EVOLUTION

INTERNATIONAL JOURNAL OF ORGANIC EVOLUTION

PUBLISHED BY

THE SOCIETY FOR THE STUDY OF EVOLUTION

Vol. V DECEMBER, 1951 No.4

THE SPECIES CONCEPT

GEORGE GAYLORD SIMPSON

The Amcrican Museum of Natural History and Columbia University,

New York, New York

Received April 24, 1951

INTRODUCTION I agree with most of what all the au-

The species concept is focal in evolu- thors just cited have said. I believe,
tionary studies and, indeed, in all bio- however, that it is possible to say much
logical thought. Its endless discussion is of this in a somewhat different and there-
sometimes boring and seemingly fruitless, fore possibly clarifying way, to combine
but is not wholly futile. In the course of some of their apparently but not real1y
such discussion the concept has been conflicting views into one consistent state-
clarified, comprehension and a consensus ment, and to add a few significant consid-
have tended to develop, and the concept erations not explicit, at least, in any of
has changed in a significant way. There their papers.
have recently been two more flurries of Parts of the discussions cited and some
attention to this perennial topic. One, of the apparent conflicts are primarily
mostly in EVOLl:TION, by Burma (l949a, semantic. By a ponderous application of
1949b), Mayr (1949), Dunbar (1950), symbolic logic, Gregg sought to show that
Elias (1950), and Gregg (1950), was the issue raised by Burma and Mayr is
originally concerned with whether the not a genuine taxonomic problem or, at
species is a "fiction" or is "objective," least, that if it does relate to a taxonomic
but also treated such matters as the rela- .problem it does so in the wrong words.
tionships of neontological and paleonto- It is, of course, important that words be
logical species concepts. The other, in used as accurately as possible and that
the J ournal of Paleontology, hy Weller they do not obscure properly taxonomic
(1949), Jeletzky (1950), Bell (1950), questions. Nevertheless, Burma and
and Wright (1950), was concerned with Mayr (as well as subsequent discussants)
the bases and practices of paleontological were considering a genuine taxonomic
systematics and also with "morphologi- problem, in words perhaps not 10gical1y
cal" versus "phylogenetic" or "natural" impeccable but, taken in context, ade-
versus "unnatural" classification. These quately performing their main semantic
papers, among others, and an attempt to function, that of communicating under-
grapple with the whole problem for a standably among colIeagues.
class in systematics have inspired the The semantics of the systematists' vo-
following remarks. cabulary is a fascinating subject, which
EVOLUTION 5: 285-298. December, 1951. 285
286 GEORGE GAYLORD SIMPSON

surely does have its own importance but taxonomic procedure thus defined as ar-
which. has the danger of merely diverting bitrary. A completely non-arbitrary clas-
attention from the systematists' proper sification is impossible. It would be pos-
business, systematics. I believe that most sible to extend discussion to such points
of the purely semantic confusion on the as the precise definition of "essential con-
present subject can be avoided if such tinuity" or other parts of these definitions,
t~~~ as "real,': "natural," or "objec- but I think their meaning will be clear to
tive, and opposite or contrasting terms, all taxonomists, now or as discussion
are not applied to taxonomic categories proceeds, and that semantics may be
or methods of classification, and if the dropped at this point.
two terms "arbitrary" and "non-arbi-
trary" are used in specially defined senses. TYPOLOGY, MORPHOLOGY, AND GENETI-
Definitions of taxonomic categories, CAL GROUPS
such as a species, specify the sort of data
The typological concept of a taxonomic
or of inferences from data that are to be
group is that the group corresponds with
used in assignment of organisms to a
an abstract or ideal morphological pat-
group ranked in that category. For in-
tern. Variation may be dealt with by a
s~ance, the category definition of a spe-
fixed or intuitive standard as to allowable
cies as a group of "actually or potentially
deviation from the pattern, in which case
interbreeding natural populations which
~he grouping is arbitrary. (It may either
are reproductively isolated from other
Include discontinuities or draw a line
such groups" (Mayr) specifies that data
across continuity.) Or, somewhat less
and inferences as to interbreeding and its
naively, at a given level, usually that of
absence are to be used. In some cases
species, the criterion of continuity in vari-
the data or inferences used will indicate
ation around the pattern may be used, a
essential continuity among the organisms
non-arbitrary procedure for that category.
to be grouped, and in other cases they
The typological concept is pre-evolu-
will indicate essential discontinuity. Un-
tionary and non-evolutionary. It still
der the preceding genetic definition, actual
underlies a great deal of taxonomic prac-
or potential interbreeding is continuity
tice but is now seldom favored in theory.
and reproductive isolation is discontinu-
Arkell (1950), an experienced paleonto-
ity. With a morphological-associational
~ogical taxonomist, seems to be accepting
definition, continuity would be overlap in
It when he says that, "Theoretically, at
variation between compared populations
least, the number of species named reflects
and discontinuity the absence of overlap.
the number of forms, and so is more or
Essential continuity or discontinuity in
less an objective matter," but he was
geographic, ecological, or temporal dis-
mainly concerned with the highly laud-
tribution has obvious meaning.
able desire to keep super-specific cate-
I propose to call taxonomic procedure
gories conveniently manageable. The
arbitrary when organisms are placed in
only serious modern theoretical support
separate groups although the information
for frankly typological taxonomy comes
about them indicates essential continuity
from those few students who believe that
in respects pertinent to the definition be-
species arise by abrupt morphological
ing discussed, or when they are placed in
a single group although essential discon- change from one "morphotype" to an-
tinuity is indicated. Conversely, proce- other, notably Schindewolf (1950).
dure is non-arbitrary when organisms are Most of the data actually used in the
grouped together on the basis of pertinent, practice of taxonomy are morphologica1.
essential continuity and separated on the It is therefore not surprising that practi-
basis of pertinent, essential discontinuity. cal taxonomists suggest from time to time
Of course no stigma is meant to attach to that classification should be morphologi-
 THE SPECIES CONCEPT 287

