995 and Schlesinger 1997.)
PAGE PROOF: 2ND PASS Global Change: Humans and Plants 439
(A) (B)
380 360
370
360 340
CO2 (ppmv)
CO2 concentration (ppmv)

350
340 320
330
300
320
310
1960 1970 1980 1990 2000
Year 280
Figure 22.2
(A) Recent increases in atmospheric CO2 (in parts per mil-
lion by volume, ppmv) recorded at the Mauna Loa observa-
tory in Hawaii. This site was chosen because it is not subject 260
1750 1800 1850 1900 1950 2000
to the variation in CO2 levels that would be found close to
Year
sources of anthropogenic emissions, but the results are simi-
lar to those found at many other sites. The actual monthly
values are shown by data points; the wiggling line is a result
of seasonal variation due to the decrease when Northern ining the direct effects of increasing CO2 levels, we dis-
Hemisphere terrestrial vegetation is most actively photosyn-
thesizing in summer and the increase when it is least active- cuss the effects they may have on global climates. We
ly photosynthesizing in winter. The long-term trend is then examine the sources of this anthropogenic atmos-
shown by the colored line, which has been statistically pheric carbon.
smoothed based on the monthly values. Data for 1958 to
1974 are from the Scripps Institution of Oceanography, and Direct Effects of Increasing CO2 on Plants
for May 1974 to 2001 from the U.S. National Oceanic and
Atmospheric Administration. (Keeling and Whorf 2001.) (B) Plants and other photosynthetic organisms use atmos-
The trend in CO2 concentrations over the past two centuries, pheric CO2 in photosynthesis, and the carbon in the
determined from bubbles of gas in ice cores from Antarctica. atmosphere is thus the basis for most food webs on
(After Vitousek 1992.) Earth. For plants that use C3 photosynthesis—that is, the
majority of photosynthesizers—the concentration of CO2
in the atmosphere is suboptimal (see Chapter 2). It is rea-
sonable to assume, therefore, that rising atmospheric
These fluxes of CO2 are natural, although they can CO2 levels might act as a “fertilizer” to stimulate plant
be affected by human activities. Two additional impor- productivity. Elevated CO2 levels can also increase the
tant fluxes in the global carbon cycle, however, are efficiency of water and nutrient use (at least for C3
entirely anthropogenic, at least on their current scales. plants) and alter patterns of carbon allocation within the
These two fluxes, both of which add CO2 to the atmos- plant body. Because different plants respond very dif-
phere, result from the cutting and burning of large areas ferently to enhanced CO2 levels, global changes in
of forests and from the extraction and combustion of fos- atmospheric CO2 concentrations might change intra-
sil fuels, including oil, natural gas, and coal. The imme- and interspecific competitive interactions among plants.
diate consequence of these additional fluxes is that Changes in competitive abilities, as well as any other
atmospheric concentrations of CO2 are rising. Before the responses that lead to differences in relative perform-
Industrial Revolution of the nineteenth century, CO2 was ance among individuals, can change the relative repro-
280 ppmv (parts per million by volume) of the atmos- ductive output of individuals, thus altering the future
phere. The concentration of CO2 is now about 370 ppmv genetic structure of plant populations. Elevated CO2 can
on average at Earth’s surface and, if we continue our cur- also affect leaf chemistry, alter local ecosystem carbon
rent practices, could reach twice preindustrial levels— cycles, and change nitrogen cycling and other ecosys-
560 ppmv—by 2065 (Figure 22.2). These higher levels of tem-level processes. Some ecologists have predicted that
CO2 will have direct effects on plants as well as indi- although there will be large positive responses to high-
rect effects mediated by changes in climate. After exam- er CO2 levels initially (due to the “fertilizer” effect), this
will be followed by adjustment and reduced responses have positive direct responses to higher atmospheric
over time, over a range of scales from individual plants CO2 levels, including higher photosynthetic rates,
to ecosystems. The details of these changing physio- greater growth, and reduced water loss, if nothing else
logical and biogeochemical responses to rising CO2 lev- changed. However, the indirect effects of the climate
els are interesting, but beyond the scope of this book. changes that are likely to occur due to elevated CO2 will
It has been predicted that C3 plants should grow probably dwarf these direct effects.
more vigorously under elevated CO2, while C4 plants
should not (see Chapter 2), and that C4 plants may there-
fore be at a greater competitive disadvantage in an Anthropogenic Global Climate Change
enhanced-CO2 world. Differences among species in
growth and other responses to enhanced CO2 would be The Greenhouse Effect
likely to change community composition, as some species Carbon dioxide molecules in the atmosphere are high-
would do better and others relatively worse. Wand et ly effective at absorbing infrared (or longwave) radia-
al. (1999) carried out a quantitative synthesis of the results tion (see Chapter 18). Visible (short) wavelengths of sun-
of published studies on the responses of uncultivated light pass through the atmosphere without being
grasses to elevated CO2. The researchers showed that absorbed by it, warming the Earth’s surface. Earth’s sur-
across all studies, both C3 and C4 plants had higher pho- face reradiates some of that energy in the longwave part
tosynthetic rates on average in response to elevated CO2 of the spectrum (see Chapter 3). Some of that reradiated
(33% and 25%, respectively). Total biomass increased for energy is absorbed and trapped by the atmosphere, par-
both C3 and C4 grasses (by 44% and 33%, respectively), ticularly by water and CO2 molecules (Figure 22.3). This
and leaf water use efficiency was also greater for both recaptured energy warms the atmosphere, which then
grass types. These findings were surprising, and show further warms the surface. The longwave energy that
that the presumed competitive disadvantage of C4 is not trapped by the atmosphere escapes to space.
species at higher CO2 levels may not exist.
In an analysis of the effects of elevated CO2 on trees
(all with C3 photosynthesis), Curtis and Wang (1998)
found that both photosynthesis and total growth
increased substantially when CO2 levels were elevated
to twice the current atmospheric levels, although stress- Sun
ful conditions such as low soil nutrient levels reduced
these effects of enhanced CO2. They also found no evi- Reflected from
dence that trees eventually would acclimate to high CO2, atmosphere
Reflected from
reducing its effect over time, as some have predicted. It earth’s surface
is probably safe to say, therefore, that many plants would
Solar energy
Water input
Radiated by
Reabsorption by vapor
atmosphere
greenhouse gases flux
into space
in atmosphere
Absorbed
by surface
Infrared radiation
Infrared from surface
radiation
from surface
lost to space
Figure 22.3
The greenhouse effect. About 70% of the incoming short- face. A small portion of the longwave energy escapes from
wave radiant energy from the sun is absorbed by Earth’s the atmosphere to space. The other half of the energy
surface; the remainder is reflected by the surface and the absorbed by the Earth’s surface is used in the evaporation of
atmosphere. About half of the absorbed energy heats the water, which adds water vapor to the atmosphere and ini-
Earth’s surface. Some of that energy is radiated by the sur- tially cools the surface. When that water vapor condenses to
face as longwave (primarily infrared) radiant energy. Most form clouds, the energy removed from the surface is trans-
of that radiated energy is reabsorbed by atmospheric gases, ferred to the atmosphere, driving the creation of storms and
which then reradiate it back to further warm the Earth’s sur- releasing precipitation.
The carbon dioxide in the atmosphere is a natural matically reduced their production, they are so stable
part of the carbon cycle. Why, then, does it threaten to chemically that the molecules already in the air will per-
change global climates? The problem is not that there are sist for a very long time.
CO2 molecules in the atmosphere, but that there has While humans produce far smaller amounts of these
been a tremendous rise in their concentrations. The four compounds than of CO2, these other greenhouse
greater the number of CO2 molecules in the atmosphere, gases are much more potent absorbers of infrared radi-
the more infrared radiation is trapped, and the more heat ation. Methane traps more than 20 times the amount of
is retained at the Earth’s surface. CO2 is called a green- heat per molecule than a carbon dioxide does, while
house gas because, like the glass in a greenhouse, it lets nitrous oxide absorbs 270 times more. Carbon dioxide
visible light energy enter the atmosphere but prevents emissions are responsible for about two-thirds of the
longwave radiation from escaping, thereby warming the anthropogenic intensification of the greenhouse effect,
Earth’s surface. The greenhouse effect is the warming with most of the remainder due to CH4 (about 19%),
of the Earth’s surface and atmosphere due to the reten- CFCs (10%), and N2O (6%).
tion of heat by the greenhouse gases in the atmosphere.
The greenhouse effect is natural and is necessary for the
continuation of life on Earth—without it, the average
temperature of the Earth’s surface would be –18°C (0°F) (A)
instead of its actual 14°C (57°F). However, most popu- 1750
lar usage of the term “greenhouse effect” refers to the
rise in temperatures resulting from the increases in
1500
Concentration CH4 (ppbv)

atmospheric CO2 and other greenhouse gases caused by
human activities. It is the increase in the greenhouse
effect that is leading to global climate change. 1250
Water vapor is the most predominant and effective
greenhouse gas, followed by CO2. Both occur in the 1000
atmosphere naturally. Because it is produced in the
largest quantities (after water vapor), carbon dioxide is 750
the most important greenhouse gas currently increasing
due to human activities, but it is unfortunately not the
500
only one. A variety of other gases also contribute to glob- 3000 2000 1000 700 500 300 200 100 0
al warming. (Anthropogenic effects on water vapor are Time (years ago)
complex and not fully understood; they are not thought
to be a major contributor to the increase in the green- (B)
house effect.) In addition to CO2 and water vapor, four 400
of the most important greenhouse gases are methane
Emissions (1012 g/yr CH4)
350
(CH4), nitrous oxide (N2O), ozone (O3), and chlorinat-
ed fluorocarbons (CFCs). The first three occur naturally, 300
but humans have greatly increased their concentrations 250
in the atmosphere. 200
Anthropogenic methane (Figure 22.4) is released in
150
large quantities by leakage during the drilling and trans-
portation of natural gas, by anaerobic prokaryotes in cat- 100
tle rumens, and in decomposing trash in landfills and 50
dumps. Increases in nitrous oxide are largely a by-prod- 0
uct of commercial agriculture, as it is released by the 1850 1900 1950 2000
microbial degradation of nitrogen-containing fertilizers; Year
nylon production is another substantial source of N2O. Figure 22.4
Increases in atmospheric concentrations of methane, a major
Ozone, O3, is produced in the lower atmosphere (the tro- greenhouse gas. (A) Atmospheric methane (CH4) levels (in
posphere) through a series of chemical reactions where parts per billion by volume, ppbv) during the past 3000
air pollution is high, particularly as volatile hydrocar- years (note that the x-axis expresses time on a logarithmic
bons and carbon monoxide in the presence of high NO. scale). Concentrations were measured from air bubbles
Some ozone is also produced naturally from compounds trapped in ice cores from Greenland and Antarctica, or from
historic and current air samples. (After Schlesinger 1997;
emitted by vegetation and by forest fires. CFCs are arti- Cicerone and Oremland 1988.) (B) Global anthropogenic
ficial compounds used primarily as refrigerants and emissions of methane (CH4,, in 1012 g/year) from all sources
propellants. While recent international treaties have dra- from 1860 to 1994. (After Stern and Kaufmann 1998.)
Many people, incidentally, confuse the intensifica- Global Climate Change: Evidence
tion of the greenhouse effect with the problem of the The evidence is clear that the concentrations of CO2,
ozone hole in the atmosphere. They are two different— methane, and certain other greenhouse gasses have been
although related—problems. The ozone hole is the loss increasing in the atmosphere. Predictions based on
of parts of the protective layer of ozone high in the strat- physics say that these increases will result in a warmer
osphere (the part of the atmosphere from 10 to 50 km Earth. But what is the evidence, if any, that Earth is actu-
above the surface), caused primarily by a class of artifi- ally warming, or that any other climate changes have
cial compounds called halogenated hydrocarbons resulted from the rising concentrations of greenhouse
(including CFCs and related chemicals). Ordinarily the gasses? Weather and climate are complex phenomena,
stratospheric ozone layer absorbs much of the high-ener- and natural variation can confound and mask general
gy ultraviolet radiation coming from the sun, prevent- trends (see Chapter 18), so sifting out the effects of any
ing it from reaching the Earth’s surface. Halogenated particular factor can be difficult.
hydrocarbons destroy ozone in the stratosphere, allow- That is why it has taken so long to nail down the
ing more ultraviolet radiation to reach the surface; this effects of CO2 on past and future climate change. By now,
effect has been particularly severe at high latitudes in however, the weight of evidence is so great that there is
the southern hemisphere. The increase in ultraviolet little scientific doubt remaining that the Earth is rapidly
radiation poses a risk of damage to many organisms, becoming warmer even if the exact magnitude is still
including humans. This phenomenon contrasts with the uncertain. We now take a brief look at some of the dif-
intensification of the greenhouse effect, which is caused ferent lines of scientific evidence that have been amassed
by a buildup of heat-trapping gases in the atmosphere, on anthropogenically caused global climate change.
with consequent increases in global temperatures. The Over the past hundred years, the average tempera-
ozone hole is connected to the greenhouse effect and the ture of the Earth’s surface has increased by 0.6°C (±
problem of global warming specifically because CFCs 0.2°C). Almost half of this change (0.2–0.3°C) has
act both to destroy stratospheric ozone and as green- occurred over the last 25 years, and temperature increas-
house gasses, and more generally because both are es may now be accelerating (Figure 22.5). Since 1980, the
anthropogenic changes in Earth’s atmosphere. Earth has experienced 17 of the 18 hottest years of the
0.9
Land and ocean
0.6
0.3
Figure 22.5
0.0
Temperature increases over the
past 120 years, expressed as dif- –0.3
ferences between annual values
and long-term mean tempera- –0.6
Deviation from mean temperature (°C)
tures (°C) for the entire Earth 0.8

