Precipitation, fire and demographic bottleneck dynamics in Serengeti tree populations

Abstract

Tree size distributions are the outcome of demographic processes and disturbance events, and size distribution analysis provides a useful tool for understanding pattern and process in tree population dynamics. Demographic bottleneck mechanisms such as fire “traps” are important for driving tree cover dynamics in savanna systems, and bottlenecks might be expected to be revealed by bimodal size distributions in savanna tree communities. We tested the relative fit of monotonic and bimodal Weibull distributions to tree height distributions across 36 0.1-ha plots over 4 years in Serengeti National Park, Tanzania, using a Bayesian analysis. The plots were subjected to two fire treatments and spanned a mean annual rainfall gradient ranging from 600 to 900 mm year⁻¹. We found that Serengeti trees are highly bimodal in their height distributions, with a pronounced gap in the 1–3 m height range, suggesting that demographic bottlenecks are a pervasive feature of this system. We also found that pre- and post-bottleneck tree densities are increasing and declining over time, respectively. Pre-bottleneck density declined with fire and increased with mean annual precipitation, and exhibited a rainfall by fire interaction, with negative fire effects becoming more important at the wet extreme of our rainfall gradient. Overall, despite the negative effect of fire on pre-bottleneck trees, the density of the latter is increasing over time, suggesting that although recruitment into larger size classes has been tightly constrained in the past, there is mixed support for a role of fire in maintaining this pattern under current burning regimes.