Effects of mammalian herbivore declines on plant communities: observations and experiments in an African savanna

Herbivores influence the structure and composition of terrestrial plant communities. However, responses of plant communities to herbivory are variable and depend on environmental conditions, herbivore identity and herbivore abundance. As anthropogenic impacts continue to drive large declines in wild herbivores, understanding the context dependence of herbivore impacts on plant communities becomes increasingly important. Exclosure experiments are frequently used to assess how ecosystems reorganize in the face of large wild herbivore defaunation. Yet in many landscapes, declines in large wildlife are often accompanied by other anthropogenic activities, especially land conversion to livestock production. In such cases, exclosure experiments may not reflect typical outcomes of human-driven extirpations of wild herbivores. Here, we examine how plant community responses to changes in the identity and abundance of large herbivores interact with abiotic factors (rainfall and soil properties). We also explore how effects of wild herbivores on plant communities differ between large-scale herbivore exclosures and landscape sites where anthropogenic activity has caused wildlife declines, often accompanied by livestock increases. Abiotic context modulated the responses of plant communities to herbivore declines with stronger effect sizes in lower-productivity environments. Also, shifts in plant community structure, composition and species richness following wildlife declines differed considerably between exclosure experiments and landscape sites in which wild herbivores had declined and were often replaced by livestock. Plant communities in low wildlife landscape sites were distinct in both composition and physical structure from both exclosure and control sites in experiments. The power of environmental (soil and rainfall) gradients in influencing plant response to herbivores was also greatly dampened or absent in the landscape sites. One likely explanation for these observed differences is the compensatory effect of livestock associated with the depression or extirpation of wildlife. Synthesis. Our results emphasize the importance of abiotic environmental heterogeneity in modulating the effects of mammalian herbivory on plant communities and the importance of such covariation in understanding effects of wild herbivore declines. They also suggest caution when extrapolating results from exclosure experiments to predict the consequences of defaunation as it proceeds in the Anthropocene.