DIFFERENTIATION DEGREE AND DEVELOPMENTAL STABILITY OF CHIONOMYS GUD (RODENTIA: CRICETIDAE) BY NONMETRIC CRANIAL TRAITS

The work describes the results of assessing developmental stability and the degree of differentiation in Chionomys gud within Protected Areas of the Central and Western Caucasus. We characterised developmental stability in Chionomys gud in different eco-geographical conditions by means of fluctuating asymmetry levels (FAnm) in a series of cranium phenes in four geographical populations. Thus we estimated the effect of environmental factors on FAnm levels. We registered the highest FAnm value for the sample from middle mountains of the Central Caucasus, i.e. the vicinities of the village Elbrus. The environment conditions of the given sample were the most stressful for the species (annual average temperature below 5°C and annual average precipitation below 1000 mm). Two-way ANOVA for factors «annual average temperature» and «annual average precipitation» verified a correlation between FAnm levels and complex effects of the two climatic factors. We evaluated the levels of interpopulation phenetic distances by the analysis of frequencies of phenes of nonmetric cranial traits in Chionomys gud populations. Maximum phenetic distances were revealed between geographically distant (over 300 km) populations of the Central and Western Caucasus. The degree of morphological differentiation of the samples from the Central Caucasus, which are apart at the absolute distance of 20–60 km, was less pronounced. Our results suggest that the level of phenetic distances is determined not only by the effect of territorial isolation, but also by climatic peculiarities of a particular locality.

In this context Protected Areas, which are under maximum protection and represent model plots, are of significant interest as test areas for long-term studies and monitoring of the nature population state. According to Zakharov et al. (2000), besides the assessment of anthropogenic impact, it is necessary to monitor natural changes of environment. Corresponding monitoring is essential for the understanding of general trends of environment changes on both local and global scales, and also for integral estimation of the environment quality and state under the combination of effects.
Priority objects under monitoring are sensitive and fragile elements of biological assemblages, i.e. rare and endemic species as their restricted distribution makes them extremely vulnerable to unfavourable natural and anthropogenic effects. Besides, endemics are often small in numbers and found as isolated populations; they prefer specific habitats, and that is probably the reason of their extinction. Chionomys gud Satunin, 1909 refers to the abovenamed species. Chionomys gud is a typically mountain mammal and endemic within a mountain area in the Caucasus; it is also a stenoecic species which occupies rocky biotopes of middle mountains.
The aim of the present work is to study the levels of fluctuating asymmetry (FA) in nonmetric cranial traits of Chionomys gud and reveal environmental factors, which determine FA, and also to assess the relative state of individual populations according to the quality of their habitats under different eco-geographical conditions of the North Caucasus.

Material and Methods
We collected the material in the Western Caucasus within Shaposhnikov Caucasus biosphere reserve (Adygea, the vicinities of the Lago-Naki Plateau) and in the Central Caucasus within the Kabardino-Balkar State High-Mountain Nature Reserve (vicinities of the village Bezengi, Ushtulu Landmark Area) and National Park «Prielbrusye» (vicinities of the village Elbrus) (Fig. 1). According to typification of altitudinal zonal spectra which was developed by Tembotov (Sokolov & Tembotov, 1989;Tembotov et al., 2001), these localities refer to different variants of altitudinal zonation and vary in climatic conditions. The Lago-Naki Plateau refers to the Kuban variant of altitudinal zonation, the village Elbrus -to the Elbrus variant of altitudinal zonation, and the village Bezengi and Ushtulu landmark area -to the Tersk variant of altitudinal zonation. Differences between variants of altitudinal zonation are determined by orographic peculiarities of the given areas, presence and location of mountain ranges which influence on distribution of arid and humid air masses, which, in turn, form typical climatic conditions of each variant of altitudinal zonation. So, air masses of the Mediterranean and Black Sea basin, and of the Atlantic Ocean determine the more humid, warm and mild climate of the Kuban variant of altitudinal zonation. The Elbrus variant is more arid and continental as compared to other variants involved, and that is due to the structure of front ranges. In the Tersk variant, the overall elevation of the front ranges is higher, and they represent a barrier to hot winds of the Caspian Depression, thus reducing the influence of the north-eastern hot winds on mountain landscapes. The Tersk variant of altitudinal zonation is correspondingly characterised by more humid and milder climate compared to the Elbrus variant.
The sampling localities are within the subalpine belt. Table 1 shows certain physiographic sampling localities. As follows from Table 1, within the studied areas the background gamma-radiation level is low (under 0.5 μ Sv/h), and normal for humans as it offers no pathological effect.  We studied 115 Chionomys gud skulls and classified them according to 27 phenes of nonmetric threshold traits.
We calculated phenetic distances as mean measures of divergence (MMD) within and between populations, and their mean standard deviations (MSD) according to Hartman. The differences were statistically significant at p < 0.05 when MMD > 2 MSD. We also used the index of mean measure uniqueness (Vasil'ev et al., 2000).
Assessing phenetic diversity of populations we applied standard indices of resemblance and identity according to Zhivotovskiy (1982). We estimated sampling resemblance by r index, and identity index I to consider significances of deviation between the pairs of samples. The I value distribution is approximately the same as χ 2 value with m-1 degrees of freedom under null-hypothesis which testifies that both samples refer to the united sampled population. We applied the table χ 2 values for assessing differences between the samples.
To determine the level of FA we calculated the average population index of FA in nonmetric skull traits (FA nm ) as the average proportion of asymmetrically expressed bilateral traits per individual (Zakharov, 1987;Markowski, 1993).
We processed the results statistically with PHEN 3.0 (Vasil'ev, 1995), PAST (Hammer et al., 2001) and STATISTICA 10 software packages. We estimated the significance of differences between the samples by the nonparametric Kruskal-Wallis test which is an alternative to one-way ANOVA (Vasil'ev, 2005). Two-way ANOVA was applied for revealing the effects of the factors.

