Trade-offs between leaf hydraulic capacity and drought vulnerability: morpho-anatomical bases, carbon costs and ecological consequences

Authors


Author for correspondence:

Andrea Nardini

Tel: +39 040 5583874

Email: nardini@units.it

Summary

  • Leaf hydraulic conductance (Kleaf) and vulnerability constrain plant productivity, but no clear trade-off between these fundamental functional traits has emerged in previous studies.
  • We measured Kleaf on a leaf area (Kleaf_area) and mass basis (Kleaf_mass) in six woody angiosperms, and compared these values with species’ distribution and leaf tolerance to dehydration in terms of P50, that is, the leaf water potential inducing 50% loss of Kleaf. We also measured several morphological and anatomical traits associated with carbon investment in leaf construction and water transport efficiency.
  • Clear relationships emerged between Kleaf_mass, P50, and leaf mass per unit area (LMA), suggesting that increased tolerance to hydraulic dysfunction implies increased carbon costs for leaf construction and water use. Low P50 values were associated with narrower and denser vein conduits, increased thickness of conduit walls, and increased vein density. This, in turn, was associated with reduced leaf surface area.
  • Leaf P50 was closely associated with plants’ distribution over a narrow geographical range, suggesting that this parameter contributes to shaping vegetation features. Our data also highlight the carbon costs likely to be associated with increased leaf tolerance to hydraulic dysfunction, which confers on some species the ability to thrive under reduced water availability but decreases their competitiveness in high-resource habitats.

Introduction

The hydraulic architecture of terrestrial plants imposes fundamental constraints on their gas exchange and productivity (Tyree & Ewers, 1991), as a consequence of the imperative need to balance unavoidable water losses during active photosynthesis with adequate water supply to the foliage (Sperry, 2000), thus preventing eventual leaf desiccation, stomatal closure and death under water stress conditions (Tyree et al., 2002; Brodribb & Cochard, 2009).

Leaf hydraulics, as the major hydraulic bottleneck of the plant hydraulic system (Yang & Tyree, 1994; Nardini, 2001), strongly limit plant productivity (Brodribb et al., 2005), competitive success (Hao et al., 2008) and distribution (Sack & Frole, 2006). The efficiency of the leaf tissues in supplying water to the evaporating surfaces of the mesophyll is quantified by the leaf hydraulic conductance (Kleaf), which is generally expressed on a leaf area basis (Kleaf_area) (Sack & Holbrook, 2006). Photosynthetic rates are positively correlated to Kleaf_area (Brodribb et al., 2005, 2010; Brodribb & Holbrook, 2007) as a consequence of tight coordination of liquid and gas phase conductances in the leaf, functionally addressed at maintaining relatively stable leaf water potential (Meinzer, 2002; Nardini & Salleo, 2003; Savvides et al., 2012).

The magnitude of Kleaf_area depends on the hydraulic properties of both the vascular pathway through leaf veins (Sack et al., 2004; Brodribb et al., 2007) and the extra-vascular pathway involving apoplastic and symplastic routes (Zwieniecki et al., 2007; Nardini et al., 2010), although the relative importance of the different routes varies from species to species and as a function of environmental conditions (Cochard et al., 2004; Lee et al., 2009; Guyot et al., 2012). Kleaf_area has been reported to be positively correlated to vein density (Brodribb et al., 2007, 2010; Boyce et al., 2009), reflecting the importance of a relatively short water pathway in the high-resistance extra-vascular compartment for achievement of high leaf hydraulic capacity, although other studies failed to confirm such a correlation (Nardini et al., 2005, 2012; Scoffoni et al., 2011).

Leaf hydraulic conductance is known to vary substantially, on both short and long time-scales, as a function of different environmental parameters such as temperature (Sellin & Kupper, 2007), light (Scoffoni et al., 2008) and water availability (Scoffoni et al., 2012), as well as during leaf ontogenesis and ageing (Salleo et al., 2002; Aasamaa et al., 2005). In particular, drought poses a serious threat to the maintenance of sufficient Kleaf and active photosynthesis, as a consequence of the relatively high drought vulnerability of leaves compared with stems and roots (Nardini et al., 2003; Choat et al., 2005; Hao et al., 2008). Leaf hydraulic vulnerability is generally quantified in terms of P50, that is, the leaf water potential (Ψleaf) value inducing 50% loss of maximum Kleaf (Brodribb & Holbrook, 2003). A substantial body of evidence points to cavitation-induced embolism of leaf xylem conduits as a major cause of Kleaf decline under drought stress conditions (Kikuta et al., 1997; Salleo et al., 2001; Nardini et al., 2003; Johnson et al., 2012). Minimum daily Ψleaf values experienced by plants have been frequently reported to be close to or even lower than values of P50 (Johnson et al., 2011), implying that leaves are frequently exposed to partial loss of Kleaf that needs to be recovered during periods of low transpiration (Lo Gullo et al., 2003; Nardini et al., 2008; Johnson et al., 2009), a process probably incurring substantial metabolic costs for refilling embolized conduits (Nardini et al., 2011).

