Adaptive shaping of the behavioural and neuroendocrine phenotype during adolescence

Abstract

Environmental conditions during early life can adaptively shape the phenotype for the prevailing environment. Recently, it has been suggested that adolescence represents an additional temporal window for adaptive developmental plasticity, though supporting evidence is scarce. Previous work has shown that male guinea pigs living in large mixed-sex colonies develop a low-aggressive phenotype as part of a queuing strategy that is adaptive for integrating into large unfamiliar colonies. By contrast, males living in pairs during adolescence become highly aggressive towards strangers. Here, we tested whether the high-aggressive phenotype is adaptive under conditions of low population density, namely when directly competing with a single opponent for access to females. For that purpose, we established groups of one pair-housed male (PM), one colony-housed male (CM) and two females. PMs directed more aggression towards the male competitor and more courtship and mating towards females than did CMs. In consequence, PMs attained the dominant position in most cases and sired significantly more offspring. Moreover, they showed distinctly higher testosterone concentrations and elevated cortisol levels, which probably promoted enhanced aggressiveness while mobilizing necessary energy. Taken together, our results provide the clearest evidence to date for adaptive shaping of the phenotype by environmental influences during adolescence.

1. Introduction

The development of a phenotype is not only determined by its genotype but also strongly influenced by the environment. By means of this developmental plasticity, organisms are capable of producing a phenotype that is adapted to prevailing environmental conditions [1,2]. Prenatal and early postnatal development are particularly sensitive to environmental inputs, which are often mediated by the mother [3,4]. As described by the concept of predictive adaptive responses, these environmentally induced adaptations do not need to provide an immediate advantage but may improve fitness during later life [5]. For example, the extent of fur development in meadow voles is influenced by maternal photoperiodic history [6], and morphological defences are maternally induced in small aquatic crustaceans according to predator cues [7].

In the same way, it has been argued that adolescence represents an additional developmental period during which phenotypes are adaptively shaped in response to ‘updated’ information about the environment [8,9]. This notion is consistent with theoretical work emphasizing the adolescent life stage as a sensitive window of enhanced plasticity [10]. Adolescence is associated with a complex interplay of alterations in endocrine systems and neural circuits [11,12]. Moreover, sexual maturation that emerges at this time plays a crucial role in the development of social behaviour. Particularly in mammalian species, adolescents become increasingly independent from their parents and are likely to disperse from their natal group [11]. Hence, the social environment changes considerably, while the nature of interactions shifts towards a reproductive context, both of which provide new sources of environmental cues [9–11]. Moreover, individuals are receiving information about the environment directly, rather than largely indirectly via parental effects. Therefore, it appears that the sensitive period for inducing adaptive responses may not be limited to the perinatal phase. Instead, adolescence seems to offer an additional opportunity for adjustments to new social situations and to correct the developmental trajectory when the environment deviates from earlier predictions [9].

Indeed, social experiences during adolescence can have substantial modulatory effects on behaviour in mammals and birds [13–15]. One of the best-studied examples of such profound shaping of behavioural phenotypes by adolescent social conditions is the guinea pig. Males living in mixed-sex pairs from early adolescence (PMs) are extremely aggressive when encountering another male and show greatly enhanced courtship behaviour in the presence of a female when compared with males living in large mixed-sex colonies during this phase of life (CMs) [8,9,16]. These essentially different phenotypes appear to be modulated by an interaction of testosterone and glucocorticoids [17,18] and have a major impact on fitness during later life. When placed into large mixed-sex groups of unfamiliar conspecifics in adulthood, colony-housed males easily integrate whereas males lacking such social experiences exhibit exaggerated endocrine stress responses and severe reductions of body weight that can even threaten their survival [8,19,20]. The low aggressiveness and courtship activity of CMs can be regarded to be part of a queuing strategy that aims at avoiding reproductive competition until they have attained a size and physical condition that allows them to successfully compete with dominant males, and thus represents an adjustment to living in socially complex groups [8,9]. The higher aggressiveness of pair-housed males, by contrast, may be adaptive when the number of competitors is limited. In this situation, it appears to be more promising not to queue but rather directly defend and fight for mating opportunities [8,9]. Taken together, we argue that the different phenotypes shaped by adolescent experiences are adaptations to particular environments in which either queuing or defending and fighting for resources leads to higher reproductive success for the developing males.