cal, in principle, as Weller (1950) has statistical methods, as if statistical meth-

recently done. But a purely morphologi- ods were not a means of reaching conclu-
cal classification would be based strictly sions about populations and hence about
on degrees of morphological difference phylogeny. I have also insistently rec-
between organisms, and this is really so ommended population concepts in taxon-
impractical that no one, not even Weller, omy, and so have many others. A few
really tries to do it consistently. It is a paleontologists have been mentioned first
commonplace that the degree of morpho- because, on the whole, paleontologists
logical difference within what everyone, have been rather slower to grasp or ac-
morphologist, geneticist, or other, calls a cept the population concept of taxonomic
single species is frequently greater than groups. Despite some conservatives and
that between what all call separate, re- reactionaries, the concept is widely ac-
lated species. It is also quite impractical cepted among neontologists (in botany,
to obtain a valid, over-all measure of e.g., Camp, 1951; or in zoology, e.g.,
total morphological difference between Mayr, 1942).
two organisms. Characters are always If classification is to start with popula-
selected, weighted, and interpreted. As tions, category definitions at and below
Wright (1950) pointed out in criticism the species level should refer to popula-
of Weller, the usual and meaningful tions which, further, should be meaningful
basis for selection, weighting, and inter- biologically. It seems to me, and few
pretation is phylogenetic. Even typo- systematists are likely now to question
logical classification, more strictly mor- this, that such groups should likewise
phological than others, requires definition have evolutionary significance. Here is
of the morpho type from characters in a the most serious fault of typological or of
group already set up on grounds not, in purely morphological definitions. Unless
practice, purely morphological. Typolog- by chance or unless a hidden genetical
ical or not, practical morphological clas- criterion is actually used, they do not
sification starts with some sort of group- define biological populations or have clear
ing and in most practice this is usually evolutionary significance.
an attempt to recognize what is (whether Attention to biologically significant
so called or not) a genetically defined populations is the basis and justification
population. Thus Bell (1950) cogently for the now usual neontological defini-
argues the value of stratigraphic evidence tions of the species category in terms of
in paleontological taxonomy because it interbreeding and reproductive isolation,
bears on pertinent and useful biological i.e. of genetical factors, like the definition
taxonomic criteria that are not morpho- already quoted from Mayr. As Wright
logical. (1950) has mentioned, the fact that a
The fundamental point here for taxon- species, as a group, is actually diagnosed
omy is the modern idea that it is popula- in morphological terms, does not conflict
tions, not specimens, that are being classi- with definition of the species, as a cate-
fied. Newell (1948) has stressed this gory, in genetical terms. The basis for
point for invertebrate paleontology in definition of a category is quite different
criticism of the practice of naming vari- from the evidence available for decision
ants which are not populations. Jeletzky as to whether a particular group of or-
(1950) also emphasizes this point of view ganisms meets that definition. And al-
and its usefulness (one might say, neces- though the evidence used is mainly mor-
sity) for phylogenetic classification, al- phological in practice, it also almost always
though his argument is greatly weakened includes other sorts of data as well: dis-
by his statement that variants are "nat- tribution or association, at least, and
ural groups" within the population and preferably also other information.
by his contrasting of phylogenetic with The genetical definition of a species as
288 GEORGE GAYLORD SIMPSON

a group of actually or potentially inter- it could theoretically apply, sometimes

