Grasslands, which represent around 40% of the terrestrial area, are mostly located in arid and semi-arid zones. Semiarid ecosystems in Africa have been identified as being particularly vulnerable to the impacts of increased human pressure on land, as well as enhanced climate variability. Grasslands are indeed very responsive to variations in precipitation. This study evaluates the sensitivity of the grassland ecosystem to precipitation variability in space and time, by identifying the factors controlling this response, based on monthly precipitation data from Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) and the Fraction of Absorbed Photosynthetically Active Radiation (FAPAR) data from the Multi-angle Imaging SpectroRadiometer-High Resolution (MISR-HR) datasets, used as proxy for productivity, at 60 grassland sites in South Africa. Our results show that MISR-HR products adequately capture the spatial and temporal variability in productivity at scales that are relevant to this study, and they are therefore a good tool to study climate change impacts on ecosystem at small spatial scales over large spatial and temporal domains. We show that combining several determinants and accounting for legacies improves our ability to understand patterns, identify areas of vulnerability, and predict the future of grassland productivity. Mean annual precipitation is a good predictor of mean grassland productivity. The grasslands with a mean annual rainfall above about 530 mm have a different functional response to those receiving less than that amount of rain, on average. On the more arid and less fertile soils, large inter-annual variability reduces productivity. Our study suggests that grasslands on the more marginal soils are the most vulnerable to climate change.

West Africa, already highly influenced by the negative effects of climate change, is additionally characterized by rapid population growth, endemic poverty, and insecurity. This is affecting the natural capital of its ecosystems and the services they provide. Natural capital accounting (NCA) provides the fundamental evidence base required for informing economics and environmental decisions, thus strengthening the conservation and management of natural resources. The objective of this study is to showcase the development and evaluation of a semi-automated NCA platform (Sys4ENCA) designed to support decision making in the context of protected areas management in a multi-level example in western Africa. The accounting results highlight that simulations at the broader scale using national public data show that the natural capital of ecosystems in western Africa depends strongly on the mean climate and its variability. Evaluating regional datasets, the simulation with the platform shows that pressure on land in combination with weak governance reduces the capability of the ecosystem to deliver the required services in a sustainable manner, i.e., in the eastern part of the Bafing-Falémé landscape, where mining and intensive agriculture are fueling loss of natural capital. The results of Tier-3 accounting using local datasets enhanced the spatial variability and highlighted additional hotspots of degradation compared to the regional results, i.e., the prospective construction of a hydro-electricity dam (Koukoutamba) in the southern part of the Moyen-Bafing National Park located in the Bafing-Falémé landscape. The Sys4ENCA platform, combined with a multi-level approach, showed itself to be a valuable tool to facilitate protected area management as it provides not only consolidated information at a local scale but also the broader context and external pressures, i.e., climate change and demand for land. Given its automatized nature, the platform reduces human errors and increases the efficiency, speed, and harmonisation of computation over long timeframes and spatial scales.