Co-ordination between leaf biomechanical resistance and hydraulic safety across 30 sub-tropical woody species

Abstract

Background and Aims

Leaf biomechanical resistance protects leaves from biotic and abiotic damage. Previous studies have revealed that enhancing leaf biomechanical resistance is costly for plant species and leads to an increase in leaf drought tolerance. We thus predicted that there is a functional correlation between leaf hydraulic safety and biomechanical characteristics.

Methods

We measured leaf morphological and anatomical traits, pressure–volume parameters, maximum leaf hydraulic conductance (K_leaf-max), leaf water potential at 50 % loss of hydraulic conductance (P50_leaf), leaf hydraulic safety margin (SM_leaf), and leaf force to tear (F_t) and punch (F_p) of 30 co-occurring woody species in a sub-tropical evergreen broadleaved forest. Linear regression analysis was performed to examine the relationships between biomechanical resistance and other leaf hydraulic traits.

Key Results

We found that higher F_t and F_p values were significantly associated with a lower (more negative) P50_leaf and a larger SM_leaf, thereby confirming the correlation between leaf biomechanical resistance and hydraulic safety. However, leaf biomechanical resistance showed no correlation with K_leaf-max, although it was significantly and negatively correlated with leaf outside-xylem hydraulic conductance. In addition, we also found that there was a significant correlation between biomechanical resistance and the modulus of elasticity by excluding an outlier.

Conclusions

The findings of this study reveal leaf biomechanical–hydraulic safety correlation in sub-tropical woody species.

INTRODUCTION

Leaves are susceptible to drought-induced desiccation (Aldea et al., 2005; Chen et al., 2017), wind injury and insect attack (Anten et al., 2010; Onoda et al., 2011). This plant organ represents the terminal component of the soil–plant–atmosphere continuum and constitutes a significant portion (up to 80 %) of whole-plant hydraulic resistance, thereby serving as a safety valve that maintains plant hydraulic functioning (Cochard et al., 2004; Sack and Holbrook, 2006). Although adequate biomechanical resistance is essential with respect to protecting leaves from biotic and abiotic damage (Peeters et al., 2007; Anten et al., 2010; Kaoru and Lourens, 2010), previous studies have revealed that enhancing leaf biomechanical resistance is costly for plant species and leads to an increase in leaf drought tolerance (Méndez-Alonzo et al., 2019; Blumenthal et al., 2020). Consequently, by examining the relationships between leaf hydraulic safety and biomechanical properties, we are likely to gain a better understanding of plant functional co-ordination under conditions of environmental stress (Méndez-Alonzo et al., 2019).

Maximum leaf hydraulic conductance (K_leaf-max) is dependent on the hydraulic structure of both the xylem and outside-xylem pathways (Nardini et al., 2012; Scoffoni et al., 2014), and affects gas exchange rates (Nardini and Luglio, 2014; Scoffoni et al., 2015). Leaf water potentials at 50 and 88 % loss of K_leaf-max (P50_leaf and P88_leaf) are common indices used to estimate leaf hydraulic vulnerability (Blackman et al., 2010), which is associated with drought-induced mortality and the distribution of plant species (Blackman et al., 2010; Nardini and Luglio, 2014). Although some studies have indicated no significant trade-off between K_leaf-max and P50_leaf (Blackman et al., 2010; Bucci et al., 2019), the existence and potential significance of this trade-off have remained controversial and warrant more detailed evaluations (Scoffoni and Sack, 2017; Bucci et al., 2019). Leaf hydraulic capacitance is defined as the change in the relative water content (RWC) against water potential change and is partially driven by the modulus of elasticity (Brodribb and Holbrook, 2003; Nadal et al., 2018). In this regard, a lower capacitance (higher modulus of elasticity) in some species indicates a more rapid stomatal closure in response to an increase in vapour pressure deficit (Fu et al., 2019; Xiong and Nadal, 2020), thus maintaining cell turgor (Bartlett et al., 2012; Nadal et al., 2018). Consequently, leaf capacitance may influence leaf hydraulic efficiency and safety (Fu et al., 2019; Xiong and Nadal, 2020).

