Leaf hydraulic vulnerability is related to conduit dimensions and drought resistance across a diverse range of woody angiosperms

Authors


Author for correspondence:
Tim Brodribb
Tel: +61 3 62261707
Email: timothyb@utas.edu.au

Summary

  • Hydraulic dysfunction in leaves determines key aspects of whole-plant responses to water stress; however, our understanding of the physiology of hydraulic dysfunction and its relationships to leaf structure and ecological strategy remains incomplete.
  • Here, we studied a morphologically and ecologically diverse sample of angiosperms to test whether the water potential inducing a 50% loss in leaf hydraulic conductance (P50leaf) is predicted by properties of leaf xylem relating to water tension-induced conduit collapse. We also assessed the relationships between P50leaf and other traits considered to reflect drought resistance and ecological strategy.
  • Across species, P50leaf was strongly correlated with a theoretical predictor of vulnerability to cell collapse in minor veins (the cubed ratio of the conduit wall thickness to the conduit lumen breadth). P50leaf was also correlated with mesophyll traits known to be related to drought resistance, but unrelated to traits associated with carbon economy.
  • Our data indicate a link between the structural mechanics of leaf xylem and hydraulic function under water stress. Although it is possible that collapse may contribute directly to dysfunction, this relationship may also be a secondary product of vascular economics, suggesting that leaf xylem is dimensioned to avoid wall collapse.

Introduction

The ability of plants to maintain hydraulic conductance under conditions of water stress is a central driver of species’ distribution patterns (Engelbrecht et al., 2007). Because physical tension increases in the xylem when leaf water potentials fall as a result of transpirational water loss, the hydraulic pathway from the roots to the shoots is exposed to stresses that can compromise the capacity of plants to transport water. Although this tension-induced loss of hydraulic conductance is often attributed to cavitation resulting from air bubbles entering the water column via pit membranes (Zimmermann, 1983; Tyree & Sperry, 1989), it may also be a consequence of xylem wall implosion and cell collapse (Cochard et al., 2004; Brodribb & Holbrook, 2005) or increased extra-xylary resistance (Brodribb & Holbrook, 2004). Hydraulic dysfunction has serious implications for plant function because photosynthesis and growth are dependent on the efficient supply of water to the sites of evaporation (Hubbard et al., 2001; Brodribb & Holbrook, 2007). The vulnerability of the hydraulic pathway to dysfunction is typically assessed as P50, or the tension required to cause a 50% decline in hydraulic conductance. In leaves, P50 has been linked to plant survival (Blackman et al., 2009; Brodribb & Cochard, 2009), and stem P50 has been shown to be adaptive across broad taxonomic groups in relation to gradients in water availability (Brodribb & Hill, 1999; Pockman & Sperry, 2000; Maherali et al., 2004).

Hydraulic vulnerability to dysfunction is a highly integrated component of a suite of physiological and anatomical traits that reflect different patterns of hydraulic response to drought. In dry climate environments, P50 in stems is correlated with the minimum seasonal water potential experienced by species in the field (Pockman & Sperry, 2000; Bhaskar et al., 2007; Jacobsen et al., 2007). Coordination between loss of leaf hydraulic conductance and the regulation of stomatal conductance also suggests that hydraulic vulnerability to dysfunction in leaves plays an important role in plant responses to short-term water stress (Brodribb et al., 2003). Others have demonstrated a strong positive correlation between wood density (WD) and P50 in stems (Hacke et al., 2001) and have emphasized the inherent costs of increased resistance to hydraulic dysfunction in terms of both xylem cell wall reinforcement and narrower xylem conduits that reduce hydraulic efficiency (Hacke et al., 2006) and affect plant growth (Poorter et al., 2010).

Most studies of the functional and ecological significance of hydraulic vulnerability have focused on stems (Hacke et al., 2001, 2009; Maherali et al., 2004). However, water transport in leaves is functionally distinct from that in stems and, because of their relatively high hydraulic resistance (Sack & Holbrook, 2006), leaves impose significant constraints on maximum stomatal conductance and photosynthetic capacity (Brodribb et al., 2005). Compared with stems, leaves are often more vulnerable to hydraulic dysfunction (Salleo et al., 2000; Brodribb et al., 2003; Choat et al., 2005a; Hao et al., 2008). They also differ in xylem structure. Unlike the xylem in stems (Wagner et al., 1998), much of the leaf xylem is not reinforced to withstand mechanical rupture under dynamic or static loads, and therefore may be vulnerable to cell collapse under negative pressure. Cell collapse has been linked to leaf hydraulic dysfunction in some conifers (Cochard et al., 2004; Brodribb & Holbrook, 2005), although cavitation in the petioles and midribs of a number of conifer and angiosperm species has also been reported (Nardini et al., 2001; Johnson et al., 2009a). The large volumes of air inside these leaves create maximum pressure differentials across xylem cell walls and create a substantial risk of xylem implosion in the leaf veins. However, this phenomenon has not been observed in angiosperms, and no studies have examined how leaf xylem anatomy may relate to drought resistance across different angiosperm species.

