Introduction

One central theme in plant ecology is the trade-offs between the efficiency of water transport and the resistance of the vascular system to embolisms and, at the leaf scale, the mechanism of the trade-off between efficiency and tolerance is widely reflected in the so-called leaf economics spectrum (Wright et al. 2004; Onoda et al. 2017; Xiong and Flexas 2018). Within the efficiency-tolerance context, fast-growing species generally combine low structural investment with high assimilation and low-stress tolerance; in contrast, the slow-growing species associated with high structure investment, low assimilation rate, and stress-tolerant leaves (Wright et al. 2004). Besides this, over the past decades, many studies have investigated another aspect of the general trade-off, i.e. the trade-offs between the efficiency of water transport and the tolerance of the vascular system from embolisms in both stem and leaf (Blackman et al. 2010; Nardini et al. 2012; Ocheltree et al. 2016). In these studies, the tolerance of the vascular system to embolism was generally quantified in terms of Px, that is the water potential values inducing x% loss of hydraulic conductance (a list of traits, symbols, and units, see Table 1).

Table 1 List of leaf traits symbols and units
Full size table

The efficiency-tolerance tradeoff hypothesis in woody stems was supported by some studies, but not others (reviewed by Gleason et al. 2016). Although there are a few studies that investigated the trade-off between efficiency and tolerance at the leaf scale, these studies also produced mixed results (Blackman et al. 2010; Nardini et al. 2012; Ocheltree et al. 2016). For instance, Blackman et al. (2010) investigated maximum leaf hydraulic conductance (Kmax) and P50 of 20 woody angiosperm species and found no significant correlation between Kmax and P50. In contrast, a significant tradeoff between Kmax and P50 was observed across six woody species (Nardini et al. 2012) as well as across nine grass species (Ocheltree et al. 2016). The reason for these discrepancies is unclear, although a recent study suggested that the efficiency-tolerance traits tradeoff is influenced by seasonal drought (Liu et al. 2021).

Due to the important role of water transport capacity within the soil–plant-atmosphere continuum in determining CO2 diffusion conductance from leaf surface to chloroplast (Brodribb et al. 2007; Xiong et al. 2017), the light-saturated photosynthetic rate (A), a key determinant of growth capacity, is assumed to be represented by Kmax. This means that if an efficiency-tolerance tradeoff exists in the water transport system of plants, it should translate to a trade-off between hydraulic safety and A. However, a recent study showed that the correlation between Kmax and A does not exist across nine grass species, which indicated that hydraulic efficiency may decouple from carbon gain efficiency (Ocheltree et al. 2016). Indeed, the hydraulic resistance within leaves represents only a part of the hydraulic resistance of the soil–plant–atmosphere continuum (Sack et al. 2003; Sack and Holbrook 2006). Therefore, even if an efficiency vs tolerance trade-off in the hydraulic pathway exists at the organ level, it may not necessarily translate into a trade-off in carbon assimilation efficiency vs drought tolerance.

Beyond hydraulic vulnerability parameters, the leaf water potential at turgor loss point (πtlp), i.e. the leaf water potential (ψleaf) at which the leaf cell turgor pressure is zero, was suggested to be strongly related to ecological drought tolerance and species distributions relative to water supply within and across biomes (Tyree and Jarvis 1982; Bartlett et al. 2012). Practically, the πtlp, together with other parameters, including the bulk modulus of elasticity (Ɛ) and osmotic potential at full hydration (π0), are typically calculated from a plot of ψleaf against water volume in drying leaves, known as the pressure–volume (PV) curve. Previous studies have suggested that plants with low πtlp tend to maintain high stomatal conductance, hydraulic conductance, photosynthetic gas exchange, and growth under low soil water availability conditions (Farrell et al. 2017; Trueba et al. 2019; Sorek et al. 2021). Importantly, while Px reflects the tolerance water transport system of entire leaves, PV curve parameters mainly reflect the ability of mesophyll tissues, where the carbon assimilation occurs, to tolerate drought (Tyree and Jarvis 1982; Bartlett et al. 2012). Therefore, if the trade-off in carbon assimilation efficiency vs drought tolerance exists, the gas exchange should tightly correlate with PV curve parameters. Yet, the trade-off in carbon assimilation efficiency vs cellular drought tolerance has not been integrated and directly compared experimentally.

