Geographical distributions of spiny pocket mice in South America: insights from predictive models

Abstract Aim Predictive models of species’ distributions use occurrence records and environmental data to produce a model of the species’ requirements and a map of its potential distribution. To determine regions of suitable environmental conditions and assess biogeographical questions regarding their ranges, we modelled the potential geographical distributions of two spiny pocket mice (Rodentia: Heteromyidae) in north‐western South America. Location North‐western South America. Methods We used the Genetic Algorithm for Rule‐Set Prediction (GARP), environmental data from GIS maps and georeferenced collection localities from a recent systematic review of Heteromys australis and H. anomalus to produce the models. Results GARP models indicate the potential presence of H. australis throughout mesic montane regions of north‐western South America, as well as in some lowland regions of moderately high precipitation. In contrast, H. anomalus is predicted to occur primarily in drier areas of the Caribbean coast and rain‐shadowed valleys of the Andes. Conclusions The models support the disjunct status of the population of H. australis in the Cordillera de Mérida, but predict a continuous distribution between known populations of H. anomalus in the upper Magdalena Valley and the Caribbean coast. Regions of suitable environmental conditions exist disjunct from known distributional areas for both species, suggesting possible historical restrictions to their ranges. This technique holds wide application to other study systems.