hide
Free keywords:
-
Abstract:
Observations have been debated as portraying a multimodal distribution of tropical tree cover, even in regions with identical mean annual precipitation (MAP). Previous studies have discussed whether such multimodality is evidence of alternative stable states, which would indicate that tropical forests may irreversibly transition to a savanna-like state when deforestation and climate forcing reach a tipping point. However, doubts have been raised regarding this interpretation. Alternative hypotheses invoke heterogeneous environmental conditions related to soil properties, climate parameters beyond MAP, or human activity. Here, we explore the possibility that the influence of multiple environmental parameters can create multimodality in monostable systems when projected onto one dimension. We show that this situation can indeed occur, even if the system’s only equilibrium state depends monotonically on the parameters, and even if the parameter values have Gaussian distributions. Such a situation would imply that tree cover may respond linearly to forcing, without any abrupt behavior, regardless of multiple peaks in the tree cover distribution. However, when considering ecologically more realistic parameterizations of tree cover, as used in process-based vegetation models, we find that multiple tree cover modes are more difficult to obtain in the absence of alternative stable states. The reason is that environmental conditions do not affect tree cover directly and independently from each other, but indirectly by affecting tree productivity and mortality. Consequently, a coexistence of forest and savanna in these models is only possible when imposing environmental parameters that are bimodal themselves. Motivated by this finding, we analyze the observed distribution of several relevant environmental parameters in South America, and find that none of them suggests a multi-modal tree cover distribution. Our results hence emphasize that possible tipping dynamics are a concern, but also call for improved estimates of tropical tree cover distribution and the role of fire-vegetation interactions.
