Environmental and geographic drivers of global bat phylogenetic diversity

Abstract

AimUnderstanding the patterns and factors that shape biodiversity is vital to conserving species. We combined open-source genetic, environmental, and geographic information to analyze bat phylogenetic diversity (PD) patterns in continuous ecoregions across the globe. This information is important for developing bat conservation strategies, and our methodology can work for any taxa with sufficient georeferenced genetic data available. LocationGlobal. MethodsAfter curating a global dataset containing 14,037 COI DNA sequences from 343 described species of bats, we calculated PD for continuous ecoregions at different spatial scales. To avoid the difficulties of using current species names, we used genetic OTUs identified by a single-locus species delimitation method to reconstruct and date a phylogeny. We then calculated PD, estimated a lineage through time plot, and used random forest predictive modeling to identify environmental and geographic predictors of PD. ResultsIn addition to current temperature, temperature during the last glacial maximum and temperature changes between the last glacial maximum and last interglacial were most closely associated with PD. However, at different spatial scales, the top variables differed slightly. When using smaller ecoregions, latitude and population density were also identified as important, though not significant. Southeast Asia and South America had the highest levels of PD, along with parts of Africa and the Himalayas. We demonstrate that, regardless of spatial scale and uneven sampling across the globe, single-locus genetic data can reflect species diversity gradients and identify predictors of PD. Main conclusionsWe show that publicly-available, single-locus data can be used to analyze large-scale evolutionary patterns and inform conservation efforts. Additionally, choice of biodiversity measure and spatial scale matter when assessing species patterns. When looking at bats, temperature variables with a historical component are most important for predicting PD broadly, but latitude and population density could also be important on smaller scales.