While a fitness benefit of CMs in large social groups has been clearly demonstrated in earlier work, it remains to be tested whether the high-aggressive phenotype of PMs indeed enhances mating success when they directly compete with an opponent for access to females. For that purpose, we established groups of one PM, one CM and two females. Owing to their social experience during adolescence, it was assumed PMs would show higher levels of aggression as well as courtship and mating than CMs. Correspondingly, we expected higher plasma concentrations of cortisol and testosterone in PMs. According to the assumption of adaptive shaping, we predicted that this high-aggressive strategy would lead to dominance of PMs over CMs, and ultimately to greater reproductive success.

2. Material and methods

(a). Animals and housing conditions

The guinea pigs were housed in four mixed-sex colonies, each kept in an enclosure of approximately 6 m², and consisting of 7–12 males, 11–16 females and their pre-weaned offspring (less than three weeks of age). Both sexes showed a graduated age structure ranging up to 20 months. At 30 (±1) days of age, 26 male subjects, each from a different litter, were randomly assigned to pair-housing or colony-housing conditions. Each PM was placed together with an unfamiliar female in an enclosure of 0.5 m². CMs were moved to one of the other colonies. The interval at which subsequent CMs were introduced into the same colony was always at least five weeks. All animals were housed under standard conditions: 12 L : 12 D cycle, temperature 21±2°C, relative humidity 50±15%. Commercial guinea pig diet and water were available ad libitum. See [17] for further details.

(b). Experimental design and procedures

At 120–125 days of age (i.e. in late adolescence), pairs of one PM and one CM were placed together with two sexually mature, non-pregnant females unfamiliar to both males to create highly competitive situations with a single male for mating access. Each of these groups was kept in an enclosure of 2 m² that contained two shelters and was located in an unfamiliar room. Females were allowed to habituate to the novel enclosure at least 12 h before the males were introduced at 09.00 (±15 min). Experimental design and schedule of assessments are depicted in the electronic supplementary material, figure S1. Social interactions were scored during the first 3 h of competition on day 1 and for 2 h each between 09.00 and 12.00 on days 2, 3 and 8. Collections of blood samples and determinations of body weights were performed 20 h before competition (day 0), 4 h after the onset of competition on day 1, and then again on days 2, 3, 8, 15 and 22. Each competition lasted for up to five weeks but was terminated earlier when aggression was too escalated or a male lost more than 7.5% of body weight between two consecutive measurements.

(c). Behavioural measures

Social interactions were recorded by an overhead camera (Sony Handycam HDR-CX6) and evaluated by a trained observer (T.D.Z.) using the behavioural analysis software Interact (v. 9, Mangold International). Focal sampling and continuous recording [21] were applied to record frequencies and latencies of approaches, agonistic behaviour (retreat; threat display, i.e. head up, curved body posture; escalated aggression, i.e. head thrust, attack lunge, chase, turn around), courtship and sexual behaviour (anogenital licking, chin-rump follow, mount, rumba, rumping), as well as sociopositive (nose contact) behaviour based on previous definitions and categorizations [16,22] (see electronic supplementary material, table S2, for detailed descriptions). The outcome of an agonistic encounter was scored by means of retreats, with the animal provoking a retreat being regarded as the winner. On that basis, the dominance status of each male was determined by calculating a dominance index as the ratio of wins to the total number of scored agonistic interactions, ranging from 0 (absolutely subordinate) to 1 (absolutely dominant) [23]. Indices were only calculated for days on which the total number of retreats exhibited by the competing males was at least twice per hour on average.

(d). Blood sampling, body weights and hormones

Blood samples were always collected simultaneously from both males of a group at 13.00 (±15 min) by puncturing the marginal ear vessels with an injection needle. All samples for determination of cortisol and testosterone levels were collected within 3 and 6 min of entering the room, respectively, which prevents the sampling procedure from influencing hormone concentrations in the sample obtained [24]. After sampling, both males were weighed, and plasma was separated via centrifugation and deep-frozen until assayed (see [17] for further details). Testosterone and cortisol concentrations were determined in duplicate using an enzyme-linked immunosorbent assay (DES6622, Demeditec Diagnostics; intra- and inter-assay CVs: 7.5% and 9.1%) and a luminescence immunoassay (RE62011, IBL International; intra- and inter-assay CVs: 3.2% and 6.1%), respectively (see electronic supplementary material, tables S3 and S4, for antibody cross-reactivities).

