All procedures were conducted in accordance with the requirements of the United Kingdom Animals (Scientific Procedures) Act 1986, under the remit of Home Office license number 30/2673. These procedures were also approved by the appropriate ethics committee at Cardiff University.

The experimental mouse line, Cdkn1c^BACx1 possesses one copy of a BAC that spans the Cdkn1c gene and two other genes, Phlda2 and Slc22a18. Cdkn1c^BACx1 were compared to their WT cage-mates in all instances. A separate reporter line Cdkn1c^BACLacZ possesses a modified version of this BAC with a β-galactosidase reporter construct inserted into the Cdkn1c locus, disrupting Cdkn1c expression (John et al., 2001; Andrews et al., 2007). Cdkn1c^BACLacZ were compared to their WT cage-mates in all instances.

Male mice were group housed from weaning at 3–4 weeks, with between three and five animals per cage. Each cage consisted of transgenic animals (Cdkn1c^BACx1 or Cdkn1c^BACLacZ) and WT litter-mates. All were housed in a 12:12 h light:dark cycle with food and water provided ad libitum except during the “competition for water access test” where all animals had restricted access to water, provided for 2 h per day, immediately following behavioral testing.

Coat condition and general appearance was monitored regularly from weaning. Occurrences of injury due to bullying/fighting (fresh wounds on the flanks or in the ano-genital region) were recorded in both the behavioral cohort and the stock cohort, to maximize observational size.

Animals were between 8 and 12 weeks at beginning of testing. For all experiments N = 48 animals were used in total: Cdkn1c^BACx1 (n = 14) and their WT cage-mates (n = 12); Cdkn1c^BACLacZ (n = 15) and their WT cage-mates (n = 8). Average number of animals per cage: Cdkn1c^BACx1 cohort = 3.9 SEM = 0.3 (7 cages); Cdkn1c^BACx1 cohort = 3.8 SEM = 0.4 (6 cages). Test order was as follows; within-cage tube test, scent marking task and water access task. Following these experiments animal’s rank stability with and without a bedding change was assessed. For analysis of within-cage measures of dominance, bedding was unchanged for the duration of each task but was changed between each task to allow for a more representative indication of rank stability over time.

Two separate tests of olfactory function were used, one using social odors and another using non-social odors.

All statistical analysis was carried out using SPSS 20.0 (SPSS, United States). For analysis of genotype on rank within a cage group, each animal’s rank within the cage was transformed to a number between 0 (least dominant animal in the group) and 1 (most dominant animal in the group), this was performed for each group for each task to allow for differences in cage group size. A non-parametric Mann–Whitney U-test was carried out with GENOTYPE as the grouping variable.

Rank stability across a change in environment (cage-bedding change) or when the environment remained stable was rated as ‘0’ for no change and ‘1’ for change in rank in a tube test. Effect of GENOTYPE was assessed using a conditional logistic regression, regressing on CAGE ID, to take into account cage group sizes and account for extreme cages. Scent marking behavior was analyzed using a linear mixed-models ANOVA, with CAGE ID as the random factor; pairwise fixed effects were then assessed by Bonferroni. Correlation between ranks within groups across difference tasks was determined using a non-parametric Spearman’s rank-order correlation and statistical difference between correlation coefficients was assessed using Fisher r-to-z transformation. For analysis of effect of genotype on likelihood to be involved in severe fighting, a chi-square cross-tabs test was carried out with GENOTYPE as rows and BITE PRESENCE as column. Olfactory function analysis was carried out using a repeated measures ANOVA with GENOTYPE as the between subject measure and ODOR PRESENCE and the within subject measure.

