Ecological drivers of avian community assembly along a tropical elevation gradient

Community assembly theory hypothesizes that two main niche-based processes act to shape composition and organization of biological assemblages: abiotic filtering and biological interactions. Here, we conducted repeated surveys of bird abundance along an undisturbed elevational gradient in the tropical Andes to investigate (1) signals of deterministic processes driving community assembly and (2) potential mechanisms by which these forces operate (temperature, habitat complexity, fruit and insect availability), while correcting for imperfect detection and modeling species abundances with N-mixture models. We observed strong signals of abiotic filtering driving functionally and phylogenetically clustered assemblages towards higher elevations, and a weaker signal of limiting similarity resulting in few overdispersed assemblages at lower elevations. Whereas the decay in species richness with increasing elevation was explained by temperature, trait and phylogenetic dispersion were explained by both temperature and vegetation structure, implying that an interplay of abiotic and biotic mechanisms determines abundance-based community structure in our montane assemblages. Interestingly, trait and phylogenetic dispersion consistently decreased until ~3000 m but increased above this elevation, highlighting a potential role of competition in resource-scarce habitats. Combined, our findings suggest abiotic filters are still the main process shaping montane biotas across elevations, whereas resource availability might act locally upon assemblages further modifying them. Our study challenges recent studies in tropical mountains that suggest that biotic filters are a stronger force than abiotic filters in shaping tropical montane assemblages, and exemplifies how accounting for imperfect detection might overcome potential biases in detecting environmental filtering signals in community assembly studies.

Methods
Bird surveys were conducted between May and July 2014 and May and October 2015 and 2016  along an extensive gradient (1350 – 3650 m asl) in Cotapata National Park, a protected area in the Andes of western Bolivia (67°43′−68°03′ W/16°05′−16°20′ S, c. 80 km NE from La Paz city). For the purpose of analysis, we defined a ‘site’ as a band with 50-m elevational change along the gradient.

We used a multi-season community N-mixture model implemented in a Bayesian framework to estimate abundance-based community diversity and structure while accounting explicitly for unobserved species and individuals. Detectability was modeled as logit-linear combination of site- or survey-specific covariates (i.e., date, start time, survey direction, and duration). Then, we calculated taxonomic, functional and phylogenetic diversity for each 50-m elevational band using the posterior samples of the N-matrix. For each posterior sample of the N-matrix (n=500), we calculated taxonomic diversity simply as species richness. Abundance-weighted functional and phylogenetic diversity were calculated with two dispersion metrics: mean pairwise dissimilarity (MPD) and mean nearest taxon distance (MNTD). Finally, we regressed taxonomic, functional and phylogeneetic diversity as a function of temperature, habitat complexity and the diversity of food resources (fruits and insects) across elevations.