Article Variation and Tradeoff of Leaf Traits of Karst Shrub in Southwest China

Long-term droughts were found to have guided the environmental selection ofshrub plant characteristics in a karst region of China, as the plants were found to have developed a set of leaf trait combinations,includinga small specic leaf area (SLA), leaf area (LA), and large leaf dry matter content (LDMC), that are known to be suitable for drought environments.Leaf traits of plants are not only the intuitive and operable taxonomic traits in plant taxonomy, but also reect the responses and adaptations of plants to their habitats. This is helpful when trying to understand the role of environmental screening and when ltering plant functional traits. The objective of this investigation was to determine the leaf trait variations, adaptations, and patterns in the shrubs from a karst region in China.Weinvestigated 11 leaf traits to quantify the variations in their trade-offs and the trait–habitat /species relationships for the shrubs at theHuanjiang karst ecosystem observation and research station, China, using multivariate analyses.There were signicant intraspecic and interspecic changes in the leaf traits ofthe shrub plants, and there were differences among the traits. Except for carbonmass, nitrogenarea, and phosphorousarea, the interspecic variations of the leaf traits were generally higher than the interspecic variation. The correlation between the leaf traits in the karst shrubs was also signicant. Species differences had a higher explanatory degree for the leaf traits than topography or soil nutrients. The ndings of this study will enhance our understanding of the variations in leaf traits in the karst shrubregions and the adaptative strategies of the plants in degraded habitats.Furthermore, these results may provide scientic information to help guide vegetation recovery programs in the karst region of southwest China.


Introduction
Plant traits refer to a series of core plant properties that are closely related to plant colonization, survival, growth, and death. They can signi cantly affect ecosystem functions and re ect the responses of vegetation to environmental changes (Wright et al., 2004(Wright et al., , 2007. Compared with most studies that are based on plant classi cations and quanti cations, plant traits have become a reliable focus when aiming to solve important ecological problems in populations at community and ecosystem scales (He et al., 2018). In recent years, research on the distribution patterns of plant traits has mainly focused on global and regional scales. At a global scale, many researchers have studied leaf morphology, structure, and physiological traits, nitrogen and phosphorus content in leaves, seed weight, tall tree, and trunk characteristics, and furthermore the famous global plant leaf economic spectrum and tree trunk economic spectrums have been developed (Wright et al., 2004(Wright et al., , 2017Reich et al., 2004Reich et al., , 2014. At a regional scale, investigations have reported the traits and functions of forest ecosystems along a 3700 km gradient in the south-north sample zone of eastern China, and developed a set of methods to scienti cally deduce the traits of organ level measurements at the community scale (Tan et al., 2017).
Trade-offs are a common balance among plants traits, and is a combination of traits formed after natural selection. In other words, species are arranged in adaptive or competitive positions along a certain ecological strategy axis (Wright et al., 2004). These trade-offs not only include those between the aboveground and underground traits of plants (Yao et al., 2010;Yang et al., 2019), but also the trade-offs between traits, such as the leaf and branch and trunk traits, leaf traits, reproductive traits and quantity, and reproductive traits and seedling leaves (He et al., 2018;Messier et al.,2017). The trade-offs between plant traits not only help to understand the differences of plant ecological strategies in different environments, but also explore the internal mechanisms of niche differentiation and species coexistence (Chave et al., 2009).
Plant traits determine the growth, reproduction, and survival of plants and play an important role in the distribution patterns of species along environmental gradients. The relationships between plant traits and the environment, are not only conducive to the study of ecosystem functions and the coexistence mechanisms of community species, but also for predicting the effects of global climate change on plant distributions . Many studies have shown that the life strategies of plants formed by different morphological and physiological traits among species, are re ected in their rapid resource acquisitions and high resource conservation, in different ecosystems and biological communities (Terashima et al., 2011). The distribution of plant traits at different scales, is the result of multi-factor ltrations from a large to small scale. Many studies have con rmed that climate factors, such as temperature, light, and precipitation, play a decisive role in the distribution of plant traits on a global or regional scale . At a medium scale, land use and disturbance play a major role in plant traits (Ocheltree et al., 2012). However, at a small or local scale, the distribution of plant traits is determined by geomorphic and soil factors (Li et al., 2014). Compared with climatic and geographical factors, studies on the tradeoffs between habitats and plant traits, have been more concerned with degraded than non-degraded ecosystems.
The Karst region in the southwest of China is approximately 540 000 km 2 , its primary forest is an evergreen broad-leaved forest and there is a seasonal rain forest. However, due to human disturbances, most of the evergreen and deciduous broad-leaved mixed forest and seasonal rainforest in the region of the limestone and dolomite have been degraded into secondary forests and shrubs (Nie et al., 2012). In recent years, with the implementation of major ecological projects such as the Grain for Green Program (GGP) by the government of the People's Republic of China, there has been a reduction in stony deserti cation and an increase in the restoration of vegetation . However, the loss of soil nutrients and soil water in the karst region is an ongoing problem because of crop cultivation by local famers (Du et al., 2015). At the same time, the rate of shrub changing successions to forest is becoming low, due to the serious seasonal droughts and the supply constraints of the soil nutrients (Tan et al., 2017). In addition, in the shrub communities in some regions, there is a serious leakage of soil nutrients and water, that is degrading them to grass communities or even bare rock deserti cation. This makes it

