Sex reversal induces size and performance differences among females of the African pygmy mouse, Mus minutoides

Bite force is a frequently used performance trait in functional morphological studies. Most of these studies have clearly demonstrated the link between osteology, muscular activity and biting performance at the intraspecific as well as interspecific levels (e.g. Herrel et al., 1999; Aguirre et al., 2002; Davis et al., 2010; Santana et al., 2010). Morphology has also been shown to be a driver of performance and fitness via intra-sexual competition (Herrel et al., 2010; Lailvaux et al., 2004). However, to our knowledge, the interactions between genetics, behaviour, morphology and performance at the intraspecific and even intra-sex levels have not yet been studied in mammals.

Here, we investigated bite force and skull morphology in the African pygmy mouse, Mus (Nannomys) minutoides Smith 1834. This species is known to have a unusual sex determination system, with three types of sex chromosomes: X, Y and a feminizing X*, producing XY males and XX, XX* females and a high proportion of sex-reversed, X*Y females (Veyrunes et al., 2010). The three types of females present no differences in external morphology or gonad anatomy (Rahmoun et al., 2014), but are known to vary in other traits: X*Y females have a higher reproductive success (Saunders et al., 2014), and have a much more aggressive behaviour (latency to attack, number of attacks; Saunders et al., 2016) than the XX and XX* females. Because biting is one of the characteristics of attacks that relates to aggressiveness, we compared the bite force of the X*Y females with that of other females and males to assess the effect of their genotypic changes on performance. In parallel, we performed measurements of morphological attributes of the skull and mandible in all genotypes. By doing so, we tried to link chromosomal differences to performance, through potential morphological and/or behavioural (aggressiveness) variations. Examining these interactions in M. minutoides could also help improve our understanding of the ecological and evolutionary implications of the emergence of this unusual sex determination system.

All mice were raised in the facilities of Montpellier University (France). The colony was established from wild-caught animals from South Africa (see details in Saunders et al., 2014). Overall, 54 mice were used in the experiments with 19 males and 35 females, all adults. Seventeen females were found to carry a Y chromosome using PCR amplification of the Sry gene (following Veyrunes et al., 2010). The 18 others did not have the Sry gene, and thus were either XX or XX*. XX and XX* females do not differ in terms of behaviour and reproductive success (Saunders et al., 2014, 2016) and were pooled in this study after checking that they did not represent distinct phenotypic groups. Recognizing all XX* females requires karyotyping, which is technically demanding; instead, we were able to identify eight of them in our dataset thanks to their genealogy. These eight individuals were compared with the other non-X*Y females, and their morphology and bite force were roughly equivalent in mean and variance (results not shown). These XX and XX* females are hereby referred to as ‘XX females’. The experimental protocol was performed in accordance with European guidelines and with the approval of the Ethical Committee on Animal Care and Use of France (no. CEEA-LR-12170).

Bite force was measured using a piezoelectric force transducer (Herrel et al., 1999; Aguirre et al., 2002). Measurements were performed prior to PCR amplification, avoiding user-behaviour bias, and by a single user, to avoid inter-user error. Each mouse bit three times in a row and the highest bite force was selected for analysis. Group means were compared using Tukey's HSD test. The mice were then euthanized by CO₂ inhalation, and kept at −20°C until further treatment.

All skulls were cleaned manually and kept in the collections of Montpellier University. Eleven landmarks were digitized on the dorsal side of the skull, and 16 on the mandible using the software tpsDig2.0 (Fig. 1A,B). Landmark data were imported and analyzed in R (R Foundation for Statistical Computing, Vienna, Austria), using the procrustes superimposition method, with functions from Claude (2008). Missing landmarks (owing to the fragility of the mandible), if any, were estimated using a tps interpolation. Centroid sizes of the skulls and mandibles were computed and used as the size variables (Bookstein, 1991). Principal component analyses (PCAs) were run on superimposed coordinates to check for shape differences between males, XX females and X*Y females. In all analyses, the PC axes retained represented 90% of shape variation. A type II ANOVA was run to test the effects of sex chromosomal group, age, size and their interactions on bite force. For shape, several models were fitted with a different sum of squares partitioning (Table 1), and MANOVAs were used to assess the influence of sex chromosomal groups on the principal components of shape variation.

X*Y females bit significantly harder than other females (Fig. 2; data are available on request from the corresponding author). They also bit harder than males, but not significantly so; bite force was similar between males and XX females. Table 2 shows the results of the Tukey's HSD test.

