Social niche shapes social behavior and cortisol concentrations during adolescence in female guinea pigs

Individualized social niches arise in social groups, resulting in divergent social behavior profiles among group members. During sensitive life phases, the individualized social niche can profoundly impact the development of social behavior and associated phenotypes such as hormone (e.g. cortisol) concentrations. Focusing on adolescence, we investigated the relationship between the individualized social niche, social behavior, and cortisol concentrations (baseline and responsiveness) in female guinea pigs. Females were pair-housed in early adolescence (initial social pair formation), and a social niche transition was induced after six weeks by replacing the partner with either a larger or smaller female. Regarding social behavior, dominance status was associated with aggression in both the initial social pairs and after the social niche transition, and the results suggest that aggression was rapidly and completely reshaped after the social niche transition. Meanwhile, submissive behavior was rapidly reshaped after the social niche transition, but this was incomplete. The dominance status attained in the initial social pair affected the extent of submissive behavior after the social niche transition, and this effect was still detected three weeks after the social niche transition. Regarding cortisol concentrations, higher baseline cortisol concentrations were measured in dominant females in the initial social pairs. After the social niche transition, cortisol responsiveness significantly increased for the females paired with a larger, older female relative to those paired with a smaller, younger female. These findings demonstrate that the social niche during adolescence plays a significant role in shaping behavior and hormone concentrations in females.


Introduction
Despite inhabiting the same physical and social environment, individuals living together in a social group differ in how they interact with one another.This unique interaction with the social environment influences the individual's fitness and constitutes the individualized social niche (Bergmüller and Taborsky, 2010;Montiglio et al., 2013;Saltz et al., 2016;Trappes et al., 2022).The individualized social niche concept is integral to understanding the social environment on the individual level (Trappes et al., 2022).Social niches of individuals that regularly interact are not independent and an individual's social niche is often influenced by characteristics such as body size, age, or behavioral traits relative to that of groupmates (Webster and Ward, 2011;Montiglio et al., 2013).The individualized social niche is not necessarily stable throughout the lifespan.When the social environment changes through, for example, changes to group membership, social niche transitions can ensue whereby behavior and physiological correlates are reorganized to a new individualized social niche (Bergmüller and Taborsky, 2010;Mutwill et al., 2020).This process by which social behavior or internal states change in response to the social environment and social interactions is referred to as social niche conformance (Trappes et al., 2022;Kaiser et al., 2024).
Depending on the social organization of a species, different individualized social niches can develop.One well-known example of a social niche is dominance status, since dominant and subdominant individuals systematically differ in their interaction with the social environment (Bergmüller and Taborsky, 2010;Montiglio et al., 2013;Briffa et al., 2015;Saltz et al., 2016;Laskowski et al., 2022;Trappes et al., 2022).Social groups are structured by dominance relationships in many species, and dominance relationships between individuals generally develop in two different ways.Put simply, the social dynamics hypothesis postulates that dominance relationships can develop via social interactions among individuals (Chase et al., 2002).Meanwhile, the prior attributes hypothesis states that dominance relationships can arise from individual differences in factors such as body size, age, or previous social experience (Chase et al., 2002).Therefore, dominance relationships are an emergent property of interacting individuals (Drews, 1993), and the phenotypes of both individuals are important when establishing dominance relationships (Wilson et al., 2011;Schneider et al., 2017).These phenotypes can subsequently be shaped by the outcome of dominance interactions, forming feedback loops which affect future dominance interactions (Sih et al., 2015;Dehnen et al., 2022;Laskowski et al., 2022).Dominance status can have consequences for behavior (Briffa et al., 2015;Niemelä and Santostefano, 2015), hormone concentrations (Creel, 2001;Creel et al., 2013), and fitness outcomes (Chelini et al., 2011;Côté and Festa-Bianchet, 2001;Pusey et al., 1997).Changes to the social environment can result in a shift in an individual's dominance status, triggering a reorganization of correlated phenotypes such as behavior, body size, and hormone concentrations (Alonso et al., 2012;Kohn et al., 2016;Snyder-Mackler et al., 2016;Rudin et al., 2017).
Dominance relationships reduce conflict, protecting group members from costly escalated aggression (Tibbetts et al., 2022).Nevertheless, establishing and navigating dominance relationships can be stressful, and glucocorticoid (e.g.cortisol) concentrations can reflect the challenges associated with dominance status (Creel, 2001;Goymann and Wingfield, 2004;Creel et al., 2013;Korzan and Summers, 2021).Controlled by the hypothalamic-pituitary-adrenal (HPA) axis, glucocorticoids fluctuate throughout the day within their baseline range but rapidly rise in response to acute challenges, with wide-ranging effects on biological processes such as metabolism, reproduction, and behavior (Sapolsky et al., 2000).The relationship between glucocorticoid concentrations and dominance status has been studied across a wide range of taxa.The direction of this relationship varies; sometimes dominant individuals have higher glucocorticoid levels (Creel et al., 1996;Fichtel et al., 2007) while other studies find that dominant individuals have lower glucocorticoid levels (Alonso et al., 2012;Williamson et al., 2019) or no relationship is detected (Sachser et al., 1998;Arce et al., 2010;Barrette et al., 2012;Zipser et al., 2013;Mutwill et al., 2021;Rystrom et al., 2022).Even within the same species, males and females can differ in the direction of this relationship (Cavigelli and Caruso, 2015;Ode et al., 2015).Generally, whether dominant or subdominant individuals have higher glucocorticoid concentrations depends on the types of social behaviors expressed, the social organization of the species or population, and the impact of status on reproductive success (Creel et al., 2013).Furthermore, when an association is detected, it is often unclear whether glucocorticoid concentrations are the cause or the consequence of dominance status (Sapolsky, 2004).It has both been shown that manipulation of dominance status causally affects glucocorticoid concentrations (Kohn et al., 2016) and that selection for individuals with high or low HPA axis reactivity predicts aggression and dominance acquisition (Pottinger and Carrick, 2001;Touma et al., 2008).
The relationship between individualized social niches and glucocorticoid concentrations is particularly interesting to investigate in adolescence.During adolescence, individuals increasingly interact with unrelated conspecifics and experience the social environment directly rather than indirectly via maternal cues.The first reproductive opportunities arise; therefore, adjusting phenotypes based on information from the current social environment carries fitness benefits (Fawcett and Frankenhuis, 2015).Additionally, the HPA axis and social behavior undergo extensive maturation during adolescence, and development is affected by experiences in this life phase (Romeo, 2018;Sachser et al., 2018).Indeed, social behavior (Zimmermann et al., 2017a;Jäger et al., 2019;Sachser et al., 2020) as well as glucocorticoid concentrations (Zimmermann et al., 2017a;Emmerson and Spencer, 2018;Mutwill et al., 2020) are shaped by social experiences during adolescence and even into adulthood.
Guinea pigs (Cavia aperea f. porcellus) are an excellent model species to investigate the effect of social interactions on phenotypes during adolescence.Guinea pigs are social mammals and have a long adolescence during which environmental cues can shape phenotypes.After weaning at 21 days of age, sexual maturity occurs at approximately 1 month in females and 70 days in males, and full body size is reached at 8-12 months (Kaiser et al., 2010).Males become socially mature enough to become dominant in complex social environments at the age of seven months (Sachser, 1986), and it is well-established that the social environment during adolescence and adulthood shapes social behavior and hormone concentrations in male guinea pigs (Lürzel et al., 2010(Lürzel et al., , 2011;;Zimmermann et al., 2017aZimmermann et al., , 2017b;;Sachser et al., 2018;Mutwill et al., 2020Mutwill et al., , 2021)).However, it is not known whether there are comparable developmental trajectories in females.Female guinea pigs form linear dominance hierarchies that are temporally stable in adulthood in larger groups, with older and larger females generally becoming dominant over smaller and younger females (Thyen and Hendrichs, 1990;Rystrom et al., 2022).In large female groups housed in the laboratory, dominance relationships are unstable early in adolescence, but stable dominance relationships emerge at approximately 60 days of age (Thyen and Hendrichs, 1990).
The aim of this study was to investigate whether social behavior and hormone concentrations are shaped by the social niche during adolescence in females.We pair-housed female guinea pigs upon weaning for the first six weeks of adolescence and observed dominance interactions in the home enclosure to determine whether the focal female became dominant or subdominant.We predicted that dominance status attained early in adolescence would be associated with behavior and cortisol concentrations.Next, to induce a social niche transition, the social partner from early adolescence was exchanged for either a much larger or much smaller female social partner.By changing the social partner in this manner, the dominance relationship between the focal female and her social partner was manipulated.The social context was also manipulated; now the two interacting females differed in prior attributes such as body weight, age, and social experience.Our hypothesis was that behavioral profiles and cortisol concentrations would be readjusted when the social niche changes later in adolescence.However, since adolescence is a sensitive life phase for the development of social behavior and HPA axis organization (Romeo, 2018;Sachser et al., 2020), we expected that the social experience from early adolescence would also have an effect on behavior and cortisol concentrations even after the social niche transition.