Research site
The research site was located in Mulian village (24.43°N-24.45°N, 108.18°E-108.20°E, 220 m above sea level), Dacai town, Huanjiang county, west of Guangxi province, which is the location of the Huanjiang karst ecosystem observation and research station, Institute of Subtropical Agriculture, Chinese Academy of Sciences (Fig. 1). According to the records of the weather station in Huanjiang county, from 1961 to 2019, the annual average temperature was 19.3 ℃, the average temperature in January was 10.1 ℃, the average temperature in July was 28.0 ℃, and the average annual precipitation was 1750 mm, and the average annual sunshine time was 4422 h. The mother rock is limestone, and the soil is dominated by dark or brown calcareous soil with developed carbonate rocks. Karst is mainly distributed in the southwest of the county. The soil layer is shallow and the slope is large, which means that soil erosion is a serious issue. Furthermore, there is a serious level of rock exposure, resulting in a severe tendency for rocky deserti cation. The representative vegetations of the study area were grasses, shrubs, and secondary forests.

Field survey
From July to September 2019, thirty 10 × 10 m sample plots were set up, according to the different terrains and shrubs, and the diameter at breast height (DBH), height, and crown width of all the woody plants with DBH ≥ 1 cm in the quadrat were investigated and the species, quantity, height, and growth status of the shrubs and herbs were recorded. At the same time, the global positioning system (GPS) (E640 + MobileMapper) was used to record the longitude, latitude, altitude, and other geographic information of the inner center of each sample square, to investigate and record the slope, slope direction, slope position, rock exposure rate, and soil thickness. Five soil samples per plot were also collected from the surface soil (0-20 cm) according to the plum-ower pattern, and the samples were fully mixed to form the sample to be measured for soil nutrients. In this research, the soil pH, soil organic carbon (SOC), total nitrogen (TN), total phosphorus (TP), total potassium (TK), available nitrogen (AN), available phosphorus (AP), and available potassium (AK) were measured using the methods previously described by Bao (2000).

Measurements of the leaf traits
Twenty-three dominant shrubs species were selected from the research plots, with 10-15 plants of each species. In the outer four directions of each shrub crown, 4 complete branches without pests, diseases, or epiphytes, were collected under full light, and 5-6 leaves were collected from each branch, resulting in 20-24 leaves being collected from each shrub. A multi-purpose leaf area meter (Regent Instruments, Quebec, Canada) was used to scan each leaf and obtain its area, aspect ratio, and other indexes. The leaves were put into the oven, killed at 105 ℃ for 20 min, dried at 85 ℃ for 48 h, and then the dry mass was measured. The carbon, TN, TP, and TK were also measured in the laboratory. The leaf traits examined were as follows: leaf area (LA, m 2 ), speci c leaf area (SLA, m 2 /kg), leaf mass per unit area (LMA, kg/m 2 ), leaf dry matter content (LDMC, mg/g), leaf carbon content (C mass, g/kg), leaf nitrogen content (N mass, g/kg), leaf phosphorus content (P mass, g/kg), and leaf potassium content (K mass, g/kg).