(terrestrial and oceanic), the Ocean
world’s oceans, and the world’s 0.5
land surfaces. Global tempera-
0.2
tures over the past 20 years were
consistently above the long-term –0.1
averages, and most of the hottest
years on record occurred recent- –0.4
ly. Temperature increases have 1.2
been greater on land than in the Land
oceans, although temperatures 0.9
are consistently increasing for
both. The long-term mean tem- 0.6
peratures were calculated from 0.3
data recorded at thousands of
worldwide terrestrial and ocean- 0.0
ic observation sites for the entire –0.3
period of the data, and by inter-
polation for areas that were not –0.6
monitored. (Unpublished data of
T. C. Peterson, U.S. National Cli- –0.9
matic Data Center/NOAA, –1.2
October 2001.) 1880 1900 1920 1940 1960 1980 2000
Year
twentieth century. Some regions have experienced much CO2 levels, and global warming? Probably the first sci-
greater warming, while others have not warmed as entist who published the prediction that CO2 from the
much, or have even cooled. The greatest warming has combustion of fossil fuels would lead to global climate
occurred in northern latitudes (between 40° N and 70° change was the Swedish chemist Svante Arrhenius in
N across Eurasia and North America). Summer tem- 1896. It was first confirmed that CO2 concentrations in
peratures in the Northern Hemisphere in recent decades the atmosphere were actually rising by the 1930s, and by
have been the warmest in the past thousand years. the 1950s there were precise measurements of the mag-
Nighttime temperatures are increasing more than day- nitude of the increase. Most scientists had accepted the
time temperatures, leading to reduced day/night dif- reality of global warming by the 1990s. A small number
ferences in many places. of scientists do not accept the generally held conclusion
In addition to direct measurements of increasing that anthropogenic rises in CO2 concentration are respon-
temperatures, indirect data also show evidence of gen- sible for global temperature increases. There is, howev-
eral global warming. Glaciers are retreating, snow cover er, agreement that adding greenhouse gases to the atmos-
is decreasing and melting earlier in the spring, and Arc- phere must ultimately warm it , and a large majority of
tic sea ice has declined both in thickness and in the area scientists today believe the evidence is strong enough
it covers. Permafrost in the Arctic is melting. In the Ant- to link recent warming to anthropogenic changes in the
arctic, the rate at which the ice shelves that ring much concentrations of greenhouse gases.
of the continent are breaking up appears to be increasing.
There is evidence of both surface melting and warming Global Climate Change: Predictions
below floating ice, and average air and water tempera- We have just taken a look at some of the actual changes
tures around the coastlines of Antarctica are becoming in the Earth’s climate that are associated with the inten-
higher as well. Sea levels globally have already risen sified greenhouse effect. What kinds of changes are pre-
between 10 and 20 centimeters in the past hundred years, dicted for the future? Climatic scientists use a variety
largely due to the expansion of water when it warms. of direct and indirect measurements to document cli-
Temperature is not the only aspect of global climates mate changes. Predictions for what future climates
that is changing. The frequency and intensity of El Niño might be like are made by constructing and refining
Southern Oscillation events (see Chapter 18) also complex computer simulation models (Box 22A).
appears to have increased over the past few decades, The popular reaction to evidence for global warm-
with an abrupt increase in the mid-1970s that has per- ing sometimes seems to be that some warm winter days
sisted since then. ENSO events have been occurring for in temperate regions would actually be pleasant and
at least hundreds of years (and maybe far longer). How- welcome. What could be wrong with getting to play ball
ever, it is hypothesized that recent global warming, by or wash your car outside in the middle of winter? To
pumping heat energy into the surface of the oceans, will answer that question, we need to consider more closely
intensify their effects. Although some scientists have pre- what global climate change may entail.
dicted that a consequence of global warming will be a A variety of changes are predicted to result from the
global increase in extreme weather, there is presently no greenhouse effect. Averaged across the entire globe, tem-
evidence for general increases in extreme events or peratures are likely to rise by 1.4–5.8°C or more over the
weather variability across the entire globe. next hundred years. To get an idea of how large such a
At a regional scale, however, there has been an change would be, consider that during the most recent
increase both in the variability of weather conditions and glacial maximum 20,000 years ago, temperatures were
in extreme events. It has not been firmly demonstrated only about 9°C cooler globally than they are today (see
that there is a tie between these ENSO-related increas- Figure 21.8). Or consider that the average temperature
es and global warming, but there are some reasons to over the year is only 3°C higher in Milan, Italy, than in
believe that there may be a connection between the two. London, England, and 4°C higher in Charlotte, North
There is evidence suggesting that heavy rainfall events Carolina, than in New York City. Clearly, what might
have become more severe regionally, particularly at seem to be a small average temperature increase can
higher latitudes in the Northern Hemisphere. Parts of mean a very different climate.
Africa and Asia have suffered from more frequent and But these temperature increases will not be evenly
severe droughts, associated in part with shifts in ENSO spread across the globe. Changes in atmospheric and
patterns. ENSO is linked to other regional circulation oceanic circulation patterns are forecast to affect tem-
phenomena that may interact with global warming to perature, rainfall, and other climatic factors differently
further change climates in complex ways that are not yet at different locations. Land areas will probably warm
fully understood. much more than oceans, and higher latitudes in the
How long have we known about the greenhouse Northern Hemisphere will warm much more than other
effect, changes in it as a result of increasing atmospheric places, particularly in the winter. Summer warming is
BOX 22A
Modeling Climate
How do scientists know what the val- entists use complex computer simula- have revealed greater variation in radi-
ues for temperatures, precipitation, and tion models, called General Circulation ation fluxes between Earth’s surface
other weather and climatic variables Models or Global Climate Models and the atmosphere than was previ-
were in the past, particularly the distant (GCMs), that incorporate existing ously thought to exist, particularly in
past? There are both direct records of weather and climate data, as well as the Tropics. Differences in cloudiness
these variables—measurements that what is known about the major process- appear to account for this variation;
have been taken and recorded around es that determine climate, to predict clouds both increase reflection of
the world—as well as indirect records climate change. These climatological incoming solar radiation (a cooling
of various kinds. Reasonably good models have been constructed using effect), and reduce infrared radiation to
instrumental records have been kept in largely physical rather than biological space (a warming effect). A complex set
many places for temperature, rainfall data, in contrast to the ecological mod- of feedbacks between Earth’s surface
and snowfall, humidity, and air pres- els discussed in Chapter 15, although and the atmosphere can affect the
sure at Earth’s surface for at least a they can incorporate some effects of degree of cloudiness.
hundred years, but some locations are vegetation. The predictions based on GCMs are
missing data entirely or for earlier Once a model is constructed, changes uncertain to various extents, due to lim-
years. For recent years, data from satel- in some factors—such as CO2 levels in itations in what is known as well as
lites and high-atmosphere weather bal- the atmosphere—can be introduced to uncertainty about what human actions
loons exist. There are also excellent see how the climate might be affected in will be taken (for example, whether
recent data for global ocean tempera- different parts of the globe. Many sci- CO2 emissions will be reduced). How-
tures. Physical, biological, and chemi- entists must work together to construct ever, these models have been continu-
cal indicators offer insight into climate a GCM, and only a small number of ally refined and improved, and they are
and weather further back in time (see major GCMs exist. These models incor- becoming increasingly accurate and
Chapter 21). Some of these indicators porate large amounts of known data and detailed. They offer a reasonable set of
include measurements of tree rings make predictions about future tempera- predictions that will help us to antici-
from trees that are hundreds or thou- tures, rainfall, snow cover, sea level, and pate changes in global climate.
sands of years old, sediments at the bot- other climatic factors.
tom of lakes and oceans, air trapped in Recent research in this area, for
Some of the elements in a Global Climate
ancient ice cores, and annual density example, has included efforts to incor- Model (GCM). The model shown incorporates
bands in long-lived corals. porate more accurate satellite data on an atmospheric and an oceanic component.
How do scientists make predictions Earth’s radiation balance into GCMs (Based on information from U.S. Global
about future climates? Atmospheric sci- (Wielicki et al. 2002). These attempts Change Research Program Office.)
expected to be particularly great in central and northern more frequent and more intense (more rainfall per event)
Asia. More hot days and periods of extended intense over most land areas. Mid-continental areas will become
heat are predicted over all land areas, and average drier, with major implications for the world’s “bread-
humidity is expected to increase. Cold waves will basket” regions (see Chapter 19). Both mean and peak
become fewer and less intense. Glaciers and ice caps will rainfall during tropical cyclones are expected to increase
continue their dramatic retreats. a great deal, although the frequency of cyclones is not
Large changes in precipitation levels and storm predicted to be greater. The effects on mid-latitude storm
severity are also predicted. Large increases in yearly intensities are still uncertain.
rainfall are expected at high latitudes across the North- Sea levels are predicted to rise between 11 and 77
ern Hemisphere. Winter precipitation is predicted to centimeters globally, due largely to the continued ther-
increase in tropical Africa and at mid-latitudes in the mal expansion of water and, to a lesser extent, to melt-
Northern Hemisphere, and summer rainfall is predicting glaciers. If this rise occurs as predicted, it will result
ed to increase in southern and eastern Asia. Rainfall will in widespread coastal flooding. The consequences of this
decrease in Australia, Central America, and southern flooding will be felt from low-lying island nations such
Africa, particularly in winter. Rainfall patterns will shift, as the Maldives to highly populated coastal regions from
becoming more variable from year to year almost every- Bangladesh to China to Florida. Most coastal cities—
where, with increased flooding in some years and which contain a large proportion of the world’s popu-
drought in others. Extreme rainfall events will become lation—will also be threatened.
Incoming solar
energy ATMOSPHERIC GCM
Outgoing heat
energy
Cirrus clouds
Transition from
Atmospheric model layers