According to the leaf economics spectrum theory, both efficiency and tolerance are related to the investment in leaf structural components. On the one hand, numerous studies have investigated the influences of anatomical traits on Kleaf and photosynthetic traits across species and/or environmental conditions (Niinemets et al. 2002; Buckley et al. 2015; Buckley 2015; Caringella et al. 2015; Tosens et al. 2016; Xiong et al. 2017). It has been suggested that Kmax is related to vein traits, including vein length per leaf area (VLA), free vein endings (FVEs), and xylem conduit numbers and diameters in both major and minor veins. In addition to leaf veins, it is now recognized that the Kmax is also highly related to the mesophyll anatomy (Rockwell et al. 2014; Buckley et al. 2015, 2017; Buckley 2015; Xiong et al. 2017), and in fact, a leaf anatomical based outside-xylem water flow model, MOFLO, has been developed by Buckley et al. (2015). On the other hand, the influences of vein traits and mesophyll anatomical traits on drought tolerance were also suggested in many studies (Blackman et al. 2010; Scoffoni et al. 2011, 2016a, 2017; Blonder et al. 2012; Binks et al. 2016). For instance, Scoffoni et al. (2016a) showed that drought-induced decline in hydraulic conductance inside leaf veins was tightly correlated with conduit numbers in the midrib but independent of conduit number and size in minor veins. Mesophyll anatomical traits change dramatically due to the dehydration introduced leaf shrinkage and thus contributed to Kox vulnerability (Blonder et al. 2012; Scoffoni et al. 2014, 2017). However, the role of the leaf anatomy in the tradeoff between efficiency and safety remains unclear.

In this study, we investigated the hydraulic efficiency and tolerance traits, gas exchange, PV traits, and leaf anatomy traits in seven species phylogenetically distant, also varying in leaf habit and drought tolerance. The aim of this study was to assess the following: (1) whether the carbon assimilation efficiency vs drought tolerance trade-off at the leaf level exists across species; and (2) what are the most general physiological and structural bases of carbon assimilation and hydraulic efficiencies and dehydration tolerance traits on leaf level.

Materials and methods

To test the tradeoffs of carbon assimilation and hydraulic efficiencies and drought tolerance/avoidance traits across different species, seven species of diverse phylogeny, origin, life form, and, particularly, their drought tolerance based on literature investigation (details see Table 2), growing in a common garden at the campus of the University of Illes Balears, Palma de Mallorca (Illes Balears, Spain) were used for the study. The species Nerium oleander L., Platanus occidentalis L., and Glycine max. L. were measured in July 2015, and the species of Populus nigra L., Ceratonia siliqua L., Ginkgo biloba L., and Gossypium hirsutum L. were measured in July 2016. In the current study, measurements were conducted to 50–60-day old G. max and 70–80-day old G. hirsutum plants. For other species, mature plants were randomly selected. There were no extreme climatic events during and before the measurements conducted and meteorological data of the site can be found at http://plantmed.uib.es/Ingles/INTRANET.html.

Table 2 The variation of leaf mass per area (LMA), leaf thickness (Tleaf), leaf vein density (VLA), and free ending vein number per leaf area (FEV) in seven species
Full size table

Gas exchange

An open-flow gas exchange system (LI-6400XT, LI-COR, Lincoln, NE, USA) with an integrated fluorescence leaf chamber (LI-6400-40, LI-COR) was used to measure leaf gas exchange and chlorophyll fluorescence simultaneously. The sun exposed leaves on the east side of canopies were selected for each species, and the CO2 response curves were measured in three to six individuals in situ. The day before the gas exchange measurement conducted, the plants were fully irrigated. The gas exchange measurement to trees was a little challenged, and a stepladder was used to raise the gas exchange system if necessary. The light intensity, block temperature, and flow rate inside the chamber were set to 1500 μmol m−2 s−1 (10% blue light), 25 ℃, and 300 μmol s−1, respectively. The average vapor pressure deficit (VPD) inside the chamber was 1.73 ± 0.4 kPa. After the leaf reached a steady state (the fluctuation of stomatal conductance gs being less than 0.05 mol m−2 s−1 during a 10-min period), the auto-progress of the CO2 response curve was adopted. The reference CO2 concentrations were subsequently set at 400, 300, 200, 100, 50, 400, 600, 800, 1000, 1200, 1500, 2000, 400 μmol CO2 mol−1 air. The CO2 response curve measurements were performed between 8:30 am and 11:30 am each day.

As described in our previous study (Xiong and Flexas 2021), the mesophyll conductance (gm) was calculated based on Fick's first law of diffusion:

$$g_{{\text{m}}} = \frac{A}{{C_{{\text{i}}} - C_{{\text{c}}} }},$$
(1)

where the net photosynthetic rate (A), and the intercellular CO2 concentration (Ci) were taken from the gas exchange measurements. And the chloroplast CO2 concentration (Cc) was estimated based on the Harley et al. (1992) method:

$$C_{{\text{c}}} = \frac{{\Gamma *\left( {J + 8\left( {A + R_{{\text{d}}} } \right)} \right)}}{{J - 4\left( {A + R_{{\text{d}}} } \right)}}$$
(2)

The electron transport rate (J) was directly taken from the Li-COR 6400 measurement. The Γ* represents the CO2 compensation point in the absence of respiration (taken as 40 μmol mol−1 in this study), and the Rd represents the day respiration rate, which was fitted from the CO2 response curve. For each data point generated, we checked whether it met the reliability criterion (10 > dCc/dA > 50), as suggested by Harley et al. (1992).

