Appendix C. Results of maximum-likelihood models controlling for spatial and phylogenetic autocorrelation and explaining lizard population density (N, lizards/ha) as a function of body mass (M), metabolic costs (Bx, where B0 incorporates the crudest and B3 the most refined environmental temperature data), and minimum productivity (Pmin).
Notes: To illustrate, in the case of the energetic use and availability models, the models take the form: log(N) = α1 log(M) + α2 log(Pmin). The AIC values for each nested model are compared to that of the non-spatial and non-phylogenetic null models (N~1) using likelihood ratio tests (LRT). All models are significantly better than the null model at P < 0.001. For spatial autoregressive models, larger absolute values of Moranís I indicate higher spatial autocorrelation. Spatial autocorrelation is significant for all models. For phylogenetic generalized-least squares models, Pagelís λ ranges between between 0 (phylogenetic independence) and 1 (complete phylogenetic determination). LRT are used to assess whether λ deviates significantly from 0 (*P < 0.05, **P < 0.01, ***P < 0.001).