A growing number of studies highlight the nontransitive properties of ejaculates when they are in competition to fertilize a female's eggs. Increasingly, these studies suggest that postcopulatory processes act as a filter against sperm from closely related males or those with similar genotypes, limiting the deleterious effects of inbreeding on offspring fitness. We investigated the potential for such postcopulatory mechanisms of inbreeding avoidance in the guppy (Poecilia reticulata), a promiscuous livebearing fish. We used artificial insemination as a method of delivering to a female the combined ejaculates from a first cousin (relatedness coefficient r= 0.125) and an unrelated male. This method of sperm delivery controls behavioral processes of pre- and postcopulatory female choice, which can bias paternity toward unrelated males. Our genetic analysis revealed no effect of parental relatedness on paternity outcomes. The observed mean paternity share for related males (0.47) and associated variance did not differ significantly from an expected binomial distribution that assumes no biased use of sperm with respect to relatedness (0.5). Although our data provide no evidence for postcopulatory mechanisms of inbreeding avoidance, the ability of female guppies to influence ejaculate transfer and retention offers an alternative and easily testable mechanism of inbreeding avoidance in this species.

Sperm competition and cryptic female choice occur when ejaculates from different males temporally overlap at the site of fertilization (Birkhead and Pizzari 2002). On one hand, sperm competition describes the contest among ejaculates from different males to fertilize a female's eggs (Parker 1970), whereas on the other, cryptic female choice describes postmating, female-mediated processes that bias paternity toward a subset of mated males (Thornhill 1983; Eberhard 1996). Although either process can consistently favor high quality (i.e., highly competitive or phenotypically attractive) males (Radwan 1998; House and Simmons 2005; Evans and Rutstein 2008), an increasing number of studies highlight the nontransitive properties of ejaculates when they are in competition to fertilize eggs (e.g., Clark et al. 1999; Birkhead et al. 2004).

A number of studies have provided evidence that genetic relatedness or genetic similarity between mating partners influences competitive fertilization success (Olsson et al. 1996; Wilson et al. 1997; Stockley 1999; Kraaijeveld-Smit et al. 2002; Mack et al. 2002; Bretman et al. 2004; Thuman and Griffith 2005; Jehle et al. 2007). Although these effects are by no means universal (e.g., see Stockley 1997; Jennions et al. 2004; Denk et al. 2005; Lane et al. 2007), the accumulating evidence that paternity success can be skewed toward unrelated or genetically dissimilar males has been taken as evidence that female multiple mating (polyandry) facilitates postcopulatory mechanisms of inbreeding avoidance (e.g., mediated by sperm-egg interactions or differential zygote mortality). Nevertheless, the majority of theses studies cannot discount precopulatory female choice or behavioral modes of postcopulatory choice (e.g., sperm ejection), which can also favor unrelated males in the competition to fertilize eggs. For example, in some species females are able to distinguish among potential mates according to kinship or genetic similarity (e.g., Roberts and Gosling 2003), and paternity skews toward unrelated males may be mediated by processes such as differential sperm retention or rejection by females (Pizzari et al. 2004). Indeed, to our knowledge, the only study to experimentally control such processes through the use of artificial insemination failed to detect an effect of mate relatedness on competitive fertilization success (paternity) in the mallard Anas platyrhynchos (Denk et al. 2005). By contrast, other studies reporting associations between competitive fertilization success and the genetic similarity of parents have invoked mate choice as a mechanism of inbreeding avoidance (e.g., Blomqvist et al. 2002), without considering, or excluding, the possibility that such results are also consistent with postcopulatory processes (see Griffith and Montgomerie 2003; Griffith 2007).

The guppy (Poecilia reticulata) offers a highly tractable study organism for disentangling precopulatory mechanisms of inbreeding avoidance from postcopulatory processes and uncovering the potential effects of genetic relatedness on competitive fertilization success. In this livebearing poeciliid fish, females are highly promiscuous (Evans and Magurran 2000; Pitcher et al. 2003) and levels of concurrent multiple paternity are among the highest reported in the family (Hain and Neff 2007). Furthermore, methods for controlled sperm delivery through artificial insemination have been developed in guppies and used in competitive fertilization trials (Evans et al. 2003; Evans and Rutstein 2008). Importantly, many fitness components in guppies decline with inbreeding, including juvenile survival (Nakadate et al. 2003), the onset of sexual maturity (Pitcher et al. 2008), reproductive rate (Zajitschek 2008), male courtship (van Oosterhout et al. 2003; Mariette et al. 2006), coloration (van Oosterhout et al. 2003; but see Zajitschek 2008), sperm numbers, and fertilization success (Zajitschek 2008). Although juvenile guppies have been shown to distinguish kin from nonkin through both phenotype matching and familiarity (Hain and Neff 2007), recent behavioral studies suggest that adult female guppies do not base their choice of mates on the basis of kinship (Viken et al. 2006; Pitcher et al. 2008). Nevertheless, the strong preference of female guppies to mate with unfamiliar males and males with novel or rare color patterns (Farr 1977; Hughes et al. 1999; Eakley and Houde 2004; Zajitschek et al. 2006) is thought to favor inbreeding avoidance (e.g., Brooks 2002). Despite this potential for precopulatory inbreeding avoidance, however, outside their sexually receptive phase (ca. 2d after parturition) female guppies are typically subject to high levels of forced mating attempts, approaching one attempt per minute in some populations (Magurran and Seghers 1994). Any precopulatory discrimination by females, which favors unrelated males during mate choice may therefore be compromised if postcopulatory mechanisms of inbreeding avoidance are absent (these mechanisms are reviewed in animals and plants by Zeh and Zeh 1997; Tregenza and Wedell 2000; Bernasconi et al. 2004).

