Biogeographic regionalization of the Amazon using highly diverse horse flies (Diptera: Tabanidae): insights from three decades of data

Delimiting biogeographic regions based on occurrence data is an interesting approach to investigating processes behind biodiversity distribution patterns. Comparing spatial scales and identifying predictor variables of biogeographic regions have wide application for biodiversity conservation. In this study we used a comprehensive database containing more than thirty years of horse fly records to estimate species richness, endemism, and species composition, and regionalize the Amazon biogeographically. We compared five spatial scales defined by grid size (1–5º), and test five hypotheses (elevation, climate, vegetation cover, and two regionalizations from the literature) to identify predictors of the biogeographic regions. Endemism, species richness and composition were predicted by different sets of predictor variables, although the models were highly dependent on spatial scale. We identified three well-defined biogeographic regions, which have been formed by a combination of geographic distance, climate and historical factors converging with some theories proposed for mammals. Our models indicated dispersal as a key factor for regionalization, as it can be constrained by a combination of climate and historical processes changing habitats over time, although this finding was highly dependent on spatial scale. We showed that horse flies are interesting models for biogeography although they have been historically neglected.