A trade-off between leaf hydraulic efficiency and safety across three xerophytic species in response to increased rock fragment content

Abstract

Limited information is available on the variation of plant leaf hydraulic traits in relation to soil rock fragment content (RFC), particularly for xerophytes native to rocky mountain areas. In this study, we conducted a field experiment with four gradients of RFC (0, 25, 50 and 75% ν ν⁻¹) on three different xerophytic species (Sophora davidii, Cotinus szechuanensis and Bauhinia brachycarpa). We measured predawn and midday leaf water potential (Ψ_leaf), leaf hydraulic conductance (K_leaf), Ψ_leaf induced 50% loss of K_leaf (P₅₀), pressure–volume curve traits and leaf structure. A consistent response of hydraulic traits to increased RFC was observed in three species. K_leaf showed a decrease, whereas P₅₀ and turgor loss point (Ψ_tlp) became increasingly negative with increasing RFC. Thus, a clear trade-off between hydraulic efficiency and safety was observed in the xerophytic species. In all three species, the reduction in K_leaf was associated with an increase in leaf mass per area. In S. davidii, alterations in K_leaf and P₅₀ were driven by leaf vein density (VLA) and Ψ_tlp. In C. szechuanensis, Ψ_tlp and VLA drove the changes in K_leaf and P₅₀, respectively. In B. brachycarpa, changes in P₅₀ were driven by VLA, whereas changes in both K_leaf and P₅₀ were simultaneously influenced by Ψ_tlp. Our findings suggest that adaptation to increased rockiness necessarily implies a trade-off between leaf hydraulic efficiency and safety in xerophytic species. Additionally, the trade-off between leaf hydraulic efficiency and safety among xerophytic species is likely to result from processes occurring in the xylem and the outside-xylem hydraulic pathways. These findings contribute to a better understanding of the survival strategies and mechanisms of xerophytes in rocky soils, and provide a theoretical basis for the persistence of xerophytic species in areas with stony substrates.

Introduction

Stony soils with abundant rock fragments (RFs, a particle > 2 mm in diameter) are widespread worldwide, especially in mountainous regions (Poesen  and  Lavee  1994, Nyssen  et al.  2002, Zhang  et al.  2016). The presence of RFs significantly affects soil hydraulic properties, including water retention, hydraulic conductivity and hydrological processes (e.g., water infiltration, storage and evaporation) (Baetens  et al.  2009, Ma  et al.  2010, Novák  et al.  2011, Tetegan  et al.  2011, Lai  et al.  2018). In general, an increase in soil rock fragment content (RFC) tends to decrease soil water availability (SWA) because of a reduction in soil water-holding capacity and hydraulic conductivity (Tetegan  et al.  2011). Recent investigations have focused on the influence of rock properties (such as primary porosity) on plant water status and drought response in rock-dominated landscapes (Korboulewsky  et al.  2020, Nardini  et al.  2021). These studies have verified that RFs affect various plant traits, including plant water consumption, biomass, water-use efficiency and even plant survival and mortality by altering SWA (Schwinning  2010, 2020, Mi  et al.  2016, Preisler  et al.  2019, Hu  et al.  2021, Zhang  et al.  2023). Furthermore, there is a growing trend to predict the vulnerability of species to water limitation by assessing hydraulic-associated traits (Bartlett  et al.  2014, Maréchaux  et al.  2015, Mitchell  and  O’Grady  2015). However, direct evidence of how hydraulic traits respond to varying RFC is lacking, limiting our understanding of plant acclimation processes in response to water stress in rock-dominated landscapes.

Among the hydraulic traits, those related to efficient water transport (K_leaf) and hydraulic safety (water potential (Ψ_leaf) inducing a 50% loss of K_leaf (P₅₀)) are particularly important (Liu  et al.  2018, Li  et al.  2020, Brodribb  et al.  2020) to allow species to adjust to habitats with different SWA (Ocheltree  et al.  2016, Yao  et al.  2021a, 2021b, Xiong  and  Flexas  2022). Interspecific studies have shown that K_leaf decreases with decreasing SWA (Scoffoni  et al.  2012, Trifilò  et al.  2016). Species with lower P₅₀ values are more resistant to water stress and could survive in drought-prone habitats (Gortan  et al.  2009, Nardini  et al.  2012, Savi  et al.  2017, Zhu  et al.  2018, Petruzzellis  et al.  2021). Nevertheless, to our knowledge, no studies have investigated how local RF conditions affect leaf hydraulic traits. Furthermore, the conclusion that lower P₅₀ values occur in water-limited habitats comes primarily from species in temperate and tropical forests (Gleason  et al.  2016). Experimental evidence shows that intraspecific variability of P₅₀ in stem xylem is generally low (Matzner  et al.  2001, Skelton  et al.  2019). For xerophytic species native to rocky regions, and at the intraspecific level, results are poor. Therefore, if these conclusions are also applicable to intraspecific situations, we would expect to observe more negative P₅₀ (high hydraulic safety) and reduced K_leaf (low hydraulic efficiency) values under high RFC conditions (H1).