The relative photosynthetic limitations of stomatal conductance (ls), mesophyll conductance (ls), and photosynthetic biochemistry (lb) were calculated according to Grassi and Magnani (2005).

$$l_{{\text{s}}} = \frac{{{{g_{{\text{t}}} } \mathord{\left/ {\vphantom {{g_{{\text{t}}} } {g_{{\text{s}}} \cdot {{\partial A} \mathord{\left/ {\vphantom {{\partial A} {\partial C_{{\text{c}}} }}} \right. \kern-\nulldelimiterspace} {\partial C_{{\text{c}}} }}}}} \right. \kern-\nulldelimiterspace} {g_{{\text{s}}} \cdot {{\partial A} \mathord{\left/ {\vphantom {{\partial A} {\partial C_{{\text{c}}} }}} \right. \kern-\nulldelimiterspace} {\partial C_{{\text{c}}} }}}}}}{{{{g_{{\text{t}}} + \partial A} \mathord{\left/ {\vphantom {{g_{{\text{t}}} + \partial A} {\partial C_{{\text{c}}} }}} \right. \kern-\nulldelimiterspace} {\partial C_{{\text{c}}} }}}}$$
(3)
$$l_{{\text{m}}} = \frac{{{{g_{{\text{t}}} } \mathord{\left/ {\vphantom {{g_{{\text{t}}} } {g_{{\text{m}}} \cdot {{\partial A} \mathord{\left/ {\vphantom {{\partial A} {\partial C_{{\text{c}}} }}} \right. \kern-\nulldelimiterspace} {\partial C_{{\text{c}}} }}}}} \right. \kern-\nulldelimiterspace} {g_{{\text{m}}} \cdot {{\partial A} \mathord{\left/ {\vphantom {{\partial A} {\partial C_{{\text{c}}} }}} \right. \kern-\nulldelimiterspace} {\partial C_{{\text{c}}} }}}}}}{{{{g_{{\text{t}}} + \partial A} \mathord{\left/ {\vphantom {{g_{{\text{t}}} + \partial A} {\partial C_{{\text{c}}} }}} \right. \kern-\nulldelimiterspace} {\partial C_{{\text{c}}} }}}}$$
(4)
$$l_{{\text{b}}} = \frac{{g_{{\text{t}}} }}{{{{g_{{\text{t}}} + \partial A} \mathord{\left/ {\vphantom {{g_{{\text{t}}} + \partial A} {\partial C_{{\text{c}}} }}} \right. \kern-\nulldelimiterspace} {\partial C_{{\text{c}}} }}}},$$
(5)

where the gt represents the total CO2 diffusion conductance (1/gt = 1/gs + 1/gm).

In addition, the maximum carboxylation rate (Vcmax), the maximum electron transport rate (Jmax), and the gm independent of chlorophyll fluorescence (gmE) were fitted from CO2 response curves (Ethier and Livingston 2004).

Leaf hydraulic vulnerability and pressure–volume curve

Kleaf vulnerability curves were measured using the evaporative flux method (Scoffoni et al. 2011). Mature and sun-exposed branches were collected from five to forty-five individuals per species in field conditions, except for G. max, and G. hirsutum, for which 30 to 40 entire plants (keeping the root system intact as possible) with at least three new fully expanded leaves were collected. Then, the branches or entire plants were covered with two layers of black plastic bags and rehydrated overnight. Branches were cut into segments with at least three leaves under deionized water and then bench dried to create a range of leaf water potentials for vulnerability curves measurement. Dehydrated branches were placed into a sealable bag for leaf water potential. Samples were allowed to equilibrate for at least 15 min before two leaves were excised and measured for initial water potential (ψinitial, the average water potential of two leaves) using a pressure chamber (PMS 1505D, PMS Instrument Co., Albany, OR, USA). If the difference in the leaf water potential of those two leaves was greater than 0.25 MPa, the shoot was discarded. The third leaf (typically the middle one) measured Kleaf.

The details about Kleaf measurements were described in our previous studies (Xiong et al. 2017, 2018; Wang et al. 2022). In brief, leaves were cut from the branches with a fresh razor blade under ultrapure water. Then the petiole was rapidly connected to silicone tubing underwater to prevent air from entering the system. The tubing system was connected to a plastic Erlenmeyer flask (250 ml) with degassed pure water on an analytical balance (ABT 320-4 M, KERN, Balingen, Germany). The leaks of the tubing system were carefully checked before the measurement was conducted (see details in Xiong et al. 2018). The light intensities (light source: APO4, Eiviled 2010, Illes Balears, Spain) on the leaf surface were 1500 ± 65 μmol m−2 s−1, as measured with a quantum sensor (Li-190R, LI-COR, Lincoln, NE, USA). Leaf temperature was controlled between 23 and 27 °C by adjusting the room air temperature using an air conditioner.

