Figure
4: Trait loadings on the leading vectors of ID (blue points)
and IDsex:mass (red points). Error bars denote
95% confidence intervals obtained using a parametric bootstrap (n=5000)
and loadings can be considered nominally significant if CI do not cross
zero (solid vertical line).
Figure 5: A graphical illustration of the relationship between
multivariate personality, mass, and sex. Each point represents an
individual’s predicted behavioural score idmaxplotted against mean weight with colours denoting assigned sex (female,
male, unknown). Behavioural scores are calculated for each individual asidmax.i T where i is
a column vector containing the best linear unbiased predictions (BLUPs)
of individual deviations from each trait mean. Also shown for
illustrative purpose are overall (grey dashed) and sex-specific
regressions of behavioural score on mean weight. Behavioural scores are
(uncertain) model predictions and no statistical inference is intended.