Susanne Zajitschek, Station d’Ecologie Expérimentale du CNRS a Moulis, USR 2936, 09200 Moulis, France. Tel.: +33 5 61 04 03 60; fax: +33 5 61 96 08 51; e-mail: firstname.lastname@example.org
The effects of inbreeding on sperm quantity and quality are among the most dramatic examples of inbreeding depression. The extent to which inbreeding depression results in decreased fertilization success of a male’s sperm, however, remains largely unknown. This task is made more difficult by the fact that other factors, such as cryptic female choice, male sperm allocation and mating order, can also drive patterns of paternity. Here, we use artificial insemination to eliminate these extraneous sources of variation and to measure the effects of inbreeding on the competitiveness of a male’s sperm. We simultaneously inseminated female guppies (Poecilia reticulata) with equal amounts of sperm from an outbred (f =0) male and either a highly (f =0.59) or a moderately inbred (f =0.25) male. Highly inbred males sired significantly fewer offspring than outbred males, but share of paternity did not differ between moderately inbred and outbred males. These findings therefore confirm that severe inbreeding can impair the competitiveness of sperm, but suggest that in the focal population inbreeding at order of a brother–sister mating does not reduce a male’s sperm competitiveness.
Although the above studies provide some evidence that inbreeding affects sperm quantity and quality, the extent to which inbreeding depression actually results in decreased fertilization success is largely unknown (but see Konior et al., 2005). For example, in Oldfield mice (Peromyscus polionotus), even though sperm count and testis mass declined with increasing inbreeding coefficient, fertility is not significantly influenced by sperm concentration (Margulis & Walsh, 2002). Thus, it is possible that fertilization success only starts to decline when sperm counts are extremely low. In humans, for example, a wide range of sperm counts are considered to be normal, and fertility is only negatively affected if sperm concentration is below a threshold of 40 × 106 mL−1 (range 0–428 × 106 mL−1, Bonde et al., 1998). However, under conditions of sperm competition, negative effects of low sperm quantity and quality could easily be amplified.
Although several studies have investigated the effects of sperm quality, either in terms of mobility (for example Birkhead et al., 1999), viability (Garcia-Gonzalez & Simmons, 2005), quantity or morphology on sperm competitiveness, or how inbreeding affects sperm traits per se (Roldan et al., 1998; Gomendio et al., 2000; van Eldik et al., 2006; Gage et al., 2006), only a few studies have looked at the effects of inbreeding on sperm competitive ability. For example, in bulb mites (Rhizoglyphus robini), inbred males (f =0.25) have lower sperm competition success than outbred males (Konior et al., 2005), whereas in Drosophila melanogaster inbred males from homozygous lines performed much worse in sperm competition than outbred males (heterozygous lines) (Hughes, 1997). Nevertheless, it is possible that mating order, female sperm selection or differences due to male sperm allocation may have influenced the outcome of sperm competition.
In this paper, we use artificial insemination to carefully control extraneous sources of variance influencing the outcome of sperm competition and ask whether inbred male guppies sire fewer offspring in direct sperm competition with outbred competitors. Guppies are livebearing fishes with a highly promiscuous mating system (Neff et al., 2008). Importantly, guppies are also known to suffer from inbreeding depression. For example, studies on inbred strains have revealed reduced survival and decreased salt tolerance (Nakadate et al., 2003), reduced male courtship behaviour (van Oosterhout et al., 2003; Mariette et al., 2006), a reduction in male ornamentation (Sheridan & Pomiankowski, 1997; van Oosterhout et al., 2003) and reductions in body size and fertility (van Oosterhout et al., 2007) compared to outbred guppies. Despite the importance of sperm competition in guppies, we know nothing about how inbreeding impacts on sperm competitiveness.
