Sperm competition and the evolution of the sperm hook in house mice


Renée C. Firman, Centre for Evolutionary Biology, School of Animal Biology (M092), University of Western Australia, 35 Stirling Hwy, Crawley, Western Australia 6009, Australia.
Tel.: +61 8 6488 2699; fax: +61 8 6488 1029; e-mail: rcfirman@cyllene.uwa.edu.au


Sperm morphology varies considerably both between and within species. The sperm of many muroid rodents bear an apical hook at the proximal end of the head. The curvature of the sperm hook varies greatly across species, however the adaptive significance of the sperm hook is currently not known. In wood mice the apical hooks intertwine to form sperm ‘trains’, which exhibit faster swimming velocities than single cells. Thus, it has been suggested that if sperm ‘trains’ were advantageous in a competitive situation, then the apical sperm hook might be an evolutionary product of selection via sperm competition. A comparative study of rodent species provided support for the hypothesis, and showed that species with higher levels of sperm competition had more reflected sperm hooks. Here, we tested this hypothesis at the intraspecific level. We quantified sperm hook morphology from seven house mouse populations, and found that interpopulation variation in hook curvature was not explained by variation in sperm competition risk. Furthermore, observations of ejaculated sperm revealed that sperm groups are not a common characteristic of mouse ejaculates. We suggest that selection for sperm attachment to the oviduct epithelium, and thus better retainment of sperm fertilizing potential, may provide a more general explanation of the evolutionary relationship between sperm competition risk and the curvature of the sperm hook among rodents, and provide a phylogenetic comparison among rodent species that supports our hypothesis.


Sperm size and shape is highly diversified across animal taxa, ranging from tiny, amoeboid-shaped cells to long, flagellate cells (Jamieson, 1999; Jamieson et al., 1999). The adaptive significance of interspecific variation in sperm morphology has been examined by theoretical models (Ball & Parker, 1996), and comparative studies of different taxonomic groups (Gomendio & Roldan, 1991; Fitzpatrick et al., 2009). Among species, different mating systems may generate selective forces that affect sperm morphology. For example, when females mate multiply (polyandry, polygamy) ejaculates from different males will often overlap in the female tract and compete for fertilizations (Parker, 1970). In this context, postcopulatory sexual selection is expected to favour ejaculate traits that provide a competitive fertilisation advantage. Thus, sperm competition is recognised as a pervasive force in the evolution of testes size and sperm production (Harcourt et al., 1981; Møller, 1988; Gage, 1994; Hosken, 1997). Consequently, testes size (relative to body size) has routinely been used as an index of sperm competition risk in comparative studies exploring the relationship between sperm competition and sperm traits, including sperm morphology (Gomendio & Roldan, 1991; Briskie et al., 1997; LaMunyon & Ward, 1999; Balshine et al., 2001; Byrne et al., 2003; Immler et al., 2007).

For species with motile sperm, the general prediction is that sperm competition will favour increased sperm length, as longer sperm are assumed to be faster and thus more competitive. However, comparative studies of vertebrate taxa present conflicting evidence, and have revealed both positive (Gomendio & Roldan, 1991; Briskie et al., 1997; Balshine et al., 2001; Byrne et al., 2003) and negative (Stockley et al., 1997) associations between sperm competition risk and sperm size. Among mammals, two comparative studies found no relationship between sperm competition and sperm length (Harcourt, 1991; Gage & Freckleton, 2003). This issue is further confounded by limited evidence that sperm size is associated with competitive fertilisation success (Gage et al., 2002; Malo et al., 2005). Indeed, contrary to an interspecific comparison of mammal species (Gomendio & Roldan, 1991), we found that male house mice with relatively shorter sperm have greater in vivo competitive fertilisation success compared to males with relatively longer sperm (Firman & Simmons, 2008a).