It is thus not surprising that several lines of evidence point to an adaptive value of low P50 values for plants growing in drought-prone habitats. As an example, Blackman et al. (2010) reported that P50 was lower in species from dry sclerophyll forests than in those from montane rainforest habitats. In a follow-up study by the same authors (Blackman et al., 2012), it was further shown that P50 was correlated to rainfall availability in 20 woody angiosperms growing in Tasmania. However, it is still not clear whether drought-vulnerability of Kleaf can influence species’ distribution even over short distances (e.g. in response to edaphic factors; Pockman & Sperry, 2000; Hao et al., 2008; Gortan et al., 2009), thus eventually playing a role in shaping vegetation composition at local scales. Moreover, the morpho-anatomical bases and eventual carbon costs associated with adaptation/acclimation of hydraulic vulnerability of leaves are still largely unexplored. According to Blackman et al. (2010), P50 is correlated to leaf xylem conduit traits with special reference to lumen diameter and wall thickness. However, no trade-off was found between Kleaf_area, P50, and morpho-anatomical traits reflecting carbon costs for leaf construction and/or generally associated with drought tolerance, such as leaf mass per unit area (LMA) (Poorter et al., 2009). Scoffoni et al. (2011) have recently highlighted the importance of small leaf dimensions and high major vein density as determinants of leaf tolerance to hydraulic dysfunction, but also in this case measurements of Kleaf_area across 10 different species failed to capture eventual trade-offs between hydraulic vulnerability and leaf construction costs. Although Kleaf_area provides an adequate measure of the ‘sufficiency’ of the leaf hydraulic system to supply water to the evaporating surface, alternative scaling methods (e.g. by leaf mass, Nardini et al., 2012) are also plausible (Tyree et al., 1998). As an example, in some species bulk water flow outside leaf veins is known to be directed toward all mesophyll tissues and not specifically to the epidermis (Zwieniecki et al., 2007). In this case, Kleaf_area would not adequately represent the sufficiency of the leaf hydraulic system to supply water to the bulk of leaf tissues, while Kleaf_mass might provide more complete information. In a recent study, Simonin et al. (2012) have shown that Kleaf_mass is negatively correlated with leaf lifespan, suggesting that leaf water use and transport efficiency are optimized across species to maximize leaf carbon gain.

In the present study, we describe novel relationships between hydraulic/physiological and morpho-anatomical traits for leaves of congeneric species of the genera Acer and Quercus. The individuals selected for the present study were growing in natural populations within a narrow geographical range but under contrasting ecological conditions mainly arising from different soil characteristics of study sites, influencing water availability and/or evaporative demand.

Our aim was to investigate the coordination, correlation, and eventual trade-offs between: leaf hydraulic traits describing efficiency of water supply and drought vulnerability; leaf water status; morpho-anatomical traits reflecting the efficiency of the water transport system; and the costs for leaf construction in terms of biomass investment.

Materials and Methods

Plant material and study sites

Measurements were performed in species of the genera Acer and Quercus (genus Acer: Acer pseudoplatanus L., Acer campestre L. and Acer monspessulanum L.; genus Quercus: Quercus petraea Matt. Liebl., Quercus pubescens Willd. and Quercus ilex L.). Three species per genus, with contrasting ecological requirements in terms of soil water availability, were selected to highlight eventual adaptive inter-specific differences of leaf hydraulic and morpho-anatomical features. In particular, A. pseudoplatanus (Aps) and A. campestre (Aca) are relatively drought-vulnerable, shade-tolerant trees (Lens et al., 2011), adapted to grow on deep soils with high water availability. Acer monspessulanum (Amo) is a drought-tolerant tree preferentially growing in sunny and warm habitats (Tissier et al., 2004). Quercus petraea (Qpe) is a mesophilous oak preferentially growing in deep soils and displaying higher drought vulnerability than congeneric Q. pubescens (Qpe) and Q. ilex (Qil), which thrive in drought-prone sites (Cochard et al., 1992; Tyree & Cochard, 1996). All species are winter-deciduous, with the exception of Qil which is an evergreen, and leaf size decreases from the drought-vulnerable to the drought-adapted species in each genus (Fig. 1).

Figure 1.

Scans of leaves from the three Acer and three Quercus species studied.

Natural populations of all the above species are found in a narrow geographical range within the Province of Trieste (north-eastern Italy). Although total annual precipitation is rather homogeneous across the province (c. 1000 mm), the area is characterized by sharp ecological gradients arising from different edaphic conditions. Areas where karstic limestone soils with low water retention capacity dominate alternate with zones where the presence of emerging flysch soils ensures higher water availability to plants. Soil water availability of study sites was quantified in terms of total available water content (AWC), as based on dedicated maps from the Regional Agency for Agriculture Development (http://www.ersa.fvg.it; see also Gortan et al., 2009).