(e). Paternity assignment

Tissue samples were collected from the ears of all parents and a total of 51 offspring, and stored in 70% ethanol until analysed. Genomic DNA was purified by first digesting the tissue samples using Proteinase K, followed by phenol/chloroform extraction and DNA precipitation with ethanol. DNA pellets were washed with 70% ethanol several times and re-suspended in TE-buffer. Fourteen microsatellites were amplified by PCR [25,26] and sequenced. Alleles were analysed using GeneMarker (v. 2.6.4, SoftGenetics). One locus (ap13pet) deviated significantly from Hardy–Weinberg equilibrium and showed an increased number of null alleles and was hence discarded. Paternities were assigned at a 99% confidence level using the likelihood-based approach implemented in Cervus (v. 3.0.7) [27]. See the electronic supplementary material for further details.

(f). Data analysis

All statistics were calculated using R (v. 3.3.1) [28]. As some measures violated assumptions for parametric tests and transformations failed to correct this, non-parametric statistics were used throughout for consistency. Pairwise comparisons between PMs and CMs were conducted using the exact Wilcoxon signed-rank test [29]. Endocrine and body weight measures were additionally analysed for changes over time using the Skillings–Mack test [30]. These measures were converted to per cent of baseline values determined 20 h before the competition to eliminate confounding effects due to differences in baseline levels. Statistical tests were two-tailed and the significance level was α = 0.05 for all comparisons. To account for multiple comparisons of repeated measures, the Benjamini–Hochberg correction was applied to control the false discovery rate at q = 0.05 [31]. We report raw p-values but indicate statistical significance based on corrected significance levels (listed in electronic supplementary material, table S5). Central tendencies and variabilities are expressed as median with first and third quartiles (Q₁–Q₃) unless otherwise noted. As some groups were terminated early, sample sizes decreased over the course of the experiment (n_d0 = 13, n_d1 = 13, n_d2 = 13, n_d3 = 11, n_d8 = 11, n_d15 = 8, n_d22 = 8). Deviations from these sample sizes are due to ties in pairwise comparisons or insufficient sample volumes. Final sample sizes are given in the Results section.

3. Results

(a). Behavioural measures

(i). First day of competition

After the males were placed together with females, PMs initiated courtship and sexual behaviour significantly earlier than CMs (W = 10, n = 11, p = 0.042). Latencies to all other recorded behavioural elements did not differ significantly between PMs and CMs (see electronic supplementary material, table S6, for all latency measures). Concerning frequencies of behavioural patterns, PMs approached the opponent (figure 1a; W = 5, n = 13, p = 0.002) as well as the females (figure 2a; W = 3.5, n = 13, p = 0.001) significantly more often than did CMs. In agonistic interactions, PMs retreated significantly less often than CMs (figure 1b; W = 14.5, n = 13, p = 0.027) and consequently attained a dominant position in 10 of 13 groups (figure 3a; W = 14, n = 13, p = 0.024). However, PMs and CMs did not differ significantly in frequencies of escalated aggression (figure 1c; W = 12, n = 11, p = 0.067) or threat display (figure 1d; W = 7.5, n = 8, p = 0.172). Regarding interactions with females, PMs exhibited courtship and sexual behaviour more often than CMs (figure 2b; W = 17, n = 13, p = 0.046), but this difference was not statistically significant after correction for multiple comparisons. Frequencies of sociopositive behaviour did not differ significantly (figure 2c; W = 17, n = 12, p = 0.079). The females did not differ in their frequency to approach PMs or CMs (figure 2d; W = 36.5, n = 13, p = 0.553).

Figure 1.

