Differences in seasonal skull dimensions and body mass between four populations of the common shrew
In our own previous work we used tooth row length measured from X-ray images as a factor to correct for individual size variation as it remained constant throughout the shrews’ lifespan once summer juveniles were fully grown at our study site in Southern Germany (Lázaro et al. , 2017). Here, to use a calliper measurement in collection specimens (in some collections X-rays were not possible) we measured mandible length as a proxy to that correction factor. However, when looking at three additional populations (Žofín, Gugny and Białowieża) we found that mandible length varied between seasons (d.f. = 186, adj.R2 = 0.19, F = 5.3, P (seas.) < 0.05, P (loc.) < 0.001, P (seas.:loc.) > 0.1). Results for size-corrected and absolute values did not significantly differ in Radolfzell. Consequently, we compared absolute values between the four populations. First we tested for the effect of sex, but found no significant effect of sex and its interactions on braincase height for all locations in the model comparisons (AIC(M1) = -79.2, AIC(M2) = -67.8; ANOVA, P> 0.5). Thus, we excluded sex from further comparisons of skull dimensions. This is interesting, as even though differences in behavior and energetic pressure should exist particularly during reproduction in the adults, and some sexual dimorphism was found in mandible morphology of S. araneus (Nováková & Vohralík, 2017).
In the final model M2 (d.f . = 200, adj.R2 = 0.78, F (season) = 155.7,F (location) =146.6, F (interaction seasons X location) = 1.3), we found a difference between seasons and locations at the factor level (P < 0.001 both), but not their interaction (P > 0.1). The Tukey test revealed a decrease in braincase height from summer juveniles to winter subadults (P< 0.001) and an increase from winter subadults to adults (P < 0.001) at each location (Fig. 2, Table 3). Braincase height values for all seasons combined were highest in Gugny, followed by Radolfzell, Žofín and Białowieża (P < 0.05 in all pair-wise comparisons, Table 3). Thus, shrews from the four populations differed in size, but the magnitude of Dehnel’s Phenomenon did not. Our analyses of data from the literature confirmed a more pronounced decline in braincase height towards Northeastern populations at a large geographical scale. However, our review also revealed large levels of variation in winter size decline between populations within small areas (e.g. northern Germany (Schubarth, 1958)) and similar decline values in widely separated populations (e.g. southern Germany and central Finland (Skaren, 1964; Lázaro et al. , 2018a)). Our four focal populations did not follow the predicted pattern but they are all situated in central Europe and habitat differences might not be strong enough to cause the variation observed at a larger scale. Interesting is that size, as measured by braincase height did not follow the expected pattern either. Sorex araneus is smaller with increasing latitude in direct contradiction to Bergmann’s rule. However, the two neighboring Polish populations differed more in size than Gugny (Northeastern Poland) and Radolfzell (Southern Germany), which were almost identical.
As previous studies had also found (Dehnel, 1949; Lázaro et al. , 2017, 2018a), braincase height changed most strongly than other skull metrics in all populations. Thus, we focus on results for braincase height here. To summarize, we only found a slight winter decrease in skull length and a spring increase in braincase width, only in Radolfzell. More details of the results for skull length and braincase width can be found in the Supporting Information (see Table S2).
Again similar to the results on braincase height, we found few differences in body mass between the more closely investigated populations (Radolfzell, Gugny and Žofín). Again we found no significant effect of sex on body mass variation between seasons (AIC(M1) = 290.3, AIC(M2) = 293.6 ; ANOVA, P > 0.1) and pooled data of males and females in all analyses. Body mass differed significantly between seasons and locations at both factor and interaction levels (M2,d.f .=116, adj. R2 = 0.88,F (seas.) = 424.2, F (loc.) = 13.8, F (seas.:loc.) = 2.8, P (seas.) < 0.001,P (loc.)>0.001, P (seas.:loc.) > 0.05). All three populations decreased from summer juvenile to winter subadult followed by a pronounced regrowth to adult (Table 3, Fig. 3, Tukey test, P < 0.001 for all populations). Body mass was similar in juveniles and adults in all populations, but winter subadults from Žofín were lighter (P < 0.001). Žofín is the only high-altitude population in our dataset. Mountain populations suffer harsher winter conditions and we expected Dehnel’s Phenomenon to be stronger in shrews at higher altitudes. The stronger body mass decline we found in Žofín supports Dehnel’s Phenomenon as a seasonal adaptation. However, we do not see a matching difference in braincase height decline. This might mean that changes in body mass are more sensitive to local environmental differences and/or current conditions. For example, there is the little evidence for winter body mass decline in Norway (Frafjord, 2008), but a 27% decline found at similar latitudes in Finland (Hyvärinen & Heikura, 1971). Alternatively, given that data from the various sites were collected during completely different years, seasonal changes in body mass may have resulted from other causes independent from Dehnel’s Phenomenon, for example, winter malnutrition or non-adaptive changes.