Deciphering environmental controls over the hysteresis of biome switches at savanna-forest boundaries

Savannas and evergreen forests are the most important tropical vegetation types in terms of area, biodiversity, total carbon stocks, and use by humans, so shifts in the distribution of these biomes have large implications for conservation, global change, and human welfare. Predicting these shifts, however, is challenging. Some modeling studies forecast widespread expansion of forest into tropical savanna, while others forecast massive collapse of tropical forest. The difficulty of predicting and managing savanna-forest transitions arises largely from complex fire-vegetation feedbacks that are influenced by humans, climate, previous fire history, and the characteristics of plant species at a site. An improved understanding of these feedbacks is necessary for developing the next generation of earth system models and for informing management of the seasonally dry tropics. The latter is particularly true in Brazil, where a failure to recognize the natural role of fire has led to policies of fire suppression in areas of native savanna, resulting in forest encroachment into many areas of natural savanna and loss of species that depend upon this habitat. In addition to providing valuable data for managing these systems, this project will promote the acceptance of the natural role of fire with an article to be prepared for a popular science magazine and a white paper to be distributed to managers of Brazilian savanna reserves and state ministers of environment. These documents will review the evidence for the natural role of fire and the consequences of fire suppression. As the project generates results, journalists will be invited to visit the study sites to inform the public of the natural role of fire in Brazilian savannas. Four US undergraduate students will be given international research experiences that will contribute directly to the research objectives of this proposal while stimulate interest and expertise in tropical ecology among US students. Understanding the impacts of fire and fire history on vegetation dynamics is important for fire management in the US.

Understanding of the drivers of species change at the forest-savanna ecotone has been hindered by strong positive feedbacks and other non-linear processes that cause complex dynamics and hysteresis (i.e. the dependency of system not only on its current environment but also on its past environment). These legacy effects confound our ability to observe vegetation responses to environmental change. The proposed research will combine field data and modeling to test for and quantify sources of hysteresis in savanna-forest dynamics. The aims of the research are (1) quantify key processes that underlie switches between savanna and forest states, (2) use this information to refine and parameterize the CLM(ED-SPITFIRE) model for simulating savanna-forest dynamics, and (3) perform simulations to understand environmental controls on the distribution of tropical savanna and forest, with emphasis on causes of hysteresis. This model represents the coupling of a demography-based ecosystem model (ED), a mechanistic model of fire occurrence (SPITFIRE), and a model of land surface processes (CLM) that can optionally be coupled to a general circulation model (CESM). Much of the data needed to refine, parameterize, and validate the model are available from previous NSF-funded research. Gaps in this information will be resolved with a fire experiment at a natural savanna-forest boundary in Brazil. In the experiment, flammability trials will be used to understand thresholds that mediate vegetation-fire feedbacks. Measurements of biomass production and turnover will quantify the relative roles of primary productivity and mean residence time in governing biome shifts. Monitoring of tree dynamics will provide a more complete understanding of the distinct roles of savanna and forest tree species in these shifts.