Aps and Aca were sampled in the forest area surrounding the Department of Life Sciences, University of Trieste (45°39′41″N, 13°47′44″E; altitude 160 m asl; AWC = 80 mm). Several Qpe individuals were growing near the village of Monrupino (45°43′00″N, 13°46′00″E; altitude 420 m asl, AWC = 77 mm). Large populations of Amo and Qpu were growing near the village of Prosecco (45°41′51″N, 13°44′31″E; altitude 260 m asl; AWC = 44 mm). Finally, Qil is among the dominant woody species in the area of Cernizza, near Villaggio del Pescatore, (45°46′49″N, 13°35′31″E; altitude 25 m asl; AWC = 42 mm).

All measurements were performed between May and July 2011. Leaf and branch samples were always collected from south-east to south-west exposed parts of the crown, and immediately transported to the laboratory while wrapped in plastic bags to prevent dehydration. Three individuals per species were sampled, growing at a maximum distance of 30 m from each other.

Pressure–volume traits

Leaves for pressure–volume analysis were sampled in the field before 0800 h and rehydrated for 60 min by immersing their petioles in deionized water. Leaves were wrapped in plastic film and inserted into a pressure chamber (3005 Plant Water Status Console; SoilMoisture Corp., Santa Barbara, CA, USA) to record their initial Ψleaf. Experiments (five or six replicates per species) continued only if Ψleaf ≥ −0.1 MPa.

Immediately after the first Ψleaf measurement, leaves were weighed on a digital balance and the plastic film was removed to allow leaves to dehydrate while resting on the balance. After 2–5 mg of water had been lost, Ψleaf was re-measured and the procedure was repeated until the relationship between 1/Ψleaf and water loss at four or five successive points became strictly linear. Pressure–volume curves were elaborated according to Tyree & Hammel (1972) to calculate leaf osmotic potential at full turgor (π0) and water potential at the turgor loss point (Ψtlp). At the end of each experiment, leaf surface area (Aleaf) was measured using a leaf area meter (Li3000A; Li-Cor Inc., Lincoln, NE, USA). Finally, samples were dried in an oven at 70°C for 48 h and their dry weight (DW) was recorded. Leaf capacitance (Cleaf) was calculated from pressure–volume curves on the basis of the slope of the relationship between Ψleaf and water loss, and normalized by Aleaf or DW, thus obtaining Cleaf_area and Cleaf_mass, respectively. These values were used to calculate Kleaf values on the basis of the rehydration kinetic technique (see next paragraph).

Leaf hydraulic conductance and vulnerability

Leaf hydraulic conductance was measured using the rehydration kinetic technique as originally described by Brodribb & Holbrook (2003). This technique is based on the measurement of leaf water potential changes before and after leaf rehydration (Ψ0 and Ψf, respectively), as well as of rehydration time (t) and leaf capacitance (Cleaf) as:

display math(Eqn 1)

Current-year shoots were collected in the field and transported to the laboratory where they were rehydrated for 1–4 h. The shoots were then bench-dehydrated until Ψleaf reached values between −0.2 and −4.0 MPa. Three leaves per shoot where wrapped in plastic film and the shoot was enclosed in a plastic bag to stop transpiration and allow equilibration of leaf water potential values. After 15 min, two wrapped leaves where used to estimate Ψ0. The experiment continued only if the difference between the Ψ0 values of these two leaves was within 0.1 MPa. The third wrapped leaf was cut while the petiole was immersed in water and rehydrated for 30–90 s before measuring Ψf. Kleaf was then calculated using Eqn (1) and values of Cleaf_area or Cleaf_mass, to obtain final values of Kleaf_area or Kleaf_mass, respectively. Maximum leaf hydraulic conductance was calculated as the average of Kleaf values recorded for well-hydrated leaves (Ψ0 ≥ −0.5 MPa). All leaves used for the rehydration kinetics experiments were measured for Aleaf and DW as described in the previous section.

Field measurements of leaf water potential

Leaf water potential was measured on selected sunny days in spring (15–20 May) and summer (12–17 July) 2011. Two leaves per individual were sampled for a total of six leaves per species. Leaves were collected between 1200 and 1400 h to estimate the minimum daily Ψleafmin). Leaves were immediately wrapped in plastic film and transported to the laboratory in a refrigerated bag, where Ψleaf was measured using the pressure chamber within 2 h of sampling.

Morpho-anatomical measurements

LMA was calculated as DW/Aleaf for all leaves used for pressure–volume analysis or Kleaf measurements. Vein density (DeV) was measured after leaf clearance and vein staining. Samples of leaf lamina (5 × 5 mm) were cut from four to six leaves per species. Samples were maintained in 1 M KOH for 5–7 d. Samples were repeatedly washed with deionized water, immersed in 0.5% toluidine blue for 1 min, and then washed again in deionized water. Leaf samples were observed under a light microscope (at ×4 and ×10 magnification) equipped with a digital camera. Leaf images were acquired and then analyzed using ImageJ (http://rsbweb.nih.gov/ij/index.html).