The use of artificial insemination makes it possible to test how inbreeding influences sperm competitiveness whilst controlling for potential biases in fertilization success due to mating order, male sperm allocation and cryptic female choice (Evans et al., 2003; Pizzari et al., 2004; Denk et al., 2005). Surprisingly, however, this approach has not been used to test the effect of inbreeding on the competitive ability of sperm. To test our prediction that inbreeding impairs sperm competitiveness, we artificially inseminated females with the sperm of two males, one inbred and one outbred. We included two levels of inbreeding in this study, as we could not predict a priori the level of inbreeding at which inbreeding depression would become apparent. Genetic consequences of inbreeding are determined by the genetic basis of the trait concerned, including the dominance variance. This is in turn influenced by the history of selection on the trait concerned and by population size, as slightly deleterious mutations are more likely to fix in small than in large populations, and by the history of inbreeding in the population (Falconer & Mackay, 1996; Frankham et al., 2002). Historical population sizes, and therefore the levels of ancestral inbreeding in a given population, have been shown to result in different magnitudes in the effects of inbreeding in guppies (van Oosterhout et al., 2007). On the one hand, a long history of inbreeding can lead to a build up of deleterious alleles and therefore to severe inbreeding depression. On the other hand, inbreeding depression will lead to the loss of deleterious alleles via natural selection. The remaining population might then express little inbreeding depression, despite (or rather because of) a long history of inbreeding, as parts of the deleterious genetic load may have been purged. Thus, we tested for differences in sperm performance in highly inbred (four generations, f =0.59) and moderately inbred (one generation, f =0.25) guppies, using animals from a known pedigree of brother–sister inbreeding, and report the effects of both long-term and recent inbreeding on fertilization success under sperm competition.
Materials and methods
The guppy population used for the experiments described in this study are descendents of animals captured from Alligator Creek, approximately 30 km south of Townsville in Northern Queensland, Australia. Guppies were introduced to Australia approximately 100 years ago, and the study population was most likely founded by animals from Guyana (Lindholm et al., 2005).
We simultaneously conducted two artificial insemination experiments, one to test the effect of a single inbreeding event and the second to test the effect of prolonged inbreeding (four generations of full-sib mating) on male sperm competitiveness. All animals used in this experiment were from a known five-generation pedigree of individuals that had been either repeatedly mated to full siblings, repeatedly outbred to unrelated individuals in the pedigree, or repeatedly outbred and then mated to a full sibling in the generation preceding this experiment (see Fig. 1).
In each experiment, we artificially inseminated females of known ancestry simultaneously with the sperm of an inbred male (coefficient of inbreeding f =0.25 and 0.59 in experiment one and two, respectively) and an outbred male (f =0). Each pair of males was selected to have age differences of less than 4 months to minimize potential age biases in fertilization success, and to ensure low coancestry with one another (coefficient of relatedness r <0.03, i.e. control and inbred males were not bred in the same patriline) to ensure maximum power to assign paternity.
Prior to this experiment, we conducted preliminary tests on third generation inbred (f =0.5) and outbred (f =0) guppies from the same pedigree to test whether sperm numbers vary between inbred and outbred males. Male guppies release sperm in bundles (spermatozeugmata) during copulation or through mechanical stimulation (Kuckuck & Greven, 1997). We did not find significant variance in sperm numbers among spermatozeugmata within ejaculates (repeated measures anova, Wilk’s Lambda = 0.95, F2,68 = 1.97, P =0.15), similar to previous findings (Kuckuck & Greven, 1997; Evans et al., 2003). Furthermore, we found no evidence for an effect of inbreeding on the number of sperm per bundle (inbred or outbred, F1,69 = 0.27, P =0.61). However, the total number of spermatozeugmata, and therefore total number of sperm, did differ between inbred and outbred animals (F1,131 = 23.62, P <0.05); outbred males had significantly more sperm than inbred males. We therefore used equal numbers of spermatozeugmata from inbred and outbred males for the artificial inseminations to ensure that rival males contributed equal numbers of sperm.
For each trial we stripped sperm from the two males in immediate succession. To avoid effects of stripping order (i.e. sperm left for a longer time in saline solution) we randomized the order in which males were stripped. To strip a male’s sperm, we anaesthetized him by submersion in ice slurry for a few seconds and placed him left side up on a black plastic slide with a small amount of saline solution (0.9% sodium chloride) on its surface. Sperm reserves were extracted under a microscope (SMZ-2T; Nikon, Tokyo, Japan) with 10× magnification, by gently brushing a blunt probe along his anterior abdomen, with his gonopodium swung forward (Mathews et al., 1997). We transferred 10 bundles together with approximately 5 μL of saline solution per male to each of four (0.5 mL) Eppendorf vials. Only the process of adding both solutions and then sucking the entire contents up again for insemination was used for mixing, as the sperm bundles are fragile structures which need to be intact to ensure fertilization (S. Zajitschek & J. Evans, unpublished data). Males were retained until the females had produced offspring and then killed and stored in 95% ethanol for molecular paternity analysis.