Group housed male mice establish a linear, transitive (Williamson et al., 2016), social hierarchy with a single dominant individual and a number of sequential (beta, gamma, delta) subordinates (Ebbesen et al., 1992; Avitsur et al., 2007; Wang et al., 2011). We carried out three tasks assessing social dominance (tube test, urine marking, and competitive access to a resource) to determine an animal’s rank within its cage group. Strikingly, given our previous findings (McNamara et al., 2017), Cdkn1c^BACx1 animals did not occupy significantly more dominant ranks than their WT cage-mates on any individual measure of the within cage social hierarchy. Specifically, there was no difference in social rank as determined by the tube test (Figure 1A; Mann–Whitney, U = 51, p = 0.59), scent marking (Figure 1B; U = 42.5, p = 0.27) or competitive water access (Figure 1C; 120 s: U = 49.5, p = 0.51; data not shown, 600 s: U = 43.5, p = 0.29).

The Cdkn1c^BACLacZ control line, where Cdkn1c was not overexpressed, was also subjected to the same battery of social dominance tests and here too there was no difference in social rank between these males and their WT cage-mates (Scent marking: U = 48, p = 0.47; Water access 120 s: U = 54, p = 0.73; Water access 600 s: U = 50, p = 0.55; Supplementary Figure S1). These results are not caused by an inability to perform the tasks, as a clear transitive hierarchy was apparent in each measure of social dominance. For instance, a linear, transitive, hierarchy was apparent for an average of 3.4/4 days groups of Cdkn1c^BACLacZ animals and their cage mates, and an average 3.3/4 days in groups of Cdkn1c^BACx1 animals and their cage mates. Similarly, a clear hierarchy was apparent in 100% of cages containing Cdkn1c^BACLacZ animals and in 85.7% of cages containing Cdkn1c^BACx1 animals in the competition for water access task. A clear hierarchy was also apparent in 71.4% cage of cages containing Cdkn1c^BACLacZ animals and in 57.1% of cages containing Cdkn1c^BACx1 animals in the scent marking task.

In a stable social hierarchy an individual’s rank in these separate measures is expected to correlate (Wang et al., 2011). However, in groups of Cdkn1c^BACx1 and their WT cage-mates the social hierarchy was not stable, as an individual’s rank in one measure of dominance did not correlate with its rank in another (Figure 2A, tube test vs. water access in the first 120 s rank, Spearman’s ρ correlation = -0.034, p = 0.88; and Figure 2B, scent marking vs. water access 600 s rank, Spearman’s ρ correlation = -0.134, p = 0.51). In contrast, groups containing the control line Cdkn1c^BACLacZ animals and their WT cage-mates showed the expected pattern, as an individual’s rank in one measure of dominance was significantly correlated with its rank in another measure. Specifically, tube test vs. water access in the first 120 s rank (Figure 2C; Spearman’s ρ correlation = 0.521, p = 0.01) and scent marking vs. water access 600 s rank (Figure 2D; Spearman’s ρ correlation = 0.665, p = 0.001). Fisher r-to-z transformations confirmed these group differences, as the correlation coefficients seen in groups of Cdkn1c^BACx1 and their WT cage-mates were significantly different from the correlation coefficients between in groups of Cdkn1c^BACLacZ animals and their WT cage-mates (Tube test vs. water access 120 s, z = 1.93, p = 0.05; Scent marking vs. water access 600 s, z = 3.06, p = 0.002).

We hypothesized that the loss of stability between different measures of social dominance may be as a consequence of a greater propensity of Cdkn1c^BACx1 animals to challenge the established hierarchy. Therefore, we would expect these animals to have a more variable rank in the cage hierarchy across time. Dominancy relationships, while generally stable, can change under pressurizing circumstances (Cohn et al., 2012). One such circumstance is when the odor cues are removed (e.g., following a bedding change), after which the hierarchy must be re-established (Gray and Hurst, 1995; Van Loo et al., 2000). However, if the social structure is generally stable, the change of bedding should not perturb an animal’s rank any greater than in an unchanging environment. We examined this using consecutive tube tests, comparing an animal’s rank before and after odor cues were removed (a cage-bedding change). When odor cues remained constant, there was no greater change in the rank of Cdkn1c^BACx1 animals or their WT cage-mates across repeated testing (Wald statistic = 0.021, p = 0.886). However, when odor cues were removed following a cage-bedding change, Cdkn1c^BACx1 animals had a significantly more variable rank compared to their WT cage-mates (Wald statistic = 3.925, p = 0.048) (Figure 3A and Supplementary Table S1).