Anatomical dimensions of leaf mesophyll and midrib xylem were measured from transverse sections taken from the middle portion of three leaves per species. In particular, midrib xylem conduit features were measured as a proxy for xylem traits of higher order veins. Leaf samples were fixed overnight at 4°C in 6% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 6.9) and postfixed for 2 h in 1% osmium tetroxide in the same buffer. The specimens were dehydrated in a graded series of ethyl alcohol and propylene oxide and embedded in araldite. Thin sections (1 μm) were obtained with an ultramicrotome (Ultracut; Reichert-Jung, Vienna, Austria), stained with equal volumes of 1% toluidine blue and 1% sodium tetraborate and observed under a light microscope (DMR5000; Leica, Wetzlar, Germany). Images were taken using a digital camera (DFC425C; Leica).

The density of midrib xylem conduits (DeC) was calculated by counting all conduits for each midrib and dividing this number by the area of midrib xylem. Lumen diameter (DiC) and wall thickness (TCW) were measured on at least 10 randomly selected midrib conduits per section. Palisade and spongy parenchyma thicknesses (TPA and TSP, respectively) were measured at 0.5–1 cm distance from the midrib. Finally, the distance of minor veins from the lower epidermis (DVE) was measured for all veins visible in the cross-sections (generally two or three veins per section), although the exact order of visible veins could not be determined.

Statistics

Statistical analysis was performed using the software SigmaStat v. 2.0 (Systat Software Inc., San Jose, CA, USA). The significance of differences between species was tested using one-way ANOVA followed by Tukey's post hoc comparisons. The significance of correlations was determined using the Pearson product-moment coefficient. Regressions or differences were considered to be highly significant if  0.05. Regressions with a P value between 0.05 and 0.2 were considered as moderately significant.

Results

Pressure–volume traits and plant water status

Pressure–volume analysis revealed several differences among Acer and Quercus species growing in sites with different water availability. Both Cleaf_area and Cleaf_mass were higher in Aps and Aca, and significantly lower in the other species, with Qil showing the lowest Cleaf_mass (Table 1). Values of Ψtlp ranged between −1.4 and −1.9 MPa in Aps and Aca and between −2.2 and −2.4 MPa in Amo and Qpe, respectively (Table 1). The lowest Ψtlp values were recorded in the two drought-adapted oaks (Qpu and Qil). Differences in terms of Ψtlp were paralleled by similar inter-specific differences in terms of π0 (Table 1).

Table 1. Leaf capacitance on a leaf area (Cleaf_area) or dry mass (Cleaf_mass) basis, leaf water potential at the turgor loss point (Ψtlp) and leaf osmotic potential at full turgor (π0) as derived from leaf water potential isotherms (pressure–volume curves) in three Acer and three Quercus species
SpeciesCleaf_area (mmol MPa−1 m−2)Cleaf_mass (mmol MPa−1 g−1)Ψtlp (MPa)π(MPa)
  1. Means are reported ± SD. Different letters indicate statistically significant differences (< 0.05) between species.

A. pseudoplatanus 624 ± 70 a12.4 ± 1.4 a−1.4 ± 0.3 a−1.0 ± 0.1 a
A. campestre 595 ± 98 a16.0 ± 2.9 b−1.9 ± 0.1 b−1.2 ± 0.1 a
A. monspessulanum 397 ± 61 b5.9 ± 1.2 c−2.2 ± 0.3 c−1.6 ± 0.2 b
Q. petraea 311 ± 18 bc4.2 ± 0.7 c−2.4 ± 0.1 c−1.7 ± 0.1 b
Q. pubescens 313 ± 41 bc3.2 ± 0.3 c−2.9 ± 0.2 d−2.3 ± 0.2 c
Q. ilex 344 ± 30 bc2.6 ± 0.2 d−2.9 ± 0.3 d−2.3 ± 0.1 c

In May, field measurements of Ψleaf did not reveal significant differences among Acer species (Fig. 2). Ψleaf above −1.0 MPa was recorded in Qpe and Qpu, while Qil showed significantly lower values (Fig. 2). In July, Aps maintained the highest Ψleaf while values for both Aps and Amo dropped to c. −2.5 MPa. Among Quercus species, Ψleaf was higher in Qpe than in Qpu and Qil (Fig. 2).

Figure 2.

Minimum daily water potential (Ψleaf) as measured in May (black columns) or July (gray columns) in three Acer and three Quercus species (Aps, A. pseudoplatanus; Aca, A. campestre; Amo, A. monspessulanum) and three Quercus (Qpe, Q. petraea; Qpu, Q. pubescens; Qil, Q. ilex). Mean values are reported ± SD. Different letters indicate significant differences between species.