The weight of water loss was recorded every 60 s, and the transpiration flux rate was calculated as the slope of the linear regression between weight and time for every 6 minutes. For each leaf, the duration of the steady transpiration was not short than 15 min. After the leaves were removed from the tubing system and kept in darkness (in Ziploc Smart Zip Freezer Bags) to equilibrate for at least 20 min and then, the final leaf water potential (ψfinal) was measured with a pressure chamber (Model 1505, PMS Instrument Company, Albany, USA). The leaf areas of the leaves were measured from digitalized pictures using Image J (https://imagej.nih.gov/ij/), and water flux rate (E) of each leaf was normalized by its leaf area Kleaf was calculated as follows:

$$K_{{{\text{leaf}}}} = \frac{E}{{\Psi_{{{\text{water}}}} - \Psi_{{{\text{final}}}} }}.$$
(6)

where the Ψwater is the water potential of distilled water (= 0 MPa).

As discussed in several previous studies (Scoffoni et al. 2012; Wang et al. 2018, 2022), dehydrated leaves may recover in leaf water potential before reaching steady-state transpiration during the evaporative flux method (EFM) measuring due to the stomata opening and the ψfinal is less negative than ψinitial. Alternatively, the transpiration rate may be sufficient for ψfinal to be driven lower than ψinitial. According to the suggestion of those studies we constructed vulnerability curves by plotting Kleaf against whichever was the lowest, ψfinal or ψinitial (‘ψlowest’), that is, the leaf water potential associated with the strongest dehydration experienced during the experiment, and each leaf was considered as a data point. In the current study, the maximum leaf hydraulic conductance (Kmax) of each species was calculated as the average Kleaf for the points above − 0.5 MPa of the vulnerability curves. Five pressure–volume curves per species were conducted to estimate osmotic potential at full turgor (π0; MPa) and at turgor loss point (πtlp; MPa), and modulus of elasticity (ɛ; MPa) as in Sack and Pasquet-Kok (2011).

Leaf vein density

Leaves were chemically cleared in 15% NaOH (w/v) and then bleached following our previous standard protocol (Xiong et al. 2018), and then stained with safranin and fast green in ethyl alcohol. After washing redundant safranin and fast green in the water, leaves were scanned to measure major vein length, and the minor vein densities were measured from pictures captured using a light microscope (U-TVO.5XC; Olympus, Tokyo, Japan). Leaf area and vein length were manually measured by using ImageJ (more details see our previous studies, Xiong et al. 2018). In this study, the major vein is defined as the sum the 1°, 2° and 3° veins and the veins of any order higher than 3° were considered minor veins. In this case, G. biloba only has the major veins (Fig. S1).

Light microscopy analysis

Small leaf discs of about 4.0 × 1.2 mm were also cut from the middle of new fully expanded leaves after the gas exchange measurement. In a vacuum chamber, the leaf discs were infiltrated with fixative 2.5% glutaric aldehyde in 0.1 M phosphate buffer (pH = 7.6) at 4 ℃, and post-fixed in 2% buffered osmium tetroxide at 20 ℃ for two h. The samples were embedded in Spurr’s epoxy resin. For light microscopy, semithin leaf cross-sections were cut using a fully automated rotary microtome (Leica RM2265, Leica Microsystems, Milton Keynes, UK). The leaf sections were stained with 1% (w/v) toluidine blue in 1% (w/v) Na2B4O7, and they were examined at 40 × and 100 × magnification with an Olympus IX71 light microscope (Olympus Optical, Tokyo, Japan). As for the transmission electron microscope (TEM), H-7650 (Hitachi—Science & Technology, Tokyo, Japan) was used for observation and photography. Three leaves per species were analyzed. The total cross-sectional area of mesophyll tissues (Smes) and intercellular air space area (Sias), the total length of the mesophyll cell wall exposed to intercellular air space (lmes), the total length of chloroplasts touching the plasma membrane appressed to intercellular air space (lc), cell wall thickness (Tcw), and the width of the analyzed leaf cross-section (L) were measured using Image J software (National Institute of Health, Bethesda, MD, USA). The volume fraction of intercellular air space (fias) was calculated as follows:

$$f_{{{\text{ias}}}} = \frac{{S_{{{\text{ias}}}} }}{{S_{{{\text{mes}}}} }}$$
(7)

Sm and Sc were then calculated as follows:

$$S = \frac{l}{L} \times F,$$
(8)

where S is Sm or Sc, l is lmes or lc, and the F is the curvature correction factor, which was measured and calculated for each species according to Tomás et al. (2013) for palisade and spongy cells by measuring their width and height and calculating an average width/height ratio. The curvature factor correction ranged from 1.09 to 1.54 for spongy cells and 1.38 to 1.61 for palisade cells.