Here, we examine how interspecific variation in the vulnerability of leaves to hydraulic dysfunction (P50leaf) is related to a number of structural and functional traits in a sample of woody temperate angiosperm species with a broad range of leaf forms and rainfall preferences. Specifically, we tested whether structural properties of the leaf xylem were related to hydraulic vulnerability, on the basis that the wall thickness and lumen diameter of leaf xylem conduits determine their capacity to resist implosion by water tension. We also tested for relationships between leaf hydraulic vulnerability and other leaf structural and functional traits that are widely recognized to influence drought tolerance and reflect plant ecological strategy.

Materials and Methods

Plant species and habitat

We sampled 20 phylogenetically disparate woody angiosperm species from montane rainforest (15 species) and dry sclerophyll forest (five species) on the island of Tasmania, in cool temperate Australia (Table 1). Nineteen of these species were evergreen, but ranged in their degree of scleromorphy (as reflected by leaf mass per unit area, LMA) from 137 g m−2 in the relatively broad leaves of the rainforest species Atherosperma moschatum to 772 g m−2 in the extremely scleromorphic needles of Hakea lissosperma (Table 1). The sample group also included the winter deciduous species Nothofagus gunnii with LMA = 102 g m−2. One of the evergreen species, Tasmannia lanceolata, was vessel-less. Climatic limits in terms of minimum water availability, as reflected by the fifth percentile of mean annual rainfall across each species’ distribution, ranged from 351 mm yr−1 for the dry forest species Bursaria spinosa to 1268 mm yr−1 for the montane rainforest species Orites diversifolia (C. J. Blackman et al., unpublished). These species are known to vary widely in leaf xylem vulnerability to hydraulic dysfunction, which, in turn, is closely correlated with the estimates of climatic limits for water availability described above (C. J. Blackman et al., unpublished).

Table 1.   Compiled data (mean ± SD) of leaf hydraulic vulnerability (P50leaf), absolute leaf hydraulic conductance (Kleaf), leaf mass per unit area (LMA) and wood density (WD) for all species sampled in the study
FamilySpeciesP50leaf (−MPa)Kleaf (mmol m−2 s−1 MPa)LMA (g m−2)WD (g cm−3)
Montane rainforest
  AsteraceaeOlearia pinifolia (Hook.f.) Benth.1.71 ± 0.032.52 ± 0.40314.1 ± 13.20.643 ± 0.016
  AtherospermataceaeAtherosperma moschatum Labill.1.48 ± 0.033.14 ± 0.45137.2 ± 0.60.558 ± 0.019
  EricaceaeCyathodes straminea R.Br.2.00 ± 0.152.91 ± 0.88199.5 ± 11.30.702 ± 0.041
  EricaceaeGaultheria hispida R.Br.1.32 ± 0.046.75 ± 1.05172.3 ± 8.10.536 ± 0.02
  EricaceaeRichea scoparia Hook.f.1.41 ± 0.093.79 ± 0.65201.2 ± 20.40.550 ± 0.015
  MyrtaceaeEucalyptus coccifera Hook.f.2.65 ± 0.158.91 ± 1.66253.4 ± 17.00.639 ± 0.012
  NothofagaceaeNothofagus cunninghamii (Hook.) Oerst.1.70 ± 0.113.88 ± 0.81186.7 ± 13.70.583 ± 0.019
  NothofagaceaeNothofagus gunnii (Hook.) Oerst.1.53 ± 0.0413.53 ± 0.62102.0 ± 4.40.643 ± 0.02
  PittosporaceaePittosporum bicolor Hook.1.87 ± 0.083.25 ± 0.67251.5 ± 18.50.730 ± 0.02
  ProteaceaeHakea lissosperma R.Br.2.85 ± 0.2413.91 ± 2.81772.4 ± 71.10.656 ± 0.022
  ProteaceaeLomatia polymorpha R.Br.1.57 ± 0.124.93 ± 0.78292.8 ± 17.70.700 ± 0.033
  ProteaceaeOrites diversifolia R.Br.1.25 ± 0.109.94 ± 3.17380.9 ± 16.40.743 ± 0.023
  ProteaceaeTelopea truncata (Labill) R.Br.1.58 ± 0.089.18 ± 1.75243.3 ± 9.00.689 ± 0.019
  RubiaceaeCoprosma nitida Hook.f.1.95 ± 0.0412.82 ± 1.70174.7 ± 17.40.720 ± 0.02
  WinteraceaeTasmannia lanceolata (Poir) A.C.Sm.1.56 ± 0.133.54 ± 0.64212.0 ± 19.70.698 ± 0.021
Dry sclerophyll forest
  AsteraceaeOlearia hookeri (Sond.) Benth.2.36 ± 0.119.83 ± 1.54227.9 ± 36.00.828 ± 0.005
  MyrtaceaeEucalyptus pulchella Desf.4.31 ± 0.368.03 ± 1.69217.8 ± 13.60.681 ± 0.045
  PittosporaceaeBursaria spinosa Cav.3.20 ± 0.083.33 ± 0.54132.1 ± 11.80.757 ± 0.019
  ProteaceaeHakea microcarpa R.Br.3.96 ± 0.2215.11 ± 3.31628.1 ± 25.80.658 ± 0.019
  ProteaceaeLomatia tinctoria (Labill) R.Br.2.08 ± 0.134.22 ± 1.56177.3 ± 10.90.769 ± 0.012