breeding organisms reproductively iso- turns out to be vague or impractical.
latedfrom other such groups is non- 3. There are many groups of organ-
arbitrary both in its inclusion and its isms, or circumstances involved in their
exclusion. Its criteria are reproductive taxonomic grouping, to which the stated
continuity and discontinuity. The group genetical definition does not apply even
defined is co-extensive with the continuity in theory.
and bounded by the discontinuity. A spe- 4. The genetical definition implies but
cies under this definition is the largest does not adequately state or overtly take
group with non-arbitrary exclusion and into consideration more definitely evolu-
the smallest group with non-arbitrary in- tionary criteria on which it does or should
clusion. By the criteria of this definition depend, criteria as to the evolutionary
and in cases to which it applies, infra- role of a lineage, to be discussed below.
specific groups are non-arbitrary as to 5. Application of this or of related evo-
what they include (being reproductively lutionary concepts of the species does not
continuous, by definition), but more or correspond with past and current usage
less arbitrary as to what they exclude in certain groups and by certain taxono-
(having boundaries without full repro- mists.
ductive discontinuity, by definition). Un- It seems to me that all these objections
der the same criteria and circumstances, have considerable force, more force than
supra-specific groups are arbitrary as to is granted them by some students whose
inclusion, because by definition they do taxonomic work is in the circumscribed
or may include two or more groups be- fields where the genetical definition is in
tween which there is discontinuity, but fact most practical or those whose inter-
non-arbitrary as to exclusion, because ests are not primarily taxonomic. Yet
their boundaries coincide with the non- I do not think that the objections invali-
arbitrary boundaries of included species. date the genetical concept or remove it
Thus under this particular concept and from a central and basic position in taxo-
in the particular cases to which it applies, nomic theory. They merely require that
the species is defined as the one taxonomic it be modified in certain applications and
category that is non-arbitrary both in that it be supplemented by other concepts
exclusion and in inclusion. This is an- to meet situations to which it does not
other way of expressing what is clearly properly or practically apply. The rest
intended by statements that the species of this paper is devoted mainly to dis-
(so defined) is the "objective" or the cussion of some desirable or necessary
"real" taxonomic unit. If my usage of modifications and supplemental concepts.
"non-arbitrary" is accepted and discus- One pertinent subject not directly dis-
sion of the meaning of "objective" or cussed, because of limitations of space
"real" is avoided, it should not be seri- and ability, is the taxonomy of asexual
ously questioned that the statement of groups, clones, apomicts, agamic com-
the first sentence of this paragraph is plexes, and the like. Some of what is
valid. Objections, which may also be said below bears inferentially on this sub-
entirely valid in their own terms, are of ject, and it has been briefly but ably con-
five principal sorts: sidered by Stebbins (1950) with refer-
1. The genetical concept of species is ences to older literature.
not the only one possible, and for certain
GENETICAL AND EVOLUTIONARY SPECIES
groups and in particular circumstances it
may be less desirable than some other. As Mayr (1946, 1950) has emphasized,
2. Application of the genetical defini- the usefulness of the genetical concept in
tion to actual cases, even those to which taxonomy and the non-arbitrary defini-
 THE SPECIES CONCEPT 289

tion of the genetical species (its "objective Camp, 1951). A rigidly genetical zoolo-
reality") are most evident in what he gist might then insist that such botanical
calls "non-dimensional species," those es- genera equal zoological species, but evi-
tablished in biotas living in one place at dently most botanists feel that in some
one time. Under such conditions, dis- way their species are analogous with zo-
continuities in morphological and asso- ological species and they can make out
ciated physiological variation are usually a good, even though not an absolutely
evident. In sexually reproducing groups clear-cut, case. (See Stebbins, 1950.)
it is almost always easy under these cir- In practice, even by zoologists who
cumstances to establish by observation, adhere strictly to genetical concepts of
experimentation, and inference which taxonomic units and who work on groups
morphological discontinuities reflect re- to which the concepts are applicable, it
productive discontinuities and to desig- is often clear that the criterion of inter-
nate these as species boundaries. breeding or its absence is not taken as
But, as Mayr has also recognized, the wholly decisive. Species may be distin-
fact that genetical species are usually guished even though they interbreed (hy-
rather obvious under these special limi- bridize) to some extent, and populations
tations does not mean that they are may be referred to a single species even
equally clear and the genetical definition though there is evidence that they are
equally adequate under other and per- not in fact interbreeding. Other criteria
haps more important conditions. Popu- are given weight additional to that of their
lations do have extension in time and evidence on interbreeding, e.g., morpho-
space and a non-dimensional taxonomy logical divergence, partial or full inter-
cannot cope with many essentials of life sterility, and especially occurrence with
and of its evolution. With extension in discontinuity in the same area.
space, the criteria of genetical continuity The genetical definition is meaningful
and discontinuity, of actual or potential because it is related to the evolutionary
interbreeding or its absence, cease in many processes that give rise to the groups
cases to be absolute and clearly non- being classified. Yet the genetical cri-
arbitrary and become merely relative. teria are not related to evolutionary
The similar and related local populations change directly but only, as a rule, by
may not in fact interbreed over a period implication. The following seems to be
of years and yet may reasonably be con- the strictly evolutionary criterion implied:
sidered as still having that potentiality. a phyletic lineage (ancestral-descendent
On the other hand, quite extensive inter- sequence of interbreeding populations)
breeding may occur between adjacent evolving independently of others, with
populations which nevertheless retain its own separate and unitary evolutionary
their own individualities, morphologically role and tendencies, is a basic unit in
and genetically, so clearly that any con- evolution. The genetical definition tends
sensus of modern systematists would call to equate the species with such an evolu-
them different species. In some groups tionary unit. Most of the vagueness and
of plants, even though species are defined differences of opinion involved in use of
and considered as genetical entities, oc- the genetical definition are clarified, at
currence of some hybridization between least, if not wholly resolved by taking the
adjacent species may be the rule rather genetical criterion, or interbreeding, not
than the exception. In such cases the as definitive in itself but as evidence on
species are in part arbitrarily bounded whether the evolutionary definition is
even though the gene flow is less between fulfilled. Thus the species as actually
than within species, and the genus may used by many progressive systematists in
become the most fully non-arbitrary unit both animals and plants does tend to
(a thought expressed in other words by approximate a unitary phyletic lineage of
290 GEORGE GAYLORD SIMPSON

