Table 6.
Results of the general linear mixed model testing for the impact of the island effect. We used the islandic variable as the dependent variable—scaled with the sympatric variable to remove bias; and we offset the covariates with the variable clade to correct for the impact of inter-clade variations. The covariates in this analysis were the four PCs extracted from the PCA, and the analysis was run under a main effect model. Significant variables are in bold.
| Items | Tests | Parameters Estimates | |||
|---|---|---|---|---|---|
| Wald χ2 | df | p | B | SE | |
| PC1 | 4.74 | 1 | 0.029 | −0.32 | 0.15 |
| PC2 | 10.49 | 1 | 0.001 | 0.40 | 0.12 |
| PC3 | 11.46 | 1 | 0.001 | −0.38 | 0.11 |
| PC4 | 0.13 | 1 | 0.715 | 0.05 | 0.14 |