Threshold-dependent plant-microbe interactions shape soil microbial life-history strategies and stability across an elevational gradient

Despite growing recognition of nonlinear plant–microbe relationships, the ecological thresholds at which plant multidimensional diversity triggers shifts in microbial life-history strategies and community stability remain poorly understood. Along a subalpine elevational gradient, we integrated plant taxonomic, functional, and phylogenetic diversity (PD) with soil microbial sequencing and employed ensemble machine learning and generalized additive mixed models (GAMMs) to quantify nonlinear and threshold responses. Bacterial communities were dominated by K-strategists (oligotrophs) and exhibited unimodal stability patterns across elevation, with both life-history strategies and community stability primarily regulated by plant functional or PD. In contrast, fungal communities displayed a gradual but incomplete shift toward K-strategists at higher elevations, while community stability increased with elevation. Fungal life-history strategies were predominantly structured by elevation-associated environmental filtering, whereas their stability was positively associated with plant functional diversity, indicating a partial decoupling between compositional strategies and temporal stability. Both microbial domains displayed pronounced nonlinear and threshold-dependent responses. Intermediate levels of plant functional attributes promoted bacterial K-strategist dominance and stability, whereas extreme trait values or nutrient enrichment reduced stability. Fungal responses were characterized by elevation-driven threshold transitions consistent with strong abiotic filtering. Variance partitioning further revealed asymmetric controls: Plant multidimensional diversity explained more variation in bacterial strategies and stability, whereas fungal strategies were more strongly constrained by abiotic factors. Together, these findings demonstrate that soil microbial assembly along environmental gradients is governed by domain-specific and threshold-driven mechanisms rather than continuous linear shifts, with important implications for predicting belowground ecosystem responses to global change.