Leaf hydraulic conductance and vulnerability

Maximum Kleaf_area ranged between 13.1 ± 1.7 and 6.9 ± 1.0mmol s−1 MPa−1 m−2 as recorded in Aps and Amo, respectively. Within each genus, Kleaf_area was significantly higher in species growing in sites with higher water availability (Aps and Qpe) than in congeneric species thriving in drier soils (Fig. 3). Qualitatively similar, but quantitatively larger differences between species were observed in terms of Kleaf_mass, which was found to range between 258.5 ± 34.2 and 59.4 ± 13.7 mmol s−1 MPa−1 kg−1 in Aps and Qil, respectively.

Figure 3.

Maximum leaf hydraulic conductance on a leaf area (a) or leaf dry mass (b) basis, as measured in three Acer (Aps, A. pseudoplatanus; Aca, A. campestre; Amo, A. monspessulanum) and three Quercus (Qpe, Q. petraea; Qpu, Q. pubescens; Qil, Q. ilex) species. Means are reported ± SD. Different letters indicate significant differences between species.

Linear relationships were recorded between Kleaf_area and Ψleaf, although the slope differed between the species studied (Fig. 4). On the basis of regression analysis, values of leaf water potential inducing 50% loss of Kleaf (P50) were calculated. In the genus Acer, P50 ranged between −1.19 and −1.89 MPa in Aps and Amo, respectively, reflecting species-specific differences in drought tolerance. Qpe turned out to be the most drought-vulnerable oak (P50 = −1.96 MPa) and Qil the most resistant (P50 = −3.50 MPa). Values of P50 were correlated to soil AWC (r = 0.799; P = 0.057), thus indicating that species growing in the driest sites displayed the highest resistance to drought-induced leaf hydraulic dysfunction, and vice versa.

Figure 4.

Changes of leaf hydraulic conductance on a leaf area basis (Kleaf_area) as a function of leaf water potential (Ψleaf) in three Acer and three Quercus species. Dotted lines indicate the Ψleaf values inducing 50% loss of Kleaf_area.

Morpho-anatomical traits

Within each genus, Aleaf decreased and LMA increased from species growing under high water availability to those growing in drought-prone sites (Table 2). Several other anatomical differences were observed between drought-vulnerable and drought-resistant species. Vein density was significantly lower in the former that in the latter. Similarly, xylem conduit density increased from drought-vulnerable to drought-resistant species in both genera, while the opposite trend was observed for xylem conduit diameter (Table 2). Also, conduit wall thickness increased in each genus from mesic to xeric species/sites. As an example, in Acer species TCW was c. 1.4 μm in Aps and c. 50% higher in Amo (c. 2.1 μm), and even larger differences were recorded between Qpe and Qil in this respect (Table 2). Although TPA, TSP and DVE showed statistically significant differences among species, no clear habitat-related trend could be observed with reference to these anatomical traits (Table 2).

Table 2. Leaf surface area (Aleaf), leaf mass per unit area (LMA), vein density (DeV), midrib xylem conduit density (DeC), diameter (DiC) and wall thickness (TCW), palisade and spongy parenchyma thickness (TPA and TSP, respectively), and distance from veins to lower epidermis (DVE) as measured in leaves of three Acer and three Quercus species
SpeciesAleaf (cm2)LMA g (m−2)DeV (mm mm−2)DeC (mm−2)DiC (μm)TCW (μm)TPA (μm)TSP (μm)DVE (μm)
  1. Means are reported ± SD. Different letters indicate statistically significant differences (P < 0.05) between species.

A. pseudoplatanus 63.0 ± 11.5 a49.1 ± 7.5 a9.5 ± 0.6 a1979 ± 414 a12.0 ± 1.0 a1.4 ± 0.1 a64.3 ± 3.5 a71.8 ± 2.6 a15.5 ± 4.0 a
A. campestre 30.0 ± 5.1 b44.2 ± 7.9 a10.0 ± 0.6 a3482 ± 306 b8.8 ± 0.4 b1.7 ± 0.3 a56.1 ± 6.8 a52.5 ± 11.9 b11.4 ± 6.9 a
A. monspessulanum 16.5 ± 2.7 c71.0 ± 8.8 b11.6 ± 0.5 b5669 ± 157 c8.4 ± 0.6 b2.1 ± 0.2 b68.6 ± 2.4 a73.4 ± 9.0 a36.0 ± 6.0 b
Q. petraea 54.4 ± 13.9 a74.3 ± 9.1 b9.9 ± 1.2 a3547 ± 101 b10.9 ± 0.4 a1.2 ± 0.3 a85.8 ± 1.3 b48.1 ± 8.1 b26.7 ± 4.2 b
Q. pubescens 25.1 ± 2.3 b96.6 ± 9.7 c11.5 ± 0.7 b2623 ± 329 b8.2 ± 0.5 b1.4 ± 0.1 a78.4 ± 9.1 ab51.6 ± 5.9 b25.5 ± 6.5 b
Q. ilex 14.4 ± 2.4 c115.2 ± 15.1 d11.2 ± 0.7 b5270 ± 122 c6.4 ± 1.1 c2.0 ± 0.1 b132.2 ± 22.3 c64.0 ± 9.0 ab33.1 ± 6.4 b