Leaf biomechanical resistance can be quantified in terms of leaf force to punch (F_p) and force to tear (F_t) per unit fracture length or per unit width, which reflect the structural and material resistance of leaves against herbivores and physical damage (Peeters et al., 2007; Anten et al., 2010). Leaf biomechanical resistance varies considerably among plant species (Onoda et al., 2011) and is associated with leaf structural features such as the thickness of the cuticle and whole leaf (Onoda et al., 2012; He et al., 2019), vein density (Kawai and Okada, 2016) and minor vein diameter (Hua et al., 2019). Moreover, these leaf structural traits have also been found to be significantly correlated with leaf hydraulic characteristics; for example, the correlation between vein diameter and P50_leaf (Blackman et al., 2010; Scoffoni et al., 2017a), and between vein density and K_leaf-max (Nardini et al., 2012; Kawai and Okada, 2016). Therefore, it is conceivable that leaf structural properties could result in the functional co-ordination between leaf hydraulics and biomechanical traits.

In this study, we measured morphological and anatomical traits, pressure–volume curves, vulnerability curves and biomechanical resistance of leaves in 30 woody species growing in a sub-tropical moist evergreen broadleaved forest. By assessing the co-ordination and/or trade-off between biomechanical and hydraulic traits, we hypothesized that higher leaf biomechanical resistance would be associated with greater hydraulic safety, lower hydraulic conductance and higher modulus of elasticity.

MATERIALS AND METHODS

Study site and plant materials

This study was conducted at the Dinghushan Forest Ecosystem Research Station (DFERS: 23°09′21″ to 23°11′30″N, 112°30′39″ to 112°33′41″E; 300 m a.s.l.), Guangdong, China. This site is influenced by a humid monsoon climate, with a mean annual temperature of 22.5 °C and average monthly temperatures ranging from 13.8 °C (January) to 28.8 °C (July). The mean annual precipitation is approx. 1900 mm, 80 % of which falls during the wet season (from April to September). The soil in this area is a lateritic type, with a pH of 4.6 (Zhou et al., 2013) and organic matter, total N and total P contents of 48.69, 1.76 and 0.29 g kg^–1, respectively (He et al., 2016).

The core area of the DFERS Nature Reserve is an old-growth sub-tropical evergreen broadleaved forest dominated by tree species of Lauraceae and Myrtaceae (Zhu et al., 2013; Li et al., 2015). For the purposes of the present study, we selected 30 common tree and shrub species (Table 1), the canopy heights of which ranged from 1.5 m to 20 m (Li et al., 2015). For each species, five healthy mature individuals were selected, for which all leaf trait measurements were carried out during the wet season, with the exception of minimum leaf water potential, which was measured in the dry season.

Table 1.

Characteristics of 30 sub-tropical woody species.

Species	Family	Code	Canopy position
Acronychia pedunculata	Rutaceae	Ape	Mid-canopy
Aidia canthioides	Rubiaceae	Aca	Mid-canopy
Alchornea trewioides	Euphorbiaceae	Atr	Understorey
Aporosa yunnanensis	Euphorbiaceae	Ayu	Mid-canopy
Ardisia quinquegona	Myrsinaceae	Aqu	Understorey
Artocarpus styracifolius	Moraceae	Ast	Mid-canopy
Blastus cochinchinensis	Melastomataceae	Bco	Understoreey
Canarium pimela	Burseraceae	Cpi	Mid-canopy
Castanopsis chinensis	Fagaceae	Cach	Canopy
Castanopsis fissa	Fagaceae	Cfi	Canopy
Clerodendrum fortunatum	Verbenaceae	Cfo	Understorey
Cryptocarya chinensis	Lauraceae	Cch	Canopy
Cryptocarya concinna	Lauraceae	Cco	Canopy
Diospyros morrisiana	Ebenaceae	Dmo	Mid-canopy
Evodia lepta	Rutaceae	Ele	Mid-canopy
Gironniera subaequalis	Ulmaceae	Gsu	Canopy
Macaranga sampsonii	Euphorbiaceae	Masa	Mid-canopy
Machilus chinensis	Lauraceae	Mch	Canopy
Mallotus paniculatus	Euphorbiaceae	Mpa	Mid-canopy
Melastoma sanguineum	Melastomataceae	Msa	Understorey
Memecylon ligustrifolium	Melastomataceae	Mli	Mid-canopy
Mischocarpus pentapetalus	Sapindaceae	Mpe	Mid-canopy
Psychotria asiatica	Rubiaceae	Pas	Understory
Sarcosperma laurinum	Sapotaceae	Sla	Mid-canopy
Schefflera heptaphylla	Araliaceae	She	Mid-canopy
Schima superb	Theaceae	Ssu	Canopy
Sterculia lanceolata	Sterculiaceae	Stla	Mid-canopy
Syzygium acuminatissimum	Myrtaceae	Sac	Canopy
Triadica cochinchinensis	Euphorbiaceae	Tco	Mid-canopy
Xanthophyllum hainanense	Polygalaceae	Xha	Mid-canopy

