SEARCH

SEARCH BY CITATION

Keywords:

  • microsatellites;
  • Mus domesticus;
  • polyandry;
  • sperm competition

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Polyandry generates selection on males through sperm competition, which has broad implications for the evolution of ejaculates and male reproductive anatomy. Comparative analyses across species and competitive mating trials within species have suggested that sperm competition can influence the evolution of testes size, sperm production and sperm form and function. Surprisingly, the intraspecific approach of comparing among population variation for investigating the selective potential of sperm competition has rarely been explored. We sampled seven island populations of house mice and determined the frequency of multiple paternity within each population. Applying the frequency of multiple paternity as an index of the risk of sperm competition, we looked for selective responses in male reproductive traits. We found that the risk of sperm competition predicted testes size across the seven island populations of house mice. However, variation in sperm traits was not explained by sperm competition risk. We discuss these findings in relation to sperm competition theory, and other intrinsic and extrinsic factors that might influence ejaculate quality.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Multiple mating by females (polyandry) is a common mating strategy observed across many taxa (Birkhead & Møller, 1998; Simmons, 2001). Females may gain direct benefits from soliciting multiple mates (Ridley, 1988; Arnqvist & Nilsson, 2000) and/or acquire genetic benefits that improve offspring fitness (Tregenza & Wedell, 1998; García-González & Simmons, 2005a; Fisher et al., 2006). It has been suggested that the frequency of polyandry within populations may be indicative of the benefits females gain (Bretman & Tregenza, 2005). For example, genetic benefit models predict that the frequency of polyandry will increase as genetic variation within the population increases (Petrie & Lipsitch, 1994). Alternatively, if polyandry facilitates the avoidance of inbreeding, the frequency of multiple mating is expected to be greater in populations with reduced genetic variation (Brooker et al., 1990).

Molecular markers are an essential tool for investigating the frequency of polyandry in natural populations (Bretman & Tregenza, 2005). Numerous studies utilizing microsatellite loci have provided evidence of mixed paternity within populations, and estimates of the number of sires contributing to litters or clutches (e.g. Baker et al., 1999; Kraaijeveld-Smit et al., 2002; Bretman & Tregenza, 2005; Dean et al., 2006; Simmons et al., 2007). However, this approach only reveals the number of contributing sires and not necessarily the number of males with which a female has mated. Thus, the frequency of multiple paternity often underestimates the true frequency of polyandry (Coltman et al., 1999). This problem has been surmounted by genotyping sperm stored in the spermathecae of wild-caught female insects (Bretman & Tregenza, 2005; Simmons et al., 2007) and by generating simulated mating frequencies from genetic data and competitive skews (Neff & Pitcher, 2002; Dean et al., 2006).

An inevitable consequence of polyandry is that sperm from rival males will overlap in the female reproductive tract and compete to fertilize available ova (Parker, 1970). The outcome of sperm competition is typically determined by bias in sperm use by the female (Eberhard, 1996), interactions between parental genotypes (Zeh & Zeh, 1996) and ejaculate characteristics that provide a competitive fertilization advantage (Parker, 1970). Thus, sperm competition is recognized as a persuasive force in the evolution of male reproductive traits (Smith, 1984; Birkhead & Møller, 1998; Simmons, 2001).

Comparative analyses across different taxa have suggested that sperm competition favours increased testes size and sperm production (Harcourt et al., 1981; Gage, 1994; Hosken, 1997; Møller, 1998), increased ejaculate volume (Ramm et al., 2005), longer (Gomendio & Roldan, 1991; Briskie et al., 1997; LaMunyon & Ward, 1999; Byrne et al., 2003) or sometimes shorter sperm (Stockley et al., 1997; Immler et al., 2007) and improved sperm fertilizing ability (Gomendio et al., 2006). Experimental evolution studies using invertebrates have confirmed that testes size and sperm numbers increase in populations subject to sperm competition (e.g. Hosken & Ward, 2001; Pitnick et al., 2001) and sperm competition trials have shown that sperm size (Gage & Morrow, 2003; García-González & Simmons, 2007), sperm motility (Gage et al., 2004) and sperm quality (García-González & Simmons, 2005b) can all influence a male’s fertilization success. Thus, selection via sperm competition may influence a range of ejaculate characteristics. However, studies that examine within-species variation in the strength of selection by sperm competition across natural populations and the resultant microevolutionary changes in sperm competition traits are rare (Ribble & Millar, 1992; Brown & Brown, 2003; Long & Montgomerie, 2005).

