Beyond the model: expert knowledge improves predictions of species' fates under climate change.

摘要: The need to proactively manage landscapes and species aid their adaptation climate change is widely acknowledged. Current approaches prioritizing investment in conservation generally rely on correlative models, which predict the likely fate of under different scenarios. Yet, while model statistics can be improved by refining modeling techniques, gaps remain understanding relationship between performance ecological reality. To investigate this, we compared standard distribution models highly accurate, fine-scale, models. We critically assessed realism each species' model, using expert knowledge geography habitat study area biology species. Using interactive software an iterative vetting with experts, identified seven general principles that explain why under- or overestimated suitability, both current predicted future climates. Importantly, found that, temperature estimates dramatically through better downscaling, many still inaccurately reflected moisture availability. Furthermore, did not account for biotic factors, such as disease competitor species, were unable presence micro refugia. Under-performing resulted divergent projections distributions. Expert regions contain refugia, even where fine-scale distributions population losses. Based results, identify four priority actions required more effective responses. This approach improving understand impacts applied broadly improve evidence base underpinning management