Statistical analysis

One-way ANOVA analysis was used to test the differences in measured traits among species. Regression analyses were performed with mean values to test the correlations between parameters, and the PCA analysis was performed using FactoMineR (Lê et al. 2008). To identify the ‘best’ vulnerability curve for each species, four different mathematical functions fit the Ψlowest and Kleaf data using orthogonal nonlinear least-squares regression approach (Wang et al. 2018). Linear \(\left( {K_{{{\text{leaf}}}} = {\text{ a }} \cdot \, \Psi_{{{\text{lowest}}}} + {\text{ b}}} \right)\), Sigmoidal \(\left( {K_{{{\text{leaf}}}} = \frac{a}{{1 + e^{{ - \left( {\frac{{\Psi_{{{\text{lowest}}}} - c}}{b}} \right)}} }}} \right)\), exponential \(\left( {K_{{{\text{leaf}}}} = \frac{a}{{1 + \left( {\frac{{\Psi_{{{\text{lowest}}}} }}{c}} \right)^{b} }}} \right)\), and logistic \(\left( {K_{{{\text{leaf}}}} = c + ae^{{ - b\Psi_{{{\text{lowest}}}} }} } \right)\) curves were fitted to these data. The curve with the lowest Akaike Information Criteria (AIC) and/or Bayesian information criterion (BIC) values were selected as the ‘best’ fit (Table S1). All analyses were performed in R version 3.5.1 (R Core Team 2018).

Results

Traits diversity

Across the 7-species, there were substantial differences in leaf mass per area (LMA), leaf thickness (Tleaf), and vein length per area (VLA; Table 2 & Fig.S1). The LMA varied from 32.7 in G. max to 151.5 g m−2 in N. oleander; Tleaf from 0.36 in G. biloba to 1.63 mm in N. oleander; VLA from 1.8 in G. biloba to 15.7 mm mm−2 in C. siliqua; and FEV from 0 in G. biloba to 8.8 mm−2 in G. hirsutum. Across the estimated species, the major and minor vein density also showed large variation. Species varied by 4.4-fold in major vein length and 4.1-fold in minor vein length (Table S2). Like the leaf veins, the mesophyll structures also showed large differences across species (Fig. S1; Table S2). In P. nigra, 40.9% of the cross-section was occupied by intercellular air space (fIAS); however, the fIAS in N. oleander was only 15.2% (Table S2). There was also a significant variation in the Sm (a minimum of 19.7 m2 m−2 in N. oleander and a maximum of 31.1 m2 m−2 in G. hirsutum), Sc (a minimum of 10.2 m2 m−2 in G. biloba and maximum of 28.8 m2 m−2 in G. max) and Tcw (a minimum of 0.197 µm in N. oleander and maximum of 0.431 µm in G. biloba).

Across all selected species, the saturated photosynthetic rate (A) varied from 7.3 µmol m−2 s−1 in G. biloba to 27.6 µmol m−2 s−1 in G. hirsutum (Table 3). Stomatal conductance to CO2 (gs) ranged 3.6-fold, and mesophyll conductance to CO2 (gm) ranged 3.7-fold. The largest gs and gm were observed in G. hirsutum, and the lowest in G. biloba. In addition, both the maximum carboxylation rate of rubisco (Vcmax) and maximum electron transportation rate (Jmax) varied greatly among species (Table 3).

Table 3 Mean values for the photosynthetic parameters at 25 ℃
Full size table

Leaf hydraulic vulnerability and pressure–volume curves

The shape of vulnerability curves for Kleaf differed substantially across species (Fig. S2, Table S1). Species varied significantly in aspects of leaf hydraulic vulnerability curve and pressure–volume curve parameters (Table 4). The maximum leaf hydraulic conductance (Kmax) varied from 5.05 to 13.5 mmol m−2 s−1 MPa−1 (Table 4). The P50 varied from − 0.72 to − 2.24 MPa in G. hirsutum and N. oleander, respectively; and in P80 from − 1.22 to − 3.97 MPa in the same two species. However, the 0.05 level of confidence intervals for both P50 and P80 were quite large. There was also a significant variation in the π0 (the highest being − 1.05 MPa in G. max and the lowest − 2.38 MPa in N. oleander), πtlp (the highest being − 1.41 MPa in G. max and the lowest -3.32 MPa in N. oleander) and Ɛ (a minimum of 8.9 MPa in N. oleander and maximum of 16.0 MPa in P. nigra).

