A Multivariate Statistical Approach to the Hutchinsonian Niche: Bivalve Molluscs of Central Canada.

The use of multiple discriminant analysis to identify the significant and independent ecological factors separating species distributions is proposed and discussed. Such an analysis was performed on 345 samples, containing a total of 10 bivalve mollusc species, from 32 lakes in Manitoba, Ontario, and Saskatchewan. Measurements of nine ecological parameters were associated with each sample. Five discriminant functions account for 95% of the among species variance, and 4 of the 5 are ecologically interpretable. Three of these, accounting for 80% of the among-species variance, are interpreted as bases of trophic, rather than physical or chemical, separation. There is separation of species on each discriminant function. The use of dissatisfied scores to classify lakes with maximum relevance to species distributions is demonstrated and discussed. A generally applicable measure of environmental heterogeneity based upon this type of analysis is proposed. The value of this type of analysis in quantifying ecological concepts derived from the Hutchinsonian niche model is discussed. An example is given of a reduced available niche resulting in the loss of two species, smaller realized niches for the remaining species, and greater niche overlap.