When females mate with more than one male in a single reproductive event, sperm from rival males are forced to compete for fertilization (Parker 1998). Theory predicts that sperm competition will favour increased spermatogenic investment, and thus can be an influential force in the evolution of testes size and sperm number (Parker 1998). Consistent with theory, experimental evolution studies have shown that sperm competition selects for larger (Hosken & Ward 2001) and more efficient testes (Firman & Simmons in press). Comparative studies across taxa have used testes size (relative to body size) as a proxy for sperm competition risk, and investigated the evolutionary associations between sperm competition and sperm morphology. These studies have provided conflicting evidence for the role of sperm competition in the evolution of sperm size. Sperm competition has been shown to have no influence on sperm length (Gage & Freckleton 2003), and to select for increased sperm length (Gomendio & Roldan 1991; Briskie et al. 1997; Byrne et al. 2003). Thus, the influence of sperm competition on sperm morphology and function remains controversial (Humphries et al. 2008).

In 2005, wild-derived house mice (n = 120) were obtained from an animal resources centre (Murdoch, Western Australia, Australia) and subjected to laboratory evolution (Firman & Simmons in press). Eight selection lines were established and maintained under either a polygamous (P) or a monogamous (M) mating regime. Historical information on the source population used to establish the selection lines, the mating design and selection regime has been described previously (Firman & Simmons in press). Following five generations of selection, eight-week-old males (n = 68) were sacrificed via lethal injection and used for sperm analyses. Males from four M lines and three P lines were used in this experiment. Sperm from the caudal epididymides were incubated at 37°C for 90 min in a culture medium (Firman & Simmons 2008a). A detailed description of the medium and culture conditions is described elsewhere (Murase & Roldan 1996). Sperm velocity was quantified using the CEROS computer-assisted sperm analysis system (CASA) (v.10, Hamiliton & Thorne Research) using standard mouse parameters (Nayernia et al. 2003). An approximately 10 µl aliquot of sperm suspension was loaded into a haemocytometer and five fields of view were scanned. The CASA gave measures of total sperm number and the percentage motile sperm, and provided velocity parameters including: average path velocity (VAP), curvilinear velocity (VCL), straight line velocity (VSL), beat cross frequency, amplitude of head displacement, linearity and straightness. Samples were maintained at 37°C and CASA scans were repeated at set intervals over 5 h. These data were use to construct a ‘percentage motile sperm’ decay curve, the magnitude of which was used as a measure of sperm longevity.

Preliminary ANOVAs showed that there was significantly more variation between individuals than within individuals in sperm number, sperm morphology measurements and sperm velocity traits (p < 0.001). As such, repeatability estimates were high (r > 0.966; Becker 1984), reflecting that we had sufficient replication to capture individual variation (table S1 in the electronic supplementary material). To summarize the variation among the highly correlated sperm velocity traits, we performed a principal component analysis, which produced a single principal component that accounted for 50 per cent of the total variation, and had a corresponding eigenvalue of 3.47 (table S2 in the electronic supplementary material). VAP, VSL and VCL contributed significantly to PC1 (hereafter referred to as composite sperm velocity) (Mardia et al. 1979) (table S2 in the electronic supplementary material). We performed statistical analyses to investigate the relationship between sperm morphology and composite sperm velocity. Our independent unit of replication was the number of lines for each selection treatment (M = 4, P = 3). Thus, we conducted nested ANOVAs with replicate line nested within selection history as a random factor. Of the five sperm components, sperm midpiece length was the only significant predictor (table 1); sperm with longer midpieces had faster swimming velocities (figure 1a). The selection history of the lines from which males were sampled also had a significant effect on sperm velocity; males from the polygamous lines had, on average, greater PC1 values (0.73 ± 0.27) compared with males from the monogamous lines (−0.67 ± 0.25) (table 1). Males from the P lines had on average higher mean values compared with males from the M lines for many of the sperm performance traits (table S1 in the electronic supplementary material). However, there was no statistically significant divergence in the total number of sperm (table 1) or sperm morphology (table S3 in the electronic supplementary material). There was a significant negative relationship between composite sperm velocity and sperm longevity (table 1; figure 1b).

Analyses across mammal taxa have lead to the intuitive conclusion that sperm competition may select for larger, faster sperm (Gomendio & Roldan 1991; Anderson & Dixson 2002; Fitzpatrick et al. 2009). However, there have been few direct empirical investigations to support this hypothesis. Here, we provide conclusive data on the relationship between sperm form and sperm function, and show that sperm midpiece length predicts sperm swimming velocity in house mice.