• Corresponding Editor: D. P. C. Peters


Do communities return to their former state when we disturb them? The answer is “surely not always,” since some disturbances may be so devastating that recovery will be impossible. If communities do recover, then how fast is that recovery? Do different subsets of species return at the same rate? Is that rate a simple exponential recovery— meaning that the change toward the original state is fastest when the community is furthest away and it slows as the community converges? Or is recovery a more dynamically complex process? These questions are theoretically interesting and practically important. The theoretical questions are if there is a particular state—some exact composition—to which a community is likely to return, if there might be several (or many) possible such states, or if community composition is essentially haphazard. The practical implication is that if disturbed ecological communities do not tend to return to a previous state, it may be impossible to undo human impacts on natural ecosystems.

We follow the fate of species assemblages following the removal of vegetation for mining. We show that these assemblages in restored subtropical coastal dune forests in South Africa do converge with a regional equilibrium state and that convergence is possible within a reasonable period. However, changes in assemblages from different trophic levels were idiosyncratic: convergence in the dung beetle assemblage did not mimic convergence for trees and birds, for example. Few of the assemblages converged exponentially, the simplest shape for the decay function. Furthermore, trends were sometimes different for different indices of community dissimilarity, suggesting that whether one accepts convergence depends, in part, on exactly what one measures.