We then inseminated the contents of the three vials each into a different replicate female. Replicate females were inseminated with the sperm from each male pair to increase the probability that every sperm mix led to the production of at least one brood. Broods within females and among females within male pair did not differ in paternity and were subsequently combined.
Each female was anaesthetized and placed ventral side up in a small polystyrene cradle under a microscope (Nikon) at 10–20× magnification (Zajitschek et al., 2006). The gonopore was opened using a hand-pulled glass pipette. The contents of the appropriate Eppendorf vial (approximately 10 μL) were inserted into the female using a hand-pulled plastic pipette. The female was then transferred into a 2-L container with aerated water, containing 1 mL Armour Coat (Aristopet, Brisbane, QLD, Australia) until she recovered.
All females were closely monitored for the next 48 h and showed no ill effects from the experimental treatments. They were then transferred into individual 5 L tanks with a layer of multicoloured gravel covering the bottom and plastic plants as hiding space for newborn fry. Animals were fed fresh brine shrimp every day, and kept on a 12 h light/12 h dark cycle at 26 °C. Tanks were checked twice a day for newborn broods (expected from 3 weeks after the insemination), and all offspring were killed shortly after birth, and preserved in 95% ethanol for storage until paternity analysis. The females were left to produce successive broods for a maximum of 3 months after the birth of their first brood (Constantz, 1989). Females that did not produce broods (N = 101) were kept for a maximum of 4 months before being returned to the laboratory stock population. Three females were inseminated with each sperm mix, and in case of death or none of the females appearing pregnant, an additional female was inseminated 3 weeks later. Females were killed after having a second brood, or after not having offspring for more than 3 months after the birth of their first brood. Males were killed after successful inseminations (at least one female appearing pregnant) and kept in 95% ethanol for genetic analyses.
We extracted DNA from the whole bodies of offspring, and from approximately 2 mg wet tissue of each female and each candidate father, using the Puregene DNA Purification kit, following the Mouse tail tissue protocol (Gentra Systems, Minneapolis, MN, USA).
We used seven markers (Table 1) which have been found to be highly variable in wild guppy populations (Becher et al., 2002; Lindholm et al., 2005; Nater et al., 2008) in a multiplex touchdown PCR setup (PCR conditions: 94 °C for 10 min, 10 cycles of 94 °C for 45 s, 70 °C for 1 min and 72 °C for 1 min; 10 cycles of 94 °C for 45 s, 64 °C for 1 min and 72 °C for 1 min; 10 cycles of 94 °C for 45 s, 70 °C for 1 min and 58 °C for 1 min; 10 cycles of 94 °C for 45 s, 55 °C for 1 min and 72 °C for 1 min; and 10 cycles of 94 °C for 45 s, 50 °C for 1 min, and final elongation at 72 °C for 1 min), with the Quiagen Multiplex PCR kit. Alleles were visualized and scored using the software genemapper (version 3.0; Applied Biosytems, Foster City, CA, USA).
Table 1. Loci used in multiplex PCRs.
Alleles (f = 0)
Alleles (f = 0.25)
Alleles (f = 0.59)
High inbreeding resulted in decreased allelic variation in paternal genotypes.
Compared to wild populations, neutral genetic diversity in introduced Australian populations is very low, indicating population bottlenecks in the past (Lindholm et al., 2005). Nevertheless, in comparison to data from the same laboratory population (Lindholm et al., 2005), the allelic diversity in our samples turned out to be even lower than expected. This affected the scoring confidence in several cases (N = 118 offspring from 31 different females), in which paternity could only be assigned with 80% instead of 95% certainty (cervus 3.0; Field genetics Ltd, London, UK).
Quantifying inbreeding effects
To examine effects of inbreeding on paternity success, we used a randomization testing approach, implemented in the PopTools Add-in (Hood, 2006) for Microsoft Office Excel (Microsoft Corporation, Redmont, WA, USA). We calculated the proportion offspring sired by the outbred male (Poutbred) for each sire pair and the mean Poutbred for each experiment. Because of the differences in numbers of dams per sire pair that successfully produced broods, we pooled the offspring from all females for each sire pair. We then used randomization to generate a null distribution for mean Poutbred based on the expectation that paternity would be equally shared (i.e. that Poutbred = 0.5) within each sire pair (family), keeping the same array of family sizes. For each family within each randomization run we used the RND function in MS Excel to assign a random number (r) between 0 and 1 to each family. We then turned this number into a number of outbred offspring by taking the critical value of the binomial distribution for family size n, probability r, and a 0.5 binomial probability of any given offspring being sired by the outbred male using critbinom(n,r,0.5). Our test statistic for each psuedoreplicate of the simulation was the difference between the observed number of outbred offspring for the experiment and the simulated number. We used the Monte Carlo function in PopTools to conduct 10 000 randomization trials for each experiment. The number of simulations in which Poutbred was of equal or greater value than the observed value was interpreted as the probability that outbred males did indeed enjoy a paternity advantage in that experiment.
