Adaptation strategy of karst forests: Evidence from the community‐weighted mean of plant functional traits

Abstract Conservative survival strategy of plants growing in harsh karst habitats is observed from the view of plant functional traits, such as morphological traits and ecological stoichiometry. However, whether the plant communities in karst forests with high species turnover adopt a conservative strategy remains undetermined. This study comprehensively investigated the characteristics of functional traits of dominant plant species in four forests (i.e. Platycarya strobilacea, Quercus fabri, Quercus variabilis, and Pinus massoniana forests) in a trough‐valley karst watershed in Northern Guizhou Province, Southwestern China to explore the adaptation strategy of karst forests at the community level. At the organ and the species levels, traits differed among species, and the leaf and the bark morphological traits and root C:N:P ecological stoichiometry presented large interspecific variations. At the community level, the P. massoniana forest presented the lowest specific root length and dry matter content and tissue density of roots, branch, twig, and bark; the Q. fabri and the Q. variabilis forests displayed low specific leaf area and high dry matter content and tissue density of roots, branch, and twig; and the Platycarya strobilacea forest exhibited high specific leaf area. The P. massoniana forest was subjected to N and P colimitation, and the three other broad‐leaved forests were limited by P supply. The community‐weighted means rather than the arithmetic means of traits were preferential to represent the trait characteristics at the community level. From the view of plant functional traits at the community level, karst forests develop multiple functional traits like low specific leaf area, high dry matter content and tissue density of leaf, roots, branch, and twig, and decrease N and P investments in leaf for a conservative survival strategy to adapt to harsh habitats.


| INTRODUC TI ON
Plant functional traits (PFTs) are the inherently physiological and externally morphological characteristics highly related to the ecesis, survival, growth, and death processes of plants (Violle et al., 2007).
Most PFT studies worldwide focus on the organ and the species levels, whereas PFT studies conducted at the community and the ecosystem levels are often underpowered Zhang et al., 2018). Furthermore, the arithmetic means of several dominant species are used to represent the community trait values. Such data analysis may bring about remarkable uncertainties, and results may not reflect the traits of a plant community. Natural plant communities are composed of species adapted to certain environments, and different species play different roles in community assembly and function exertion (Grime, 1998;Huston, 1997). Arithmetic mean trait values evidently fail to consider the complexity of species composition, community structures, and functions in complex natural plant communities Muscarella & Uriarte, 2016;Wright et al., 2004). Besides, no criterion is available in the selection of dominant species and individuals, for example, the number of species that should be chosen. Thus, PFT investigations that consider species composition, community structures, and functions at the plant community level must be conducted.
Karst, an extremely unique geomorphology that has resulted from the solvation of carbonatite (limestone and dolomite) is sporadically ubiquitous in the global land area but widespread around the southern United States, Mediterranean coasts of Europe, and Southwestern China (Jiang et al., 2014). In Southwestern China, vegetation degradation happens everywhere due to the fragility of karst ecosystems and intensive human disturbances. The forest restoration of degraded vegetation has become an environmental topic in karst regions. PFTs and the trait-based community ecology theory (a theory using trait-based approaches to determine community composition, structures, and functions) can reveal the adaptation strategies of vegetation in different restoration stages and environmental habitats and evaluate the restoration effects of different modes (Hedberg et al., 2013;Lavorel & Garnier, 2002;Pywell et al., 2003;Roberts et al., 2010;Sandel et al., 2011). Existing research on PFTs in Southwestern China indicates that plants grow in a plateau-surface, peak-clum depression, and peak-forest plain karst morphological terrains with harsh habitats (e.g., high temperature, water shortage, and shallow soils) exhibit low leaf area (LA), specific leaf area (SLA), and fine root-specific length (SRL), high leaf dry matter content (LDMC), and leaf tissue density (LTD). Plant growth is limited by N and P supply, and the interspecific variations of PFTs are generally large (Jiang et al., 2016;Liu et al., 2014Liu et al., , 2015Liu et al., , 2019Pang et al., 2019;Pi et al., 2017;Yang et al., 2020;Zhong et al., 2018). As a result, the conservative survival strategy with low growth rate and high resource utilization of karst plants is commonly observed (Tang et al., 2016).
However, most previous PFT studies in karst areas focus on leaf traits, and traits of other organs (root, branch, trunk, and bark) are rarely reported Yang et al., 2020;Zhong et al., 2018 Zhong et al., 2018). Zhang et al. (2018) have also found that the CWM (calculated on the basis of the relative biomass) and the arithmetic mean values of C:N:P ecological stoichiometry in China's forests differ remarkably, and the former is better to represent the ecological stoichiometry at the community level. Therefore, the CWM of traits of the leaf together with other organs may reflect the community trait characteristics and reveal the adaptation strategy of karst plants at the community level.
In the present study, three natural secondary forests and an   (Table 1). Each woody plant with a diameter at breast height (D) ≥1 cm was recorded with species identity (botanical nomenclature was based on Chen, 1982Chen, -2004, D (measured using a diameter tape), height (measured using a telescopic rod and a steel tape), and canopy width (canopy projection width, measured using a steel tape). The total biomass of each individual was estimated using biomass allometric models (Table S1). The biomass of tree species with ≥15 individuals in each plot was estimated using their own biomass allometric models, and the biomass of other tree and shrub species was estimated using universal allometric models (Liu et al., 2020).

