Introduction:
Fire is near ubiquitous in the tropical savanna biome and is fueled by an understorey of light- demanding C4 grasses. Fire limits tree growth and promotes grass growth which in turn increases the probability of fire spread, thus creating a positive feedback process. As a corollary, fire suppression has been shown to increase woody basal area at the expense of grasses. These data have been interpreted using various models to claim that in the absence of fire at high rainfall, trees outcompete grasses and a forest ecosystem would emerge. We tested this claim for the African continent by simulating tree-grass competition and fire impacts on vegetation using a process-based model.
Material and Methods:
The model simulates the growth and decomposition of a hypothetical tree and grass as influenced by soil water content. The accumulated basal area is further modified by (i) competition for light (aboveground) and soil resources (belowground) and (ii) basal area removal in a fire. To estimate the unknown parameters which determine the strength of competition and fire impacts in the model, we calibrated the model using observed woody basal area and ecosystem states (forest/savanna) at 1287 sites across Africa.
Results and discussion:
After calibration, the model reproduced the observed distribution of forest and savanna ecosystem states across Africa without invoking fire impacts, even at high rainfall sites. However, fire may be necessary to explain some variation in the patterns of observed woody basal area. While fire feedbacks undeniably modify tree-grass ratios, their mere presence does not imply that they heavily influence the emergent ecosystem state. We suggest that models forecasting the effect of altered precipitation on ecosystem state must strive for better representation of tree-grass competition for water.