Male–male interactions. Frequencies of (a) approaching opponent, (b) retreat, (c) escalated aggression and (d) threat display per hour of observation on days 1 (first 3 h), 2, 3 and 8 (2 h each) after onset of competition. Values are presented as box plots with medians, first to third quartiles, tenth to ninetieth percentiles, and outliers (n_d1 = 13, n_d2 = 13, n_d3 = 11, n_d8 = 11). Statistics: Wilcoxon signed-rank test (two-tailed) with Benjamini–Hochberg correction: *p = 0.027, **p ≤ 0.009, ***p = 0.001.

Figure 2.

Male–female interactions. Frequencies of (a) approaching females, (b) courtship and sexual behaviour, (c) sociopositive behaviour and (d) approach by females per hour of observation on days 1 (first 3 h), 2, 3 and 8 (2 h each) after onset of competition. Values are presented as box plots with medians, first to third quartiles, tenth to ninetieth percentiles and outliers (n_d1 = 13, n_d2 = 13, n_d3 = 11, n_d8 = 11). Statistics: Wilcoxon signed-rank test (two-tailed) with Benjamini–Hochberg correction: *p = 0.014, **p ≤ 0.008, ***p ≤ 0.001.

Figure 3.

Dominance relationships and reproductive success. (a) Dominance indices 3 h after onset of competition (day 1) for each group (A–M). Index is based on outcomes of agonistic encounters and ranges from 0 (absolutely subordinate) to 1 (absolutely dominant). (b) Number of sired litters and offspring within each group (A–M). Statistics: Wilcoxon signed-rank test (two-tailed).

(ii). Days 2 and 3 of competition

On the two subsequent days, frequencies of approaching the opponent (figure 1a; d2: W = 5, n = 13, p = 0.002; d3: W = 0, n = 11, p = 0.001) and females (figure 2a; d2: W = 3.5, n = 13, p = 0.001; d3: W = 0, n = 11, p = 0.001) remained significantly higher in PMs compared with CMs. Concerning agonistic interactions, PMs exhibited significantly more escalated aggression, including chases, head thrusts and attack lunges, which were rarely displayed by CMs (figure 1c; d2: W = 4.5, n = 12, p = 0.004; d3: W = 2, n = 11, p = 0.004). Consequently, PMs still retreated significantly less often than CMs (figure 1b; d2: W = 7, n = 12, p = 0.009; d3: W = 0, n = 10, p = 0.002) and were clearly dominant in most groups (electronic supplementary material, figure S7; d2: W = 7.5, n = 11, p = 0.021; d3: W = 0, n = 8, p = 0.008). Threat display, however, became virtually absent (figure 1d) and could thus no longer be reliably analysed. In interactions with females, PMs showed significantly more courtship and sexual (figure 2b; d2: W = 7.5, n = 13, p = 0.005; d3: W = 0, n = 11, p = 0.001) as well as sociopositive (figure 2c; d2: W = 0, n = 12, p < 0.001; d3: W = 1, n = 9, p = 0.008) behaviour than their opponents, as CMs seldom interacted with females. As opposed to the initial day, the females approached PMs significantly more often than they approached CMs (figure 2d; d2: W = 3, n = 13, p = 0.001; d3: W = 3.5, n = 11, p = 0.006).

(iii). Day 8 of competition

Agonistic encounters between males became less frequent than in the first week, and PMs and CMs did not differ significantly in frequencies of approaching the opponent (figure 1a; W = 11, n = 11, p = 0.053), retreats (figure 1b; W = 11, n = 10, p = 0.105) or escalated aggression (figure 1c; W = 11, n = 10, p = 0.104). Accordingly, no significant difference in dominance was found between PMs and CMs (electronic supplementary material, figure S7; W = 3.5, n = 6, p = 0.219). In male–female interactions, PMs still approached females significantly more often than did CMs (figure 2a; W = 6, n = 11, p = 0.014) and females also continued to approach PMs significantly more often than they approached CMs (figure 2d; W = 3, n = 11, p = 0.005). However, PMs and CMs no longer differed significantly in frequencies of courtship and sexual (figure 2b; W = 8, n = 8, p = 0.188) or sociopositive (figure 2c; W = 8, n = 9, p = 0.094) behaviour.

