Results of the generalized additive models selected by shrinkage method of the relationship between diet overlap and different explanatory variables.

Models are written Ot = α+s1 (X1,t)+s2 (X2,t)+s3 (X3,t)+…+εt, with si, a nonparametric smoothing function specifying the effect of the covariates Xi on the dependent variable O for year t; α, intercept; and ε, stochastic noise term. The estimated degrees of freedom (edf) for each explanatory variable is indicated as is the significance (** p.Value <0.01, * <0.05 and · <0.10). Variables with edf = 0.00 were shrank by the fitting procedure and thus effectively removed from the formulation. See Fig 2–3 for the model fit and for the confidence intervals of the retained variables.
Results of the generalized additive models selected by shrinkage method of the relationship between diet overlap and different explanatory variables.