Table 2. The eight different models based on the density of Hypostomus punctatus at the five sampling sites as the response variable, with all the
possible combinations of independent variables (density of Parotocinclus maculicauda, and first and second PCA axes) were fitted and compared
using Akaike’s Information Criterion and the relative weight (wi) for each model. The best model, with the highest wi value, includes the predictors P.
maculicauda density and the second PCA axis.