LAUSR

Abstract The recently described trait‐based approach is becoming widely popular for a mechanistic understanding of species coexistence. However, the greatest challenge in functional analyses is decomposing the contributions of different ecological and evolutionary processes (e.g., niche‐based process, neutral process, and evolutionary process) in determining trait structure. Taking rodents (Rodentia) in the Hengduan Mountains as our study model, we aim to (1) quantify the vertical patterns of functional structure for head–body length (HL), tail/body ratio (TR), animal component in diet (ACD), and all traits; (2) disentangle the relative importance of different assembly processes (environment, space, and phylogeny) in structuring trait dispersion; and (3) assess the feasibility of Bergmann's rule and Allen's rule along elevational gradient. Our results have suggested that the vertical functional structure pattern varied across these three traits, indicating distinct functional roles in the community assembly process. These nonrandom vertical patterns of HL, TR, and terminal ACD have demonstrated these traits were dominated by different ecological process along environmental gradient. In variance partitioning, high proportion of the spatial variations in trait dispersion was explained by environmental and spatial models, which have provided supporting strong evidence for niche‐based and neutral processes in leading species coexistence. Although the three traits all exhibited apparent phylogenetic signals, phylogenetic relationship within community failed to predict the spatial variations of functional dispersion, confirming the enormous inference of phylogenetic signals in predicting trait structure. By assessing the vertical patterns of HL and TR at order and family levels, we argued that functional adaptation along an environmental gradient is a surrogate of series of complex processes (e.g., environmental filtering, interspecific interaction, and neutral dispersal) acting on multiple functional axes, which results in inconsistence with the empirical rules along elevational gradient.