Fig. 3

Principal component graphs of biologically relevant ‘levels’. The planes spanned by the first two principal components (PCs) from the habitat utilization (A), leaping performance and posture (B) and morphology (C) datasets are shown with grey symbols representing the mean values of the species and black symbols representing the mean values of the categorical variables. The grey lines connect the species means to highlight cross-level differences. Variable loadings of the corresponding PC analysis (D, E, F) are found right to the PC graph. Loadings of continuous variables represent Pearson’s correlation coefficients and can range from − 1 to + 1. The closer a point falls to the circle’s margin the better is its variable’s representation in this 2D subspace (falling on the margin indicates total representation). Loadings of categorical variables (indicated by asterisks) represent R² values from an ANOVA and can range from 0 to + 1. The point labels in panel F correspond to the numbered morphological variables in Fig. 1C (the point for IMI referring to the effective length measurements used to compute the index)