4. Discussion
4.1 Decrease in skull and thus overall size over time (1953-2004)
Skull dimensions of juvenile and adult S. araneus significantly decreased during the study period (Table 1). Skull length, the often-used proxy for overall body size, decreased significantly in juveniles between 1953 and 2004 and in adults between 1970 and 2004. Skull height and width also decreased significantly in juveniles and adults. This outcome adds to the very mixed picture of the change in body size in mammals under global climate change. The analysis of 50 species of rodents from seven families revealed no significant temporal change in body size (represented mainly by skull dimensions) in 29 species, a significant decline in 13, and an increase in size in eight species (Nengovhela et al. 2020). Of the three carnivore species in which skull size was measured in the second half of the 20th century and analysed in the context of climate change, skull size showed a significant increase in two species and a significant decline in the other (Yom-Tov et al. 2008, 2010a, 2010b). Thus, S. araneus in Białowieża adds to the number of rather rare cases of decreased body sizes among mammals.
The decrease in body size in S. araneus in Białowieża is in contrast with the observed increased body size (body and foot length) inSorex cinereus in Alaska during the second half of the twentieth century along with increasing temperature. In S. cinereus, the increase was presumably due to higher food availability in winter as a result of improved weather conditions for its prey (Yom-Tov and Yom-Tov 2005). We argue that the smaller size of S. araneus in summer could be an effect of increasing drought leading to reduced food availability. The negative impact of decreased soil moisture on the abundance of epigeic and endogeic invertebrates has been well documented (Coyle et al. 2017; Singh et al. 2019). S. araneus strongly depend on the abundance of food because their metabolic rates measured under natural conditions equals 258% of the value predicted for an average mammal of the same body mass (Ochocińska and Taylor 2005). Drought leading to soil moisture deficits may specifically impact earthworms, the main food source of S. araneus . S. araneusprefer humid habitats with humus-rich soil, as in our study plot, which harbour large numbers of earthworms (Shchipanov et al. 2019).
In support of this, the decrease in skull length in adults between 1970 and 2004 was significantly correlated with the increase in soil moisture deficit, which started to increase in the Białowieża Forest in the 1970s and accelerated in the 1980s (Figures 4e and S2e) as a result of decreasing precipitation (Figures 4d and S2d) and increasing temperatures (Figures 4C, S2a, S2b). The decrease in skull height in both juveniles and adults was correlated with an increase in the mean lifetime temperature (Table 2) and was stronger after 1970, which coincided with the inflection point in precipitation and soil moisture deficit.
There were no regular measurements of the soil water level in the Białowieża Forest before 1985, and those that existed were not comparable with later measurements. However, between 1985 and 2001, the water level decreased in mixed deciduous biotopes by 40 cm. This resulted from an increase in temperature, a decrease in rainfall, and winters without snow and thus a lack of water accumulation at the beginning of the vegetation period (Pierzgalski 2002). The progressive decrease in the soil water level was certainly magnified by the intense drainage work on peatlands in the eastern Belorussian part of the Białowieża Forest in the late 1950s. The average biomass of earthworms in the upper soil layer in the mixed deciduous areas of the Białowieża Forest was 42.2 g m-2 between 1997 and 2000 and depended on soil moisture: it was highest in spring, decreased to 22% of the spring values in summer, and increased to 41% in autumn (Kowalczyk et al. 2003). This pattern followed the annual course of soil humidity in the Białowieża mixed deciduous forest (Pierzgalski 2002). Earthworms were not available in the forest in winter (Kowalczyk et al. 2003).
While the increasing lifetime temperature affected both juveniles and adults, resulting in lower skull heights, the adults were additionally affected by the conditions in the preceding winter. Milder minimum temperatures in February and decreasing number of days with snow cover in the coldest months of January and February resulted in adults with lower skull heights (Table 2). The lack of snow cover leads to freezing of the upper soil layer and substantially reduces the abundance of various arthropods, which are important components of the winter diet inS. araneus (Templer et al. 2012). Snow generates a relatively mild and stable subnivean microclimate in the litter layer where the shrews live (Coulianos and Johnels 1962; Churchfield et al. 2012) that is not only favourable for their prey but may also diminish the costs of thermoregulation and consequently the food requirements of the shrews (Aitchison 1987). However, the importance of the insulative value of snow for adult S. araneus is less convincing, as concluded by the weak relationship between skull height and the number of days with snow cover at least 5 cm thick in January and February (Table 2).
Skull dimensions in juvenile S. araneus from the northeastern part of European Russia were correlated with temperature and precipitation from 1976 to 2003, but in contrast to our results, there was no clear directional trend to the change (Poroshin et al. 2010). This might have been caused by the lack of clear change in weather parameters and/or the shorter time period covered in that study than in ours.
