We studied the seasonal variation of adaptive humoral immunity (AHI) in northern red-backed vole (Clethrionomys rutilus Pallas, 1779, RBV) and grey-sided vole (C. rufocanus Sundevall, 1846, GSV) in Tomsk region of Western Siberia. Immunoresponsiveness (IR) to sheep red blood cells was assessed by the number of antibody-producing cells in the spleen. The use of a generalized linear model to analyze the effects of species, sex, year of research, and season of withdrawal of individuals from nature on IR showed a significant effect of species identity, season of animals capture and the interaction of species with season. The RBV demonstrated higher immune responses during a year, and both species had higher IR in winter. Suppression of IR in spring was greater, started earlier and lasted longer (March–May) in GSV. In RBV, immunosuppression was restricted to April. The significant negative within year correlations of IR with body mass and masses of reproductive organs in GSV a tradeoff between AHI and growth and reproduction processes. A reason for the difference between species in the seasonal variation of AHI may be related to the difference in tropho-energetic requirements of each vole species. GSV is a predominantly herbivorous rodent and its thermoregulation seems less efficient than of RBV. The deeper spring immunosuppression in GSV may explain in part its higher mortality during the season of colds.

We caught voles to assess immunoresponsiveness from September to June in 2016–2017 and in 2017–2018 within one habitat type (mixed coniferous – parvifoliate forest: upper layer: Picea obovata + Betula pendula + Abies sibirica+ Populus tremula+ Pinus sibirica ; middle layer: Prunus padus + Sorbus aucuparia sibirica; undergrowth: Ribes rubrum + Ribes nigrum + Rosa majalis.+ Rubus idaeus). Temperature conditions at the voles’ trapping site (t C°) averaged for every 10 days during 2016–2017 and 2017–2018 are represented in Fig. 1. The temperatures were measured from October to May in the open air, in the litter and in the soil at a depth of 15 cm every three hours by autonomous recorders DS1921G-F5 (Maxim Integrated Products, USA). Data on temperature were provided by S.I. Gashkov.

During the period of our study of immunoresponsiveness (2016–18), the relative abundance (N individuals per 100 trap-days) of the red-backed vole continued to increase in 2016, the growth stopped in 2017, and in 2018 there was a slight decrease in the number. In the gray-sided vole, a slow increase in the relative abundance occurred in 2016–17; the growth stopped in 2018 (Fig. 2). In the period 2016–18, both populations demonstrated abundances close to long-term averages.

Animals we removed from nature were kept individually in cages (25×40×12 cm) under the natural photoperiod at a temperature of +8 – +10o С from October to April, and at natural ambient temperatures for the remainder of the study. Food (oats, apples, grass), water and nesting material were provided ad libitum.

We weighed the animals immediately after capture within an accuracy of 0.1 g and estimated the masses of the reproductive organs, the testes in males and the uterus with ovaries in females, with an accuracy of 0.001 g after killing no later than one week after capture. To assess the between-month variation of the masses of reproductive organs in order to increase the sample size, we used data for a longer period: for red-backed vole from 2016 to 2018 (82 males and 59 females), for grey-sided vole from 2016 to 2021 (93 males and 80 females).

Measurement of adaptive humoral immunoresponsiveness. We immunized voles from 8 to 9 am 36 hours after capture. Tests of the duration of the glucocorticoid response to manipulations similar to trapping disturbance and to injection of ACTH in a congeneric species, Clethrionomys glareolus, showed that the effect of the stress factors ends within a day (Zavyalov et al., 2003; Rogovin and Naidenko, 2010). We did not treat pregnant females. To assess adaptive humoral immunity, we used local hemolysis in a liquid medium (Cunningham, 1965). We estimated the number of antibody-producing cells (APC) of the spleen that were formed in response to the introduction of a non-replicating antigen, 0.5 ml of 2 % of SRBC suspension injected intraperitoneally. On the 5th day, the voles were killed by cervical dislocation. The suspension of spleen cells was prepared as described by Moshkin et al. (1998). The reaction mixture consisted of 500 µl spleen cell suspension, 500 µl washed SRBC (4 x 109, erythrocytes/ml) and 500 µl lyophilized guinea pig serum (Biomed, Perm, Russia) resolved with 1 ml of isotonic sodium chloride solution. Cunningham chambers were prepared from glass microscope slides, loaded with 200 µl of reaction mixture (2 chambers per individuals), and incubated for 2 h at 37°C before hemolysis zones were counted. We calculated and then averaged the number of hemolysis zones (antibodies-producing splenocytes) for each individual. We used two indicators of immune activity. To eliminate within and significant between species differences in body masses, the total number of APCs in the spleen was divided by individual body mass (Novikov et al., 2010). Estimation of the number of APCs per unit of body mass has an advantage over the estimate per unit of spleen mass, since splenomegaly is often observed in forest voles (10,6 % in red-backed vole and 7,7 % in grey-sided vole in our study). An increase in the mass of the spleen in such cases is not associated with an increase in the number of APCs (Kravchenko, unpublished). To analyze correlations of immunoresponsiveness with body mass and with masses of reproductive organs, we used the absolute number of APCs.

Statistical procedures

All statistical analyses were performed using STATISTICA v. 10.0 (StatSoft Inc., USA). To evaluate effects of a set of independent variables on the magnitude of humoral immunoresponsiveness to SRBC (the number of antibody-producing splenocytes per unit body mass), we applied Generalized Linear Model (GLZ) analysis with the normal condition and Log link function. The intensity of humoral immune response was assigned as a dependent variable. Predictors included four categorical variables (species, sex, year of research, and season of withdrawal of voles with four gradations: winter, spring, summer and fall). Two-way interactions between predictors were also considered. The significance of predictors and their interactions was tested using Wald statistics and Likelihood ratio test type III. To choose the best model among candidate models we applied the Akaike Information Criterion (AIC; Akaike, 1973). The models were considered to have substantial support if ΔAIC (the difference between the AIC score of the given model and the minimal AIC score) was less than 2.0.

For a more detailed study of within year variation of the immunocompetence of each of the vole species, we studied the monthly variation of immunoresponsiveness to SRBC. The monthly variation of body mass and mass of reproductive organs (testes in males, and uterus + ovary in females) was also considered. We used Kolmogorov-Smirnov test and visual examination of histograms to examine distributions. The Levene’s test was used to check homogeneity of variances. Because the assumption of normality and homogeneity of variances was unconfirmed for the within year variation of immunoresponsiveness in the body mass and mass of reproductive organs we used Kruskal-Wallis ANOVA with multiple (post-hoc) comparisons of average ranks for each pair of groups (normal z-values were computed for each comparison, as well as post-hoc probabilities corrected for the number of comparisons for a two-sided test of significance (Siegel and Castellan, 1988). We also used Mann-Whitney U-test for independent pair comparisons. Spearman’s Rank Order Correlation Coefficient (Rs) was used to measure linkage of continuous non-normally distributed data. Since our analyses were based in part on data from a different number of years (we analyzed the between-month variation in the mass of the reproductive organs of voles over a longer time interval, than immunoresponsiveness), we indicated the sample sizes in the headings to the table and figures. Tests were two-sided, with a significance level < 0.05.