Using patterns of variability, to test for multiple community states on rocky intertidal shores

Predictions based on theory of multiple stable states suggest that larger perturbations should lead to more unpredictable patterns of succession. This prediction was tested in the Gulf of Maine using data from 60 intertidal plots of varying size that were experimentally cleared of the rockweed Ascophyllum nodosum and from 14 benchmark sites from throughout the Gulf Rockweed was removed from the experimental clearings ranging from I to 8 in in diameter in 1996 and data collected in 2004 were used to test effects of clearing size and location on divergence and variability in species composition. Benchmark data were collected in 2005, and the 14 sites were from a dataset on 53 sites throughout the Gulf of Maine. The selected sites were randomly chosen from all sites with > 80% canopy cover by A. nodosum and were expected to be similar to uncleared control plots from the experiment. Experimental removal of A. nodosum resulted in clearings at 12 sites within 4 bays. Abundances of gastropods, barnacles, mussels, and fucoid algae and the percentage cover of barnacles, mussels, fucoid algae, bare space, and other species were sampled. CAP and PERMDISP analyses revealed significant differences in multivariate dispersion and variability with both clearing size and location. Variability generally increased with clearing size and location effects were related to the north-south positioning of the sites. Benchmark sites were similar to the experimental control plots but as variable as the largest clearings. Results suggest that succession in larger clearings has been more unpredictable than in small clearings. The pattern of variability in the experimental clearings is consistent with the predictions of multiple stable states. However, the large amount of variation among the benchmark sites was due to mussels and was unexpected. This unexpected variability underscores the importance of sampling benchmark sites as part of experiments. (c) 2006 Elsevier B.V. All rights reserved.