(b). Endocrine and body weight measures

(i). Plasma testosterone

Social experiences during adolescence influence testosterone secretion (see Discussion), and accordingly, PMs had significantly lower testosterone levels than CMs prior to the competition (PMs: 2.4 ng ml⁻¹, Q₁–Q₃: 1.5–3.1 ng ml⁻¹; CMs: 6.3 ng ml⁻¹, Q₁–Q₃: 4.1–8.2 ng ml⁻¹; W = 8, n = 13, p = 0.006). The competition significantly affected testosterone levels of PMs and CMs (figure 4a; PMs: χ² = 31.56, p < 0.001; CMs: χ² = 14.54, p = 0.024; d.f. = 6, n = 13), but the responses differed considerably. During the first 4 h, testosterone values of PMs increased markedly to about 288% (Q₁–Q₃: 188–322%), whereas CMs showed a decrease to about 85% (Q₁–Q₃: 44–101%). As a result, PMs had significantly higher relative testosterone concentrations than CMs (W = 0, n = 12, p < 0.001), even though absolute values did not differ significantly (W = 24, n = 12, p = 0.254; see electronic supplementary material, table S8, for all absolute values). On the following 2 days, testosterone levels of PMs remained substantially elevated, while levels of CMs further decreased. Consequently, PMs had significantly higher relative testosterone values than CMs (d2: W = 1, n = 11, p = 0.002; d3: W = 2, n = 11, p = 0.003). The same was found for absolute values (d2: W = 4, n = 11, p = 0.007; d3: W = 10, n = 11, p = 0.042), although the difference was not statistically significant on day 3 after correction for multiple comparisons. Subsequently, testosterone levels of PMs returned to baseline, but relative concentrations were still significantly higher in PMs than in CMs on days 15 and 22 (d8: W = 16, n = 11, p = 0.147; d15: W = 0, n = 7, p = 0.016; d22: W = 3, n = 8, p = 0.039). Absolute testosterone concentrations, however, did not differ between PMs and CMs during the final weeks (d8: W = 15, n = 11, p = 0.123; d15: W = 16, n = 7, p = 0.813; d22: W = 6, n = 8, p = 0.109).

Figure 4.

Endocrine measures. Median (first to third quartiles) of (a) plasma testosterone and (b) plasma cortisol concentrations on days 1, 2, 3, 8, 15 and 22 after onset of competition. Values are presented as percentage of baseline values determined 20 h before onset of competition (day 0). Numbers above the x-axis denote sample size. Statistics: changes over time, Skillings–Mack test: d.f. = 6, n = 13; testosterone: PMs: p < 0.001, CMs: p = 0.024; cortisol: PMs: p < 0.001, CMs: p < 0.001; pairwise comparisons, Wilcoxon signed-rank test (two-tailed) with Benjamini–Hochberg correction: *p ≤ 0.039, **p ≤ 0.003, ***p ≤ 0.001.

(ii). Plasma cortisol

One day before the competition, cortisol values did not differ between PMs and CMs (PMs: 229 ng ml⁻¹, Q₁–Q₃: 171–271 ng ml⁻¹; CMs: 290 ng ml⁻¹, Q₁–Q₃: 238–373 ng ml⁻¹; W = 37.5, n = 13, p = 0.6). When placed together with females, both responded with a significant increase in circulating cortisol (figure 4b; PMs: χ² = 27.87, p < 0.001; CMs: χ² = 28.23, p < 0.001; d.f. = 6, n = 13). However, PMs showed a greater elevation within the first 4 h and a delayed decrease afterwards compared with CMs. As a result, PMs had consistently higher relative cortisol levels than CMs, but these differences were not statistically significant after correction for multiple comparisons (d1: W = 15, n = 13, p = 0.033; d2: W = 29, n = 12, p = 0.47; d3: W = 7, n = 11, p = 0.019; d8: W = 3, n = 9, p = 0.02; d15: W = 1, n = 6, p = 0.063; d22: W = 3, n = 8, p = 0.039). For absolute concentrations, see the electronic supplementary material, table S9.