In drought-prone habitats, the midday minimum daily Ψ_leaf value of plants often approaches or even falls below the P₅₀ value (Johnson  et al.  2009), suggesting that their leaves often experience substantial K_leaf loss (Nardini  et al.  2008, 2011, Johnson  et al.  2009). Furthermore, the presence of RFs could exacerbate water stress in xerophytic species in drought-prone regions. Therefore, it is important to investigate the relationships between hydraulic traits in xerophytic species in order to elucidate their acclimation processes to elevated RFC. Previous studies have revealed a trade-off between leaf hydraulic safety (P₅₀) and efficiency (K_leaf) (Choat  et al.  2012, Nardini  et al.  2012, Ocheltree  et al.  2016, Scoffoni  and  Sack  2017, Scoffoni  et al.  2017), suggesting that leaf hydraulic efficiency is compromised for the sake of safety in coping with variable water stress. However, contrasting results have emerged from other investigations, with some suggesting a lack of substantial associations (Xiong  and  Flexas  2022), or others showing positive correlations between leaf hydraulic safety and efficiency (Yao  et al.  2021a). These discrepancies may be because of species difference, as inter- and intra-specific species differ in their hydraulic responses to environmental change, potentially leading to different and conflicting conclusions. However, to date, no research has attempted to determine whether there is a trade-off between leaf hydraulic efficiency and safety within single species when adapting to varying SWA. Therefore, to better understand how xerophytic species adapt to water stress in rocky environments, controlled experiments are needed to accurately explore the leaf hydraulic response to changes in RFC.

If there is a trade-off between hydraulic efficiency and safety of xerophytic species in adapting to increasing RFC, it is important to elucidate the drivers behind such a relationship. A recent study by Nardini  (2022) has shown that leaf mass per area (LMA) is mechanistically correlated to physiological traits that confer drought tolerance. During drought, K_leaf is influenced by both the xylem (K_x) (Scoffoni  et al.  2011, Buckley  et al.  2015) and/or the outside-xylem compartment (K_ox) (Gascó  et al.  2004, Yao  et al.  2021a). The accumulation of xylem embolism in the vein system is the main reason for the decrease in K_x (Nardini  et al.  2001, Scoffoni  et al.  2017, Trifilò  et al.  2021). Species with higher major leaf vein density (VLA_major) typically exhibit increased resistance to embolism (as evidenced by more negative P₅₀ values) because of the presence of numerous alternative pathways for water movement across the leaf. Consequently, a higher density of VLA_major serves to mitigate the negative effects of drought on K_x (Nardini  et al.  2003a, Scoffoni  et al.  2011). Furthermore, VLA_major is inversely correlated with leaf area (Scoffoni  et al.  2011, Mauri  et al.  2020), and this is often coupled with high LMA because of the dense vascular structure (Blonder  et al.  2011, Nardini  et al.  2012, John  et al.  2017). Recent research has indicated that lower K_leaf can also be attributed to reduced minor vein density (VLA_minor) (Yao  et al.  2021b). Therefore, modifications of the leaf vein system associated with leaf embolism resistance might translate into leaf tissue density, which could have a direct effect on LMA (Wang  et al.  2021).

Another important determinant of K_leaf decline is the loss of cell turgor (Ψ_tlp), which induces cell and leaf shrinkage (Canny  et al.  2011) and leads to reduction in K_OX by hindering apoplastic water movement in the mesophyll (Scoffoni  et al.  2014, Trifilò  et al.  2016, Abate  et al.  2021). Therefore, a more negative Ψ_tlp, which reduces the risk of cell shrinkage, can help to maintain a relatively constant K_OX under drought conditions (Farrell  et al.  2017). Furthermore, other traits related to pressure–volume (P–V) curve, such as leaf capacitance (C_leaf), cell osmotic potential (π₀) and cell wall elastic modulus (ε), could also influence the variation in Ψ_tlp and might further translate into K_leaf drop (Bartlett  et al.  2012, Turner  2017). Some studies have shown significant correlations between leaf Ψ_tlp and P₅₀ in different species in habitats that differ in long-term SWA (Blackman  et al.  2010, Nardini  et al.  2012, Nardini  and  Luglio  2014). Thus, Ψ_tlp and associated cellular traits provide an indirect linkage between P₅₀ and LMA. Despite extensive investigations of trade-off correlations involving K_leaf, P–V traits, P₅₀ and LMA under different environmental conditions (Zhu  et al.  2013, Gleason  et al.  2016, Louis  et al.  2018), the influence of RFs on these traits remains unclear, especially for xerophytic species native to arid regions. In general, leaves with high LMA were more frequently observed in nutrient- and water-limited habitats (Nardini  2022). As LMA comprehensively reflects variation in leaf structure (mainly including leaf vein traits and cell characteristics, reflected by P–V traits in the present study) in response to environmental change (Poorter  et al.  2009), we hypothesized that the increasing RFC would directly affect the LMA, which is mechanistically linked to leaf vein traits, outside-xylem tissue traits (i.e., Ψ_tlp) and the maintenance of K_x and K_ox, ultimately driving the trade-off correlations between hydraulic efficiency and safety (H2).