Trait correlations

Several significant correlations were observed between leaf functional and structural traits. Kleaf_area was inversely correlated to leaf vein density (Fig. 5a) and positively correlated to Aleaf (Fig. 5b); moreover, Kleaf_area was inversely correlated to the density of midrib xylem conduits as well as to the thickness of conduit walls, while it was positively correlated to the diameter of conduits (Table 3). Kleaf_mass was also inversely correlated to vein density (Fig. 5c) as well as to LMA (Fig. 5d), while it was positively correlated to DiC (Table 3).

Figure 5.

Relationships between leaf hydraulic conductance on a leaf area basis (Kleaf_area) and (a) vein density and (b) leaf surface area, as well as between leaf hydraulic conductance on a dry mass basis (Kleaf_mass) and (c) vein density and (d) leaf mass per unit area (LMA). Each point represents mean values for single species (Aps, Acer pseudoplatanus; Aca, A. campestre; Amo, A. monspessulanum; Qpe, Quercus petraea; Qpu, Q. pubescens; Qil, Q. ilex). The regression lines are reported together with r2 and P values (Pearson product moment correlation).

Table 3. Correlation matrix showing r and P values (Pearson product moment correlation) for correlations between leaf anatomical traits (midrib xylem conduit density (DeC) and diameter (DiC), wall thickness (TCW), vein density (DeV) and distance from veins to lower epidermis (DVE)), leaf hydraulic conductance on a leaf area (Kleaf_area) or dry mass (Kleaf_mass) basis, and leaf tolerance to hydraulic dysfunction (P50)
  De C Di C T CW De V D VE K leaf_area
  1. Significant relationships are highlighted in bold.

Di C =   −0.674      
=   0.142      
T CW =   0.839 =   −0.709     
=   0.037 =   0.115     
De V = 0.608 =   −0.795 = 0.598   
= 0.200 =   0.059 = 0.210   
D VE =   0.759 = 0.519 =   0.839 =   0.756   
=   0.080 = 0.292 =   0.037 =   0.082   
K leaf_area =   −0.642 =   0.874 =   −0.638 =   −0.866 = 0.496 
=   0.169 =   0.023 =   0.173 =   0.026 = 0.317 
K leaf_mass = 0.608 =   0.778 = 0.391 =   −0.877 =   −0.849 =   0.726
= 0.200 =   0.069 = 0.444 =   0.022 =   0.033 =   0.102
P 50 = 0.422 =   0.771 = 0.287 =   −0.685 =   −0.659 = 0.524
= 0.404 =   0.073 = 0.582 =   0.133 =   0.154 = 0.286

Values of Ψtlp were found to be positively correlated to P50 (Fig. 6a) but inversely correlated to Kleaf_mass (Fig. 6b), while Ψmin values were inversely correlated to Kleaf_area (Fig. 6c). In turn, P50 was negatively correlated to Kleaf_mass (Fig. 7a) and vein density (Table 3) but positively correlated to LMA (Fig. 7b) and DiC (Table 3).

Figure 6.

Relationships between: (a) leaf water potential at the turgor loss point (Ψtlp) and leaf water potential inducing 50% loss of leaf hydraulic conductance (P50); (b) Ψtlp and leaf hydraulic conductance on a dry mass basis (Kleaf_mass); (c) minimum seasonal water potential (Ψmin) and leaf hydraulic conductance on a leaf area basis (Kleaf_area). Each point represents mean values for single species (Aps, Acer pseudoplatanus; Aca, A. campestre; Amo, A. monspessulanum; Qpe, Quercus petraea; Qpu, Q. pubescens; Qil, Q. ilex). The regression lines are reported together with r2 and P values (Pearson product moment correlation).

Figure 7.

Relationships between leaf water potential inducing 50% loss of leaf hydraulic conductance (P50) and leaf hydraulic conductance on a dry mass basis (Kleaf_mass, (a) or leaf mass per area (LMA) (b). Each point represents mean values for single species (Aps, A. pseudoplatanus; Aca, Acer campestre; Amo, A. monspessulanum; Qpe, Quercus petraea; Qpu, Q. pubescens; Qil, Q. ilex). The regression lines are reported together with r2 and P values (Pearson product moment correlation).

Negative correlations were observed between xylem conduit diameter and density, thickness of conduit walls, as well as vein density (Table 3). LMA was moderately correlated to vein density (r = 0.69, P = 0.126) but not to leaf thickness, calculated as TPA + TSP (P = 0.470).

