Specialist resources are key to improving small mammal distribution models



Small ground-dwelling mammals can have complex ecological relationships with environmental factors that limit the usefulness of coarse data in predictive species distribution models. We investigated the relative importance of available abiotic and biotic, landscape- and quadrat-scale data for predicting the distributions of four small mammals using data at three resolutions: 150 m, 500 m and 1000 m. At 150 m, the inclusion of landscape-scale data to a climate-only model improved the predicted occurrence of the wet heath specialist and woodland generalist, but not the two dense understorey species. Limited improvement was obtained with the inclusion of available quadrat-scale data (possibly because of missing or insufficiently detailed descriptive variables). As the models of best fit were re-applied to lower-resolution environmental data (500 m and 1000 m), the variance explained decreased for the wet heath specialist and two dense understorey species. These trends corresponded with reduced variance explained predominantly by biotic variables or abiotic landscape variables respectively. In contrast, the resolution of environmental data had no effect on the woodland generalist species distribution models, indicating the habitat for this more mobile species was sufficiently represented at the lowest resolution (1000 m). These results highlight the potential value of landscape- and finer-scale variables in modelling the distributions of small mammals. Where such variables are unavailable, higher-resolution climate data could better represent resource availability (indirectly) or suitable microclimates (directly), especially for more vulnerable, above-ground nesting species. We encourage the collection of additional detailed and high-resolution environmental information to facilitate the development of more accurate models of the extent and distribution of small mammals.