Water supply and demand remain balanced during leaf acclimation of Nothofagus cunninghamii trees


Author for correspondence:
Timothy J. Brodribb
Tel: +61 362261707
Email: timothyb@utas.edu.au


  • Higher leaf vein density (Dvein) enables higher rates of photosynthesis because enhanced water transport allows higher leaf conductances to CO2 and water. If the total cost of leaf venation rises in proportion to the density of minor veins, the most efficient investment in leaf xylem relative to photosynthetic gain should occur when the water transport capacity of the leaf (determined by Dvein) matches potential transpirational demand (determined by stomatal size and density).
  • We tested whether environmental plasticity in stomatal density (Dstomata) and Dvein were linked in the evergreen tree Nothofagus cunninghamii to achieve a balance between liquid and gas phase water conductances. Two sources of variation were examined; within-tree light acclimation, and differences in sun leaves among plants from ecologically diverse populations.
  • Strong, linear correlations between Dvein and Dstomata were found at all levels of comparison. The correlations between liquid- and vapour-phase conductances implied by these patterns of leaf anatomy were confirmed by direct measurement of leaf conductance in sun and shade foliage of an individual tree.
  • Our results provide strong evidence that the development of veins and stomata are coordinated so that photosynthetic yield is optimized relative to carbon investment in leaf venation.


Higher rates of photosynthesis are inevitably linked to higher rates of transpirational water loss from leaves as a result of the exchange of CO2 and water through stomatal pores. However, transpiration carries substantial energetic costs involved in constructing and maintaining the specialized xylem tissue that supplies water to the leaves, thereby creating a trade-off between transpirational cost and photosynthetic benefit (Cowan & Farquhar, 1977). Evidence of this trade-off is apparent in many aspects of plant form and function, from the behaviour of stomata (Farquhar et al., 1980) to the architecture of the water transport system. However, the complicated process of calculating costs and benefits associated with transpiration has prevented a predictive analysis of how whole plants should adapt to environmental conditions. Leaves provide a simplified set of cost/benefit parameters that could provide a quantitative insight into how plants optimize the balance between transpirational costs and photosynthesis.

Within leaves, the major cost of transpiration is in the provision of water supply to evaporating tissues, and here vein density (measured as the total length of leaf vascular tissue per unit leaf area; Dvein) is crucial because of its role in determining the efficiency of water transport (Brodribb et al., 2007; Boyce et al., 2009; McKown et al., 2010). Additional costs are associated with regulating the demand for water, as a consequence of the construction, maintenance and leakiness (Jordan et al., 2008) of stomata. This raises the expectation that plants should coordinate water supply and demand (Kuppers, 1984; Meinzer & Grantz, 1991; Brodribb & Feild, 2000) by maintaining a balance between Dvein and stomatal density (Dstomata) to maintain homeostasis in leaf water content. Furthermore, it should be possible to extend stomatal optimization theory (Cowan, 1986) to predict the magnitude of leaf anatomical responses to changes in key parameters such as light and humidity.

Within leaves, the responses of Dvein and Dstomata to a range of environmental cues raise the possibility that supply/demand coordination exists. For example, leaves expanded under low light produce fewer stomata and lower gas exchange rates than leaves grown in the sun (Ashton & Berlyn, 1994; Poole et al., 1996; Kürschner, 1997). There is also evidence that plants exhibit plasticity in vein density in response to different light environments and other environmental factors (Wylie, 1951; Uhl & Mosbrugger, 1999 and references therein; Zwieniecki et al., 2004), suggesting that leaves are able to control the expression of minor vein development to suit the photosynthetic and hydraulic demands (Nardini et al., 2005; Sack et al., 2005). The production of relatively high Dvein in sun leaves should result in a higher hydraulic efficiency than shade leaves, but this should come at a cost proportional to Dvein (Brodribb et al., 2007; McKown et al., 2010).

Despite evidence that the density of veins and stomata are correlated among genotypes (Oguro et al., 1985; Tanaka & Shiraiwa, 2009) and respond similarly to changes in light, it remains to be seen whether plasticity in Dvein and Dstomata leads to coordinated modification of the demand and supply of water in the leaf. Furthermore, it is unknown whether the similar responses of veins and stomata occur as a result of co-ordinated development, or merely a correlated response through pleiotropy (the control of two or more traits by the same genes), or secondary association via a different functional trait, such as leaf expansion. One means of discriminating between these alternative explanations is to test whether the responses to different inductive processes are quantitatively equivalent, and whether there is evidence for independent control of stomatal and vein density. Given the prominence of Dstomata as a proxy for leaf function, and the emerging potential of Dvein for functional reconstruction of living and fossil plants, it is important to understand the linkage between these distinct leaf characters during acclimatory and adaptive modification of the leaf.