Animals and housing conditions
Animals used in this experiment were bred at the Department of Behavioural Biology in Münster from a breeding program of multicolored shorthaired guinea pigs (Cavia aperea f. porcellus).Breeding groups consisted of one male with two to three females and their preweaned offspring.Therefore, parentage was known for all individuals.In total, 48 focal females and 75 partner females were included in this experiment.
Throughout the experiment, the focal female was housed with her partner in an enclosure measuring 1 m by 0.5 m, with a wall height of 0.5 m.Walls were constructed of wood with an opaque red plastic section at the bottom (approximately 7 cm in height).The floor was covered with wood shavings (Tierwohl Super, J. Rettenmaier & Söhne GmbH + Co KG, Rosenberg, Germany), and food (Höveler Meerschweinchenfutter 10700, Höveler Spezialfutterwerke GmbH & Co. KG) and water with ascorbic acid was available ad libitum.Hay was replenished daily.Two small shelters were placed on opposite ends of the enclosure; one was made from wood and one was made from red transparent plastic.These shelters were rectangular (9 cm by 20 cm) with a height of 13 cm.As a door, one of the smaller walls was missing and only one female comfortably fit within each shelter.Lights were on daily from 7:00-19:00 and room conditions were kept at 20-24 • C and 40-70 % humidity.
Partner females were housed in one large group when not paired with a focal female.The size of this enclosure fluctuated based on the number of partner females within.Various shelters were provided, and all other enclosure and room conditions were comparable to the experimental enclosures.Some partner females were used with two different focal females.If so, they were housed in the large female group for at least one week before being housed with the second focal female.

