The importance of scale in testing the origins of alternative community states

The possibility that different species assemblages may represent persistent alternative community states remains largely unexplored by experimental ec ologists because of a variety of conceptual and experimental problems. We d iscuss some of the conceptual roadblocks to experimentation and propose sev eral avenues for attacking the problem experimentally. We address the conce ptual issues involved in (1) the blurring of the distinction between the pr ocesses that initiate the switch among alternative states and the positive- feedback processes that maintain those states, and (2) the role of spatial scale in initiating the switch. We suggest that the switch between alternative states requires, first, a di sturbance that removes species involved in the positive feedbacks needed fo r maintenance and, second, the arrival of other individuals that initiate t he switch to the alternative assemblage. The removal of the species that ma intain the system must be large enough and over a long enough time to allow the arrival and establishment of members of the alternative assemblage, an d so we hypothesize that the switch among alternative states is scale depen dent. This scenario suggests that the switch among alternative states can be inve stigated experimentally through the manipulation of the scale of the distur bance and of the arrival of members of the alternative state. Small-scale d isturbances should consistently fail to initiate a switch, while larger-sca le events should initiate a switch at least part of the time. We also note that in some cases the scale of disturbance and/or the arrival of recruits cannot be manipulated or controlled and suggest that several approaches oth er than factorial experiments with ANOVA, such as spatial autocorrelation m ethods, may be useful. We illustrate the potential and the difficulties of various approaches by d iscussing two systems in eastern North America that may contain alternative states. Mosaics of mussel beds and algal beds occupy rocky coasts from New England northward, and patchworks of forests and heathlands occur in easte rn Canada and in the Appalachian highlands. While the study of alternative states in the marine system can be approached experimentally, the scale of disturbance required to switch forests to heathlands is too large for exper imentation and must rely on the use of other approaches.