solid to vapor
Stratus clouds Atmosphere
(Temperature, winds,
and precipitation)
Precipitation and Cumulus clouds
Snow cover evaporation Stratus
Evaporative clouds
and heat energy
exchanges Precipitation and
evaporation
Runoff Winds
Soil Sea ice
moisture Heat and salinity
Ocean exchange
(Currents, temperature,
Geography and salinity)
Land surface
(Snow cover, vegetation, reflectivity,
topography, and land use) Vertical
overturning
OCEAN Ocean model
Ocean bottom GCM layers
topography
Biotic Consequences of Climate Change persers will clearly suffer if environmental changes
The consequences of regional and global changes in cli- reduce or eliminate the animals they depend on. Climate
mate are likely to have large, though only partially pre- change is predicted to alter ecosystem cycling for many
dictable, effects on plants. The details of these changes major nutrients. Soil respiration and nitrogen mineral-
are not yet known, but parts of the bigger picture are ization rates, for example, have been shown to increase
beginning to come into focus. Some species, ecosystems, substantially when ecosystems are warmed experimen-
and vegetation types will benefit from these changes and tally. Some models of climate change predict increases
increase, while others will lose ground, declining or dis- in fire frequency and severity in many ecosystems.
appearing (Figure 22.6). While there are models that The distributions of many plant species are expect-
make specific predictions about particular regions, we ed to change, resulting both in shifts in the locations of
will focus here on the general nature of the changes we plant communities and in different combinations of
can expect. species growing together in the communities of the
Individual plant physiology, growth, reproduction, future. The increases in productivity resulting from high-
and allocation are all likely to be affected by changing er CO2 levels may be amplified in some ecosystems by
climates. Interactions of plants with other species—com- warmer climates and longer growing seasons. Other
petitors, pollinators, herbivores, and pathogens—will ecosystems will be affected by increased drought, which
probably change as well. For example, plants that are will cancel out the positive effects of higher CO2 levels
dependent on highly specialized pollinators or dis- on photosynthesis and growth. Changes in temperatures
Figure 22.6 (A) Current biome distribution

(A) Current distribution of the
world’s biomes, as derived
from current climatic data
using the MAPSS model
(Mapped Atmosphere Plant
Soil System). (B) Predicted
altered distribution of biomes
75 to 100 years in the future,
based on one set of scenarios
for global CO2 increases.
According to this model, the
early stages of global warming
will lead to increases in the pro-
ductivity and density of forests
worldwide, as increased carbon
dioxide levels act as a fertilizer.
Continued elevated tempera-
tures, however, will strain
water resources, in time pro-
ducing drought-induced stress
and broad-scale die-offs of veg-
etation, with associated increas-
es in wildfires. (Based on mod-
els by Neilson 1995 and data
from http://www.fs.fed.us/
pnw/corvallis/mdr/mapss/)
(B) Predicted distribution, 2070 – 2099
Tundra Tropical broadleaf forest

Taiga/tundra Savanna/woodland
Boreal conifer forest Shrub/woodland
Temperate evergreen forest Grassland
Temperate mixed forest Arid lands
and rainfall regimes are expected to have major conse- by 70% over the oldest accurate historical records, appar-
quences for agricultural regions. Some highly produc- ently as a result of colonization by lower-elevation
tive areas are predicted to become unproductive, while species; at the same time, glaciers there have lost 30–40%
some regions that are currently marginal agriculturally of their surface area. Likewise, butterflies, birds, and
will become highly productive. For example, there are other species that interact closely with plants have been
expected to be declines in the agricultural productivity moving upward in elevation and poleward (reviewed
of the U.S. Midwest and increases in that of the grain belt by Hughes 2000).
of central Canada. The crops that are usually grown in Reductions in available habitat as elevation increas-
particular areas may change substantially, with grain es, coupled with species being forced upward by warm-
belts shifting and frost-sensitive crops being grown at ing, are likely to result in at least regional extinctions of
higher latitudes. some alpine species. Changes in the vegetation of the
As we have seen, atmospheric CO2 levels have fluc- San Francisco Mountains in northern Arizona, for exam-
tuated widely over the course of Earth’s history (see ple, were modeled by Francis (1999). Currently, this veg-
Figure 21.6), and global and regional climates have fluc- etation varies from sage shrublands at the base of the
tuated widely as well (see Figure 21.8). Biomes and the mountains, through Pinus ponderosa (Ponderosa pine,
species inhabiting them have changed dramatically Pinaceae) forests at mid-elevations, to spruce-fir forests,
over the eons. Many plant species have moved as cli- to tundra at the summit. The predicted effects of climate
mate has changed. After the most recent glaciation in change include a substantial spread of the sage shrub-
North America, for example, species moved northward lands, the movement of ponderosa pine forests to high-
to recolonize areas that became deglaciated as temper- er elevations, and the loss of spruce-fir forests and tun-
atures warmed (see Figure 21.10). The biggest difference dra (Figure 22.7).
between the global changes of the past and those occur- Range shifts, range contractions and expansions,
ring now is that the current changes are happening at and changes in phenological patterns are likely to occur,
much faster rates—far greater, ecologists predict, than as changes take place not only in average temperatures,
the rates at which many species can migrate or adapt. but in the length of growing seasons, time of snowmelt
Many of the past changes were on a regional or conti- or even the existence of snowpack, the duration of
nental scale, rather than the current global scale, so drought and extreme heat events, fire regimes, and soil
refuges may be scarcer today: all species everywhere moisture availability at different times of the year. Fur-
are more likely to experience at least some of the effects thermore, the ecology of organisms at all levels, from
of climate change. populations to ecosystems, will be affected by changes
Habitat reduction and fragmentation are likely to in many of these factors at the same time. Due to the
make it more difficult for species to migrate in response combined effects of multiple stresses, species currently
to climate change. Parks and preserves that currently threatened or endangered are more likely to go extinct,
protect endangered populations, species, or communi- and additional species are likely to become endangered.
ties may become ineffective refuges: as climatic condi- Finally, changes in land forms are likely to have sub-
tions change, such refuges may no longer be as suitable stantial effects on vegetation at the landscape and
for those organisms or systems. If species are unable to regional levels and on the global persistence of particu-
reach new favorable habitat, or if habitat in newly suit- lar groups of species dependent on affected habitats. Sea
able environments is not protected, those species will level rises and increased rainfall have already caused
disappear. Species will almost certainly not move togeth- many seacoasts to experience increased flooding, storm
er, but will migrate (or fail to migrate) individually (see damage, and erosion (Figure 22.8). These processes are
Chapter 21). As a result, many vegetation and commu- likely to accelerate over the coming decades. Freshwa-
nity types may disappear and be replaced with new ter marshes close to coastal areas may be affected by salt-
species combinations having new properties. water incursion. Highly diverse and productive coastal
Arctic and boreal forest species, facing the greatest ecosystems, such as salt marshes and mangrove forests,
degree of climate change, may find themselves with no will be subjected to negative and sometimes disastrous
place to go as warming shrinks the extent of suitable effects (including erosion and submersion). Small islands
habitat. Alpine species forced to move to higher altitudes close to sea level may disappear entirely as oceans rise,
may face similar problems. Many plant species are taking their endemic and rare plant and animal species
reported to be moving to higher elevations in moun- with them (as well as the homes of the people living
tainous areas, including trees that are becoming estab- there). Even far from coastlines, increases in storm inten-
lished above current tree lines. In Europe, current plant sity, sometimes coupled with the effects of deforestation,
species richness on 30 peaks in the Alps has increased are expected to result in major disturbances to plant
(A)
Figure 22.7
(A) The San Francisco peaks are a volcanic range in
northern Arizona. The highest point is at 3850 meters
elevation. (Photograph by S. Scheiner.) (B) Current and
predicted vegetation patterns on the San Francisco
Mountains. Currently, a small area of alpine tundra
exists at the highest elevation with spruce-fir forests at
slightly lower elevations. Intermediate elevations are
dominated by ponderosa pine forests. The lower eleva-
tions are covered by sage shrubland and pinyon-
juniper woodland. The predicted changes are based on
a climate change model (see Box 22A). Rising tempera-
tures will result in the spread of the sage shrublands,
the movement of ponderosa pine forests to higher ele-
vations, and the loss of spruce-fir forests and tundra.
(After Francis 1999.)
(B) Current vegetation Predicted vegetation
Grass/bare soil
Shrubland
Pinyon-juniper woodland
Ponderosa pine forest
Aspen forest
Spruce/fir forest
Alpine tundra
Mobile
New Orleans
Houston
Tampa
Below 1.5 meters

1.5 – 3.5 meters Miami
Above 3.5 meters
150 km
Figure 22.8
fpo
Lands vulnerable to sea level rise along the Gulf Coast of the
southern United States. The map shows land below the 1.5-
timeters since 1929. Approximately 58,000 km2 of land along
the Atlantic and Gulf coasts lie below the 1.5-m contour.
m and 3.5-m contours (elevations above sea level in 1929), Those above the 3.5-m contour would be vulnerable after a
which are vulnerable under different models of future cli- period of several centuries with a greater sea level rise. Ris-
mate change. A rise in sea level sufficient to inundate lands ing sea levels will threaten, drastically alter, or destroy many
at the 1.5-m contour at high tides appears to be most likely coastal ecosystems, such as highly productive salt marshes
to occur in the next 120 years, but has a 1% chance of occur- in many areas, and the Florida Everglades in the southern
ring in the next 60 years. Sea level has risen about 20 cen- part of Florida. (After Titus and Richman 2001.)
communities by causing flooding in river valleys and 7000
Carbon emitted (1012 g C/yr)

landslides, mudslides, and avalanches on steep slopes.
6000
Anthropogenic Effects 5000
on the Global Carbon Cycle 4000 From fossil fuels