In the control group, both Cdkn1c^BACLacZ and their WT cage-mates also displayed no difference in change in rank when the environment remained stable (Wald statistic = 0.010, p = 0.922; Figure 3B and Supplementary Table S2). However, in contrast to the Cdkn1c^BACx1 animals, the Cdkn1c^BACLacZ mice also showed no difference in change in rank following a bedding change (Wald statistic = 0.665, p = 0.415). This indicates that in the absence of odor communicants indicating the dominant animal Cdkn1c^BACx1 animals are more likely to change rank.

The maintenance of social status within a specific territory is an important aspect of social behavior. Both wild-caught and laboratory mice establish territories (Anderson and Hill, 1965; Hurst, 1987; Fuxjager et al., 2010). Territorial ownership is communicated via ‘scent marks’ composing of major urinary proteins (Hurst et al., 1993, 2001). This scent marking behavior positively influences male reproductive success (Thonhauser et al., 2013), demonstrating a clear advantage upon which selective pressures may act. Scent marking communicates a dominant animal’s territory and scent-marking activity is altered in the presence of a dominant resident (Drickamer, 2001). An general assessment of the scent marking test indicated approximately 30% increase in the level of scent marking by Cdkn1c^BACx1 animals and their WT cage-mates in comparison to Cdkn1c^BACLacZ animals and their WT cage-mates, although this failed to reach significance [t(54) = -1.85, p = 0.07] (Figure 4A). A more detailed examination of these data revealed that WT animals increased scent marking toward (i.e., when paired with) a Cdkn1c^BACx1 male, compared to a control Cdkn1c^BACLacZ male [t(19) = 1.96, p = 0.074] (Figure 4B and Supplementary Figure S3) or toward another WT cage-mate [t(13) = 0.006, p = 0.94] (Figure 4C and Supplementary Figure S3). This indicates that the presence of Cdkn1c^BACx1 animals may elicit a differential territorial behavior response in WT cage-mates.

Importantly, the effects on social stability are unrelated to general olfactory function per se, as there was no difference between Cdkn1c^BACx1 and WT animals in time spent exploring a social odor (male urine) [t(24) = -0.29, p = 0.79, no odor present; t(24) = -1.41, p = 0.18, odor present] (Figure 5A). Similarly there was no difference between Cdkn1c^BACx1 and WT animals in ability to detect a non-social odor (food) [latency to sniff: t(22) = -0.783, p = 0.44] (Figure 5B).

A stable social hierarchy normally benefits group-housed animals (Ebbesen et al., 1992; Moles et al., 2006; Sa-Rocha et al., 2006; Cordero and Sandi, 2007) and the more frequent rank fluctuation of Cdkn1c^BACx1 animals appeared to have a consequence for fitness. We observed significantly more signs of severe in-cage fighting (fresh cuts along flanks or in ano-genital region observed on at least one occasion) in cages containing animals over-expressing Cdkn1c and their WT cage-mates (42% of animals, n = 50) compared to cages of Cdkn1c^BAClacZ animals and their WT cage-mates (no observed occurrences, n = 56) and cages containing only WT litter-mates of Cdkn1c^BACx1 animals (no observed occurrences, n = 7). Within groups of Cdkn1c^BACx1 and their WT cage-mates signs of severe fighting were not different by genotype (Fischer’s exact test, p = 0.26; Supplementary Figure S2), indicating that presence of a Cdkn1c^BACx1 in a group had negative effects on physical fitness for all animals within the social group.