(iii). Body weights

One day before the competition, body weights of PMs and CMs did not differ significantly (PMs: 754 g, Q₁–Q₃: 706–838 g; CMs: 799 g, Q₁–Q₃: 778–856 g; W = 29, n = 13, p = 0.266). Both showed a significant decrease in response to the competition (electronic supplementary material, figure S10; PMs: χ² = 35.11, p < 0.001; CMs: χ² = 40.0, p < 0.001; d.f. = 6, n = 13). However, the acute decline within the first 4 h was significantly greater in PMs than in CMs (W = 5, n = 13, p = 0.002). At later time points, no significant differences were found between PMs and CMs (d2: W = 27, n = 13, p = 0.216; d3: W = 27, n = 11, p = 0.638; d8: W = 23, n = 11, p = 0.413; d15: W = 11, n = 8, p = 0.383; d22: W = 7, n = 8, p = 0.148).

(d). Reproductive success

In 12 of 13 groups, at least one female was impregnated and gave birth afterwards. This resulted in 19 litters with a total of 51 offspring (median litter size: 3; range: 2–5). Parentage was successfully assigned for all offspring. Multiple paternity was detected in one case; all other 18 litters were fathered by a single male. In sum, PMs accounted for 76.3% of all litters (14.5 of 19) and 80.9% of all offspring (41 of 51). When comparing male reproductive success within each group (figure 3b), PMs sired significantly more offspring (W = 12, n = 12, p = 0.034) and also tended to sire more litters (W = 12.5, n = 11, p = 0.075) than CMs.

In order to determine factors that are involved in mediating reproductive success, both males of a group were categorized according to their dominance index on the first day and their initial body weight. The comparison of reproductive success among these categories revealed that dominant males fathered significantly more litters (W = 0, n = 11, p < 0.001) and offspring (W = 0, n = 12, p = 0.001) than their subordinate competitors. By contrast, the body weight prior to the competition had no significant effect on reproductive success (litters: W = 28.5, n = 11, p = 0.79; offspring: W = 30, n = 12, p = 0.525).

4. Discussion

(a). Is adolescence a time window for adaptive shaping of the phenotype?

There is an increasing number of studies showing remarkable phenotypic plasticity during adolescence [9,32]. These findings raise the intriguing hypothesis that adolescence represents a sensitive period for adjusting the phenotype to environmental conditions [8–10]. The social modulation of behavioural phenotypes in male guinea pigs is a likely candidate for an adaptive response to the adolescent environment. Owing to the social environment experienced during adolescence, males develop distinct phenotypes that are assumed to adjust the animals to different environmental conditions (i.e. population densities) [8,9]. In fact, it has long been recognized that CMs are better adapted to complex group-living conditions. When introduced into large mixed-sex colonies, their low aggressiveness enables them to integrate in a non-stressful way, whereas the elevated aggression of PMs prevents integration and causes intense physiological stress responses, suppression of behaviour and possibly even death if the males are not removed from the environment [8,19]. Although reproductive success has not been determined experimentally, these results strongly suggest that the queuing strategy of CMs is superior to the high-aggressive strategy of PMs in such a situation [8,9]. However, a full factorial experimental design is necessary for a comprehensive understanding of the functional consequences of the phenotypic shaping [33,34].

Accordingly, this study investigated whether the high aggressiveness of PMs confers a fitness advantage in a low-density social setting in which a male directly competes for mating access, because in this situation overt fighting rather than queuing appears to be more promising to enhance reproductive success. In agreement with our expectations, the high-aggressive strategy of PMs led to significantly higher reproductive success under these conditions. Admittedly, we did not measure lifetime reproductive success but argue that the ability to reproduce during the competition represents a reliable fitness estimate. Hence, the high-aggressive strategy of PMs clearly improved fitness in an environment with a limited number of competitors.

Combining the present results with the previous knowledge that the queuing strategy of CMs increases fitness in larger social groups, we provided compelling evidence that the phenotypes of male guinea pigs are adaptively shaped during adolescence so as to be adjusted to different environmental conditions. Therefore, we can exclude a so-called silver spoon effect of the adolescent social environment, which would be supported when a particular phenotype is advantageous regardless of the environmental conditions in which they are tested [34]. Rather, the different phenotypes appear to represent adaptations to the environment presumably predicted by the social experiences during adolescence [8,9], and thus support an environmental matching effect [33,34].

