The assembly rules proposed by Diamond (1975) sparked a series of studies that have either challenged or corroborated his ideas about how natural communities are structured. However, most of these studies have focused on the role of competition and whether the entire species community is non-randomly structured. An alternative way to determine the community structure is to understand the patterns observed in the association of each pair of species. These associations can be positive (species that co-occur more than expected due to chance), negative (checkerboard pattern, one species generally excludes the other one) or random (no pattern observed). In this study, we analyzed the patterns of species co-occurrence in a dataset including almost 130 species of fishes in approximately 10,000 lakes distributed in more than 100 watersheds in Ontario, Canada. We examined hypotheses of whether there are significant non-random relationships in species pairs of fishes and whether these relationships are consistent across different watersheds. Using incidence matrices, we used null models to test these hypotheses.
Results show many pairs of species that exhibit non-random relationships in their patterns of co-occurrence. Although some species pairs do not co-occur in some watersheds due to biogeographic differences, many of these significant relationships are consistent across watersheds as shown in our meta-analysis. Our work both extends methodological approaches to testing the assembly of ecological communities, but also demonstrates that these temperate lake fish communities show strong evidence of repeatable, non-random composition.