separate evolutionary role even though in like those (by any definition) of neontol-
both cases outbreeding, hybridization, ogy do occur in paleontology, but that
may occur and in some groups of plants actual practice regarding them may be
this is widespread and usual. Emphasis more difficult or, at least, necessarily
on unitary evolutionary role may even somewhat different in paleontology and
resolve the theoretical difficulty of de- that there also occur in paleontology taxo-
fining species in asexually reproducing nomic groups to which no strictly neon-
groups. tological species concept can properly be
This redefinition, or shift of emphasis, applied.
or revealing of the implicit basis of much That paleontological data and materials
modern evolutionary taxonomy, intro- are different from neontological is well
duces the element of time into the concept known and sometimes overstressed. Di-
of species, even in the so-called non- rect genetical methods are unavailable in
dimensional situation. It designates the paleontology, but they are very rarely
species, including the "non-dimensional" used in neontological taxonomy. The
species, as a unit which has been evolving paleontologist usually has parts, only, of
separately, or which will do so, or, as a the organisms concerned, but the neontol-
rule, both. Decision that populations ogist commonly uses parts, only, of recent
will evolve separately involves prediction. organisms. Different parts may be avail-
Such points as wide geographical discon- able or used in the two cases, but infer-
tinuity (especially with a strong inter- ences from them regarding populations
vening barrier), morphological diver- may nevertheless be closely analogous or
gence, sympatric occurrence without actually identical. Nearly or exactly the
interbreeding, and intersterility are clearly same general sorts of data, morphological,
items of evidence for this sort of pre- distributional, and associational, are fre-
diction. Their bearing seems to me more quently used in the practice of paleon-
meaningful in evolutionary terms than in tological and neontological taxonomy.
the definition of actually or potentially (Fuller discussion of these points was
interbreeding populations, although of given in Simpson, 1943.)
course the evolutionary species usually is The "non-dimensional" species is en-
also such a group. The special impor- countered more frequently in paleontology
tance of intersterility, even though no than in neontology, in spite of the fact
modern taxonomist makes it an absolute that paleontology is inherently more mul-
requirement for specific separation, is, tidimensional than neontology. The ne-
for instance, evident in this context: inter- ontologist is seldom forced to confine
sterility makes the prediction of separate himself to collections from one locality,
evolutionary roles certain. and is never justified in doing so unless
forced. Much paleontological taxonomy
THE SPECIES IN PALEONTOLOGY: DATA is necessarily and properly based on
AND DISCONTINUITIES quarry collections or mass collections from
one local stratum, associations without
Part of the endless discussion on species appreciable dispersion in space or time
concepts is concerned with the relation- and ideally non-dimensional. In such
ship between neontological and paleonto- cases neontological concepts and defini-
logical species. Opinions vary from the tions of genetical and evolutionary species
view that the two usually are quite dif- apply without modification. (Even Elias,
ferent (e.g. Elias, 1950) to the view that 1950, outspoken opponent of the current
they are usually essentially the same or rapprochement of neontological with pale-
that one is only an extension of the other ontological systematics and of both with
concept (e.g. Mayr, 1950). Both views genetics, admits that in such cases paleon-
are correct in the sense that species just tologists "may be obliged to resort to
 THE SPECIES CONCEPT 291