Results and Discussion
Sex differences by FA level were insignificant, but in all the samples FA nm was higher in females than in males, suggesting that females are more responsive to the environment. As there were no significant differences, we had to pool female and male samples for further analysis. Table 2 and Table 3 present higher FA values in the animals from the vicinities of the village Elbrus in the Central Caucasus (CC) and Lago-Naki Plateau in the Western Caucasus (WC).
Among all the studied areas the first sampling locality (village Elbrus) is characterised by severe climatic conditions (annual average temperature of 3.8°C, and annual average precipitation -920 mm). On the contrary, the second sampling locality (Lago-Naki Plateau) is the warmest and with maximum precipitation (annual average temperature of 5.7°C, and annual average precipitation -1485 mm).
We registered the lowest FA nm values in Chionomys gud from Ushtulu landmark area. As FA nm values serve to make an indirect assessment of the conditions required for animal development, so we can suggest from the data obtained that Chionomys gud habitat is more favourable within Ushtulu Landmark Area (Table 1).
On the whole, we revealed statistically significant differences by FA nm level only between a pair of samples Elbrus -Ushtulu (Table 3). The FA nm level in the sample from the vicinity of the village Elbrus shows higher stress load on the individuals.
It is our opinion that in absence of the anthropogenic factor insignificant differences of the value of this parameter in the studied samples are determined by climatic factors. Based on two-way ANOVA we calculated the impact of climatic factors: annual average temperature (factor gradation: below or above 5°C) and annual average precipitation (below or above 1000 mm).   Table 4 presents that neither the factor «annual average temperature» nor the factor «annual average precipitation» affects FA nm levels. We revealed significant correlation only within interaction of the given factors.
Thus, among the given localities the climatic conditions of the village Elbrus (annual average temperature below 5°C and annual average precipitation less than 1000 mm) present a stressing environment for Chionomys gud.
When assessing the FA nm level of the skull in the authochton from the middle mountains of the Caucasus, it is pertinent to show the results which we obtained in the same localities for such a widespread species of the Caucasus as Apodemus uralensis Pallas, 1811 (Tembotova & Amshokova, 2018). It is notable that in A. uralensis a minimum FA nm accounts for 21.9, and the maximum FA nm is 25.8, whereas in C. gud these values are 26.5 and 32.0 correspondingly. It is evident that the FA nm is higher in C. gud. An important point is that in A. uralensis, under the absence of significant differences between the samples, the maximum FA nm is in the animals from Ushtulu, and the minimum FA nm is in the animals from Bezengi. On the contrary, C. gud sample from Ushtulu is characterised by the lowest FA nm value, and its sample from the village Elbrus has the highest FA nm value. We can suggest that ecological preferences of the species determine the revealed interspecific differences, and correspondingly, different responses to the same environment.
The second task of the work is to assess the degree of phenetic differentiation in C. gud samples of different geographical distance. Such an assessment is also essential as we conduct the studies on an authochton of middle mountains and a stenoecic species which occupies rocky habitats of subalpine meadows. To do this, we calculated mean phenetic distances between the samples. The comparison of individuals by sex revealed a slight, but statistically significant difference (p < 0.05) only in two populations from the vicinities of the village Elbrus and Lago-Naki Plateau. As there were no significant differences in other populations, we pooled male and female samples for further analysis. Table 5 and Table  6 present the results of interpopulation comparison.
The comparison of three C. gud samples from the Central Caucasus showed that distances varied from 0.034 to 0.108. We revealed a minimum distance (MMD = 0.034 ± 0.016, p < 0.05) between the populations of Ushtulu -Bezengi, which are about 20 km apart (absolute distance) in two adjacent gorges (Fig. 1). The differences between the given samples were registered in the frequency of 4 traits out of 27. Table 6 shows that the animals had a high frequency of traits 2 and 26 from the vicinities of the village Bezengi; a high frequency of traits 9 and 15 was found in the animals from Ushtulu landmark area.
From the above, we can suggest that in the Central Caucasus the factor of geographical distance (absolute distance of 20-60 km) has a slight effect on the phenetic distance between the samples, although three samples are from different gorges.
We estimated the effect of geographical distance (absolute distance of 240-310 km) by comparison of C. gud samples of the Central Caucasus with samples of the Western Caucasus. The data obtained are indicative of significant isolation of the Lago-Naki sample from all the samples of the Central Caucasus (Table 5). We found a maximum phenetic distance MMD = 0.148 ± 0.012 (χ 2 = 127.0; df = 26; p < 0.001) by comparison the Lago-Naki population with the Bezengi population. In this case we registered differences in the frequency of 10 traits, 8 of which were more often in the Lago-Naki population. Phenetic distances between pairs of samples from Lago-Naki -village Elbrus and Lago-Naki -Ushtulu were less significant, and accounted for MMD = 0.084 and MMD = 0.096, correspondingly. Compared to the samples of the Central Caucasus, the sample of the Western Caucasus is characterised by a higher frequency of the following traits: FFracan (-), FTmacan, FBsla, FMblg (M 2 ), FMblg (M 3 ), FPC (M 2 ), FHgla (Table 6). The correlation analysis between phenetic distances of all four samples, one of which is from the Western Caucasus, and with an absolute distance of 20-300 km, showed a mean positive correlation (Spearman 0.54), but the value is insignificant.
The comparison of populations by index of resemblance and criterion of identity (Table  7) is consistent with the data obtained. We revealed maximum values of index of resemblance between the populations of the village Elbrus-Ushtulu and Ushtulu -Bezengi in the Central Caucasus. It is evident that values of index of resemblance gradually decrease under comparison of the populations from the Central Caucasus with the population from the Western Caucasus.
The criterion of identity revealed significant differences in all pairwise comparisons, and maximum values, which essentially exceeded the table χ 2 values (Table 7), were registered under comparison of the Central Caucasus populations with the Western Caucasus population.
The assessment of correlation between presence/absence of qualitative cranial traits and climatic factors (temperature and precipitation) of C. gud habitat is as follows: there is a significant negative correlation of moderate strength between temperature and precipitation. Fig. 3 presents the strength level of significant correlation (p < 0.05) between expression of all threshold cranial traits of four samples and climatic characteristics. The presence of 41% of traits correlates with temperature factor (strength of the relationship is moderate for 7 traits, and it is weak for 4 traits). 33% of the traits correlates with precipitation (strength of the relationship is maximum for 1 trait, it is strong for 1 trait, it is moderate for 1 trait, and it is weak for 6 traits).
We also assessed the differences of phene presence in the gradient of temperature and precipitation. There were no differences in the proportion of phenes registered in less humid habitats of C. gud (village Elbrus and village Bezengi, 47% of phene presence) and more humid habitats (Ushtulu and Lago-Naki Plateau, 48% of phene presence).