The sperm of many muroid rodent species have a complex morphology consisting of a falciform head bearing ≥1 apical hooks (Breed, 2004). To date, few investigations have explored the adaptive significance of the muroid sperm hook (Roldan et al., 1992). It has been suggested that the hooks may facilitate sperm transport within the female tract (Smith & Yanagimachi, 1990), or penetration of the ova (Flaherty et al., 1983). Alternately, the evolution of this sperm morphology may be an adaptation to sperm competition (Moore et al., 2002; Immler et al., 2007). In the wood mouse sperm hooks intertwine to form aggregated groups of 50–100 cells (Moore et al., 2002). These sperm ‘trains’ exhibit faster swimming velocities than single cells, suggesting that cooperative behaviour may be beneficial in sperm competition (Moore et al., 2002). In support of this hypothesis, a comparative study of muroid rodents revealed that species with relatively larger testes, indicative of stronger selection from sperm competition, have more curved hooks (Immler et al., 2007). House mice (Mus musculus and Mus domesticus) bear an apical sperm hook, which has been suggested to facilitate the formation of motile sperm groups (Immler et al., 2007). Here, we used an interpopulation approach to test the hypothesis that the sperm hook is an adaptation to sperm competition in house mice. In the quest for understanding the relationship between sperm competition and the evolution of sperm traits, female reproductive characteristics are often overlooked. We provide an alternative explanation for the evolution of the muroid sperm hook, and include a comparative analysis to investigate the evolutionary relationship between oestrus duration and sperm hookedness among rodents.

Materials and Methods

Sperm competition risk among populations of house mice

Previously, we sampled seven island populations of house mice (Mus domesticus) located along the coast of Western Australia. We corrected for interpopulation differences in the power to detect multiple paternity (Neff & Pitcher, 2002), and generated expected levels of multiple paternity for each population: Carnac: 0.526; Whitlock: 0.166; Boullanger: 0.313; Rat: 0.711; Beacon: 0.112; Dirk Hartog: 0.477; Thevenard: 0.796 (Firman & Simmons, 2008b). For most of our populations we have just a single measure of multiple paternity (Firman & Simmons, 2008b). Thus, we acknowledge that our data only present a snapshot of the long evolutionary histories of these populations. Nevertheless, we found that variation in sperm competition risk among populations predicted variation in testes size (Firman & Simmons, 2008b).

Sperm head morphology

We examined the morphology of the sperm head from samples of sperm from the seven island populations that varied in sperm competition risk (males/population = 25; cells/male = 10). Full details of the capture, handling, and processing of males are provided elsewhere (Firman & Simmons, 2008b). We performed an in vitro analysis of mouse sperm motility to explore the relationship between sperm head morphology and sperm swimming speed, and to obtain sperm from which to take images. Briefly, the caudal epididymes of sexually mature males were repeatedly cut with fine scissors, and placed in 1 mL of culture medium for sperm capacitation (Murase & Roldan, 1996). Following a 90-min incubation (37 °C), sperm swimming velocity (average path velocity, VAP) was quantified using the CEROS computer-assisted sperm analysis system (v10; Hamilton and Thorne Research, Beverley, MA, USA) (Firman & Simmons, 2008b). We were unable to obtain measures of sperm velocity for the Thevenard Island population. During these analyses, we observed two motile sperm ‘groups’, which were formed by cells collectively adhering to epididymal tissue.

