Pplications of The Compensating Pressure Theory of Water Transport
Pplications of The Compensating Pressure Theory of Water Transport
Pplications of The Compensating Pressure Theory of Water Transport
APPLICATIONS
MARTIN J. CANNY
Biology Department, Carleton University, 1125 Colonel By Drive, Ottawa, Canada K1S 5B6
Some predictions of the recently proposed theory of long-distance water transport in plants (the Compensating Pressure
Theory) have been verified experimentally in sunflower leaves. The xylem sap cavitates early in the day under quite small
water stress, and the compensating pressure P (applied as the tissue pressure of turgid cells) pushes water into embolized
vessels, refilling them during active transpiration. The water potential, as measured by the pressure chamber or psychrometer,
is not a measure of the pressure in the xylem, but (as predicted by the theory) a measure of the compensating pressure P.
As transpiration increases, P is increased to provide more rapid embolism repair. In many leaf petioles this increase in P is
achieved by the hydrolysis of starch in the starch sheath to soluble sugars. At night P falls as starch is reformed. A hypothesis
is proposed to explain these observations by pressure-driven reverse osmosis of water from the ground parenchyma of the
petiole. Similar processes occur in roots and are manifested as root pressure. The theory requires a pump to transfer water
from the soil into the root xylem. A mechanism is proposed by which this pump may function, in which the endodermis
acts as a one-way valve and a pressure-confining barrier. Rays and xylem parenchyma of wood act like the xylem parenchyma
of petioles and roots to repair embolisms in trees. The postulated root pump permits a re-appraisal of the work done by
evaporation during transpiration, leading to the proposal that in tall trees there is no hydrostatic gradient to be overcome in
lifting water. Some published observations are re-interpreted in terms of the theory: doubt is cast on the validity of measurements of hydraulic conductance of wood; vulnerability curves are found not to measure the cavitation threshold of water
in the xylem, but the osmotic pressure of the xylem parenchyma; if measures of xylem pressure and of hydraulic conductance
are both suspect, the accepted view of the hydraulic architecture of trees needs drastic revision; observations that xylem
feeding insects feed faster as the water potential becomes more negative are in accord with the theory; tyloses, which have
been shown to form in vessels especially vulnerable to cavitation, are seen as necessary for the maintenance of P, and to
conserve the supplementary refilling water. Far from being a metastable system on the edge of disaster, the water transport
system of the xylem is ultrastable: robust and self-sustaining in response to many kinds of stress.
Key words: embolism refilling; hydraulic architecture; hydrostatic gradient; pressure chamber; reverse osmosis; root
pressure; starch sheath; tissue pressure; transpiration; tyloses; ultrastability; water pump; xylem-feeding insects.
STATEMENT
A plant organ is like a mass of osmometers (cells)
confined in a box (epidermis, bark). When supplied with
1 Manuscript received 17 September 1997; revision accepted 5 March
1998.
Symbol convention: Italic symbols are used for primary measured
quantities, plain text symbols for derived quantities.
The author thanks Margaret McCully and Steve Vogel for helpful
discussions and criticism of the manuscript, Adam Baker for making
the plate, and the Natural Sciences and Engineering Research Council
of Canada for an operating grant.
897
898
AMERICAN JOURNAL
OF
BOTANY
[Vol. 85
Figs. 14. Figs. 12. Preparations of a strand of xylem in the petiole of a sunflower leaf, frozen intact during active transpiration, and viewed
(still frozen) in the cryo-scanning electron microscope. 1. Living cells are grey, containing white lines of solute crystals in a matrix of black ice.
Some of the vessels contain black ice and were filled with sap when frozen, others contain gas and were embolized when frozen. Bar 5 100 mm.
2. View inside an embolized vessel in a preparation like that in Fig. 1. The vessel was refilling with drops of water entering through pits from
adjacent parenchyma cells. Bar 5 10 mm. Figs. 34. Starch sheaths in the petiole of a hollyhock leaf. Transverse sections of petioles of adjacent
leaves harvested on a day of bright sunshine. Iodine stain. 3. At 0645. 4. At 1930. Bars 5 100 mm.
July 1998]
CANNYAPPLICATIONS
899
900
AMERICAN JOURNAL
Shift of emphasisThese observations and the induced hypothesis of refilling vessels change somewhat
my view of how the compensating pressure theory works.
In the original formulation (Canny, 1995) the main focus
was on raising the pressure in the vessels to stop the
water threads breaking. Now a major effect of the compensating pressure appears also to be the provision of the
steady influx of supplementary water to the vessels from
a water reservoir, refilling embolized vessels fairly quickly when the threads do break. This view of the xylem
parenchyma, as providing a constant low-level supplement to the transpiration stream, stimulates useful insights into the operation of a number of other regions of
the plant, some of which will be briefly discusssed.
