Spatially explicit estimates of global population potential for leopard (Panthera pardus)

Abstract Many studies have assessed local densities of threatened carnivore species, but few have synthesized existing data into range-wide estimates based on habitat potential. Identifying where observed densities differ from expected values may help focus conservation attention. Herein, we related published leopard density estimates and a random selection of 0 values from across their extirpated range ( N = 209) to environmental and anthropogenic drivers. We then produced a spatially explicit estimate of expected density throughout the contemporary range of this species. We compared 16 candidate models that weighed density estimates by the inverse of their standard error, controlled for study design effects, and included spatial covariates at their optimal scale as determined by Akaike’s Information Criterion. Expected leopard density ranged from 0 to 8 individuals/100 km 2 and was positively and strongly related to the proportion of protected area at the 5-km scale, less so to the proportion of developed land cover (5-km), and negatively associated with human density (1-km), temperature (20-km), and longitude. Prediction error ranged from 0 to 12 leopards/100 km 2 , being greatest in India and its bordering countries. Overall, our model produced an expected value of 131,300 (42,692–428,498 95% CI) leopards globally. We predicted high expected densities in southern and eastern Africa where many published studies originate. Central and West Africa also hold high potential but lack estimates. Fine-scale variables and legacies of armed conflict that have depleted biodiversity should be considered when evaluating contemporary species status, and our predictions may highlight areas where conservation action could be most effective.