Contrasting phylogenetic and functional trait responses in post-fire karst forests: Stochasticity and environmental filtering.

High-severity fires fundamentally alter community assembly trajectories, a process particularly critical in fragile karst ecosystems. Elucidating the mechanisms driving early post-fire assembly is vital for identifying constraints on ecological recovery. Here, we investigated plant community assembly in burned karst sites in central Guizhou Province, China, surveying vegetation 0.5-5 years following a high-severity forest fire. By integrating null models, Mantel tests, and random forest analyses, we quantified shifts in community characteristics and the drivers governing assembly processes. The results show that high-severity fire exerts a persistent suppressive effect on species composition. After five years, communities still differed significantly from unburned reference communities in taxonomic and functional composition. Ecological strategies shifted towards more conservative syndromes, whereas phylogenetic and functional recovery were not fully concordant. Stochastic processes, including homogenizing dispersal and ecological drift, accounted for more than 93% of inferred phylogenetic assembly processes, while several functional traits were significantly clustered, consistent with environmental filtering. These traits had Blomberg's K < 1, indicating weaker phylogenetic signal. Soil pH, phosphorus (P), and potassium (K) were among the abiotic variables most strongly associated with assembly patterns. These findings highlight the multidimensional and process-divergent nature of post-fire assembly in karst ecosystems, offering empirical guidance for targeted restoration and long-term resilience planning.