Nonstationary community theory: Rising to the challenge of long-term environmental change

Background/Question/Methods

Community ecology theory has progressed from models of unstructured populations focusing on point equilibria to a sophisticated array of models and modes of analysis for communities structured in time and space.  Yet a constant in all of these developments is that the physical environment, even when highly stochastic, repeats in time with long-run stable statistical frequencies.  These are so-called stationary environments.Yet the environment on Earth is not like this. Stationarity is but a crude approximation with unknown applicability. With community ecology seeking robust empirical tests of its key hypotheses, and anthropogenic factors rendering stationarity a faint hope indeed, there is an urgency to move beyond the convenient assumptions of the past to community theory that adequately accounts for long-term change.  However, stripping away the stationarity assumption casts doubt on many traditional concepts and modes of analysis. For instance, how are we to define species coexistence, and investigate it in models, when long-term change is built in?  To begin to answer these questions, I studied traditional community models for stationary environments, and removed the stationarity assumption.  I began with the much studied lottery model and broadened the investigation to consider variations on the traditional lottery assumptions.

Results/Conclusions

I found that the effects of nonstationary environments depend on just which life-history parameters are most sensitive to changes in the physical environment, how these parameters are linked to interactions between individuals, within and between species, and whether these parameters respond differentially between species to environmental change. Theory developed for stationary environments could be reinterpreted to provide predictions for nonstationary conditions: long-term change that increased average fitness differences between species predicted rapid simplifications of ecological guilds to single species.  In the lottery model, nonstationary change in fecundity that did not distinguish species allowed the dynamics of a guild to remain stationary even though the environmental inputs where not stationary.  This outcome leads to the concept of buffering of long-term change by the interactions between the organisms. It was possible to classify the kind of change that would be buffered, and to characterize the kind of interactions between organisms that buffer change.  In some cases, buffering is partial, for instance, some deviations from lottery assumptions permitted a kind of quasi-stationary coexistence when all species had the same nonstationary changes in fecundity.  If buffering does not allow quasistationary coexistence to occur on some spatial case, diversity cannot be maintained.