and industrial
The initially pleasant image you may have had of glob- 3000 sources
al warming meaning milder days in winter may begin
to look less appealing upon seeing more of the details of 2000
the changes that may be in store. Where does the anthro- 1000 From land-
pogenic carbon responsible for these changes originate? use change
We mentioned earlier that increases in atmospheric CO2 0
1750 1800 1850 1900 1950 2000
are caused mainly by two factors: deforestation (the
destruction of forests) and the combustion of fossil fuels Year
(Figure 22.9). Where are these things happening, and Figure 22.9
who is causing them? To fully investigate these ques- Carbon emissions to the atmosphere (in 1012g C/year) from
tions, we would need to discuss not only the science fossil fuel combustion and other industrial sources (green
involved, but also the social, political, and economic line) from 1750 present, and carbon flux to the atmosphere
from land-use change (primarily deforestation; black line)
bases for anthropogenic CO2 emissions. These topics can from 1850 to the present. (Data from Houghton and Hackler
be treated to only a limited extent in a plant ecology text- 2001; Marland et al. 2001.)
book. Here we mention some of the factors that may be
involved in order to suggest some of the possible direc-
tions you may want to take in your future thinking and
reading on this issue.
If new trees are planted where a forest is logged, the
Deforestation new trees will accumulate carbon and net ecosystem
While trees naturally accumulate carbon in wood and production (see Chapter 15) is likely to be positive. If the
release it when they eventually die, human activities are forest is replaced by buildings and pavement, that obvi-
dramatically accelerating the rate at which this is hap- ously will not happen. Removing forests and replacing
pening. Cutting down a tree affects atmospheric CO2 them with agricultural lands, tree plantations, parking
levels in two ways. First, if the tree is not replaced with lots, or barren wastelands has enormous ecological
another tree, the CO2 it would have taken up will remain implications beyond the effects on the carbon cycle, and
in the atmosphere rather than being fixed. Second, if the we will briefly consider some of these implications
tree is burned or allowed to decompose, much of the car- toward the end of this chapter.
bon accumulated over its long life—which can be Two factors act directly to counter the emission of
decades or centuries for canopy trees—is released to the CO2 to the atmosphere caused by forest destruction: for-
atmosphere suddenly or over a relatively short time. est regeneration and higher photosynthetic rates due
(Herbaceous plants, in contrast, rapidly exchange car- to the “fertilization” effect of higher CO2 concentrations
bon with the soil and the atmosphere.) in the atmosphere. Forest regeneration has, in some
If trees are logged to build things that remain intact cases, been dramatic. Large parts of the northeastern
for a long time (a good violin, perhaps, or a building that United States, for example, were heavily logged and
lasts a long time, rather than a stack of bulk mail solici- cleared for farming in the eighteenth and nineteenth cen-
tations), then some of the carbon in the wood will be pre- turies. The abandonment of these farms led to the regen-
served in the structure for decades or even centuries, eration of large areas of secondary forest (see Figure
rather than being immediately released to the atmos- 13.1), sufficient to act as a substantial carbon sink. How-
phere (although the carbon in roots, branches, and leaves ever, these lands have recently once again become
that remain when the trunk is harvested will decompose sources rather than sinks for atmospheric CO2, both
quickly, even when the wood is used in something long- because many of these forest ecosystems have reached
lasting). If trees are logged to make newspaper, dispos- maturity and their rate of carbon accumulation is declin-
able chopsticks, or other short-lived items, the postdis- ing (see Chapter 15), and because of renewed forest
posal fate of those items (whether they are burned, clearing for both lumber and suburban development.
stored in a landfill, or recycled) will determine the fate Cutting trees for firewood and burning trees to clear
of that carbon. The fate of the organic carbon in the for- land have been anthropogenic sources of carbon input
est soil will depend on a number of factors, including to the atmosphere throughout human history. These
what happens to the land after the forest is logged. activities were sufficiently limited before about 1800 that
they had little or no measurable effect on atmospheric a geological blink of an eye. The extractable pool—the
CO2 concentrations (see Figure 21.6B). That is no longer portion feasible to get at—of carbon contained in Earth’s
the case, however, and the effects of these activities are fossil fuels is about 4 × 1018 g C, which is about one-tenth
particularly severe for tropical forests. Today, about 1.1 of the total active pool of carbon at or near the Earth’s
× 1015 g C is emitted to the atmosphere from the burn- surface. The combustion of fossil fuels releases approx-
ing of tropical forests each year, and almost twice that imately 6 × 1015 g C per year. A little over half of this car-
amount is released from all tropical deforestation activ- bon remains in the atmosphere, adding about 1.5 ppmv
ities. A gross total of 3.4 × 1015 g C is produced by trop- CO2 each year to the atmosphere. Some of the rest of the
ical forest destruction if decomposition of uncombusted CO2 from fossil fuels enters the oceans, while some
material is included (Fearnside 2000). The total net emis- remains unaccounted for. Not until the middle of the
sion of carbon to the atmosphere resulting from tropical 20th century did the carbon released to the atmosphere
land use changes (particularly destruction of forests) is by the combustion of fossil fuels become greater than the
equal to approximately 29% of the total global anthro- carbon released by the cutting or burning of forests, but
pogenic emissions from all fossil fuel combustion and now it represents about 70% of all anthropogenic CO2
nontropical land use changes combined (Fearnside released.
2000). Tropical forest clear-cutting has been carried out CO2 emissions differ enormously among the coun-
both by people living in tropical countries and by cor- tries of the world, whether we look at the total amount
porations owned by people in developed countries. released into the atmosphere by each country or con-
Corporations based in Japan, Europe, and North Amer- sider how much, on average, each person in the coun-
ica are all active in tropical deforestation. try is responsible for producing (Figure 22.10). The high-
Not all deforestation has occurred in the Tropics, est total emissions per country come from the United
however. In many cases, developed countries have States, but China is a remarkably close second. U.S.
cleared large areas of their own forested land. An exam- emissions have been growing rapidly for the past hun-
ple is the temperate rainforests of the northwestern U.S. dred years, while those from China have been growing
and British Columbia in Canada, which have experi- very steeply since the late 1960s (although the most
enced large-scale deforestation and clear-cut logging in recent data for China suggest a potential slowdown).
the past 20 years (see Figure 19.11). (Where these old- The United States and China maintain these high emis-
growth forests are replanted or undergo secondary suc- sions by very different means. The United States has a
cession, the young trees will once again serve as sinks fairly large population and very high rates of energy
for CO2, although the removal of old-growth forests use per person. China has an enormous population, but
results in other ecological effects.) One important reason the amount of energy used on average per person is
why forest destruction is concentrated in developing quite modest (Figure 22.10C).
countries today is that forests in the wealthy industrial The top ten countries in total fossil fuel carbon emis-
countries have already been altered or destroyed. Both sions as of 1998, from the highest in descending order,
Europe and Japan, for example, have very few areas that are the United States, China, the Russian Federation,
are not heavily affected by human use. In Japan, virtu- Japan, India, Germany, the United Kingdom, Canada,
ally all lowland areas are either farmed or occupied by Italy, and Mexico. This list contains a mix of developed
human structures; on the mountainsides, there are exten- and developing countries in North America, Europe,
sive forests, but most of these are regularly logged. In and Asia. Most industrialized countries in Europe have
the Swiss Alps, livestock are grazed even on some of the shown dramatic increases in energy efficiency and, con-
highest peaks. In Spain, the highest elevations contain sequently, both decreased per capita emissions (such as
some of the greatest plant species diversity, not because those for Germany and Poland; Figure 22.10C and D)
those areas are inherently more species-rich, but because and decreased total emissions (Figure 22.10A and B).
more productive lowland areas are heavily used for agri- Modest per capita increases in countries with large and
culture. rapidly growing populations, such as India and Brazil,
over the past 15 years (Figure 22.10D) have led to steep
Fossil Fuel Combustion increases in those countries’ total emissions (Figure
The burning of fossil fuels provides the second major 22.10B). On the other hand, one could argue that, from
anthropogenic contribution to atmospheric CO2. The the standpoint of reducing human suffering, some coun-
carbon contained in fossil fuels accumulated over eons, tries actually produce too little CO2 from fossil fuels. The
as it was taken up from an ancient atmosphere by organ- tiny and erratic emissions of countries such as Congo are
isms now long extinct, over the course of millions of a reflection of disaster, war, deprivation, and human
years. Vast amounts of carbon in the form of fossil fuels misery. For other countries, low emissions may reflect
are being taken from deep in the earth and released in a tiny population, particularly one that is largely non-
(A) (B)
1500 300
U.S.A.
India
Total fossil fuel CO2 emissions × 103
Total fossil fuel CO2 emissions × 103

1250 250
1000 200
750 150 Poland
China Brazil
500 100
Colombia
Congo (× 10)
250 Germany India 50
0.0 0.0
1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000 1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000
Year Year
(C) (D)
6.0 3.5 Poland

U.S.A.
5.5
3.0
5.0
Per capita fossil fuel CO2 emissions
Per capita fossil fuel CO2 emissions
4.5 Germany
2.5
4.0
3.5 2.0
3.0
2.5 1.5
2.0
1.0 Colombia
1.5
India Congo
Brazil
1.0
China 0.5
0.5
India
0.0 0.0
1950 1960 1970 1980 1990 2000 1950 1960 1970 1980 1990 2000
Year Year
Figure 22.10
f
CO2 emissions resulting from the combustion of fossil fuels.
(A,B) Total emissions (in 1012 g C/year) for various countries
from 1800 to 1998. Note that the scales on the y-axes of the
industrialized, with few automobiles and trucks, little
industry, and largely nonmechanized agriculture.
two graphs are very different because the total amounts dif- Thus, the anthropogenic contribution to atmos-
fer by almost an order of magnitude; the data for India are pheric CO2 is highly unequal. Almost any way one
included in both graphs to show these scale differences. The
values for Congo were multiplied by 10 so that they would
measures them, U.S. emissions are disproportionately
be visible. (C, D) Per capita emissions (in 106 g C/year) for large. The combination of a fairly high per capita emis-
various countries from 1950–1998. Note that the scales on sion rate with a fairly large population makes U.S. emis-
the y-axes of the two graphs are different; data for India are sions the largest source of anthropogenic carbon input
included in both graphs to show these scale differences. to the atmosphere. The United States has contributed
(Data for Germany include combined totals for the former
German Democratic Republic and Federal Republic of Ger-
about 30% of the anthropogenic increase in atmospher-
many between 1946 and 1990.) (Data from Marland et al. ic CO2 to date, and currently adds 25% of the carbon
2001.) emitted per year. While only 5% of the world’s popula-
BOX 22B
Daily Human Activities and CO Generation 2
Every time you breathe you release CO2 phere a short time before by living The table on the right gives approx-
into the atmosphere. The amount varies plants. Human use of fossil fuels, how- imate amounts of CO2 released into the
from person to person, of course, ever, increases atmospheric releases of atmosphere by some common human
depending on an individual’s weight, CO2. activities. All of the numbers are gen-
activity level, fat:muscle ratio, and other The amount of CO2 removed from eral approximations, made using vari-
factors; but on average, in a year each the atmosphere by one mature tree ous assumptions; the amount of CO2
person respires 365 kg of CO2. annually is about 9 kg; based on this actually produced varies as conditions
But human respiration is not really figure, it takes 1,430 trees to use up the change. The purpose of these numbers
adding to atmospheric CO2. As with all CO2 produced by one sport utility vehi- is not to provide exact figures, but to
animals, the carbon respired comes cle of the largest class (e.g., Lincoln make general comparisons among dif-
from carbon removed from the atmos- Navigator). ferent activities.
tion lives in the United States, it accounts for 25% of world’s energy. Furthermore, U.S. rates of fossil fuel con-
the global daily usage of petroleum (about 3 billion of sumption per unit of Gross National Product (a meas-
the more than 12 billion liters used globally per day), ure of economic productivity) per person are very high.
and about 25% of the world’s energy consumption. What is producing all of that anthropogenic carbon?
The current rate of CO2 emission in the United States Americans use almost 1.3 billion liters of petroleum per
is 5.43 metric tons of carbon per year per person. Can day in their personal vehicles. Americans now purchase
this be accounted for by factors such as the high stan- more light trucks (pickup trucks, sport utility vehicles,
dard of living in the United States, the need for home and minivans) than cars for personal transportation, and
heating during the winter in much of the country, or these light trucks on average use far more petroleum
transportation needs in the large geographic area it than cars do with consequently greater CO2 emissions
encompasses? Not readily. The U.S. per capita emis- (Box 22B). Transportation of all sorts (commercial as well
sion rate is far higher, for example, than that in Sweden, as individual) is responsible for about a quarter to a third
a colder country with a comparable standard of living. of the total U.S. emissions. The rest come from indus-
Sweden’s average annual emission rate is 1.50 metric trial and commercial sources (about half of the total) and
tons of CO2 per person, and, as in many European coun- residences (about a fifth of the total). Box 22B lists some
tries, this rate has declined steadily (from a peak of 3.13 common activities and their resulting fossil fuel carbon
in 1970), while emissions in the United States, after a emissions.
modest drop during the 1970s, otherwise continued to Different fossil fuels release very different amounts
climb (Figure 22.10C). U.S. emissions are also higher of carbon for each unit of energy generated—coal, for
than those in Australia (4.88) and Argentina (1.03), example, produces much more atmospheric carbon for
which, like the United States, have populations spread the same amount of energy yield than natural gas. Some
over large geographic areas. sources of electricity other than fossil fuels and the burn-
Average emission rates per person must be inter- ing of plant material produce no carbon when generat-
preted cautiously, however, for many reasons—for ing energy: nuclear power, hydroelectric power, solar
example, the fuel used by ships and planes visiting power, and wind and geothermal energy all release no
Aruba, a popular tourist destination, is considered to be carbon. (They may cause other environmental problems,
part of the per capita fossil fuel use by the people of that however.)
small island country, leading to a figure deceptively sim- It is easy to see from Box 22B that even modest
ilar to that of the United States. increases in per capita fossil fuel use can have dramatic
One argument in defense of the U.S. CO2 emission consequences for a country’s total emissions if the coun-
rate is that although that country is responsible for a try has a large population. If everyone in China kept just
large proportion of the world’s anthropogenic carbon one more light bulb on for 16 hours a day, for example,
emissions, it also provides the greatest number of goods 364 billion (or 364 × 109) kg more carbon would be added
and services to world markets. But the European Union to the atmosphere per year. If every family in India and
produces almost as great an amount of goods and serv- China had a refrigerator powered by fossil fuel-gener-
ices while consuming substantially less (16%) of the ated electricity—surely a modest and reasonable
Approximate amounts of CO2 produced per year by various human activities
Activity Annual CO2 production
Production of 100 kilowatt hours of electricity