Island systems offer a unique opportunity to investigate microevolutionary change driven by selective processes. Island populations are discrete, genetically closed entities that respond quickly to selection (Millien, 2006). For example, insular rodent populations exhibit very rapid rates of microevolution, with morphological divergence principally driven by selection rather than founder or random events (Pergams & Ashley, 2001). Nonendemic house mouse (Mus domesticus) populations persist on islands along the coast of Western Australia (Abbott & Burbidge, 1995). In nature, female house mice are actively polyandrous, and produce litters sired by more than one male (Dean et al., 2006). Furthermore, the rate of multiple mating within natural mouse populations is estimated to range between 45% and 70%, creating the potential for varied sperm competition risk among populations (Dean et al., 2006).

Here, we investigated the microevolutionary response of testes size and sperm traits to sperm competition risk among seven island populations of house mice. We genotyped pregnant females and their offspring at microsatellite loci and applied two different methods to obtain estimates of the rates of multiple mating by females within populations. Multiply sired litters occurred in all populations. We calculated the expected frequency of multiple paternity within populations and applied this as an index of the risk of sperm competition. We found that males from populations under high risk of sperm competition had larger testes than males from populations under low risk of sperm competition. Variation in sperm traits across populations was not explained by sperm competition risk. Our results show that sperm competition is an important selective force in the evolution of testes size in the house mouse.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Sampling

House mice (M. domesticus) were trapped in Elliot small mammal traps (baited with peanut butter and rolled oats) during the breeding seasons of seven island populations located along the coast of Western Australia (Carnac: September 2004; Whitlock: October 2006/2007; Boullanger: October 2005/2006/2007; Rat: November 2004; Beacon: June 2005/2006; Dirk Hartog: August 2005/2006; Thevenard: January 2004) (Table 1). Many of these populations were established by stowaways on wrecked ships in the 1600s; the most recent establishment occurred during the 1980s (Abbott & Burbidge, 1995). Rat Island and Beacon Island occur in the Abrolhos Archipelago, and Whitlock Island and Boullanger Island are located in the same group (Jurien Bay). Intermittently, humans inhabit Rat Island, Beacon Island and Thevenard Island. Consequently, the populations on these islands are semi-commensal.

Table 1.   Locations and characteristics of seven island populations of house mice.
IslandLocationPopulation sizePopulation densityMale lengthFemale lengthLitters (n)Litter size
  1. Mean (±SE) male and female lengths (mm) and litter sizes are presented. Population density is given as the number of animals caught per trap night.

Carnac32°07′S, 115°39′E2630.64083.6 ± 0.776.4 ± 2.0226.0 ± 0.3
Whitlock30°19′S, 114°59′E1110.35684.1 ± 1.082.5 ± 2.3107.6 ± 0.5
Boullanger30°19′S, 115°00′E3260.15283.5 ± 0.781.2 ± 1.4307.5 ± 0.3
Rat28°42′S, 113°47′E7720.53373.1 ± 0.672.1 ± 1.3215.7 ± 0.3
Beacon28°30′S, 113°47′E1290.34978.2 ± 0.782.6 ± 1.2185.6 ± 0.5
Dirk Hartog25°05′S, 113°03′E670.10475.7 ± 0.674.6 ± 0.8135.8 ± 0.3
Thevenard21°27′S, 114°59′E2170.30154.6 ± 0.870.5 ± 1.1144.8 ± 0.4

Fifty sexually mature males were collected from each population and transported to the University of Western Australia (except for Whitlock Island, in which n = 26). All females caught in traps were killed and dissected. Ear tissue and embryos were stored separately in 100% ethanol for genotyping and paternity analysis. Only embryos large enough to avoid contamination from maternal tissue were included in the study.

