Clonal diversity driven by parasitism in a freshwater snail

Authors

  • Y. Dagan,

    1. Department of Zoology, George S. Wise Faculty of Life Sciences, Tel Aviv University, Tel Aviv, Israel
    Search for more papers by this author
  • K. Liljeroos,

    1. EAWAG, Swiss Federal Institute of Aquatic Science and Technology, Dübendorf, Switzerland
    2. ETH-Zürich, Institute of Integrative Biology (IBZ), Zürich, Switzerland
    Search for more papers by this author
  • J. Jokela,

    1. EAWAG, Swiss Federal Institute of Aquatic Science and Technology, Dübendorf, Switzerland
    2. ETH-Zürich, Institute of Integrative Biology (IBZ), Zürich, Switzerland
    Search for more papers by this author
  • F. Ben-Ami

    Corresponding author
    1. Department of Zoology, George S. Wise Faculty of Life Sciences, Tel Aviv University, Tel Aviv, Israel
    • Correspondence: Frida Ben-Ami, Department of Zoology, George S. Wise Faculty of Life Sciences, Tel Aviv University, Tel Aviv 6997801, Israel.

      Tel.:+972 3 6406080; fax: +972 3 6405347; e-mail: frida@post.tau.ac.il

    Search for more papers by this author

Abstract

One explanation for the widespread abundance of sexual reproduction is the advantage that genetically diverse sexual lineages have under strong pressure from virulent coevolving parasites. Such parasites are believed to track common asexual host genotypes, resulting in negative frequency-dependent selection that counterbalances the population growth-rate advantage of asexuals in comparison with sexuals. In the face of genetically diverse asexual lineages, this advantage of sexual reproduction might be eroded, and instead sexual populations would be replaced by diverse assemblages of clonal lineages. We investigated whether parasite-mediated selection promotes clonal diversity in 22 natural populations of the freshwater snail Melanoides tuberculata. We found that infection prevalence explains the observed variation in the clonal diversity of M. tuberculata populations, whereas no such relationship was found between infection prevalence and male frequency. Clonal diversity and male frequency were independent of snail population density. Incorporating ecological factors such as presence/absence of fish, habitat geography and habitat type did not improve the predictive power of regression models. Approximately 11% of the clonal snail genotypes were shared among 2–4 populations, creating a web of 17 interconnected populations. Taken together, our study suggests that parasite-mediated selection coupled with host dispersal ecology promotes clonal diversity. This, in return, may erode the advantage of sexual reproduction in M. tuberculata populations.

Introduction

The widespread occurrence of sexual reproduction in eukaryotes is surprising, given the many disadvantages of sexual reproduction compared with asexual reproduction and self-fertilization. Outcrossing sexual individuals must invest limited resources in finding mates, and they risk sexually transmitted diseases. Compared with asexuals, sexual populations have a lower per-capita rate of reproduction due to the production of males (Maynard Smith, 1978). One of the most prominent hypotheses to explain the ubiquity of sex is the Red Queen hypothesis (Clarke, 1976; Jaenike, 1978; Hamilton, 1980; Bell, 1982). It postulates that parasites rapidly evolve to disproportionately infect the most common host genotypes, giving rare host genotypes an advantage. This creates time-lagged, negative frequency-dependent selection on the hosts. Under this model, there is an advantage to sexual reproduction, because it results in the production of genetically variable progeny, some of which may have rare genotypes and thus evade infection. In other words, some sexually produced offspring may be more resistant to local parasites, but susceptible clonal lines will never produce resistant offspring. Therefore, co-evolution with parasites should favour genetic polymorphism (Paterson et al., 2010; Schulte et al., 2010; Bérénos et al., 2011) and clonal diversity (Lively & Howard, 1994; Lively & Apanius, 1995; Ebert & Hamilton, 1996).

