Results of the regression relationships between functional diversity and species diversity for the linear, power, asymptotic, logistic and through the origin models for each functional classification scheme (DS, DSH and DSHG).

DS: diet × size; DSH: diet × size × home range; DSHG: diet × size × home range × gregariousness.
R2: % variance explained by the models (least square fit), NLL: negative log-likelihood, AIC: Akaïke information criterion, BIC: Bayesian information criterion.
Larger R2, smaller NLL, AIC, BIC indicate better fit (italic).
*: Note that linear through origin models can have a negative R2 when the variation around the regression line is greater than the variation around the mean. This occurred for the DS classification for which we set R2 = 0 (rather than negative) to follow the usual convention. In view of this, the log-likelihood methods (NLL, AIC, BIC) were preferable to least-square fits for model comparisons.