Experimental procedure
For this experiment, adolescent female guinea pigs were pair-housed with another adolescent female for six weeks, and then a social niche transition was induced by pairing the focal female with a different partner female (for overview see Fig. 1).When the focal female was weaned (21 ± 3 days), she was transferred from her natal group to a new enclosure with another female of a similar age.The maximum age difference between the two females was five days, and the partner female was also directly transferred from her natal group to the enclosure with the focal female.The two females were weighed and simultaneously placed in their new shared enclosure at 9:00 ± 15 min.The social behavior was recorded for three hours immediately after transfer (d1), for one hour on the third day (d3), and for one hour weekly for the following five weeks (for more information see Methods subsection Social behavior observations).Additionally, cortisol concentrations were measured at 13:00 ± 15 min on d1 and d3.Baseline cortisol along with cortisol response one and two hours after the onset of a psychological stressor was measured via a cortisol response test (CRT) on the day prior to beginning the experiment (CRT1) and the day before the social niche transition (CRT2).Cortisol concentrations were determined from blood samples (for more information see Methods subsection Cortisol response test (CRT)).Body weight of the focal female was recorded after each blood sample.The partner female also experienced a modified CRT on d0 so that both females had a challenging experience prior to meeting one another since stress is known to affect the outcome of dominance interactions (Cordero and Sandi, 2007).In the modified CRT, partner females were placed singly in a CRT arena for two hours, but no blood or saliva samples were taken.Body weight of the partner female was recorded at the beginning and end of the modified CRT1 on d0, on d1 directly before being housed with the focal female, in the afternoon of d1, and the day prior to the social niche transition when the focal female underwent the CRT2.
The social niche transition was triggered on the day after the CRT2 by changing the housing partner of the focal female.At the social niche transition, the age of the focal females was 63 ± 4 days.The new female partner was either markedly older and larger (52 ± 21 days older, 220 ± 137 g heavier) or younger and smaller (35 ± 9 days younger, 193 ± 100 g lighter) than the focal female.This forced the focal females into a specific dominance status, with 42 out of the 44 focal females acquiring the dominance status expected due to disparities in age and weight.This created a full-factorial design whereby some females maintained their previous dominance status but with a new partner and other females were forced to transition to a dominance status that they had not yet experienced.The focal female was placed simultaneously with the new partner female in an enclosure that was new for both individuals at 9:00 ± 15 min.Social behavior was observed for three hours immediately after transfer (d1), for one hour in the afternoon of the third day (d3), and for one hour weekly until the third week.Additionally, cortisol concentrations were measured at 13:00 on d1 and d3.Baseline cortisol along with cortisol responsiveness one and two hours after the onset of a psychological stressor was measured via a CRT both one week (CRT3) and three weeks (CRT4) after the social niche transition.Focal females were weighed directly before being placed with the partner female on d1 and after each blood sample collection.All partner females were weighed directly before being placed with the focal female, in the afternoon of the first day, and on the final day of the experiment.
Estradiol concentration was measured to control for any possible effect of estradiol on cortisol concentrations.This was done by analyzing saliva samples that were collected directly after baseline cortisol (in CRTs) and cortisol measurements on d1 and d3 after the initial pair formation and social niche transition (for more details see Methods subsections Cortisol response test (CRT) and Statistical analysis).
Of the 48 focal females, 44 completed the experiment because four females became ill in the initial social pairs and were excluded from the experiment.Of the 44 females who completed the experiment, 41 had full datasets.This is because one female did not meet our criterion for establishing a dominance relationship in the first pair (see Supplement), one female achieved the status opposite of what was intended in the second pair, and one female both did not meet our criterion for establishing a dominance relationship in the first pair and achieved the status opposite of what was intended the second pair.In total, 25 females became dominant and 17 females became subdominant in the first pair.For the initial social pair formation, focal females were pair-housed with an adolescent female partner for six weeks.Then, a social niche transition occurred in which focal females were housed with either a larger and older or a smaller and younger female partner for three weeks.This social niche transition manipulated the dominance status of the focal female.Videos of social behavior in the home enclosures were recorded on d1 (3h), d3 (1h) and weekly (1h).Cortisol response tests (CRT1-4) occurred the day before the start of the initial social pair formation (CRT1), on the day before the social niche transition (CRT2), approximately one week after the social niche transition (CRT3), and again three weeks after the social niche transition (CRT4).Baseline cortisol concentration was additionally measured on d1 and d3 after both the initial social pair formation and social niche transition.
19 females became dominant while 22 became subdominant in the second pair (for an overview of sample size for each treatment, see Fig. 1).

Social behavior observations
Social behavior was observed using video recordings of the home enclosure that were taken using a camera mounted approximately 1 m above the enclosure.These videos were recorded for the first three hours on d1, for one hour in the afternoon of d3, and one hour each week thereafter (mornings, between 8:00 and 12:00).For all recordings aside from d1, fresh hay was added to the enclosure and a small piece of fruit or vegetable was placed in the food bowl to encourage social interactions between the focal female and her partner.After recording, videos were briefly checked.If the animals were inactive, the video recording was repeated at an appropriate time.
The behavioral coding software Interact (Interact, Lab Suite Version 2017, Mangold International GmbH) was used to count behaviors.All behaviors are defined in the ethogram in Table S1, and definitions are adapted from previous work with cavies (Rood, 1972;Kaiser et al., 2015;Sangenstedt et al., 2018).Behaviors observed included approach, follow, sniffing behaviors (body sniff, nasal sniff, anal sniff), aggressive behaviors (brawl, chase, curved body posture, fixate, flip around, head thrust/bite, head up, kick/urine spray, rumba), play behavior (frisky hop, head twitch, run), and retreat.While rumba is generally seen as a courtship behavior in males, we observed it often in dominance interactions and have here included it as an aggressive behavior (for more information see Supplementary Note).Additionally, the amount of time that the two females spent together inside one shelter was counted.
Video analysis was done in successive blocks.First, all videos from week 4 through week 6 after initial social pair formation were analyzed.In case of a missing video, week 3 was also analyzed (N = 2).It was necessary to start analyzing behavior from the first social pairs before the end of the experiment in order to manage the social niche transition for the remaining focal females with the goal to balance the treatments and prevent full siblings from being in the same treatment.Next, all videos from week 3 after the social niche transition were analyzed.Finally, all videos from d1 (of initial social pair formation and social niche transition) were analyzed.Videos in each block were analyzed in a random order and the individual identity of the focal female was blinded, although the observer was aware which female in the pair was the focal female.After the social niche transition, it was not possible to blind the observer to dominance status due to the size difference of the focal and partner females.All videos were analyzed by one observer who was extensively trained in guinea pig social behavior.Furthermore, intraobserver reliability was assessed by re-analyzing one randomlyselected video from 15 females from the first block (week 4 through 6 after initial social pair formation) after the completion of all video analysis (see supplement for more information).

