2.3 The choice of months representing the three age categories
We compared skulls in three age categories of individuals that characterize their entire lifetime: fully grown juveniles, size-decreased subadults, and regrown adults (Figure 3). To test the homogeneity of each age category, we used an ANCOVA model with the month as a factor, the time (year) as a covariate, and the interaction between month and year. The skull height of juvenile individuals differed between July and August (F 1,201=8.08,P =0.0049), and the interaction was significant (F 1,201=8.12, P =0.0048) in the period between 1953 and 1981, when individuals from both months were measured. Thus, we further analysed July and August juveniles separately.
The interaction between month and year was not significant in the analogous ANCOVA model for skull height in subadults from December, January, and February (F 2,82=0.75,P =0.476). However, skull height had not fully decreased in December and was higher than in January and February when we removed the interaction from the model (P =0.0159 and P =0.0426, respectively). The skull height did not differ in the model that included only the month and year for skull height in subadult shrews from January and February (F 1,67=0.55,P =0.461), and we used only these months for further comparisons (Figure 3). In these two months, the skull height was lowest in the lifespan of S. araneus (Dehnel 1949; Pucek 1970). Moreover, the lack of an interaction between month and year in December, January, and February indicated that skull height reached a minimum in January irrespective of the year. In adults (Figure 3), neither the year × month interaction nor the month was significant in the model for skull height (F 4,104=0.88, P =0.476;F 4,104=0.88, P =0.477). The month was also nonsignificant (F 4,108=1.67, P =0.163) when the interaction was removed from the model. Thus, we retained all months from April to August in the final analyses of adults.
We ran the same tests for skull width and length, which varied much less between seasons (Lázaro et al. 2017). The year × month interaction was not significant in any age category, and there were no significant differences between collection months. See Figure 4 for an overview of analysed S. araneus data.