neontological concepts of taxonomic and peculiarly paleontological situation

terms.") discussed below.
Discontinuities are more frequent and In such a case, the preferred practical
of more varied sorts in paleontology than procedure is:
in neontology. This has certain disad-
1. To consider the two (or more) lots
vantages for paleontological theory and
of associated specimens as samples of
interpretation, but it also has some prac-
different local populations and to derive
tical advantages, Discontinuities of ob-
from them estimates of morphological
servation, only, due to inadequate sam-
variation in those populations.
pling of local populations or inadequate
2. If the population estimates indicate
distribution of sampling stations, occur in
no significant mean difference, to consider
hath fields but are generally harder to fill
the samples as representative of essen-
in when paleontological. Discontinuities
tially a single population and hence taxo-
of record, that is, in the organisms ac-
nomic group.
tually present and available for sampling
3. If the population estimates indicate
in the field, are a particular paleontologi-
significant mean difference but overlap in
cal problem and may concern both time
range of variation, to consider the sam-
and space. When samples have been
ples as drawn from different subspecies
obtained from different localities or hori-
of one species.
zons, rocks and fossils intermediate be-
4. If the population estimates indicate
tween them may not exist. Such discon-
no overlap in range of variation (for at
tinuities are, as of now, facts in nature.
least one well-defined character), to con-
Their use to delimit taxonomic groups is
sider the samples as drawn from different
non-arbitrarv, by definition. Yet they do
species.
not necessarily coincide with any particu-
Species recognized in this way are non-
lar sort of discontinuity that existed when arbitrary in exclusion and inclusion by
the organisms were alive. Hence their combined morphological and distribu-
relationship to the sorts of units defined tional criteria. They are morphologically
in neontology may be and remain am- similar to most genetical and evolutionary
biguous. species. In many cases they will in fact
The special questions involved in suc- be genetical or evolutionary species, but
cession or sequence will be discussed sep- under the stated conditions it is virtually
arately, but it should be noted here that impossible to <'Ietermine this equivalence
paleontological samples discontinuous in with any high degree of probability.
space are often also discontinuous in time
and that the possibility can seldom be THE SPECIES IN PALEONTOLOGY:
discarded. With such samples of similar SUCCESSION
organisms it is always difficult and it Succession on a small scale and in-
may be quite impossible to determine volving short periods of time occurs in
whether: some neontological data and there in-
volves some special taxonomic problems,
(a) They represent local populations
but on the whole succession is distinc-
that were genetically continuous, and tively paleontological.
hence infraspecific groups by genetical Discontinuities of observation and of
and evolutionary definition. record are frequent in paleontological
(b) They represent separate phyletic study of successive populations. They
lineages, and hence distinct genetical and frequently correspond with discontinuities
evolutionary species. of time, already mentioned. Diastems,
(c) They represent ancestral and de- geologically brief intervals of non-deposi-
scendent populations, and hence a special tion (with or without erosion), are abun-
292 GEORGE GAYLORD SIMPSON

dant in most stratigraphic sequences. delimitation of .morphological-distribu-

They represent local discontinuities in tional species which may approximate,
time, but may be considered taxonomi- but cannot usually be clearly equated
cally unimportant if there was no sig- with, genetical-evolutionary species. This
nificant change in the populations being greatly simplifies paleontological proce-
studied or if intervening fossils of the dures and in many particular cases it
same or closely similar populations are averts the special taxonomic' problems
available from other localities. Larger inherent in continuity of succession.
and regional stratigraphic unconformities Essential continuity in sequences long
are also, although less, common and they enough to involve significant progression
usually represent taxonomically signifi- or diversification of populations is far
cant discontinuities in time. from universal in paleontology. It is,
Discontinuities in succession may also however, frequent and becomes steadily
be caused by migration, by change of more so as collecting becomes more ex-
(biotic, and commonly of correlated strati- tensive. The special problems involved
graphic) facies, or, frequently, by a com- therefore do have great and increasing
bination of both (see especially Bell, 1950, practical importance. They are of su-
also Newell, 1948). Such discontinuities preme importance for paleontological
often coincide with discontinuities in time taxonomic theory. No one seriously
but, as Bell has stressed, they need not doubts that the whole of life has factually
do so. For instance figure 1 shows dia- been a continuum of populations when
grammatically a situation in which there the whole sequence is considered, in spite
is a discontinuity of facies, with fossil of the innumerable discontinuities in the
populations as limited by facies successive record.
wherever found, but really contempora- The genetical-evolutionary concept of
neous and without discontinuity (or, in- species is applicable as between different
deed, regional succession) in time. phyletic branches, evolving lineages, espe-
'When discontinuities in succession are cially if they are contemporary but also
present in the data, they may be dealt if they are not. Thus in figure 2A, a, b,
with in practice as outlined above for and c are three different species, by
paleontological discontinuities in general. genetical or, more clearly, by evolutionary
They may similarly permit non-arbitrary definitions, although a and bare contern-