We used trapping data to estimate population sizes and densities. Population size was calculated via a population depletion method, given by the equation N = −(a/b), where a is the intercept and b the slope of cumulative catch numbers plotted against catch per unit effort (i.e. number of mice caught/number of traps). Capture rate was used to determine population densities (i.e. density = number of mice caught per trap night).

Genotyping

DNA was extracted from 1 mm3 of female ear and embryo tissue using the EDNA HISPEX extraction kit (Fisher Biotec, Subiaco, WA, Australia). The samples were screened using 11 microsatellite loci: D4Mit1 +D10Mit14 + D13Mit1 + D18Mit17; D6Mit138 + D11Mit4 +D14Mit132 + D16Mit1 and D1Mit17 + D2Mit1 +D15Mit13 (Dietrich et al., 1992; Dean et al., 2006) (Appendix S1 in the online Supporting Information). Due to low statistical power to detect multiple paternity within populations from the Whitlock, Boullanger and Beacon islands, these populations were screened at five additional loci: D8Mit267 + D11Mit29 + D12Mit4 + D15Mit174 +D17Mit51 (Dietrich et al., 1992; Dean et al., 2006) (online Appendix S1). After doing so, the Beacon Island population still had low statistical power to detect multiple paternity, and thus was genotyped at four more loci: D1Mit322 + D2Mit277 + D8Mit121 + D17Nds2 (Dietrich et al., 1992; Dean et al., 2006) (online Appendix S1). Labelled forward primers were obtained from Geneworks (Hindmarsh, SA, Australia) (FAM) and Applied Biosystems (Foster city, CA, USA) (NED, PET, VIC) and unlabelled primers from Geneworks. Primers were multiplexed in 10 μL reactions in a PTC-0200 DNA Engine (Geneworks). Reactions contained 5 or 6 μL of Qiagen Multiplex Kit, 0.25 μm of forward labelled primer, 0.25 μm of reverse primer and ∼200 ng of template DNA. The thermocycling profile for all loci was 5 min denature at 95 °C, 50 cycles of 90 °C for 20 s, 55 °C for 20 s and 72 °C for 30 s, followed by 72 °C for 3 min. PCR products (1.5 μL) were run on a ABI3730 Sequencer, sized using Genescan-500 LIZ size standard and genotyped using Genemapper software (v3.0; Applied Biosystems).

Paternity analysis

The number of sires per litter was estimated by allele counting, and with the software package gerud (Jones, 2001). Allele counting involved identifying and subtracting the maternal allele from each embryo and inferring a paternal allele. At each locus, homozygotes were assumed to have one maternal allele and one paternal allele, and heterozygotes with the same genotype as the mother were taken to have one contributing paternal allele (Dean et al., 2006; Simmons et al., 2007). The greatest number of nonmaternal alleles at a single locus was taken as an estimate of the number of paternal alleles. The total number of paternal alleles was divided by two (as each sire could donate two different alleles), giving a conservative estimate of the number of males contributing to a litter. gerud uses multiple loci simultaneously, removes maternal alleles from offspring genotypes and then simulates all possible paternal genotypes before calculating the combinations of these genotypes that yield the fewest possible number of males that could have contributed to the observed offspring genotypes (Jones, 2001). We performed analyses to assess the statistical power of the microsatellite loci to detect multiple paternity within each population (PrDM) and used these values to calculate the expected frequency of multiple paternity for each population (Neff & Pitcher, 2002).

The frequency of multiple paternity within each population was determined as the proportion of litters that were sired by more than one male. The frequency of multiple paternity provided a conservative estimate of the frequency of polyandry, and an index of the risk of sperm competition within populations.