Dominance status determination
A rank index was calculated to assist in the determination of the dominance status of the focal females (see supplementary material).The rank index was calculated as the proportion of total retreats in which the partner retreated from the focal female.A retreat was broadly defined as: A female moves away from another female so that she maintains a distance of more than one body length; this behavior is shown either after an interaction of the females or after an approach of one of the females involved.A rank index was calculated from each video observation for each female and had a range between 0 (focal female completely subdominant) and 1 (focal female completely dominant).
To assess the outcomes of dominance interactions, the rank index was inverted and used as a measure of submissive behavior.Submissive behavior was analyzed on d1 of initial social pair formation to determine whether the dominance relationships observed in weeks four, five, and six were already present on d1.Submissive behavior was also analyzed on d1 and wk3 after the social niche transition to determine whether the initial dominance status attained early in adolescence affected outcomes of future dominance interactions with a different social partner.

Cortisol response test (CRT)
The cortisol response test is described in detail elsewhere (Rystrom et al., 2022).Briefly, the focal females are placed alone in a novel enclosure and baseline cortisol as well as the cortisol response after one and two hours is measured.The focal female was collected from her home enclosure and a baseline blood sample was taken within three minutes of entering her housing room by pricking a blood vessel in the ear with a sterile blood lancet (Solofix® Blutlanzetten, B. Braun Melsungen AG, Melsungen, Germany) and collecting approximately 150 μl of blood in two capillary tubes (Capillary tubes for microhaematocrits, 100 μl, Paul Marienfeld GmbH & Co KG, Lauda Königshofen, Germany).
Afterward, a saliva sample was taken by inserting a cotton bud (Sterile applicators, Carl Roth GmbH + Co. KG, Karlsruhe, Volume 13 ml) into the mouth of the focal female and encouraging her to chew on it by slowly twirling the cotton bud.A complete saliva sample consisted of two saturated cotton buds.The female was then weighed and placed into the CRT arena located in a different housing room.The CRT arena was m 2 , built similarly to the housing enclosures, and contained food and water but no shelter.Exactly one hour after the onset of the test, the focal female was removed from the arena, the response blood sample was taken within three minutes, and the female was weighed and returned to the enclosure.This procedure was again repeated exactly one hour later so that each female had three blood samples taken (baseline cortisol concentration, cortisol response after one hour, and cortisol response after two hours), one saliva sample at the first sampling timepoint (estradiol), and was weighed three times.
Blood samples were immediately prepared and frozen at − 20 • C (for more details see Rystrom et al., 2022).Cortisol concentrations of the samples were determined using enzyme-linked immunosorbent assays (Cortisol ELISA, RE52061, IBL International GmbH, Hamburg, Germany) once all samples were collected.Details regarding the analysis of cortisol concentrations in guinea pigs including validation data is provided in the supplement.The intra-assay variances were on average CV = 2.98 % and the inter-assay variances were on average CV = 3.51 %.
Saliva samples were also immediately prepared.The cotton bud tips were cut off and placed into a 1.5 ml Eppendorf tube which had a hole punctured in the bottom and was stacked within a 2 ml Eppendorf tube.These stacked tubes were then centrifuged at 13000 rpm for 10 min so that the saliva was extracted from the cotton bud and collected in the ml Eppendorf tube.The saliva was then pipetted into a sterile 1.5 ml Eppendorf tube and centrifuged at 13000 rpm for 3 min.This was repeated until no visible pellet remained.The saliva samples were then frozen at − 20 • C. All saliva samples were analyzed upon completion of the experiment.
The estradiol concentration in the saliva was analyzed with an invitro 17β-estradiol saliva luminescence immunoassay (Estradiol ELISA, RE62141/RE62149, IBL International GmbH, Hamburg, Germany) as described by IBL International GmbH (2016).This ELISA is usually used for the estradiol analysis of human saliva but has been standardized and used for guinea pig saliva at the Department of Behavioural Biology in Münster.Details regarding the analysis of estradiol concentrations in guinea pigs including validation data is provided in the supplement.The intra-assay variances were CV = 9.97 % and the inter-assay variances were CV = 11.30%.

Statistical analysis
Data analysis was carried out with R version 4.1.1(R Core Team, 2021).In general, two-tailed tests were used and significance was set at P < 0.05.For simple associations between parameters such as behavior, body weight, and mass with dominance status in the initial social pairs, non-parametric Wilcoxon signed rank tests were used.Paired Wilcoxon tests were used to analyze behavior on d1 in the two social pairs.Effect sizes were calculated as r using the rstatix package (version 0.7.0;Kassambara, 2023).Linear models and linear mixed-effect models (package lme4 version 1.1-27.1;Bates et al., 2015) were used to assess the relationship between cortisol concentrations and dominance status as well as the relationship between behavior after the social niche transition and dominance status.Continuous explanatory variables were meancentered and categorical explanatory variables were contrast coded.Therefore, in the model output, the intercept indicates the grand mean and the estimate for categorical variables indicates the difference between the two levels.ANOVA type III tables were generated with the lmerTest package (version 3.1-3; Kuznetsova et al., 2017) for mixedeffect models and the car package (version 3.0-11; Fox and Weisberg, 2019) for linear models.For linear models, effect sizes were calculated as partial-Eta squared (η2) using the effectsize package (version 0.8.5;Ben-Shachar et al., 2020).The performance package (version 0.7.3;Lüdecke et al., 2021) was used to verify model assumptions.
Adjusted repeatability of rank indices calculated for weeks 4-6 after initial social pair formation was estimated using the rptR package (Stoffel et al., 2017).Number of parametric bootstraps was set to 1000 and permutation was set to 500.Behavior after the social niche transition on d1 and week 3 was analyzed with linear models.Separate models were fit for each behavioral category."Follow" was not analyzed for week 3 because this behavior was observed in few females at this timepoint.
For the data from the CRTs, in order to select relevant variables, linear mixed-effect models were fit with one of the cortisol measurements (baseline, responsiveness after one hour, and responsiveness after two hours) as the response variable and body weight, age, and estradiol concentration as explanatory variables.Data from all four CRTs were included, and animal identity was included as a random effect.Both age and body weight had a significant effect on at least one of the cortisol measurements while estradiol concentration did not significantly affect any of the cortisol measurements (see Table S2).Therefore, estradiol concentration was excluded as an explanatory variable for all subsequent models.Age was included as an explanatory variable to control for its potential effect in only the model using data from CRT1.For all models with cortisol concentrations as a response variable, body weight was included as an explanatory variable to control for any possible effect.
Linear models were fit for the data from CRT1 and CRT2 separately.Dominance status was included as an explanatory variable, and for the model using the CRT2 data, the respective cortisol concentration measured at CRT1 was included as an explanatory variable.To analyze the cortisol concentrations after the social niche transition, linear mixedeffect models were fit with the data from CRT3 and CRT4 combined.Explanatory variables included dominance status in the initial social pair, dominance status after the social niche transition, and all interactions between CRT and the two dominance statuses.CRT (CRT3 or CRT4) was included as an explanatory variable to control for any effect.
Cortisol concentrations on d1 and d3 were analyzed separately for initial social pair formation and social niche transition.Data from d1 and d3 was combined into one model with cortisol concentration as the response variable.For the initial social pair formation model, explanatory variables included day (1 or 3), dominance status, and the interaction effect between dominance status and day.For the social niche transition model, both dominance statuses were included as well as day and all interactions between dominance statuses and day.Post hoc paired Wilcoxon tests were used to test whether cortisol concentrations on d1 and d3 were significantly different from baseline cortisol as measured on d0.

