Across the globe, mobile species are key components of ecosystems. Migratory birds and nomadic antelope can have considerable conservation, economic or societal value, while irruptive insects can be major pests and threaten food security. Extreme weather events, which are increasing in frequency and intensity under ongoing climate change, are driving rapid and unforeseen shifts in mobile species distributions. This challenges their management, potentially leading to population declines, or exacerbating the adverse impacts of pests. Near-term, within-year forecasting may have the potential to anticipate mobile species distribution changes during extreme weather events, thus informing adaptive management strategies. Here, for the first time, we assess the robustness of near-term forecasting of the distribution of a terrestrial species under extreme weather. For this, we generated near-term (2 weeks to 7 months ahead) distribution forecasts for a crop pest that is a threat to food security in southern Africa, the red-billed quelea Quelea quelea. To assess performance, we generated hindcasts of the species distribution across 13 years (2004-2016) that encompassed two major droughts. We show that, using dynamic species distribution models (D-SDMs), environmental suitability for quelea can be accurately forecast with seasonal lead times (up to 7 months ahead), at high resolution, and across a large spatial scale, including in extreme drought conditions. D-SDM predictive accuracy and near-term hindcast reliability were primarily driven by the availability of training data rather than overarching weather conditions. We discuss how a forecasting system could be used to inform adaptive management of mobile species and mitigate impacts of extreme weather, including by anticipating sites and times for transient management and proactively mobilising resources for prepared responses. Our results suggest that such techniques could be widely applied to inform more resilient, adaptive management of mobile species worldwide.
Read full abstract