ROOTS, ROOT EXUDATIONS, AND
ROOT PRESSURE
There is good evidence that a process similar to that
found in the leaves is operating also in roots. McCully,
Huang, and Ling (1998) followed changes of embolisms
in the vessels of transpiring corn roots, frozen intact in
the field. The technique used was the same cryo-scanning
microscopy of the fully hydrated tissues and assessment
of the contents of the individual vessels. Early in the day
no root vessels were embolized. Embolisms appeared at
sunrise, increased to a plateau (;75%) during the middle
of the day, and fell during the afternoon or evening during active transpiration, to reach zero again by dusk. In
embolized vessels at all times water could be seen entering through the pits in the vessel walls from the adjacent
parenchyma and from branch roots (Fig. 2). Again, a reverse osmosis of water squeezed from the xylem parenchyma by the tissue pressure confined within the mechanical barrier of the endodermis seems a likely expla-
OF
BOTANY
[Vol. 85
nation. This is a manifestation of the well-known but inadequately explained phenomenon of root pressure. A
possible source of the pressure, though not an explanation
of the pressure-driven xylem flux of water, was identified
in Canny (1995) as the tissue pressure generated by phloem and parenchyma cells confined within the stele. The
additional hypothesis of the reverse osmosis of cellular
water into the vessels, which refills embolisms, is a step
towards explaining the water flux observed in excised
roots. We shall return to this question presently.
Recent work has shown that there are at least two other
exudations produced by roots, which may be generated
by tissue-pressure-driven reverse osmosis, but operating
in the cortex, not the stele. The first of these to be published was the finding of liquid water (with varying concentrations of solutes) in some of the intercellular spaces
of the root cortex (Canny and Huang, 1993). The intercellular solutions were found at all times of the day, and
in all ages of root, and were present even at times of
considerable water stress. Such accumulations of solution
outside cells have been observed by us constantly since,
in many hundreds of root samples examined with the
CSEM (e.g., McCully, 1994) and come to be regarded as
a normal component of root tissues. More recently, much
larger accumulations of liquid water were recorded, filling some of the large aerenchyma spaces in corn roots
(Van der Weele, Canny, and McCully, 1996; Watt et al.,
1996).
The second kind of expressed water is that found exuding from field-grown corn roots at night (McCully,
1995). Convex drops of water were seen (again frozen
preparations in the CSEM) on the epidermis and root
hairs and filling much of the space between soil particles
in the early morning. By midday the liquid had gone, and
the rootsoil interface contained only air among the plant
tissues and the soil particles. Without attempting to formulate detailed mechanisms for these two kinds of exudation, it is tempting to couple them with the supplementary water as examples of pressure-driven reverse osmosis from root cells. But if cortical cells may be considered as confined and under pressure, the epidermal
cells are certainly not so, and the source of pressure for
them is less clear (but see below).
Hypothesis of the root water pumpThe logic deriving from the compensating pressure theory demands that
there must be a water pump in roots. The reasons for this
are several. First, with the pressure in the xylem conduits
not far from zero bar absolute, a pump is needed to take
water from the soil where the water potential may be
several bars negative, and transfer it to the conduits. With
the simpler cohesion theory no such pump was necessary,
and none was looked for. Pressures in the xylem were
believed to be far enough negative to extract water from
even quite dry soils. Without this comfortable simplicity,
some role for the living cells of the root in extracting
water from the soil is required.
Second, the compensating pressure theory requires a
supply of supplementary water to refill embolisms in the
roots. In the stem and leaves this is stored locally, used
during transpiration, and recharged at night from below.
But in roots, though the cortex may act as a temporary
store, the ultimate source of the supplementary water for
July 1998]
CANNYAPPLICATIONS
901
902
AMERICAN JOURNAL
OF
BOTANY
[Vol. 85
Fig. 5. Diagram of the features of a root required for the operation of the water pump. The thickened cell walls of the endodermis and the
central xylem vessel are shown black; cytoplasm, pink; vacuoles and the sap of the central xylem vessel, blue. The solute concentration in the
vacuoles increases progressively from the cortex (u, v), through the endodermis (w), the pericycle (x), and the xylem parenchyma (y). In other
words, the water concentration in these vacuoles decreases along this path. The plasmodesmata between the endodermis and the pericycle constitute
the one-way valve that permits the inward diffusion of water (red arrows) down a concentration gradient of water, but restrict the outward (pressuredriven) flow of water. Pressure builds up in the stele, and drives water from the stelar cells into the vessel (black arrows). For details see Appendix
3 and text.