Here, we examine the magnitude and direction of plasticity of Dvein and Dstomata to test the hypothesis that changes in the architecture of leaf xylem and stomatal density should be coordinated such that hydraulic and stomatal conductivities remain balanced. Using a common temperate tree species (Nothofagus cunninghamii), we compared within plant responses to sun and shade with responses to the disparate environmental cues embodied in differences in full sun leaves within populations, among populations and within plants. The comparison between sun and shade leaves on the same plants reflects the effects on development of a simple inductive cue, whereas the other comparisons are more complex, integrating adaptive and developmental responses to the macro- and micro-environment. In particular; within plant responses are purely developmental, responses among plants within populations reflect a combination of genetic differences and plastic responses to microsite variation, and variation among populations reflects adaptation combined with plastic responses to the macro-environment. Finally, we examined the possibility that plasticity in Dvein enables maximum photosynthetic yield relative to leaf xylem investment. Examining sun/shade plasticity in a single tree, we extended the gas exchange optimality criteria of Cowan & Farquhar (1977) to the relationship between Dvein and assimilation and compared observed plasticity in Dvein with the predicted optimal condition where photosynthetic yield per unit investment is at a maximum.

Materials and Methods


Leaves were collected in 2008 from eight naturally occurring populations of Nothofagus cunninghamii (Hook.) Oerst. in Tasmania. Nothofagus cunninghamii is a common evergreen tree that dominates cool-temperate rainforest in Tasmania. The sampled populations were ecologically diverse, ranging from tall lowland cool-temperate rainforest to subalpine shrubland (Supporting Information Table S1). From each population, we collected one upper branch growing in full sun from each of nine trees. From five of the populations we also collected one branch in full shade and two branches in full sun – one from low on the tree and one from high on the tree – from an additional tree (making a total of 10 trees sampled from these populations). The branches were chosen so that we could be confident that the leaves on the shade branches developed in the shade under a closed canopy and those on the sun branches developed in the sun. We measured vein density, stomatal density, guard cell length and leaf area from leaves randomly sampled from the branches previously described.

Sample preparation and measurement

We measured vein density (Dvein) from leaf paradermal sections prepared as follows: a sharp, double-sided razor was used to remove the upper epidermis and palisade mesophyll from whole leaves. The leaf was then placed in commercial household bleach (50 g l−1 sodium hypochlorite and 13 g l−1 sodium hydroxide) until clear, rinsed in water, and replaced in water for c. 20 min to allow residual bleach to dissipate. The resulting section was stained with 1% toluidine blue for c. 2 min, rinsed gently but thoroughly in water, and mounted on microscope slides in phenol glycerin jelly. Dvein and the longest distance between veins were measured using ImageJ (http://rsbweb.nih.gov/ij/) from digital photomicrographs of the paradermal sections at ×25 magnification. For assessment of Dvein, three fields of view were measured from each leaf, while the maximum distance between veins was measured in 10 randomly selected areoles per leaf. In each areole the vein edge was traced in ImageJ and then pen size was increased until no white space remained within the areole. The maximum pen size was converted to the maximum inter-vein distance (div). For each tree sampled for sun and shade comparisons, we prepared sections from five shade leaves, five lower sun leaves and five upper sun leaves. In addition, from all eight populations one sun leaf was prepared from each of five other plants, and three leaves from another tree.

Leaf areas were measured from the prepared whole-leaf paradermal sections by scanning the leaves at 300 pixels per inch with a flatbed scanner, then measuring the area with image analysis (ImageJ). Three replicate measurements of each leaf were made by rescanning the leaves at different orientations and positions on the flatbed scanner.

Stomatal density (Dstomata) was measured on abaxial cuticles (stomata were absent from adaxial surfaces of leaves) prepared from the same leaves on which Dvein and leaf area were measured. The cuticles were prepared by dismounting the paradermal section by gently heating it on a hot plate, cutting it lengthways, rinsing one half in warm water to remove residual jelly, and then soaking in warm 10% aqueous Cr2O3 until clear, rinsing thoroughly, staining with dilute (< 0.1%) crystal violet, rinsing, if necessary cleaning with a single-hair paintbrush, and then mounting on microscope slides, in phenol glycerin jelly. The remaining half leaf was remounted for future reference. Dstomata was also measured on older cuticle preparations from the same populations. These preparations included three leaves from each tree from each population. The trees included all those from which measurements of Dvein were made. In all cases, stomatal densities were measured from digital photomicrographs of the cuticle preparation at ×50 magnification (giving 20–50 stomata per field of view) using the counting tool in ImageJ. At least three fields of view were measured from each section.