A field experiment was conducted to examine the effects of increasing RF addition treatments (0, 25, 50 and 75%, ν ν⁻¹) on leaf hydraulic efficiency and safety traits, as well as their relationship in three xerophytic species. The experiment allowed for the simultaneous assessment of responses in hydraulically related traits within and between species as RFC increased. The aim of this study was to answer the following questions. (i) How do leaf hydraulic-related traits of xerophytic species respond to increasing RFC? (ii) Is there an inherent trade-off between leaf hydraulic efficiency and safety in these three xerophytes? (iii) What are the contributing factors that influence this trade-off among xerophytic species in response to changes in RFC?

Materials and methods

Study site and experimental design

The RF addition experiment was carried out in a common garden located on Jingzhou Hill in Maoxian County, Sichuan, China (31°70′N, 103°87′E) at an elevation of 1637 m. The site experiences a mean annual temperature of 15.6 °C (Hu  et al.  2021) and an annual potential evaporation of 1332 mm. The mean annual precipitation is 494.8 mm, with ~80% of falling during the growing season (May–October) (Bao  et al.  2012). The soil at the site is classified as Cinnamon, with a coarse texture that extends to a depth of 50–70 cm (FAO-UNESCO  1988, Bao  et al.  2012). The site was previously used for cultivating potatoes and celery until April 2016, which meant a 2-year interval before the of the current study began.

This study focuses on the three most common xerophytic species found in the dry valley of the Minjiang River (31°42′N, 103°53′E, altitude range of 1600–1920 m). Sophora davidii is a species that typically fixes nitrogen and has strong drought tolerance (Li  et al.  2009). Bauhinia brachycarpa is legume that relies on arbuscular mycorrhizal fungi for growth but does not have the ability to fix nitrogen (Li  et al.  2010). Cotinus szechuanensis is a slow-growing species that is characterized by a deep root system (Hu  et al.  2021). The Minjiang River dry valley is a typical arid ecosystem with low rainfall (400–700 mm), high evaporation rate (1400 mm), high temperature and a strong Foehn effect (Yang  et al.  2020). Vegetation development in this region is hindered by heightened RFC-induced water deficits caused by abundant RFs throughout soil profiles (Li  et al.  2008, Wu  et al.  2008, Xu  et al.  2008).

In February 2018, a completely randomized block experiment was conducted at the study site. The experiment comprising 12 treatment combinations involving four RFC levels (0, 25, 50 and 75% volumetric contents, ν ν⁻¹) paired with three species. Each treatment included five replicates, resulting in a total of 60 plots. The RFC setting was based on a previous investigation in an arid valley (Bao  et al.  2012), where RFC ranges from 1 to 65% (ν ν⁻¹). Each plot was represented by a pit measuring 1 m in length, 1 m in width and 0.5 m in depth, with a distance of 0.5 m between adjacent plots (Figure S1 available as Supplementary data at Tree Physiology Online). Schist was chosen as the designated RF for this study because of its prevalence in the region. RFs were extracted from the pit material, along with additional local deposits, to ensure an adequate supply. Rocks were crushed into fragments of 1–2 cm in diameter. Twelve samples were used to determine the porosity (V_rock), density (d_rock) and available water content (AWC). The V_rock was measured using the water displacement method (Hughes  2005). To obtain the AWC, sequential measurements of Ψ_rock and corresponding weight were taken during controlled dehydration. After each dehydration step, samples were sealed in the sample holder for 1 h before Ψ_rock was measured. Experiments were stopped at Ψ_rock ≤ −5 MPa and samples were oven dried at 70 °C for 72 h to obtain dry weight (DW). The AWC for plant uptake was calculated as the difference between rock water content values at −0.2 and −1.5 MPa (Nardini  et al.  2021). The resulting d_rock was 2.56 ± 0.03 g cm⁻³, V_rock was 1.45 ± 0.38% and AWC averaged 6 mg g⁻¹ (Hu  et al.  2021). To achieve the experimental design, soils in each plot were excavated, sieved through a 2-mm mesh and mixed uniformly with RFs (1–2 cm in diameter). The mixture was then reintroduced into each plot, and polyethylene film was affixed to the plot walls to prevent external disturbances. It is worth noting that the plot bottoms were left unlined to facilitate natural drainage. The entire experimental setup procedure was concluded by April 2018. In August 2022, we selected 52-month-old plants of each species with a height of 1.5–1.8 m and an average canopy size of 0.9–1.2 m to investigate the lasting effects of RFC on leaf hydraulic-related traits. Please refer to Figure S2, available as Supplementary data at Tree Physiology Online, for further information on the study’s timeline.

Leaf water potential

In August 2022, we measured the predawn and midday leaf water potentials (Ψ_leaf) of five fully expanded leaves from five individuals of each species at each RFC level using a pressure chamber (Model 3115 Portable Field Plant Water Console, Santa Barbara, CA, USA). The measurements were taken on a sunny day at 04:00–06:00 h and 12:00–14:00 h, respectively (22 August 2022). To prevent water loss in transpiring samples, we followed the precautions described by Turner  (1988). Briefly, leaves were collected and immediately placed in a sealable bag that had been previously flushed with air to ensure 100% humidity. After 30 min of equilibration, Ψ_leaf was measured according to the method described by Yang  et al.  (2021).

