Introduction

In litter-bearing mammals, the maternal energy requirements during lactation are extremely high and increases with the number of offspring in the litter (Millar 1975, 1977; Gittleman and Thompson 1988). Milk production is closely related to maternal body weight and fecundity (König and Markl 1987; König et al. 1988; Kenagy et al. 1989; Guerra 2001). The maternal body resources serve as an additional source of energy under nutritional insufficiency during lactation. In small rodents, the energy equivalent of decrease in body weight by 1 g averages 19.3 kJ in lactating females (Weiner 1987). Moreover, maternal resource deficits can result in delayed growth and poor survival of offspring (König et al. 1988; Barnard et al. 1998; Ricceri et al. 2007; Hudson and Trillmich 2008; Andersen et al. 2011; Pan et al. 2014). Indeed, the positive association between maternal body mass and litter size and the negative association between offspring body mass and litter size are well documented (Falconer 1960; Cameron 1973; Millar 1983; Myers and Master 1983; Sikes 1998). In this regard, offspring fitness essentially depends on the maternal physical state (Nazarova and Evsikov 2012).

Some fitness-related characteristics, such as the nutritional state of pregnant females, embryo mortality, litter size, and sex ratio, vary with population size, and these characteristics affect the development of life history traits in individuals of the following generation through the maternal effect (Evsikov et al. 2000, 2008; Nazarova and Evsikov 2008, 2012). Small mammals that experience drastic population fluctuations demonstrate sex-biased maternal investment, which depends on the mother’s body conditions and food availability (Shibata and Kawamichi 2009). Offspring number in the litter and intra-uterine sex ratio affect the probability of adjacent fetuses being of the opposite sex, which further influences many aspects of the development of offspring (Clark et al. 1992; Curtis 2010; Zhang et al. 2011; Nazarova and Evsikov 2012; Szenczi et al. 2013) as well as the physiology (König and Markl 1987; Kenagy et al. 1989; Guerra and Nunes 2001) and behavior of mothers (McGuire and Bemis 2007). Litter sex ratio may affect serum testosterone levels and aggressiveness in male offspring due to the intrauterine position effects (Clark et al. 1992; Szenczi et al. 2013). Prenatally androgenized males show morphologically altered reproductive organs and accentuated reproductive and aggressive behaviors (Ryan and Vandenbergh 2002). After birth, litter size and sex composition are important factors determining the degree of sibling competition for limited maternal resources and the development of aggressive behavior and social dominance (Namikas and Wehmer 1978; Mendl and Paul 1990, 1991; Benus and Henkelmann 1998; Dupont et al. 2015).

Over the past decade, interest in the ontogenetic nature of consistent individual behavioral and neurophysiological variations in mammals has increased (Mendl and Paul 1990, 1991; Stamps and Groothuis 2010; Trillmich and Hudson 2011; Rödel et al. 2017). Studies on rodents and lagomorphs have suggested that individual and social behaviors in adulthood depend on the quality of maternal care (Ryan and Wehmer 1975; Benus and Henkelmann 1998; Dimitsantos et al. 2007; Hudson and Trillmich 2008; Cameron 2011; Rödel et al. 2017). The frequency of tactile contact with offspring and the amount of time mothers spend in the nest are labile signals that convey reliable information regarding the local ecological conditions and exert a programming effect on the physiology and behavior of the offspring (MacLeod et al. 2007; Coutellier 2012). Variations in the frequency of these behaviors during the neonatal period produce a lasting effect on the expression of genes that control physiological functions and social behaviors in adulthood (Weaver et al. 2004).

We examined the effects of the early postnatal environment on social behavior in water vole [Arvicola amphibius (Linnaeus, 1758)], a common mammal in West Siberia, whose populations are subject to severe multiannual fluctuations in numbers (Evsikov and Moshkin 1994; Evsikov et al. 2001). In rodents, social behavior plays pivotal roles in the regulation of space use and reproductive potential (Johnsen et al. 2019). In water voles, male spacing patterns are determined mainly by the spatial and temporal distribution of reproductively active females and availability of optimal habitats (Evsikov et al. 1997, 2017; Muzyka et al. 2010). Male home ranges significantly overlap with one another and with one or more female home ranges, suggesting the presence of polygynous or polygynandrous mating system. Female home ranges are smaller and relatively more isolated from one another (Stoddart 1970; Bragin et al. 2004). The frequency of intrapopulation aggressive interactions consistently varies with the male androgen status over a population cycle, with males at the peak population phase being the most aggressive (Evsikov et al. 1997). Serum testosterone levels in adult males were the highest during the peak phase and the lowest during the decline phase, varying more than tenfold between the two periods (Moshkin et al. 1984). At a high population density, rodent individuals that are more aggressive toward conspecifics may have a selective advantage (Chitty 1967; Evsikov et al. 1997).