Because maximum stomatal conductance is determined not only by Dstomata, but also by stomatal aperture size (Parlange & Waggoner, 1970), we also measured pore length to determine if the relationships of Dstomata to Dvein may be biased by variation in stomatal size.

Hydraulic and stomatal function of sun and shade leaves

A single adult tree of N. cunninghamii was used to examine acclimation between vein anatomy, leaf hydraulic conductance (Kleaf) and gas exchange. The tree was located on campus at the University of Tasmania in Hobart (147°18′30′′E, 42°54′10′′S) growing with tree ferns that shaded the lower canopy. During spring (October 2010) five fully sun-exposed and five deeply shaded branches were sampled at c. 11:00 h. Branches were immediately bagged and transferred to the laboratory where they were recut underwater, leaving a small segment bearing c. 20 well-hydrated, healthy, even-aged leaves. These small branchlets were connected to a flowmeter to measure the transpirational flux (Brodribb & Holbrook, 2006). Laboratory conditions were controlled at 22°C and 50% relative humidity and a fibre-optic light source was used to provide 800 μmol quanta m−2 s−1 at the leaf surface. Leaf temperature was monitored by two K-type thermocouples pressed against the abaxial surface of the leaf. During the establishment of a transpirational steady state, a stirred layer was maintained around the leaves by an electric fan. After 3 min at a maximum transpirational steady state, branchlets were quickly removed, immediately wrapped in plastic and foil and transferred to a pressure chamber (PMS, Albany, OR, USA) where leaf water potential was measured. The leaf hydraulic conductance was calculated as the ratio of transpirational flux to leaf water potential, and standardized to the viscosity of water at 20°C. Logged readings of humidity, leaf temperature and transpirational steady state were used to calculate mean stomatal conductance for each branchlet. Following Kleaf measurements, a subsample of two leaves from each measured branchlet was obtained and counts of Dvein and div carried out as detailed in the section ‘Sample preparation and measurement’.

Optimal vein density

To test whether the observed changes in vein density between sun and shade were optimal in terms of carbon investment, we established the relationship between assimilation rate (A) and stomatal conductance (gs) under sun (1500 μmol quanta m−2 s−1) and shade (150 μmol quanta m−2 s−1) conditions for canopy leaves of the same individual used above for hydraulic measurements. We then used a leaf hydraulic model to express gs and A in terms of Dvein according to the constraint of maintaining homeostasis in leaf water potential. Finally, we used the same economic model employed to determine optimum gs by Cowan & Farquhar (1977) to calculate optimum Dvein for N. cunninghamii leaves in the sun and shade. Rather than the optimization criterion being constant ∂E/∂A we defined constant ∂Dvein/∂A as the optimal vein investment. Clearly, this dynamic definition of optimality does not take into account important contributing factors to long-term optimality (such as leaf lifespan), but we considered this to be a good first approximation for optimal Dvein assuming vein costs increase linearly with Dvein (McKown et al., 2010).

Gas exchange:  The relationship between A and gs was determined by fitting curves to instantaneous gas exchange data collected from the same N. cunninghamii tree used in the previous section. Measurements of the dynamic response of stomata to sun/shade transitions were made on fully expanded canopy leaves inside the cuvette of a Li-6400 (Licor, Lincoln, NE, USA) gas exchange apparatus. Temperature, vapour pressure and CO2 concentrations were maintained at 22°C, 1.5 kPa and 380 μmol mol−1 CO2, respectively, while leaves were exposed to light transitions from 1500 μmol quanta m−2 s−1 (sun equivalent) to 150 μmol quanta m−2 s−1 (shade/sunfleck equivalent). Leaves were initially allowed to reach equilibrium stomatal conductance in the dark followed by illumination at 1500 μmol quanta m−2 s−1 and then 150 μmol quanta m−2 s−1 while the dynamic response of stomata was recorded. Final measurements of dark respiration and photosynthesis under 50 μmol mol−1 CO2 were made in order to determine the CO2 compensation point and maximum carboxylation capacity. Four light transitions were performed on each of five individual shoots and A, gs and internal CO2 concentration logged every 5 s. Nonlinear and linear portions of the gs vs A function could be differentiated empirically, with a carboxylation limited portion of the curve fitted using the Caemmerer & Farquhar (1981) model of photosynthesis and a linear portion at low light where electron transport is likely to limit A. The transition from linear to nonlinear portions of the A vs gs function was determined manually by maximizing r2 values for the two partial regressions.