Leaf biomechanical resistance and structural traits

For each assessed specimen, we sampled 2- to 3-year-old canopy branches (approx. 35 cm long and 5–10 mm in basal diameter) with attached healthy mature leaves using a retractable long-reach pruner (with a maximum length of 18 m). The collected samples were enclosed in black plastic bags, together with wet filter papers, and transported to the laboratory for further measurements. Twenty leaves from each individual were used to measure leaf biomechanical resistance employing a digital force gauge with a precision of 0.001 N (HADPI, Leqing, China). The force to tear (F_t), i.e. the force that splits a leaf in two, was assessed by applying a gradually increasing force on a rectangular (5 × 25 mm) area in the central part of a leaf, avoiding the midrib (Onoda et al., 2011). In turn, the force to punch (F_p) was defined as the force required for a 0.6 mm needle to penetrate the leaf mesophyll. Major leaf veins were avoided while performing these measurements. For determinations of leaf mass per unit area (LMA), 20–30 leaves were sampled from each individual. Total leaf area was measured using a leaf area meter (Li-3000A; LiCor, Lincoln, NE, USA), and the leaves were then oven-dried at 70 °C for 72 h to determine LMA (g cm^–2), which was calculated as the ratio of the leaf dry mass to the lamina area after removal of the petiole.

Leaf structural traits were measured for two leaves per individual. Images were obtained using a Leica DM 2500 light microscope with ×10 or ×40 objectives, and subsequently analysed using ImageJ software (National Institutes of Health, Bethesda, MD, USA). Vein density higher than the third order was measured as the minor vein density (Kawai and Okada, 2016). Prior to obtaining vein images using a light microscope, we chemically cleared leaves in 5 % NaOH and bleach following standard procedures (Scoffoni et al., 2015). Leaf cross-sections were prepared from the centre of each leaf to determine the thickness of the cuticle, epidermis, hypodermis, and palisade and spongy mesophyll (Fletcher et al., 2018). Further cross-sections were obtained from the base of the midrib to determine conduit diameter, wall thickness and vessel density (Scoffoni et al., 2017a). The original biomechanical resistance and anatomical data of the leaves have been provided in Supplementary data Tables S1 and S2.

Leaf pressure–volume relationships

Leaf pressure–volume (P–V) curve measurements were performed following previously described procedures (Tyree and Hammel, 1972; Nardini et al., 2012). Five healthy leafy branches were collected from five individuals of each species between 05.00 and 07.00 h, during which time leaf water potential is higher than –0.3 MPa (leaves were assumed to be water saturated; Zhu et al., 2019). The basal ends of branches were immersed in water and recut. The entire branches were then covered with large black plastic bags with the cut ends maintained underwater, and transported to a nearby laboratory within 1 h. It should be noted that this short rehydration process was adopted to prevent leaf water loss during transportation, and is unlikely to have resulted in oversaturation. For the purposes of P–V measurements, we selected a single distal mature leaf from each branch, which was allowed to dry at room temperature on a lab bench (approx. 25 °C). Leaf weight and water potential were successively and repeatedly measured using a 1/10⁴ precision scale (Toledo XS204; Mettler, Zurich, Switzerland) and a pressure chamber (PMS 1550D; Corvallis, OR, USA), respectively. Having completed measurements, the leaf was oven-dried at 70 °C for at least 72 h prior to measuring the corresponding oven-dry mass. In order to fit curves (Supplementary data Fig. S1) and calculate parameters such as leaf water potential at the turgor loss point (TLP) and modulus of elasticity (ε), we used a leaf P–V relationship analysis program (Schulte and Hinckley, 1985; Zhu et al., 2019). In this study, ε was calculated as the slope of pressure potential against the RWC before reaching the turgor loss point (Bartlett et al., 2012).