Statistical analysis
The mean and standard deviations for each trait were calculated, and the differences of each trait between the different species were compared using the independent sample t test. Coe cients of variation (CV) were used to calculate the degrees of variation for each trait. The Pearson correlation test was used to analyze the functional correlations both between and within species. Using principal component analysis (PCA), the covariance matrix among the traits was determined after traits had been log-transformed. The effects of the topography, soil nutrients, and species on the trait variations were examined by one-way analysis of variance. The general linear model (GLM) was used to detect the effects of single factors, double factors, and multi-factor interactions. All statistical analyses were calculated using SPSS 17 software.
There were signi cant intraspeci c and interspeci c changes in the leaf traits of the shrub plants in the karst areas, and there were differences among the different traits (Fig. 2). From the average value of the species, the intraspeci c variation of the N area and K area was large, while that of the C mass was the smallest. LA had the largest interspeci c variation, while the P area had the smallest interspeci c variation. Except for the C mass, N area, and P area, interspeci c variations of the leaf traits were generally higher than interspeci c variations.
3.2. Trade-offs in the leaf traits of shrub plants There were signi cant positive or negative correlations among the leaf traits ( Fig. 3; Table 2). LA was signi cantly correlated with SLA, P mass, and K mass. Except for C mass and K mass, the SLA was signi cantly correlated with the other traits. Signi cant correlations were also detected between LMA and LDMC, P mass, P area, and K area. The LDMC was only signi cantly correlated with the P area. There were signi cant autocorrelations between the N mass, P mass, K mass, and N area ( Table 2). The PCA of the traits showed six independent axes of leaf trait variation ( Table 3). The rst three principal components collectively accounted for 70.00% of the total variation, whereas the other three principal components accounted for a further 24.32%. The rst axes, which was predominantly related to SLA and P mass, together accounted for 36.64% of the total variation, whereas axes 2 and 3 together accounted for 33.36% of the total trait variation, indicating that LDMC, K mass, and K area dominated these axes.
Axis 4, which accounted for 9.37% of all trait variation, was dominated by LA, C mass, N mass, and N area, whereas axes 5 and 6 were dominated by C mass and LA, respectively.  Table 1 for the meaning of the abbreviated terms.

Multi-factorial control of leaf trait variation
Factor analysis of the variance showed that species differences had a high explanatory degree for leaf traits ( Table 4). The R 2 value between species and leaf traits ranged from 0.32 to 0.86. Among these traits, the species highly explained the variation of LA, P mass, K mass, and K area. The effects of topography on the variations of the leaf traits was small, but the effects on the K area were relatively large (Table 4). Similarly, the different soil nutrients had small effects on the variation of leaf traits. The interactions between the topography and soil nutrients only had a small effect on most trait variation than that of topography and soil nutrients separately (Table 4).

Variations in the leaf traits and its control factors
In this study, it was found that 11 leaf traits from the 10 shrub plants in the karst area had different degrees of variation within and between the species (Fig. 2). Previous studies have shown that interspeci c variations play a dominant role in the variation of plant functional traits, but increasing evidence has shown that intraspeci c variation is not negligible (Albert et al., 2010;Jiang et al., 2016). The average intraspeci c variation of the leaf traits in this research was 23%, which was like that of previous investigations (Zhong et al., 2018). The range of intra-specialty variation of the karst plant traits was also lower than that of the non-karst plants, which may re ect the small levels of morphological plasticity in the harsh habitat conditions (Auger et al., 2013).
The interspeci c variations of the plant traits differed with the different environmental conditions (Zhong et al., 2018). Our data showed that the range of interspeci c variations for the 11 traits ranged from 2.9-121.3%, and the variations for the LA were signi cantly higher than those for the other traits. Plant traits are jointly determined by genetic factors and environmental conditions (Yao et al., 2010). Among species with different genetic backgrounds or different taxon, the leaf traits between the species vary more than those of a branch, which indicated that the performance of the leaf was relatively unstable among species.
On the other hand, variations of the leaf and root traits also varied with the habitat conditions, to adapt to their living environment (Yang et al., 2019). In this research, we also found that the leaf traits were affected by the topography and soil nutrients. However, the extent to which environmental factors had an effect was less than that of species groups (genetic background) (Table 4)