Hydraulic model:  A previous study of extant species from across the evolutionary spectrum of vascular plants (Brodribb et al., 2007) demonstrated that the hydraulic conductance of mesophyll tissue in leaves is conservative, and hence that Kleaf could be defined by Eqn 1.

image(Eqn 1)

where dm = π/2 (div2 + dy2)1/2; assuming hydraulic flow from the vein endings to the sites of evaporation in the leaf is apoplastic and that mesophyll cells are capsular in shape, div is the longest horizontal distance from the vein terminals to the stomata and dy is the distance from the vein terminals to the epidermis. dm, div and dy are expressed in μm.

Vein density has a major influence on dm because higher Dvein yields shorter hydraulic distances between the vein and stomata. As a result of the fundamentally similar geometry in reticulate veined leaves, the relationship between Dvein and div is highly conservative across species, enabling div and hence Kleaf to be expressed in terms of Dvein (Eqn 2) (Brodribb & Feild, 2010),

image(Eqn 2)

A mean vein-epidermal thicknesses for N. cunninghamii of 82 μm was used for the parameter dy.

Under steady-state gas exchange, the water flow into leaves (F) and the vapour flow out of leaves (E) are equal and can be summarized by Eqns 3 and 4.

image(Eqn 3)
image(Eqn 4)

where F and E are hydraulic and evapo-transpirational fluxes (mmol m−2 s−1), ΔΨleaf is the water potential gradient within the leaf (MPa), υ is the leaf to air vapour pressure deficit (kPa) and boundary layer effects are assumed to be relatively small.

Rearranging Eqns 1–4 enables gs to be expressed in terms of Dvein. We parameterized equations with values that are typical for temperate-tropical environments; 2 kPa for υ and 0.25 MPa for ΔΨleaf (Brodribb & Holbrook, 2003). Finally, we used the relationship between gs and A described in the gas exchange methodology to express A in terms of Dvein under the conditions of water potential homeostasis.

Data analysis

For all analyses, data were tested for heteroscedascity and normality of residuals. No transformations were deemed necessary. Probabilities were adjusted for multiple comparisons with the Dunn–Sidak method (Sokal & Rohlf, 1995).

Overall relationship between traits among sun leaves:  The variation in sun leaves was analysed using bivariate, hierarchical analyses of variance, optimized using restricted maximum likelihood (REML), and implemented in asreml (Gilmour et al., 1995). This tool identifies variance components (including covariances between traits), which means that it is possible to identify variances and covariances uniquely attributable to different levels of a sampling hierarchy. We used bivariate REML to identify correlations between traits at the among populations, among plants within populations, and among leaves within plants levels. The method is extremely robust with regard to imbalanced sampling, and the estimated correlations are independent of other levels of the sampling hierarchy, unlike conventional correlations which are biased by the effects of variation at lower levels of the hierarchy.

The bivariate analyses were performed on pairwise combinations of Dstomata, Dvein and leaf area. The analyses followed a model with populations, plants within populations and leaves within plants as random effects for each trait. To minimize the effects of rounding errors, each variable was standardized to a variance of 1. The significance of correlations was tested using the likelihood ratio test, in which the whole model is compared to one in which the relevant correlation is constrained to be zero (Stram & Lee, 1994).

For each trait, least squares means were calculated for plants and populations. These were calculated using univariate models, the former with leaf within plant as a random effect and plant as a fixed effect, the latter with leaf within plant and plant within population as random effects, and population as a fixed effect. These analyses were implemented in jmp7 (SAS Institute Inc., Cary, NC, USA). Conventional means were calculated for individual leaves.

Analyses were also performed on the inverse of leaf area and the inverse of the square of leaf area (which are parameters that, under an assumption of passive responses to leaf expansion, would be expected to be correlated with Dstomata and Dvein, respectively). The partitioning of variance components and correlations of these parameters with Dstomata and Dvein were very similar to those of leaf area, except that (as expected) correlations showed reversed signs.

Effect of sun/shade on traits:  The effects of sun and shade on Dstomata, Dvein, guard cell length and leaf area were assessed using two-way factorial fixed effect analysis of variance with site and light conditions (sun or shade) as the two factors, and based on leaf means of the parameters. The sun leaves included leaves from both upper and lower sun branches, because preliminary analysis showed that there were no differences in these traits between leaves from these two areas, either overall or within any site. These analyses were implemented in jmp7.