Vulnerability to hydraulic dysfunction

For each species, leaf hydraulic vulnerability curves were constructed by measuring the percentage loss of leaf hydraulic conductance from maximum values (Kmax) in leaves rehydrated from a range of leaf water potentials (Ψleaf). For the purposes of these curves, Kleaf was measured by assessing the kinetics of Ψleaf relaxation upon leaf rehydration (Brodribb & Holbrook, 2003). Briefly, hydrated branches from three individuals of each species were cut early in the morning and immediately bagged to arrest water loss. Having transported them to the laboratory, the branches were allowed to desiccate slowly at light intensities sufficient to ensure light-induced hydraulic function (c. 50 μmol quanta m−2 s−1) over a maximum of 48 h or until the percentage loss of Kleaf approached 100%. Initial Ψleaf was determined by measuring leaves neighbouring the sample leaf in a Scholander pressure chamber (PMS, Albany, OR, USA). The sample leaf was then cut under water and allowed to rehydrate for a period of between 30 and 300 s depending on the initial Ψleaf. Final Ψleaf was measured with the pressure chamber, and Kleaf was calculated from the ratio of the initial to final Ψleaf and the capacitance of the leaf:

image(Eqn 1)

[Ψo, initial leaf water potential (MPa); Ψf, final leaf water potential (MPa); T, duration of rehydration (s); Cleaf, leaf capacitance (mmol m−2 MPa−1) calculated from Eqn 2]. Leaf vulnerability was analysed by plotting Kleaf against Ψo. We tested the possibility that Kleaf may have changed during rehydration (Brodribb & Holbrook, 2006) by comparing calculated Kleaf values after rehydration times ranging from 30 to 300 s. Calculated Kleaf remained indepen-dent of rehydration time in all species, confirming the suitability of rehydration for the measurement of vulnerability. The vulnerability of leaf hydraulic conductance to decreasing water potential (P50leaf) for individual species was defined as the Ψleaf value at which Kleaf had declined by 50% from the mean maximum rate (Kmax) for each species. For each species, P50leaf was calculated by fitting a three-parameter sigmoidal regression function of the form, Kleaf (%) = 100/[1 + ea(Ψleafb)], to the Kleaf vs Ψo data.

Leaf capacitance was measured directly by rehydrating leaves connected to a flow meter that measured the volume of water entering the leaf through the petiole. Here, leaf capacitance was calculated as the volume of water taken up by the leaf during a transition from Ψo to Ψf:

image(Eqn 2)

F, total water uptake into the leaf during rehydration adjusted for leaf area (mmol m−2 s−1) and temperature following Brodribb & Holbrook (2006); Ψo, initial leaf water potential (MPa); Ψf, final leaf water potential (MPa); T, duration of rehydration (s)]. Leaf samples were collected from fully expanded sun-exposed shoots. Leaves were always rehydrated from initial water potentials that fitted within the range before incipient loss of conductance. The mean leaf capacitance for each species was then used in Eqn 1 to calculate Kleaf.

Pressure–volume traits

For each species, one leaf from each of six neighbouring plants was sampled for the determination of leaf turgor dynamics from pressure–volume (PV) analysis (Tyree & Hammel, 1972). For species with small leaves or reduced leaf petioles, small terminal shoots were sampled. Branches were cut under water in the morning and rehydrated until Ψleaf was > 0.05 MPa, after which leaves were detached and allowed to slowly desiccate on the laboratory bench. During this process, the leaf weight (weighed to 0.0001 g) and water potential were measured periodically.

