Elephants and other large herbivores reduce the growth and survival of adult trees, and often are regarded as critical in maintaining landscape heterogeneity in African savannas. However, in other terrestrial ecosystems, anti-herbivore defenses can buffer against top-down control and thereby offset herbivory as a driver of productivity and plant population dynamics. Our study addressed two questions: 1) do well-defended adult trees exhibit dampened responses to the loss of megaherbivores, relative to congeners that do not invest heavily in anti-herbivore defenses?; and 2) do differences in anti-herbivore defenses employed by adults of different tree species affect the response of tree populations to early demographic stages (i.e., tree seeds and seedlings)? We combined a long-term herbivore exclusion experiment in Laikipia, Kenya with ant removal manipulations and cafeteria-style feeding trials on elephants to show that ant symbionts reduced palatability of their host tree (Acacia drepanolobium) relative to other Acacia that were not defended by ants.
Differences in tree palatability were manifested across broad spatial and temporal scales: productivity of A. drepanolobium did not change directionally following the long-term exclusion of megaherbivores, while that of its congeners increased significantly over 15 years across two properties spanning ca. 250 km2. Further, analyses of life table response experiments (LTREs) demonstrated that rates of population change in A. drepanolobium were driven largely by seedling production and recruitment. Thus, for this unpalatable tree, inconspicuous consumers (i.e., rodents) imposed strong demographic filters via seed and seedling predation, presumably because adult trees were unpalatable to larger, more conspicuous mammalian herbivores. Our results highlight how ant symbionts can regulate the productivity of savanna trees, thereby sustaining a suite of community- and ecosystem-level processes in African savannas.