Despite a growing body of theoretical support, empirical evidence for functional consequences of phenotypic plasticity during adolescence has been limited. In one study, social defeat during adolescence seemed to increase coping effectiveness of male rats in response to comparable stressful situations in later life [13]. Recent work in zebra finches [14,35] revealed that social experiences during adolescence shaped the development of males' sexual and aggressive behaviour in ways that remarkably resemble the observed effects in guinea pigs: pair-housed birds developed a more aggressive phenotype, whereas group-housed individuals were less aggressive and therefore integrated more easily into groups of unfamiliar conspecifics. The authors also argued that these behavioural modifications may represent adjustments to different social environments [14,35], though the experiments did not assess reproductive success as a fitness measure. The current results build upon and extend previous findings, and together strongly support the conclusion that adolescence is a time window for adaptive shaping of the phenotype. Moreover, the social modulation of behavioural phenotypes during this life stage might be a general mechanism in group-living animals to alter and canalize behaviour in an adaptive fashion, so that earlier influences on development can be complemented and readjusted, if necessary, to meet current environmental conditions [8,9].

(b). Underlying mechanisms

(i). Behavioural mechanisms

Consistent with previous studies [16,36], PMs displayed much more aggression towards the male opponent than did CMs right from the initial hours of the competition. In social systems, aggressive behaviour plays an essential role in establishing dominance relationships that ultimately aim to reduce agonistic encounters in competition for limited resources [37]. Hence, the high aggressiveness allowed most PMs to immediately become dominant over their colony-housed opponents. In guinea pigs, as in many other species, dominant males have preferential access to mating opportunities [38,39]. Accordingly, dominant males sired more offspring than their subordinate competitors in all groups (apart from one group that produced no offspring at all). Notably, this effect was independent of the males' body weights. Thus, the high-aggressive phenotype of PMs that had been shaped by social conditions during adolescence led to higher reproductive fitness when later competing with a single male for mating access.

In addition to greater aggression directed towards the male competitor, PMs also engaged in more social interactions with females. PMs initiated courtship and sexual activities earlier and displayed them more intensely than CMs from the onset of the competition, which is well in line with previous observations [16,36]. Likewise, PMs exhibited more sociopositive behaviour towards females than did CMs during the following days. The females themselves showed no differences in approaching the males on the first day of the competition. However, they approached PMs more frequently than CMs during subsequent observations. This probably indicates that females developed a preference for PMs, which, however, was not predetermined but rather resulted from the males' different behavioural strategies to cope with the competition. In a variety of species, including wild cavies, females select mates based on particular phenotypic characteristics that are likely to indicate fitness benefits for themselves or their offspring, such as courtship activities, dominance status and other related traits [40,41]. Hence, the females might have played a considerable role in determining male reproductive success by preferentially mating with PMs, which attained a dominant status owing to their strategy of high aggressiveness and courtship activities. Interestingly, these results again closely parallel those observed in zebra finches, in which males housed in mixed-sex pairs during adolescence were more aggressive towards male conspecifics as well as more attractive to unfamiliar females than group-housed males [14].

(ii). Neuroendocrine mechanisms

Our understanding of how behavioural and neuroendocrine processes mediate the shaping of phenotype by social experiences during adolescence has improved greatly in recent years [9]. As predicted by the challenge hypothesis, social interactions stimulate testosterone secretion in males of various species [42,43], including guinea pigs [17]. Interactions with conspecifics—particularly courtship and agonistic encounters—are generally more frequent under group-living conditions. Consequently, circulating testosterone is notably higher in CMs compared with PMs [17], as was the case in this study for baseline levels determined prior to the competition. Testosterone is an important component in regulating hypothalamic–pituitary–adrenal (HPA) function during perinatal and adolescent development [12] and has an inhibitory effect on acute cortisol responsiveness in male adolescent guinea pigs [18]. Thus, taken together, the enhanced level of social interactions within a group triggers testosterone secretion, which, in turn, reduces the cortisol responsiveness of CMs during adolescence. As acute increases in plasma glucocorticoids promote aggressive behaviour in several species [44,45], the magnitude of HPA responsiveness is assumed to be part of the mechanisms underlying the different levels of aggressiveness observed in PMs and CMs [9].