All methods used in this experiment were approved by the UNSW Animal Care and Ethics Committee (Approval number 04/66).
Our genetic analyses enabled us to assign paternity to 517 of 548 offspring (94.3%), with at least 95% confidence for 73% of offspring, and 80% confidence for 94.5% of offspring (see Table 2 for details on each experiment). Nevertheless, as small brood sizes might skew the picture of relative paternity, we excluded all broods with fewer than six offspring from the Monte Carlo analyses (i.e. three broods from experiment 1 and 21 broods from experiment 2, see also Table 2).
Table 2. Numbers of animals used in both experiments, and outcomes of paternity analyses.
The range of offspring numbers varied between 2 and 20 in experiment 1 (mean = 9.91 per female, SD = 5.96), in experiment 2 brood sizes ranged from 1 to 30 offspring (mean = 7.94, SD = 6.23). In experiment 2, seven families with 6–38% of their offspring unassigned were included in the analyses; we treated the unassigned offspring as nonexisting. Numbers reported in this table are based on all animals, including families and offspring that were not used in the analyses.
In experiment 1, in which sperm from highly inbred (focal) males were inseminated in equal numbers to those from outbred males, we found that inbred males were significantly less successful in gaining paternity than their outbred rivals (experiment 1: randomization test, P <0.05, see Fig. 2). By contrast, experiment 2 revealed no significant effect of inbreeding on sperm competitiveness when levels of inbreeding in focal males were reduced to f =0.25 (experiment 2, randomization test, P =0.22, Fig. 2).
Effects on family size
The total number of offspring per inbred/outbred sire pair was not related to the proportion of offspring sired by the outbred male, providing no support for the idea of beneficial effects of outbred sperm on brood sizes (regression, reduced data set: high inbreeding F1,5 = 0.02, P =0.89; low inbreeding F1,20 = 0.88, P =0.36; see Fig. 2, entire dataset including small broods: high inbreeding F1,6 = 0.95, P =0.37; low inbreeding F1,25 = 0.02, P =0.89). The mean number of offspring per dam did not significantly differ between the two experiments (Students t-test with unequal variances assumed: t64 = 1.17, P =0.25; see also Table 2), indicating that the low paternity share of highly inbred males in experiment 1 did not reduce mean brood size significantly.
We show that high levels of inbreeding reduce male fertilization success under sperm competition, but that moderate levels of inbreeding do not. In our high-inbreeding experiment, outbred males sired a significantly higher proportion of offspring than inbred males. In guppies, inbreeding has been shown to severely affect ornamentation (Sheridan & Pomiankowski, 1997; van Oosterhout et al., 2003), behaviour (van Oosterhout et al., 2003; Mariette et al., 2006) and salt tolerance (Nakadate et al., 2003), but little is known on the effects of inbreeding on sperm traits in this species (but see preliminary data on sperm count, and Pitcher et al., 2008). Inbreeding depression can have deleterious effects on fitness in wild populations, as the original habitat in Trinidad, Guyana and Venezuela as well as in feral populations worldwide comprise small pools within rivers and streams separated by riffles and waterfalls, which can get cut off from the main water bodies during the dry season (Magurran, 2005). This isolation can increase the probability of inbreeding within populations, and over time may have contributed to the purging of severely detrimental alleles, as indicated by the apparent lack of inbreeding depression found here and elsewhere (van Oosterhout et al., 2007) and may help explain the absence of inbreeding avoidance at low levels of inbreeding (Evans et al., 2008; Pitcher et al., 2008; Zajitschek & Brooks, 2008).