The turgor loss point (TLP) was determined from the inflection point of the graph of 1/Ψleaf vs the relative water content. The calculation of osmotic potential at full turgor (Ψπ) followed Kirkham (2005). The bulk modulus of elasticity (ε), meanwhile, was calculated from the change in pressure potential at full turgor:

image(Eqn 3)

[P, pressure potential; R, relative water content; V, volume of symplastic water] (Schulte & Hinckley, 1985).

Water potentials

Predawn and midday water potentials were measured during late summer 2008. Measurements were made on three replicate samples collected from montane rainforest species co-occurring on Mount Wellington and from the dry forest species Hakea microcarpa (Table 1). Rainfall for Mount Wellington is spread fairly uniformly throughout the year (Bureau of Meteorology; http://www.bom.gov.au). However, because samples were collected at the end of several years of below-average rainfall, midday water potentials at this time were measured as an estimate of the minimum water potential (Ψmin) experienced by plants in the field. To determine the extent to which these environmental conditions impacted leaf hydraulics in situ, Ψmin was transposed to each species’ vulnerability curve, and the percentage loss (if any) of Kleaf was calculated. The distance (MPa) between Ψmin and the hydraulic limit (P50leaf) was defined as the safety margin for each species.

Leaf xylem dimensions

Anatomical dimensions of the leaf xylem of the minor veins were measured from transverse sections using a Nikon DS-L1 digital camera connected to a light microscope at ×100 magnification amplified by a 2.5× magnification tube. Leaf sections from three individuals of each species were cut using a freeze-microtome, stained with 5% toluidine blue, and mounted on glass microscope slides in phenol glycerine jelly. Lumen breadth (b) and wall thickness (t) were measured on all or, at most, five adjacent conduits from each of three minor veins per leaf section per species, giving a total of c. 45 conduits per species (see Fig. 2). Both b and t were measured by adjusting the microscope focus up and down and clearly defining the thickest part of the helically thickened cell wall; b was calculated as the average of the maximum and minimum diameters of each lumen. Leaf minor veins were defined as the highest vein order that retained clearly distinguishable xylem and phloem anatomy. Xylem conduits associated with the leaf minor veins were carefully distinguished from cells associated with vein endings, which were enlarged and comprised of sclereids in some species. In hydrated leaves, the conduits in the leaf minor veins were very close to circular in cross-section, with relatively minimal contact between cells, and hence did not conform well to the double-wall bending model for stem xylem (Hacke et al., 2001). The theoretical collapse pressure for these conduits (ρcr) was therefore calculated using Timoshenko’s equation (Eqn 4) for pipe buckling under pressure:

image(Eqn 4)

[E, elastic modulus (MPa); v, Poisson ratio; b, conduit lumen diameter; t, wall thickness]. There are no estimates for the radial elastic modulus of leaf xylem for angiosperms, and estimates for wood xylem vary tremendously, with upper values as high as 3000 MPa (Bergander & Salmen, 2000). However, direct experimental analysis of wood fibres subjected to transverse compression suggests that the elastic modulus of lignified woody cells falls in the range 50–70 MPa (Shiari & Wild, 2004). Because the xylem conduits in leaf minor veins are generally helically rather than more uniformly thickened, they are likely to have relatively low elastic moduli. We estimated the elastic modulus for the current sample of angiosperms by calculating Eqn 4 using a range of E values until ρcr corresponded to a 1 : 1 relationship with P50leaf. The Poisson ratio for lignin (0.28) was used (Innes, 1995).

LMA and WD

LMA and WD were measured for each species in order to test for links between leaf vulnerability and ecologically relevant plant traits. LMA was measured in 5–20 mature, fully expanded, sun-exposed leaves within the most recent growth cohort, from five individuals of each species. Leaf areas were measured as projected areas with a flatbed scanner and image analysis software (Image J; National Institutes of Health, Bethesda, MD, USA). Leaves were then placed in an oven at 70°C for at least 3 d, and LMA was calculated as the ratio between the dry leaf mass and leaf area.

WD was measured on five individuals of each species. From each species, five 4-cm-long wood samples were taken from 3–5-yr-old stems, as inferred from counts of growth rings. The pith and bark were removed from each sample, and fresh volume was measured by water displacement. The sample mass was determined after drying for at least 3 d at 70°C, and WD was calculated as the ratio between oven-dried mass and fresh volume (g cm−3).

Trait correlations

Linear regression analyses were used to examine key interspecific trait correlations (Sigmaplot; SPSS Inc., Chicago, IL, USA). Regressions or differences were considered to be significant if  0.05.

