A high-resolution model of bat diversity and endemism for continental Africa

Abstract

Bats are the second-most species-rich mammal group numbering more than 1270 species globally. Our knowledge of their geographic distributions and diversity patterns however is very limited – possibly the poorest among mammals – mainly due to their nocturnal and volant life history, and challenging fieldwork conditions in the tropics where most bat species occur. This knowledge gap obscures the geographic extent of ecosystem services provided by bats (i.e. pollination, seed dispersal and insect control), translates into inefficient conservation policies, and restricts macroecological analyses to coarse spatial resolutions. In contrast to the currently prevailing method of estimating species distributions using expert-drawn range maps, correlative species distribution models (SDMs) can provide estimates at very fine spatial grains and largely account for widespread sample bias as well as the prevalent Wallacean shortfall in species occurrence data. Very few such studies have hitherto been published that cover a large and complete taxonomic group with fine resolution at continental extent. Using an unparalleled amount of occurrence data, the MaxEnt algorithm and tailored solutions to specific modelling challenges, we created SDMs for nearly all 250 African bat species to explore emerging diversity patterns at a resolution of 1km2. Predicted species richness generally increases towards the equator conforming to expectations. Within the tropical area of elevated richness, several pronounced richness peaks and lows stand out, hinting at a complex interplay of determining factors. Richness gradients are often steep, decreasing strongly away from streams, and especially so in savanna biomes. Species richness also seems positively associated with rugged terrain, in particular at lower elevations. Centres of endemism are found primarily at low latitudes near major elevational ranges. Overlap with hotspots of species richness is rather low, and confined to five or six topodiverse, relatively low lying areas between western Guinea and the East African coast. Several poorly sampled regions are identified that may represent rewarding future survey targets. Our results demonstrate the value of stacking SDMs to infer plausible continent-wide diversity gradients at a spatial resolution fine enough to directly inform conservation policies and to open up new avenues in macroecological research.