A 50-μL aliquot of the sperm suspension was fixed in a 4% formaldehyde solution. Sperm smears were prepared on slides and stained with Coomassie brilliant blue (Firman & Simmons, 2008b). Images of stained sperm were captured and formatted so that the sperm hook was oriented to the left-hand side of the image (Fig. 1). We used geometric morphometric analysis (GMA) to quantify variation in mouse sperm head size and shape (Zelditch et al., 2004). We placed 13 landmarks around the periphery of each sperm head using tpsDig 2w32 (http://life.bio.sunysb.edu/morph; F. James Rohlf, Department of Ecology and Evolution, Stony Brook University, Stony Brook, NY, USA) (Fig. 1). The points that clearly marked the junction where the sperm midpiece meets the sperm head (1, 13), the terminal end of the sperm hook (4), and the terminal end of the bulbous region of the sperm head (12) were assigned as fixed landmarks (Fig. 1). The remaining landmarks were assigned as semi-sliders. We generated relative warp scores and centroid scores using the software package tpsrelww32. Partial warp scores were generated that described deviations from the consensus shape (Fig. 2), and thus variation in sperm head morphometry. Partial warp scores were then subject to relative warp analysis (corresponding to a principal component analysis) that reduced the multivariate shape to a few variables that allow differences in shape among individuals to be examined (Zelditch et al., 2004). Centroid size provides a measure of the size of the structure independent of shape and is calculated as the square root of the summed square distances between each landmark and the centroid of the structure being measured (Zelditch et al., 2004).

Figure 1.

 Schematic diagram of a M. domesticus sperm head with 13 landmarks placed around its periphery for geometric morphometric analysis. Landmarks 1, 4, 12 and 13 were assigned as fixed, and the remaining landmarks were assigned as semi-sliding.

Figure 2.

 The consensus shape (centre) from geometric morphometric analysis is shown with partial warps as vectors with their origin as the consensus position of each landmark. Thin-plates splines show variation in shape along the first relative warp, with extreme negative scores shown to the left, and extreme positive scores shown to the right.

Observations of ejaculated sperm

We adopted the same methodology as Moore et al. (2002) to investigate the behaviour of ejaculated house mouse sperm. We established three experimental treatments: (i) male and female mating pairs were observed continually, and the female was euthanized immediately following the deposition of a mating plug (0–10 min treatment, n = 4); or male and females were paired, and the female was inspected half hourly for the presence of a mating plug, and the females were euthanized; (ii) 0–1 h (n = 4), and (iii) 1–2 h (n = 4) postcoitus. In all treatments the ejaculate was gently ‘milked’ from the reproductive tract, which was then flushed with phosphate buffered saline. We prepared two separate 20 μL aliquots of each ejaculate to assess sperm behaviour. Using a Leica DME light microscope (×100 magnification) we observed each preparation for approximately 5 min, and recorded short clips using a Panasonic NV-GS250 camera attached to the Leica microscope (the sperm clips are available online in the supplementary material at http://www.eseb.org/JEB-2009-00207%20movies/).

Oestrus duration and sperm hookedness

We conducted a comparative analysis to investigate the evolutionary relationship between oestrus duration and sperm hookedness among rodents. We used sperm hook angle data from Immler et al. (2007), and obtained oestrus duration data from published sources (Table S3). There are limited oestrus data available from the published literature. Thus, our sample size was restricted to 11 at the genus level (Table S3). Mean species testes mass and body mass data were obtained from references within Immler et al. (2007), and were included in the model. Utilizing an established phylogeny (Immler et al., 2007), we adopted a generalised least squares approach and tested for correlated evolution between the continuous traits oestrus duration/testes mass/body mass, and sperm hook angle (Pagel, 1997, Freckleton et al., 2002). We set the branch length to one, and calculated lambda (λ) as a measure of phylogenetic correlation (Pagel, 1999).

Results and discussion

It has been suggested that the rodent sperm hook has evolved to facilitate the formation of sperm trains or groups that enhance a male’s competitive ability (Moore et al., 2002). Support for this hypothesis comes from a recent study of muroid rodents, which revealed that species with relatively larger testes, a proxy for the strength of selection from sperm competition, have longer, more reflected sperm hooks (Immler et al., 2007). Comparative studies provide insights into historic evolutionary pathways, and may predict selective forces that are operating at the intraspecific level. However, frequently intraspecific patterns of sperm competition and sperm morphology do not reflect interspecific patterns. Additionally, sperm traits evolve rapidly within species. Thus, there is potential for interpopulation variation in sperm morphology to arise quickly (Miller & Pitnick, 2002). Previously, we found significant variation in the risk of sperm competition among seven island populations of house mice (Firman & Simmons, 2008b). Here, we utilized a sophisticated morphometric analysis technique to quantify sperm head shape, and explored the evolutionary relationship between sperm competition and sperm morphology at the intraspecific level.

