Determinants of patchiness of woody vegetation in an African savanna

How is woody vegetation patchiness affected by rainfall, fire and large herbivore biomass? Can we predict woody patchiness and cover over large-scale environmental gradients? We quantified variation in local patchiness as the lacunarity of woody cover on satellite-derived images. Using Random Forest regression we analysed how both average woody cover and its patchiness depend on annual rainfall, fire frequency and grazer and browser metabolic biomass densities. Fire frequency and rainfall were the clearest predictors, whereas effects of large herbivores on woody vegetation were smaller and more complex. Under low rainfall conditions (500 mm·yr−1) trees had less total cover and were more regularly spaced across the landscape (lower patchiness). Woody cover and vegetation patchiness increased with rainfall whereas fire frequency decreased woody cover and further increased patchiness. These results suggest a switch from competition between neighbouring trees under low rainfall conditions towards benefits among neighbours with increasing fire frequencies. Whereas overall woody cover and its patchiness are two independent aspects of savanna woody vegetation patterns, both need to be investigated to obtain a good understanding of the functioning and diversity of savanna ecosystems