Pressure–volume curve traits

To determine the P–V curve traits, we randomly selected five leafy branches from five individuals for each RFC level within each species (one individual per plot). The branches were cut in deionized water to prevent air entry. Subsequently, the leaves were recut underwater and left to rehydrate overnight. After complete rehydration, leaf area (A_leaf) was determined using a leaf area meter (Li3100C; Lincoln, NE, USA), and then the initial Ψ_leaf was measured by a pressure chamber (Model 3115 Portable Field Plant Water Console, Santa Barbara, CA, USA). The experiments started once the leaf water potential (Ψ_leaf) reached a value of ≥−0.2 MPa. Following the first Ψ_leaf measurement, leaves were weighed on a digital balance (with a precision of 10⁻⁴ g). After losing 2–5 mg of water, Ψ_leaf was remeasured and the process was repeated until a strictly linear relationship between 1/Ψ_leaf and water loss at four or five successive points. The leaves were then dried in an oven at 70 °C for 48 h to determine their DW. The P–V curves, turgor loss point (Ψ_tlp), modulus of elasticity (ɛ) and related parameters were calculated using an Excel sheet developed by Sack  and  Pasquet-Kok  (2011). The osmotic potential of the leaf at full turgor (π₀) was estimated by calculating the negative value of the y-intercept of the linear regression of the P–V curve under the Ψ_tlp. The bulk leaf capacitance before turgor loss (C_leaf) was calculated from P–V curves based on the slope of the relationship between Ψ_leaf and water loss, and normalized by A_leaf and DW.

Leaf hydraulic conductance

The hydraulic conductance of the entire leaf, including xylem conduits and mesophyll components, was measured using the rehydration technique outlined by Brodribb  and  Holbrook  (2003). A total of 600 branches were collected (10 branches × 5 individuals × 3 species × 4 RFC levels) at predawn. The branches were enclosed in black plastic bags and quickly transported to the laboratory. The branches were then dehydrated through bench drying, resulting in a range of leaf water potential values (−0.2 to −4.0 MPa). After a minimum of 0.5 h of dark treatment to ensure stomatal closure and leaf water potential equilibration, we selected two leaves from each branch to measure the initial leaf water potential (Ψ₀, MPa) using a pressure chamber. The adjacent leaves were then cut underwater and rehydrated for specific durations (t, 30–300 s). Following this, the leaves were extracted and dried to determine the leaf water potential (Ψ_f, MPa) after hydration.

Following rehydration, the leaves were removed and dried to measure the leaf water potential (Ψ_f, MPa). K_leaf was then calculated using the following formula:

where C_leaf represents the bulk leaf capacitance before turgor loss, Ψ₀ represents the initial leaf water potential and Ψ_f corresponds to the water potential of the rehydrated leaf. In this study, the maximum K_leaf was calculated as the mean value obtained from well-hydrated leaves (Ψ₀ > −0.5 MPa). This value was used to calculate area-based (K_area) and mass-based (K_mass) measurements according to C_area and C_mass, respectively. However, it is important to note that this estimation method has limitations. This is because in some species, leaf tissue is hydraulically compartmentalized, and only a portion of the total leaf water is well-connected to the transpiration stream. This can be reflected by the dynamic capacitance, but not by the C_leaf (Blackman  and  Brodribb  2011). To determine the leaf water potential at 12, 50 and 88% loss of hydraulic conductance (P₁₂, P₅₀, P₈₈), leaf vulnerability curves were fitted for each species at varying RFC levels using a Weibull model (Ogle  et al.  2009) and the ‘fitcond’ function of the ‘fitplc’ package (Duursma  and  Choat  2017) in the R Statistical Programming environment.

Leaf mass per area and leaf vein length measurements

The leaves used to establish the P–V curve were also used to calculate the LMA, which is determined by the ratio of DW to A_leaf. Vein length per leaf area (VLA) was measured following the methodology outlined by Scoffoni  et al.  (2014). Specifically, for each individual, three leaves were chemically bleached in 5% NaOH and then strained with safranin and fast green. The leaf vein system characteristics were assessing using an image processing system. The system consisted of a microscope (Olympus BX53F, Japan), a camera (Toupcam, Hangzhou ToupTek Photonics Co., Ltd, China) and a computer equipped with analysis software (ToupView, Hangzhou ToupTek Photonics Co., Ltd). The leaf vein images were analyzed using WinRhizo 2020 (Regent Instrument, Canada). The length of the leaf veins was divided into major vein length (VLA_major), which includes the first- to third-order leaf veins, and minor vein length (VLA_minor), which includes veins higher than third order. VLA_major and VLA_minor were calculated as the vein length per unit area. The total VLA was calculated as the sum of VLA_major and VLA_minor. Figure S3, available as Supplementary data at Tree Physiology Online, presents the leaf vein systems for the three species with different RFC.