The contribution of maternal body weight, litter size, and litter sex composition to the development of water vole male behavior has not been studied to date. A better understanding of the ontogenic factors driving the development of behavioral phenotypes is paramount to elucidate the mechanisms underlying the population dynamics and phenotypic plasticity of this rodent species. Here, we examined the effects of natural variability in the early developmental environment on the behavior of adult water vole males through a correlational study. We hypothesized that maternal traits associated with reproductive effort, including the body mass of the mother and the size and sex composition of the litter in which the males were born and weaned, affect the development of social behaviors in male offspring. Thus, our aim was to reveal maternal and litter characteristics that potentially affect male adult body weight, androgen status, and social behavior. Based on literature on litter-bearing mammals, we predicted that individual variability in aggressiveness in adulthood is positively correlated with inter-sibling competition for maternal resources in early life. We expected that males from large litters or from male-biased litters would become more aggressive due to the experience of sibling competition in early life or the intrauterine position effects. Specifically, we explored (1) the associations among body mass, androgen status, and social behavior in adult males and (2) the effects of maternal body mass and litter size and litter sex composition at birth on male social behavior in adulthood.

Materials and methods

Animals

The study included 48 sexually mature males (9–15-months old, which is the age range of sexual maturity in A. amphibius males) from a colony of water voles maintained at the Institute of Systematics and Ecology of Animals SB RAS (Novosibirsk, Russian Federation). The colony was founded in 1984 from animals captured near Novosibirsk (10 males and 8 females) and has since been regularly (every 1–3 years) supplemented with individuals from the same population to limit inbreeding. Based on pedigree information, the average inbreeding coefficient is 5.07%.

Animals were housed in individual stainless steel cages (27 × 48 × 25 cm3) filled with hay and maintained under a natural photoperiod at an ambient temperature of 18–25 °C. Food (carrots, steamed grains, and fresh grass) and water were provided ad libitum. Females of this species exhibit induced estrus. Estrus incidence in females occurs within 2 weeks of male–female cohabitation (Nazarova et al. 2007). Female body mass at mating; litter size at birth; and body mass of offspring at birth, at 1, 2, and 3 weeks of age (weaning occurred at 21 days), and in adulthood after reaching sexual maturity (9–15-month old) were measured for all animals. All individuals were sexed and marked individually after birth.

Procedures and samples

Social behavior tests were performed from April to May 2016, from 10.00 to 14.00 h (NOVT UTC + 7 time zone) in a neutral arena. The size of the testing arena was small (50 cm diameter and 30 cm wall height) to enhance encounters and, possibly, aggression. A total of 24 tests were conducted on 48 males. Each vole was used for testing once. Pairs of males were selected randomly, without taking into account the characteristics of the mother or litter. For each test, a pair of unfamiliar animals, who were not close relatives (siblings or half-siblings), was used. For identification, the males were dorsally marked with a water-based, non-toxic paint. Individuals were randomly placed in opposite halves of the arena, which was divided in the middle by a removable transparent partition. The partition was lifted after a habituation period of 2 min, and the frequencies of all encounters initiated by each the dyad member were videotaped for 10 min (Martin and Bateson 1993).

The frequency of social interactions was recorded for each animal in the dyads (Kudryavtseva et al. 2014; Lee et al. 2019): (1) amicable contacts (approaching without aggressive contact: naso-nasal, naso-genital, or naso-bodily sniffing); (2) indirect aggressive interactions, or threat behavior (aggressive demonstrations: tooth chattering or tail rattling), and (3) direct aggressive interactions (fights, attacks, bites, and lunges). To characterize social interactions, we used the sum of values for the behavioral patterns of each category. Teeth chattering and tail rattling were counted as the number of completed behavioral acts (from the beginning of the series to the end). This method has been previously used (Evsikov et al. 1997).

