Detailed analyses of morphological trends in four populations
To assess the differences in braincase height, skull length, braincase width, brain mass and body mass between stages of Dehnel’s Phenomenon and locations, we used ANOVA for each of the five metrics. Dehnel’s Phenomenon-stage was treated as a factor with three levels (summer juvenile, winter subadult, adult). We first assessed the effect of sex on our models, even though in our previous work we found no significant influence of sex on the seasonal changes of these variables (Lázaroet al. , 2017, 2018a). For each response variable (braincase height, skull length, braincase width, brain mass, and body mass) we compared two models using ANOVA: (M1) included season, location and sex and their interactions as explanatory variables. We removed sex from the second model (M2). We based our model selection on Akaike’s information criterium (AIC) and chose M1 as final model for each metric only if it revealed a lower AIC value and the difference between the two models was significant. With the final model for each metric, we used Tukey tests to perform pair-wise comparisons between the factor levels.