The evolution of polyandry: an examination of the genetic incompatibility and good-sperm hypotheses

Authors


Leigh W. Simmons, Evolutionary Biology Research Group, Department of Zoology, The University of Western Australia, Nedlands, WA 6009, Australia. Tel.: +8 9380 2221; fax: +8 9380 1029; e-mail: lsimmons@cyllene.uwa.edu.au

Abstract

I have examined the adaptive significance of polyandry using the Australian field cricket Teleogryllus oceanicus. Previous studies of polyandry have examined differences in offspring production by females mated multiply to a single male or females mated multiply to different males. Here I combine this approach with a study of parentage of offspring produced in the later group. Females mated to two different males had a higher proportion of their eggs hatching than did females mating twice with a single male. Offspring fitness parameters were not effected. There was little evidence to suggest that females elevate their hatching success via fertilizing their eggs with sperm from genetically compatible males. Although the average paternity points towards random sperm mixing, there was considerable individual variation in sperm competition success. Patterns of parentage were consistent across females mating twice or four times. Sperm competition success was not related to offspring viability or performance. Thus, the notion that competitively superior sperm produce competitively superior offspring is not supported either. The mechanism underlying increased hatching success with polyandry requires further study.

Introduction

A recurrent question in evolutionary biology has been why females should mate with so many different males. Traditionally it was thought that whereas male fitness should be an increasing function of the number of mating partners obtained, female fitness should be maximized with only one or a few matings (Bateman, 1948; Arnold & Duvall, 1994). Nevertheless, female acceptance of matings from multiple partners (polyandry), as well as multiple matings with the same partner, are taxonomically widespread phenomena (Thornhill & Alcock, 1983; Ridley, 1990; Birkhead & Møller, 1992; Hunter et al., 1993; Keller & Reeve, 1995; Eberhard, 1996). Evolutionary biologists have therefore sought adaptive explanations for the evolution of polyandry.

Where females obtain material benefits from males at copulation, or where female reproductive success is limited by the availability of viable sperm supplies, the occurrence of polyandry represents no theoretical problem. For example, Ridley (1988) found that for 58% of insect species studied, females ran out of sperm if not allowed to remate, and that in almost all species studied female fecundity was elevated through multiple mating. In their recent meta-analysis of the insect literature, Arnqvist & Nilsson (2000) found that the average net fitness gain from polyandry was as high as 30–70%. The effect of polyandry on fitness was independently significant for taxa with and without nuptial feeding, and arose because of positive effects of polyandry on egg production and fertility.

It is difficult to envisage how polyandry can be favoured in the absence of material benefits and this has led to the development of a number of theoretical models based on the potential for females to obtain genetic benefits from their behaviour (Keller & Reeve, 1995; Zeh & Zeh, 1996, 1997; Yasui, 1998; Jennions & Petrie, 2000; Tregenza & Wedell, 2000). Genetic benefit models can be broadly categorized into four groups. First, females may accept matings from additional males when their previous mates are deemed to be of inferior genetic quality (Thornhill & Alcock, 1983; Simmons, 1987b; Kempenaers et al., 1992; Graves et al., 1993). The model is essentially a ‘good genes’ model of sexual selection in that by mating with males of superior quality, females accrue indirect fitness benefits for their offspring.

Second, by mating with multiple males, females may increase the genetic diversity of their progeny. Genetic diversity arguments are particularly relevant to colonies of social insects where benefits can range from a more efficient division of labour (Robinson, 1992), to a reduced susceptibility of colonies to parasites (Schmid-Hempel, 1998; for recent reviews see Cole & Wiernasz, 1999; Jennions & Petrie, 2000). More generally, it has been proposed that a greater genetic diversity of offspring may reduce the risks of reproductive failure for females faced with variable and unknown future environments (so called ‘genetic bet-hedging’Watson, 1991). Apart from the fact that meiosis and recombination should generate sufficient genetic diversity within offspring produced by monandrous females (Williams, 1975), Yasui (1998) shows how bet-hedging is unlikely to provide the necessary selective advantage to favour the evolution of polyandry.

