Collaborative Research: The Role of Spatial Interactions in Determining the Distribution of Savanna and Forest

Princeton University, Princeton NJ

Investigators

Abstract

Ecologists have long emphasized climate as primary factor determining ecosystem dynamics at large scales, but the relationship between climate and vegetation is not always deterministic. Where vegetation-environment feedbacks are significant (e.g., in savannas), predicting responses to climate can be especially difficult. Recent work suggests that feedbacks with fire may make savannas much more common than they would be were their distributions solely determined by climate; as a result, savanna and forest responses to global change may thus be drastic, sudden, and difficult to foresee. However, existing work does not explain why savanna is spatially aggregated with savanna and forest with forest -- a pattern that indicates that spatial processes may also play a role in determining ecosystem responses to climate. Here the investigators will consider the impacts of those spatial interactions between savanna and forest on their distributions and on their potential responses to climate and land-use change, both in the past and into the future. The investigators have identified three hypotheses to explain the spatial aggregation of savanna and forest: H1) that savanna and forest are bistable, and that spatial structure in initial conditions (as a result of past climate) determines their distributions, H2) that savanna and forest are bistable, but that spatial processes within savanna (e.g., fire spread) result in spatially structured distributions, and/or H3) that nearest neighbor interactions between savanna and forest change their distributions on long time scales, impacting their long-term stability. These hypotheses are variously supported in the empirical literature, and existing work has not attempted to disentangle these processes. Results will allow the research team to generate informed theoretical and empirical predictions about the past and future distribution of savanna and forest globally. The proposed work will also generate novel mathematical results. Possible outcomes of theoretical, spatial-stochastic models include a) savanna and forest coexistence in landscapes, b) forest exclusion by savanna, c) savanna exclusion by forest, or d) alternative stable states in biome savanna/forest dominance. The last outcome would be unlikely in a homogeneous spatial stochastic model, where the winning biome is decided by the direction of movement of the biome boundary (i.e., the front), but most closely resembles real biome distributions. Spatial stochastic model results will be reconciled with observations using theoretical and simulation modeling.