In the bulb mite R. robini (Konior et al., 2005) and in D. melanogaster (Hughes, 1997), when a female copulates with both an inbred and an outbred male, the inbred males suffer reduced paternity. However, in these studies female post-copulatory choice and the relative quantity of sperm inseminated by rival males may have influenced patterns of paternity. Our ability to experimentally control the relative amount of sperm inseminated from rival males whilst concomitantly limiting the opportunity for post-copulatory female choice suggests instead that the effects of inbreeding on paternity under sperm competition may be due, at least in part, to reduced sperm quality. Sperm quality parameters such as swimming velocity, percentage motility and viability (the proportion of live sperm in an ejaculate) can be key determinants of sperm competitiveness (Birkhead et al., 1999; Donoghue et al., 1999; Gage et al., 2004; Denk et al., 2005; Garcia-Gonzalez & Simmons, 2005) and are known to be susceptible to genetic stress through inbreeding (Roldan et al., 1998; Margulis & Walsh, 2002; Denk et al., 2005; Gage et al., 2006; Fitzpatrick & Evans, 2009). For example, inbreeding has been shown to have deleterious effects on both sperm viability and the percentage of motile sperm in several mammal species (Gomendio et al., 2000; van Eldik et al., 2006). Consistent with our findings, these two previous mammal studies revealed no inbreeding depression in sperm traits in animals with low (f <0.02) inbreeding coefficients (Gomendio et al., 2000; van Eldik et al., 2006). A similar reduction in sperm performance (e.g. motility or viability) under high, but not low to moderate, inbreeding could be responsible for the observed reduction in sperm competitiveness in our experiment 1, but not in experiment 2. Moreover, the fact that only high levels of inbreeding were found to affect sperm competitiveness might indicate that the population under investigation has purged the most detrimental mutational load due to historical inbreeding events (i.e. genetic bottleneck during the introduction to Australia, Lindholm et al., 2005), and that the observed reduction in fertility at high levels of inbreeding is caused by epistatic interactions of groups of mildly deleterious alleles, which are only expressed as multiple homozygotes.
Although the sperm traits that confer a competitive fertilization advantage have yet to be identified in guppies, the observation that both sperm swimming speed and the proportion of viable sperm in the ejaculate positively covary with male ornamentation (Locatello et al., 2006; Pitcher et al., 2007), which in turn predicts sperm competitiveness (Evans et al., 2003), suggests that these traits may be particularly susceptible to the effects of inbreeding. Under semi-natural conditions and in populations with known pedigrees, elevated inbreeding coefficients have often been associated with measurable fitness declines, whereas lower levels of inbreeding result in an apparent lack of any deleterious effects (Gomendio et al., 2000; van Eldik et al., 2006). Even though the degree of inbreeding depression varies widely in the wild (reviewed in Crnokrak & Roff, 1999; Keller & Waller, 2002), negative effects of inbreeding are manifest in small, fragmented populations, where inbreeding becomes chronic after several generations (Frankham et al., 2002).
In summary, we report the first direct experimental evidence that inbreeding influences paternity success exclusively through its detrimental effects on the competitive performance of sperm. The negative impacts of inbreeding on sperm quality have been investigated in a range of species (Roldan et al., 1998; Trouve et al., 1999; Gomendio et al., 2000; Margulis & Walsh, 2002; Gage et al., 2006), but even though sperm traits themselves can be sensitive to genetic stress at low levels of inbreeding, our findings suggest that these effects may not translate into measurable declines in reproductive fitness. This may explain why post-copulatory mechanisms of inbreeding avoidance at low levels of parental relatedness (r = 0.125) are absent in this population (Evans et al., 2008). Interestingly, we also found that inbreeding influences sperm production. According to lottery-based models of sperm competition (Parker, 1998), reduced numbers of sperm are expected to result in reduced paternity. As such, the detrimental effects of inbreeding on sperm competitiveness, through its effects on sperm quality, are likely to be further compounded by declines in sperm production, thus further aggravating the deleterious effects of inbreeding on male reproductive fitness.
Funding was provided by an Australian Research Council grant and fellowship to RB. SZ was supported by an EIPRS scholarship and an emergency grant from Joyce Vickery Scientific Research Fund, supplied from the Linnean Society of New South Wales. JPE was supported by an ARC Fellowship. We thank Jari Garbely for his assistance with the genetic analyses. Comments from Ulrika Candolin, Felix Zajitschek, Barabara Tschirren, Erik Postma, Gunilla Rosenqvist, Bob Wong, John Fitzpatrick and two anonymous reviewers greatly improved the manuscript.