Male reproductive traits

Males were held in a constant room temperature (22 °C) for 7 days before being killed and dissected. Testes and epididymis weights were recorded before the caudal epididymes 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), the sperm suspension was transferred to a prewarmed tube and swimming sperm were filmed using an Hitachi HV-C20E/K-SA camera (North Ryde, NSW, Australia) fixed to a compound microscope. The VHS recordings were converted into digital clips, and sperm motility quantified using the CEROS computer-assisted sperm analysis (CASA) system (v10; Hamilton and Thorne Research, Beverly, MA, USA) with parameters set at frame rate, 60 Hz; minimum contrast, 80; minimum cell size, 20 pixels; minimum progressive average path velocity (VAP), 50 μm s−1 and VAP cut-off 7.4 μm s−1. Rapid cells were defined as cells with VAP > progressive cell VAP. Cells with slow motility were set as static and not included in the analysis. The CASA provided measures of percentage of motile sperm, percentage of progressive sperm, percentage of rapid sperm and VAP (μm s−1). Motility scans at set intervals over 5 h provided a motility decay curve, the magnitude of which was used as a measure of sperm longevity. We were unable to obtain motility data for the Thevenard population.

At the time of sperm capacitation (i.e. 90-min incubation), 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 to determine the percentage of acrosome-reacted cells (viable) (Larson & Miller, 1999). One hundred sperm were examined for the presence of the characteristic dark acrosomal crescent. Images of stained sperm enabled measurements of sperm tail length and total sperm length (×400 magnification) and head length, head width and midpiece length (×1000 magnification) using the image analysis application ImageJ (v1.32). Sperm numbers were obtained by counting sperm cells in an Improved Neubauer haemocytometer. Body weight (F1,323 = 14.97, < 0.001) and testes weight (F1,323= 4.33, = 0.038) were both predictors of epididymal sperm number (whole model: F2,323 = 21.40, < 0.001).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Population characteristics

An analysis of island location and body length revealed a predicted biogeographical cline in body size (Bergmann’s Rule, Freckleton et al., 2003) (Table 1). Males from populations on islands at higher latitudes and cooler temperatures were larger than males from populations at lower latitudes (F1,5 = 28.93, r2 = 0.852, = 0.003). Female size showed a similar trend (F1,5 = 2.50, r2 = 0.577, = 0.175). Moreover, there was a tendency for larger females to have larger average litter sizes (F1,5 = 4.28, r2 = 0.461, = 0.093). The level of genetic diversity was consistently low across populations (Table 2). The average observed and expected heterozygosity across populations at 11 microsatellite loci was HO = 0.34 ± 0.04, HE = 0.35 ± 0.04, compared with that recorded for an Australian mainland population where HO = 0.55 (Dean et al., 2006).

Table 2.   Observed and expected heterozygosities of island populations of house mice for 11 microsatellite loci.
IslandMean NAMin. NAMax. NAMean HOMean HEMean FIS
  1. Values in parentheses include *five or †nine additional loci. NA refers to the number of observed alleles.

Carnac2.64 ± 0.39150.37 ± 0.030.36 ± 0.080.12 ± 0.03
Whitlock*2.18 ± 0.35 (2.44 ± 0.27)1 (1)5 (5)0.38 ± 0.08 (0.42 ± 0.07)0.42 ± 0.07 (0.41 ± 0.06)0.23 ± 0.08 (0.19 ± 0.05)
Boullanger*2.36 ± 0.41 (2.56 ± 0.33)1 (1)5 (5)0.28 ± 0.07 (0.35 ± 0.07)0.29 ± 0.08 (0.35 ± 0.06)0.09 ± 0.03 (0.10 ± 0.02)
Rat2.73 ± 0.43160.39 ± 0.060.41 ± 0.060.15 ± 0.04
Beacon†2.00 ± 0.19 (2.05 ± 0.17)1 (1)3 (4)0.29 ± 0.07 (0.37 ± 0.05)0.26 ± 0.07 (0.37 ± 0.04)0.15 ± 0.03 (0.15 ± 0.02)
Dirk Hartog2.91 ± 0.34150.41 ± 0.070.48 ± 0.060.20 ± 0.05
Thevenard2.91 ± 0.28140.46 ± 0.060.49 ± 0.060.24 ± 0.04