Whether the shade-induced changes in Dstomata and Dvein were consistent with the variation among sun plants was tested as follows. The relationship between Dstomata and Dvein induced by shade for each of the five plants for which shade leaves were available was estimated as the change in Dvein divided by the change in Dstomata (i.e. the slope of the relationship). The slope of the relationship between Dvein and Dstomata for sun leaves was estimated using standardized major axis regression (a form of model 2 regression), implemented in smatr (Daniel S. Falster, David I. Warton and Ian J. Wright; Macquarie University, Sydney, Australia). This method was used because, unlike conventional model 1 regression, it provides slopes that are relatively unaffected by the strength of the correlation. Such slopes were calculated at the among population, among plants within populations and among leaves within population levels based, respectively, on population least squares means, residuals from a one-way analysis of variance among sites using least squares means for plants, and residuals from one-way analysis of variance among plants using leaf means.


Variation among sun leaves

In sun leaves, 38–52% of the observed variation in stomatal density, vein density and leaf area occurred among populations, but there was also large variation among plants with populations (30–35% of the total variation), and less variation (9–16%) among leaves within plants (Fig. 1; Table S2). The total variation in guard cell size was relatively small compared with the other traits (coefficient of variation of only 0.06), and it showed much higher contributions at the leaves within plant level than the other traits (Table S2).

Figure 1.

Percentage of variation (± SE) attributable to different levels of the sampling hierarchy, for leaf traits within Nothofagus cunninghamii. Values for the inverse and square root of the inverse of leaf area are very similar to those given for leaf area.

Stomatal density and vein density were very strongly correlated at the among population level (= 0.99), and less strongly (> 0.5) but still significantly at the among plants within populations and among leaves within plant levels (Table 1; Fig. 2). Stomatal density and vein density were both strongly correlated with leaf area traits at the among population level (r2 > 0.5), but not at the leaf within plant level. In each case, the relationships among populations were approximately linear with intercepts significantly greater than zero (< 0.01). At the plant within population level, Dstomata and Dvein showed strongly contrasting correlations with leaf area (= 0.17 vs = 0.53, respectively). Furthermore, the regressions of vein density on 1/√leaf area, stomatal density on 1/leaf area and vein density on stomatal density all had significant positive intercepts (< 0.01; Fig. 2).

Table 1.   Relationships in Nothofagus cunninghamii between pairs of traits in sun leaves at population, plant within population and leaf within plant levels and contrasting sun and shade leaves of the same plants
 Stomatal density and vein densityVein density and leaf areaStomatal density and leaf area
± SESlope± SESlope± SESlope
  1. Within-stratum correlations, slopes of the relationships and tests of significance (***, < 0.001; **, < 0.01) are given where applicable. The slope for sun vs shade leaves was calculated as the mean of the ratio of the differences in each trait.

Among populations0.99 ± 0.050.016***−0.85 ± 0.21−0.025**−0.82 ± 0.24−1.59**
Plants within population0.61 ± 0.130.019**−0.53 ± 0.15−0.029**−0.17 ± 0.19−1.51 NS
Sun leaves within plants0.50 ± 0.140.015***−0.26 ± 0.16−0.032 NS0.12 ± 0.132.18 NS
Sun/shade responses 0.014*** 0.065 NS 2.90 NS
Figure 2.

Associations between traits in sun leaves within Nothofagus cunninghamii. For each pair of traits, scatter plots of population means with a linear regression (closed circles and line) and leaves (grey crosses) are given on the left, and within-stratum correlations (± SE) are given on the right. (a) Vein density vs stomatal density (regression: vein density = 3.4 + 0.0160 × stomatal density; r2 = 0.99), (b) vein density vs 1/leaf area (regression: stomatal density = 233 + 5919 × 1/leaf area; r2 = 0.53), and (c) stomatal density vs 1/√leaf area (regression: vein density = 5.4 + 26.9 × 1/√leaf area; r2 = 0.61). Results of significance tests for correlations are given (***, < 0.001; **, < 0.01; *, < 0.05; NS, > 0.05).

Responses to light environment

Dstomata and Dvein, minor vein thickness and guard cell size were all very significantly greater (< 0.001) in sun leaves than in shade leaves (Fig. 3). These differences were similar at all sites, as indicated by the lack of a significant interaction between the site and sun/shade effects. The effects of shade on Dstomata, Dvein and minor vein thickness were large (Figs 3, 4), whereas the changes in stomatal size (as measured by guard cell size) were small. Leaf area was much greater in shade leaves at the lowest altitude site (Black River) but differed little (and not significantly) at the other sites. This altitude-dependent response to shade is consistent with that observed by Hovenden & Vander Schoor (2006). For all five plants, the change in Dvein in response to shade was in proportion to the change in Dstomata, as predicted from the overall relationships among sun plants. This is indicated by the close similarity of slopes for the sun/shade response to the slope of the overall relationship both among populations, among plants within populations and among leaves within plants (Fig. 5; Table 1). Z-tests comparing these slopes showed no significant differences among them. By contrast, the relationships between stomatal or vein density and leaf area were not consistent among these different levels of response, with different slopes apparent.