Third, it has been proposed that polyandry can facilitate both sperm competition and/or cryptic mechanisms of female choice that allow females to obtain indirect fitness benefits for their offspring (Harvey & May, 1989; Eberhard, 1996). In their study of adders, Madsen et al. (1992) found that the proportion of stillborn young produced by a female was negatively correlated with the number of different mating partners she had experienced. Madsen et al. (1992) suggested that if sperm that are more successful in competition were also more effective in producing viable offspring, then polyandry would increase the average viability of offspring. Parker (1992) noted that Madsen et al.′s (1992) conclusions demanded a correlation between sperm competitive success and offspring viability. Although Yasui (1997) has shown how such a ‘good-sperm’ model could theoretically explain the evolution of polyandry, as yet there is no empirical evidence that offspring of low viability come from sperm that fare badly under sperm competition.

Alternatively, females could obtain fitness benefits for their sons if males who were successful in sperm competition sired sons who were themselves successful in sperm competition (Curtsinger, 1991; Keller & Reeve, 1995). The ‘sexy-sperm’ hypothesis relies on heritable variation in sperm competitive ability and predicts a genetic correlation between sperm competitiveness and degree of polyandry. Although there is some evidence to suggest that there can be genetic variance in sperm competitive ability (Clark et al., 1995; Radwan, 1998) and female mating frequency (Solymar & Cade, 1990), there have been no tests of the predicted genetic correlation between polyandry and sperm competitiveness.

Finally, Zeh & Zeh (1996, 1997) note that for a variety of reasons males may vary in their genetic compatibility with females. Genetic incompatibility can arise through intragenomic conflict, the occurrence of selfish elements, or through inbreeding (see also Tregenza & Wedell, 2000). If females are unable to assess male genotypes prior to copulation, they could exploit post-copulatory mechanisms such as sperm competition and/or sperm selection to screen males after copulation. Thus, Zeh & Zeh (1997) argue that by increasing the genetic diversity of sperm within her sperm stores, a female can ensure that only those sperm from compatible males achieve fertilization. There is evidence that fertilization success is dependent on an interaction between male and female genotypes (Olsson et al., 1997; Clark et al., 1999). Moreover, the genetic incompatibility hypothesis could explain the reduction in stillborn young with increasing polyandry observed in adders (Madsen et al., 1992). Zeh’s (1997) own work with pseudoscorpions also shows that polyandrous females have reduced embryonic mortality compared with monandrous females, and the hypothesis has been invoked to explain a number of cases in which elevated offspring viability has been observed for polyandrous females (Olsson et al., 1994; but see Stockley, 1997; Keil & Sachser, 1998; Watson, 1998; Kempenaers et al., 1999). Nevertheless, these studies are confounded by potential material benefits obtained from copulation, because either they do not control for the numbers of matings performed by females or they are correlational (Newcomer et al., 1999).

There have been just two experimental studies of the genetic incompatibility hypothesis. First, using the field cricket Gryllus bimaculatus, Tregenza & Wedell (1998) reported an increased proportion of eggs hatching for females mated with multiple males compared with females mating the same number of times with the same or fewer males. Second, Newcomer et al. (1999) adopted the same experimental design to confirm the effect of polyandry on embryo failure noted earlier by Zeh (1997) in the pseudoscorpion Cordylochernes scorpioides. Interestingly, behavioural studies of both the cricket (Bateman, 1998) and the pseudoscorpion (Zeh et al., 1998) show that females prefer to copulate with novel males than with previous copulation partners, thereby ensuring the acquisition of sperm from multiple partners. Nevertheless, neither of these studies has demonstrated the critical assumption of the genetic incompatibility hypothesis, that viable offspring result from eggs being fertilized by one or a few compatible males.