Our geometric morphometric analysis returned 22 relative warps that explained among-population variation in sperm head morphology. The first relative warp (RW1) explained 34% of the variance in sperm head shape, and described a narrowing of the head and a more pronounced hook (Fig. 2). The second (20%), third (14%) and fourth (11%) relative warps explained lesser amounts of the variance (thin-plate splines of the extremes of RW2, RW3 and RW4 are provided in the online appendix). Preliminary anova revealed that there was significantly more variation between individuals than among individuals in centroid size (F174, 1575 = 11.222, P < 0.001), RW1 (F174, 1575 = 8.032, P < 0.001), RW2 (F174, 1575 = 2.921, P < 0.001), RW3 (F174, 1575 = 3.437, P < 0.001), and RW4 (F174, 1575 = 1.420, P < 0.001). Additionally, there was significant among-population variation in sperm head size (centroid scores: F6, 168 = 38.32, P < 0.001), and the relative warp scores describing sperm head shape (RW1: F6, 168 = 69.16, P < 0.001; RW2: F6, 168 = 8.96, P < 0.001; RW3: F6, 168 =17.46, P < 0.001; RW4: F6, 168 = 4.23, P < 0.001).

We used measures of the frequency of multiple paternity and scores on RW1 to assess the relationship between sperm competition risk and sperm hook morphology among populations of house mice. Sperm competition risk did not explain a significant proportion of the among population variation in sperm head size (F1, 5 = 0.074, P = 0.797), or sperm hook curvature (RW1: F1, 5 = 1.647, P = 0.256). Additionally, among populations sperm head size (F1, 4 = 0.120, P = 0.747) and hook curvature (RW1: F1, 4 < 0.001, P = 0.991) were not related to sperm swimming velocity (VAP). Because the unit of analysis in our study is individual islands, our statistical power to detect a relationship between sperm morphometry and risk of sperm competition is low. We note that we had enough power to detect covariation between sperm competition risk and testes size (Firman & Simmons, 2008b), so that any effect of sperm competition on sperm head morphometry must be considerably smaller than that on sperm productivity. We calculated the effect sizes (Pearson’s r) and their 95% confidence intervals (CI) to better appreciate the potential magnitude of the effect size of sperm competition risk on RW1 (r = 0.50 [–0.41, 0.91]), and RW1 on VAP (r = 0.16 [–0.68, 0.81]) (Nakagawa & Cuthill, 2007). The 95% CI on the effect size are broad, but in both cases asymmetrical about zero, suggesting caution in accepting the null hypothesis of no effect. Sperm competition risk did not explain the variation in the second, third or fourth relative warps (analyses provided in Table S1 of the online appendix). Furthermore, using testes size as a proxy for variation in sperm competition risk among island populations yielded quantitatively similar results (analyses provided in Table S2 of the online appendix).

We retrieved ejaculates from mated females to determine whether the house mouse sperm hook facilitates the formation of sperm groups. We observed active aggregations of sperm that could be described as motile groups in two of the ejaculates that were retrieved immediately following copulation (clips A and C, the sperm groups appear approximately two minutes into each clip). It was difficult to elucidate exactly how the groups were formed, but appeared to be via 30 or more sperm adhering to each other along the head-midpiece region of the cell (as also described by Immler et al., 2007). We did not observe motile sperm groups in the ejaculates that were retrieved 0–1 or 1–2 h postcoitus. Thus, while we found some evidence for the formation of motile sperm groups, we can conclude that they are not a general characteristic of mouse ejaculates. Additionally, we found no evidence that the apical sperm hook functions specifically to facilitate the formation of sperm groups. Given that aggregated sperm undergo the acrosome reaction and lose their fertilising potential (Moore et al., 2002), the adaptive significance of sperm groups is puzzling. It may be that sperm groups in house mice are merely an incidental by-product of sperm adhering together in such a way that they attain progressive motility. Indeed, while the sperm hook may facilitate the formation of sperm trains in wood mice (Moore et al., 2002), there is currently no evidence to suggest that sperm trains are in fact advantageous in sperm competition.