Table 4 Maximum leaf hydraulic conductance (Kmax), leaf water potential at 50% and 80% decline of leaf hydraulic conductance (P50 and P80), osmotic potential at full turgor (π0) and at turgor loss point (πtlp; MPa), and modulus of elasticity (ɛ)
Full size table

Trait correlations

We first analyzed the correlations among tolerance traits and found that Kleaf vulnerability traits (P50 and P80) were weakly correlated with pressure volume traits across selected species (Fig. 1). Then, the correlations among efficiency traits were analyzed, and it was found that A was strongly correlated with both gs (r2 = 0.88; p = 0.002; Fig. 2d), gm (r2 = 0.86; P = 0.002; Fig. 2e) and Vcmax (r2 = 0.96; p < 0.001; Fig. 2f). However, Kmax was decoupled from gas exchange traits except for gm (r2 = 0.73; p = 0.014; Fig. 2c) across species. Here we also estimated the influences of leaf anatomy on gm and Kmax, the results showed that Kmax was decoupled from VLA, VLAmajor and VLAminor across species; however, gm was strongly affected by Sc and Tcw (Fig. 3). The correlation analysis between efficiency and tolerance traits showed no correlations between A and P50 (r2 = 0.2; p = 0.315) or πtlp (r2 = 0.31; p = 0.199) across species (Fig. 4). Similarly, no significant correlations were found between Kmax and P50 (r2 = 0.06; P = 0.61) as well as πtlp (r2 = 0.40; p = 0.144). However, we found that P50 positively correlated with VLAmajor (r2 = 0.63; p = 0.03), but not with VLA (r2 = 0.074; p = 0.554) and VLAminor (r2 = 0.002; p = 0.934) across species (Fig. 5a–c). Moreover, cell wall thickness (Tcw) had significant impacts on pressure–volume parameters (Fig. 5d, e, f). The π0 (r2 = 0.84; P = 0.004), πtlp (r2 = 0.82; P = 0.005) and ɛ (r2 = 0.62; p = 0.037) significantly correlated with Tcw.

Fig. 1
figure 1

Correlations among pressure–volume curve traits and leaf hydraulic vulnerability curve traits. Lines were fitted using a linear model, and the shade areas around lines indicate the 95% of confidence intervals. π0, osmotic potential at full rehydration; πtlp; turgor loss point; ɛ, modulus of elasticity; P50, leaf water potential at a 50% of maximum leaf hydraulic conductance lost; and P80, leaf water potential at an 80% of maximum leaf hydraulic conductance lost. The full names for each of the species are provided in Table 2

Full size image
Fig. 2
figure 2

Correlations among efficiency traits. The correlation between traits was fitted using the linear model, and the shade areas around lines indicate the 95% of confidence intervals. A, light-saturated photosynthetic rate; Kmax, maximum leaf hydraulic conductance; gs, stomatal conductance; gm, mesophyll conductance; and Vcamx, maximum carboxylation efficiency

Full size image
Fig. 3
figure 3

Impacts of leaf anatomical traits on mesophyll conductance (gm) and maximum leaf hydraulic conductance (Kmax). VLA. Leaf vein density; VLAmajor, major vein density; VLAminor, minor vein density; fIAS, fraction of leaf mesophyll volume occupied by intercellular air space; Sc, total chloroplast surface area exposed to intercellular air space per unit of leaf surface area; and Tcw, cell wall thickness

Full size image
Fig. 4
figure 4

Efficiency vs safety across species. a, b Correlations of leaf water potential at a 50% of maximum leaf hydraulic conductance lost (P50) to light-saturated photosynthetic rate (A) and maximum leaf hydraulic conductance (Kmax); and c, d correlations of osmotic potential at the turgor point loss point (πtlp) to A and Kmax

Full size image
Fig. 5
figure 5

Influences of leaf a vein density (VLA), b major vein density (VLAmajor), and c minor vein density (VLAminor) on the leaf water potential at a 50% of maximum leaf hydraulic conductance lost (P50); and the influences of cell wall thickness (Tcw) on d osmotic potential at full rehydration (π0); e turgor loss point (πtlp) and f modulus of elasticity (ɛ)

Full size image

The principal component (PCA) analysis results are shown in Fig. 6 and Table S3. The first PCA axis accounted for 52% of the total variation and showed strong loadings on LMA, VLAmajor, fIAS, gas exchange, leaf hydraulic vulnerability, and pressure–volume traits. The second axis, which accounted for 24.5% of the total variation, had strong loadings on Kmax, VLA, VLAminor and mesophyll structural traits (Fig. 6a). PCA ordination using all species demonstrated variable combinations of leaf hydraulic, gas exchange traits, leaf anatomy, leaf hydraulic vulnerability, and pressure–volume traits (Fig. 6b).

