Protected areas alone rarely predict mammalian biodiversity across spatial scales in an Albertan working landscape

Abstract Protected areas (PA) networks are promoted globally as an effective method of biodiversity conservation and are increasingly implemented to meet the Convention on Biological Diversity Aichi objectives. However, whether PA networks actually conserve biodiversity, and how surrounding landscape features impact their efficacy, is controversial. We used a landscape approach to test whether mammalian biodiversity is maintained locally in PA networks, and whether landscape disturbance in the surrounding matrix detrimentally impacts biodiversity. We measured mammalian biodiversity using camera traps and functional diversity metrics, an approach that could be broadly applied to PA networks in working landscapes globally. We used generalized linear models to relate mammalian biodiversity metrics to natural, anthropogenic, and protected habitats across a range of 20 spatial scales to encompass increasing amounts of matrix. Biodiversity metrics increased with proportion of natural habitats at small scales and decreased with anthropogenic disturbance at large scales surrounding PAs. We found the conservation value of PAs are largely determined by the natural habitat remaining undisturbed, and the degree and type of disturbance in the surrounding working landscape. Implementing protected areas in degraded ecosystems, without addressing that degradation, will likely not achieve mammalian biodiversity conservation goals. We suggest, to achieve Aichi objectives, PAs placed in areas of high natural habitat and mitigated development will provide the best value for mammalian biodiversity conservation.