Power to detect multiple paternity

We performed analyses to assess the power of the microsatellite loci to detect multiple paternity within each population. The programme by Neff & Pitcher (2002) employs four parameters to calculate the probability of detecting multiple paternity (PrDM): (1) the number of descriptive loci; (2) the number and frequency of alleles at each loci; (3) the number of offspring per litter; and (4) the number of potential sires and their reproductive skew. The average litter size of each population and the average reproductive skew generated by gerud was used in the analysis of each population. For each of the seven populations, power analyses were conducted on the original 11 microsatellite loci. These analyses revealed that the Carnac (0.605), Rat (0.536), Dirk Hartog (0.646) and Thevenard (0.539) islands had PrDM values > 0.5, whereas the Whitlock (0.465), Boullanger (0.323) and Beacon (0.133) islands had PrDM values < 0.5. Thus, females and their offspring from the Whitlock, Boullanger and Beacon Island populations were genotyped at five additional loci. These additional microsatellite loci largely improved the probability to detect multiple paternity for the Whitlock (0.602) and Boullanger (0.640) Island populations, but only made a small improvement for the Beacon Island (0.261) population. Thus, the Beacon Island population was genotyped at four more loci, which raised the PrDM value and made it comparable with the other six populations (0.498). Although the inclusion of the additional microsatellite loci increased the power to detect multiple paternity within the Whitlock, Boullanger and Beacon Island populations, the observed frequency of multiple paternity did not change in these populations. For all seven populations, the expected frequency of multiple paternity corrected for the power to detect multiple paternity was calculated following Neff & Pitcher (2002): (observed frequency of multiple paternity) (PrDM)−1 (Table 3).

Table 3.   Paternity analysis of island populations of house mice using the allele counting and gerud methods.
IslandAllele countinggerudExpected MP
MPNo. of siresInformative lociNAMPNo. of siresSire combinationPaternity skew
  1. Both methods gave the same frequencies of multiple paternity (MP). A maximum of two sires were observed within litters showing multiple paternity (except Rat Island, in which there was evidence of three sires by allele counting). For allele counting, the number of informative loci and the number of paternal alleles (NA) are given. For gerud, the mean ± SE number of sire combinations and paternity skew are given. Expected MP was calculated by (observed MP) (PrDM)−1. Three populations were genotyped at *five or †nine additional loci to increase the power to detect mixed paternity.

Carnac0.31821–2 3 or 60.318221.9 ± 4.50.61 ± 0.020.526
Whitlock*0.10021–23 or 60.100222.0 ± 0.00.70 ± 0.030.166
Boullanger*0.20021–33, 6 or 90.200215.7 ± 1.80.64 ± 0.030.313
Rat0.38131–23, 4 or 60.381225.9 ± 8.90.58 ± 0.010.711
Beacon†0.0562130.056216.0 ± 0.00.63 ± 0.030.112
Dirk Hartog0.30821–3 3, 6 or 90.308217.0 ± 6.00.56 ± 0.010.477
Thevenard0.42921–3 30.429228.0 ± 4.10.60 ± 0.020.796

Multiple paternity

The allele counting method and gerud gave the same estimates of the observed frequency of multiple paternity (Table 3). A maximum of two sires was observed within litters showing multiple paternity, except for one litter from the Rat Island population, in which there was evidence of three sires by allele counting. The paternity skew values given by gerud suggest that for litters with multiple paternity, a single male sired the majority of pups. Although the range in the observed frequency of multiple paternity among populations was clearly considerable (6–43%), this variation failed to reach statistical significance (Monte Carlo simulation, = 0.175 ± 0.002). The observed frequency of multiple paternity across all litters (26%) was slightly higher than that reported for mainland populations (23%) (Dean et al., 2006).

All analyses were conducted on the expected frequency of multiple paternity that incorporated PrDM. Population density did not predict multiple paternity (F1,5 = 0.31, = 0.604). Furthermore, the frequency of multiple paternity did not differ between commensal island populations and noncommensal populations (χ2 = 0.074, d.f. = 2, < 0.99). The coefficient of inbreeding (FIS) (F1,5 = 0.69, = 0.445), average litter size (F1,5 = 3.28, = 0.130) and number of litters sampled (F1,5 < 0.01, = 0.966) also did not predict the frequency of multiple paternity among populations.