An alternative hypothesis has been proposed for the evolution of the apical rodent sperm hook. It has been suggested that the hook may facilitate the passage of sperm along the female reproductive tract by enabling them to ‘dig in’ the points of the hook (Smith & Yanagimachi, 1990). A study of mouse sperm in situ in the oviduct revealed that sperm with beating flagella adhere to the epithelium of the lower isthmus via the hooked region of the head (Suarez, 1987). Mammalian sperm depend on limited energy supplies to travel to the site of fertilization and penetrate the ova (Suarez et al., 1991; Strauss et al., 1995; Ho & Suarez, 2001). In mice the window between copulation and ovulation may be as long as eight hours so that sperm have to retain their position in the reproductive tract prior to ovulation (Rugh, 1968). Thus, sperm may conserve their energy reserves and fertilizing ability prior to ovulation by attaching to the oviduct epithelium. Therefore, if more reflected sperm hooks results in better attachment to the female tract, and thus better energy conservation, then we might expect to observe an evolutionary relationship between the interval between copulation and ovulation and sperm hookedness.

Although oestrus usually extends beyond ovulation, one would expect the interval between copulation and ovulation to be greater in species with longer oestrus periods. Indeed, sperm lifespan correlates positively with oestrus duration among mammals (Parker, 1984; Gomendio & Roldan, 1993). We performed a species comparison to test whether oestrus duration predicts sperm hookedness among muroid rodents. After controlling for phylogeny, our analysis revealed that, together with testes mass (slope b = 0.07, t = 2.71, P = 0.030) and body mass (slope b = –0.09, t = 3.10, P = 0.017), oestrus duration (slope b = 0.04, t = 2.40, P = 0.047) correlated positively with sperm hookedness (λ = 6.86 × 10–5, n = 11). Thus, we provide evidence that supports our prediction that the sperm hook may function to facilitate the attachment of the sperm to the oviduct epithelium, which in turn may facilitate the retainment of sperm fertilising potential. Females with longer oestrus periods would be expected to have higher mating frequencies, and thus generate higher levels of sperm competition; this may explain the proximal relationship between relative testes size and sperm hookedness. We investigated whether oestrus duration predicted relative testes size within our data set. The analysis (λ = 6.61 × 10–5, n = 11) revealed no relationship between oestrus duration and testes mass (slope b = 0.29, t = 1.39, P = 0.201), most likely due to the strong effect of body mass (slope b = 1.00, t = 7.52, P < 0.001) in our small sample. Future research should explore the relationship between the duration of the female receptive period and sperm hookedness utilising empirical methods, and firmly establish the function of the sperm hook and its role in facilitating the retainment of sperm fertilising potential.

In conclusion, we present an investigation of the role of the sperm hook in facilitating the formation of sperm groups in house mice, and a comparative analysis to account for the evolution of the sperm hook among muroid rodents. Our in vivo observations of mouse sperm behaviour revealed no evidence that motile sperm groups form via the sperm hook. Moreover, among populations the rate of mixed paternity within litters did not explain variation in the curvature of the hook, suggesting that variation in sperm competition risk is not influencing the evolution of sperm head morphology among the seven populations of house mice sampled here. More research is required to properly elucidate the role of the muroid sperm hook in facilitating the retainment of sperm motility prior to ovulation. In so doing, we may then establish whether sperm competition has contributed to the divergence in sperm head morphology across muroid rodents.


We thank A. Denhlom for collecting the sperm images, M. Werner for the construction of Fig. 1, and J. Fitzpatrick for assistance with the phylogenetic analyses. This study was approved by the UWA Animal Ethics Committee (300/100/299), and funded by the Australian Research Council.