Leaf hydraulic conductance and vulnerability curves

We measured leaf hydraulic conductance (K_leaf) using the rehydration kinetic method, and constructed leaf vulnerability curves by measuring K_leaf at different leaf water potentials (Brodribb and Holbrook, 2003). The collection of leafy branches (approx. 35 cm long and 5–10 mm in basal diameter) was similar to that described for the measurements of leaf P–V relationships. These branches were allowed to dry out slowly on a laboratory bench to obtain a range of water potentials. Prior to the determination of K_leaf, the branch was wrapped in a black bag for approx. 1 h to ensure all attached leaves showed a similar water potential (the difference between apical and basal leaves was <0.1 MPa). Having measured the initial water potential of the equilibrated leaves (Ψ₀), neighbouring leaves were cut underwater and rehydrated for 10–200 s (t). Note that under conditions of lower leaf water potential, the rehydration time is longer, and we performed an equilibration procedure by placing rehydrated leaves in a sealed bag for approx. 15–30 min prior to measuring Ψ_f. Leaf hydraulic conductance was calculated in accordance with the following equation (Brodribb and Holbrook, 2003; Johnson et al., 2018):

$${\displaystyle \begin{array}{l}{\displaystyle K_{\mathrm{l}\mathrm{e}\mathrm{a}\mathrm{f}}=(C_{\mathrm{l}\mathrm{e}\mathrm{a}\mathrm{f}}/t)\times \ln ({\mathrm{\Psi }}_0/{\mathrm{\Psi }}_{\mathrm{f}}),}\end{array}}$$

where C_leaf is the leaf hydraulic capacitance (mmol m^–2 MPa^–1) prior to (C_leaf-FT) or after the turgor loss point, Ψ_o is the leaf water potential before partial rehydration (MPa), Ψ_f is the leaf water potential after partial rehydration, and t is the duration of rehydration (s). C_leaf is calculated from the slope of the change in RWC per leaf area against water potential range as follows (Brodribb and Holbrook, 2003):

$${\displaystyle \begin{array}{l}{\displaystyle C_{\mathrm{l}\mathrm{e}\mathrm{a}\mathrm{f}}=\mathrm{\Delta }\mathrm{R}\mathrm{W}\mathrm{C}/\mathrm{\Delta }\psi \times (\mathrm{D}\mathrm{W}/\mathrm{L}\mathrm{A})\times (\mathrm{W}\mathrm{W}/\mathrm{D}\mathrm{W})/\mathrm{M},}\end{array}}$$

where DW is the oven-dry mass (g), LA is the leaf area (m²), WW is the mass of water at saturation (g) and M is the molar mass of water (g mol^–1). Leaf vulnerability curves (Supplementary data Fig. S2) were fitted using a three-parameter sigmoid model in SigmaPlot 12.5 (Systat Software Inc., San Jose, CA, USA). Maximum leaf hydraulic conductance (K_leaf-max) was estimated from the vulnerability curves and calculated as the leaf water potential equal to zero (Nardini et al., 2012). Leaf water potentials at 50 and 88 % loss of K_leaf-max (P50_leaf and P88_leaf, respectively) were calculated to estimate the vulnerability of leaves to hydraulic dysfunction (Scoffoni et al., 2014). In a previous study, Blackman and Brodribb (2011) developed a further method for determining C_leaf according to the volume of water taken up by the leaf during the transition from the initial to final leaf water potential by using a flowmeter. They suggested that C_leaf calculated based on traditionally assessed leaf P–V relationships may overestimate K_leaf. Accordingly, this methodological issue should be taken into consideration. At the same study site, Liu et al. (2019) used the evaporative flux method (Sack and Scoffoni, 2012; Scoffoni et al., 2016) to measure leaf hydraulic conductance in the xylem (K_leaf-x) and outside-xylem (K_leaf-ox) for 15 woody species in common with this study (Supplementary data Table S1). These published data were collected and incorporated into the regression analysis in this study.

