For savanna trees, the transition from juvenile to adult stage is a key demographic process influencing savanna tree cover and shifts in alternative stable states between woody and grass dominance. In African savannas, wildlife browsing can directly suppress juvenile trees. The widespread replacement of wildlife by cattle is presumed to cause indirect positive effects on woody saplings by reducing the competitive effects of grasses on trees. However direct empirical evidence of wildlife- vs. cattle-induced changes in tree demography have been rare. In our study system, browsing of the dominant tree, Acacia drepanolobium, is reduced to varying degrees by four species of ant-mutualists, and tree growth is also limited by intraspecific competition with neighboring trees. Here we report the results of a 10 year experiment designed to quantify the effects of wild herbivores, cattle, grasses, ant occupancy, and neighboring trees, on acacia saplings in a Kenyan savanna. Saplings of varying ant occupancy growing under four long-term herbivore regimes (wild herbivores, cattle, cattle + wild herbivores, and no large herbivores) were cleared of surrounding grass or left with the surrounding grass intact. Tree neighborhood density was estimated by counting the number of trees >2 meters in height within a 3m radius of each sapling. After ten years ant occupancy, height, and the basal stem diameters were re-measured.
Grass-removal had a positive effect on both basal diameter and height across all herbivore treatments except the no-herbivory treatment. Mortality rates were low (8% all trees) and did not differ significantly across herbivore treatments (Pearson’s, Χ2 = 4.86, df = 6, p > 0.5). Grass removal and wildlife presence both increased the proportion of trees that “escaped” from the grass layer (attained a height >70 cm; functionally reached the next demographic stage). This positive wildlife effect may have been mediated by tree density; there was a negative relationship between neighborhood tree density and sapling size, and tree density was lowest in plots with wildlife. Across all plots, most trees that escaped (~76%) were occupied by one of the two most aggressive ant mutualists (Crematogaster mimosa and Crematogaster nigriceps). These results provide evidence that while the effects of moderate utilization by cattle (i.e., the cattle treatment) were negligible, simulated heavy grazing (i.e., grass removal treatment) increased sapling growth and recruitment regardless of wildlife. Despite loss of sapling biomass to wildlife browsing, recruitment rates remained relatively high, possibly due to reduced interspecific competition from neighboring trees.