Statistical analyses
We used R 4.1.0 (R Core Team 2020) for statistical analyses. First, to explore how the colour parameters were inter-related, we performed correlations between UV chroma, carotenoid chroma and brightness both at the individual level and at the nest level (the latter using mean values of colour parameters). Second, we fitted three linear mixed models with a normal distribution of errors using the lmer function in the “lme4 ” package (Bates et al. 2015) to determine the relationships between body mass and each of the three colour parameters. The models included as fixed effects the average body mass of the brood (= among-nest effect), the deviation from the average body mass of the brood (= within-nest effect), and their interaction. We included in addition year (2017, 2018 and 2019), nestling sex, brood size, and the interactions between year and nestling sex, average body mass and year, average body mass and brood size, and nestling sex with the deviation from the average body mass. Backward elimination for non-significant interactions (α = 0.05) was used to build the minimal models. We also included nest ID as a random intercept, and the interaction between nest ID and the deviation from the average body mass (= within-nest effect) as a random slope.