Discussion

Our data reveal novel correlations and trade-offs between leaf hydraulic efficiency and safety across closely related species growing under contrasting ecological conditions. In particular, experimental results suggest that leaf resistance to drought-induced hydraulic dysfunction is key to plant survival and competition even over limited geographical ranges. Moreover, our findings highlight that increased resistance to hydraulic failure is based on a suite of morpho-anatomical traits which imply substantial carbon investment in leaf construction and translate into higher costs of leaf water use (Simonin et al., 2012), and presumably reduced competitive ability in high-resource habitats.

Kleaf_area values recorded in Acer and Quercus species were consistent with those previously reported for species of the same genera (Sisó et al., 2001; Sack et al., 2003) and well within the range reported for other woody angiosperms (Sack & Holbrook, 2006). In accordance with previous studies (Gortan et al., 2009), we found Kleaf_area to be higher in plants growing in mesic habitats than in those from dry ones. High Kleaf_area is considered adaptive in humid habitats, where competition for light and nutrients is probably favored by high photosynthetic and growth rates, which are not limited by water availability. Because stomatal aperture and photosynthesis are constrained by Kleaf_area (Brodribb et al., 2005), high values of this parameter in mesophilous Acer and Quercus species probably allow these plants to successfully colonize areas with high water availability. The possible functional advantage of lower Kleaf_area in the xerophilous Acer and Quercus species is more difficult to capture. Sinclair et al. (2008) have suggested that low Kleaf_area in drought-resistant soybean (Glycine max) lines might increase their stomatal sensitivity to vapor pressure deficit (VPD), thus limiting transpiration under high evaporative demand and promoting a conservative use of soil water. However, it is not clear how conservative use of soil water in a natural environment could prevent out-competition by species adopting a ‘water-spending’ strategy. It might also be hypothesized that species growing in drought-prone sites, being exposed to low water potential values (see Fig. 2), experience stomatal closure induced by decreased turgor level. In this case, the functional advantage of high Kleaf_area would be overcome by stomatal response to cell water status. Indeed, a clear relationship appeared to exist between Ψmin values experienced by different species and their corresponding Kleaf_area (Fig. 6c). An alternative hypothesis is that low Kleaf in drought-adapted species is a consequence of morpho-anatomical modifications designed to increase tolerance to leaf hydraulic dysfunction, but previous studies have failed to identify clear trade-offs between leaf hydraulic safety and efficiency.

Plants growing in dry sites are exposed to low water potential, and leaves have been reported to be very vulnerable to drought-induced hydraulic dysfunction. We found the most negative P50 values in species growing in areas with low soil AWC (see 'Results'). This finding suggests that leaf hydraulic vulnerability can limit the distributional range of plants over much narrower scales than those previously reported (Blackman et al., 2012), thus contributing to shaping vegetation composition and landscape traits. In particular, our data suggest that local distribution of congeneric Acer and Quercus species as related to edaphic factors is at least partially driven by the differential tolerance of their leaf hydraulic system to low Ψ values.

According to previous findings, P50 was coordinated with other physiological traits conferring drought resistance, and specifically with Ψtlp (Fig. 6a) (Blackman et al., 2010; Scoffoni et al., 2012). In our as well as in previous studies, the relationship between the two parameters was not very different from 1 : 1. This observation, together with the well-known impact of reduced Kleaf_area on stomatal aperture (Trifilò et al., 2003), might suggest that coordination between P50 and Ψtlp is designed to induce stomatal closure in response to early cavitation events in the leaves (Salleo et al., 2000), thus reducing the transpiration rate before the occurrence of complete turgor loss. In this sense, leaves might be seen as ‘hydraulic fuses’ (Johnson et al., 2011) protecting living cells from approaching a critical water status. Alternatively, the relationship might suggest that turgor-induced changes in the hydraulic properties of the extra-vascular water pathway (Brodribb & Holbrook, 2006; Kim & Steudle, 2007) act together with leaf xylem embolism (Nardini et al., 2003; Johnson et al., 2012) in determining the overall drop of Kleaf_area during leaf dehydration.