Here I adopt the experimental approach of Tregenza & Wedell (1998) to examine potential genetic benefits of polyandry in the Australian field cricket Teleogryllus oceanicus. Moreover, I combine this approach with an examination of paternity within polyandrous females. This combined approach allows an assessment of both the genetic incompatibility hypothesis and the good-sperm hypothesis for the evolution of polyandry. The genetic incompatibility hypothesis predicts that paternity will be biased toward males with genotypes that maximize the hatching success of a polyandrous female’s eggs. The good sperm hypothesis predicts that males successful in sperm competition should sire offspring of superior viability and/or performance.

Materials and methods

The crickets were third and fourth generation of a population collected from mangrove flats in and around Cairns, Northern Queensland, Australia. Crickets were reared in a constant temperature room maintained on a 12:12 h light:dark cycle and a temperature of 25 °C. Food (cat chow) and water were provided ad libitum. Experimental crickets were separated from stock culture as penultimate instar nymphs and housed in single sex cultures.

Allozyme markers

Phenotypic variation at the beta-esterase (β-est) locus was used as a genetic marker in this study. Enzymes were extracted by grinding equal volumes of tissue (hind femur muscle for adults or entire nymphs) and buffer (0.2 M tris-HCl, 0.25 M sucrose, 0.1% mercaptoethanol and 0.02% bromophenol blue, at pH 8). Alleles were scored using starch-gel electrophoresis with tris-EDTA-borate buffer. Four alleles (a–d) were identified based on migration distance, both anodally (a) and cathodally (b–d), through the gel. Preliminary studies showed that genotype frequencies of offspring of known parentage did not deviate significantly from that expected based on mendelian inheritance. Thus, allozyme phenotypes represented underlying heritable genotypes.

Manipulation of polyandry in doubly mated females

Prior to experiments, all adult males and females were screened for their β-est genotype using a randomly acquired hind femur; crickets often autotomize a hind limb in response to predators and will do so when handled. Three females of identical genotype were chosen to mate with two males. To induce experimental variation between male and female genotypes, males were chosen so that they were either of the same genotype as the females, or they were of different genotypes. Moreover, the two males allocated to a group of three females could be of the same genotype, or they could be of different genotypes to one another. One female was mated twice to one of the males. The second female was mated twice to the second male. Finally, the third female was mated twice, once with each of the males. Thus, all females mated twice. Two females mated monandrously and one female mated polyandrously.

Matings were staged in clear plastic containers (8 × 8 × 6 cm). Typical of most field crickets, male Teleogryllus attach an externally visible spermatophore and then remain with the female while sperm are transferred (Loher & Rence, 1978; Hockman & Vahed, 1997). During this so called guarding period, males produce a new spermatophore. Males were left to guard females for 60 min before being removed. The spermatophore was also removed from the females genital opening at 60 min in order to standardize the amount of ejaculate they received (preliminary observations confirmed no significant variation between males in the numbers of sperm transferred after 60 min, L.W. Simmons, unpublished). Females were left for 24 h before remating, either with the same male or the alternate male. After their two matings, females were provided with a Petri dish containing damp sand and allowed to oviposit for 7 days. Eggs were rinsed from the sand and a random sample of 200 eggs per female was placed onto moist filter paper and incubated at 25 °C. The numbers of offspring that hatched were counted. A random sample of 50 offspring per female was placed into a 5-L plastic box with fly screen lid. Families were reared to adulthood as for stock cultures, recording the number of adults eclosing, the egg-to-adult development time, and the realized adult size measured as the width of the pronotum.

When two males differed in their β-est genotype it was possible to unambiguously assign parentage to each of the two mates of a polyandrous female. In these cases (n=17 females) a hind tibia was used to genotype each adult offspring. Furthermore, a second box of 50 randomly selected offspring was established when nymphs originally hatched. These offspring were reared for just 10 days to allow them to reach a size sufficient for genotyping. All nymphs and adult tissue was frozen at −70 °C prior to allozyme electrophoresis. Paternity estimates were based on the genotypes of 583 nymphs with a mean family size of 34 ± 4 nymphs. Paternity estimates from adult offspring were based on genotypes of 229 individuals with a mean family size of 13 ± 1 individuals.