Mid-day leaf water potential and hydraulic safety margin

Mid-day leaf water potentials were measured between the hours of 13.00 and 15.00 on consecutive sunny days in November during the dry season. Two healthy canopy leaves were collected from each individual, the leaf water potentials of which were immediately determined using a PMS pressure chamber (PMS, Corvallis, OR, USA). The leaf hydraulic safety margin (SM_leaf) was calculated as the difference between mid-day leaf water potential in the dry season and P50_leaf (Johnson et al., 2016)_.

Statistical analysis

Linear regression analyses were performed using mean values to determine correlations between parameters among species using the ‘lm’ function in R 3.6.1. We conducted a path model analysis to evaluate the correlations among the ‘biomechanics’ latent variable (estimated from F_t and F_p), K_leaf-max, the modulus of elasticity (ε) and –P50_leaf using AMOS 22 software (AMOS Development Corporation, Spring House, PA, USA). Path coefficients were estimated via maximum-likelihood procedures. Better fitting models were those characterized with a probability level of the model chi-squared statistics (χ ², P > 0.05), higher comparative fit index and goodness-of-fit index, and reduced root mean square residuals (Jackson et al., 2009; Gleason et al., 2018).

RESULTS

Correlations between leaf biomechanical resistance and hydraulic traits

Across the 30 sub-tropical species in the present study, both F_t and F_p were negatively correlated with P50_leaf and P88_leaf, and were positively correlated with SM_leaf (Fig. 1; Table 2), indicating that species with higher leaf biomechanical resistance tended to exhibit a greater leaf hydraulic safety. Even excluding Memecylon ligustrifolium that had exceptionally high maximum biomechanical resistance (F_t = 1.11 kN m^–1), only the relationship between SM_leaf and F_p became weakened (Supplementary data Fig S3, r = 0.33, P = 0.07). There was, however, no significant relationship between K_leaf-max and leaf biomechanical traits (Fig. 2C. D). Additionally, in the case of 15 tree species, we analysed the correlation between leaf biomechanical resistance and leaf hydraulic conductance in the xylem and outside-xylem (K_leaf-x and K_leaf-ox, respectively) using published data from the same study site (Liu et al., 2019), and found that both F_t and F_p were negatively related to K_leaf-ox, rather than to K_leaf-x (Fig. 2; Supplementary data Fig. S4). Furthermore, we found that leaf biomechanical resistance was decoupled from the turgor loss point (TLP) across species, and was positively correlated with the leaf modulus of elasticity when the outlier M. ligustrifolium was excluded from the analysis (Fig. 3).

Fig. 1.

Correlations between leaf biomechanical resistance and (A, B) leaf water potential at 50 % loss of hydraulic conductance (P50_leaf) and (C, D) leaf hydraulic safety margin (SM_leaf) across 30 sub-tropical woody species. F_t, force to tear; F_p, force to punch. Note that these correlations are analysed by excluding Memecylon ligustrifolium (Mli) in Supplementary data Fig. S3.

Table 2.

Pearson correlations among leaf hydraulic and biomechanical traits across the 30 sub-tropical forest species.