Using natural gas 64 kg (141 lbs)
Using oil 84 kg (186 lbs)
Using coal 104 kg (230 lbs)
Using hydroelectric power 0 kg (0 lbs)
Common human activities and appliancesa
60 W incandescent light bulb (16 hours a day, 365 days) 364 kg (803 lbs)
Pentium III computer and 17-inch color monitor (on 24 hours a day, 8,232 kg (18,148 lbs)
no “energy saver” features)
Home heating/hot water using natural gas (based on average usage per U.S. home 4,113 kg (9,068 lbs)
in U.S.; varies with ambient and indoor temperature, etc.)
Home heating/hot water using fuel oil (based on average usage per home 9,525 kg (21,000 lbs)
in the northeastern U.S.)
Color TV (on 8 hours a day, which is average for U.S. households) 848 kg (1,869 lbs)
Refrigerator (average size, 10 years old) 939 kg (2,070 lbs)
Refrigerator (0.6 m3/20 cu ft, new, energy-efficient) 522 kg (1,150 lbs)
Clothes dryer (52 uses) 261 kg (575 lbs)
Toaster (365 uses) 20 kg (44 lbs)
Single round trip, New York to San Francisco on commercial airline 330 kg (728 lbs)
Vehicle emissions: Conventional gasoline enginesb
Small car (4 cylinders, 1.7 L engine; e.g., Honda Civic, Toyota Corolla) 5,171 kg (11,400 lbs)
Family vehicle (6 cylinders, 3 L engine; e.g. Ford Taurus wagon) 8,074 kg (17,800 lbs)
Minivan (6 cylinders, 3.3 L engine; e.g., Dodge Caravan) 8,528 kg (18,800 lbs)
Full-size van (8 cylinders, 5.4 L engine; e.g., Ford Econoline) 11,521 kg (25,400 lbs)
Sports car (8 cylinders, 5.7 L engine; e.g., Chevrolet Camaro) 8,255 kg (18,200 lbs)
Midsize SUVs (sport utiltiy vehicle; 4WD, 6 cylinders, 4 L engine; e.g., Ford Explorer) 9,344 kg (20,600 lbs)
Largest SUVs (4WD, 8 cylinders, 5.4 L engine; e.g., Lincoln Navigator) 12,973 kg (28,600 lbs)
Vehicle emissions: Electric vehiclesb,c
Ultra-compact (Ford Th!nk) 4082–5008 kg
(9,000–11,040 lbs)
Midsize SUV (Ford Explorer; U.S. Postal Service vehicles) 7,080–8450 kg
(15,600–18,630 lbs)
Vehicle emissions: Compressed natural gas vehiclesb
Small car (Honda Civic, 4 cylinders, 2.2 L engine) 4,536 kg (10,000 lbs.)
Midsize car (Toyota Camry, 4 cylinders, 2.2 L engine) 5,715 kg (12,600 lbs)
Luxury car (Ford Crown Victoria, 8 cylinders, 4.6 L engine) 7,983 kg (17,600 lbs)
Full-size van (Ford Econoline E250, 8 cylinders, 5.4 L) 10,977 kg (24,200 lbs)
a
Electricity is assumed to be generated by a coal-fired power plant, unless otherwise noted.
b
Assumptions: Based on 45% highway driving, 55% city driving, 15,000 annual miles (24,000 km), and automatic transmission.
c
Values assume electricity generated from coal or natural gas. Assumptions for CO2 equivalents are taken from the U.S. DOE GREET
(Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation) model, a “wells to wheels” model that includes energy
used in extracting the fuel (available on the Web from the U.S. Department of Energy), where electricity generated comes from liquid
and compressed natural gas (methane); several estimates are also provided where electricity is generated from less efficient coal, but
for those estimates only CO2 directly expended, and not CO2 equivalents, are included.
increase in the standard of living—well, you can project acids. High levels of NH4+ and NO3– in fog may also
the consequences (easily estimated; see Box 22B). directly harm trees.
Without question, there would be economic and While recent efforts to curb air pollution have been
other costs to changing our current patterns of fossil fuel quite successful in limiting SO2, nitrogen-containing pol-
consumption. But what are the alternatives? The con- lution has increased, becoming a greater problem. Nitro-
sequences of our current actions—or lack of actions— gen from anthropogenic sources enters the atmosphere
may be far more complex and expensive than many in various forms, primarily NOx, but also N2O5, HNO3,
have anticipated. and NH3. Humans have doubled the rate at which nitro-
gen enters the terrestrial nitrogen cycle each year since
Acid Precipitation the 1850s—and the rate continues to rise (Figure 22.11).
This additional nitrogen has strong effects on ecosystems
and Nitrogen Deposition at local to regional scales.
Two other human-induced phenomena affecting plants Anthropogenic nitrogen comes from a wide variety
are acid precipitation and nitrogen deposition. Like ris- of sources: automobile exhaust, industrial manufacture,
ing CO2 levels, these anthropogenic changes in biogeo- power plants, and agricultural fertilizers and animal
chemical cycles affect plants, but they have effects main- wastes. This broad range of sources is one of the main
ly on a regional scale. However, because they affect so reasons it is much more difficult to control nitrogen pol-
many different regions, and because their effects are so lution than sulfur pollution. Sulfur is produced by far
profound, we mention them here. Acid precipitation has fewer kinds of sources, primarily the combustion of coal
been a cause for environmental concern for decades, and other fossil fuels (particularly by power plants) and
while nitrogen deposition has gained widespread atten- some industries. However, the largest contribution of
tion only in recent years. The two are closely related to anthropogenic N to the terrestrial nitrogen cycle comes
each other. from the industrial fixation of atmospheric N for agri-
Acid precipitation refers to precipitation that is cultural fertilizer (currently 80 × 1012 g/year). Eventual-
more acidic than normal. Normal precipitation is slightly, substantial quantities of this N are volatilized or trans-
ly acidic because CO2 dissolved in water droplets in ferred as particulate matter (dust and other small
the atmosphere forms carbonic acid. Acid precipitation particles) into the atmosphere as NOx. The use of com-
is caused primarily by two common components of air mercial N fertilizers is increasing rapidly: the amount of
pollution, SO2 and NOx (which includes NO and NO2), N in chemical fertilizers applied from 1980 to 1990, for
that are produced by industrial and automotive emis-
sions. These gases undergo chemical reactions in the
U.S. annual nitrogen oxide emissions (1012 g N/yr) (

atmosphere to form sulfuric acid (H2SO4) and nitric acid 160 20
(HNO3). In solution, these acids produce highly acidi-
18
fied (very low pH) rain and snow. 140
Acid precipitation harms vegetation directly by 16
damaging foliage and other plant parts. Indirect effects 120
)
caused by acidification of the soil and surface water have 14