July 1998]
CANNYAPPLICATIONS
903
The supplementary water is a more abundant and general manifestation of the water postulated (and rather deviously demonstrated) by Munch (1930) to arise from the
phloem. For Munch it was the superfluous water from
the translocated sugar solution, left over after the sugar
had been used in a sink. This water would be returned
to the plant body and the source by transfer to the xylem.
Milburn (1996) has revived the idea and proposed that
this released water helps to repair embolisms.
The phloem was recognized as a likely source of pressure (Canny, 1995). A fairly direct measure of the tissue
pressure that exists in the phloem, and which could drive
some of the reverse osmosis, was made by Buttery and
Boatman (1966) in the bark of rubber trees. Manometers
inserted into the bark were readily sealed into the pool
of exuding latex and gave reproducible measures of the
pressure driving the latex out of the laticifers. They record pressures in the range 19 to 113 bar (10.9 to 11.3
MPa) morning and evening. The high values at dawn fell
by ;4 bar (0.4 MPa) at times of high evaporative demand. This fall is a direct measure of the compensating
pressure P.
Experiments. Revive and exploit this system to investigate the behavior of P.
The hydrostatic gradientOnce the possibility is admitted of a water pump in roots, which might deliver
water into the xylem conduits against a pressure of several bars, it becomes possible to think in a new way about
the movement of water up trees. Since the first mention
of this question it has been a commonplace that the water
must be lifted up the tree against the force of gravity, that
1 bar (0.1 MPa) of tension must be applied to lift the
water through each 10 m, or 10 bar (1 MPa) for the tallest
(100 m) trees. I find no mention of the fact that the water
is already up the tree, that the tree is in fact a standing
tank of static water surrounding a few conduits in which
water moves. The static water, contained in living cells,
fibers, intercellular spaces, and the smallest conduits,
must be a continuum, even if it contains islands of dry
material and gas. Isolated regions of static water cannot
persist. So the continuum of static water has a weight,
and the force exerted by this weight is proportional to
the height. A standing tank of water with a pipe running
up it from bottom to top has a gradient of positive pressure increasing downwards, not a tension in the pipe increasing upwards. Everywhere the pressure in the pipe
equals the pressure in the tank. To suck water from the
top of the pipe requires no more force than to suck water
904
AMERICAN JOURNAL
OF
BOTANY
[Vol. 85
VULNERABILITY CURVES
As explained in Canny (1995) the results of almost all
experiments are interpretable equally by the cohesion theory and the compensating pressure theory. As an example
of alternative explanation I take the experiments of Pockman, Sperry, and OLeary (1995) who tried to show that
the pressure chamber did indeed measure tension in the
xylem by inducing known amounts of water stress in
branches. They compared the effects of the centrifugal
stress generated by spinning branches with two other
stress treatments, drying in air and external air pressure
applied around the branch. They measured the effects of
these stresses by estimating the hydraulic conductances
of the branches. Their argument is that the hydraulic conductance gives a measure of the cavitations produced in
the conduits by the stresses. The stresses of centrifugation
and drying act by moving or extracting water, and air
pressure acts by forcing air into the conduits through
pores in pit membranes. The results are displayed in the
form of vulnerability curves, plots of the percentage
loss of conductance against the stress (Fig. 6). The three
stresses are measured as pressures in MPa, centrifugal
stress calculated from the angular velocity, drying stress
measured by the pressure chamber on leaves taken from
the branches, and pressure stress by the air pressure applied in a jacket around the branch.
To understand the procedure it is necessary to consult
an earlier paper (Sperry and Saliendra, 1994) describing
the apparatus (Fig. 7). Air pressure was applied in a jacket around the body of the branch, while one end was
connected to a reservoir supplying solution for the conductance measurement under small pressure determined
by the elevation of the reservoir. Note the air vent, which
allowed escape of bubbles driven through the branch by
the air pressure, and the notches that admitted air to the
branch. An initial flow measurement was made with an
air pressure of 11 bar (1100 kPa) . The pressure was
July 1998]
CANNYAPPLICATIONS
905
HYDRAULIC ARCHITECTURE
Just as the pressure chamber does not measure tension
in the xylem, so the traditional way of measuring hydraulic conductance does not measure conductance because the activity of the living cells alters the conductance during the measurement. If all the measurements
we have of pressure gradients are wrong, and all the measurements of hydraulic conductance are wrong, what to
we truly know about the hydraulic architecture of trees?
