Increase in leaf organic acids to enhance adaptability of dominant plant species in karst habitats

Abstract Estimation of leaf nutrient composition of dominant plant species from contrasting habitats (i.e., karst and nonkarst forests) provides an opportunity to understand how plants are adapted to karst habitats from the perspective of leaf traits. Here, we measured leaf traits—specific leaf area (SLA), concentrations of total carbon ([TC]), nitrogen ([TN]), phosphorus ([TP]), calcium ([Ca]), magnesium ([Mg]), manganese ([Mn]), minerals ([Min]), soluble sugars, soluble phenolics, lipids, and organic acids ([OA])—and calculated water‐use efficiency (WUE), construction costs (CC), and N/P ratios, and searched for correlations between these traits of 18 abundant plant species in karst and nonkarst forests in southwestern China. Variation in leaf traits within and across the abundant species was both divergent and convergent. Leaf [TC], [Ca], [Min], [OA], and CC were habitat‐dependent, while the others were not habitat‐ but species‐specific. The correlations among [TN], [TP], SLA, [TC], CC, [Min], WUE, [OA], and CC were habitat‐independent, and inherently associated with plant growth and carbon allocation; those between [CC] and [Lip], between [Ca] and [Mg], and between [Mg] and [WUE] were habitat‐dependent. Habitat significantly affected leaf [Ca] and thus indirectly affected leaf [OA], [Min], and CC. Our results indicate that plants may regulate leaf [Ca] to moderate levels via adjusting leaf [OA] under both high and low soil Ca availability, and offer new insights into the abundance of common plant species in contrasting habitats.


| INTRODUC TI ON
Karst is a unique ecological system, which is defined as a landscape formed by dissolution of soluble rocks with rocky soils, caves, sinkholes, and lacking surface stream (Geekiyanage et al., 2019;Williams, 2008). Karst habitats are fragile and vulnerable, with high concentrations of calcium ([Ca]) and high pH in their shallow soils (Wei et al., 2018). One of the largest karst ecosystems is located in subtropical mountainous regions of southern and southwestern China (Yuan, 1991), exhibiting remarkably high species richness and endemism, contributing significantly to the floristic diversity of China (Wei et al., 2018), due to the fine-scale heterogeneity of hydrogeology, topography, and associated water availability influenced by a monsoon climate (Guo et al., 2017). Being important hot spots of biodiversity and endemism (Clements et al., 2006), the karst ecosystems in China are under threats from human disturbance and global change which weaken their stability and accelerate rocky desertification (Lian et al., 2015;Tian et al., 2017).
A variety of plant functional traits have been considered to be related to dynamics of plant communities and functions of forest ecosystems (Boukili & Chazdon, 2017;He et al., 2019;Kunstler et al., 2016). Some plant traits are associated with factors that drive plant diversity and community assembly (Adler et al., 2013;Kunstler et al., 2016); similarity of leaf traits may increase competition among coexisting dominant tree species (Kraft et al., 2015). Many important leaf traits are used to assess plant adaptability and growth in different environments. For example, a high specific leaf area (SLA, Balachowski & Volaire, 2018;Hamann et al., 2018;Lambers & Poorter, 1992) and high leaf nutrient concentrations (Lambers & Poorter, 1992;Zhang et al., 2018) reflect a high capacity for plant growth. Some plant chemicals (secondary metabolites) are helpful to enhance adaptability to stressful environments. Soluble phenolics (SP, Karabourniotis et al., 2014) are related to plant defense under biotic stresses; soluble sugars (SS) serve as osmotic solutes to acclimate to water deficits (Galiano et al., 2017;Kuang et al., 2017); organic acids (OA) are important to sustain cellular functions under drought (Farooq et al., 2009).
Comparison of the differences in leaf traits and their effects on dominant plant species from contrasting habitats gives an opportunity to understand how dominant plants adapt to different habitats (Geekiyanage et al., 2018). Soil properties, such as soil pH values, water availability, and Ca concentrations, significantly differ between karst and nonkarst habitats (Hao et al., 2015), and these can substantially affect plant growth (Burstrom, 1968;Kinzel, 1983). However, there are a few plant species abundant in both karst and nonkarst forests in southwestern China (Zhu et al., 1998), despite species composition being notably different between the habitats (Fu et al., 2015). In this study, we measured  (Hao et al., 2015).