The experiments described above gave information on sperm utilization by females and its association with offspring performance following double matings. In a second experiment I examined sperm utilization and offspring performance when 32 females were each allowed to mate with four different males. Matings were performed as described above with the exception that each female was provided, and mated with, all four of her mates on a single day. Male genotypes were chosen so that one of the females four mates, the focal male, could be assigned parentage unambiguously. The position of the focal male in the mating sequence was assigned at random. Hatching success was not assessed in this experiment. Three random samples of 50 offspring from each female were established in 5-L boxes. Two boxes of offspring were frozen at 10 days of age and the other was reared to adulthood. The number of offspring reaching adulthood was determined as well as the egg-to-adult development time and final adult size of each offspring. Finally, all adult offspring were genotyped. Paternity estimates were based on a total of 2239 10-day-old nymphs, with a mean of 66 ± 6 nymphs per family. Paternity estimates from adult offspring were based on a further 599 individuals with a mean of 21 ± 2 per family.

Results

Testing the genetic incompatibility hypothesis

Females who mated with two different males had a significantly higher proportion of their eggs hatching than did females who mated twice with the same male (mean ± SE for polyandrous females 0.55 ± 0.02 vs. 0.48 ± 0.02 for monandrous females; t=2.45, d.f. 96, P=0.016; statistics performed on arcsin transformed proportions). There was no significant variation in measures of offspring performance between females mating with the same male and those mating with different males (monandrous vs. polyandrous: proportion surviving, 0.28 ± 0.02 vs. 0.27 ± 0.03, t91=0.37; P=0.71; development time, 76.9 ± 1.6 days vs. 76.6 ± 1.8 days, t89=0.155; P=0.88; adult pronotum width, 5.67 ± 0.04 mm vs. 5.56 ± 0.06 mm, t89=1.47; P=0.15, d.f. for latter comparisons were reduced because of low survival of some families). I examined embryo mortality using clutches of eggs derived from 13 non-experimental polyandrously mated females. Two days after being laid the eggs had begun to absorb water and swell. After 1 week embryos with clearly defined eye-spots were visible. All eggs that hatched had done so within 2 weeks. There was significant variation in the proportion of eggs reaching each stage: although 93.6 ± 2.8% of eggs took up water and began development, only 81.7 ± 4.5% reached the stage where eye-spots were visible and only 61.4 ± 5.5% hatched (repeated measures ANOVA on arcsin transformed proportions, between subjects F12,26=1.52; P=0.18, within subject F2,24=33.03; P < 0.001). The greatest mortality was between the eye-spot stage of development and hatching. Thus, consistent with previous studies the effects of polyandry on female fitness appear to be associated with prehatching embryo viability.

Males may exhibit intrinsic variability in their fertility so that the hatching success of polyandrous females reflects the average fertility of her two mates or the fertility of her most fertile mate. Following Tregenza & Wedell (1998) I looked for an association between the proportion of eggs hatching for females mated to two males and the mean proportion of eggs hatching for females mated monandrously to these same males. There was no correlation between the average hatching success of two monandrous females and the hatching success of polyandrous females mated to the same males (r25=0.052, P=0.79) or the hatching success of the monandrous female mated to the most fertile male and that of the polyandrous female (r25=0.138, P=0.49). Thus increased hatching success of polyandrous females does not result from the superior fertility of one of the males.

Genetic incompatibility can arise because of selfish elements or cytoplasmic endosymbionts that disturb optimal sex ratios (Zeh & Zeh, 1996). Typically these have their effect via the killing of sperm or embryos that would normally give rise to males. Thus, if the avoidance of this form of genetic incompatibility were responsible for the increased hatching success of eggs in this study, we would expect to see variation in the adult sex ratios of families produced by monandrous and polyandrous females. However, the sex ratio did not differ between mating protocols (polyandrous females produced 47 ± 2% male offspring compared with 48 ± 2% male offspring for monandrous females; t86=0.46, P=0.65). Overall, the offspring sex ratio did not deviate from 1:1 (t87=1.41, P=0.16).