Figure 3.

The effect of shade on leaf traits of a single Nothofagus cunninghamii plant from each of five populations (arranged in order of increasing altitude). Least squares means (± SE) for shade (shaded columns) and sun (open columns) leaves are shown for each plant. For stomatal density (a), vein density (b), vein thickness (c) and guard cell length (d), there were significant differences between sun and shade, and among populations (< 0.001), and no significant interaction between the two factors (> 0.05). For leaf area (e), there was a highly significant interaction (< 0.001) between the two factors, but when the Black River plant was excluded, the shade and interaction effects were insignificant (> 0.05).

Figure 4.

Representative paradermal sections (a, b) and cuticle preparations (c, d) of sun (a, c) and shade (b, d) leaves of Nothofagus cunninghamii. Note the higher density of veins and stomata in the sun leaves. Bar, 400 μm.

Figure 5.

The effect of shade on vein density, stomatal density and leaf area, showing pairwise comparisons within a single Nothofagus cunninghamii plant from each of five populations for sun (open symbols) and shade (closed symbols) leaves, with standard errors. Lines connect means from the same plant. Leaf means from all sun leaves are shown as grey circles, with the linear regressions through these leaf scores indicated by a dashed line. (a) Vein density vs stomatal density; (b) stomatal density vs leaf area; (c) vein density vs leaf area.

A strong relationship between the spacing of veins (a determinant of hydraulic conductance) and Dstomata (an index of transpiration rate) was demonstrated by the highly significant linear correlations between 1/div and Dstomata (Fig. 6). Correlations were observed in both sun (= 21 422x + 53.783; r2 = 0.45) and shade (y =26 163− 23.085; r2 = 0.66) sampled leaves, and in both cases regression intercepts were not significantly different from zero (Fig. 6a). Despite the different ranges of 1/div and Dstomata in sun and shade leaves, the slopes of the two regressions were similar.

Figure 6.

Regressions of leaf means of stomatal density on 1/interveinal distance for sun (open symbols, dashed line) and shade (closed symbols, unbroken line) leaves of Nothofagus cunninghamii. Both regressions had highly significant slopes (< 0.01 in both cases) but neither had an intercept significantly different from 0, indicating a strong proportionality between vein spacing and stomatal density.

Acclimation of anatomical and hydraulic properties

Measurements of liquid- and gas-phase leaf conductances from sun and shade leaves of an N. cunninghamii tree showed that stomatal conductances in shade leaves were on average 70% lower (30.0 ± 11 compared with 101 ± 16 mmol m−2 s−1, respectively; =5) and Kleaf was 62% lower (3.24 ± 0.6 compared with 8.55 ± 1.1 mmol m−2 s−1 MPa−1, respectively; = 5) than in sun leaves. Among all leaves from sun and shade, gs and Kleaf were strongly correlated (Fig. 7a) with a nonsignificant intercept. The corresponding anatomical variation between sun and shade leaves in this individual was similar to that observed across populations, with a 35% reduction of Dvein in shade leaves relative to sun leaves and a 10% decline in the distance from veins to epidermis (113 ± 15 to 127 ± 16 μm, respectively). Although vein spacing was strongly correlated with stomatal conductance, both Kleaf and gs were lower than predicted according to the expected proportionality with 1/div (Fig. 7b).

Figure 7.

(a) Measured hydraulic and stomatal conductances in sun (open) and shade (closed) leaves from an individual tree of Nothofagus cunninghamii. Data for individual leaves as well as means for sun and shade (± SD) are shown. (b) Relationships between vein spacing and stomatal conductance and Kleaf (insert). Dotted lines show expected proportional relationships. All differences between sun and shade are highly significant (< 0.01).


From the dynamics of stomatal responses to sun and shade light intensities (1500–150 μmol quanta m−2 s−1), we defined the optimum sun and shade gs (58 and 96 mmol m−2 s−1, respectively) from the inflection point of the relationship between A and gs. Using a hydraulic model to determine the influence of Dvein on A, we found a similar inflection point was evident, reflecting the transition from light to CO2 limitation of photosynthesis (Fig. 8). The vein densities at the two inflection points are theoretically the optimum values for the two light intensities (approximating sun and shade). The observed shift in Dvein from sun to shade (8.9 to 5.7 mm mm−2) was similar to the shift in optimum Dvein, although in both sun and shade leaves we found Dvein was slightly higher (20–35%) than the optimum value.

