Fig. 3. Individual differences in niche centre between the short- and long-term assays of food preferences. Correlations have been estimated with a bi-variate Bayesian Mixed Models including period (short-term vs long-term) as fixed effect and individual as random effect. Correlations are the mode and HPD intervals (within brackets) estimated by means of a bi-response Bayesian mixed model of PCoAs in the short- and long-term assays. Dots with no line indicate individuals for which no long-term assay was conducted.