Measurement of SWC

Soil water content (SWC) was measured using neutron probes (CNC503DR) at 10 cm intervals within the soil profile from 0 to 50 cm. For a detailed procedure, refer to Hu  et al.  (2021). In brief, SWC measurements were conducted every 3 days from July to September 2018, followed by weekly measurements from October 2018 to March 2019. The data collected in 2021 were used to investigate the changes in SWC across soil profiles under different RFC conditions, as shown in Figure S4 available as Supplementary data at Tree Physiology Online.

Statistical analysis

To examine variations in leaf traits among different RFC levels, one-way ANOVA analysis was conducted. Multiple comparisons were carried out using the least significant difference (LSD) test with a significance threshold of P < 0.05. Regression analysis was performed using mean values to investigate correlations between parameters. Spearman correlation analysis was used to explore the coordination and trade-off relationships between efficiency (K_leaf), P–V traits, safety (P₁₂, P₅₀ and P₈₈) and structural variables (LMA, VLA). The ‘corrplot’ package in R (Wei  and  Simko  2021) was utilized for this purpose. Partial least squares path model (PLS-PM) was implemented to examine the direct and indirect effects of rock fragments (RFC), LMA, P–V traits (characterized by Ψ_tlp), leaf vein traits (characterized by VLA, VLA_major, VLA_minor) and K_leaf (characterized by K_area) on leaf hydraulic safety (characterized by P₅₀). The model’s goodness-of-fit (GOF) index, which should exceed 0.7 for reliable results, was used for evaluation. The PLS-PM model was constructed using the R package ‘plspm’ (v 0.5.0) (Gaston  et al.  2023).

Results

Response of leaf traits to increased RFC

The predawn and midday Ψ_leaf values of the three species consistently decreased as RFC increased (Figure 1). Predawn Ψ_leaf ranged from −0.54 MPa in 0% RFC to −1.62 MPa in 75% RFC (Figure 1a), whereas midday Ψ_leaf values were even more negative, ranging from −1.22 MPa in 0% RFC to −2.81 MPa in 75% RFC (Figure 1b).

Figure 1.

Response of leaf water potential (Ψ_leaf) to RFC in three xerophytic species. (a) Predawn Ψ_leaf, (b) midday Ψ_leaf. Mean ± SE (n = 5). Different lowercase letters indicate significant differences (LSD test, P < 0.05) among RFC treatments.

The leaf structural traits of the three species showed significant changes in response to variations in RFC, except for A_leaf (Table 1). Leaf mass per area, VLA and VLA_major consistently increased from 0 to 75% RFC across all three species. In the case of VLA_minor, S. davidii showed an increasing trend, whereas no clear trend was observed for C. szechuanensis and B. brachycarpa (Table 1). However, A_leaf did not exhibit a distinct trend in response to RFC among the three species. Moreover, significant interspecific differences were observed in all structural traits (P < 0.05; Table 1).

Table 1.

Effects of RFC and species on leaf area (A_leaf), LMA, leaf VLA, major vein length per leaf area (VLA_major) and minor vein length per leaf area (VLA_minor). Mean ± SE (n = 5).

Table 2 shows that for all three species, there was a clear increase in C_area, C_mass and ε, whereas π₀ and Ψ_tlp became more negative with increasing RFC in relation to P–V curve parameters. Significant interspecific differences were observed in P–V traits, except for Ψ_tlp (Table 2). Figure 2 illustrates that leaf hydraulic conductance, as measured by K_area and K_mass, was higher in soil with low RFC and decreased continuously as RFC increased. Interspecific differences were significant in K_area and K_mass (Figure 2).

Table 2.

Effects of RFC and species on leaf capacitance area based (C_area) and mass based (C_mass), modulus of elasticity (ɛ), osmotic potential at full turgor (π₀) and at turgor loss point (Ψ_tlp). Mean ± SE (n = 5).

Figure 2.

Maximum leaf hydraulic conductance on a leaf area (K_area) (a) and leaf dry mass (K_mass) (b) basis, as measured in three species varied with soil RFC. Mean ± SE (n = 5). Different lowercase letters indicate significant differences (LSD test, P < 0.05) among RFC treatments.

To assess the vulnerability of the leaf hydraulic system, vulnerability curves were constructed for the three species, showing K_area decline in relation to leaf water potential (Figure 3). Under high RFC conditions, all species displayed greater resistance to embolism compared with lower RFC conditions, with more negative values of P₁₂, P₅₀ and P₈₈ observed in high RFC conditions.

Figure 3.

Changes of leaf hydraulic conductance on a leaf area basis (K_area) as a function of leaf water potential (Ψ_leaf) in S. davidii (a–d), C. szechuanensis (e–h) and B. brachycarpa (i–l), varied with soil RFC. Dotted lines, respectively, indicate the Ψ_leaf values inducing 12% (P₁₂), 50% (P₅₀) and 88% (P₈₈) of K_area decline.