We evaluated the potential for postcopulatory inbreeding avoidance in guppies using artificial insemination to simultaneously inseminate equal numbers of sperm bundles from related (first cousin) and unrelated males into sexually receptive virgin females. The coefficient of relatedness for related males in our experiment (r= 0.125) is at the higher end of the range of pairwise relatedness values found within pools in natural Trinidadian populations (mean of all pairwise relatedness values = 0.099; Hain and Neff 2007) and is, therefore, the kind of value for relatedness between the more closely related male–female pairs that might encounter one another in nature. Moreover, inbreeding depression is well documented at similar levels of parental relatedness in natural populations (cases reviewed by Hartl and Clark 1989; Lynch and Walsh 1998) and it has been argued that postcopulatory processes of inbreeding avoidance can be extremely subtle (e.g., Kempenaers 2007).

Our method of artificial insemination has the advantage over natural matings of controlling male sperm allocation (i.e., the relative contribution of sperm from rival males is held constant), mating order, and selective sperm retention by females, all of which are thought to influence competitive fertilization success in guppies (Evans and Magurran 2001; Pilastro et al. 2004). Our study therefore explores the potential for physiological postcopulatory processes (e.g., gamete interactions, sperm-female reproductive tract interactions and preparturition differential embryo survival) in influencing the relative paternity success of genetically related and unrelated males (see also Denk et al. 2005).



All animals used in this experiment were from the F2 generation of a breeding design that was initiated from 20 parental generation male–female pairs. All parental generation animals were first-generation laboratory born descendents of wild-caught fish collected from Alligator Creek (30 km south of Townsville) in Queensland, Australia. Recent genetic analyses indicate that this population was founded ca. 1910 from a source of wild guppies from Guyana (Lindholm et al. 2005).

We mated two to four F1 sons and daughters from each of the 20 parental generation pairs to randomly allocated F1 mates descended from other pairs. From the resulting F2 generation, we were then able to identify male–female pairs that were first cousins (r= 0.125), and a second male that was unrelated to both the male and the female in this pair. This kind of triplet is the unit of replication in our study, and we generated 20 independent triplets. Where there were same-sex full-siblings available for all three animals in a given triplet, we repeated the triplet to insure against any failed artificial inseminations. Where multiple repeats of a triplet produced progeny, we pool the data to avoid problems of pseudoreplication.


Male guppies produce sperm packaged in bundles (spermatozeugmata). In each trial, ejaculates were stripped from the two males (Matthews et al. 1997) and equal numbers of spermatozeugmata (containing approximately equal numbers of sperm—see Evans et al. 2003) were artificially inseminated into a six-month-old virgin female using a machine-pulled micropipette following the methods in Evans et al. (2003). Briefly, in each trial 10 spermatozeugmata (containing approximately 2.7 × 105 sperm cells) were obtained from each of the two males, gently mixed in a physiological saline solution (0.9% NaCl) and inseminated simultaneously into the anesthetized female using a machine-pulled micropipette (penetration depth approximately 2 mm). The number of sperm from each male was within the natural range of ejaculate size in this species (Pilastro and Bisazza 1999). Following each artificial insemination, a small clipping was taken from each male's caudal fin to provide tissue for subsequent DNA extractions. For their part, females were revived immediately after insemination and isolated individually in 5-L containers until they produced their first brood. After giving birth, a small clipping was taken from each female's caudal fin for the paternity analysis (see below). Offspring were too small to survive fin clipping and were therefore killed in a lethal dose of the anaesthetic MS222 for tissue extraction from whole bodies. Broods comprising fewer than three offspring were excluded from the paternity analyses. In total, n= 18 families were analyzed (mean number of offspring per brood = 16.4, SD = 9.4; range 3–41).