Recent evidence suggests a relationship between the tendency of body size to decline over time and metabolic rates in small mammals. In rodents that do not use torpor, there is a significant negative correlation between the temporal body mass change and basal metabolic rate (BMR; Villar and Naya 2018). Species with a high BMR have become smaller in the last six decades. The size decrease in S. araneus , with their lack of ability to hibernate, and the BMR in juvenileS. araneus, which equals 278% of the value predicted for an average mammal of the same body mass (Ochocińska and Taylor 2005), fits this pattern. It is likely that the relationship between the body size decrease and high metabolic rate is generally true for soricine shrews. Three out of four Sorex species (not including S. cinereus ) had a decreased body mass during the 20thcentury, and two of them had a significantly decreased body mass (Naya et al. 2017). However, data from different ages and thus Dehnel stages were pooled in this study, potentially affecting the results and hiding the effects of seasonal body size changes.
4.2 Changes in the magnitude of Dehnel’s phenomenon (1955-1985)
We found that the decline in skull height from summer to winter inS. araneus from the Białowieża Forest became stronger between 1955 and 1985 (Figures 6a and b). We standardized skull height with skull length in the analysis of Dehnel’s phenomenon to describe changes in relative skull height. Relative skull height is strongly correlated with brain size (Bielak and Pucek 1960; Lázaro et al. 2018). Thus, our results also reflected changes in the magnitude of brain size changes. Initially, the skull height of juveniles did not differ between July and August from 1955 to the 1970s (Figure 6b). The skull height in these two months then diverged over the following years (Figure 6b). The smaller skull height in August than in July after 1970 meant that the decrease in skull height started earlier. The magnitude of the decrease was also higher because the relative skull height of winter subadults decreased even more over time (Figure 6).
What is the environmental background of these temporal changes? The increasing magnitude of Dehnel’s phenomenon was largely due to the significant decline in the skull height in January and February subadults (Figure 6). Unfortunately, the numbers of available subadult specimens each year were too low to statistically correlate them with weather conditions. The significant decrease in January temperatures until 1972 might explain the increasing magnitude of Dehnel’s phenomenon, but January temperatures did not change significantly later on. A seasonal decrease in skull height in S. araneus is thought to be anticipatory of harsh winter conditions. The question then arises whether weather conditions in summer and autumn can have direct or indirect impacts on the skull height in winter. Mean daily July temperatures did not change from 1955 to 1985, but August temperatures significantly increased in this period. One may speculate that increasing temperatures and possibly drought could decrease the availability of earthwormsand other invertebrates in August and speed up the decrease in skull height over time (as discussed above), leading to a decrease in skull height in winter. Lower amount of invertebrate food in summer could lead to a greater reduction of resources in winter. Although we did not find a significant change in the soil moisture deficit in August between 1955 and 1985, the sharper skull height decrease with the increasing mean temperature in the driest quarter of the year in a geographic comparison of S. araneus (Lázaro et al. 2021) supported the hypothesis that soil moisture deficit might be an important factor for this species. To elucidate this, the role of summer and winter availability of food in Dehnel’s phenomenon should be tested under controlled laboratory conditions.
We also found a significantly greater skull height regrowth from 1955 to 1985, which resulted from a steep decrease in skull height in the subadults, while adults did not change (Figure 6). Almost nothing is known about the drivers of this regrowth, which are presumably different from those of the decrease and are likely associated with preparation for the terminal reproductive period in shrews (Lázaro et al. 2019, 2021). Comparison of the skull height regrowth in different S. araneus populations revealed only one correlation with weather parameters, namely, a positive correlation with precipitation during the warmest quarter of the year (Lazaro et al. 2021). This also points at changes in food availability.
In summary, we documented a decrease in overall size as represented by absolute skull height and length in the 52-year (1953-2004) skull series and identified increasing temperature and drought as the main correlates of the decrease. Interpreting the results regarding Dehnel’s phenomenon, which increased in magnitude because of a sharp size decrease in winter subadults, was more difficult. Our dataset about winter subadults stopped shortly before the time point when climate change became significant, and it would have been interesting to see how subadult size developed after this. There are strong indications that there are important effects of climate on food availability that indirectly cause trends in both skull size and the magnitude of Dehnel’s phenomenon. However, this must be clarified experimentally.
With continued climate change, all the parameters that we found to be of importance for the size of the common shrew and its unique winter adaptation are expected to increase even more strongly. The shrew’s distribution ranges may be affected by this, as many regions may become too hot and/or dry for these and other cold-adapted animals.