NO3– in ice (ppb) (
even greater negative consequences for plants. The 100 12

release of ions harmful to plant growth in acidified soil
may cause toxic reactions. More broadly, as soils become 80 10
more acidic, leaching of cations can reduce nutrient 8
60
availability or cause imbalances in the ratios of different
nutrients, with detrimental effects for many plant species 6
40
(see Chapter 4). Die-offs of Acer saccharum (sugar maples, 4
Sapindaceae) in Quebec, Canada, have been attributed 20
to deficiencies of soil K and Mg caused by acid rain. For- 2
est declines in central Europe have been attributed to 0 0
acid rain depleting levels of soil Mg relative to N. 1750 1800 1850 1900 1950 2000
)
Plants can also absorb SO2 and NO2 gases directly Year
from the atmosphere. In Germany, damage to conifer Figure 22.11
foliage seems to be due to direct uptake of anthropogenic Concentrations of nitrate (NO3–) deposited in ice in Green-
N, which results in leaching of Mg and other cations land (in ppb of ice) from the late eighteenth century to the
late twentieth century (green line), and annual emissions of
within foliage tissues. SO2 and NO2 can also be deposit- nitrogen oxide (NOx, in 1012 g/year) by fossil fuel combus-
ed in particulate form (dryfall) on leaves, soil, and other tion in the United States (black line). (After Schlesinger 1997
surfaces and then later react with rainfall to produce and Mayewski et al. 1990.)
instance, was at least as great as the amount applied iments in bogs at four sites in Western Europe showed
before that in all of human history. that increases in nitrogen deposition reduced the growth
In addition to its contribution to acid rain, nitrogen of the dominant Sphagnum moss species. This reduction
pollution returns to the ground as both wet deposition was a result of shading by increased cover of taller vas-
and dry deposition—dissolved in water and as partic- cular plants, and led to changes in community composi-
ulate matter. Current nitrogen inputs in the northeast- tion (Berendse et al. 2001). Similarly, the widespread
ern United States and central Europe, which suffer from replacement of Calluna vulgaris (heather, Ericaceae) in East
the greatest nitrogen deposition, are 10–50 kg Anglia, U.K., and in other locations by dense stands of
N/ha/year, which is between 5 and 20 times natural lev- grass seems to be due to the indirect effects of nitrogen
els. Nitrogen is limiting to plant growth in many com- deposition on competitive interrelationships in heath-
munities (see Chapter 4), so extra nitrogen in the form lands. Interestingly, grazing may reduce nitrogen-induced
of nitrogen deposition initially acts as a fertilizer, lead- competitive replacements by grasses in species-rich veg-
ing to increased growth. But excess nitrogen deposi- etation in the Netherlands (Kooijman and Smit 2001).
tion can exceed the ability of the system to use the nitro- It is difficult to separate the effects of nitrogen dep-
gen, leading eventually to nitrogen saturation of the osition on species losses from the effects of other envi-
system. Its effects are nonlinear: negative effects may not ronmental factors that have also changed over the same
be manifested as nitrogen continues to accumulate until time period. Simply observing nitrogen deposition and
a critical threshold is reached. At that point, the effects species losses occurring together does not mean that
of additional N rapidly become detrimental to many nitrogen deposition necessarily caused the declines.
plants, soil bacteria, and fungi. These effects include However, experiments have provided evidence of a
leaching of nutrients (such as calcium) from the soil as direct link between atmospheric nitrogen deposition and
a result of increased acidity, loss of fine roots, steep changes in species composition in heathlands and cal-
declines in mycorrhizal fungi and consequent reductions careous grasslands in the Netherlands. Aerts and
in plant uptake of phosphorus, and increased leaching Berendse (1988) used field experiments to demonstrate
of N into ground and surface water due to increased bac- that increased nitrogen input in amounts similar to that
terial nitrification and reduced plant uptake. of anthropogenic deposition led to declining cover of
Ecosystem responses to nitrogen deposition are Erica tetralix (Ericaceae; a dwarf shrub, formerly domi-
uneven. Different systems vary greatly in their respons- nant) in wet heathlands and increasing cover of Molinia
es, depending on soils, plant species composition, cli- caerulea (Poaceae; a perennial grass replacing Erica).
mate, and other factors. Areas in the United States that Changes in species composition in these communities
show evidence of nitrogen saturation and consequent have been attributed to the effects of increased nitro-
declines in forest ecosystems include the Catskill Moun- gen (from anthropogenic deposition) on competitive
tains north of New York City, the Great Smoky Moun- interactions in these naturally low-fertility habitats
tains in Tennessee, high-elevation forests in the eastern (Aerts 1999; Aerts et al. 1990). Together with other evi-
Colorado Rockies, and the San Gabriel and San dence, studies such as these have implicated atmos-
Bernardino Mountains outside of Los Angeles. pheric N deposition in changes in community compo-
Nitrogen deposition can also reduce plant tolerance sition and losses of plant biodiversity in the Netherlands,
of other stresses, including frost, periods of drought, United Kingdom, and elsewhere in Western Europe.
excessive heat, and herbivory. Changes in insect her-
bivory patterns in forests and other plant community Declining Global Biodiversity and Its Causes
types can be sensitive indicators of nitrogen deposition About 1,500,000 living species of all kinds have been sci-
effects. Changes in community composition, with the entifically described and named worldwide, including
disappearance of sensitive species, have been attributed about 250,000 plants. It is estimated that there are at least
to the effects of nitrogen deposition in various parts of 5 million living species that have not yet been described,
Europe. Mosses and lichens appear to be especially vul- and there may be many times that number. Only a hand-
nerable to nitrogen deposition, often showing signs of ful of species have been well studied—primarily those
stress first. In the United Kingdom, sensitive Sphagnum that are economically important (domesticated species,
moss species have declined in the Pennine Mountains diseases, and pests). Even among most of these “well-
due to the toxic effects of NH4+ and NO3– deposition, studied” species, little is known of their ecology. The
and the loss of many species of mosses and lichens in evolutionary relationships, genetics, physiology, and
Cumbria since the 1960s also appears to be a conse- other aspects of most nonhuman organisms are almost
quence of nitrogen deposition. entirely unknown. At the very least, there is a wealth of
Changes in community composition, and the losses understanding to be gained from greater knowledge of
of some species, may also be the result of changes in com- these living things and the communities with which they
petitive interactions due to higher nitrogen input. Exper- interact.
Biodiversity, short for “biological diversity,” refers 1.2

to all of the populations, species, and communities in a
defined area (which may be local, regional, continental, 1.0
or include the entire globe) and includes the range of
genetic variation in these living things. In contrast to
the more specific term “species diversity,” the term “bio- 0.8
Relative effect
diversity” was coined to emphasize the many complex
kinds of variation that exist within and among organisms
0.6
at different levels of organization. We can speak, for
instance, of the biodiversity of major human crop species,
which is currently greatly threatened. As a small num- 0.4
ber of commercial varieties of domestic crops have been
widely planted everywhere on Earth, many local vari-
0.2
eties have been abandoned. These local varieties, culti-
vated and selected sometimes over thousands of years
for geographically specific desirable traits, may possess 0
Changes in Climate Nitrogen Biotic Atmospheric
genes that enable them to withstand insects, drought, and land use change deposition exchange CO2
other stresses—genes that are missing in the more pro-
ductive but potentially vulnerable commercial strains. If Figure 22.12
we fail to maintain these ancient crop races, these invalu- Average estimated relative effects of the most important cur-
able genetic storehouses will be lost forever. rent threats to global biodiversity, projected for the next 100
years. Land use changes (such as habitat loss and deforesta-
A range of anthropogenic factors currently threat- tion) are predicted to have the greatest harmful effects on
ens biodiversity from local to global scales. It has been biodiversity, followed by climate change, nitrogen deposi-
difficult to accurately assess current rates of outright tion, the incursion of invasive exotic species and other biotic
species extinction except in some very limited cases. It exchanges, and direct effects of rising atmospheric CO2 lev-
is even more difficult to accurately quantify species’ els. (After Sala et al. 2000.)
declines, and impossible to obtain precise numbers on
overall rates of loss of genetic or other biodiversity. Not
surprisingly, we have better data for large, attractive, (see Chapter 17). Figure 22.13 is a graphic illustration
obvious, and useful organisms than for others—we have of forest loss and fragmentation in Central America.
much better population data for grizzly bears and sugar Skole and Tucker (1993) quantified changes in forest
maples, for example, than for mycorrhizal fungi, even fragmentation in the Brazilian Amazon over a ten-year
though mycorrhizal species are critical to the existence period (1978–1988) using satellite images. By the end of
of many plants (see Chapter 4). One of the difficulties that period, almost three times more forested area con-
is that most people would not be aware of or recognize sisted of forest fragments—patches of forest less than
the decline of important mycorrhizal fungi. Likewise, 100 km2, surrounded by cleared land—than at the
declines or even extinctions of many nondomesticated beginning. The size and shape of cleared patches also
plant species would be likely to go unnoticed in many changed dramatically during that period due to trans-
parts of the world. formations in the economy and human demographics.
Clearings changed from smaller and more irregular
Habitat Fragmentation and Loss patches to larger and more continuous areas, with neg-
Threats to global, regional, and local biodiversity include ative implications for the viability of the forest ecosys-
a range of factors. While the specific threats to biodi- tems (Peralta and Mather 2000).
versity differ among biomes (Sala et al. 2000), the most These changes are important because habitat frag-
important factor threatening plants and plant commu- mentation can have many different negative effects (Lau-
nities globally is changes in land use, which are result- rance 2000a; see Chapter 17). For example, fragmenta-
ing in the destruction, degradation, and fragmentation tion can change forest microclimates as wind, light
of habitats (Figure 22.12). Besides being entirely de- levels, temperatures, and runoff increase at the edges of
stroyed, habitats can be degraded by many factors, the forest, while humidity decreases. Because of the
including losses of critical species, damage to the soil diminished area of the forest environment and the
(such as erosion or nitrogen deposition), overgrazing, altered conditions within it, deep-forest species can suf-
and changes in disturbance regimes caused by humans fer reduced survival and reproduction, while popula-
and their domesticated animals. Fragmentation of nat- tions of species adapted to edge conditions, such as
ural communities by roads, agriculture, or human set- many invasive plants, can grow. Increases in herbivory
tlements is changing many natural habitats worldwide and disease and declines or extinctions of some native
Figure 22.13
Forest fragmentation in Central America between 1950 and
1985. Deforestation and fragmentation have increased in the
Belize years since 1985, although some forests are now protected in
1950 national parks, most notably in Costa Rica. Not only has the
total area of intact forest declined dramatically, but the
Honduras
remaining forest fragments are small, often isolated from
one another, and may experience large edge effects. Similar
Nicaragua patterns are repeated for forests in many other parts of the
world, and over a range of scales.
Guatemala
El Salvador Costa Rica Panama
lion km2 (1.4% of Earth’s dry land surface); originally,
these communities covered an area of 17.4 million km2.
Together they have lost 88% of their original vegetation
1970
cover. Twenty of these hotspots are in tropical forests
or mediterranean climate regions. Nine of them are on
oceanic islands or archipelagos, where long isolation has
led to the evolution of endemic species. Many of the oth-
ers are isolated in other ways.
Not only rare species and rare communities are
threatened, however. Less diverse but unique commu-
nities, sometimes with unusual organisms, including
serpentine grasslands (see Box 15A), parts of the North
American deserts, and some salt marshes, have also been
1985
subjected to various threats not only to their extent, but
to their continued existence. In some cases, even origi-
nally very common species and ecosystems have been
reduced to the point of becoming endangered, or have
disappeared altogether. Climate change and human land
uses have dramatically increased the extent of newly
created depauperate deserts (with little vegetation of any
kind, and with few uniquely adapted desert species) at
the expense of fragile grasslands and shrublands. In the
Sahel region of Africa, south of the Sahara, vast areas
of once productive grasslands supporting large popu-
lations of people and animals have become shifting
sands. In Lagan in central China (near the headwaters
of the Yellow River), extensive grasslands have long sup-
plant species may result from the increased incursion of ported humans and their grazing animals. They have
edge effects into forests and other habitats. now become barren wastelands supporting little vege-
Many habitats around the world are threatened by tation, with recurrent sandstorms and vanishing rivers,
human activities. When these are rare or unique local lakes, and pastures, apparently due to a combination
habitats, reducing or destroying them may harm the of climate change and intensive land use (particularly
only populations of a particular species in a region, or overgrazing as a result of human population pressure).
may harm the only remaining population of that species Tropical forests are among the best-known threat-
anywhere on Earth. Unique organisms and communi- ened biomes. About half of the species on Earth live in
ties are found in many locations, and human encroach- tropical rainforests, even though they now encompass
ment is, in one way or another, threatening many of only about 7% of terrestrial land cover. Many of these
these places. Twenty-five biodiversity hotspots—threat- species are found only in very limited areas. For exam-
ened areas with very high species diversity—have been ple, organisms in the seasonal tropical forests of the large
identified worldwide. These areas together contain island of Madagascar, off the east coast of southern
133,000 plant species, or about half of the world’s plant Africa, have evolved in isolation for 121 million years,
species (Figure 22.14, Table 22.1). The five most species- leading to some of the world’s highest levels of plant
rich of these hotspots support 20% of all plant species. diversity and endemism. Madagascar has an estimated
Taken together, the 25 hotspots currently occupy 2.1 mil- 10,000 plant species, of which about 80% are endemic.
Table 22.1 “Hotspots” of biodiversity
Original extent Remaining primary Area protected Number Number of