Male reproductive traits

To summarize the variation among the 12 highly correlated sperm traits, we applied a principal components analysis (PCA). We extracted three components that collectively accounted for 61.3% of the variation in sperm quality (online Appendix S2). Variables that were at least 0.7 times as large as the largest eigenvector were considered to have contributed significantly to that PC (Mardia et al., 1979). The first principal component (PC1) accounted for approximately 30% of the variation and was weighted predominantly by measures of sperm motility (percentage motile, rapid and progressive sperm and sperm velocity). The second principal component (PC2) accounted for 17% of the variation, and described sperm length (tail and total sperm length). The third principal component (PC3) explained 13% of the variation and was weighted by sperm head length, sperm midpiece length and sperm longevity (online Appendix S2).

To account for the geographical cline in body size, male body weight was included as a covariate in analyses of testes size and sperm number (Table 4). ancova revealed significant among-population variation in testes size (F6,318 = 17.00, < 0.001), sperm number (F6,318 =30.81, < 0.001) and sperm quality (PC1: F5,263 = 3.31, = 0.007; PC2: F5,263 = 3.85, = 0.002; PC3: F5,263 = 60.51, < 0.001) (online Appendix S3). After controlling for body size (F1,4 = 29.66, = 0.001), a significant proportion of the among-population variation in mean testes size was predicted by the expected frequency of multiple paternity (F1,4 = 8.90, = 0.041; Fig. 1). However, among-population variation in relative epididymal sperm number (F1,4 = 0.004, = 0.952) or sperm quality (PC1: F1,4 = 0.33, = 0.60; PC2: F1,4 = 0.101, = 0.767; PC3: F1,4 = 1.59, = 0.041) were not predicted by the expected frequency of multiple paternity. Qualitatively similar results were obtained for the observed frequency of multiple paternity (analyses not shown). Island latitude, population density and the coefficient of inbreeding did not account for among-population variation in testes size, sperm number or sperm quality indices (online Appendix S4).

Table 4.   Mean ± SE male body weight (g), testes weight (mg) and epididymal sperm number of island populations of house mice (n = 50, except for Whitlock where n = 26).
IslandBody weightTestes weightNo. of sperm
Carnac22.0 ± 0.5159.5 ± 3.417.7 ± 1.0
Whitlock20.9 ± 0.7143.6 ± 5.87.2 ± 0.7
Boullanger19.4 ± 0.4125.6 ± 3.611.6 ± 0.8
Rat15.9 ± 0.4116.0 ± 2.313.7 ± 0.9
Beacon18.2 ± 0.4111.8 ± 2.014.3 ± 0.9
Dirk Hartog15.8 ± 0.3123.1 ± 2.116.5 ± 1.0
Thevenard15.1 ± 0.3121.6 ± 2.64.5 ± 0.4
image

Figure 1.  Residual testes weight (±SE) as predicted by the expected proportion of multiply sired litters, among seven island populations of house mice (n = 50, except for Whitlock where n = 26).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

We used microsatellite loci to genotype pregnant females and their offspring, and obtained estimates of the levels of multiple paternity within seven island populations of house mice. We utilized two conservative methods for calculating the observed frequency of multiple paternity, both of which gave identical estimates for each population. Multiple paternity was evident in all populations, the frequency of which ranged from 6% to 43% of litters. Initial power analyses of our genetic data revealed that the Whitlock, Boullanger and Beacon Island populations had less power to detect multiple paternity than the Carnac, Rat, Dirk Hartog and Thevenard Island populations. Thus, we genotyped the low-power populations at additional loci, and then calculated the expected frequency of multiple paternity by incorporating the power of the microsatellite loci to detect multiple paternity within each population (Neff & Pitcher, 2002). These analyses suggest that the frequency of multiple paternity among these island populations is likely to range from 16% to 79%. Within multiply sired litters, a maximum of two sires were detected. However, paternity skews generated by gerud suggested that one male sired most of the offspring. This pattern of paternity is commonly observed in mammals, and may reflect a paternity bias toward the male mating closest to the time of ovulation (Lacey et al., 1997; Gomendio et al., 1998). Our laboratory studies of house mice show that paternity is skewed toward the first male to mate, when females mate with two males (Firman & Simmons, 2008a) and when they mate with three males (Firman & Simmons, 2008b). Given the relatively small litter sizes, and the estimates of competitive skew found here, Dean et al.’s (2006) simulations suggest that the detection of more than two sires within a mixed paternity litter is unlikely. Thus, although the genetic data suggest that females mated with just two males, the actual number of mates, and thus the intensity of sperm competition, is likely to be higher.