Trait correlations

The positive correlation between Ψ_tlp and P₅₀ was observed across all three species (Figure 4). Further analysis was conducted to investigate the effects of LMA and VLA on K_area, Ψ_tlp and P₅₀, revealing strong influences of LMA and VLA on these parameters (Figure 5). Specifically, Ψ_tlp and P₅₀ showed positive correlations with LMA and VLA (Figure 5a, b, d and e), whereas K_area displayed negative correlations with both LMA and VLA (Figure 5e and f).

Figure 4.

Relationships between: leaf water potential inducing 50% loss of leaf hydraulic conductance (P₅₀) and maximum leaf hydraulic conductance on area (K_area) (a) and mass basis (K_mass) (b); turgor loss point (Ψ_tlp) and maximum K_area (c), maximum K_mass (d). Lines were fitted using a linear model and reported together with r² and P values. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001.

Figure 5.

Relationships between LMA and leaf water potential inducing 50% loss of leaf hydraulic conductance (P₅₀) (a), turgor loss point (Ψ_tlp) (b) and maximum leaf hydraulic conductance (K_area) (c), as well as relationships with VLA and P₅₀ (d), Ψ_tlp (e), K_area (f).

Figure 6 shows the correlation analysis of hydraulic-related traits, which revealed some species-specific differences. Across all three species, K_leaf had significant negative relationships with P–V traits and leaf safety (Figure 6), respectively. Additionally, P–V traits showed a significant positive correlation with leaf safety. In the case of S. davidii, all leaf structural traits showed strong correlations with K_leaf, P–V traits and leaf safety (Figure 6a). However, for C. szechuanensis, neither A_leaf nor VLA_minor showed significant relationships with other traits (Figure 6b). Similarly, for B. brachycarpa, only VLA_minor did not display significant correlations with the other traits (Figure 6c).

Figure 6.

Spearman correlation coefficients between leaf hydraulic-related traits on S. davidii (a), C. szechuanensis (b) and B. brachycarpa (c). K_area: maximum area-based leaf hydraulic conductance; K_mass: maximum mass-based leaf hydraulic conductance; π₀ (−MPa): leaf osmotic potential at full turgor; Ψ_tlp (−MPa): turgor loss point; C_area: area-based bulk leaf capacitance before turgor loss; C_mass: mass-based bulk leaf capacitance before turgor loss; ɛ: modulus of elasticity; P₁₂: Ψ_leaf values inducing 12% loss of leaf hydraulic conductance; P₅₀: Ψ_leaf values inducing 50% loss of leaf hydraulic conductance; P₈₈: Ψ_leaf values inducing 88% loss of leaf hydraulic conductance; A_leaf: leaf area; VLA: total leaf vein density; VLA_major: major leaf vein density; VLA_minor: minor leaf vein density. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001.

Response mechanism of leaf hydraulic efficiency and safety to increasing RFC

The PLS-PM analysis showed the direct and indirect effects of RFC, LMA, P–V traits and VLA on K_leaf and leaf hydraulic safety (GOF > 0.7) (Figure 7). The results indicate that leaf hydraulic safety was significantly associated with VLA and Ψ_tlp (Figure 7). Additionally, LMA was found to have a negative impact on K_leaf across the three species (Figure 7). In S. davidii, K_leaf was significantly influenced by VLA and Ψ_tlp (Figure 7a), whereas in C. szechuanensis and B. brachycarpa, only Ψ_tlp had a positive effect on K_leaf (Figure 7b and c).

Figure 7.

The PLS-PM describing the direct and indirect effects on leaf hydraulic efficiency (K_leaf) and safety in S. davidii (a), C. szechuanensis (b) and B. brachycarpa (c). Solid line and dotted line indicated the relationships are significant (P < 0.05) and insignificant (P > 0.05), respectively. Numbers adjacent to the arrows are standardized path coefficients. The proportion of variance explained (R²) appears in each response variable in the model.

Discussion

Response of leaf hydraulic traits to increasing RFC

Predawn and midday Ψ_leaf variations can indicate plant responses to changes in SWC (Williams  and  Araujo  2002). Recent research (Huang  et al.  2023) has shown that the RFs used in this study could potentially provide some water to plants. In our investigation, however, both predawn Ψ_leaf and midday Ψ_leaf showed significant declines as RFC increased (Figure 1). This suggests indicating that soils with higher RFC have a lower water holding capacity (Figure S4, available as Supplementary data at Tree Physiology Online), resulting in limited SWA for plants grown under high RFC conditions (Hu  et al.  2021, Huang  et al.  2022).

Leaf hydraulic safety (P₅₀) and hydraulic efficiency (K_leaf) are recognized as key physiological traits for plant acclimation to diverse SWA conditions (Nardini  and  Luglio  2014, Scoffoni  et al.  2017). In this study, we extended the assessment of P₅₀ and K_leaf to leaves of three xerophytic species in response to increasing RFC. We observed a consistent intraspecific trend across all three species, where P₅₀ became more negative (Figure 3) and K_leaf decreased gradually (Figure 2) (H1), which indicating that the xerophytic species improve their drought tolerance by sacrificed hydraulic transport capacity. This finding demonstrates the impressive adaptability of K_leaf and P₅₀ in xerophytic species to increasing RFC, which is in contrast to recent research indicating that site water availability had low influence on P₅₀ in Fagus sylvatica L. at stem level (Weithmann  et al.  2022). Unfortunately, in this study, the K_leaf value was calculated using leaf bulk capacitance instead of dynamic capacitance. This may lead to bias because of the potential hydraulic compartmentalization in the leaf tissue (Blackman  and  Brodribb  2011). This is an important limitation of our study and implies that our results cannot be compared directly with the data measured by other standard methods (Scoffoni  et al.  2012, Hernandez-Santana  et al.  2016).