Compliance with ethical standards
All procedures complied with the regulations covering animal experimentation within Germany (Animal Welfare Act) and the EU (European Communities Council DIRECTIVE 2010/63/EU) and were approved by the local and federal authorities (Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen "LANUV NRW", reference number: 84-02.04.2015.A439).

Dominance status after initial pair formation
The rank index calculated four, five, and six weeks after the pair formation in early adolescence was significantly repeatable (R adj = 0.643, SE = 0.076, 96 % CI: [0.458, 0.758], P < 0.001).Focal females who became dominant or subdominant did not initially differ in their body weight (Wilcoxon test, N dom = 25, N sub = 17, r = 0.116, P = 0.471).Furthermore, within each pair, the initial difference in body weight and age between the focal female and the partner female was not associated with the dominance status achieved by the focal female (Wilcoxon test, N dom = 25, N sub = 17, body weight: W = 254, r = 0.207, P = 0.193, age: W = 200, r = 0.050, P = 0.755).

After the social niche transition
Behavior in the three hours directly after social niche transition was affected by the direction of the social niche transition as well as the dominance status experienced previously in the initial social pair (Table 1).The direction of the social niche transition had a significant effect on the amount of aggressive behavior observed (Table 1a; Fig. 3a); focal females who were paired with a smaller, younger partner female were significantly more aggressive than the focal females who were paired with a larger, older partner.Furthermore, the previous dominance status of the focal female from her first social pair significantly affected how often the focal female followed (Table 1c; Fig. 3c) and sniffed (Table 1e; Fig. 3e) her partner.Females who had previously been subdominant both followed and sniffed their new partner significantly more often than females who had previously been dominant (Fig. 3e).
Three weeks after the social niche transition, only aggressive behavior was significantly affected by the dominance status triggered by the social niche transition (Table 2a), with aggressive behavior observed more often in females paired with a smaller, younger partner than females paired with a larger, older partner (Fig. 4a).Dominance status attained in the early adolescence pair significantly affected outcomes of dominance interactions after the social niche transition.Females who had been dominant in the initial social pair and later became subdominant after the social niche transition were significantly more submissive than females who were subdominant in the initial social pair and later became dominant after the social niche transition.However, these proportions of submissive interactions were relatively intermediate compared to the females who had always been dominant (lowest proportion of submissive outcomes) and the females who had always been subdominant (highest proportion of submissive outcomes).This was observed both one day (Table 1f, Fig. 3f) and three weeks after the social niche transition (Table 2e, Fig. 4e).

Behavioral response to initial pair formation and social niche transition
Social behavior was observed for the three hours directly after social pair formation in early adolescence and social niche transition later in adolescence.Focal females were overall more active directly after the social niche transition than when the first social pair was formed in early adolescence.This included approach (Paired Wilcoxon test, N = 44, V = 12.5, r = 0.836, P < 0.001), follow (Paired Wilcoxon test, N = 44, V = 1, r = 0.854, P < 0.001), sniff (Paired Wilcoxon test; N = 44, V = 0, r = 0.870, P < 0.001), aggression (Paired Wilcoxon test, N = 44, V = 32, r = 0.788, P < 0.001), and play (Paired Wilcoxon test; N = 44, V = 33, r = 0.803, P < 0.001).Meanwhile, focal females spent more time within one shelter with their partner upon the initial social pair formation in early adolescence (Paired Wilcoxon test; N = 44, V = 754, r = 0.511, P < Table 1 Social behavior of the focal female directly after social niche transition.Dominance status in initial social pair and after social niche transition were contrast coded, and reference levels are "subdominant".Estimates were obtained using the summary() function and ANOVA type III tables were generated using the Anova() function from the car package.Effect size was calculated as partial η2 using the eta_squared function in the effectsize package.Differences were considered significant at p < 0.05 (bold); 0.05 ≤ p < 0.1 was considered a trend (italic).

0.001).
In the three hours after pair formation in early adolescence, females did not form a dominance relationship reflective of the stable relationship observed later in these pairs (i.e., proportion of submissive behavior did not significantly differ between focal females who were dominant and focal females who were subdominant in weeks 4-6; Wilcoxon test; N dom = 25, N sub = 17, W = 158, r = 0.176, P = 0.267).However, already on the first day after social niche transition, focal females who were partnered with a larger, older female were significantly more submissive than focal females who were partnered with a smaller, younger female (Table 2e, Fig. 4e).