Fig. 6
figure 6

Principal component analysis (PCA) of gas exchange, leaf hydraulic, hydraulic vulnerability, press-volume, and leaf anatomical traits. The symbols used for each trait represented in the axis titles are provided in Table 1 and the full names for each of the species are provided in Table 2

Full size image

Discussion

Hydraulic vulnerability traits and pressure–volume traits

Across species, the parameters (P50 and P80) of Kleaf vulnerability curves were correlated strongly with the parameters (π0, πtlp, ɛ) of pressure–volume curves, as observed in previous studies (Blackman et al. 2010; Scoffoni et al. 2012; Powell et al. 2017; Figs. 1, 6a; Fig. S6). Our results together with earlier studies suggest that Kleaf vulnerability traits and pressure–volume traits are co-selected in species with large drought tolerance. The embolization of the xylem conduits has been widely believed to be the major cause of hydraulic decline during drought, and Kleaf vulnerability curve was widely used in evaluating the capacity of leaves to avoid xylem embolism in the past (reviewed by Scoffoni et al., 2012). The tight correlations between Kleaf vulnerability traits and pressure–volume traits were hypothesized to arise because cells maintaining turgor at more negative water potentials could preserve cell integrity and, therefore, confer resistance to Kleaf decline (Blackman et al. 2010; Scoffoni et al. 2016a). However, several recent studies highlighted that the decline of Kleaf during dehydration is caused by the decrease of outside xylem hydraulic conductance (Kox) rather than the hydraulic conductance inside xylem (Kx) (Trifiló et al. 2016; Scoffoni et al. 2017). If this were the case, then the Kleaf vulnerability parameters would mainly reflect the drought tolerance of mesophyll tissues as pressure–volume parameters do. Unfortunately, we cannot test this question as we did not separate the contributions of Kx and Kox on Kleaf vulnerability in this study.

Hydraulic conductance and gas exchange efficiency

Although Kmax and gas exchange traits varied greatly among species with different growth habits, no significant correlations between Kmax and gas exchange traits except gm were observed in the current study (Fig. 2). This result consisted with a previous one on nine C4 grasses (Ocheltree et al. 2016); however, it was opposite to other studies which observed strong correlations between Kmax and gas exchange traits explained as coordinated evolution of these processes (Brodribb et al. 2007; Scoffoni et al. 2016b). Lacking a correlation between Kmax and A in the present study unlikely results from the smaller set of species because the variability in both A and Kmax among our species was similar to that in the previous studies (Brodribb et al. 2007; Ocheltree et al. 2016; Scoffoni et al. 2016b).

The coordination theory of Kleaf and gas exchange is based on the following two assumptions: (i) leaf is the bottleneck of plant hydraulics system, and (ii) stomatal resistance is the major limiting factor to A (Brodribb et al. 2007; Scoffoni et al. 2016b). During photosynthesis, the ability to keep the stomata opening to capture CO2 depends on the plant’s capacity to replace that water lost through stomata. Hence, the whole plant hydraulic conductance should match gs to maximize photosynthesis. By performing a literature synthesis analysis, Sack and Holbrook (2006) summarized that leaves contributed on average to about 30% of whole plant hydraulic resistance and suggested that leaf is the bottleneck of plant hydraulic system, although with large variations across species and environmental conditions (Sack and Holbrook 2006). Conversely, many direct investigations suggested that the root is the major bottleneck in the plant hydraulics (Steudle 2000; Domec et al. 2009), also varying greatly across species and growth conditions. For instance, due to the existence of strong aerenchyma and apoplastic barriers in roots, the radial, as well as the axial water transport capacities in roots, were strongly limited in wet habit species (Steudle 2000; Ranathunge et al. 2005; Kotula et al. 2009). Therefore, the variable contributions of leaves and roots to plant hydraulic resistance between species may be one of the reasons for the decoupled correlation between Kmax and gas exchange traits. Further studies are needed to check the contributions of roots and leaves on plants hydraulic systems across species with variable growth conditions. In the current study, a tight correlation between A and gs indicated that gs could be one of the major photosynthetic limiting factors. However, the limitation analysis showed that the relative contributions of gs, gm, and biochemistry factors to A varied widely across species, and the A of the selected species was largely constricted by biochemical factors despite gs and gm contributing nearly half of the limitation to A in some of species (Fig. S3). While in gymnosperms (represented here by Ginkgo), despite gs and gm are generally far larger photosynthetic limitations than biochemistry, and in angiosperms, all three limitations are similar on average (Gago et al. 2019), biochemical limitations tend to be larger in crops (Nadal and Flexas 2018, 2019). In the present study, we include two true crops, cotton and soybean) plus four fast-growing woody species often used as crops (Populus, Ceratonia, Nerium, and Platanus), which could be the predominance of biochemical limitations in this survey. In fact, the other species (Ginkgo) present the lowest biochemical limitations among the seven studied species. Regardless of the reason, the reality is that in the species selected, the stomatal limitation is not the largest limitation to photosynthesis, being thus another possible explanation for the coordination between Kleaf and gas exchange.