Traits	F t	F p	K leaf-max	K leaf-mass	P50leaf	P88leaf	Ψ md	SMleaf	TLP	ε
F p	0.88***
K leaf-max	–0.11	–0.01
K leaf-mass	–0.42*	–0.31	0.80***
P50leaf	–0.45*	–0.41*	0.37*	0.43*
P88leaf	–0.49**	–0.42*	0.26	0.42*	0.82***
Ψ md	–0.15	–0.11	0.03	0.15	0.24	0.20
SMleaf	0.42*	0.38*	–0.37*	–0.39*	–0.95***	–0.77***
TLP	–0.06	0.04	0.17	0.36*	0.28	0.24	0.34	–0.18
ε	0.25	0.05	–0.48**	–0.46**	–0.52**	–0.49**	–0.04	0.52**	–0.12
C leaf-FT	–0.08	–0.00	0.52**	0.36	0.40*	0.34	0.09	–0.38*	0.27	–0.69***

Traits abbreviations: F_t, force to tear, kN m^–1; F_p, force to punch, kN m^–1; K_leaf-max, maximum leaf hydraulic conductance, mmol m^–2 s^–1 MPa^–1; K_leaf-mass, leaf hydraulic capacity on a dry mass basis; P50_leaf, leaf water potential at 50 % loss of hydraulic conductance, MPa; P88_leaf, leaf water potential at 88 % loss of hydraulic conductance, MPa; Ψ_md, mid-day leaf water potential in the dry season, MPa; SM_leaf, leaf hydraulic safety margin, MPa; TLP, water potential at turgor loss point, MPa; ε, bulk modulus of elasticity before the turgor loss point, MPa; C_leaf-FT, leaf water capacitance before the turgor loss point, mmol m^–2 MPa^–1. Statistically significant coefficients are highlighted in bold (*P < 0.05; **P < 0.01; ***P < 0.001).

Fig. 2.

Correlations between leaf biomechanical resistance and (A, B) K_leaf-ox values for the 15 woody species derived from a recently published study conducted at the same study site (Liu et al., 2019) and (C, D) maximum leaf hydraulic conductance (K_leaf-max) across 30 sub-tropical woody species. F_t, force to tear; F_p, force to punch. The species codes are shown in Table 1.

Fig. 3.

Correlations between leaf biomechanical resistance (F_t) and (A) the bulk modulus of elasticity before the turgor loss point (ε) and (B) leaf water potential at turgor loss (TLP) across sub-tropical woody species. Note that Memecylon ligustrifolium (triangle, Mli) is excluded from this correlation analysis.

Correlations among leaf hydraulic efficiency and safety

We detected a weak relationship between K_leaf-max and P50_leaf across the 30 woody species (Fig. 4), indicating a hydraulic efficiency–safety trade-off in the leaves. In contrast, we observed no significant correlation between K_leaf-max and P88_leaf (Fig. 4). A lower ε (higher C_leaf-FT) was, however, found to be associated with higher K_leaf-max and less negative P50_leaf (Table 2).

Fig. 4.

Correlations between maximum leaf hydraulic conductance (K_leaf-max) and (A) leaf water potential at 50 % loss of hydraulic conductance (P50_leaf), and (B) leaf water potential at 88 % loss of hydraulic conductance (P88_leaf) across 30 sub-tropical woody species.

Path model analysis

Path model analysis revealed correlations among leaf biomechanical resistance (biomechanics), leaf hydraulic conductance (K_leaf-max), modulus of elasticity (ε) and leaf hydraulic safety (–P50_leaf; Fig. 5). A higher biomechanical resistance made a significant contribution to a more negative P50_leaf (path coefficient 0.37), whereas leaf biomechanical resistance showed no appreciable correlation with K_leaf-max. A higher hydraulic safety (more negative P50_leaf) did not lead directly to a lower K_leaf-max (path coefficient –0.21). In addition, ε was found to have a significantly negative effect on K_leaf-max and a positive effect on –P50_leaf (path coefficients were –0.40 and 0.45, respectively).

Fig. 5.

Path model analysis of the relationships linking leaf biomechanical resistance (biomechanics) and maximum leaf hydraulic conductance (K_leaf-max), the modulus of elasticity (ε), and leaf water potential at 50 % loss of hydraulic conductance (–P50_leaf). Arrows indicate the proposed links between variables. Standardized path coefficients are shown on the arrows. Dotted and solid lines indicate non-significant and significant paths, respectively (*P < 0.05, **P < 0.01). Probability level of the model chi-squared statistics (χ ², P > 0.05), the comparative fit index (CFI), goodness-of-fit index (GFI) and root mean square residual (RMR) are model fit indicators.