of primary vegetation (km2) (km2) of plant endemic species
Location vegetation (km2) (% of original extent) (% of hotspot) species (% of global plantsb )
Tropical Andes 1,258,000 314,500 (25.0) 79,687 (25.3) 45,000 20,000 (6.7%)
Mesoamerica 1,155,000 231,000 (20.0) 138,437 (59.9) 24,000 5,000 (1.7%)
Caribbean 263,500 29,840 (11.3) 29,840 (100.0) 12,000 5,000 (2.3%)
Brazil’s Atlantic forest 1,227,600 91,930 (7.5) 33,084 (35.9) 20,000 7,000 (2.7%)
Choco/Darien/western 260,600 63,000 (24.2) 16,471 (26.1) 9,000 2,250 (0.8%)
Ecuador
Brazil’s Cerrado 1,783,200 356,630 (20.0) 22,000 (6.2) 10,000 4,400 (1.5%)
Central Chile 300,000 90,000 (30.0) 9,167 (10.2) 3,429 1,605 (0.5%)
California Floristic Province 324,000 80,000 (24.7) 31,443 (39.3) 4,426 2,125 (0.7%)
Madagascara 594,150 59,038 (9.9) 11,548 (19.6) 12,000 9,704 (3.2%)
Eastern Arc and costal forests 30,000 2,000 (6.7) 2,000 (100.0) 4,000 1,500 (0.5%)
of Tanzania/Kenya
Western African forests 1,265,000 126,500 (10.0) 20,324 (16.1) 9,000 2,250 (0.8%)
Cape Floristic Province 74,000 18,000 (24.3) 14,060 (78.1) 8,200 5,682 (1.9%)
Succulent Karoo 112,000 30,000 (26.8) 2,352 (7.8) 4,849 1,940 (0.6%)
Mediterranean basin 2,362,000 110,000 (4.7) 42,123 (38.3) 25,000 13,000 (4.3%)
Caucasus 500,000 50,000 (10.0) 14,050 (28.1) 6,300 1,600 (0.5%)
Sundaland 1,600,000 125,000 (7.8) 90,000 (72.0) 25,000 15,000 (5.0%)
Wallacea 347,000 52,020 (15.0) 20,415 (39.2) 10,000 1,500 (0.5%)
Philippines 300,800 9,023 (3.0) 3,910 (43.3) 7,620 5,832 (1.9%)
Indo-Burma 2,060,000 100,000 (4.9) 100,000 (100.0) 13,500 7,000 (2.3%)
South-central China 800,000 64,000 (8.0) 16,562 (25.9) 12,000 3,500 (1.2%)
Western Ghats/Sri Lanka 182,500 12,450 (6.8) 12,450 (100.0) 4,780 2,180 (0.7%)
Southwest Australia 309,850 33,336 (10.8) 33,336 (100.0) 5,469 4,331 (1.4%)
New Caledonia 18,600 5,200 (28.0) 526.7 (10.1) 3,332 2,551 (0.9%)
New Zealand 270,500 59,400 (22.0) 52,068 (87.7) 2,300 1,865 (0.6%)
Polynesia/Micronesia 46,000 10,024 (21.8) 4,913 (49.0) 6,557 3,334 (1.1%)
Totals 17,444,300 2,122,891 (12.2) 800,767 (37.7) * 133,140 (44%)
a
Madagascar includes the nearby islands of Mauritius, Reunion, Seychelles, and Comoros.
b
The total number of plant species globally is here estimated to be ~300,000.
*These totals cannot be summed due to overlapping between hotspots.
Madagascar also has some of the world’s most highly ing. Some scientists believe that global warming may
threatened habitats. Although forest destruction began have intensified these climatic effects.
many years ago, it accelerated greatly in the twentieth It is difficult to get precise measurements of how
century. Between 1960 and 1990, slightly less than half much deforestation has already occurred, or of current
the forested areas of Madagascar were destroyed. Today, regional rates of deforestation. Satellite imaging has
less than 10% of the original (pre-eighteenth-century) greatly aided our ability to track deforestation, particu-
forests remain. larly in combination with new methods for image analy-
Deforestation is occurring in tropical rainforest habi- sis. Estimated rates of deforestation differ among
tats worldwide, and is threatening or leading to the regions. Deforestation and logging have been acceler-
extinction of many species. Other forests, including high- ating in the Amazon in the 1990s (Figure 22.15). Similar
ly diverse dry tropical forests, are also severely threat- rates of deforestation are occurring in tropical Africa
ened. Climatic effects sometimes interact with other fac- (1.3–3.7 million ha/year: Boahene 1998). While the rain-
tors to intensify damage to tropical forests and other forests of Southeast Asia (Cambodia, Indonesia, Laos,
biomes. For instance, the 1997–1998 El Niño event Malaysia, Myanmar, Thailand, and Vietnam) cover a
caused uniquely ferocious and damaging fires in the much smaller area in the decade from the mid-1970s to
rainforests of the island of Borneo, exacerbating the dam- the mid-1980s, almost as much rainforest there was lost
age already done by extensive logging and land clear- each year as in the Brazilian Amazon. Vast areas of
Philippines
Caucasus South-Central
Mediterranean
California China
Basin
Floristic
Province Caribbean Indo-Burma
Brazil’s Polynesia/
Mesoamerica Micronesia
Cerrado
Tropical Western Western
Polynesia/ Choco/ Andes African Ghats and
Micronesia Darien/ Forests Sri Lanka Wallacea
Western Sundaland New
Ecuador Central Brazil’s
Succulent Caledonia
Atlantic
Chile
Forest Karoo Eastern Arc Madagascar Southwest
and Coastal Australia
Cape Floristic Forests of New Zealand
Province Tanzania/Kenya
Figure 22.14
Twenty-five regions of the world identified as biodiversity
hotspots have exceptional densities of species diversity and Invasive Species and Other Threats to Biodiversity
endemism. The hotspots are located within the colored
regions, but may make up only parts of those regions. (After Other major global threats to biodiversity include over-
Myers et al. 2000.) exploitation of species by hunting, fishing, or collect-
ing and the spread of invasive exotic species. Overex-
ploitation more obviously threatens animals than plants,
ancient forest were converted to agriculture or cut for but plants are not immune to this threat. First, logging
timber. Similar rates of forest loss continue there to the and removal for other reasons have greatly reduced the
present day. If current rates of deforestation continue, populations of some plant species. Several plant taxa,
the world’s tropical forests will disappear within 100 particularly orchids and cacti, are threatened by ama-
years, existing only in tiny, isolated fragments (see Fig- teur collectors (Orlean 2000). While collecting and sell-
ure 22.13). ing plants taken from natural populations is illegal in
many areas, steep prices and minimal fines conspire to
maximize these depredations. Second, many plant
2.0
species are threatened indirectly by overexploitation of
other species. Killing major predators, for example, can
Deforestation rate (106 ha/yr)
lead to large rises in herbivore populations, resulting

1.5 in overgrazing or overbrowsing. The extensive loss of
understory vegetation in forests all over the northeast-
1.0 ern United States in recent years is due in large part to a
tremendous rise in populations of white-tailed deer. The
rise in deer populations is in turn the result of a number
0.5
of factors, from suburbanization patterns that fragment
forests and increase edge habitats, which favor the deer,
0.0 to declines in predators and hunting (hunters are unwel-
1990–91 1992–94 1995–98
come in suburban backyards). The rise in deer numbers
Time period
threatens forest regeneration because deer browsing kills
Figure 22.15 tree saplings.
Deforestation rates in the Brazilian Amazon have accelerated The spread of exotic and invasive species is increas-
during the 1990s. Forest losses are expressed in millions of
hectares a year. Deforestation rates in other parts of the ingly threatening ecosystems all over the world (see
Amazon basin have also been increasing. (Data from Lau- Chapter 14). This phenomenon has serious conse-
rance 2000b.) quences both for humans and for the preservation of
native species and natural communities (Lodge 1993). 6.0
Global human population (billions)

Human activities such as deliberate and accidental
species introductions, forest fragmentation, logging, and 5.0
other disturbances have all been factors in the invasion
and spread of non-native species. Dukes and Mooney 4.0
(1999) raise the provocative suggestion that global cli-
mate change, including nitrogen deposition and rising 3.0
atmospheric CO2, may preferentially enhance the suc- Industrial
cess of invaders. 2.0
Revolution
Exotic plants have invaded forests in many parts c. 1750
1.0
of the world; these species may have substantial effects
on the retention and restoration of native species. New
0
Zealand, for example, now has at least as many natu- 1 500 1000 1500 2000
ralized exotic plants as native species (Allan 1961; Webb Year
et al. 1988). These introduced species have modified the Figure 22.16
vegetation of large areas in New Zealand. While native Worldwide human population growth, estimated over the
species have been not been driven to extinction there by past 2000 years.
the exotic species, the native plants appear to have been
locally displaced and diminished in abundance by the
invaders in many cases. Recent growth in human populations has been
Grasslands, deserts, and other plant communities much greater in developing than in industrialized
have also been greatly altered by invasive species. Inva- regions, and tends to be highest in the Tropics, where a
sive plant species can decrease the diversity of native large proportion of the people are living at or below sub-
plants and disrupt ecosystem processes, altering nitro- sistence levels. Population growth has slowed in many
gen cycling, fire regimes, and other fundamental char- of the developed, industrialized countries. As a result of
acteristics of ecosystems (Higgins et al. 1999; Vitousek these population growth patterns, increasingly greater
et al. 1996; Mack and D’Antonio 2001). In addition, bio- proportions (and absolute numbers) of the world’s peo-
logical invasions by exotic plants can exact enormous ple live in developing countries, and an increasingly
economic costs, including decreases in timber growth large proportion of the world’s people are children who
rates and forest value. These ecological and economic live in poverty in those countries (Figure 22.17).
impacts can be so great that some ecologists have urged This rapid human population growth could have
recognition of biological invasions as one of the most large environmental consequences (such as increased
important forms of global change affecting ecosystems resource use, habitat destruction, and species extinc-
today. tions) as well as economic and political effects and impli-
cations for global stability. But how much of the ongo-
Human Populations and Land Use Patterns ing deforestation and other damage to natural
The number of people in the world has grown dramat- communities can we actually ascribe to human popu-
ically since the Industrial Revolution, and is continu- lation densities and population growth? How accurate
ing to grow rapidly (Figure 22.16). The consequences is it to assume that increasing human populations direct-
of global increases in the human population have been ly affect forests or any other plant communities? Human
a cause for alarm, debate, and controversy for many societies are complex systems, and it is not surprising
years—as exemplified by the wide range of reactions that human populations affect the environment in indi-
to the much-disputed book by Paul Ehrlich, The Popu- rect and multifactorial ways. It is an oversimplification
lation Bomb (1968). Some of the predictions made regard- to assume that human population size is directly corre-
ing human population growth have been seriously lated with damage to the environment. Instead, both
flawed and alarmist, but attempts at outright dismissal local and international economic and political forces
of the concerns about its consequences have also been interact with population size to influence changes in
highly flawed. Much of the disagreement over and inter- land use (Lambin et al. 2001).
est in the implications of human population growth has Social scientists who seek to characterize and quan-
concerned the effects it might have on food supplies, and tify the interactions between humans and their environ-
more generally on human societies. Less popular atten- ments disagree about the extent to which different fac-
tion has focused on the possible environmental effects tors, including population size, are responsible for
of growing human populations, although both envi- environmental problems such as deforestation. On the
ronmentalists and social scientists have expressed con- one hand, there is solid evidence that across a wide num-
cern about these effects. ber of developing countries, population growth results
110 Industrialized Developing