Results

For all species, the response of leaf hydraulic conductance to decreasing water potential was sigmoidal, with an initial plateau, followed by a decline in Kleaf to a minimum value close to zero as Ψleaf declined (Fig. 1 shows two vulnerability curves typical of species with low and high leaf hydraulic vulnerability; Supporting Information Fig. S1 shows vulnerability curves for the remaining species). Among the sampled species, mean P50leaf ranged from −1.25 MPa in Orites diversifolia to −4.3 MPa in Eucalyptus pulchella (Table 1).

Figure 1.

 Percentage loss of leaf hydraulic conductance from maximum (Kmax) as leaf water potentials decline for two representative species with high (Atherosperma moschatum, a) and low (Eucalyptus coccifera, b) leaf hydraulic vulnerability. Curves fitted are sigmoidal functions. Solid vertical lines indicate the water potential at 50% loss of Kleaf (P50leaf).

At the cellular level, a strong and highly significant linear relationship (r2 = 0.81) was found between mean P50leaf and the mean ratio (t/b)3 for cell dimensions in the leaf minor veins for each species. A wide range of values was found for this ratio, with high P50leaf associated with narrower, thicker walled cells (Fig. 2). Across species, mean wall thickness and lumen breadth were unrelated. Of these two cell dimensions, P50leaf was only correlated significantly with lumen breadth (r= 0.25; < 0.05). Conduit dimensions in the leaf minor veins varied across species, with values for lumen breadth ranging from 2.78 ± 0.4 μm in Richea scoparia to 5.38 ± 1.4 μm in Orites diversifolia and conduit wall thickness ranging from 0.44 ± 0.05 μm in Richea scoparia to 0.8 ± 0.2 μm in Coprosma nitida (Table S1). Substituting the mean ratio (t/b)3 for each species in Eqn 4 to calculate the theoretical collapse pressure for conduits in the leaf minor veins of each species, a strong correspondence with P50leaf was found if the cell wall elastic modulus was set at 200 MPa (Fig. 3).

Figure 2.

 Relationship between leaf hydraulic vulnerability (P50leaf) and the conduit dimensions that dictate cell vulnerability to collapse, assuming that conduit geometry in the minor veins approximates a cylindrical tube. Data points are mean ± SD across all surveyed cells. Regression line and significance level are shown (***, < 0.001). Inset images show detail of the conduit dimensions in the leaf minor veins for two species typical of high (Atherosperma moschatum; i) and low (Hakea microcarpa; ii) leaf hydraulic vulnerability. Bars, 10 μm.

Figure 3.

 Regression lines through the theoretical collapse strength of leaf xylem conduits (ρcr) calculated for each species (Eqn 4) using estimates of cell wall elastic modulus (E) of 100 and 300 MPa for conduits in the leaf minor veins. Data points with error bars show the variation of ρcr calculated for each species using E = 200 MPa. Using this value of E, the theoretical collapse strength was found to closely correspond to a 1 : 1 relationship with P50leaf across species.

The in situ minimum water potential (Ψmin) varied among the 16 species examined, and was also correlated significantly with P50leaf (Fig. 4; r2 = 0.48; < 0.01). The safety margin between Ψmin and P50leaf also correlated with P50leaf (r= 0.82; < 0.001), and was generally lower in species that were more vulnerable to hydraulic dysfunction (Fig. 4). However, according to predicted reductions in Kleaf at Ψmin for each species, more vulnerable species were not more exposed than less vulnerable species to hydraulic dysfunction at Ψmin (Fig. S2). The species that was predicted to suffer the greatest loss in Kleaf (27.8%) under drought conditions observed in the field was the moderately vulnerable species Olearia pinifolia (Fig. S2).

Figure 4.

 Relationship between leaf hydraulic vulnerability (P50leaf) and seasonal minimum water potential (Ψmin) for a subset of 16 species. Data points are mean ± SD. Regression line and significance level are shown (**, < 0.01). For each species, the difference between P50leaf and Ψmin was defined as the safety margin (MPa) before significant hydraulic dysfunction occurred under drought conditions (dashed line).

Significant correlations were found between P50leaf and the osmotic and bulk elastic characteristics of the leaves. Strong positive linear relationships between P50leaf and both TLP (Fig. 5a) and osmotic potential at full turgor (Ψπ; Fig. 5b) were evident across species means; however, P50leaf was only weakly positively correlated with leaf bulk modulus of elasticity (ε; Fig. 5c). A trade-off between leaf hydraulic vulnerability to dysfunction and absolute leaf hydraulic conductance (Kleaf) was not observed across the species sampled (Fig. 6a). In addition, P50leaf was not correlated significantly with either WD (Fig. 6b) or LMA (Fig. 6c). Further to this, neither WD nor LMA was correlated significantly with the climatic limits in terms of minimum water availability across each species’ distribution (Fig. S3).

Figure 5.