A major difficulty in testing the Red Queen hypothesis is finding suitable host–parasite systems, where sexual and asexual forms coexist and are able to displace one another. In fact, most empirical tests of the Red Queen hypothesis have been conducted using organisms in which sex is either triggered environmentally or behaviourally coupled to asex (sperm-dependent asexuality: Lively et al., 1990; Tobler & Schlupp, 2005; cyclical parthenogenesis: Decaestecker et al., 2007; Wolinska & Spaak, 2009; hermaphrodism: Schulte et al., 2010; Morran et al., 2011). Sexuals cannot be driven to extinction by asexuals in those organisms. Evidence for the maintenance of sexual reproduction by parasitism, particularly in populations where both sexuals and asexuals coexist, is limited to one snail-trematode model system (Potamopyrgus antipodarum: Lively, 1987; Dybdahl & Lively, 1998; Jokela et al., 2009). The persistence of sex in such populations via Red Queen dynamics is based on the premise that novel progeny generated by sex are more likely to escape attack by coevolving parasites.

Lively & Howard (1994) showed that a highly diverse asexual population may also express variation in resistance that will allow some host lineages to evade parasites and fuel negative frequency-dependent selection. In this case, if the genetic variability of asexuals is high, sex may be unable to provide any benefit against parasites (Lively & Howard, 1994). In strictly clonal populations, parasite-mediated selection might promote high clonal diversity (Ebert & Hamilton, 1996), whereas if the risk of infection is low, clonal diversity is expected to be determined by interclonal competition, which should select for lower clonal diversity. Recently, King et al. (2011) examined the contribution of parasites to host clonal diversity in natural populations of P. antipodarum and found support for the prediction of Lively & Howard (1994). It should be emphasized that clonal diversity can also be maintained regardless of parasitism, through genetic drift, gene flow, selection and processes that determine the origin of new asexual clones.

The present study investigates whether clonal diversity is associated with the frequency of parasitism. We predict that in populations with high risk of parasites, higher clonal diversity will be found, because of negative frequency-dependent selection (Paterson et al., 2010; Schulte et al., 2010; Bérénos et al., 2011). This assumes that parasites are coevolving locally and can adapt to locally common host genotypes. It also assumes that there is standing genetic variation in the host population, so that rare resistant host genotypes can increase in frequency. This would in turn lead to fluctuations in clonal frequencies that would on average maintain higher clonal diversity than in populations without parasite-driven negative frequency-dependent selection, although epidemiological feedbacks could lead to stable polymorphism (Lively, 2010). However, if co-evolution is not local, then locally driven correlations between clonal diversity and parasitism would not be observed. This can happen, for example, due to very high gene flow among parasite populations (Lively, 1999). Ultimately, the likelihood of parasite local adaptation is largely dependent on differences in dispersal rates between the parasite and its hosts (Gandon, 2002; Gandon & Michalakis, 2002), and the effect of migration on parasite local adaptation depends on the geographical mosaic of selection (Gandon & Nuismer, 2009).

A parasite's life cycle and dispersal modes are key determinants of its ability to adapt to locally common host genotypes and of the likelihood of local cycling (Louhi et al., 2010). Particularly in snail–trematode host–parasite systems, where free-living stages of the parasite are lacking or have limited mobility, parasite gene flow is often mediated by the host. In many such systems, the trematodes cycle through three different hosts: snails, fish and birds. As the vagility of the first intermediate host (snail) is low (Myers et al., 2000), parasite gene flow is determined by the host with the highest dispersal rate (e.g. birds, Prugnolle et al., 2005). Furthermore, the wider the geographical range of the host, the weaker the strength of parasite local adaptation, because exposure to several hosts results in the accumulation of different cercarial lineages over time and increased gene flow among these parasite lineages due to interbreeding (Lajeunesse & Forbes, 2002; Rauch et al., 2005). Taken together, we predict that in the absence of parasite gene flow that promotes local adaptation, parasites should select for host resistance or other host life-history traits (reviewed in Sorensen & Minchella, 2001). In essence, the parasites become one of many selective factors affecting the host population density, which is why we propose to test an alternative prediction that infection prevalence should be positively correlated with snail density (density-dependent transmission, Anderson & May, 1979), and that denser populations are more likely on average to be more genetically diverse than sparse populations.