| Study site, species, and sampling
This study was conducted in Guizhou Province of southwestern China (103°36′-109°35′E, 24°37′-29°13′N), which has a typical karst distribution accounting for approximately 74% of its total area (Zhu, 1993). Characterized by a plateau monsoon humid climate, Guizhou has a mean annual temperature and a mean annual precipitation of 15.5°C and 1,400 mm, respectively, and has typical subtropical karst forests (Tian et al., 2017). Soils in karst forests in this province are generally developed from dolomite and/or limestone, with pH values varying from 6.3 to 7.8 (Wang et al., 2004), and soil [Ca] of 10.6 ± 6.3 mg/g (Zhang et al., 2014). In this study, the karst and nonkarst forests were uniformly selected, with soils developed from limestone and from granite, respectively.
Leaves of 18 common plant species abundant in both karst and nonkarst forests throughout this province (Huang et al., 1988) ( Table 1) were sampled for analysis. Each species was sampled from six forests (three karst and three nonkarst). A minimum of three mature trees per species were sampled per forest. At least 10 mature, fully expanded, and healthy leaves were collected per tree. To minimize the influence of tree age, the individual trees of the species were of similar accounts of growth rings, which was determined by tree core extracted using an increment borer (Ф 5.15 mm, Haglöf, Sweden), in both karst and nonkarst habitats. After sampling, the leaves were stored in ice bags and transported back to the laboratory.

| Leaf trait measurements
In the laboratory, the leaves were cleaned and divided into two parts. One part was used to measure leaf area (Li-COR LI-3000C, Inc., Lincoln, Nebraska, USA) and then dried to constant weight (65℃ for 72 hr) for SLA calculation. The other part was freeze-dried, ground, and then used for chemical analyses. Leaf [TC] and [TN] were measured with an elemental analyzer (Isoprime 100, Elementar Isoprime, South Manchester, UK). Leaf [TP] was determined via molybdenum-antimony colorimetry after digestion by sulfuric acid (Murphy & Riley, 1962 Poorter et al. (1997) and Blainski et al. (2013). Briefly, a part of leaf powder, about 1.0 g, was extracted with a mixture of chloroform:methanol:water (2:2:1; v:v:v). The chloroform phase was used to determine leaf [Lip] from the residue weighed after evaporation  (Poorter et al., 1997). The water/methanol phase was used to determine [SS] and [SP] using the anthrone and Folin-Ciocalteu method, respectively (Poorter et al., 1997). Leaf [lignin] was determined after chloroform:methanol:water extraction and 3% HCl extraction (Poorter et al., 1997).
were determined according to Cataldo et al. (1975). Another part of leaf powder, about 0.10 g, was combusted in a muffle furnace at 550°C for 6 hr. The ashes after combusted consist of minerals (all mineral nutrients in leaves), oxides (derived from OA), and nitrate (Poorter et al., 1997 , and ash alkalinity via the following equations: Leaf CC was calculated according to Poorter et al. (1997): Water-use efficiency of the plant species was calculated based on leaf δ 13 C values (Ehleringer & Cerling, 1995;Farquhar et al., 1982), which were determined using a mass spectrometer (Thermo Finnigan,

| Statistical analyses
All statistical analyses were conducted using R software (version 4.0.2). The units and key statistic summary of each leaf trait are provided in Table 2. Prior to multivariate analysis, the traits were checked for approximate normality (Shapiro-Wilk test). Those that did not follow normality were log 10 - ([TP] traits between the habitats were tested using general linear mixed effect models (GLMEMs), with species and individuals as the random effects (Crawley, 2007), and determined using Tukey's HSD post hoc tests and conducted by the lsmeans function in lsmeans package after processing GLMEMs (Lenth, 2016). The correlations between leaf traits across the habitats were estimated by the pc function and idaFast function in R package pcalg (Kalisch et al., 2012). Briefly, we used the pc function to estimate the equivalence class of a directed acyclic graph (DAG) based on the PC algorithm; then, we used the idaFast function to calculate the coefficient of each pathway in DAG. Pearson's coefficients (r) were calculated to test the correlations of the traits between the habitats. We tested the paths of effects of habitat on those traits that differed between habitats via structural equation model (SEM) by the sem function in lavaan package. Briefly, we built an a priori model using these leaf traits affected by habitat. After running the a priori model, all nonsignificant paths were removed (p > .05) and we ran this new model again. The ratio of chi-square to degrees of freedom (chi-square/DF, ≤ 2, p > .05), comparative fit index (CFI, ≥0.95), and root mean squared error of approximation (RMSEA, 0 ≤ RMSEA ≤ 0.05) were used to assess the goodness of the final model when chi-square/DF ≤2 (p > .05) (Schermelleh-Engel et al., 2003). The significance was set at p < .05.

| Variations of leaf traits
All the studied leaf traits were affected by species (Figure 1). The variations in the leaf traits were inconsistent among species (Figure 2).