Measurements of C throughout the canopies of trees
with the pressure chamber have produced the broad generalization that C becomes more negative along the outward path from the trunks through branches of smaller
diameters to twigs and leaves (e.g., Borchert, 1994). The
gradient of C is seen as a manifestation of the pressure
gradient needed to drive the water flow along the outward
path. It agrees well with Zimmermanns demonstration
that the leaf specific conductance (i.e., the flow through
a portion of the wood pathway divided by the area of the
leaves it supplies) decreases along the same path, and so
does the diameter of the largest vessels (Zimmermann,
1983). This concept of tree hydraulic architecture has
been elaborated to map the pressure gradients necessary
to account for the distribution of leaf specific conductances in a number of different tree types (e.g., Tyree and
Ewers, 1991). The orthodox explanation of the gradient
of C is that there are junction constrictions where each
branch or twig joins a larger one, which would generate
an increase in the pressure gradient across the junction,
and negative pressure would increase outwards in the
canopy (Tyree and Ewers, 1991).
On the compensating pressure theory the gradients of
906
AMERICAN JOURNAL
OF
BOTANY
[Vol. 85
XYLEM-FEEDING INSECTS
As emphasized previously (Canny, 1995; Crews et al.,
1998), one of the strongest arguments against large negative xylem pressures is the extraction of xylem sap by
many species of insect. Mittler (1967) calculated the pressure that a leafhopper generated to suck sap through its
stylets at the rate it was excreted as 23 bar (2300 kPa),
and Raven (1983) estimated from the dimensions and
properties of the muscles of the cibarial pumps, a maximum suction of 3 bar (300 kPa). A detailed study of the
feeding of the leafhopper Homalodiscus on four plant
species (Andersen, Brodbeck, and Mizell, 1992) produced results that the authors have great difficulty in reconciling with the cohesion theory. These leafhoppers
sucked and exuded sap at the rate of up to 0.7 cm3/h
through stylet canals 12 mm wide, implying a velocity of
43 cm/sec and a pressure gradient along the 2 mm of
stylet of 0.48 bar (48 kPa). Feeding rates were strongly
correlated with the organic nitrogen in the sap, which
varied on a diurnal cycle, and was highest in the afternoon. The authors were puzzled to find that the feeding
rate increased at this time, just when the pressure chamber required the strongest balancing pressure. Feeding
rate increased as C became more negative down to
218 bar (21.8 MPa). In plants stressed beyond 218 bar
(21.8 MPa), when C reached values down to 232 bar
(23.2 MPa), the feeding rate was reduced.
By now, the reader will easily make the translation to
the alternative interpretation, that down to C values
equivalent to the threshold of xylem parenchyma osmotic
pressure (.18 bar[1.8 MPa]) the pressure in the conduits
is kept close to zero, any embolisms are quickly refilled,
and the leafhoppers are easily able to pump the sap with
its enhanced organic nitrogen at maximum rates. In the
stressed plants, as the compensating pressure runs out and
sap pressures fall to 23 bar (2300 kPa) and beyond, the
rate of extraction slows, embolisms cannot be refilled,
and excretion stops.
TYLOSES, TERMINATING VESSEL FUNCTION
In the frozen preparations of sunflower petioles, which
provided the data for the embolism studies (Canny,
1997a, b), tyloses were frequently observed. A full account of them is given in Canny (1997c). They were
forming in, and filling, the oldest, thick-walled vessels at
the inner ends of the xylem arcs, farthest from the phloem. These were just those vessels that were shown to be
most vulnerable to embolism during the first water stress
imposed by transpiration early in the day. In terms of the
compensating pressure theory these tyloses are readily
interpretable as a device to preserve tissue pressure, prevent its dissipation in compressible gas spaces, and conserve supplementary water by eliminating superfluous
sinks for it. Those vessels farthest from the pressurizing
phloem would be the most vulnerable to embolism. The
strategy to put them out of action, replace them with incompressible parenchyma, and form new ones near to the
phloem from the cambium, is entirely consistent with the
requirements of the theory and with the tactics of meeting
these requirements by pressurizing and supplying supplementary water with the available tissues.
July 1998]
CANNYAPPLICATIONS
907
908
AMERICAN JOURNAL
of stored water to transpiration in Scots pine. Plant, Cell and Environment 2: 309317.
WATT, M., C. M. VAN DER WEELE, M. E. MCCULLY, AND M. J. CANNY.
1996. Effects of local variations in soil moisture on hydrophobic
deposits and dye diffusion in maize roots. Botanica Acta 109: 492
501.