To test these predictions, we sampled 22 natural populations of the freshwater snail Melanoides tuberculata throughout Israel. We examined the relationship between male frequency (an indicator of sexual reproduction in M. tuberculata, Livshits et al., 1984; Heller & Farstey, 1990), infection prevalence, snail density and two indices of clonal diversity.

Materials and methods

Host–parasite system

Melanoides tuberculata is a cerithiodean gastropod typically found in warm temperate to tropical freshwater bodies as well as in shallow slow-running water, especially on soft mud and sand substrata (Livshits & Fishelson, 1983; Dudgeon, 1986). Mean size at first reproduction varies considerably (7.0–16.4 mm, Berry & Kadri, 1974; Ben-Ami & Hodgson, 2005; Facon et al., 2005), and adult snails can reach a size of 20 mm (Pointier et al., 1993). It is native to North Africa and the Middle East (Pointier, 1999; cf. Facon et al. 2003), as evident from fossil records dating back to the Pleistocene (Tchernov, 1973; Sivan et al., 2007), but has invaded North and South America, East Africa, Southern Asia and Australia (Ben-Ami, 2006). Melanoides tuberculata is ovoviviparous. The eggs contain large amounts of glycogen and protein yolk (Hodgson et al., 2002), and embryos derive little direct nutrition from the mother after the eggs are formed (Ben-Ami & Hodgson, 2005). Formerly thought to be obligately asexual (Jacob, 1957, 1958; Berry & Kadri, 1974), evidence suggests that sexual (diploid) forms may coexist with asexual (polyploid) forms (Samadi et al., 1999). Moreover, sex in M. tuberculata plays a crucial role in its ability to invade new ecosystems, because it amplifies the effect of multiple introductions of the snail by generating novel trait combinations (Facon et al., 2005, 2003). The frequency of fertile males can reach up to 66% particularly in Israel (Livshits & Fishelson, 1983; Heller & Farstey, 1990; Hodgson & Heller, 1990; Ben-Ami & Heller, 2005, 2007, 2008).

Melanoides tuberculata is the first intermediate host for several trematodes that have important public health and agricultural implications (e.g. eye fluke, Radev et al., 2000; Dzikowski et al., 2003; Ben-Ami et al., 2005; Yousif et al., 2010; reviewed in Pinto & de Melo, 2011). Among these trematodes is Centrocestus sp., the most abundant digenetic trematode of M. tuberculata in Israel (Ben-Ami & Heller, 2005). The parasite develops parthenogenetically within the snail and sterilizes it, and cercariae liberated from infected snails encyst on the gills of fish. The larval worms become adults after being consumed by waterfowl or waders, wherein they reproduce sexually and their eggs are released with the definitive host's faeces (Farstey, 1986). These eggs hatch into free-swimming miracidia that penetrate snail tissue, thus completing the parasite's life cycle. Previous studies have shown that the parasite is capable of sterilizing up to 84% of infected female snails (Ben-Ami & Heller, 2005). Other commonly found castrating trematodes of M. tuberculata in Israel include Philophthalmus sp., which requires an aquatic bird as a definitive host (Ben-Ami, 2006), and Transversotrema patialense, which necessitates a marine or freshwater fish as a definitive host (Ben-Ami et al., 2005).

Study sites and data collection

A survey of 22 independent M. tuberculata populations was carried out in Israel during the summer of 2011. We sampled populations from streams, ponds and springs located along the Mediterranean Coast, in the Judean Desert and the Negev, and in the Jordan Valley and other inner valleys. We randomly collected 100 snails from each population and transported them alive to the laboratory, where they were measured and sexed based on the colour of their gonads (Heller & Farstey, 1989). There is no evidence of differential capture probability between male and female snails, or among clones/morphs. Trematode prevalence was determined by examining the gonad and digestive gland of 35 snails under a light microscope, because castrating trematodes tend to occupy the gonadal space of the snail host visceral mass. There is also no evidence of differential capture probability between infected and uninfected snails, or among snails infected by different trematode species. We used 100 snails for sex determination (and only 35 snails for estimating infection prevalence) to get more accurate estimates of sexual reproduction, because male frequencies were low. Snail density was estimated by collecting and counting all snails from five randomly selected squares of 25 × 25 cm (average number of snails from five squares). The snails used for estimating density were returned to their respective habitat within a few hours after sampling. Approximately twenty snails from each population were either snap-frozen at −80 °C or kept alive for genotyping.

