Identifying Ecological Knowledge and Research Gaps via the African Database of Savanna Protected Areas (ADSPA)

Aim: Despite their extent (40° of latitude and 50° of longitude), research in African savannas is dominated by a few heavily studied areas. We gathered data from African savanna protected areas to (i) evaluate their contributions to the primary literature, (ii) identify environmental groupings with respect to climate, soils, and landscape variables, and (iii) analyze the determinants of tree cover and fire within groupings. Location: Africa. Methods: We extracted climate, soil, topography, hydrology, elephant, fire, and tree cover data from polygon boundaries for 244 African savanna protected areas. The polygon layers and data were assembled into a novel geodatabase: African Database of Savanna Protected Areas (ADSPA). Cluster analysis identified natural bioclimatic groupings and structural equation modelling was used to analyse and compare the drivers of fire and tree cover within and across clusters. Results: Previous literature disproportionately focused on a few savannas: 46% of savanna publications came from 2% of protected areas. Cluster analysis identified five bioclimatic groups: (1) African hot mesic savannas, (2) African cool mesic fertile savannas, (3) West African hot semi-arid savannas, (4) Southern African semi-arid savannas, and (5) Kalahari arid savannas. Current savanna science in protected areas is biased toward the Southern African semi-arid and African cool mesic fertile savannas, while hot mesic, hot semi-arid, and arid savannas are underrepresented. Climate and soils were strongly associated with tree cover and fire across protected areas, but no significant biome-wide effects of fire on tree cover emerged. However, tree cover was negatively related to fire in the hot mesic savanna cluster. Main Conclusions: Significant biogeographic and ecological variation within African savannas highlights the need for research across the entire breadth of the biome, especially West Africa. We stress the need for spatially explicit, Africa-wide, data on mammalian herbivore biomass to better assess the importance of this variable for savanna functioning.