Finally, the genetic incompatibility hypothesis specifically predicts that females mate polyandrously so that they can bias paternity, either via the incitement of sperm competition or through mechanisms of cryptic female choice, toward individual males who are genetically compatible (Zeh & Zeh, 1997; Yasui, 1998). There was no variation in the hatching success of polyandrous females that could be attributed to the number of alleles their two mates shared at the β-est locus (Table 1). Neither was there any variation in hatching success that could be attributed to the number of alleles monandrous females shared in common with their single mate (Table 1). Thus, there is no evidence for incompatibility at the β-est locus making it a reliable marker for assigning paternity in this study. I calculated paternity skew for polyandrous females as ∑Pi2 (after Starr, 1984) where Pi is the proportion of offspring sired by the ith male (paternity skew for doubly mated females ranges from 0.5 when both males share equally in parentage, to 1.0 were one male sires all offspring). If females increase hatching success via the avoidance of fertilizations with an incompatible male, there should be a positive association between hatching success and paternity skew. However, there was no such relationship (r15=0.166, P=0.52).

Table 1.   Proportion of eggs hatching in relation to the relative genotypes at the β-est locus of a monandrous female and her single mate, and the relative genotypes of two males mating with a polyandrous female. Thumbnail image of

Testing the good-sperm hypothesis

Variation in the outcome of sperm competition is depicted in Fig. 1, both for the situation when females mated with two males and when they mated with four males. The basic pattern of sperm utilization was one of sperm mixing. When two males competed the second male sired an average proportion of 0.46 ± 0.06 offspring, which did not differ from a random expectation under sperm mixing of 0.50 (t16=0.66, P=0.52). Nevertheless, there was high variation in the outcome of sperm competition (sample size corrected CV=53%) (Fig. 1a). When a male competed with three other males, the average proportionate fertilization success did not differ from an expectation of 0.25 (0.30 ± 0.04; t31=1.21, P=0.24). Nevertheless, an examination of the distribution of paternity values in Fig. 1b shows that 31% of males obtained either no fertilizations (n=2) or very low proportions of fertilizations (<0.1, n=8), whereas 9% of males gained almost complete paternity despite females having mated with three other males. The variation in paternity was greater when four males competed (sample size corrected CV=82%; a formal statistical comparison is not appropriate because the logarithms of the data were not normally distributed, Zar, 1984). There was no significant variation in paternity associated with a male’s position (male 1–4) in the female’s mating sequence (r30=0.044, P=0.81).

Figure 1.

 Variation in paternity outcome for Teleogryllus oceanicus. In (A) females mated with two males and values are the proportion of offspring sired by the second male to mate, or P2. In (B) females mated with four males and values are the proportion of offspring sired by the focal male.

The good-sperm hypothesis predicts that males who are successful in sperm competition convey superior heritable viability to their offspring. The association between success in sperm competition and embryo viability can be examined using data from doubly mated females. A measure of the relative fertilization success of two competing males is available from females that mated polyandrously. In addition, a measure of the hatching success of eggs fertilized by each of the two males in the absence of competition is available from those females that mated monandrously with them. If males successful in sperm competition sire embryos of higher pre-hatching viability we would expect a positive association between the difference in hatching success of eggs between females mated monandrously to two males, and the relative proportion of offspring sired when the males were in competition. The data in Fig. 2 fail to provide support for the good-sperm hypothesis. This analysis rests on the assumption that the observed proportion of offspring sired by each male estimated from 10-day-old offspring reflects none random fertilization, rather than random fertilization followed by differential embryo mortality. There is good evidence to support this assumption. If observed paternity variation resulted from random fertilization followed by differential mortality, females with high paternity skew should have lower hatching success, which was not the case (see above).