 Relationships between leaf hydraulic vulnerability (P50leaf) and cell turgor traits derived from pressure–volume analysis. P50leaf correlated significantly with both the turgor loss point (TLP, a) and osmotic potential at full turgor (Ψπ100, b), but only weakly with the bulk modulus of elasticity (ε, c). Data points are mean ± SD for (a) and (b), and species’ means for (c). Regression lines and significance levels are shown (***, < 0.001; *, < 0.05).

Figure 6.

 Absolute leaf hydraulic conductance (Kleaf) (a), wood density (WD) (b) and leaf mass per unit area (LMA) (c) as a function of leaf hydraulic vulnerability (P50leaf). Data points are mean ± SD. Regression lines and significance levels are shown (ns, > 0.05).

Discussion

The strong correlation between P50leaf and the theoretical predictor of xylem collapse across our broad sample of species indicates an important link between xylem anatomy and P50leaf that may be either causal, or at least indicate a highly conservative margin between the pressure causing cell collapse and P50leaf. Furthermore, our results suggest that P50leaf is coordinated with a suite of other leaf traits related to drought resistance. However, P50leaf was unrelated to leaf hydraulic conductance and other important traits widely recognized as indicators of ecological strategy (e.g. WD and LMA). Amongst our sample of woody plants, P50leaf therefore represents an important trait dimension describing plant ecological strategy in relation to drought resistance, independent of traits correlated with carbon economy. This is, to our knowledge, the first study to examine directly how interspecific variation in P50leaf is coordinated with both leaf structural traits and other functional traits relevant to land plant ecology.

Leaf vulnerability and cell collapse

The response of leaf hydraulic conductance to declining leaf water potentials across our sample of species was typical of other species (Brodribb & Holbrook, 2003; Hao et al., 2008) in following a sigmoidal trajectory, whereby the percentage loss of conductance increased as Ψleaf approached TLP. The most conventional explanation for drought-induced dysfunction is xylem cavitation and subsequent conduit embolism (Nardini et al., 2001; Salleo et al., 2001; Johnson et al., 2009a). However, the strong and significant correlation between P50leaf and (t/b)3 in our sample of plants adds weight to other studies suggesting that conduit collapse may also contribute significantly to decreasing Kleaf as leaf water potential declines. In studies of conifers, evidence of progressive leaf xylem cell collapse was shown to mirror leaf hydraulic dysfunction (Cochard et al., 2004; Brodribb & Holbrook, 2005). These authors also showed that the theoretical collapse pressure for these conduits, calculated from equations that describe cell wall resistance to bending (Hacke et al., 2001) and pipe buckling under tension (Timoshenko, 1930), corresponded to actual cell collapse observed during leaf dehydration.

The relationship between the theoretical collapse pressure (ρcr) for the conduits of our sampled angiosperms and the measured pressure at which losses in Kleaf occurred, depends on the elastic modulus used to calculate ρcr. The estimates of elastic modulus (E) used in our calculations are in broad agreement with estimates of E for xylem conduits in the leaves of a number of conifer species (Cochard et al., 2004; Brodribb & Holbrook, 2005). If E for xylem in the leaf minor veins across our sample of species was on the order of 100–300 MPa (as determined here), conduits in the minor veins would begin to collapse when water potentials in these conduits decline to c. P50leaf. In transpiring leaves, the water potential in these minor veins will be higher than that of the mesophyll, but lower than that in the major veins and petiole. As such, water potential in these minor veins may differ from the bulk leaf water potential reflected in P50leaf. This discrepancy is difficult to estimate, but may not be large, given that Sack et al. (2005) suggested that resistances upstream and downstream of the minor veins were comparable in magnitude. Further to this, a preliminary survey of leaf xylem dimensions across different plant groups indicates that the ratio of cell wall thickness to cell lumen breadth is constant across different vein orders (C. J. Blackman et al., unpublished), suggesting that xylem throughout much of the leaf is equally vulnerable to collapse-induced dysfunction. Such collapse-induced dysfunction has important implications for the physiology of dehydrated leaves, particularly in terms of hydraulic recovery. For example, hydraulic recovery from wall collapse could occur more readily than recovery from xylem embolism when pressures are largely negative (Tyree & Yang, 1992), which may help explain the rapid recovery of Kleaf observed in plant species exposed to moderate drought stress (Lo Gullo et al., 2003; Blackman et al., 2009).