Genotypic analyses

We performed cellulose acetate electrophoresis as described by Richardson et al. (1986) and by Hebert & Beaton (1989), with modifications by Fox et al. (1996) and Jokela et al. (2009). We identified nine polymorphic loci (six dimeric enzymes: AA1, AA2, IDH1, IDH2, PEP-D and 6PGD; and three monomeric enzymes: MPI, PGM1 and PGM2). We observed up to three bands per locus in most snails (85.3%), four bands in another 12.0% of the snails and five bands in the remaining 2.7% of the snails. When deemed necessary, we reran electrophoresis and performed blind double-reading of genotypes. Occasionally one of the nine loci did not produce a readable banding pattern. In a few cases, two of the nine loci were unreadable. In such cases, the multilocus genotype was scored if unambiguous assignment was possible. If not, or if more than two loci were unreadable, the individual was not used in any analyses. In total, we assigned 182 different genotypes to 375 snails (five snails could not be scored due to ambiguousness, and seven snails had unreadable banding patterns).

We used GENODIVE to calculate two clonal diversity indices that are commonly used in the literature (Meirmans & Van Tienderen, 2004). First, we computed clonal richness, that is, the number of genotypes found in a population. This index is essentially the Stoddart diversity index (Stoddart & Taylor, 1988). We then computed the effective number of genotypes, that is, the number of genotypes found in a population corrected for sample size. These two diversity indices are appropriate for comparing species diversity between ecological units, when it is not feasible to quantify the degrees of dissimilarity between the genotypes (Kosman & Leonard, 2007).

We used MLGsim 2.0, an updated version of MLGsim (Stenberg et al., 2003), to calculate the probability that identical multilocus genotypes (MLGs) at 9 allozyme loci indeed belong to the same genotype. This improved version was used because our analyses were based on 375 snails, whereas MLGsim has a maximum sample size limit of 200 individuals. Initially the program calculates Psex (likelihood) values for genotypes that are found more than once, that is, the probability that these genotypes occur the observed number of times in an asexual population with the observed allele frequencies, assuming Hardy–Weinberg and linkage equilibria. Thereafter, the program applies a Monte Carlo simulation to obtain the critical values of Psex for three significance levels (0.05, 0.01 and 0.001), thus identifying the genotypes that are significantly overrepresented and therefore likely to be members of the same clone (Stenberg et al., 2003).

Statistical analyses

All statistical tests were carried out using IBM SPSS Statistics for Windows version 21. Mean values are shown with ± SE. Frequency data were arcsine(square root)-transformed, and densities were log-transformed before being analysed. All probabilities are two-tailed. We used linear regression to determine the relationship between infection prevalence, male frequency, snail density and each of the two clonal diversity indices. We also used binary logistic regression to regress clonal diversity on infection status. In both regression analyses, we initially controlled for snail size by entering it as a covariate into the model. We then removed snail size from all models because it was nonsignificant. The presence/absence of fish was recorded for all sites (and entered into nonsignificant regression models), because fish may serve as intermediate hosts for several trematodes, including Centrocestus sp. As in the study Ben-Ami & Heller (2005), sites were grouped into three habitat geography categories (valley, coast or desert) and into three habitat types (stream, pond or spring). These categorical predictor variables were ‘dummy’-recoded into uncorrelated dichotomous variables (contrasts), before entering them into nonsignificant regression models.