| Vegetation survey and biomass estimation
In each forest, the species chosen for PFT measurements accounted for not less than 90% of the total forest biomass. According to biomass distribution patterns among species in the four karst forests, nine species, that is, P. strobilacea (accounting for 60.92% of the forest biomass), P. massoniana (16.65%), Albizia kalkora (7.52%), and Platycladus. orientalis (7.13%) in P. strobilacea forest; Q. fabri (57.55%), Quercus acutissima (24.22%), Camellia japonica (5.23%), and P. massoniana (3.56%) in Q. fabri forest; Q. variabilis (95.09%) in Q. variabilis forest and P. massoniana (86.20%) and Lindera glauca (3.80%) in P. massoniana forest, were chosen. Fresh masses of leaf, root, branch, twig, and bark samples were weighed using an electronic balance (accurate to 0.001 g). Bark thickness (BaT, mm) and LT (mm) values were measured using an electronic Vernier caliper (accurate to 0.01 mm). The LA, fine root length, and volume were scanned using the WinFOLIA multipurpose leaf area meter (Regent Instruments, Canada) (Yang et al., 2020;Zhong et al., 2018). The volumes of coarse and medium roots, branch, twig, and bark samples were determined using the drainage method, and those of leaf samples were obtained as the product of LA and LT (Cornelissen et al., 2003). All samples were dried at 85°C

| Measurement of morphological traits
for 72 h in an oven to determine their dry masses. The values of morphological traits were calculated as shown in Table S2.

| Determination of elemental contents
After morphological trait measurements, 5 leaves, 5 roots (mixed with coarse, medium, and fine roots), and 5 branch samples of each species were selected. All plant samples were powdered and sieved through a 0.2 mm sieve. The contents of total C (TC) and total N (TN) of the leaf (LC and LN), root (RC and RN), and branch (BrC and BrN) were determined using the Vario MACRO Cube (Thermo Scientific, Germany), and those of total P (TP) of the leaf (LP), root (RP), and branch (BrP) were determined using the iCAP 6300 ICP-OES Spectrometer Analyzer (Thermo Scientific, USA).

| Data analysis
In accordance with empirical studies (He et al., 2019;Zhang et al., 2018), the relative biomass (i.e., the biomass of one species as a percentage of the total forest biomass in each plot) was used to extrapolate PFTs from the species level to the community level. The CWM of a single trait was treated as the average trait value in the community, and was calculated using the following equation: where CWM x is the CWM for trait x; s is the number of species, which accounts for not less than 90% of the total biomass in the forest Bi × ti ∕Bs, community; B i is the relative biomass of the ith species in the forest community; t i is the trait value for the ith species, and B s is the biomass percentage of the chosen species in the forest community.
The coefficients of interspecific variation (standard deviation

P. massoniana
In general, the morphological traits of leaf and bark presented large interspecific variations as shown by large coefficients of interspecific variation; and those of roots, branch, and twig showed small interspecific variations as indicated by small coefficients of interspecific variation ( Table 2). The maximum coefficient of interspecific variation was BaT (96.89%. Table 2). SLA (60.05%) and LT (50.23%) also presented relatively large coefficients of interspecific variation (Table 2). CRDMC presented the minimum coefficient of interspecific variation (9.52%, Table 2).

| Ecological stoichiometry of plant species
Ecological stoichiometry differed among plant species ( Figure 2B,  presented intermediate ecological stoichiometry (Table 3).

| CWM of plant functional traits
At the community level, the P. massoniana forest presented low SRL, DMC, and TD of roots, branch, twig, and bark, LN and LN/P ( Figure 3, Table S3). Among the three broad-leaved forests, the P.  Table S3). Otherwise, the CWM of PFT values were preferential to represent the trait characteristics at the community level ( Figure 4, Table S3). Abbreviatins: LT, leaf thickness; LTD, leaf tissue density; LDMC, leaf dry-matter content; SLA, specific leaf area; CRTD, coarse root tissue density; CRDMC, coarse root dry-matter content; MRTD, medium root tissue density; MRDMC, medium root dry-matter content; FRTD, fine root tissue density; FRDMC, fine root dry-matter content; SRL, fine root-specific length; BrTD, branch tissue density; BrDMC, branch dry-matter content; TTD, twig tissue density; TDMC, twig dry-matter content; BaT, bark thickness; BaTD, bark tissue density; BaDMC, bark dry-matter content.   Xi et al., 2011). The characteristics of root, stem, branch, and twig traits are rarely investigated Pi et al., 2017;Yang et al., 2020;Zhong et al., 2018), values found in previous studies (Jiang et al., 2016;Liu et al., 2014Liu et al., , 2015Xi et al., 2011). Leaf traits present large and branch and twig traits present small interspecific variations in previous studies and the present study. Interspecific variations in root traits often exhibit high uncertainties due to complex and diverse belowground habitats (Comas & Eissenstat, 2004;Westoby & Wright, 2006). In the present study, the morphological traits of roots display small interspecific variations compared to those of leaf, whereas the root C:N:P ecological stoichiometry displays large interspecific variations compared to leaf and branch C:N:P ecological stoichiometry. Minimal attention has been paid to bark traits. We have investigated the BaT, BaTD, and BaDMC of species in karst vegetation and found that bark traits present large interspecific variations compared to roots, branch, and twig traits.