Safety vs efficiency tradeoff

In the current study, we investigated the leaf level safety and efficiency tradeoffs by measuring leaf hydraulic vulnerabilities, pressure–volume curves, carbon assimilation efficiency and water transport efficiency. No significant tradeoffs between hydraulic safety and efficiency traits on leaf level were observed in our study (Fig. 4), and this result contrasts with several previous studies (Nardini et al. 2012; Ocheltree et al. 2016). Some species seem to have low carbon assimilation efficiency and low drought tolerance, which cannot be considered a strict tradeoff. However, it is true that we did not find any species with large efficiency and large tolerance, for which some tradeoff-like mechanisms may still operate at that level. Furthermore, the principal component analysis confirmed that efficiency traits were largely independent of safety traits (Fig. 6). Further analysis showed that carbon assimilation efficiency traits (gas exchange trait) are related to Sm, Sc, FVE, and VLAminor, while drought tolerance traits are more related to VLAmajor, Tcw, LMA, and thickness of up and low epidermis (Fig. 6); the lack of tradeoffs between efficiency and tolerance traits across species may be thus explained by different leaf anatomical features setting ones and others.

The correlation between Kleaf and VLA was often estimated in previous studies, and a higher VLA is predicted to have both higher Kx and Kox: the former by providing more parallel flow paths through the vein system, and the latter by decreasing horizontal path length for water transport from the veins to the sites of evaporation (Brodribb et al. 2007; Buckley et al. 2015). In the current study, Kmax was independent of VLA, VLAmajor, and VLAminor, which was in contrast with some of previous studies (Sack and Frole 2006; Brodribb et al. 2007), but consistent with others (Nardini et al. 2014; Caringella et al. 2015; Xiong et al. 2015). In fact, beyond VLA, Kleaf is also influenced by many other anatomical traits, such as leaf thickness, mesophyll tissue thickness, the size of bundle sheath cells and bundle sheath extensions, and biochemistry traits, including aquaporins mediated membrane permeability (Sade et al. 2014; Secchi and Zwieniecki 2014; Buckley et al. 2015; Caringella et al. 2015). However, we did not estimate the bundle sheath and bundle sheath extension traits nor biochemical traits. Further research is needed to uncover the role of leaf vein traits in determining Kleaf by choosing a plant set with a background of other similar traits. Leaf vein traits have also been suggested to influence leaf hydraulic vulnerability. Leaf dehydration may lead to embolism in the vein xylem, leading to a decline of Kx and thus to a decline Kleaf. In the present study, we found that P50 tightly correlated with VLAmajor (Fig. 3), which agrees with cavitation events shown to occur in leaf petioles or midribs using a range of hydraulic measurements and visualization approaches (Blackman et al. 2010; Brodribb et al. 2016; Scoffoni et al. 2016a).

The investigation of the influences of mesophyll anatomical traits on photosynthetic, hydraulic, and pressure–volume traits revealed that mesophyll traits strongly influence gm and pressure–volume traits; however, the Kmax is independent of mesophyll traits. Indeed, the strong influences of Sc and Tcw on gm observed here have been reported by many previous studies (Evans et al. 2009; Tomás et al. 2013; Tosens et al. 2016; Xiong et al. 2017), and our results further demonstrate that the higher gm in two crops may relate to their low Tcw and high Sc. Recently, the question of how mesophyll structural traits influence water transport inside leaves has been discussed (Buckley et al. 2015; Xiong et al. 2017). Our results showed no clear correlations between mesophyll anatomical traits and Kmax. Indeed, the water transport pathways in mesophyll tissues are complex, and many other traits, including the permeability of membranes and vein features, might play a role in Kmax. Pressure–volume traits were significantly affected by mesophyll cell wall thickness, for instance, both π0 and πtlp were positively correlated with Tcw, and ɛ negatively correlated with Tcw. Our results support that the cell wall plays an important role in drought tolerance and photosynthesis, which was also highlighted in recent studies (Nadal et al. 2018; Roig-Oliver et al. 2020). However, in the present, we could not find the negative correlation between A and ɛ described by Nadal et al. (2018), likely because in our species set, A scaled negatively with Tcw, as often observed, but ɛ also scaled negatively with Tcw (Fig. 5f). The latter observation is novel and implies that species with thicker cell walls may have instead more elastic tissues (and probably cell walls). This may reflect that while cell walls are important for both drought tolerance and photosynthesis, the cell wall components conferring one or another may differ, as recently shown by Roig-Oliver et al. (2020).

In summary, we have shown that no strict trade-off exists between carbon assimilation efficiency and drought tolerance at the leaf level across species differing in phylogeny, origin, life form, and drought tolerance. Several relationships previously shown in woody angiosperms were not consistent across woody, herb, and crop species in this study. For instance, the Kmax was decoupled from photosynthetic rate, P50, and leaf vein density. Further analysis indicated that different leaf anatomical traits in determining efficiency and tolerance traits might explain the observed lack of tradeoffs on leaf scale.