No major morphological differences appeared between the three groups,which overlapped in morphospace (Fig. 1C,D, Table 1).

X*Y females displayed a significantly greater centroid size than males and XX females (Table 2), even independently of age (Table 1). MANOVAs did not reveal any effect of the group on shape; however, they showed significant allometry shared by all groups (Table 1).

There was a significant size–bite force correlation (r≈0.6, d.f.=52, P<0.01 for both bones) when considering the data as a whole (i.e. not taking the groups into account). A relationship was also found between age and bite force (r=0.3, d.f.=49, P<0.05). In the linear model created, the type II ANOVA revealed that, once age and size were taken into account, bite force differences were no longer significant between the groups (F=1.28, d.f.=2, den. d.f.=42, P>0.2), although the tendency for X*Y females to have higher bite force than XX remained. In this model, age had no significant effect, and size remained the main explanatory variable.

We showed that in the African pygmy mouse, X*Y females bite harder than the XX females and, more surprisingly, also bite harder than males (although not significantly so). Limited shape variation between the sex chromosomal groups (Fig. 1C,D) shows that skull and mandible shape are not at the root of the higher bite force of X*Y females. It also suggests that the presence of the Y and/or X* chromosomes does not significantly influence shape. However, without considering the groups, we found a positive relationship between size and bite force, as shown in previous studies (Herrel et al., 2001; Lailvaux et al., 2004; Freeman and Lemen, 2008). Furthermore, X*Y females displayed a greater skull size, which appears to explain part of their increased bite force (Tables 1, 2). Performance changes independent of size but owing to behavioural differences between the groups are subtle and not significant, suggesting that aggressiveness has a much lower impact on performance variation than morphological determininants (here size).

Hormones are also known to influence performance, as demonstrated by hormonal and molecular screening of elite female athletes, which detected sex development disorders with hyperandrogenism (e.g. XY females) in an occurrence 200 times greater than in the general population (e.g. Fénichel et al., 2013). In our model, however, the M. minutoides X*Y females harbour typical ovarian anatomy (no ovotestes) and a normal anogenital distance, suggesting a low level of circulating androgen (Rahmoun et al., 2014), observations that have been recently confirmed by preliminary hormonal assays on testosterone (F.V., unpublished data). Therefore, the increased skull size of X*Y females appears to be the main cause of their higher bite force.

Although not much is known about the ecology of M. minutoides (Britton-Davidian et al., 2012), the advantages of having higher performance may be twofold. Higher bite force may increase the range of available food resources (Aguirre et al., 2003) by allowing the X*Y individuals to feed on harder food. Second, higher aggressiveness coupled with stronger bites may have a role in intra-sex competition: X*Y females may increase their reproductive success by taking dominance over other females via attacks (Lailvaux and Irschick, 2006). This has been shown for lizards, in which dominance, territorial defense and the outcome of fights are all linked to performance (Herrel et al., 2010; Lailvaux et al., 2004).

Chromosomal changes in the X*Y females may thus give them a dominant position, consistent with the hypothesis that an increase in fitness explains the maintenance of these females, despite the costly loss of the unviable YY embryos (Saunders et al., 2014). Considering our results and those of Saunders et al. (2014, 2016), it appears that the X*Y females obtain selective advantages through many aspects of their fitness: reproductive success, behaviour and whole-organism performance. The effect of the X* chromosome thus goes well beyond feminization, and is at the origin of a complex and multi-factorial X* syndrome that extends to life history traits, behaviour (Saunders et al., 2014, 2016), size variation and performance (present study). Interestingly, these X*Y females were characterized as ‘super females’ because they were shown to have a better reproductive output than the XX females (Saunders et al., 2014), but in addition, they present hyper-masculinized traits with an enhanced aggressiveness, larger skulls and stronger bite forces than males. It is notoriously hard to assign performance modifications to naturally occurring genetic changes, but this model offers a fantastic framework with which to further connect performance, morphology, behaviour and sex chromosomes.

The authors thank the CECEMA animal facility. We also thank Paul A. Saunders for his comments on the manuscript. We are grateful to A. Herrel and one anonymous reviewer for their insightful comments. This publication is a contribution of the Institut des Sciences de l'Evolution de Montpellier (UMR 5554 – UM2+CNRS+IRD+EPHE) No. ISEM 2017-055.