Figure 2.

 A comparison between the relative ability of two males to fertilize eggs that successfully hatch when mating twice with monandrous females and their abilities to sire offspring when in competition with each other within a polyandrous female. The data are presented as the hatching success of females mated to male 2 minus the hatching success of females mated to male 1, and the proportion of offspring sired by male 2 minus the proportion sired by male 1. There was no significant relationship (F1,15=0.48, P=0.50).

The predicted association between sperm competitive success and offspring viability can also be tested by comparing the relative post-hatching performance of offspring sired by each male, for both two male and four male sperm competition trials. First, if there were a correlation between sperm competitiveness and offspring viability, males who were successful in sperm competition should have a disproportionate number of offspring in the adult population because of their higher viability. Thus, a plot a paternity at offspring hatching vs. paternity when offspring reach adulthood should yield a slope that is significantly greater than one (H0 slope=1.0). However, for both two male (0.89 ± 0.11, t15=1.00, P=0.33) and four male (0.99 ± 0.12, t24=0.08, P=0.933) trials, a male’s paternity contribution to adult offspring did not deviate from that expected based on his paternity estimated when offspring hatched (Fig. 3).

Figure 3.

 The relationship between paternity estimated from the genotypes of offspring estimated 10 days after they hatch and at the time of adult eclosion for females mating with (A) two males or (B) four males. Both relationships are strong and significant (two males: F1,15=70.39, r2=0.91, P < 0.001; four males: F1,24=68.28, r2=0.86, P < 0.001). The slopes do not deviate from an expectation of 1.0 so that the representation of adult offspring in the population is in proportion to the fertilization success of their fathers (see text).

Finally, I examined the relative egg-to-adult development time and realized adult body size of offspring in relation to their fathers success in sperm competition, measured at offspring hatching. For each individual adult offspring I calculated the deviation from the family mean development time or body size and then calculated average values for offspring sired by the second male in two male sperm competition trials, or the focal male in four male sperm competition trials. If success in sperm competition were associated with offspring performance, offspring sired by successful males should have negative deviations from the family mean development time (faster development) and positive deviations from the family mean pronotum width (larger adult offspring). The reverse should be true for males unsuccessful in sperm competition. However, there was no significant relationship between a males success in sperm competition and the development speed of his offspring (Fig. 4) or their realized adult size (Fig. 5). Thus, there is no support for a correlation between sperm competitiveness and offspring viability or performance.

Figure 4.

 A test of the relationship between a male’s success in sperm competition and the egg-to-adult development time of his offspring relative to those sired by alternative males (A) for sperm competition trials involving two males (r11=–0.273, P=0.37), and (B) for sperm competition trials involving four males (r19=0.173, P=0.45) (see text for more details).

Figure 5.

 A test of the relationship between a male’s success in sperm competition and the realized adult size (pronotum width) of his offspring relative to those sired by alternative males (A) for sperm competition trials involving two males (r11=0.466, P=0.11), and (B) for sperm competition trials involving four males (r19=0.003, P=0.99) (see text for more details).

Discussion

Consistent with previous experimental manipulations of polyandry (Tregenza & Wedell, 1998; Newcomer et al., 1999), I found that female T. oceanicus had an increased proportion of viable eggs when mating with different males compared with females mating the same number of times but with a single male.

Genetic incompatibility?

