Savanna vegetation is variable, and predicting how water, nutrients, and chronic disturbances interact to determine vegetation structure in savannas represents a challenge. Here, we examined in situ interactions among rainfall, soils, grasses, fire, and elephants that determine tree layer responses to resource gradients in Kruger National Park in South Africa, using 363 long-term monitoring sites throughout the park. Grass biomass increased with rainfall and on nutrient-rich clay soils. Fire frequency, too, increased with rainfall. Conversely, tree density was greater on sandier soils, where water infiltrates more readily, and in areas where the maximum interval between fires was longer, irrespective of average fire frequency. Elephant density responded positively to tree density, but did not contribute significantly to decreasing tree density. Savanna vegetation structure was reasonably predictable, via a combination of rainfall (favoring grasses), soil (sandy soils favoring trees), and fire (limiting trees until a longer interval between fires allows them to establish). Explicit consideration of bottom-up and top-down interactions may thus contribute to a predictive understanding of savanna vegetation heterogeneity.

Rainfall variability, including drought, consistently translates into variation in grass productivity in savanna ecosystems. These variations in grass availability can impact grazer populations, which can in turn shape grass responses to drought. However, despite clear evidence that grazers move freely in response to forage and water availability, drought ecologists have mostly failed to consider how these landscape‐scale processes influence vegetation responses to drought. Here, we ask how rainfall variability and drought, in interaction with population dynamics and movement of large grazers, shape grass layer responses in extant savanna landscapes in Kruger National Park in South Africa. We found that grass biomass was highly responsive to rainfall variability; notably, grass declines during drought were especially severe—even more severe than previously documented elsewhere. Grazing probably contributed, exacerbating drought effects in droughted regions and spatially extending them into non‐droughted refugia, where grass biomass also declined. Moreover, the existence of these refugia potentially prevented a grazer population crash. Synthesis. Our current focus on the physiological effects of drought is probably insufficient for understanding vegetation responses to drought. A renewed focus on landscape processes, including animal movement, will be critical to predicting savanna responses to increasingly frequent extreme events.

White rhinoceros (rhinos) is a keystone conservation species and also provides revenue for protection agencies. Restoring or mimicking the outcomes of impeded ecological processes allows reconciliation of biodiversity and financial objectives. We evaluate the consequences of white rhino management removal, and in recent times, poaching, on population persistence, regional conservation outcomes and opportunities for revenue generation. In Kruger National Park, white rhinos increased from 1998 to 2008. Since then the population may vary non-directionally. In 2010, we estimated 10,621 (95% CI: 8,767–12,682) white rhinos using three different population estimation methods. The desired management effect of a varying population was detectable after 2008. Age and sex structures in sink areas (focal rhino capture areas) were different from elsewhere. This comes from relatively more sub-adults being removed by managers than what the standing age distribution defined. Poachers in turn focused on more adults in 2011. Although the effect of poaching was not detectable at the population level given the confidence intervals of estimates, managers accommodated expected poaching annually and adapted management removals. The present poaching trend predicts that 432 white rhinos may be poached in Kruger during 2012. The white rhino management model mimicking outcomes of impeded ecological processes predicts 397 rhino management removals are required. At present poachers may be doing “management removals,” but conservationists have no opportunity left to contribute to regional rhino conservation strategies or generate revenue through white rhino sales. In addition, continued trends in poaching predict detectable white rhino declines in Kruger National Park by 2016. Our results suggest that conservationists need innovative approaches that reduce financial incentives to curb the threats that poaching poses to several conservation values of natural resources such as white rhinos.