DISCUSSION

Co-ordination between leaf biomechanical resistance and hydraulic safety

The findings of previous studies on sub-tropical woody species have indicated that higher leaf biomechanical resistance is associated with tougher leaves and thus a higher construction cost (He et al., 2019), which was also evidenced by the positive correlations between LMA and F_t and F_p of the present study (Supplementary data Table S3). However, in the present study, we found that none of the measured leaf anatomical traits was associated with both biomechanical resistance and P50_leaf (Supplementary data Table S3), which appears to contrast with our assumption that leaf structural features would contribute to significant correlations between leaf biomechanics and hydraulic safety. Nevertheless, when we excluded the outlier M. ligustrifolium from the analyses, we found that the leaf modulus of elasticity was negatively correlated with P50_leaf and positively correlated with F_t (Fig. 3). This correlation accordingly indicates that the rigidity of mesophyll cell walls is potentially a key factor determining the co-ordination between leaf biomechanical resistance and hydraulic safety in most sub-tropical woody species. Compared with the other species assessed in the present study, M. ligustrifolium was characterized by a notably higher F_t with more negative P50_leaf, which may conceivably be attributable to the densely arranged filiform sclereids throughout mesophyll tissue (Supplementary data Fig. S5). This species often occurs in the forest understorey, exhibiting long leaf longevity and low photosynthetic capacity (Zhu et al., 2013; He et al., 2019). The combinations of these functional characteristics are significant for its persistence in low light habitats. Accordingly, we suggest that more species and detailed analyses of leaf structural traits (cell wall thickness) are necessary to clarify such co-ordination.

Our results also implied that enhanced resistance to insect herbivory is associated with an increase in the hydraulic safety margin during the dry season (Fig. 1). Sub-tropical evergreen broadleaved forests in southern China, which are acknowledged to play an important role in current global carbon cycling (Yu et al., 2014), have been subjected to an intensification of drought stress during the past three decades (Zhou et al., 2011). In this regard, climate drying is a primary factor contributing to tree mortality and shifts in the community compositions of these sub-tropical forests (Zhou et al., 2011, 2013). Moreover, other studies in the same region have reported an increase in the frequency of insect pest outbreaks, which have the effects of suppressing tree growth and increasing mortality (Zhang et al., 2006; Chen et al., 2017). Indeed, climate change is predicted to contribute to an increase in tree mortality via drought, heat stress and insect attacks, with manifold interactive influences on forest ecosystems (Anderegg et al., 2015; Choat et al., 2018). The co-ordination between leaf biomechanical resistance and hydraulic safety may thus improve our understanding of the demographic patterns of sub-tropical forest species in response to climate change.

Leaf biomechanical resistance is correlated with outside-xylem hydraulic conductance

Contrary to our expectations, we detected no significant correlation between K_leaf-max and leaf biomechanical resistance (Fig. 2). Consistently, Kawai et al. (2016) reported decoupled correlation between leaf hydraulic efficiency and biomechanical resistance across eight Fagaceae species growing in temperate forests in Japan. This may be because leaf hydraulic conductance was heavily influenced by vein characteristics (Scoffoni et al., 2017a; Bucci et al., 2019), which could change independently of the many other factors controlling biomechanical resistance (Trifiló et al., 2016; Scoffoni et al., 2017b). On the other hand, we suspect that this apparent lack of correlation could be attributable to the fact that K_leaf-max is co-determined by both xylem and outside-xylem pathways and that our measurements of leaf biomechanical resistance were obtained mainly from the leaf proportion lacking main veins. In this regard, it should be noted that the rehydration kinetic method used in the present study was unable to distinguish the K_leaf values of different parts of leaf tissue (i.e. xylem and outside-xylem pathways). We did, however, detect a correlation between leaf biomechanics and outside-xylem leaf hydraulic conductance (K_leaf-ox values were from a published paper at the same site; Fig. 2), which could reflect the fact that outside-xylem characteristics influence leaf hydraulic efficiency (Trifiló et al., 2016; Scoffoni et al., 2017b), and also contribute to biomechanical investment (Onoda et al., 2015). Further attention should thus be focused on the outside-xylem pathways of leaves with respect to determining the correlation between leaf hydraulics and biomechanics in more species and biomes.

