Integrating climate and species modelling to predict migratory and nomadic movements of birds in sub-Saharan Africa

The dynamic distributions of migratory and nomadic species pose unique challenges to population monitoring and management, particularly as extreme weather events drive rapid and unprecedented shifts in their distributions. With these events intensifying globally under climate change, effective management is essential, given that these highly mobile species can be pests, invasive species, or hold considerable conservation and economic value. Near-term forecasting of upcoming shifts in migratory and nomadic species distributions is urgently required to inform more effective, proactive management. However, the development of such forecasts has been constrained by the lack of tools for modelling species distributions with dynamic environmental variables, which are necessary to capture fine-scale changes. This thesis introduces and applies an innovative tool for dynamic species distribution modelling to evaluate a near-term forecasting system designed to improve the management of the nomadic red-billed quelea Quelea quelea, the most destructive crop pest bird in sub-Saharan Africa.

To advance existing tools, the dynamicSDM R package for species distribution modelling at high spatiotemporal resolution was developed. The package incorporates novel functions that explicitly consider the temporal dimension across key modelling stages, utilise Google Earth Engine to process high-resolution datasets, and offer flexibility for broad application. Using the package, dynamic species distribution models were developed to forecast red-billed quelea distribution shifts under near-term forecasts of ecoclimatic conditions across Southern Africa. To assess the performance of near-term distribution forecasting, hindcasts that replicated the forecasting system in a historical period (2004–2016) were generated and compared to historical observations of quelea distribution. The results demonstrated that red-billed quelea distributions could be accurately hindcast at high resolution with seasonal lead-times (two weeks to seven months ahead), including during extreme drought. To assess the utility of near-term distribution forecasts for adaptive management, a spatially explicit individual-based model was developed to simulate nomadic quelea populations under different management strategies. Simulations demonstrated how management efficiency and effectiveness could be enhanced by using near-term forecasts to implement proactive practices, compared to existing reactive approaches. Together, this research shows that by utilising emerging spatiotemporal datasets and modelling techniques, we can anticipate near-term species distributions to inform adaptive management of migratory and nomadic species under global change.