Figure 8.

Modelled relationships between vein density and assimilation for sun (bold curve) and shade conditions (thin curve). The inflection point in both curves is attributable to the transitions from light-limited to diffusion-limited photosynthesis, and this transition point defines the optimum gas exchange and vein density for sun and shade conditions. Symbols indicate mean (± SD) measured values from sun (closed circle) and shade leaves (open circle) from a single tree. Although 20–35% higher than optimum, the observed shift in Dvein was similar in magnitude to the predicted values.


It is well known that the densities of stomata and leaf veins (Dstomata and Dvein) are plastic, particularly in response to light. Our data illustrate for the first time that the responses of Dstomata and Dvein to environmental conditions achieve a homeostatic balance between hydraulic and stomatal conductivities. Proportionality between Dstomata and Dvein was strong at different levels of the genetic hierarchy ranging from within plants to within and among populations (Table 1). Furthermore, we confirmed the functionality of the observed anatomical connection between veins and stomata by demonstrating that conductances to liquid and water vapour in sun and shade leaves of N. cunninghamii remain tightly correlated (Fig. 7) and proportional to assimilation rate, thereby maintaining optimal vein allocation relative to photosynthetic gain.

Optimization and coordination of water supply and demand in response to irradiation level

Studies of leaf gas exchange demonstrate that angiosperms dynamically regulate the apertures of their stomata to minimize transpirational costs while maximizing CO2 assimilation (Cowan, 1986). Our results comparing sun and shade leaves show that these optimization principles extend to developmental control of veins and stomata in the leaf (Fig. 8). Reduced energy for electron transport means that shade leaves achieve much lower photosynthetic yields than sun leaves, and coordination between photosynthetic rate and stomatal conductance (Wong et al., 1979) results in much lower stomatal conductances in shade leaves. Accordingly we found N. cunninghamii shade leaves produced on average 30% fewer stomata than sun leaves of the same plants. Assuming both stomatal and vein synthesis incur significant costs to the plant, then the most efficient use of resources would occur if Dvein was reduced to match the lower Dstomata in the shade, and this is what we observed in the plasticity of these parameters within a single tree (Fig. 8). Although the sequence here is described as a stomata-first response to light intensity, the veins may equally be the primary sensor, or both cell types simultaneously, without altering the significance of the resultant coordination.

Other papers have demonstrated that vein density is related to maximum photosynthesis, stomatal anatomy and transpiration in sun leaves, and that shade-adapted species tend to produce lower vein density (Schuster, 1908; Wylie, 1951; Nardini et al., 2005; Sack et al., 2005). However, the data here provide strong evidence that, at least in N. cunninghamii, reduced demand for water in the shade is quantitatively matched by a reduction in hydraulic supply associated with a substantial reduction in Dvein. Measured conductances in sun and shade leaves confirmed that parallel acclimation of vein and stomatal density to light resulted in a very close coordination between water supply and demand (Fig. 7a). Optimal allocation of carbon to leaf veins occurs if Dvein and maximum assimilation rate vary in proportion, thereby maintaining a constant ratio of yield to xylem investment (Cowan & Farquhar, 1977). We found that the observed changes in stomatal and vein density resulted in a photosynthetic rate that remained close to the optimal value in terms of instantaneous cost vs benefit (Fig. 8). The observed correlations between vein density, stomatal density and leaf gas exchange therefore invoke a developmental coordination of tissues responsible for transpiration and water delivery to achieve optimal carbon allocation (Figs 6, 8).