Although no relationship was observed between leaf hydraulic vulnerability and Kleaf_area in the present or in previous studies (Blackman et al., 2010; Scoffoni et al., 2011, 2012), our data support the existence of a trade-off between Kleaf_mass and P50, suggesting that increased tolerance to hydraulic dysfunction comes at the cost of increased biomass investment in the leaf hydraulic system (Fig. 7). In turn, both Kleaf_mass and P50 were correlated to LMA, suggesting that the shift of both parameters toward lower values might be driven by similar morpho-anatomical changes. In particular, P50, Kleaf_mass and LMA were found to be correlated to vein density, while LMA was not correlated to leaf thickness. Data reported in Fig. 5a and Table 3 suggest that construction of a dense vein network increased carbon costs associated with water transport (lower Kleaf_mass) while improving leaf tolerance to low Ψleaf values. This view is supported by published analyses of relative densities of different leaf tissues. Castro-Díez et al. (2000) have reported a tight correlation between LMA and the proportion of minor vein tissue (xylem and sclerenchyma) in transversal leaf sections of 52 European woody species, while no relationship was observed between LMA and leaf thickness or the proportion of air spaces in the lamina. Poorter et al. (2009) further calculated the density of leaf vasculature (1.40 g cm−3) to be c. 5-fold higher than that of mesophyll cells (0.31 g cm−3). In agreement with the above findings, recent modeling efforts have shown that high vein density implies high carbon investment in venation and results in high LMA values (Blonder et al., 2011) These data indicate a possible role for increased vein density in driving the increase of LMA and decrease of Kleaf_mass recorded in the present study in drought-adapted Acer and Quercus species.

Increased leaf vein density has been suggested to confer tolerance to mechanical damage (Sack et al., 2008) and embolism-induced blockage of the vascular pathway (Scoffoni et al., 2011), by providing alternative water flow pathways in a redundant network (Corson, 2010). Our data are in accordance with the above findings and confirm that high vein density is a trait conferring drought resistance to leaves of woody plants growing in soils with poor water availability. The strong negative correlation (data not shown) between vein density and Aleaf recorded across our study species further suggests that reduction of leaf size favors the appearance of high vein density (Zwieniecki et al., 2004; Sack et al., 2012) and supports recently reported correlations between leaf size and hydraulic vulnerability (Scoffoni et al., 2011).

Leaf hydraulic vulnerability across Acer and Quercus species was also closely correlated to xylem conduit diameter, but not to the thickness of conduit walls. In turn, conduit diameter was inversely correlated to wall thickness, density of conduits in midrib xylem, and vein density. In other words, low P50 values were associated with smaller and more numerous conduits, as already reported for stem xylem in some studies (Maherali et al., 2006; Lens et al., 2011). A reduction of conduit diameter implies a reduction of water transport efficiency (Giordano et al., 1978). Such a reduction might have been partially counterbalanced by increased conduit and/or vein density. Overall, the observed changes in leaf vein architecture, together with high TCW of small conduits, might provide an explanation for the association between high LMA, low Kleaf_mass, and increased tolerance to leaf hydraulic dysfunction.

Xylem conduit diameter apparently constrained Kleaf_area as well (Table 3), and the negative correlation between vein density and xylem conduit diameter is probably the basis for the recorded negative correlation between Kleaf_area and vein density (Fig. 5a). This finding is unexpected, given that previous studies have highlighted the major contribution of high vein density to the progressive increase of Kleaf_area during angiosperm evolution (Brodribb & Feild, 2010; Brodribb et al., 2010). However, we note that previously reported positive correlations between Kleaf_area and vein architecture were observed in a range of vein densities from 1 to 25 mm mm−2. In the relatively narrow range of vein densities recorded in our study species, and more typically found in eudicots (5–15 mm mm−2; Boyce et al., 2009), previous studies (Scoffoni et al., 2011) have failed to identify clear relationships between this anatomical trait and leaf hydraulic capacity, suggesting that in some cases other traits such as conduit diameter or the efficiency of the extra-vascular water pathway might represent more important constraints on leaf hydraulics than vein density. Similar considerations apply to the lack of correlation between Kleaf_area and the distance vein–epidermis, which apparently contrasts with previous findings (Brodribb et al., 2010). Again, this apparent inconsistency probably derives from the fact that in our study species DVE ranged between 16 and 36 μm, while previous analyses had much larger DVE intervals, from a few tenths of a micrometer up to 1 mm (Brodribb et al., 2007, 2010).

In conclusion, our data highlight the role played by leaf hydraulic efficiency and safety in constraining plant species’ distribution. These functional traits are likely to play an important role in shaping vegetation composition and distribution at different spatial scales. Increased hydraulic safety of leaves is key to a plant's adaptation to drought-prone sites, but comes at the cost of increased carbon investment in leaf construction, probably as a consequence of smaller and denser conduits and veins, as well as increased conduit wall thickness. These morpho-anatomical changes translate into increased LMA and, most importantly, decreased Kleaf_mass, which probably makes drought-resistant species less competitive in high-resource habitats, from which they are excluded. The trade-off between leaf hydraulic safety and efficiency, when expressed on a dry mass basis, calls for future studies at larger scales, addressed at understanding biophysical and metabolic challenges to leaf construction in xerophilous plants, and related benefits in terms of increased drought resistance.

Acknowledgements

This study was funded by the University of Trieste (Finanziamenti per la Ricerca di Ateneo, 2009). We are grateful to M. Codogno for useful information about the ecology of the species studied, and to T. J. Brodribb for discussion with G.P. on preliminary data. We would like to thank three anonymous reviewers for constructive comments that helped improve the manuscript.

Ancillary