TA B L E 3 Ecological stoichiometry (mean
Overall, most PFT studies worldwide focus on several dominant or model species and ignore the complex species composition and the community structure in natural plant communities. Thus, whether the conclusions derived from such studies are applicable to complex natural plant communities remains to be verified Wright et al., 2004). The connection of individual-level PFTs with community structures, processes, and functions becomes a hot and difficult topic in this research field (Kunstler et al., 2016;Reichstein et al., 2014). In recent years, some plant ecologists F I G U R E 3 PCA showing the distribution of the morphological traits (a) and ecological stoichiometry (b) among different types of karst forest in Northern Guizhou Province, Southwestern China. Axis1 accounted for 78.57% (a) or 72.10% (b) of the variables, and Axis2 accounted for 14.48% (a) or 26.66% (b) of the variables. LT, leaf thickness; LTD, leaf tissue density; LDMC, leaf dry-matter content; SLA, specific leaf area; CRTD, coarse root tissue density; CRDMC, coarse root dry-matter content; MRTD, medium root tissue density; MRDMC, medium root dry-matter content; FRTD, fine root tissue density; FRDMC, fine root dry-matter content; SRL, fine root specific length; BrTD, branch tissue density; BrDMC, branch dry-matter content; TTD, twig tissue density; TDMC, twig dry-matter content; BaT, bark thickness; BaTD, bark tissue density; BaDMC, bark dry-matter content; LC, leaf total carbon content; LN, leaf total nitrogen content; LP, leaf total phosphorus content; LC/N, leaf carbon-nitrogen ratio; LC/P, leaf carbon-phosphorus ratio; LN/P, leaf nitrogen-phosphorus ratio; RC, root total carbon content; RN, root total nitrogen content; RP, root total phosphorus content; RC/N, root carbon-nitrogen ratio; RC/P, root carbon-phosphorus ratio; RN/P, root nitrogen-phosphorus ratio; BrC, branch total carbon content; BrN, branch total nitrogen content; BrP, branch total phosphorus content; BrC/N, branch carbon-nitrogen ratio; BrC/P, branch carbon-phosphorus ratio; BrN/P, branch nitrogenphosphorus ratio successfully extrapolated PFT characteristics from the organ and the species levels to community and ecosystem levels on the basis of relative biomass or individual number (especially the former) of species in a plant community (Ali et al., 2017;Zhang et al., 2018).
Karst forests are known for their rich species composition and high interspecific variations in PFTs (compared to non-karst forests in the same climate zone). Thus, the direct use of the arithmetic mean traits to represent the community traits is inappropriate. In the present study, we have calculated the species biomass-weighted mean community traits and found that CWM traits are preferential to represent the traits at the community level, which are indicated by high biases between CWM and arithmetic mean traits (Figure 4, Table S3).
The four karst forests have slightly lower community LN contents (12.54-17.72 mg g −1 ) and significantly lower LP contents (0.82-0.85 mg g −1 ) than plants in China (LN: 18.6 mg g −1 ; LP: 1.21 mg g −1 ) and in the world (20.09 and 1.77 mg g −1 ), indicating that karst forests and plants present low LN and LP contents (especially the latter) (Han et al., 2005;Reich & Oleksyn, 2004). According to Koerselman and Meuleman (1996) and Tessier and Raynal (2003) The forest restoration of degraded vegetation, such as grasslands, tussocks, and shrublands created by intensive human disturbances, has become a formidable task in karst regions in Southwestern China, and the increases in the biodiversity and the C storage are often used to evaluate the restoration success Ni et al., 2015). PFTs and the trait-based community ecology theory provide another pathway to predict the success of restoration efforts and the prospects of local vegetation restoration (Hedberg et al., 2013;Lavorel & Garnier, 2002;Pywell et al., 2003;Roberts et al., 2010;Sandel et al., 2011). In the present study, the P. massoniana forest is a fast-growing forest, which can rapidly increase the local vegetation coverage and the C storage. The P. strobilacea forest presents relatively high SLA and low DMC and TD of roots, branch, and twig

| CON CLUS IONS
Overall, the CWMs rather than the arithmetic means of PFTs were preferential to represent the trait characteristics at the community

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
The data are available in the manuscript and supplementary materials.