Tropical forest diversity, environmental change and species augmentation: After the intermediate disturbance hypothesis

It is not simple to predict how environmental changes may impact tropical forest species diversity. Published hy- potheses are almost invariably too incomplete, too poorly specified and too dependent upon unrealistic assumptions to be useful. Ecologists have sought theoretical simplicity, and while this has provided many elegant abstract concepts, it has hindered the attainment of more practical goals. The problem is not how to judge the individual hypotheses and arguments, but rather how to build upon and combine the many hard-won facts and principles into an integrated science. Controversy is inevitable when the assumptions, defini- tions and applications of a given hypothesis are unclear. Elegance, as an end in itself, has too often been used to justify abstract simplification and a lack of operational definition. Clarifying and combining hypotheses while avoiding assump- tions provides a potentially more useful, if less elegant, stand- point. An appraisal of Connell's intermediate disturbance hypothesis, and its application to long-term observations from a Ugandan forest illustrates these concerns. Current emphases encourage ecologists to exclude consideration of environmental instability and non-pristine ecosystems. In reality, many environmental changes and eco- logical processes contribute to both the accumulation and erosion of diversity, at all spatial and temporal scales. Site histories, contexts, long-term processes, species-pool dyna- mics, and the role of people require greater emphasis. These considerations reveal that many environmental changes, even those associated with degradation, can lead to a transient rise in species densities. Drawing on related studies, such as forest yield prediction, suggests that the formulation and calibration of simulation models provides the most tractable means to address the complexity of real vegetation. Simulation-based approaches will become increasingly useful both in unifying the study of vegetation dynamics and in providing improved predictive capacity. Quantification of the processes, scales and sensitivities of the dynamics of tropical forest communities remains a major challenge.