Five microsatellite markers were used to assign paternity: TTA (GenBank accession number AF164205; JS Taylor unpubl. cited in Lindholm et al. 2005), Pret46 (Watanabe et al. 2003), Pr39, Pr 80 (Becher et al. 2002), and Pooc G10 (Parker et al. 1998). PCR products were labeled with fluorescent markers and run on an Applied Biosystems 3730 DNA Analyzer (Applied Biosystems, Foster City, CA). Paternity was assigned to offspring according to allele sharing between the two putative sires, mother and offspring. The five loci were moderately variable, ranging from 3 to 9 alleles in our sample. Paternity assignment (where scored) was unequivocal because each individual offspring could only have been sired by one of the two males in the triplet. Of the total 296 offspring produced in this experiment, across 18 broods, one of the males could be unambiguously eliminated as sire for 253 offspring (85.5%) based on shared unique alleles (within the triplet) between the sire and offspring. In the remaining cases, the genotypes of the two males in the triplet were not sufficiently different from one another at these five loci to permit unequivocal scoring of paternity.


The proportion of offspring sired by the related male (termed Prelated) was estimated from his paternity share. We used randomization tests to compare the observed mean paternity share of related males (Prelated) to the null expectation of 0.5, which assumes no biased use of sperm. Randomization tests were conducted using the Poptools (version 2.7) add-in ( (Hood 2006) to Microsoft Excel (2007). For all tests null expectations were generated for the 18 families by estimating the critical value for the appropriate binomial distribution with the relevant properties (e.g., 0.5 paternity share of the male who was related to the dam—i.e., Prelated) for a probability given by a random number. We then estimated the experiment-wide properties for each set of psuedoestimates (e.g., mean Prelated) and resampled this property 10,000 times to obtain an expected distribution of the property.


Our results revealed no evidence that fertilization success was skewed toward unrelated males. The mean paternity share of the male that was related to the dam (Prelated) was 0.47 (SE = 0.06), which does not differ from a null (0.5) expectation (randomization test, P= 0.35, see Fig. 1). This result remained not significantly different from the null expectation when all unassigned offspring were assigned to either the unrelated (P= 0.3) or related (P= 0.14) male. This indicates that related males were systematically neither more nor less successful than unrelated males in siring offspring. Furthermore, the variance in the paternity distribution (0.07) did not differ significantly from the null expectation under a binomial distribution with mean of 0.5 (null mean, P= 0.11) or 0.47 (observed mean, P= 0.12). This indicates that there is no strong signature of biased paternity within broods.

Figure 1.

Observed and expected distribution of paternity by the related male (Prelated, proportion of offspring in each brood sired by the related male). (A) Observed number of families and (B) Mean expected number of families from 1000 randomizations based on observed family sizes and a binomial distribution in which probability of paternity by related male of each offspring was 0.5.


Following artificial insemination by a first cousin and an unrelated male we found no evidence that competitive fertilization success was skewed toward genetically unrelated male guppies. Although this finding contrasts with a number of studies that report fertilization biases in favor of unrelated or genetically dissimilar males (Olsson et al. 1996; Wilson et al. 1997; Stockley 1999; Kraaijeveld-Smit et al. 2002; Mack et al. 2002; Bretman et al. 2004; Thuman and Griffith 2005; Jehle et al. 2007), it accords with several others that have similarly failed to reveal evidence for postcopulatory inbreeding avoidance mechanisms (crickets: Jennions et al. 2004; squirrels: Lane et al. 2007; shrews: Stockley 1997).

Like Denk et al.'s (2005) work on mallards, our experimental approach to preclude precopulatory mate choice and behavioral processes of postcopulatory selection revealed no explicit evidence for postcopulatory processes of inbreeding avoidance, at least in the current focal population and under the degree of parental relatedness imposed in our study. Unlike Denk et al. (2005), however we did not evaluate how the relative quality of competing ejaculates influenced sperm competitiveness. Denk et al. (2005) found that both long-term sperm swimming velocity (measured > 8 min of sperm being active) and motility (the proportion of motile cells in a sample) predicted competitive fertilization success, as estimated by assigning paternity to the resultant broods. Similar results linking relative sperm performance traits to competitive fertilization success have been reported in other taxa (e.g., fowl: Birkhead et al. 1999; Froman et al. 2002; fish: Gage et al. 2004; insects: García-González and Simmons 2005). Thus, it is possible that intrinsic differences in the competitive ability of competing male ejaculates would have masked any subtle forms of postcopulatory sperm choice based on relatedness. Nevertheless, in guppies, postcopula processes other than those due to such intrinsic male effects are thought to constitute important sources of variance in male reproductive success. For example, Pilastro et al. (2004) found that females differentially handle ejaculates at mating according to their perception of male attractiveness, so that higher numbers of sperm are retained following copulations with highly ornamented males. This effect was independent of male effects (e.g., strategic sperm investment), which were vigorously controlled in the experiment. The domestic fowl Gallus gallus domesticus, similarly exhibits cryptic female preferences mediated by differential handling of sperm from preferred and nonpreferred males (Pizzari and Birkhead 2000), and females of its relative the red junglefowl retain fewer sperm following copulations with brothers compared to nonsib matings (Pizzari et al. 2004). If similar active modes of postcopulatory inbreeding avoidance operate in guppies, sperm choice may be unnecessary. Thus, although female guppies apparently do not discriminate among potential mates on the basis of kinship (Viken et al. 2006; Zajitschek 2008), their ability to assess relatedness through familiarity and phenotype matching (Hain and Neff 2007) may enable them to differentially retain sperm from related or unrelated males, a possibility that clearly warrants further investigation.