Arnqvist & Nilsson’s (2000) meta analysis of the insect literature revealed a general effect of polyandry on the hatching success of eggs. The effect in part may arise if sperm stores become depleted with continued bouts of egg deposition (Ridley, 1988). Nevertheless, Arnqvist & Nilsson (2000) recognize that genetic incompatibility between mates can contribute to this general effect and that experimental manipulations are required to disentangle direct benefits from potential genetic benefits. In this and previous experimental studies of polyandry, the amount of ejaculate received was controlled across mating treatments by providing females with an equal number of matings. It is of course possible that males show variation in their ability to fertilize eggs, through variation in sperm numbers within ejaculates and/or fertilization capacity. Polyandry could thus bestow direct benefits by guarding against male infertility. If this were the case, providing females with equal numbers of matings from a single male would not adequately control for variation in the quality/quantity of sperm received. Nevertheless, experimental studies of polyandry show that males do not exhibit intrinsic variation in their ability to induce high hatching success (Tregenza & Wedell, 1998; Newcomer et al., 1999; this study). Potential variation in genetic compatibility has thus become a default explanation for observed benefits of polyandry (Tregenza & Wedell, 2000).

Yasui (1998) notes that if females are to benefit from polyandry by avoiding genetic incompatibility, they must have a mechanism whereby they can bias fertilization towards compatible males. If this were not so, polyandrous females would have a certain proportion of their offspring inviable while the same proportion of monandrous females would have all their offspring inviable. On average there would be no net fitness advantage to polyandry. The data for T. oceanicus show that females did not increase the hatching success of their eggs by skewing paternity toward one of their mates, a result that is inconsistent with the genetic incompatibility hypothesis for the evolution of polyandry.

Alternative explanations

Gonadotropic substances are incorporated into the ejaculates of Teleogryllus and other field crickets and exert their influence on female reproduction via an increase in the numbers of eggs produced (Stanley-Samuelson et al., 1986; Simmons, 1988). Although such substances can be detrimental to female lifespan, in general the direct benefits associated with increased offspring production more than outweigh reduced longevity, at least for moderate levels of polyandry (Arnqvist & Nilsson, 2000). Although dismissed in previous studies, it is at least possible that direct benefits represent an alternative explanation for viability differences between offspring of polyandrous and monandrous females. It seems reasonable to assume that prehatching embryo viability is dependent on the amount of resources laid down in the egg during its development. Gonadotropic substances are known to influence vitellogenisis and oviposition (Stanley-Samuelson & Loher, 1986). If males vary in the composition or potency of their gonadotropic compounds this could lead to variation in the quality of eggs produced by females and thus the viability of those eggs. Seminal fluid constituents can vary greatly across ejaculates produced by the same male, and can vary independently of sperm numbers (Cook & Wedell, 1996; Wedell & Cook, 1999; see review in Simmons, 2001).

Tregenza & Wedell (1998) considered and rejected the hypothesis that variation in seminal fluid products could explain increased hatching success in their study of G. bimaculatus because females mated twice with a single male did not have a lower hatching success than females mated four times with a single male. Thus, quantity of ejaculate substances had no apparent effect. First, in contrast with Tregenza and Wedell’s study, using the same population of G. bimaculatusSimmons (1988) found that the hatching success of females mated with a single male increased linearly as the number of matings with that male increased from a single mating, to two matings, to 4–6 matings. Thus the quantity of ejaculate obtained from a single male did influence hatching success. Moreover, if males show intra-ejaculate variation in the quality of seminal products, this could have an effect on egg quality and embryo viability independent of the quantity of seminal products received. Thus, polyandrous mating could reduce the variance in seminal products acquired by females and thus contribute directly to fitness.

It is also a possibility that increased embryo viability represents a maternal effect. Simmons (1987a) found that female G. bimaculatus allowed to choose their mates invested differentially in offspring production compared with those allocated a single male; females allowed to choose amongst 10 potential mates laid more eggs and a greater proportion of their eggs than did females allocated a single male. Differential reproductive investment in eggs after mating with high quality males has been demonstrated in butterflies (Wedell, 1996) and recently work with birds has found that differential investment in eggs following matings with attractive males can increase offspring condition (Cunningham & Russell, 2000). Thus, it is at least possible that experimentally induced monandry in otherwise choosy females results in the withholding of investment in eggs until females have had the opportunity to sample multiple males.