" "
"
..
. . "...
"
"
" "
"
" " " " ..
" " .. . . ..
" . .. ".. ..... ... .. ..
" .. , .
. ..
""""
" "
"
."
"."
" "
" " " " " " " " " " " " "
",
" " "
" " "
. . "
" " ."
" "
" "
" ". "
" "" "
,. " " " " " ""
" " ". " " " " "
"" " " " " " " " " " " " " " "" " " " " " " "" "" " " " .. " " " " " " " " "
. :::: : :::: :::::::6: : ::: :: : :: :::s: :::::::::.: : :.&': ::: : :'. ::: ::: ::
"
""
" """
""""""" "
"" "
" " " " " " " " " " " " " " " "" "
" " "" " "" ""
" ,,"""""
" " .. "
"""" " " " " "
""""
""
"""
" " " "" "
"""
" "
" "" .. ..
" "" " " " " " "" " "" " " " " " "" " " " " " " " "" " "" " " " " " " " "" " " " " " " " " " " "
- 0- -
- - -_
- -......:,
--
- . ····b .' '" ..... ···:·b·············· ...
• .!-."""" " """ " """" "." " """" •• """"" " •• "" """"" """ "",,""
" ••• " •• '." ":." •• ,," ••••••••••••••••••••••• 'C••••
t. ••••
.........
........ ..
-=-----=--0 - - - - = - - - 0 - - · . . . . . . . . . :::: '.: :::::'0::::: ::::::
--0--

- - --0-- -- 0 -- -- --0-- - - - - 0-

FIG. L Diagram of change of facies and fossil succession. Broken lines and dots
represent two different rock facies each with a characteristic fossil, a and b. Although
b everywhere occurs above a in any given local section, the two are, in fact,
contemporaneous.
 THE SPECIES CONCEPT 293

porary with each other but not with c. tary species until it divides and then to
Serious problems in theory, and usually consider the descendent branches as spe-
also in practice, arise rather regarding the cies distinct from each other and from
parts of such a pattern that cannot be the single ancestral line, as diagrammed
distinguished as separate branches. in figure 2D. This grouping meets an
One possible solution, diagrammed in evolutionary definition of species, al-
figure lB, is to recognize central lines though delimitations between adjacent
as species and to distinguish branches as species are arbitrary by genetical criteria.
other species. This procedure is "cor- It is, however, both undesirable and im-
rect" from an evolutionary point of view, practical. It frequently happens that a
or, better, the species so designated do population undergoes no essential change
fulfill the proposed evolutionary defini- even though a branch, a separate species,
tion even though their delimitation is has arisen from a part of it. -E.g. in fig-
genetically arbitrary at the points of ure 2D, d and e may be genetically and
branching. For rather small groups un- morphologically identical in all essentials.
der exceptionally favorable circumstances It is then not meaningful taxonomy to
the procedure is also practicable and is designate them as separate species. An
actually used. Its practicability depends, even more serious objection is practical:
however, on recognition of an essentially the pattern of branching in a paleonto-
unchanging central line, G, and main logical sequence is gradually discovered,
branches, band c. It is, however, more perhaps never fully known, and generally
usual even within rather small groups depends as much on opinion as on un-
and universal within really large groups equivocal data. The taxonomy of long-
and long sequences for all lines to evolve known species would be changed every
materially. Then it is not practical tax- time a new branch was discovered or
onomy to designate the whole of anyone inferred and would be excessively and
line as a single species. and there is no unnecessarily subject to personal dis-
meaningful criterion for designating agreement. Moreover, a phyletic line
"main" or "central" lines and branches. may change radically between branches
Thus the four alternatives of figure 2C (say within e of figure 2D), and it is
are all equally acceptable interpretations then not useful taxonomy to classify it
of the same phyletic facts as in figure 2A, as the same thing throughout.
in terms of main lines and branches, if The difficulties involved here are merely
all lines are undergoing progressive obscured by the presence of phyletic
change. The only reasonable criterion of branching. They arise, regardless of
choice would be designation of certain whether or not branching occurs, from
terminal branches as more important, or the problem of classifying ancestral and
somehow definitive, than others. A logi- descendent stages in a continuously evolv-
cal extreme would be, for instance, to ing population. Such a population may
take H 01110 sapiens as the supreme species be diagrammatically represented, as in
and to consider its ancestry, from the figure 3A, by a curve of variation (both
beginning of life (or even before) as the genetical and morphological), moving
main line, not specifically separable from through time and also being displaced as
H. sapiens. This arrangement has in its genetical and morphological characters
fact been seriously proposed by a philoso- change. A cross-section represents the
pher (Miller, 1949). Taxonomists will population at a particular instant in time,
surely agree that this result and the whole as it would be represented by a fossil
procedure involved are impractical if not sample from a single horizon. Such a
absurd. cross-section is a genetical non-dimen-
Another possible approach is to recog- sional species, as seen both in neontology
nize each evolutionary lineage as a uni- and in paleontology.
294 GEORGE GAYLORD SIMPSON

