23The drivers of community coexistence are known to vary with environment, but their consistency across 24 latitudes and scales, and resulting conservation implications, remain little understood. Here, we 25 combine functional and phylogenetic evidence along elevations to document strong biotic constraints 26 on coexistence in avian communities in both benign (tropical low elevations) and severely harsh 27 (temperate/polar highlands) environments. Assemblages in both are marked by high assemblage 28 functional uniqueness, whereas in tropical highlands and temperate/polar low elevations there is strong 29 functionally redundancy and pronounced environmental constraints. Only in harsh environments is 30 phylogeny an effective surrogate for functional assemblage structure, reflecting nuanced shifts in the 31 position, shape, and composition of measured multivariate trait space along gradients. Independent of 32 scale and latitude, high elevation assemblages emerge as exceptionally susceptible to functional change. 33 34 42 (10-12) are expected to increase divergence (i.e., overdispersion) of trait characteristics among 43 community members, particularly for closely related species, environmental constraints (or filters) that 44 select for common phenotypes might decrease trait divergence and enhance trait similarity (i.e., 45 clustering) (13). The latter are expected to dominate toward harsh and less stable environments (13, 46 14), whereas biotic constraints, particularly competitive exclusion, should be more prominent in 47 productive and stable settings (15, 16). Despite their key role for gauging the functional consequences of 48 climate change, whether and how these processes and resulting patterns hold from local to 49 regional/global scales, where macroevolutionary constraints and contingencies on clade functional 50 space emerge, remains largely unclear (17-19) or reliant on phylogenetic proxies of uncertain surrogacy 51 value (20-22). 52 53 Here, we use the natural experiment provided by elevational gradients of the world's main mountain 54 regions replicated along latitude (23, 24), a time-calibrated phylogeny (25), comprehensive trait data 55 (26), and elevational distributions (24) addressing nearly all extant bird species, to test how functional 56 structure varies from benign to harsh conditions across local, regional, and global scales. For 8,410 57 assemblages along elevations from sea level to 7,340 meters, we estimate prevalence of assembly 58 mechanisms using dendrogram-based functional diversity (FD), hypervolume-based functional diversity 59 (FDH; Supplementary Material), and species' local functional distinctness (FDI) which captures the 60 distinct contribution species make to the total functional diversity of a local assemblage (27). We use 61 their species richness-controlled values, cFD and cFDI (standardized effect sizes), supported by quantile 62 scores and associated p-values, to distinguish overdispersion from clustering. To investigate the 63 evolutionary underpinnings of species coexi...