Co-variation of different leaf traits
In the process of plant growth and long-term adaptation to the environment, the traits showed a certain correlation with the comprehensive action of physiological, phylogenetic, and environmental factors, and nally formed a series of optimal combinations of functional traits adapted to speci c environments (Kerkhoff et al., 2006). In this study, it was found that the correlations between leaf traits in the karst shrubs was signi cant. For examples, we detected signi cant correlations between the SLA and LA at the species level (Table 2), which was consistent with the ndings of previous but recent investigations (Yang et al., 2019;Cornelissen et al., 2003).
LMA and LDMC can both represent the utilization of environmental resources by plants to a certain extent, and are closely related to a plants adaptation strategies to the environment, and can re ect the adaptation characteristics of plants to different habitats (Cornelissen et al., 2003).The positive correlation between the two has been universally con rmed (Wilson et al., 1999). The results of this study were also in agreement with previous research ( Table 2). In general, species distributed along the LMA-LDMC axis (the regression line between LMA and LDMC) have different utilization methods of the environmental resources, especially those distributed at both ends of the resource utilization axis. Plants with high LMA and LDMC usually live in poor environmental conditions (for example, drier or colder), and have strong resistances to oppression . They maintain their growth and development by accumulating captured resources. Plants with low LMA and LDMC, on the contrary, generally live in a superior environment and have high production capacity, but poor resistance to harsh environments .
In addition, the nutrient content of the leaves is related to the utilization of plant resources or the strategies of plant adaptation to the environment. Generally, the photosynthetic capacity and respiratory consumption of plants with high nutrient content in leaves, especially those with high N content, are usually strong, and they adapt to the environment through rapid nutrient cycling. Those with low nutrient content in their leaves, with low photosynthetic capacities, will survive through rapid nutrient cycling (Wright et al., 2004(Wright et al., , 2007Feng et al., 2010). Here, we found that there was no signi cant correlation between LMA, LDMC, and the nutrient contents in a single area (except for P mass). With the increase of plant LMA and LDMC, the nutrient content per unit area constant, which is not consisted with some previous research (Jiang et al., 2016;Zhong et al., 2018). Previous studies indicated that when the N and P areas are constant, the N and P masses will decrease with increasing leaf thickness (LMA and LDMC), but this decrease will be offset by the increases of N and P areas. from the increase of the LMA and LDMC. Therefore, N mass and P mass have no signi cant relationship with LMA and LDMC . Here, our data showed that N mass was not signi cantly related with LMA and LDMC, while the P mass not (Table 2) The eld site in this research. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors. The eld site in this research. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.

Figure 2
Coe cient of interspeci c/intraspeci c variations of the leaf traits in the karst shrubs. Note: See Table 1 for the meaning of the abbreviated terms.

Figure 2
Coe cient of interspeci c/intraspeci c variations of the leaf traits in the karst shrubs. Note: See Table 1 for the meaning of the abbreviated terms. Color map of the correlations (r, left) and correlation probability (p, right) between leaf traits. Note: See Table 1 for the meaning of the abbreviated terms. Color map of the correlations (r, left) and correlation probability (p, right) between leaf traits. Note: See Table 1 for the meaning of the abbreviated terms.  Trait dimensions from the rst four principal component (PC) analysis. Note: See Table 1 for the meaning of the abbreviated terms. Trait dimensions from the rst four principal component (PC) analysis. Note: See Table 1 for the meaning of the abbreviated terms.