Baseline cortisol concentrations
Baseline cortisol was measured throughout the experiment: at each cortisol response test as well as on the first and third day after pair formation in early adolescence and social niche transition later in adolescence.The only time when dominance status was significantly associated with baseline cortisol was six weeks after the initial pairs were formed (CRT2; Table 3), when the baseline cortisol concentration of dominant females was significantly higher than that of subdominant females (Table 3).
Cortisol concentration significantly changed from day 1 to day 3 both after the initial pair formation and the social niche transition.Cortisol concentration significantly increased in the first three days after the initial social pair formation (Table 4; Fig. 5).Cortisol concentration measured on d3 but not d1 was significantly higher than that measured on d0 (Paired Wilcoxon tests; d0-d1: N = 43, V = 320, r = 0.282, P = 0.066; d0-d3: N = 44, V = 254, r = 0.424, P = 0.005).Meanwhile, in the three days after the social niche transition, cortisol concentration significantly decreased (Table 4; Fig. 5).Cortisol concentrations at day 1 were significantly higher than those measured at the same time the day before (Paired Wilcoxon test; N = 43, V = 96, r = 0.694, P < 0.001), and cortisol levels had returned to the baseline level by day 3 (Paired Wilcoxon test; N = 43, V = 442.5,r = 0.018, P = 0.915).

Cortisol responsiveness
At the first cortisol response test, cortisol responsiveness after one or two hours did not differ between focal females who would later become dominant and focal females who would later become subdominant (Table S3).When measured again six weeks after the initial social pair formation, cortisol responsiveness after one and two hours were not significantly affected by dominance status (Table 3).
When measured one and three weeks after the social niche transition, cortisol responsiveness after one hour was not significantly affected by the dominance status in the previous social pair, the current dominance status triggered by the social niche transition, or any interactions between dominance status and CRT (Table 5b).
Cortisol responsiveness after two hours was significantly affected by the interaction between the dominance status triggered by the social niche transition and time (Table 5c).The cortisol response of females paired with a larger, older partner and subsequently became subdominant increased from the first to third week after the social niche  (f) is represented as the proportion of total retreats in which the focal female retreats from her partner.Data is grouped by the dominance status after the social niche transition (x axis) as well as the dominance status in the initial social pair (color).Barplots represent the median for each group.Summary of statistical results is given above each individual figure within the plot.NS: P ≥ 0.1; (*): 0.5 ≤ P < 0.1; *: P < 0.5; **: P < 0.1; ***: P < 0.001.See Table 1 for detailed statistics.transition while the cortisol response of females paired with a smaller, younger partner and subsequently became dominant slightly decreased (Fig. 6).

Discussion
These results explore the behavioral and hormonal consequences of social niche transitions in adolescent female guinea pigs.After a social niche transition occurred, aggressive behavior was rapidly and completely reshaped.Focal females with smaller and younger partners were more aggressive than focal females with larger and older partners immediately after the social niche transition.Submissive behavior, on the other hand, was not entirely reshaped.The focal female's previous dominance status from the initial social pair modulated the outcome of dominance interactions with a new partner even three weeks after the social niche transition.Finally, when females underwent a social niche transition in which their new social partner was a larger, older female, the focal female was forced to become subdominant and the magnitude of cortisol responsiveness to a stressor increased.

Social niche specialization and conformance
When social pairs were formed in early adolescence, dominance status attained was not predicted by cortisol concentrations or modest differences in body weight or age.Furthermore, none of the social behaviors observed upon introduction predicted dominance status.Aside from interactions with siblings and parents in the natal group, these females were initially naïve in social contexts and gained social experience by interacting with their new social partner.When social behavior was again observed four, five, and six weeks after pair formation, behavioral differences between dominant and subdominant individuals had developed.Therefore, stable dominance relationships were observed later but were not established upon introduction.This suggests that developing individual differences in social behavior (Stamps and Groothuis, 2010;Sachser et al., 2020) and their effect on social interactions (Bergmüller and Taborsky, 2010;David et al., 2011;Montiglio et al., 2013;Briffa et al., 2015) contributed to the development of dominance relationships, in line with the social dynamics hypothesis (Chase et al., 2002).With regard to the social niche hypothesis, in this social context, the individualized social niches of females in early adolescence became specialized over time (Bergmüller and Taborsky, 2010;Montiglio et al., 2013;Trappes et al., 2022).
In contrast, females established a clear dominance relationship already on the first day after the social niche transition, with larger and older females becoming dominant over smaller and younger females.This is in line with the prior attributes hypothesis since females rapidly conformed to the appropriate social niche when there were pronounced differences in body size and age (Chase et al., 2002;Montiglio et al., 2013).Therefore, we conclude that different social mechanisms contributed to the development of dominance relationships depending on the social context.This agrees with findings from the literature: Studies assessing predictors of dominance status across multiple social groups have observed that the dominance relationships in some social groups are structured by intrinsic attributes such as age or body size while others are not, for example in reindeer (Rangifer tarandus; Holand et al., 2004), Phayre's leaf monkeys (Trachypithecus phayrei crepusculus; Lu et al., 2016) and rhesus macaques (Macaca mulatta; Kohn et al., 2016).

Shaping of the behavioral profile by dominance status
Regarding social behavior, there were two major findings.First, in both the initial social pairs and after the social niche transition, dominant females were consistently more aggressive than subdominant females.Furthermore, there was no lingering effect of dominance status from the initial social pair on aggressive behavior toward the new social

Table 2
Social behavior of the focal female three weeks after the social niche transition.Dominance status in initial social pair and after social niche transition were contrast coded, and reference levels are "subdominant".Estimates were obtained using the summary() function and ANOVA type III tables were generated using the Anova() function from the car package.Effect size was calculated as partial η2 using the eta_squared function in the effectsize package.Differences were considered significant at p < 0.05 (bold); 0.05 ≤ p < 0.1 was considered a trend (italic).partner after the social niche transition.In a similar study with female rhesus macaques (Macaca mulatta), adult social groups were manipulated and changes in social behavior were correlated to the change in dominance status (Kohn et al., 2016).Dominant females were more aggressive, and when dominance status changed, aggression was rapidly reshaped.Indeed, females across a wide range of species flexibly adjust  2 for detailed statistics.