Pressure–volume traits have been increasingly used as functional indicators of drought tolerance (Sack  et al.  2003, Bartlett  et al.  2012, 2014). Ecologically, Ψ_tlp is known to be strongly correlated with SWA (Bartlett  et al.  2012), and species from drier habitats usually exhibit lower Ψ_tlp values (Lenz  et al.  2006, Merchant  et al.  2007, Mitchell  and  O’Grady  2015). This study found that there was a significant RFC-dependent plasticity across the three species. Specifically, π₀ and Ψ_tlp became more negative, whereas C_leaf and ε gradually increased (Table 2). Overall, from an intraspecific variation perspective, the hydraulic traits of all three xerophytic species adjusted to improve drought tolerance under increasing RFC. To date, the intraspecific variability of leaf hydraulic-related traits in response to environmental variation is still not well understood (Pritzkow  et al.  2020). Despite the application of different measurement principles to estimate leaf hydraulics relative to standard approaches (Blackman  and  Brodribb  2011, Scoffoni  et al.  2012), and the corresponding methodological limitations, to our knowledge, this is the first experimental exploration into how intraspecific variation in K_leaf and P₅₀ of xerophytic species reacts to varying RFC. Therefore, in comparison with earlier studies involving multiple species, our findings serve to elucidate the drought acclimation strategies of xerophytic species to increasing RFC. Moreover, they contribute to a deeper understanding of the correlations between leaf hydraulic efficiency and safety in rocky environments.

Clear efficiency–safety trade-off with RFC in xerophytic species

Correlations between leaf hydraulic efficiency and safety can lead to different or even contrasting conclusions when examined across multiple species, closely related species and within single species. These conclusions may present negative, null or even positive relationships (Nardini  et al.  2014, Pritzkow  et al.  2020, Yao et al. 2021a, 2021b). In addition, these relationships are affected by climate, site-specific water availability and species variability (Scoffoni  et al.  2017, Pritzkow  et al.  2020). In this study, we extended the correlation between hydraulic efficiency and safety to leaves of xerophytic species responding to increased RFC. It is worth noting that there is a significant interspecific difference in K_leaf and P₅₀ (Figures 2 and 3). However, the trade-off between hydraulic efficiency and safety in response to increasing RFC displayed remarkable similarity across the three xerophytic species (Figures 4 and 6). Therefore, a pronounced and consistent trade-off between hydraulic efficiency and safety emerged across xerophytic species in response to increasing RFC (Figure 4). The improved leaf safety of xerophytic species in coping with high RFC conditions comes at the cost of hydraulic efficiency. On the other hand, this trade-off appears to be an important element in shaping the distribution of xerophytic species (Engelbrecht  et al.  2007, Nardini  et al.  2012). Specifically, xerophytic species from the same habitats tend to adopt similar hydraulic acclimation processes to cope with environmental water stress, regardless of their unique ecological traits.

Drivers of the trade-off between hydraulic safety and efficiency in xerophytic species

In adapting to increasing RFC, while our evidence supports a clear trade-off between leaf hydraulic efficiency and safety in xerophytic species adjusting to increasing RFC, the drivers behind the observed variation in these traits appears to differ among species (Figure 7). As hypothesized, the similarity in LMA and VLA_major response patterns, but disparity in A_leaf and VLA_minor (Table 1) suggest that changes in leaf structure may be the drivers in establishing the correlations between leaf hydraulics and safety (Nardini  2022).

In this study, A_leaf decreased slightly with increasing RFC for S. davidii and B. brachycarpa, but LMA and VLA_major increased gradually from 0 to 75% RFC across three species (Table 1). These changes are consistent with the trend toward smaller and tougher (high LMA) leaves that are typically found in drier and nutrient-limited sites (Poorter  et al.  2009, Nardini  et al.  2014). The relationship between VLA_major and P₅₀ was negative across the three species (Figure 6). This finding is consistent with previous studies, which suggested that an abundance of major veins can facilitate alternative water pathways in situations where some veins are partially blocked by embolism. This mechanism helps to mitigate the adverse effects of drought on leaf vein hydraulic conductance (K_x) (Nardini  et al.  2003b, Scoffoni  et al.  2011). Interestingly, our study revealed a positive correlation between VLA_minor and P₅₀ in S. davidii (Figure 6a), highlighting the significant role of the minor vein system in shaping leaf hydraulic safety. This observation could explain why S. davidii is more drought tolerant than the other two species (Figure 3). Furthermore, a negative correlation between leaf area and VLA_major was observed in S. davidii and B. brachycarpa (Figure 6a and c), resulting in increased LMA values (Blonder  et al.  2011, Nardini  et al.  2012). Our study also demonstrates this indirect impact of LMA on leaf hydraulic safety (Figure 7) (Nardini  2022). In summary, RFC significantly influences LMA and leaf vein systems, which, in turn, regulate the trade-off dynamics between hydraulic safety and efficiency in xerophytic species (Figure 7).