Significant trade-off between leaf hydraulic efficiency and safety

We found that K_leaf-mass was also significantly and positively correlated with P50_leaf (Table 2), which is consistent with the findings of previous studies that leaf hydraulic efficiency–safety trade-off is associated with the carbon investment of different leaf tissues (Simonin et al., 2012; Nardini and Luglio, 2014). In addition, our results showed that there was a weak trade-off between K_leaf-max and P50_leaf across sub-tropical woody species (Fig. 4), which was found to be associated with the modulus of elasticity (Fig. 5; Table 2). Because a high C_leaf-FT (low ε) is favourable for high K_leaf-max, it could contribute to protecting leaves from a rapid decrease in water potential and buffer the loss of hydraulic conductance (Xiong and Nadal, 2020). On the other hand, a high ε (more thicker mesophyll cell walls; Peguero-Pina et al., 2017) could prevent leaf shrinkage and cell dehydration under low water potentials, resulting in high hydraulic safety (Niinemets, 2001; Scoffoni et al., 2014; Trifiló et al., 2016; Méndez-Alonzo et al., 2019). Moreover, we did not find a significant trade-off between K_leaf-max and P88_leaf (Fig. 4), which implies that leaf hydraulic trade-off may exist in the outside-xylem pathways. Because P88_leaf indicates the irreversible hydraulic threshold at which most veins are embolized (Scholz et al., 2014), in contrast to 50 %, loss of leaf hydraulic conductance is more likely to be caused by leaf shrinkage of the outside-xylem pathways (Scoffoni et al., 2014, 2017b).

It should be noted that there are several different methods to investigate the decline of hydraulic efficiency with increasing water stress, including the rehydration kinetics (RK) method (Brodribb and Holbrook, 2003), the evaporative flux method (Sack and Scoffoni, 2012), the optical vulnerability (OV) technique (Brodribb et al., 2016) and high-resolution computed tomography (microCT; Bouche et al., 2016). In a previous study, Brodribb et al. (2016) suggest that RK and OV methods measured in four angiosperm species produce similar P50_leaf. However, other studies indicate that P50_leaf determined from the RK method was significantly higher than that measured from microCT in a conifer species (Pinus pinaster; Bouche et al., 2016) and a herb species (Triticum aestivum; Corso et al., 2020). Therefore, it would be necessary to investigate leaf hydraulic trade-off in both xylem and outside-xylem pathways by using a appropriate method in future studies.

Conclusion

The findings of this study provide insights into the biomechanics–safety–efficiency–elasticity relationships in leaves (Fig. 5). A high biomechanical resistance was found to be associated with considerable hydraulic safety (more negative P50_leaf and greater SM_leaf), and was decoupled from hydraulic efficiency (K_leaf-max) rather than K_leaf-ox. Moreover, we observed that the low modulus of elasticity was correlated with high hydraulic efficiency and low safety, contributing to a weak trade-off between hydraulic safety and efficiency. Further studies are warranted to examine similar trait correlations in a broader range of plant groups and habitats, and to assess how these relationships shape the ecological strategies of plant species.

SUPPLEMENTARY DATA

Supplementary data are available online at https://academic.oup.com/aob and consist of the following. Table S1: mean values of leaf biomechanical and hydraulic traits. Table S2: mean values of leaf morphological and anatomical traits. Table S3: relationships between leaf structural traits and leaf hydraulic and biomechanical traits. Figure S1: leaf pressure–volume curves. Figure S2: leaf vulnerability curves. Figure S3: correlations between leaf biomechanical resistance and hydraulic safety when excluding Memecylon ligustrifolium (Mli). Figure S4: correlations between leaf biomechanical resistance and leaf hydraulic conductance in the xylem. Figure S5: leaf cross-section image of Memecylon ligustrifolium.

FUNDING

This study was supported by the Natural Science Foundation of Guangxi Zhuang Autonomous Region (2017GXNSFBA198188), the Bagui Young Scholarship awarded to S.-D.Z. and a scientific research fund of Guangxi University (XTZ160182).