Although we showed a strong relationship between P50leaf and (t/b)3, there was no evidence of a direct link between cell collapse and hydraulic dysfunction in leaves. Alternatively, it may be that the linkage between these traits is the result of an evolved coordination between xylem structural strength and P50leaf. This would suggest that a safety factor is maintained between cavitation-induced hydraulic dysfunction and conduit wall collapse in leaf xylem. Maximum economy in terms of vein synthesis will be achieved if the wall reinforcement of xylem conduits in leaf veins is the minimum required to avoid implosion at negative pressures sufficient to cause conduit cavitation. Hacke et al. (2001) used this economic argument to explain their observed relationship between P50 and xylem implosion resistance in woody stems. Our results would be consistent with such a process if our estimate of the elastic modulus of the conduit walls was incorrect, leading to biased estimation of collapse pressure, or if the water potential in the minor veins is considerably less negative than Ψleaf.

Other studies have emphasized the role of xylem cavitation and subsequent conduit embolism in driving reduced leaf hydraulic conductance (Nardini et al., 2001; Bucci et al., 2003; Johnson et al., 2009a). Johnson et al. (2009a), for example, provided good evidence that linked leaf hydraulic dysfunction to cavitation events in a small number of conifer and angiosperm species. However, these authors detected cavitation solely in the leaf midrib where, relative to the minor veins, xylem is more structurally reinforced by surrounding fibres and may thus be more prone to cavitation than collapse. We therefore propose that future studies aimed at identifying the processes that drive leaf hydraulic dysfunction need to carefully examine different vein orders and how they behave under drying conditions.

Leaf vulnerability and traits related to drought

Our result of a strong relationship between P50leaf and the minimum water potential experienced by plants in the field (Ψmin) is consistent with previous studies in stems (Pockman & Sperry, 2000; Bhaskar et al., 2007; Jacobsen et al., 2007). Because Ψmin in the current study was largely measured in co-occurring species, interspecific variation in Ψmin, and therefore P50leaf, might be related to variation in rooting depth that reflects different water foraging strategies. Although this is typical of co-occurring species with a broad range of vulnerabilities in Mediterranean climates (Pockman & Sperry, 2000), here Ψmin was measured in co-occurring species from montane rainforest where high annual rainfall greatly reduces the probability of strong moisture gradients developing through the soil profile. Under such mesic conditions, high drought resistance is unlikely to be advantageous. However, we measured Ψmin at the end of a multi-year period of below-average rainfall and thus some species may have been affected by soil drying. Furthermore, such episodic drought may have contributed over time to the ecologically anomalous co-existence of the observed broad range of tolerances to water stress in this generally high-rainfall site. Equally, the observed variation in Ψmin among these species might be related to variation in stomatal sensitivity to leaf water potential, whereby the stomata of species with xylem that is more resistant to drought-induced dysfunction remain open at lower water potentials, thereby allowing Ψmin to fall below that of more sensitive species.

Interestingly, we also observed a significant relationship between P50leaf and the hydraulic safety margin (Ψmin – P50leaf) across the species examined. This indicates that species with more negative P50leaf in the current sample tend to exhibit broader safety margins from hydraulic dysfunction under drought conditions than more vulnerable species. Although Meinzer et al. (2009) reported similar results between hydraulic safety and daily minimum stem water potential, these authors argued that large safety margins were a consequence of high sapwood capacitance and low WD. Despite this, we found no evidence to suggest that species with more vulnerable leaves and narrower safety margins were more susceptible to hydraulic dysfunction during drought (or had lower WD). However, some of the species sampled are likely to experience reduced Kleaf in the field, and may operate at water potentials sufficiently low to induce some hydraulic dysfunction. Regular loss of Kleaf has been reported in several species (Nardini et al., 2003; Johnson et al., 2009b), suggesting that plants are able to readily reverse leaf hydraulic dysfunction. Indeed, rapid recovery of Kleaf is likely if the principal cause of leaf hydraulic dysfunction is cell collapse in minor veins. If, on the other hand, slowly reversible cavitation is the process of leaf hydraulic dysfunction, this would suggest that a number of these species on Mount Wellington are at the limit of their ecological tolerance, and that further climatic change towards increasing drought severity may create conditions for catastrophic hydraulic decline, resulting in local population extinction.

Our results place P50leaf at the centre of a suite of coordinated leaf traits related to drought resistance. We observed significant relationships between P50leaf and pressure–volume traits that describe the behaviour of leaves during increasing water deficit. Among these was a significant relationship between P50leaf and the water potential associated with TLP. This result suggests that, despite the strong link between P50leaf and xylem anatomy, leaf hydraulic dysfunction may be partly the consequence of reduced extra-xylary conductance as TLP is approached (Brodribb & Holbrook, 2004; Knipfer & Steudle, 2008). Further to this, previous studies have reported a complex association between hydraulic dysfunction, stomatal closure and TLP, suggesting that leaf hydraulic vulnerability influences the timing of stomatal closure during water stress (Brodribb et al., 2003). However, it is clear that species with inherently more negative TLPs consistently show greater tolerance of lower leaf water potentials (Kubiske & Abrams, 1994; Bucci et al., 2004) and greater ecological amplitude in terms of minimum water availability (Lenz et al., 2006).