The good-sperm hypothesis

It has been proposed that females could influence paternity and thereby gain genetic benefits for their offspring, via polyandry and the incitement of sperm competition. Under the good-sperm hypothesis, sperm that are successful in sperm competition give rise to offspring of superior viability and fitness (Yasui, 1997). This hypothesis is problematic. As Parker (1992) notes, the same deleterious mutations that affect aspects of the individuals total phenotype would have to affect the individuals gametes. Whether gamete phenotype is controlled by its haploid genotype or that of its diploid parent is controversial (Erickson, 1990). Studies of Drosophila provide little evidence for haploid expression (Erickson, 1990). Nevertheless, it could be that males of high viability are able to produce greater numbers of sperm and/or seminal fluid products that enhance fertilization success, thereby indirectly generating a relationship between success in sperm competition and offspring viability (Parker, 1992; Yasui, 1997). Whatever the proximate mechanism however, this study provides no support for the good-sperm hypothesis; offspring viability and performance were not associated with their father’s success in sperm competition. This is the first attempt to empirically test the good-sperm hypothesis and more experiments are needed before a general assessment of its utility can be made. This study provides a protocol by which such work can proceed.

General implications for cryptic female choice

Eberhard (1996) used work on field crickets to illustrate a potential mechanism for cryptic female choice. One gonadotropic substance present within the seminal fluid of crickets increases the rate of contraction of the muscles in the region of the spermathecal duct (Kimura et al., 1988). Further, muscular contractions of the spermathecal duct are required to transport sperm from the spermatheca to the site of fertilization (Sugagawa, 1993). Eberhard (1996) thus suggested that female crickets may respond differentially during the processes of sperm storage and use, thereby biasing paternity towards certain males.

The end product of all proposed mechanisms of cryptic female choice is variation in the paternity of independent offspring (Eberhard, 1996). This study found considerable variation in paternity outcome when females mated with two or four males. However, this variation was not associated with offspring fitness. The data presented here therefore provide little support for the notion that females can gain genetic benefits for their offspring via cryptic female choice. That is not to say that sperm competition cannot yield indirect genetic benefits for females. It may be that fitness is expressed in the adult male offspring via their own success in sperm competition. Thus, a ‘sexy-sperm’ hypothesis may still prove to provide fitness benefits that contribute to the evolution of polyandry in crickets (Keller & Reeve, 1995). Such a process would not require the evolution of any direct mechanism of cryptic female choice as envisaged by Eberhard (1996), only that females mate multiply.

Patterns of sperm utilization

The patterns of sperm utilization reported here are interesting from the perspective of sperm competition mechanisms. Typically sperm competition experiments involve two males and conclusions concerning the outcome of sperm competition in natural populations are based on this often artificial mating pattern. In reality, many more than two males can be in competition and Zeh & Zeh (1994) found that the basic pattern of sperm competition can alter, depending on how many males are competing; in pseudoscorpions last male sperm precedence breaks down when females mate with a third male. This study shows that the basic pattern of sperm mixing typical of crickets (Backus & Cade, 1986; Simmons, 1987b; Gregory & Howard, 1994; Sakaluk & Eggert, 1996) persists at least for four matings. A recent review suggests that in general, patterns of sperm utilization in insects do not vary with the numbers of males involved (Simmons, 2001).

In conclusion, female T. oceanicus clearly benefit from polyandry through the increased viability of their eggs. There is little evidence to suggest that this arises because of the avoidance of genetic incompatibility. Neither is there evidence to suggest that females can gain viability benefits for their offspring via the incitement of sperm competition and a good-sperm process. The observed elevations in embryo survival reported in this and previous studies of polyandry are intriguing, but the mechanism behind this effect requires further study.

Acknowledgments

I thank Kylie Shau-Gaull, Oliver Berry and Julie Wernham for their help with this study and Mike Johnson for generously providing laboratory space and equipment. Tom Tregenza and Nina Wedell provided valuable discussion and comments on an earlier draft. This research was supported by the Australian Research Council.

Ancillary