Polyandry may allow females to avoid reproductive failure (Tregenza & Wedell, 1998) or to accrue additive genetic benefits that improve offspring fitness (García-González & Simmons, 2005a; Fisher et al., 2006). It has been suggested that the frequency of polyandry will be related to population levels of variation for fitness-related genes (Petrie & Lipsitch, 1994; Bretman & Tregenza, 2005). For avian colonies, it is predicted that polyandry would be more frequent in populations in which there are greater densities of breeding males, and thus high variation in male genetic quality (Petrie & Lipsitch, 1994). However, intraspecific investigations testing this prediction have produced inconsistent evidence (reviewed in Griffith et al., 2002). Island populations of mammals are recognized as having low levels of genetic diversity/heterozygosity, typically due to a small number of founding individuals and high levels of inbreeding (Frankham, 1997; White & Searle, 2007). Thus, the intensity of sexual selection and the frequency of polyandry are predicted to be lower among island populations compared with among conspecific mainland populations (Petrie & Lipsitch, 1994). A comparative study across 54 passerine species provided support for this hypothesis. Insular, island populations of passerine birds exhibited lower levels of extra-pair paternity than equivalent mainland populations (Griffith, 2000). Alternatively, when polyandry facilitates the avoidance of inbreeding, multiple paternity is expected to occur more frequently in populations with low genetic variation (Brooker et al., 1990). In our study, the level of genetic variation within island populations of house mice was consistently low compared with mainland populations (Dean et al., 2006). However, low genetic diversity did not inhibit our ability to detect multiple paternity; the average frequency of multiple paternity across all litters was higher than that reported for litters from more genetically diverse mainland populations (Dean et al., 2006). In house mice, polyandry facilitates inbreeding avoidance by enabling paternity to be biased towards unrelated males (Firman & Simmons, 2008a). Thus, greater rates of multiple mating might be expected in relatively inbred populations. In this study, the coefficient of inbreeding did not explain the frequency of multiple paternity among the island populations. However, if polyandry facilitates bias sperm use against related individuals, then it would reduce population inbreeding levels. In this case, the relationship between population level of inbreeding and of multiple mating might be obscured by a ‘cancelling out’ effect. Nevertheless, five of the seven islands had higher observed frequencies of multiple paternity than the average reported for more outbred mainland mouse populations (Dean et al., 2006). Therefore, the possibility that elevated levels of polyandry in island populations of mice may reflect a strategic behavioural adaptation of inbreeding avoidance by females warrants further study, particularly given the difficulties inferring the number of males with which females mate.

Polyandry results in covert competition between the ejaculates of males (Parker, 1970) and sperm competition is recognized as a persuasive force in the evolution of testes size and sperm number (e.g. Hosken & Ward, 2001; Pitnick et al., 2001). We conducted multiple tests (i.e. testes size, epididymal sperm number and three principal components) and used the frequency of multiple paternity as an index of the risk of sperm competition. We found that males from populations under a higher risk of sperm competition had larger relative testes size than males from populations under a lower risk of sperm competition. Parker’s ESS models of ejaculate evolution assume that sperm competition conforms to a raffle, with the relative number of sperm from each competitor being the primary determinant of success (Parker, 1998). Thus, when sperm competition risk is high, selection is predicted to favour increased investment in sperm production by increased testes mass. This prediction has been supported largely by correlations between the risk of sperm competition and testes size across taxa (Harcourt et al., 1981; Møller 1998; Gage, 1994; Hosken, 1997) and adjustments in ejaculate expenditure when males are presented with a risk of sperm competition (Gage, 1991; Lacey et al., 1997; Simmons et al., 1999; for mice see Preston & Stockley, 2006; Ramm & Stockley, 2007). Studies of natural populations of cliff swallows have shown that polyandry occurs at higher frequencies in larger colonies (Brown & Brown, 1996) and males from large colonies have larger relative testes size compared with males from smaller colonies (Brown & Brown, 2003). Here, utilizing the same intraspecific approach, we also provide support for sperm competition theory, showing that variation in sperm competition risk predicts variation in testes size across natural populations of house mice.

