Elevational patterns of phylogenetic and morphological structure at local and regional spatial scales
Resulting from interactions between stochastic and deterministic processes in heterogeneous habitat, phylogenetic and trait-based structure of most assemblages exhibited random dispersion in two datasets. However, significant linear patterns of SES.MPD, SES.PWsize, SES.PWshape in LSD and SES.PWshape in RED have implied that the relative importance of deterministic processes (i.e., environmental filtering and competitive exclusion) varied along the elevational gradient. We estimate that apparent elevational patterns of community structure in LSD are resulting from significant niche separation among lineages. Early study has mentioned that three families (Muridae, Cricetidae and Sciuridae) in Rodentia have acted key roles in assembling rodent communities in the HMs (Du et al. 2017). Therein, long-tailed murine species have occupied the complete gradient, whereas short-tailed species of Cricetidae mainly distribute at medium and higher elevations. In addition, hylacolous sciurine species mainly survive in broad-leave and coniferous forests ranging from mid-low to mid-high elevations, except for Marmota himalayana surviving in alpine desert steppe.
    In contrast, non-significant linear elevational patterns of SES.MPD and SES.PWsize in RED are possible resulting from higher environmental heterogeneity and enlarged species component. In RED, the horizontal extent of each elevational band approach 9 degrees and 14 degrees at longitudinal and latitudinal directions. Resulting from extraordinarily neighboring topological and climatic heterogeneity in HMs, regional slice with a 100m-elevation range contains enormous subareas and microhabitats, which has harbored mass of rodent species without substantial overlap in distribution. This artificial treatment might slightly influence the pattern of species diversity, but greatly affect phylogenetic and functional diversity pattern, especially the loss or gain of rare species (Mi et al. 2012). To some extent, this offers an interpretation for the consistent patterns of species richness but distinct phylogenetic and morphological community structure patterns in two datasets.
    According to best predictive model selection, we have detected that different facets of community structure performed distinct dependence to environmental variables, and the degree of environmental dependence was much lower at regional scale. Obviously, due to higher level of environmental heterogeneity, enlarged species component and less accurate treatment in extracting environmental variables, the interpreting power of climate predictive models deserve to sharply decline.