The weak, but significant, relationships between P50leaf and both osmotic potential at full turgor (Ψπ100) (Fig. 5b) and bulk modulus of elasticity (Fig. 5c) suggest that increased leaf hydraulic vulnerability scales with the leaf tissue properties outside the xylem. Although low osmotic potentials at full turgor are entirely compatible with low leaf water potentials, and may enable water uptake from drying soil, both increases and decreases in bulk elastic modulus have been attributed to aiding survival during drought (Roberts et al., 1980). The negative relationship between leaf hydraulic vulnerability and bulk elastic modulus (ε) supports data suggesting that high ε confers greater drought resistance by allowing large adjustments of leaf water potential with only small changes in leaf water content (Niinemets, 2001).

Leaf vulnerability as an independent trait dimension

Many previous studies of interspecific variation in stem P50 have pointed to the possible costs and trade-offs associated with increased resistance to hydraulic dysfunction. For example, a trade-off between xylem safety and efficiency has been demonstrated in the stems of a number of arid-land shrub species (Hacke et al., 2009) on the basis that vulnerability to cavitation is correlated with the diameter of the largest pores (with the least hydraulic resistance) in intervessel pit membranes (Choat et al., 2005b). However, this relationship is inconsistent across species (Maherali et al., 2004). According to our data, it is also inconsistent in leaves. Indeed, there is no reason to expect such a trade-off in leaves, considering that Kleaf and P50leaf are coordinated with different aspects of leaf structure and function. Kleaf is related to traits such as vein density and hydraulic architecture (Sack & Frole, 2006; Nardini et al., 2008; Brodribb & Feild, 2010) that influence water flux and gas exchange, whereas P50leaf is associated with a different suite of traits related to the ratio of xylem conduit wall thickness and lumen breadth (t/b) and pressure–volume characteristics (Sack & Holbrook, 2006). Although not significant, we observed a trend towards increased resistance to hydraulic dysfunction and increased hydraulic conductance in leaves. This pattern may reflect the variety of leaf hydraulic pathways and structures that influence conductivity in different species. For example, a high concentration of water-conducting sclereids embedded in the mesophyll of the two Hakea species in the current study may account for the very high Kleaf values recorded for these species (Brodribb et al., 2010), which were also characterized by low leaf hydraulic vulnerability.

Unlike stem xylem hydraulic vulnerability, we found no correlation between P50leaf and WD across our sample of species, which was not surprising considering that WD also relates to a number of variables that describe wood mechanical properties (Chave et al., 2009). More surprising was the lack of a significant relationship between P50leaf and LMA, considering that LMA tends to increase with leaf density (Niinemets, 1999). LMA varies across a broad range of biomes (Reich et al., 1999) and is coordinated with a suite of traits related to carbon economy. It is also widely thought to be negatively related to precipitation and linked to a number of leaf tissue traits related to drought survival (Niinemets, 2001). Our results therefore suggest that neither WD nor LMA is a good predictor of resistance to hydraulic dysfunction under water stress in the species examined here. However, it has long been argued that scleromorphy (the underlying trait assessed by LMA) is a direct response, not to dry climates, but rather to the need for the protection of long-lived leaves against a diverse array of damage types, especially under resource-poor conditions (Turner, 1994; Jordan et al., 2005). Thus, in our species, variation in LMA may instead partly reflect the evolution of scleromorphy within the regional flora in response to nutrient-deficient soils (Hill, 1998). We also found that neither LMA nor WD was correlated significantly with each species’ climatic limits in terms of water availability. Following this, care needs to be taken when applying functionally correlative traits, such as LMA and WD, to analyses aimed at understanding both ecosystem functioning and the combined effect of climate change and increased drought on different plant communities.

Conclusion

Our results point to leaf hydraulic vulnerability to water stress (P50leaf) as a key functional trait strongly linked to leaf structure, anatomy and ecological tolerance. In particular, we provide evidence across a diverse group of species of a strong relationship between the conduit dimensions (t/b)3 in leaf minor veins and P50leaf. Further work is necessary to determine whether this correlation is causal or a secondary product of vascular economics. We also show strong relationships between P50leaf and traits within the leaf symplast, suggesting that processes in the mesophyll may also contribute to leaf hydraulic dysfunction. Further examination of these relationships will advance our understanding of plant physiological responses to water stress.

Acknowledgements

The Australian Research Council provided support in the form of a grant (DP0878342) to T.J.B. and G.J.J. and a post-graduate scholarship to C.J.B.

Ancillary