However, sperm competition risk did not account for variation in ejaculate characteristics, such as epididymal sperm number, sperm size and sperm quality. Ejaculate quality has been shown to be influenced by many factors and to vary considerably both within and between individuals. We can eliminate the possibility that seasonal breeding patterns contributed to variation in semen quality (Penfold et al., 2000; de Haas van Dorsser & Strick, 2005), as sampling was conducted during the reproductive period of each population. Reduced nutrient and water consumption can depress male reproductive function (Keast & Marshall, 1954; Cook, 1991; Nelson, 1993). Thus, variation in local environmental conditions among islands, such as food and water availability, may have induced trade-offs between general and reproductive maintenance and influenced seminal parameters. The prevalence of viruses and pathogens varies considerably between Australian island populations of house mice (Moro et al., 1999, 2003). Not surprisingly, health status also affects semen quality (Kortet et al., 2004; Aitken et al., 2004). Additionally, confounding factors that we could not control, such as differences in male mating history (Huber et al., 1980; Bissoondath & Wiklund, 1996), age (Huber et al., 1980; de Haas van Dorsser & Strick, 2005; Crosier et al., 2007) and social status (Koyama & Kamimura, 1999; Rudolfsen et al., 2006), may account for variation in epididymal sperm numbers, both within and between island populations. Finally, high levels of inbreeding have been shown to impair sperm quality in populations of mammals (Roldan et al., 1998; Gomendio et al., 2000; Gage et al., 2006). In our study, the coefficient of inbreeding did not explain variation in male reproductive traits among the island populations. However, we cannot reject the possibility that inbreeding has influenced the evolution of sperm quality within these populations. Here, the inbreeding coefficient was derived from individuals genotyped at only 11, 16 or 20 microsatellite loci, which may not provide a reliable estimate of the degree of inbreeding within each population. The application of large numbers of loci (> 30) is necessary to obtain an accurate measure of genome-wide heterozygosity within individuals and the estimation of rigorous inbreeding coefficients for populations (Gage et al., 2006).

Unfortunately, we do not have population data beyond what was acquired in the present study. Thus, we cannot state whether or not these levels of multiple paternity would be repeatable and acknowledge that we can only present a snapshot of the long evolutionary history of these populations. Nevertheless, we have shown that multiple paternity within litters is a common feature of natural populations of house mice. Moreover, population variation in the rates of mixed paternity litters, and thus the risk of sperm competition for males, are significant predictors of testes size variation across seven geographically isolated island populations. We hereby provide the first intraspecific demonstration that sperm competition can drive macroevolutionary changes in mammalian testes size. We hope our study inspires others to address this issue using larger sample sizes, and to assess whether this intriguing pattern of multiple paternity and testes size occurs in other mammalian species.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

For access to the islands and assistance in the field, we thank T. Friend and T. Button from the Department of Environment and Conservation Western Australia (WA); A. Darbyshire and G. Meinema from Department of Fisheries WA; and W. Gibb (UWA). Many thanks to M. Love and the other volunteers who assisted with fieldwork. The sperm images were prepared by A. Denholm, and P. Mose formatted the sperm motility clips. We are grateful to M. Jennions for providing valuable comments on the manuscript. This study was approved by the UWA Animal Ethics Committee (300/100/299) and funded by the Australian Research Council.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Appendix S1 Characterization of 20 microsatellite loci used for paternity analysis for seven island populations of house mice

Appendix S2 Principal components analysis of sperm traits of island populations of house mice

Appendix S3 Mean (±SE) sperm traits and PC values of island populations of house mice (n = 50, except for Whitlock where n = 26)

Appendix S4 Analyses of testes size, sperm number and principal components (PC) with island latitude, population density and the coefficient of inbreeding (FIS)

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

FilenameFormatSizeDescription
JEB_1612_sm_appendices.doc130KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.