Like other studies that have estimated competitive fertilization rates from paternity data in a similar context (e.g., Wilson et al. 1997; Stockley 1999; Kraaijeveld-Smit et al. 2002; Mack et al. 2002; Bretman et al. 2004; Thuman and Griffith 2005; Jehle et al. 2007), we attempted to sample all of the offspring produced by each female for our genetic analyses. Nevertheless, we cannot rule out the possibility that differential embryo viability (or indeed differential postparturition offspring survival) accounted for some of the variance in paternity. For example, it is possible that some zygotes failed to develop or were reabsorbed by the female during gestation, or that some offspring did not survive after parturition (e.g., through filial cannibalism, which is known in fish; DeWoody et al. 2001). To the extent that these or other postzygotic processes selectively target embryos from related parental crosses, paternity distributions may not accurately reflect those at fertilization. Indeed, the inequality between competitive fertilization success and paternity outcomes, arising from differential (postzygotic) selective processes, presents a persistent problem for those working on sperm competition in internal fertilizers (see García-González 2008). Nevertheless, it is unlikely that these effects would have changed our conclusions, as any biased mortality toward related parental crosses would have shifted the distribution of paternity in favor of unrelated males (i.e., this would have increased the likelihood of us detecting an effect). Furthermore, selective mortality arising from consanguineous fertilizations would result in smaller broods compared to unrelated parental crosses. To test for such an effect, we compared the mean brood size arising from the current investigation (9.31 ± 0.74 SE) with those from a previous experiment involving double matings with nonrelatives (7.90 ± 0.97 in Evans and Magurran 2001). We found no significant difference in brood size between these groups (t-test: t53= 1.17, P= 0.247). Thus, although differential mortality is unlikely to have changed our conclusions, such effects may constitute important sources of variance in paternity in previous studies that have invoked differential fertilization success as a mechanism of inbreeding avoidance (see discussion by Olsson et al. 1999).

Although we found no evidence for differential fertilization success with respect to parental relatedness, it is possible that such effects may occur in different contexts or under different levels of parental relatedness. A particularly clear case in which such selection is thought to occur is where mating occurs in a contact zone between two sympatric, but morphologically similar species (e.g., Veen et al. 2001). It is possible that in our focal guppy population, selection favoring postcopulatory mechanisms of inbreeding avoidance is relatively weak, or that levels of relatedness were not sufficiently high to have triggered such mechanisms. Inbreeding depression and avoidance are by no means ubiquitous, and when inbreeding depression is minimal it has been suggested that both sexes may actually raise their inclusive fitness by mating with relatives (Parker 1979; Lehmann and Perrin 2003; Kokko and Ots 2006). Furthermore, in Japanese quail both sexes prefer first cousins as mates over unrelated or full-siblings (Bateson 1982), possibly reflecting optimal outbreeding by both males and females (Bateson 1983). However, we found no evidence for a bias in paternity toward first cousins, and in fact guppies are generally susceptible to strong inbreeding depression (Shikano and Taniguchi 2002; Nakadate et al. 2003; van Oosterhout et al. 2003). Indeed, our own work on this population has shown inbreeding depression for traits such as courtship behavior (Mariette et al. 2006), fertility, and juvenile mortality (Zajitschek 2008; R. Brooks and E. Postma, unpubl. data). Furthermore, inbreeding depression can be manifested at similar levels to those imposed here (Hartl and Clark 1989; Lynch and Walsh 1998), and claims regarding biased paternity toward unrelated males have been made at comparable levels of relatedness between mates (Blomqvist et al. 2002; but see Griffith and Montgomerie 2003). Nevertheless, we advocate further studies that investigate the effect of increasing levels of parental relatedness on competitive fertilization success (e.g., Mack et al. 2002), as well as those focusing on other potential mechanisms of inbreeding avoidance (e.g., differential sperm handling by females).

Associate Editor: T. Chapman


We thank A. Rutstein for assistance with the paternity analyses, and T. Chapman, T. Pizzari, F. García-González, P. Stockley, and an anonymous referee for comments that helped improve the manuscript. This research was supported by grants and fellowships to JPE, RCB and SCG from the Australian Research Council.