Conclusions
The studies on population differentiation of Chionomys gud in the Central and Western Caucasus by means of phenetics present the following data.
The range of phenetic distances in C. gud accounted for 0.034-0.148. In the Central Caucasus, despite the fact that all three samples are from three parallel gorges, the distance factor has a minimal effect on the phenetic distance between the samples (absolute distance of 20-60 km). The phenetic distances are more pronounced between geographically distant (over 300 km) populations of Lago-Naki Plateau and the village Elbrus. According to the published data, such values of phenetic distances are common in populations which were isolated by landscape-ecological barriers over prolonged periods (Vasil'ev et al., 2000). The correlation analysis of phenetic distances of four samples which are located 20-300 km apart, and one of them is from the Western Caucasus, revealed the presence of a moderate positive correlation (Spearman 0.54), although it was insignificant.
However, we suppose that not only the factor of territorial isolation but also climatic characteristics exert influence on the level of phenetic distances. So, the village Elbrus has more severe climatic conditions with an annual average temperature below 5°C and an annual average precipitation below 1000 mm, whereas Lago-Naki Plateau is characterised by an annual average temperature above 5°C and an annual average precipitation above 1000 mm. The data on fluctuating asymmetry provide support for this view.
The level of fluctuating asymmetry in C. gud populations from different eco-geographical conditions of the Caucasus accounted for 26.5-32.0%, which is significantly higher than in Apodemus uralensis (21.9-25.8%) (Tembotova & Amshokova, 2018), which is a widespread species of the North Caucasus, and the middle mountains of the Central Caucasus represent stress conditions for it.
We revealed the highest values of fluctuating asymmetry in C. gud from the village Elbrus, suggesting that development of animals in the given population is under definite stress. The climate of the village Elbrus is the most severe of all the studied localities. Two-way ANOVA showed that neither the annual average temperature nor the annual average precipitation, when taken separately, had a significant effect on the FA nm level, but the FA nm was correlated at a statistically significant level under the interaction between the factors «annual average temperature» and «annual average precipitation». The conditions for C. gud habitats are the most optimal in Ushtulu landmark area as the lowest FA nm values were registered at this locality. So, we suggest that the conditions with an annual average temperature below 5°C and an annual average precipitation within 1000-1100 mm are optimal for the species.