WESTGATE, M. E., AND E. STEUDLE. 1985. Water transport in the midrib
tissue of maize leaves. Plant Physiology 78: 183191.
WHITE, P. R. 1938. Root pressurean unappreciated force in sap
movement. American Journal of Botany 25: 223227.
ZIMMERMANN, M. H. 1983. Xylem structure and the ascent of sap.
Springer, Berlin.
APPENDIX 1
THE PETIOLE AS A SOURCE OF SUPPLEMENTARY WATER
Consider a petiole of one of the sunflower leaves used in Canny
(1997a), with radius 4 mm and ;150 vessels whose median diameter
is 40 mm. The question is: what volume of water must be provided to
maintain a supplementary water supply for the day, and is it reasonable
to suppose that the petiole parenchyma can supply it by a modest volume shrinkage? If DP is the pressure difference between the parenchyma and the vessel lumen, and Lp is the hydraulic conductivity of the
plasma membranes of the cells surrounding a vessel, the volume needed
for unit length is
Lp 3 total vessel surface 3 DP 3 time.
We must select likely values for Lp and DP. In Canny (1997a) the value
chosen for the former was from the midrange of those measured for
midrib tissue of corn leaves by Westgate and Steudle (1985), Lp 5 1027
m3m22s21bar21 [1026 m3m22s21MPa21]. The value of DP must be
greater than the vapor pressure of water but need not be very large. In
the middle of the day the difference is reversed, because water flowed
out of the vessels. Suppose an average of 1/10 bar (10 kPa). Then for
1 m of petiole with 150 vessels of diameter of 40 mm and an 8-h day
the volume becomes,
1027 3 150p 3 40 3 1026 3 28 800 3 0.1 5 5.4 3 1026 m3.
This would require a 5% shrinkage in the diameter of the petiole.
Assuming a transpiration of 100 g/d from a leaf with 1 dm2 of lamina
and 20 cm of petiole, the supplementary water would be ;1% of the
transpiration flow.
It is likely that Eqs. 1 and 2 are not applicable in very small spaces.
(1) must overestimate the flow in spaces so narrow that some of the
volume is occupied with vicinal water. Poiseuille flow develops in pipes
100 times longer than their diameter, which is roughly the ratio in the
desmotubule channels (below).
We are seeking the pore size that will effectively restrict JvP relative
to JvD. Suppose we choose to set JvD 5 100 JvP. Most of the variables
on the two sides cancel, and we are left with
r2 5 8hD DC/100 DP.
[Vol. 85
r2 5 8hD DC/5.5 3 106 DP.
(4)
BOTANY
r 2 5 8 3 20(2.5 3 10 29 ) 3 400/(5.5 3 10 6 3 10 6 )
APPENDIX 2
JvP 5 Np r4 DP/8h.
OF
(3)
Before proceeding we must get the units right. The units of JvP are
volume flow units, m3/sec, while those of JvD are mass transfer units,
e.g., moles/sec. To equate the two fluxes they must be measured in the
same units, which can be achieved by multiplying JvP by the number of
moles of water in a cubic metre, 5.5 3 104. With this change, Eq. 3
becomes
July 1998]
CANNYAPPLICATIONS
522 pmolh21plasmo21.
This may be compared with a calculated possible diffusive flux.
Calculated water flux through the endodermal valveFrom Eq. 2 in
Appendix 2 and continuing with the notation used there
JvD 5 Npr2 D DC/l.
Values of the variablesWe have already (Appendix 2) settled on
values for D 5 2.5 3 1029 m2/s, and DC 5 400 mol /m3. The value for
l was not needed in Appendix 2, but can be taken from Tyree (1970)
as 5 3 1027 m. For the pore radius r, the argument of Appendix 2 was
that values of r less than 5 nm would turn the endodermal plasmodesmata into the one-way valve for water movement. The value of JvD
depends very critically on the value of r. The estimate of Terry and
Robards (1987) for the Abutilon nectary trichomes was that in each
plasmodesma there was an annulus of 1020 cylindrical pores each with
r 5 1.5 nm.
CalculationInserting these values in Eq. 2 gives a flux of
JvD 5 1 pmolh21plasmo21
while the generous maximum value of r 5 5 nm gives
J 5 10 pmolh21plasmo21.
ConclusionThe proposed mechanism is at the threshold of possibility and deserves more detailed consideration. If the endodermal plasmodesmata have channels somewhat larger than 1.5 nm, the diffusive
flux with the assumed values for the other variables would comfortably
supply water at the middle range of measured rates. To achieve fluxes
909