Testing non-equilibrium theories in savannas: 1400 years of vegetation change in Tsavo National Park, Kenya.

Non-equilibrium processes in savannas are poorly understood because little data is available on how tree abundance has changed over long periods of time. Here I analyse fossil pollen and charcoal from Kanderi Swamp in Tsavo National Park, Kenya in order to investigate patterns of vegetation change over the past 1400 years. I compare the results with the predictions of cyclical, phase and transition and stochastic non-equilibrium models. Fourier analysis of the pollen data provides evidence of cyclical changes in tree abundance with a periodicity of 250–500 years, but the pollen data also suggests that the vegetation surrounding Kanderi swamp remained in a prolonged woodland phase, as predicted by phase and transition models, from approximately 770–1520 a.d. In addition, the results provided evidence of much shorter term changes in tree/grass abundance, perhaps reflecting stochastic variations in rainfall. The charcoal data suggests a link between vegetation change and fire history, possibly mediated through the effects of rainfall variation on the accumulation of combustible biomass in the herb layer. There was also evidence of extreme fires occurring twice in the past 1400 years. The results suggest that at the local scale, the vegetation near Kanderi Swamp has been mainly influenced by localised disturbance, rather than climatic shifts or the Rinderpest epidemic, and that elements of cyclical, phase and transition and stochastic models are needed in order to fully understand the observed pattern of vegetation change. [Copyright &y& Elsevier]