Range geometry and socio‐economics dominate species‐level biases in occurrence information

Aim Despite the central role of species distributions in ecology and conservation, occurrence information remains geographically and taxonomically incomplete and biased. Efforts to address this problem, such as targeted data mobilization and advanced distribution modelling, all crucially rely on a solid understanding of the patterns and determinants of occurrence information. Numerous socio-economic and ecological drivers of uneven record collection and mobilization among species have been suggested, but the generality of their effects remains untested. Here, we provide the first global analysis of patterns and drivers of species-level variation in different metrics of occurrence information. Location Global, including separate analyses for six zoogeographical realms. Methods We evaluated three alternative metrics of occurrence information: (1) the record count per species, (2) the coverage of a range with records and (3) the geographical bias in how the records represent different range parts. To this end, we developed scale-independent metrics of range coverage and geographical record bias. We applied the three metrics to 2.8 million point-occurrence records and extent-of-occurrence range maps of 3625 mammalian species. We used multi-model inference to evaluate 13 putative drivers of species-level variation in data availability. Results All three metrics of occurrence information revealed severe species-level biases. These data limitations were mainly linked to range size and shape, and the within-range geography of socio-economic conditions. Species attributes related to detection and collection probabilities, such as body size or diurnality, were remarkably weak predictors of record count and range coverage. Main conclusions Species-level biases in mobilized occurrence information hamper its broader application in basic and applied biodiversity research. To successfully account for these limitations, the site-specific socio-economic constraints to record collection and mobilization, rather than species-specific constraints to detection, should be explicitly incorporated into ecological models. Furthermore, our results strongly suggest that range-restricted species should be prioritized in future data mobilization efforts.