Understanding what governs the distribution, diversity and abundance of species has been a central theme of ecology for decades. In semi-arid landscapes, termites are one of the most conspicuous and ubiquitous modifiers of habitats. Through their foraging and mound-building activities, termites directly modify the basal components of the ecosystems and indirectly change the foraging patterns of vertebrate herbivores.  In so doing, they exert strong direct and indirect effects on entire communities. The objective of our study was to examine the combined effects of termites and vertebrate herbivores on plant and invertebrate communities in an East African savanna.

Results/Conclusions Using observational data, large-scale experimental manipulations, and isotopic analysis, we found that proximity to termite mounds was the strongest predictor of the fruiting status of /Acacia drepanolobium/.  /Acacia/ trees growing adjacent to termite mounds were more likely to reproduce than those growing further away, in off-mound soils (F_1,451 = 8.9107; P< 0.01).  Although vertebrate herbivores preferentially used termite mounds as demonstrated by dung deposits, long-term exclusion of mammalian grazers did not significantly reduce /A. drepanolobium /fruit production.  Leaf N concentration was significantly affected by proximity to termite mounds but unaffected by exclusion of vertebrates.  In addition to effects on the dominant tree species, termites modify the distribution of grasses and herbaceous species and, in so doing, affect the density and distribution of herbivorous and predaceous invertebrates.  The density and species diversity of invertebrates was significantly higher on termite mounds than off mounds (paired t_28 = 4.5; P < 0.001).  Invertebrate diversity, however, was often lower on mounds than off-mounds, suggesting that invertebrates may partition mound habitat.  Our results demonstrate that termites directly affect plants by enhancing soil quality and indirectly affect entire communities through altering the density and foraging patterns of both vertebrate and invertebrate herbivores.  Teasing apart these direct and indirect effects is critical to our understanding of East African savannahs in particular, and to our understanding of food web dynamics in general.