• A

,, \
,- \

II ",
I
I "
I '

, ,
I
I
"
I
I
I ,-,
'0
I
\
\
\
I \
I \
I \
,
. c:
I
I •
\
\
\
. \
I I
"
I
,
I,
\
, ''
c
\ I
'-' \
,,'

,, ' \
: d ,~
\
, ,
o '_.'

FIG. 2. Classification of successive populations with phyletic sequence and branching.

In each diagram, the time sequence is from bottom to top and the solid lines represent
phyletic descent. The broken lines enclose phyletic segments classified by various
methods as distinct species. A, branching phyletic sequence with three clearly distinct
evolutionary species, a-c. B, similar but more branched sequence with main lines and
branches classified as species. C, four different possibilities of designating main lines
and branches in the same phyletic sequence as in A. D, sequence as in A. with species
boundaries set at points of branching. See discussion in text.
 THE SPECIES CONCEPT 295

I
Range of Range of I
I
a C I
I
---_L- ..
I
I
I
I
c
I

I
Time b
cross - section
: population at

I 3 given times

a
~~~~~~~~------------------
A
- Genetical and morphological variation ___

B
FIG. 3. Phyletic evolution and classification of successive populations. Phyletic
sequence in a varying population is represented by a normal curve moving through
time and changing in mean character, and by the solid generated thereby. A, sequence
without branching. B, branched sequence. See discussion in text.

The whole sequence of populations, a In practice, the paleontologist calls a

to c in figure 3A, is genetically continu- and c different species if, as in figure 3A,
ous and it fulfills the conditions of both the inferred ranges of variation do not
genetical and evolutionary definitions of overlap. They are not different species
a species, as previously discussed. By by the widely accepted genetical criteria
these concepts, it is a single taxonomic or by the proposed evolutionary criteria
group, defined as a species. Yet with the discussed above. The comparison is clari-
passage of time and continuation of pro- fied in figure 3B in which speciation (in
gressive evolution, c has become quite the neontological sense) is represented as
different from a. For purposes of evolu- also having occurred. In this diagram, c
tionary study and of practical application and d are different species by any current
to stratigraphy, it is essential that a dis- usage; explicitly they are different geneti-
tinction be made between these popula- cal-evolutionary species and also different
tions, which are different. species in current paleontological practice.
296 GEORGE GAYLORD SIMPSON

But a and c are parts of a single genetical- criterion. It is in such situations that
evolutionary species, although called dif- the frequent occurrence of discontinuities
ferent species in paleontological practice. of record, the absence of part of the se-
The paleontologist thus uses the desig- quence a-c, is practically useful in pro-
nation "species" for two sorts of entities viding a means of separating a and c.
which are radically and fundamentally in- This is still a separation in. what was
congruent. The only way in which the a continuum, but it is non-arbitrary (by
species category might be defined so as the special definition of that word in this
apparently to include both sorts of entities paper) as regards the actually available
would be to abandon any evolutionary materials being classified.
significance for taxonomy and to use Since paleontologists are applying the
purely morphological criteria. But this designation "species" to two fundamen-
is not a useful solution. The general tally dissimilar sorts of taxonomic cate-
undesirability and impracticality of purely gories, it would appear logical that they
morphological taxonomic concepts have confine that name to one of them and use
been sufficiently emphasized above. a different name for the other. This has
Moreover, the whole sequence of organ- also been suggested, but it runs up against
isms represented in figure 3B cannot be another serious practical difficulty: the
classified at all, in morphological or any paleontologist often does not know and
other terms, if the pattern in time, i.e. has no way to determine which of the
the evolutionary situation, is ignored. As two basically different sorts of groups
static, separate pictures, the morphologi- called "species" he has before him.
cal difference between a and c and that It is a common situation to have two
between c and d are of the same sort, discontinuous paleontological samples such
but within the pattern of the whole group as a and c of figure 4A. (It has been
in time, even the morphological relation- noted that if a and c are discontinuous in
ships are not the same in the two cases, space, the possibility that they are also
for a and c are morphologically (as well different in time can seldom be ruled
as genetically) continuous through inter- out.) By applying the practical methods