Testosterone measurement

Blood (100–200 μL) was collected from the retro-orbital sinus a week before the experiment between 10.00 and 14.00 h (NOVT UTC + 7 time zone) to determine serum testosterone level. The blood samples were centrifuged for 15 min at 1500 g, and the serum was stored at − 20 °C until further assays.

Serum testosterone level was measured using an ELISA kit (#X-3972, Vektor-Best, Novosibirsk, Russia), following the manufacturer’s instructions. The sensitivity of the assay was 0.2 nmol·L−1, and the intra-assay coefficient of variation was < 8%. Testosterone level was determined based on a calibration curve after measuring the optical density of the solution in the microplate wells at 450 nm using the PowerWave XS2 spectrophotometer (Winooski, USA). The testosterone assay for A. amphibius was validated by testing for parallelism using two-fold serial dilutions of unextracted serum over a dilution range from 1:1 to 1:64. The slopes of the lines were compared and were found to not differ significantly (t8 = 1.520, p = 0.084) following log-logit data transformation (Chard 1987). Serum testosterone levels were measured in 41 of the 48 males. The volume of serum collected from seven males was insufficient for analysis.

Statistical analysis

All analyses were conducted using IBM SPSS v.27 (SPSS, Chicago, IL, USA). For the analysis of pre-weaning growth, we used a linear mixed model (LLM). Assumptions of homoscedasticity and normality of residuals were confirmed using diagnostic plots. Non-normally distributed behavioral data were subjected to square-root transformation to approximate a normal distribution. Litter size and age were included in the model as covariates. Age × litter size interaction and a random slope for week (age) were included in the model to assess the differences in litter size-dependent offspring growth. Litter identity was included as a random intercept to account for the non-independence of offspring within a litter. Maternal identity was not included in the analysis because each litter was produced by a unique parental pair.

We used the LLM to evaluate the effect of early rearing environment (litter size at birth, litter sex composition at birth, and maternal body mass) on the frequency of different social behaviors. For each category of social behavior, the random effect of litter identity was evaluated to avoid the over-representation of individual litters. Considering the possible interdependence of the behavior of interacting males, dyad identity was included in the model. All data are expressed as mean ± standard error of the mean (SE). The level of significance was set at an alpha value of 0.05.

Results

Effects of early rearing environment on adult body mass and androgen status

The mean litter size at birth was 4.73 ± 0.24 (min–max: 1–7); sex ratio at birth (% males) was 66.49 ± 4.01 (min–max: 33.3–100); and postpartum maternal body mass was 158.10 ± 5.29 g (min–max: 102.4–237.1). There were no significant correlations between litter size and sex ratio (r = − 0.28, n = 26, p = 0.167), litter size and maternal body mass (r = 0.21, n = 26, p = 0.309), and sex ratio and maternal body mass (r = − 0.21, n = 26, p = 0.309).

All these characteristics can potentially affect adult body mass and androgen status in males. The LLM with litter identity as the random intercept and postpartum maternal body mass, litter size, and sex composition as the predictors showed no effects of these factors on adult body mass (litter size: F1,22.95 = 0.170, β = 2.673 ± 6.479 SE, p = 0.684; sex ratio: F1,24.39 = 3.163, β = 0.748 ± 0.421 SE, p = 0.088; and maternal body mass: F1,18.38 = 0.311, β = 0.159 ± 0.285 SE, p = 0.584). The variance explained by the random factor was not significant (χ2 = 1.667, p = 0.096). Serum testosterone levels in adult males were not depend on litter size (F1,14.50 = 3.401, β = − 0.875 ± 0.474 SE, p = 0.086), sex ratio (F1,19.02 = 0.018, β = − 0.004 ± 0.033 SE, p = 0.894), maternal body mass (F1,9.27 = 0.983, β = 0.019 ± 0.019 SE, p = 0.347). There was no significant effect of male body mass on testosterone levels (F1,27.91 = 2.641, β = 0.022 ± 0.014 SE, p = 0.115).