Table 3
Statistical analysis of cortisol concentrations six weeks after the formation of the initial social pairs (CRT2).Dominance status (subdominant or dominant) was contrast coded, and subdominant is the reference category.Continuous fixed effects (body weight and cortisol concentration at CRT1) were mean centered for better interpretability of main effect estimates.Estimates were obtained using the summary() function from the lmerTest package and ANOVA type III tables were generated using the Anova() function from the car package.Effect size was calculated as partial η2 using the eta_squared function in the effectsize package.Differences were considered significant at p < 0.05 (bold).their aggressive behavior to match changing social conditions (Bowler et al., 2002;Pusey and Schroepfer-Walker, 2013;Stockley and Campbell, 2013;Taylor et al., 2014).
The second major finding is that the dominance status attained in early adolescence had a persistent effect on dominance interactions after the social niche transition, even when controlling for dominance status after the social niche transition.This could be due to winner-loser effects (Hsu et al., 2006;Lehner et al., 2011), whereby females who retreat from their partner are more likely to retreat in future encounters.Winnerloser effects on dominance status acquisition have been demonstrated for example with male shore crabs (Carcinus maenas).Males who had won many encounters during a social disturbance event proceeded to acquire an even higher dominance status when original social groups were reformed (Tanner et al., 2011).However, since dominance status in the initial social pairs was not manipulated but rather emerged organically, it cannot be ruled out that the propensity to behave submissively was an intrinsic behavioral trait.
In addition, dominance status predicted the frequency of several other social behaviors.As observed in weeks four through six in the initial social pairs, dominant females approached their partner more often.Contests are often won by the initiator (Jackson, 1991;Kar et al., 2016), and dominance acquisition has been predicted by behaviors such as proactivity (David et al., 2011) and exploration (Favati et al., 2014;Fox et al., 2009;Kaiser et al., 2019;but see Devost et al., 2016).Therefore, the female who most often initiated contact might have become dominant.After the social niche transition, previously subdominant females sniffed and followed their new social partner more often.There was also a tendency for females that changed dominance status to follow their new partner more often and sniff their partner more often three weeks after the social niche transition.A possible explanation is that subdominant females and females with an unstable dominance status are more motivated to thoroughly assess new social partners.It has been previously shown that subordinate male mice assess unfamiliar odor cues longer than dominant male mice do (Garbe and

Table 4
Cortisol concentrations on the first and third day after initial pair formation and social niche transition.Body weight was mean-centered for better interpretability of main effect estimates.Day and dominance status were contrast coded, and reference levels are "day 1" and "subdominant".Estimates were obtained using the summary () function and ANOVA type III tables were generated using the anova() function from the lmerTest package.Effect size was calculated as partial η2 using the eta_squared function in the effectsize package.Differences were considered significant at p < 0.05 (bold).4); baseline cortisol increased from day 1 to day 3 after the initial social pair formation and decreased from day 1 to day 3 after social niche transition.There was no significant effect of dominance status (see Table 4).
T.L. Rystrom et al. Kemble, 1994), and that male and female dogs of low status sniff urine of unfamiliar dogs longer than high status dogs do (Lisberg and Snowdon, 2009).

Social factors affected cortisol concentrations
There were three significant findings indicating that cortisol concentrations were affected by social factors.First, cortisol concentrations responded differently to the initial pair formation and the social niche transition.Upon the formation of the initial social pairs, the main challenge was likely the removal from the natal group.Our results align with previous findings that cortisol levels of female guinea pigs increase upon separation from the mother and remain heightened for days (Hennessy et al., 2002).Meanwhile, the main challenge after the social niche transition was likely the social instability induced by the change of social partner.It is well-established that cortisol levels are higher during periods of social instability (Sapolsky, 1982;Sachser et al., 1998;Nuñez et al., 2014), and cortisol concentrations rise during social confrontations with unfamiliar female guinea pigs (Glenk et al., 2018).Since dominance relationships were quickly settled after the social niche transition, the first few highly socially-active hours post-introduction were likely the most challenging, with cortisol levels decreasing

