Abstract
Relationships between animal distribution patterns and measured physical variables were investigated in a shallow, mesohaline fjord by means of principal components analysis and multiple regression analysis. Twentyfour ~tations were sampled in March, June and October 1973. In all three data-sets the first principal component, accounting for 20-26 % of the total faunistic variation, correlated with the silt and clay content. Depth and distance to the fjord mouth correlated with one or several of the first five principal components, except for the March survey, in which depth did not correlate with any of the components. The first five principal components together accounted for between 49 and 57 % of the total variation. The principal components analyses show that distribution patterns, even though correlated with the same environmental parameters, change throughout the year. On the basis of the nature of these changes, it is hypothesized that they are, at least partly, due to interspecific interactions. Immediately after the main period of larval settling (summer and early autumn) the species have their widest distribution in the fjord. Due to interspecific interactions, among other things, post settling mortality is non-random, and the distributions of most species are hereby limited and changed. Seasonal variations in distributional overlap between species, species richness, evenness and \gB-diversity arc consistent with the hypothesis. The role of stochastic variation is discussed, and it is concluded that community structure, even though influenced by both biotic and abiotic conditions, is not exactly predictable ftom these due to environmental and demographic stochasticity.