S. salpa abundance analysis.

Results of S. salpa abundance model selection with a Generalized Linear Model (GLM). The selected family distribution for the GLM was a negative binomial (with a logarithmic link function) owing to the overdispersion present in the data set. We present the significance of dropping each variable one by one from the full model. The full model was Abundance ∼ Zone+Year+Zone × Year. Akaike’s Information Criterion (AIC) was used to select the best model, evaluating the trade-off between model parsimony and goodness of fit. The lower the AIC the better the model.