Rates, patterns and causes of vegetation change in the Gran Sabana, Venezuela
Final Report Abstract
Savannas are defined as tropical ecosystems with a continuous grass layer and a discontinuous tree layer. At high rainfall sites savannas border onto forests, or exist in a mosaic of forest and savanna. In this study we investigated whether climate, fire or management practices can explain changes in the cover of forest and savanna vegetation in tropical regions with seasonal climates. Understanding these controls is critical for predicting future patterns of carbon storage in tropical systems. Our study system, the Gran Sabana, Venezuela, is a high rainfall site where savanna and forest form a mosaic. Previous work is ambiguous as to the role of edaphic factors, climate, fire and management controls on savanna - forest dynamics. We used remote sensing, stable carbon isotope methods to describe both recent and longer term changes in forest and savanna cover. We used simulation modelling to help us interpret vegetation dynamics. We developed a new vegetation map that provides a detailed estimate of the area covered by the regions’ vegetation formations. We furthermore use statistical methods to illusrate that the a series of interacting environmental factors control the distribution of forest and grassland in the Gran Sabana. In some environmental settings forest or grassland appear deterministic, in other setting both forest and grassland can occur. We further created a time series of vegetation state in the Gran sabana and performed a land cover change analysis on this time series. This analysis revealed that the trajectories of change were local and idiosyncratic and that overall no clear trend existed. This contrasts with anecdotal accounts which suggest that forest loss is prevalent. Stable isotope analysis allowed us to reconstruct vegetation change over longer time horizons. Here we report that many grasslands were formally forests. We also found that the rate of change from forest to grassland has been remarkably constant over the last 10 000 years. Vegetation modelling allowed us to illustrate that fire and climate ensure that vegetation state can be bimodal (either forest or savanna) at many sites in the Gran Sabana. Yet soil fertility was found to be an important regulator of vegetation state. Furthermore we found that simulations that assumed that fire adapted trees are absent from the Gran Sabana better matched tree cover estimates derived from satellite imagery. This study has used a range of methodologies to explore controls on forest-grassland-savanna mosaics. All perspectives were consistent with the hypothesis that forest and savanna are alternative stable states in this region, while at the same time revealing that vegetation has been changing slowly and in the direction of forest loss over the last 10 00 years.
