Summary results of model selection for the relations between fish metrics and environmental gradients reflected in each of the two components extracted from principal component analyses of regional (R) and local (L) variables.

For each response variable (fish metric), the table provides the cumulative Akaike weight (Σwi), calculated by summing the Akaike weights of the N best approximating models (Δi ≤2), Nagelkerke r-square of the top-ranked model (R2), and model-averaged estimates of GLM coefficients (unconditional 95% confidence intervals in parentheses) for the gradients included in the model (and quadratic term). Coefficients where the 95% confidence interval does not overlap zero are represented in bold.