In agreement with the stimulating effect of social interactions on testosterone secretion [42,43], circulating testosterone was substantially elevated in PMs following the onset of competition so that the relative elevation was significantly higher than in CMs—especially during the initial days of the encounter when dominance relationships were established. Winning (i.e. dominating) a competition is often associated with increased testosterone levels, which appear to facilitate an enhanced ability to win future encounters [46]. Moreover, testosterone has a key function in promoting courtship and mating [47]. Hence, the increased testosterone levels of PMs seem to have promoted dominance acquisition as well as reproductive activities in the observed situation.

The competition caused significant increases in cortisol levels of males from both adolescent social conditions. However, PMs showed consistently higher elevations of cortisol levels than did CMs (although this difference was not statistically significant after correction for multiple comparisons). The combination of this enhanced HPA activity and the significant elevation of testosterone in PMs is likely to be the neuroendocrine basis of their high-aggressive strategy. Aggression as well as courtship activities during mate competition and the related surges in testosterone are associated with high energetic costs [48]. As glucocorticoids are crucially involved in regulating essential metabolic processes [49], the high cortisol levels observed in PMs probably not only triggered aggression but also mobilized the required energy. In support of this assumption, PMs lost considerably more body weight than CMs during the initial hours, when cortisol and testosterone levels were most elevated and aggressive and courtship activities were highest. The release of glucocorticoids is often viewed only as a response to adverse conditions, often with pathological consequences. It is important to note, however, that elevated glucocorticoid levels have significant adaptive functions and may not necessarily indicate a stressful situation, but rather be a response to increased metabolic demands required for behavioural activity [50]. As found in this study, enhanced HPA responsiveness may be beneficial under certain conditions by ultimately increasing reproductive success. Still, prolonged elevations of glucocorticoids can also have detrimental effects, such as suppressing immune function [49], which may affect health. Therefore, it is possible that the prolonged HPA activity of PMs has long-term fitness costs, which may eventually cancel out initial benefits. However, it cannot be concluded from this study whether this also applies for the observed situation.

(iii). Sensitive period for the shaping process?

Consistent with earlier findings, we showed that the phenotypic differences resulting from social experiences beginning at one month of age have emerged by late adolescence at about four months of age. PMs do not differ from CMs in circulating testosterone or HPA responsiveness before attaining sexual maturity at two to three months of age [17]. In addition, the striking effects of adolescent social experiences on phenotypic development can be observed even when the different housing conditions are imposed as late as two months of age [36] and persist until full adulthood when males remain in their housing conditions [19]. Hence, the period prior to sexual maturity appears to be less important in shaping reproductive strategies. However, based on the current knowledge, it remains unclear whether the sensitive period is restricted to adolescence or also extends to later life stages.

5. Conclusion

The development of behavioural and neuroendocrine phenotypes is profoundly shaped by the social environment during adolescence. As previously demonstrated, the exposure to large mixed-sex colonies during this life stage leads to a phenotype of low aggressiveness and low cortisol responsiveness as part of an effort-saving queuing strategy that facilitates integration into social groups. In this study, we showed that a strategy of high aggressiveness and high cortisol responsiveness that is shaped by pair-housing conditions during adolescence confers a reproductive benefit when competing directly with an opponent for mating access. In summary, these data provide the clearest evidence to date for adaptive shaping of the behavioural and neuroendocrine phenotype by environmental influences during adolescence.

Ethics

All procedures complied with the regulations covering animal experimentation within the EU (European Communities Council Directive 2010/63/EU) and were approved by the local and national authorities (LANUV; reference number: 84-02.05.20.12.117).

Data accessibility

Data are available from the Dryad Digital Repository: http://dx.doi.org/10.5061/dryad.fb1s4 [51].

Competing interests

The authors declare that they have no competing interests.

Authors' contributions

T.D.Z. participated in the design of the experiment, collected the data, carried out statistical analyses, prepared the initial draft and created the figures. All other authors critically revised the manuscript for important intellectual content. M.B.H. provided conceptual and statistical input. S.K. and N.S. conceived of the study, designed the experiment and supervised the work. All authors gave final approval for publication.

Funding

This research was funded by grants of the German Research Foundation (DFG) to N.S. and S.K. (FOR 1232, Sa389/11-2, Ka1546/9-1).