Social behavior of males in dyadic tests and its correlssation with body mass and androgen status

The LLM with dyad identity and litter identity as the random factors and body mass and serum testosterone level as the predictors showed no dependence of the frequency of amicable contacts or direct aggression on individual body mass or testosterone level (amicable contacts: βbodymass = 0.001 ± 0.003 SE, F1,38 = 0.013, p = 0.909 and βtestosterone = 0.019 ± 0.035 SE, F1,38 = 1.590, p = 0.215; direct aggression: βbody mass = 0.001 ± 0.002 SE, F1,37.24 = 0.065, p = 0.800, and βtestosterone = 0.015 ± 0.240 SE, F1,31.20 = 0.011, p = 0.918). The frequency of threats was negatively correlated with serum testosterone level (β = − 0.102 ± 0.036 SE, F1,37.75 = 8.032, p = 0.007) but not correlated with body mass (β = 0.002 ± 0.003 SE, F1,34.33 = 4.032, p = 0.052).

Effects of early rearing environment on the development of social behavior in males

We examined whether the maternal nutritional state (body mass) and litter size and sex composition at birth affected the development of male social behaviors using an LLM with litter identity and dyad identity as the random factors (Table 1). The frequency of direct aggression was correlated with the size of the litter into which the males were born and weaned. With the increase in the number of offspring per litter, the frequency of agonistic contacts significantly increased (Fig. 1). The frequency of different social behaviors was not correlated with litter sex ratio at birth or postpartum maternal body mass.

Table 1 Summary of the results of linear mixed models assessing the effect of litter size and sex ratio at birth, postpartum maternal body mass, litter identity (IDlitter) and dyad identity (IDdyad) on male social behavior in dyadic tests
Full size table
Fig. 1
figure 1

Relationship between the frequency of direct aggression and litter size at birth in adult (9–15 month-old) male water voles (A. amphibius). Size of circle is proportional to the number of individuals

Full size image

Preweaning body mass and litter size

The LLM with litter identity as the random intercept and age as the random slope showed a highly significant negative effect of the litter size × age interaction on offspring body mass (F1,163.64 = 31.252, β = -0.861 ± 0.154 SE p < 0.001), indicating that the preweaning growth closely depends on the number of offspring at birth (Fig. 2). The random effect of litter identity was not significant (χ2 = 0.860, p = 0.390).

Fig. 2
figure 2

Relationships between the body mass and litter size at birth in male water voles (A. amphibius) at different ages

Full size image

Discussion

Rearing environment and adult body mass

We examined whether the early postnatal environment produces a long-term effect on male–male interactions. First, we evaluated the effects of rearing conditions on adult body mass and testosterone level as the possible morphological and endocrine drivers of male social behavior, respectively. Our results indicate that neither litter size and sex ratio at birth nor maternal mass significantly affect male body mass in adulthood.

Adult body mass, testosterone level, and aggressive behavior

In this study, the frequency of aggressive behaviors against the same-sex conspecifics did not depend on male body mass. This result is consistent with previous reports (Krebs 1970; Krebs et al. 1977) that neither body mass nor testosterone level affected the aggressive behavior of adult Microtus townsendii males, as assessed in a neutral arena dyadic test. Surprisingly, in the present study, testosterone levels in adult males were significantly but negatively correlated with the frequency of threats in the dyad tests, but not with the frequency of direct aggressive and amicable contacts. Contrary to these findings, Moshkin et al. (1984) reported that serum testosterone levels in males from a natural population showed a weak but significantly positive correlation with wounding level. Of note, however, the authors did not include the frequency of threats in the ethogram of aggressive behavior. In wild male water voles in the population decline phase, testosterone propionate injection increased both body mass and aggressiveness (Moshkin et al. 1984).

Despite numerous studies in this area, however, the association between testosterone level and aggressive behavior raises many questions. Testosterone can underlie behavioral responses, although the social context may add to the complexity of this association (Gleason et al. 2009). The correlation between blood testosterone level and aggressiveness can be positive, negative, or even null in different rodent species, depending on their social structure (Trainor and Marler 2001; Gromov and Voznessenskaya 2013). In humans, a very weak correlation between aggressiveness and baseline testosterone level and a strong correlation between aggressiveness and high testosterone level, measured after social interactions, have been reported (Geniole et al. 2019). Therefore, we propose that the inverse association between the basal testosterone level and frequency of threats observed in the present study can be explained by the higher reactivity of the hypothalamic–pituitary–gonadal axis to social stimuli. We intend to test this hypothesis in future research.