nations nations
100
90
80
70
Age (years)
60
50
40
30
20
10
0
200 100 0 100 200 300 400 500 600
Millions of people Millions of people
Figure 22.17
Global age pyramid. The total number of people, in millions, in each age category,
from birth to 110 years, in industrialized and in developing nations. The total number
of people in developing nations is far greater than in industrialized nations. The dis-
tribution of ages is also different, with relatively more children and more people at
younger ages in the developing countries. The “baby boom” of people now in their
thirties to early fifties is clearly evident as a bulge at those ages in the industrialized
nations, but not in the developing nations. The number of very old people (over 90) is
very small for both groups of nations; the maximum human age (not clearly visible
here) is not different for the two groups. (Data from United Nations records.)
in greater deforestation (Ehrhardt-Martinez 1998). This ally by large multinational corporations). The third
has also been demonstrated at a global scale, although major agent of forest destruction was the very small-
the negative relationship between human populations scale clearing of forests for farming (slash-and-burn agri-
and declining forest area may have lessened in recent culture) by long-term residents as well as by impover-
years, at least in some countries (Mather and Needle ished immigrants to the region from other parts of the
2000). country. Thus, Parayil and Tong found that factors other
In studies focused at the scale of particular countries than population growth were the major direct contribu-
or regions, there have been contradictory results. In two tors to plant community destruction at a regional scale.
studies of the effects of human population density on Patterns of forest loss and fragmentation are also
deforestation in the Amazon basin of Brazil, its effects changing (Laurance 2000b). In past decades, many of the
were at best indirect. Pfaff (1999) found that in the Brazil- largest-scale forest losses in the Amazon basin occurred
ian Amazon, population density did not directly deter- at its southern and eastern edges. The development of
mine tropical forest clearing. Rather, many factors were major new highways, however, is now bringing settlers
implicated, from land characteristics and distance to and loggers directly into the heart of the central Ama-
markets to government development projects. Howev- zon forest, leading to great increases in fragmentation
er, Pfaff also found that while a static measure of popu- and forest loss (Figure 22.18; Laurance et al. 2001). This
lation size was not a good predictor of deforestation, the observation reinforces Pfaff’s argument that the distri-
first settlers in a region had a far greater effect on forest bution of human populations is at least as important as
destruction than later immigrants. He concluded that the absolute number of people in a country. The open-
the timing and distribution of population growth were ing of new roads in forest areas also was found to be the
more important than its absolute magnitude in deter- most important factor in deforestation in the rainforest
mining its impact. of Cameroon, in central Africa (Brown and Ekoko 2001).
In the second study, Parayil and Tong (1998) found In that study, multiple agents (from the government to
that population size and growth were not directly corporate loggers to local village interests and needs)
responsible for deforestation in the Amazon basin. and causes of deforestation acted synergistically to the
Instead, multiple agents were involved, operating at dif- detriment of the natural systems.
ferent scales. Major agents responsible for deforestation Patterns that are broadly similar in their complexi-
included large-scale clearing of forests to create cattle ty, but different in important details, have been found in
ranches and logging of forests for lumber (both gener- human-environment interactions in the United States.
30 nities and conversion to human-dominated landscapes.

They lead directly to far greater per capita fuel con-
sumption, because suburban residents have no choice
25 but to commute long distances to work, to shop, to
attend school, and for social and recreational opportu-
nities. The most fuel-efficient vehicle driven by a com-
20
Percent deforested
muter with a 100-mile round trip to work, or by a sub-

urban parent who racks up 25,000 miles per year in
15
everyday activities, will have far more impact than even
Paved highways a very large sports utility vehicle driven very few miles
by a person living in a compact neighborhood where it
10 is possible to walk or drive short distances for most
All roads activities. Recent efforts to build traditional towns and
and highways
neighborhoods once again offer the opportunity to
5 reverse some of these effects (Langdon 1997).
0 A Ray of Hope?
0 –10 11–25 26 – 50 51– 75 76 –100
Distance from road (km) Is all gloom and doom? Is the world going to hell in a
handbasket? There are clearly reasons to think so. How-
Figure 22.18 ever, there are also people working mightily to change
Effect of proximity to roads on deforestation in the Brazilian
Amazon. The percentage of the original closed-canopy forest
the statement so often heard that “all this will happen
that was destroyed by 1992 is shown as related to distance if the current rates of [fill in your favorite bad thing]
from paved highways and from all roads and highways. remain unchanged.” Perhaps some of these things can
Forests close to any roads, particularly paved highways, are be changed. Human behavior is not immutable, and if
far more likely to be lost than are those farther away. (After current practices are altered, some of the consequences
Laurance et al. 2001.)
of our actions can also be changed.
Although there are many factors acting to continue
our current momentum, it is certainly not impossible,
In the past half-century, two distinct patterns of chang- for instance, to begin to bring carbon emissions in the
ing residential land use have had large effects on the United States into line with the kinds of reductions that
reduction and fragmentation of wildlands. Both are, on have been accomplished in Western Europe. If we are
the surface, seemingly innocuous, and stem in part from concerned about the environmental damage being
the desire of people to get out of the cities and live in caused by unrestrained suburban sprawl in North
closer contact with nature. The first of these is the rise of America, there are workable, popular, and economical-
suburbanization—the establishment of vast low-densi- ly feasible alternatives to current environmentally cost-
ty areas surrounding metropolitan centers, particularly ly land development patterns (for example, see Duany
in North America. The consequence has been a shift in and Platt-Ziederbeck 2001). People are working in many
population from cities to “metropolitan regions” countries to reduce carbon emissions, stop deforestation
sprawled over thousands of square kilometers. This shift and habitat fragmentation, and restore damaged or
has changed land that was once agricultural or occupied destroyed ecosystems. We have the option of support-
by natural communities into a mix of homes on large ing efforts to protect species and habitats locally and
plots of land, with commercial development at some dis- around the world, and of changing our own habits and
tance. The second, more recent change has been the practices to reduce our personal impact on the environ-
incursion of vacation and retirement homes into wilder- ment. We also have the option of acting politically to
ness areas, particularly in coastal, montane, and desert alter current patterns of environmental destruction. The
areas (Bartlett et al. 2000). While this land use does not consequences of doing nothing should be obvious.
consume as much land as suburban development, it can
cause fragmentation of natural areas, and results in Summary
changes in fire regimes, erosion, and incursions of inva-
sive plant and pest species. Earth is experiencing one of the most rapid periods of
Local planning and policy decisions that create sub- change in its history. The global carbon cycle is central
urban development patterns, in contrast to traditional to many of these changes. Anthropogenic increases in
towns or cities, have other unintended consequences atmospheric CO2 and other greenhouse gases are expect-
in addition to the direct destruction of natural commu- ed to cause global climate changes that include warmer
temperatures, changes in precipitation patterns, and pos- areas becoming wetter and others drier, and both flood-
sibly changes in other climatic factors such as storm ing and drought will increase.
intensity and frequency. The increase in atmospheric Other anthropogenic global changes include habi-
greenhouse gasses is caused by a number of factors, chief tat destruction and fragmentation, nitrogen deposition,
among them fossil fuel combustion and deforestation. and the spread of exotic invasive species. Nitrogen dep-
There is now strong evidence that many of the predict- osition and habitat decline and fragmentation will result
ed climate changes are already occurring. Predictions for in the decline or extinction of sensitive species. Togeth-
the future, taking current trends into account, are for er, these changes will result in major shifts in species’
warmer climates everywhere, with the greatest changes distributions, changes in plant community composition
in polar regions. Rainfall patterns will shift, with some and ecosystem productivity, and declining biodiversity.
Additional Readings
Classic References Jefferies, R. L. and J. L. Maron. 1997. The embarrassment of riches: Atmos-
pheric deposition of nitrogen and community and ecosystem processes.
Arrhenius, S. 1918. The Destinies of the Stars. G. P. Putnam’s Sons, New
Trends Ecol. Evol. 12: 74–78.
York.
Hardin, G. 1968. The tragedy of the commons. Science 162:1243–1248.
Woodwell, G. M. 1970. Effects of pollution on the structure and physi-
ology of ecosystems. Science 168: 429–433. Langdon, P. 1997. A Better Place to Live: Reshaping the American Suburb.
University of Massachusetts Press, Amherst.
Naeem, S., F. S. Chapin III, R. Costanza, P. R. Ehrlich, F. B. Golley, D. U.
Contemporary Research Hooper, J. H. Lawton, R. V. O’Neill, H. A. Mooney, O. E. Sala, A. J. Sym-
Aber, J. D. 1992. Nitrogen cycling and nitrogen saturation in temperate stad and D. Tilman. 1999. Biodiversity and ecosystem functioning: Main-
forest ecosystems. Trends Ecol. Evol. 7: 220–223. taining natural life support processes. Issues in Ecology No. 4. Ecological
Society of America, Washington DC.
Hughes, L. 2000. Biological consequences of global warming: is the sig-
nal already apparent? Trends Ecol. Evol. 15: 56–61. Vitousek, P. M, H. A. Mooney, J. Lubchenco and J. M. Melillo. 1997.
Human domination of Earth’s ecosystems. Science 277: 494–499.
Mack, R. N., D. Simberloff, W. Mark Lonsdale, H. Evans, M. Clout and F.
A. Bazzaz. 2000. Biotic invasions: Causes, epidemiology, global conse-
quences and control. Ecol. Appl. 10: 680–710.

Ecology of Plants PDF

Uploaded by

Copyright:

Available Formats

Ecology of Plants PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ecology of Plants PDF

Uploaded by

Copyright:

Available Formats

What is the relationship between ecology and plants?

What is the relationship between ecology and plants?

What are the three components that ecology is built upon?

What are the three components that ecology is built upon?

PAGE PROOF: 2ND PASS

1 The Science of Plant Ecology

The Genesis of Scientific Knowledge

One reason the models often

(A) (B) (C)

Colder 1 Year 10 Years Figure 1.3

Duration of ice covert (days)

dence regarding the validity of a theory, dif-

106 Figure 1.4

slightly longer distances (100 m) and times (100 years). At yet

10–6 10–4 10–2 100 102 104 106

North-facing; Cold air

in that habitat. The amount of light available for photo-

Grana stacks (B)

Fructose Glucose To sucrose,

average than respiratory losses, however, resulting in a

loss of carbon and energy from the plant, as there is no +

(A) Figure 2.6

0.6 (After Mohr and Schopfer 1995.)

0.2 leaves to mature plants with many leaves, this shift

100 Chlorophyll b ( entially captured by the light reactions. Perhaps para-

80 100 green wavelengths are particularly ineffective for pho-

active radiation, or PAR. The amount of usable light

0.6 uptake is driven by a concentration gradient of CO2, set

Photo to come afternoon Figure 2.7

ing are continually in motion, widening and narrow-

Variation in Photosynthetic Rates within

(A) Mesophyll cells

other enzymes of the Calvin cycle) in the bundle sheath

sheath cells of C4 plants is an order of magnitude high-

70 with great efficiency, the ratio of CO2 fixed to water lost

it takes additional energy in the form of ATP to run the

Vacuole Starch Vacuole

(C) Figure 2.16

Relative abundance key

50 C4 grasses Other research has documented changes with alti-

their superior water use efficiencies, most closely asso-

(A) Epidermis Palisade cells (B) Epidermis

compensation point in Viola pubescens and Tiarella cordi- 2500

Incident photon flux (400–700 nm) (µmol m–2 s–1)

incoming light, regardless of the direction from which

Do Sun and Shade Adaptations Exist within Species?

In the widespread arctic/alpine species Oxyria digyna

Adapting to Life on Land

Simple perforation plate (E)

–150 –100 –50 –10 –4 –2 0

–2.3 –1.9 MPa –2.3

Roots: – 0.2 to – 0.4 MPa

Moist soil: – 0.1 MPa

Bundle sheath Mesophyll Xylem

(B) Figure 3.9

100 agricultural species, often close their stomata during the

Stomatal density Maximum

Water storage tissue

Cuticle Epidermis Stomatal crypt Open stomata