The effects of environmental variables on total incubation (%) and female share of incubation (%).

Analysis for the full day (0–24 h), daytime (6–18 h) and night (18–6 h) data are shown separately.Notes.The full models included time period, ambient temperature, breeding site (mainland, island) and time period × temperature as fixed terms. The effect of temperature was estimated separately for each population by a random slope term. Nest ID was in the models as a random intercept term to control for pseudoreplication. Temperature was a second degree orthogonal polynomial. The significance of each predictor was assessed by eliminating it from the full model and comparing the fit of the two models using likelihood ratio test. Population effect was tested in two ways: (i) by removing the random intercept and slope term from the model, (ii) by replacing the random intercept and slope term with a random intercept term in the full model and removing this term. Temperature was tested by removing temperature, period × temperature and the random slope term for temperature from the model. The slope difference for temperature between populations was tested by removing the random slope term and keeping only the random intercept term in the model. The quadratic effect of temperature was tested by replacing the second degree orthogonal polynomial term with a linear term.