Litter size and male–male interactions

Among the factors associated with the family environment, only litter size at birth appeared to significantly affect adult male–male social interactions. The frequency of aggressive interactions in the dyadic tests was positively correlated with litter size at birth. This result is consistent with previous reports in mice, rats, and pigs. As such, males reared in a larger litter were more scent marked (laboratory mice, Mus musculus: Gibson et al. 2015) and achieved significantly higher social interaction and aggressiveness scores (laboratory rats, Rattus norvegicus: Deitchman and Lavine 1977; domestic pigs, Sus scrofa domesticus: D'Eath and Lawrence 2004; laboratory mice: Ryan and Wehmer 1975; Barnard et al. 1998) than those reared in a smaller litter. In mink, Neovison vison, biting among kits increased around the age of weaning, particularly in large litters (Clausen and Larsen 2015).

Our results may be explained by the well-documented link between the size and quality of litter as well as the quality of mother–pup interactions in rodents (Champagne et al. 2003; Cameron 2011). Tactile stimulation is an important mediator of the effects of maternal care on the behavioral development of offspring. In mice, mothers rearing smaller litters spent more time in the nest, more time licking individual pups, and more time nursing the litter than mothers rearing larger litters (Priestnall 1972). Pine (Microtus pinetorum), prairie (Microtus ochrogaster), and meadow (Microtus pennsylvanicus) vole mothers rearing large litters spent less time in the nest than those rearing small litters (McGuire and Bemis 2007). In rats, males raised in larger litters with less maternal care were more successful in competing for food and were more aggressive toward humans in adulthood than males raised in smaller litters (Seitz 1954). The frequency of maternal licking was inversely correlated with the neural system activation mediating fearfulness in the reared offspring (Caldji et al. 1998). Therefore, the less the tactile contact with the mother, the more stress-reactive the offspring and the more aggressive the adult males.

Offspring growth and litter size

Body mass of preweaned males and size of the litter in which they were born were negatively correlated, and this association became stronger with increasing male age. Our results confirm the previous reports that increased fecundity is associated with the decreased amount of resources devoted to each pup in litter-bearing mammals (Falconer 1960; Smith and Fretwell 1974; Sikes 1998; Hudson et al. 2011), suggesting the presence of a trade-off between offspring number and quality (Pianka 1976).

Life history behavioral adaptation

In mammals, reproduction is energetically costly (Gittleman and Thompson 1988). Previous studies have found that the availability of maternal milk and presence of siblings are important postnatal environmental factors that promote the initiation of solid food intake, growth, and the development of phenotypic behaviors (Thiels and Alberts 1986; Hudson et al. 2011; Gibson et al. 2015). We assume that behavioral adaptations help resolve the mother–offspring conflict for resources during lactation. In water voles, lactation leads to the depletion of maternal reserves (Nazarova and Evsikov 2008). At weaning—the time of a sharp decline in maternal investment—there may be competition among siblings for the mother’s milk and nesting resources, particularly in large litters. This competition may promote the earlier initiation of weaning, farther dispersal from the natal territory, and development of more active competitive behaviors that persists in later life. Indeed, nutritional deficits during infancy were associated with earlier weaning in rats (Babicky et al. 1973; Thiels and Alberts 1986).

Previously, we found a positive effect of litter size on the exploratory behavior of preweaning water voles, suggesting that the higher exploratory activity of sucklings from large litters can facilitate the initiation of solid food intake and thus help reduce the competition for maternal resources and preserve the maternal body condition (Lyubaya and Nazarova 2011). Currently, little is known regarding whether the associations among maternal behavior, timing of weaning, and offspring sex in litter-bearing mammals (Curley et al. 2009). Some of these factors may explain the correlation between male aggressiveness and litter sex ratio observed in the present study.

Maternal influences on population dynamics

The change in litter size is an important driver of the water vole population cycle. Litter size is greater in the population peak phase and smaller in population decline phase (Evsikov et al. 2008). Based on the results obtained, we conclude that the complex conditions during early development associated with maternal fecundity and maternal care produce a lasting effect on the social behavior of adult male water voles. According to our previous findings, male reproductive success depends on social behavior in water voles. The frequency of sniffing by males is positively correlated with the incidence of estrus in females after pairing (Nazarova et al. 2007). Regarding aggression, the olfactory preference for males is non-linear: the most and the least aggressive males are the least attractive to females (Evsikov et al. 2006). Thus, the dependence of male adult behavioral phenotype on early postnatal conditions may be significant in the regulation of reproductive potential and population dynamics of water voles.