vening populations and c and d are not. previously summarized, the paleontolo-
(One might say here, in line with Dun- gist can readily draw population infer-
bar, 1950, that c and d are continuous ences from these samples, find that
through the sequence c-a-d, hut the fact variation probably did not overlap in the
that this involves a reversal in direction populations, and define them as different
of time still makes an essential difference "species." However, he does not know
from the sequence a-c or c-a, which is in what sense they are different species,
consistent in the direction of time.) because he does not know whether the
In the situation represented in figure relationship is as in figure 4B or as in
3A, the desirable and indeed necessary figure 4C, and unless other crucial popu-
taxonomic separation of a and c, whether
lations can be sampled he may have no
they are called species or by some other
conclusive way of finding out. In dealing
category term, is arbitrary, because
with different samples from closely simi-
through intervening stages they are con-
tinuous by all meaningful criteria. The lar populations, this is one of the com-
placing of an intermediate population, such monest situations in the practice of
as b, in one category or the other is, of paleontology.
course, also arbitrary. When the data In such cases, a distinction cannot be
really reflect the continuity of the se- made in practice between "species" in the
quence, intermediate populations must basic genetical or evolutionary sense and
often be placed by rule of thumb rather in the sense of subdivisions in a continu-
than by any more positive and meaningful ous ancestral-descendent line. I do not
 THE SPECIES CONO~PT 297

1
Time
A
IA +- space-and morpholoGY_
A B
I
I

A
"A
',"
\ I
, 1

a //1
\ 1
, 1
, 1

, 1
" \1
I
I
I
I
I
I
I
I

c
FIG. 4. Diagram illustrating problems of interpretation of two samples of related
fossil organisms separated in space and (or, and possibly) in time. The variable popu-
lations represented by the samples are represented by normal curves. A, the given
situation. B, interpretation as a single phyletic sequence, in which a and c represent
the same species by genetic definition. C, interpretation as a branching sequence, in
which a and c are different species by genetic or evolutionary definition. See discussion
in text.

here favor or propose a special term for method of approach to other points will
the latter sort of taxonomic group. I do probably be evident to most taxonomists.
maintain that it is desirable and useful to
realize that these are two quite different LITERATFRE CITED
things, and that the "species" of paleon- ARKELL, W. J. 1950. A classification of the
tological taxonomy may be of either sort. Jurassic ammonites. Jour. Paleont.. 24 :
354-364.
There are many other pertinent and BELL, W. C. 1950. Stratigraphy: a factor in
interesting points that might be consid- paleontologic taxonomy. Jour. Paleont., 24:
ered, such as the problem of dual, partly 492-496.
coincident evolutionary species a~c and BURMA, B. H. 1949a. The species concept:
a-d in figure 3B, the uses of evolutionary a semantic review. Evolution, 3: 369-370.
1949b. Postscriptum. Evolution, 3:
acceleration and appearance of key char- 372-373.
acters for separating ancestral and de- CAMP, W. H. 1951. Biosysternaty. Brittonia,
scendent species, or the relationships of 7: 113-127.
subdivisions of choroclines and chrono- DUNBAR, C. O. 1950. The species concept:
further discussion. Evolution, 4: 175-176.
clines. This paper is, however, suffi- ELIAS, M. K. 1950. Paleontologic versus ne-
ciently long already, and the possible ontologie species and genera. Evolution,
extension of its general point of view and 4: 176-177.
298 GEORGE GAYLORD SIMPSON

GREGG, J. R. 1950. Taxonomy, language and in invertebrate paleontology. Jour. Paleont.,

reality. Amer. Nat., 84: 419-435. 22: 225-232.
JELETZKY, J. A. 1950. Some nomenclatorial SCHINDEWOLF, O. H. 1950. Grundfragen der
and taxonomic problems in paleozoology. Palaontologie, Schweizerbart, Stuttgart.
Jour. Paleont., 24: 19-38. SIMPSON, G. G. 1943. Criteria for genera,
MAYR, E. 1942. Systematics and the origin of species, and subspecies in zoology and pale-
species. Columbia Univ. Press, New York. ontology. Ann. New York Acad. Sci., 44:
- - . 1946. The naturalist in Leidy's time and 145-178.
today. Proc. Acad. Nat. Sci. Philadelphia, STEBBINS, G. L., JR. 1950. Variation and evo-
98: 271-276. lution in plants. Columbia Univ. Press,
1949. The species concept: semantics New York.
versus semantics. Evolution, 3: 371-372. WELLER, J. M. 1949. Paleontologic classi-
MILLER, H. 1949. The community of man. fication. Jour. Paleont., 23: 680-690.
Macmillan, New York. WRIGHT, C. W. 1950. Paleontologic classifica-
NEWELL, N. D. 1948. Infraspecific categories tion. Jour. Paleont., 24: 746-748.