Macroecological conclusions based on IUCN expert maps: A call for caution

Aim: International Union for Conservation of Nature (IUCN) expert maps are increasingly used in macroecological research. However, they have not been produced for this purpose. Macroecological insights based exclusively on this type of data could therefore be misleading. Here we compare, for a large taxonomic group (bats) and an entire biogeographical realm (Africa), the species-specific discrepancies between IUCN maps and species distribution models (SDMs) that approximate the complete geographical range of species. We then examine the implications for a typical macroecological analysis that explores environmental correlates of species richness. Location: Continental Africa. Time period: Around 2000. Taxa: Bats (Chiroptera). Methods: We measure disagreement between IUCN expert maps and SDMs at both the species (geographical ranges) and the aggregated level (range size-frequency distributions and species richness). We further quantify the difference in absolute and relative weight assigned to three variables hypothesized to drive species richness: primary productivity, climatic seasonality and environmental heterogeneity. Results: Location, shape and size of individual species' ranges, derived richness patterns and range size-frequency distributions differ substantially. SDMs predict larger and more complex geographical ranges, and species' range sizes vary less. The spatial congruence of richness hotspots among both datasets is only 42%. These discrepancies are large enough to alter the absolute explanatory power of environmental correlates, whereas the redundancy in the variation explained increases markedly when richness is inferred using SDM-based estimates of complete species ranges. Main conclusions: IUCN expert maps differ considerably and systematically from SDMs built to estimate complete species ranges, primarily because of their intentional greater sensitivity to geographical sampling bias. This property is desirable in a conservation context but unwanted in most macroecological analyses. We therefore caution against the use of IUCN expert maps in macroecology and recommend at least gauging the robustness of results using additional range estimates designed to approximate the complete geographical range of species.