| Effects of habitat on the correlations among leaf traits
The correlations among leaf traits were either habitat-independent or habitat-dependent for the common trees (Figure 5a

| Pathways via which habitat affected leaf trait variations
The effects of habitat on leaf [OA], [Min], and CC were indirect, via affecting leaf [Ca] (Figure 6a). In addition, leaf CC was decreased by both leaf [Ca], [Min], and [OA], but lower for species from karst than those from nonkarst habitats (Figure 6a, b). Furthermore, leaf

| Intraspecific and interspecific variations in leaf traits
A variety of leaf traits reflects adaptation of plants to a specific environment (Hazen et al., 2018), but some exhibit substantial phenotypic plasticity in many plants (Bjorkman et al., 2018;Russo & Kitajima, 2016 [Ca], and leaf CC (Figures 1 and 4), but species had significant effects on all leaf traits studied here (Figure 1). The inconsistency between habitats and across the common plants in this study (Figures 1, 2, and 4) segregates the importance of intraspecific (Kraft et al., 2014) and interspecific variations (Albert et al., 2011). Although the dissimilarity of leaf traits decreases competition among coexisting plant species (Kraft et al., 2015), the fine-scale diversity of hydrogeology, topography, and associated water availability influenced by a monsoon climate (Guo et al., 2017)  nonkarst: 18 ± 4.8, Geekiyanage et al., 2019), and moderate nutrient retrieval by plant growth in karst habitats (Liu et al., 2015). Part of our first aim of this study is to assess whether leaf traits are affected by habitat and species. We found that five leaf traits were significantly affected by habitat, while all leaf traits were affected by species.
There are three mechanisms that may explain the negative correlations between SLA and WUE: (a) CO 2 supply at sites of carboxylation may be decreased due to a longer internal CO 2 diffusion pathway in thicker leaves (Hultine & Marshall, 2000;Prieto et al., 2018); (b) densely packed mesophyll may reduce the conductance of mesophyll to CO 2 in thicker leaves (Prieto et al., 2018;Tomás et al., 2013); and (c) more enzymes related to photosynthesis in thicker leaves may increase the demand for CO 2 (Hultine & Marshall, 2000;Prieto et al., 2018). The negative correlations between leaf nutrients (N and P) and WUE may result from plants enhancing mass flow of nutrients by increasing transpiration and enhancing uptake of mobile nutrients, and plants with high leaf nutrient concentrations increasing stomatal conductance and photosynthetic activity (Field et al., 1983;Prieto et al., 2018). Although there were some habitatdependent correlations between leaf traits, for example, positive cor- Abbreviations of all leaf traits are provided in Table 2 F I G U R E 3 Correlations (Pearson's correlation coefficients, r) between the traits of 18 common tree species from karst (K) and nonkarst (NK) habitats. The abbreviations of the tree species are indicated in Table 1. Abbreviation of all leaf traits is provided in Table 2 | 10285 TANG eT Al.
ion balance and to decline the restriction of excess [Ca] on plant growth (Figure 7, White & Broadley, 2003), while those growing in nonkarst habitats need an amount of Ca to maintain normal physiological functions, for example, preventing an efflux of potassium and decreasing turgor (Bressan et al., 1998;Burstrom, 1968). Therefore, plants in nonkarst habitats might enhance Ca via increasing OA in leaves (Figure 7). The role of leaf OA regulating the level of Ca may allow species to dominate in both karst and nonkarst habitats.
F I G U R E 4 Differences in leaf traits of the dominant plant species between karst and nonkarst habitats. Different lowercase letters indicate significant differences between habitats based on linear mixed effect models (post hoc Tukey test, p < .05). The absence of lowercase letters indicates that the effect of habitat was not significant. Boxes in each boxplot show the first and third quartiles and the median; the upper and lower whiskers indicate the largest and smallest values away from 1.5*IQR (interquartile range) of the third quartiles and first quartiles, respectively; black points in each figure are values that fell outside the whiskers. Abbreviation of all leaf traits is provided in Table 2 F I G U R E 5 Correlations of the studied leaf traits derived from the idaFast function in karst (a) and nonkarst (b)

| High leaf [OA] is a consequence of high leaf [Ca] in karst
Calcium is a plant macronutrient, while abundant Ca has adverse effects, for example, affecting ion uptake by roots (Kinzel, 1983). We found that the effect of habitat on leaf traits was associated with the differences in leaf [Ca] between the habitats (Figure 6 Poorter & Bergkotte, 1992

| CON CLUS IONS
We quantitatively assessed the variations and causal effects of leaf traits of plant species common in both karst and nonkarst habitats.
We showed that the variations in leaf traits within and across the common plant species were both divergent and convergent between the habitats, and the correlations between leaf traits were either dependent or independent of habitat.

ACK N OWLED G M ENTS
This study was supported by the subproject of the National Major

Scientific Research Project of China (No. 2013CB956701) and Youth
Innovation Promotion Association of Chinese Academy of Sciences.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data are available from the Dryad Digital Repository https://doi. org/10.5061/dryad.x95x6 9pjc (Tang et al., 2021).