Model selection statistics for the hypotheses considered ordered by small-sample Akaike Information Criteria differences from the top model (ΔAICc).

Legend: Model terms are: enhanced Babine smolt abundance at the full Skeena (Sm(f)) and Babine Lake (Sm(b)) scale, sea surface temperature (SST), and an interaction between the two (x). “Null” is the null hypothesis, “LogL” is log likelihood, and “wi” is Akaike model weight. The evidence ratio [ratio of wi values (for the best model divided by another model's wi)] is a measure of how much less likely a model is compared to the top model given the data and set of models considered. The number to the left of each model corresponds to the numbering of hypotheses in the main text, and “inf” means to infinite.