Print PageA recurring question in community ecology is: why do savannas have so few trees? In other words, what keeps trees from colonizing and growing in the open patches of grassland, which are theoretically capable of becoming dense woodland? Theories of tree limitation in savannas have focused either on bottom-up constraints, such as water limitation, and top-down processes, such as limitation by fire. These theoretical models invariably come with implicit assumptions about how recruitment processes (the demographic stages of limitation) should shift across resource and disturbance gradients, although these assumptions have rarely been tested in natural systems. For example, bottom-up limitation is predicted to result in poor seed germination and seedling establishment, and establishment should vary across space with the intensity of competition for water. In contrast, top-down limitation by fire is predicted to result in a significant pool of saplings in the herbaceous layer, but poor recruitment during the sapling to adult phase. As a result, top-killed resprouting individuals are predicted to be common in the herbaceous layer, and the spatial location of new recruits should track areas of low fire intensity. Here we describe a unique fire manipulation study designed to address the relative strength of bottom-up versus top-down factors on the regulation of tree recruitment in savannas across a soil moisture gradient. The study comprises 40 0.1-ha plots spanning a 200-km moisture gradient in Serengeti National Park, Tanzania. Each year, half the plots are burned and half are protected from fire. We mapped the spatial location of every seedling, sapling, resprout and adult tree annually, and tested hypotheses about the importance of spatial processes related to water availability and fire at each demographic stage.
Results/Conclusions
Our findings suggest that the strength of recruitment limitation may shift across the soil moisture gradient, from regulation at the seedling establishment stage by water availability at low rainfall, to regulation at the sapling to adult stage by fire at high rainfall. Furthermore, a significant decoupling occurs between the species composition in the canopy and the understory, highlighting the unexpected importance of seed dispersal by large mammals. Finally, our results suggest that spatial patterns within savannas, determined by tree neighborhoods, have subtle but important influences on recruitment that should be integrated into models of savanna dynamics. In conclusion, tradeoffs between bottom-up and top-down mechanisms operate across large environmental gradients and at smaller scales determined by the spatial distribution of adult trees.
