Predicting the effects of land cover change on biodiversity in Prairie Canada using species distribution models

Land cover change is the largest direct driver of global biodiversity loss but often the relationships between habitats and species occurrence are unknown. The conservation community requires tools to assess variation in biodiversity related to land cover for maximizing return on investment. Our objectives were to 1) develop a biodiversity mapping and assessment tool at a fine spatial scale for terrestrial vertebrates, and 2) test how much biodiversity is conserved by retaining natural habitats within agricultural landscapes. We built species distribution models for amphibians, birds, mammals, and reptiles (329 species, > 1.2 million observations) within Prairie Canada. Predicted biodiversity within 805 m x 805 m sites ranged from 0 to 238 species (66 +/- 0.1). The proportion of annual cropland at a site had the largest negative effect on biodiversity among predictors. Using simulations of land cover change, we predicted that conserving 20 % of natural habitats would conserve an average of 26.5 % of maximum species richness in fields with annual cropland and 74.3 % of maximum species richness in fields with tame grass (perennial cropland). Our tool predicted that fields with conservation easements (n = 312) had more species (114 +/- 2) and natural habitat (48 +/- 1 %) compared to nearby unprotected sites (82 +/- 3 species; 32 +/- 2 % natural habitat). Our results highlight the importance of retaining natural habitats, including wetlands, grasslands, and forests within farms to support biodiversity. In addition, our predictions can be used to target areas for conserving and restoring habitats.