Macroecological explanations for differences in species richness gradients: a canonical analysis of South American birds

Aim To evaluate how spatial variation of species richness in different bird orders responds to environmental gradients and determine which order level trait best predicts these relationships. Location South America. Methods A canonical correlation analysis was performed between the species richness in each of 17 bird orders and eight environmental variables in 374, 220 × 220 km cells. Loadings associated with the first two canonical variables were regressed against six order‐level predictors, including diversification level (number of species in each order), body size, median geographical range size and characteristics included in the model to control Type I error rates (the phylogenetic relationship among orders and levels of local‐scale spatial autocorrelation). Results Richness patterns of 14 bird orders were highly correlated with the first canonical axis, indicating that most orders respond similarly to energy‐water gradients (primarily actual evapotranspiration, minimum temperature and potential evapotranspiration). In contrast, species richness within Trochiliformes, Apodiformes and Galliformes were also correlated with the second canonical variable, representing measures of mesoscale climatic variation (range in elevation within cells, minimum temperature, and the interaction term between them) and landcover (habitat diversity). We also found that total diversification within orders was the best predictor of the loadings associated with the first canonical axis, whereas body size of each order best predicted loadings on the second axis. Conclusion Our results broadly support climatic‐related hypotheses as explanations for spatial variation in species richness of different orders. However, both historical (order‐specific variation in speciation rates) and ecological (dispersal of species that evolved by independent processes into areas amenable to birds) processes can explain the relationship between order level traits, such as body size and diversification level, and magnitude of response to current environment, furnishing then guidelines for a further and deeper understanding of broad‐scale diversity gradients.