The individualistic and community-unit concepts as falsifiable hypotheses

This study reformulates the community-unit and individualistic concepts of plant communities as hypotheses concerning the distribution of species' boundaries along a gradient. These hypotheses are tested by an analysis of deviance on data derived from a direct-gradient analysis of a freshwater marsh plant community in Breckenridge, Quebec, Canada. Boundaries are clustered at certain intervals along the gradient (p<0.001), contradicting the individualistic hypothesis. Upper boundaries are not consistently clustered at the same intervals as lower boundaries (p<0.001), contradicting the community-unit hypothesis. Thus, neither of the two usual models of community structure explain the patterns found in Breckenridge Marsh, suggesting that the historical dichotomy is too limited. Hypotheses of pattern should be tested using inferential statistics. Hypotheses of mechanism should be tested by experimentation. The way out of the community-unit vs. individualistic community debate is to deny the dichotomy and to consider multiple working hypotheses of community structure.Nomenclature as in Gleason & Cronquist (1963), Manual of vascular plants of notheastern United States and adjacent Canada, Boston, unless otherwise indicated except for the the grasses which are from Dore & McNeill (1980), Grasses of Ontario, Quebec.The statistical advice of Dr L. Lefkovitch is gratefully acknowledged. We also thank Mme. Claire Gauthier for the use of her property, C. Kettle for assistance with data collection as well as Robin Day and Dr P. Catling for assistance with identification of species. This research was supported by a National Sciences and Engineering Research Council of Canada operating grant.