Leaf Functional Trait Variation and Environmental Filtering across Hierarchical Levels in Complex Karst Peak–Depression Landscapes

The karst peak–cluster depression landscape is characterized by pronounced topographic and edaphic heterogeneity, resulting in a high degree of spatial variability in resource availability. Such a complex environmental mosaic influences plant functional traits and their adaptive strategies. However, the patterns of trait variation and their underlying environmental drivers in karst peak–cluster depressions remain poorly understood. In this study, we investigated a 15-ha forest dynamics plot in the Nonggang Nature Reserve, Guangxi. We used linear mixed models (LMMs) to partition variance, principal component analysis (PCA) to examine trait covariation, and redundancy analysis (RDA) combined with hierarchical partitioning to evaluate the effects of topographic and edaphic factors on leaf functional traits at the community, species, and intraspecific levels. The results showed that trait variation was mainly driven by interspecific differences, with part of the variation attributable to evolutionary history. Morphological traits and physiological traits exhibited relatively higher intraspecific and interspecific variation, respectively. The trait covariation patterns revealed that several leaf chemical traits exhibited atypical modes of coordinated variation. The explanatory power of topographic and edaphic factors for trait variation differed among ecological levels, being highest at the species level, followed by the community and intraspecific levels. Among these, the soil carbon-to-nitrogen ratio and slope were identified as the main drivers of species-level variation. Overall, our findings reveal functional shifts in plant ecological strategies in complex karst landscapes and emphasize the differential influence of topographic-soil factors across hierarchical levels, providing empirical evidence for understanding plant adaptive mechanisms along multidimensional environmental gradients.