Table 5
Statistical analysis of cortisol concentrations after the social niche transition (data from CRT3 and CRT4).Body weight was mean-centered for better interpretability of main effect estimates.CRT, dominance status in initial social pair, and dominance status after social niche transition were contrast coded, and reference levels are "CRT3" and "subdominant".Estimates were obtained using the summary() function and type three anova tables were generated using the anova() function from the lmerTest package.Effect size was calculated as partial η2 using the eta_squared function in the effectsize package.Differences were considered significant at p < 0.05 (bold); 0.05 ≤ p < 0.1 was considered a trend (italic).Num DF for all predictors of all models = 1.Color indicates dominance status in the initial social pairs.Bars represent medians for each dominance status after social niche transition.There was a significant interaction effect between CRT and dominance status after social niche transition (see Table 5 for statistics); the cortisol response of individuals who were paired with a larger, older female in the social niche transition and subsequently became subdominant increased over time relative to that of individuals paired with a smaller, younger female in the social niche transition and subsequently became dominant. thereafter.
Second, the dominance status attained in the initial social pairs was correlated to baseline cortisol concentration.Six weeks after the initial social pair formation, dominant females had a significantly higher baseline cortisol concentration than subdominant females.Higher concentrations of cortisol are often correlated to higher levels of aggression (Haller, 2014), and in groups of size-matched individuals such as the female pairs early in adolescence, the individuals with higher circulating cortisol levels might be able to quickly mobilize the energy needed to express aggressive behavior.Since cortisol is a metabolic hormone, the higher baseline cortisol concentrations measured in dominant females might indicate that a higher metabolic rate is necessary to support energetically-demanding behavior such as aggression (Milewski et al., 2022).Furthermore, since dominance relationships were not immediately settled after the initial pair formation, instability of the dominance relationship, challenges by subdominant females, and the overall uncertainty of social interactions may have resulted in a stressful and more energetically-demanding social environment for dominant females (Bergman et al., 2005;Carvalho et al., 2018).Meanwhile, in social groups where dominance status is clearly established by body size or age such as after the social niche transition, social interactions are more predictable, and dominant individuals will not carry the stress of policing others (Creel et al., 2013).
Third, manipulating the social niche triggered a shift in cortisol responsiveness.Females who were partnered with a larger female and subsequently became subdominant after the social niche transition did not have a significantly higher cortisol responsiveness than females who were partnered with a smaller female and subsequently became dominant.Rather, cortisol responsiveness changed over time, and the direction of this change was modulated by the new social niche.In addition to manipulating the dominance status of the focal female, the social niche transition introduced a systematic difference in body size between her and her partner.Therefore, it is unclear through which mechanism the social niche transition triggers a shift in cortisol responsiveness.It is conceivable that living with a larger, older, dominant partner female shaped the activity and foraging patterns of the focal female (Gerrish and Alberts, 1995;Galef and Giraldeau, 2001), in which case a shift in energetic requirements and metabolism could modulate HPA axis activity (Haase et al., 2016;Jimeno et al., 2018;Milewski et al., 2022).Alternatively, the social stress from becoming subdominant to a larger female could have triggered the observed shift in cortisol responsiveness.Research on primate species indicates that differences in stressorinduced glucocorticoid concentrations emerge after dominance status is attained (Sapolsky, 2004;Cavigelli and Chaudhry, 2012;Michopoulos et al., 2012;Kohn et al., 2016).Notably, no correlation between dominance status and cortisol response to a psychological stressor has been detected in guinea pigs (Zipser et al., 2013;Mutwill et al., 2021;Rystrom et al., 2022).However, in these studies, dominance rank was observed in larger social groups rather than being manipulated in pairs as in the present study.Few other studies attempt to link stressor-induced glucocorticoid concentrations to dominance status in non-primate species (but see Pravosudov et al., 2003;Lindström et al., 2005;Poisbleau et al., 2005;Rubenstein, 2007).By systematically manipulating the social niche during adolescence, these results demonstrate that changes to the social niche causally affected the developmental trajectory of cortisol responsiveness.These results therefore support the hypothesis that the social niche plays an important role in shaping hormonal parameters.

Fig. 1 .
Fig.1.Experimental design.For the initial social pair formation, focal females were pair-housed with an adolescent female partner for six weeks.Then, a social niche transition occurred in which focal females were housed with either a larger and older or a smaller and younger female partner for three weeks.This social niche transition manipulated the dominance status of the focal female.Videos of social behavior in the home enclosures were recorded on d1 (3h), d3 (1h) and weekly (1h).Cortisol response tests (CRT1-4) occurred the day before the start of the initial social pair formation (CRT1), on the day before the social niche transition (CRT2), approximately one week after the social niche transition (CRT3), and again three weeks after the social niche transition (CRT4).Baseline cortisol concentration was additionally measured on d1 and d3 after both the initial social pair formation and social niche transition.

Fig. 2 .
Fig. 2. Behavior from weeks four, five, and six combined after initial social pair formation for focal females who became dominant (N = 25) and subdominant (N = 17) in their adolescent social pairs.Plotted are counts of behaviors for 3h of observation per individual with bars representing medians.Dominant and subdominant females significantly differed in the amount of aggressive behavior and approaches; dominant females were more often aggressive (Wilcoxon test, W = 317, P = 0.008) and approached their partners more often (Wilcoxon test, W = 316, P = 0.008).Dominant and subdominant females did not differ in the amount of play behavior (Wilcoxon test, W = 244.5,P = 0.419) or sniffing behavior (Wilcoxon test, W = 213.5,P = 0.989).

Fig. 3 .
Fig.3.Social behavior of the focal female observed in the first three hours after the social niche transition.Plotted are counts of behaviors (a-e); submissive behavior (f) is represented as the proportion of total retreats in which the focal female retreats from her partner.Data is grouped by the dominance status after the social niche transition (x axis) as well as the dominance status in the initial social pair (color).Barplots represent the median for each group.Summary of statistical results is given above each individual figure within the plot.NS: P ≥ 0.1; (*): 0.5 ≤ P < 0.1; *: P < 0.5; **: P < 0.1; ***: P < 0.001.See Table1for detailed statistics.

Fig. 4 .
Fig. 4. Social behavior of the focal female three weeks after the social niche transition from one hour of observation.Plotted are counts of behaviors (a-d); submissivebehavior (e) is represented as the proportion of total retreats in which the focal female retreats from her partner.Data is grouped by the dominance status after the social niche transition (x axis) as well as the dominance status in the initial social pair (color) Barplots represent the median for each group.Summary of statistical results is given above each individual figure within the plot.NS: P ≥ 0.1; *: P < 0.5; ***: P < 0.001.See Table2for detailed statistics.

Fig. 5 .
Fig. 5. Baseline cortisol concentration on the first and third day after (a) initial social pair formation and (b) social niche transition.Lines connect data points from each individual.Bars indicate medians, which are grouped by dominance status in the initial social pairs.Focal females who became dominant or subdominant after the social niche transition are plotted separately in (b).Baseline cortisol concentration significantly correlated with day (see Table4); baseline cortisol increased from day 1 to day 3 after the initial social pair formation and decreased from day 1 to day 3 after social niche transition.There was no significant effect of dominance status (see Table4).

Fig. 6 .
Fig.6.The change in cortisol responsiveness at two hours [ng/ml] from one (CRT3) to three (CRT4) weeks after the social niche transition.Individual data points are plotted, grouped by dominance status after social niche transition.Color indicates dominance status in the initial social pairs.Bars represent medians for each dominance status after social niche transition.There was a significant interaction effect between CRT and dominance status after social niche transition (see Table5for statistics); the cortisol response of individuals who were paired with a larger, older female in the social niche transition and subsequently became subdominant increased over time relative to that of individuals paired with a smaller, younger female in the social niche transition and subsequently became dominant.