Leaf hydraulic safety is closely related to outside-xylem tissue traits such as Ψ_tlp. Therefore, the variation in P–V traits can also determine the trade-off relationship between leaf hydraulic efficiency and safety (Scoffoni  et al.  2014, Trifilò  et al.  2016, Abate  et al.  2021). In addition, Brodribb  et al.  (2016) observed a decline of 50–66% in K_leaf as Ψ_leaf approached Ψ_tlp. Similar results were observed in Arabidopsis thaliana (Scoffoni  et al.  2018), eight species with different life forms (Scoffoni  et al.  2017) and Caragana species (Yao  et al.  2021a). These results suggest that there is a trade-off between leaf safety and hydraulic efficiency, which can also be regulated by P–V traits. In the present study, a more negative Ψ_tlp resulted in a more negative P₅₀ in high RFC conditions (Figure 7). More negative Ψ_tlp could provide wider range of midday Ψ_leaf at which the leaf maintains basic function and remain turgid (Bartlett  et al.  2012, Yao  et al.  2021b). Additionally, our study revealed a significant intraspecific variation of Ψ_tlp, for instance with B. brachycarpa approaching 2 MPa (Table 2). Across all three species, we found that Ψ_tlp was closely related to π₀ and ε (Figure 6), which is consistent with previous studies indicating that differences in Ψ_tlp are primarily driven by parallel variation in π₀ and partially by modifications of ε (Bartlett  et al.  2012, Turner  2017). The mechanistic link between Ψ_tlp and π₀ has been well understood (Turner  2017); the negative relationships between ε and Ψ_tlp might reflect coordination between P–V traits, suggesting that when fully hydrated, cells with low π₀ develop high turgor pressure, possibly requiring mechanical reinforcement of cell walls (Mitchell  et al.  2008, Savi  et al.  2017).

Additionally, it is worth noting that LMA had a significant impact on K_leaf in all three species (Figure 7), suggesting that the trade-off between hydraulic efficiency and safety is also regulated by LMA (Nardini  2022). Unfortunately, our current study did not measure additional mesophyll anatomical parameters, such as mesophyll cell layers, cell wall thickness and mesophyll cell size (John  et al.  2013), or specific LMA constituents, such as leaf density and thickness (Nadal  et al.  2023). This limits our capacity to further clarify the regulatory mechanism of the trade-off between leaf efficiency and safety in xerophytic species. In addition, in terms of the experimental design, it is important to consider that the bottom of the pits was not lined with plastic film. This allowed plant roots to grow below the rock layer, which could have interfered with our conclusions. However, based on the monitoring of fine-root tissue density and biomass accumulation fraction vertical distribution (0–70 cm) (Figure S5 available as Supplementary data at Tree Physiology Online), it is believed that these plants were still limited after 4 years because of the presence of RFs in the top 0.5 m of soil the profile. However, it is important to exercise caution when drawing conclusions about the relationship between VLA and K_leaf (Figures 6 and 7), as K_leaf in this study represents the entire leaves rather than being divided into K_x and K_ox. Therefore, to further clarify the underlying reasons for this trade-off relationship, future studies should focus on a wider range of leaf anatomical and structural traits, as well as more detailed K_leaf parameters (Buckley  et al.  2015, Sonawane  et al.  2021).

Conclusions

The study revealed notable variations in leaf hydraulic traits across three xerophytic species as RFC increased. A clear trade-off was found between leaf hydraulic efficiency and safety as RFC increased for these species. This trade-off likely results from changes in both the xylem and the outside-xylem hydraulic pathways. These findings highlight the important role of the trade-off between leaf hydraulic efficiency and safety in the acclimation of xerophytic species to drought stress, particularly in highly rocky soils. Importantly, our findings highlight the role of leaf vein traits and P–V traits in determining the dynamic trade-off between leaf hydraulic efficiency and safety in xerophytic species. This has valuable implications for studying intraspecific variability in hydraulic traits, while also emphasizing the influence of localized soil rock conditions in elucidating the geographic distribution patterns of these species. Future studies on plant acclimation and their responses to climate change should consider RFC variations more thoroughly.

Conflict of interest

None declared.

Funding

The National Natural Science Foundation of China (No. 32271654).

Authors’ contributions

X.Z., W.B. and F.L. conceived the study and charted the experimental plan. X.Z., S.M., H.H. and L.H. performed the experiments. X.Z., S.M., H.H. and L.H. carried out the field experiment and performed most of the trait measurements. X.Z. provided statistical analysis. X.Z., W.B. and F.L. interpreted the data. X.Z. wrote the manuscript with the help of all authors.

Data availability statement

All study data are included in the article and/or Supplementary data, and all raw data are available from the corresponding author upon reasonable request.