It is noteworthy that, although the reduction in Kleaf from sun to shade was proportional to the reduction in stomatal conductance (Fig. 7), Kleaf in the shade was lower than expected given the reduction in Dvein (Brodribb et al., 2007). This lower than expected Kleaf may be a consequence of a decline in the conductivity of the leaf xylem itself, or in the conductivity of the mesophyll connection between veins and the sites of evaporation. Changes in leaf thickness cannot explain the lower than expected Kleaf in shade leaves because the vein-to-epidermal distance in shade leaves was c. 10% shorter than in sun leaves, meaning shade leaves should have displayed slightly higher, not lower than expected Kleaf (Noblin et al., 2008). Alternatively if the distribution of resistances within the venation had changed such that major veins had substantially lower conductances in the shade, then this also could explain low Kleaf in the shade (McKown et al., 2010). However, this would have required a very large change in the conductances of the midrib and first vein orders, and we noted only minimal changes in the diameter of these lower vein orders between sun and shade (data not shown). A 15% reduction in minor vein diameter (Fig. 3c) was observed between sun and shade leaves, and this may have contributed to reducing Kleaf relative to vein density, but most studies show that the mesophyll hydraulic pathway dominates leaf resistance (Sack & Holbrook, 2006). Under these circumstances it seems most likely that reduced mesophyll conductance in shade plants may augment the effect of lower Dvein. Decreased mesophyll hydraulic conductance in the shade may have been a result of differences in aquaporin density or mesophyll anatomy, but considering that all leaves were measured at light intensities > 500 μmol quanta m−2 s−1 it is assumed that aquaporin production was at maximum levels in both sun and shade leaves (Cochard et al., 2007).

Within and among population responses in sun leaves

Strong patterns of correlation between stomatal density and vein density were not only evident between sun and shade leaves of N. cunninghamii but also among the sample of fully sunlit leaves. Highly significant correlations between Dvein and Dstomata of sun leaves at population, plant within population and leaf within plant levels reinforce the argument for vein–stomatal coordination. In particular, the virtually identical slopes in the stomatal density/vein density relationship regardless of whether the variation was induced by shade or by other factors that vary among and within populations provide evidence that multiple extrinsic factors induce parallel responses in stomatal and vein densities. It is difficult to explain such parallel responses except in terms of coordinated responses, particularly given that the lack of within plant association between leaf size and vein density suggests that variation in vein density is not a passive result of differential whole-leaf development (Fig. 2). It is not clear what the specific drivers of the among and within population variation were, but there were major environmental differences among the sample sites, as a result of both altitude, which spanned c. 1100 m, and other factors including geology and regional climatic patterns. Several factors that vary with altitude, such as decreasing atmospheric pCO2, increasing irradiation (Körner, 1999) and possibly increasing vapour deficits (e.g. Leuschner, 2000), may have been implicated. A combination of low ambient pCO2 and high vapour deficit at high altitudes should favour increased stomatal and vein densities by increasing demand for water for a given rate of CO2 uptake (Gale, 1972; McElwain, 2004). Indeed, systematic increases in stomatal density with altitude have been observed, including in one Nothofagus species (Kouwenberg et al., 2007). As little of the observed variation in stomatal density and leaf size among populations along an altitudinal gradient appears to be genetically based (Hovenden & Vander Schoor, 2006), the strong associations between leaf area, vein density and stomatal density at the population level may be largely the consequence of correlated morphological acclimation.

Developmental coordination

The coordinated plasticity in veins and stomata seen here was only partially related to leaf size (Figs 3, 5), meaning that either a common growth hormone stimulates both vein and stomatal development, or one developmental process affects the other. There is a possibility that auxin might act as a coordinating hormone in this respect. Auxin is fundamentally important for the differentiation and patterning of leaf veins (Sachs, 1993) and there is evidence that mutations affecting auxin carriers also result in abnormal stomatal differentiation (Mayer et al., 1993; Spitzer et al., 2009). A common role for auxin in vascular and stomatal development would allow integration of these two developmental processes, but as yet no direct evidence associates auxin in stomatal developmental processes. However, high light intensity appears to directly up-regulate stomatal differentiation (Boccalandro et al., 2009), and it is possible that increased epidermal porosity in expanding leaves might increase auxin fluxes in the leaf lamina and simulate branching of minor veins. Clearly this subject is ripe for investigation.


We found very strong coordination of vein density and stomatal density, in spite of independent control of these parameters, implying that some linking developmental process may guide the optimal development of veins to match the stomatal demand for water. Similar patterns of coordination occur within and between N. cunninghamii plants, emphasizing the fundamental nature of this developmental organizing principle. Other plant species are known to show differing degrees of plasticity in vein and stomatal characters, including patterns opposite to those shown here (Schuster, 1908), and future research will show whether coordination is a universal characteristic or an ecologically variable trait. Given the strong emphasis on stomatal and vein densities as proxies of plant function and atmospheric conditions, it is critical to understand the degree of independence of the two leaf attributes. In this context, our data provide a basis for understanding and predicting the coordinated plasticity of stomatal and vein density traits in response to environment.


This work was supported by ARC Discovery Grant DP0559226 to T.B. and G.J. We thank Scott McAdam and Madeline Murphy for collecting material, and preparing and measuring cuticles and paradermal sections.