Results

We used total infection rate (i.e. infection rate by all parasite species combined, hereafter referred to as infection prevalence) as an estimate of risk of infection (King & Lively, 2009; King et al., 2011). Although trematode communities may consist of several coexisting species, co-infections by two parasites are rare and usually the community has one common species and several rare (Kuris & Lafferty, 1994; Louhi et al., 2013). In fact, we did not find snails simultaneously infected by two or more parasite species. In our samples, the most common species varied among populations. For example, Centrocestus sp. was the most common parasite in four populations, and Philophthalmus sp. and T. patialense were abundant in one population each. Two unidentified trematodes dominated in the remaining five parasitized populations. Both identified and unidentified trematodes had severe effects on host fitness, resulting in host castration (F. Ben-Ami, unpublished data).

The frequency of M. tuberculata males varied between 0 and 23.0%, with a mean ± SE of 2.2 ± 1.1% males per population (Fig. 1a). Twelve populations had no males, suggesting that the populations are strictly clonal, and eight additional populations were predominately clonal (i.e. ≤ 3% males). Hence, only the remaining two populations can truly be considered mixed populations, with both sexual and asexual snails. Infection prevalence varied between 0 and 85.7%, with a mean ± SE of 17.5 ± 5.5% infected snails per population (Fig. 1b). Eleven populations were parasite-free. Snail density varied between 0.2 and 107.3 snails per sampling square (25 × 25 cm), with a mean ± SE of 18.5 ± 5.8 snails per sampling square (Fig. 1c). Infection prevalence could not explain the variation in male frequency (= 0.200, df = 21, R2 = 0.01, = 0.66; Fig. 2a), nor could it explain the variation in snail density (= 0.276, df = 20, R2 = 0.014, = 0.61; Fig. 2b). Snail density did also not explain the variation in male frequency (= 1.617, df = 20, R2 = 0.078, = 0.22; Fig. 2c). Degrees of freedom for tests involving snail density are one less, because the density in one population could not be measured. The presence/absence of fish, habitat geography and habitat type did not significantly improve these regression models (> 0.44 when entering these categorical variables into the models, coefficients of all categorical variables were nonsignificant, > 0.13).

Figure 1.

Distribution across populations of (a) male frequency (b) infection prevalence (c) snail density, and (d) number of host genotypes (nominal and effective). Nominal refers to the actual number of genotypes found in a population, whereas effective refers to the same number corrected for sample size.

Figure 2.

Relationship between (a) infection prevalence and male frequency (b) infection prevalence and snail density, and (c) snail density and male frequency. In Fig. 2a, data point (0, 0) has been enlarged to highlight the fact that nine populations were parasite-free and had no males.

Clonal diversity varied considerably among host populations, with some populations consisting of only two host genotypes, whereas in others every single host individual had a different genotype (Fig. 1d). The number of individual snails per genotype across all host populations varied between 1 and 36, with a mean ± SE of 2.1 ± 0.3 individuals per genotype. Seventeen of 182 host genotypes were shared by two populations; two additional genotypes were shared by three populations; and one genotype was present in four populations (Simpson/Nei index when pooling all populations together = 0.99, Nei, 1987). In comparison with the mean number of individual snails per genotype, only four of the twenty shared genotypes were sufficiently common in terms of total number of snails (13–36); and none of these shared common genotypes was over-infected in comparison with the mean infection prevalence. We obtained Psex values for 14 of the 20 shared host genotypes that were significant at < 0.001 (10 genotypes), < 0.01 (one genotype) and < 0.05 (three genotypes). Thus, only six of the 20 shared host genotypes as observed here could actually be different genotypes (but the resolution of the method does not allow distinguishing among them).

Infection prevalence explained 20.7–26.3% of the variation in clonal diversity (< 0.04, Table 1, Fig. 3), whereas snail density could not explain this variation (> 0.44, Table 1, Fig. 4). Habitat geography, but not the presence/absence of fish and habitat type, improved these latter regression models. However, they remained nonsignificant (> 0.18 when entering these categorical variables into the models explaining snail density). Due to the non-normal distribution (skewness) of infection prevalence among populations, we also regressed clonal diversity on infection status (0 for parasite-free populations, 1 for parasitized ones). Based on the Nagelkerke pseudo R2, we found that infection prevalence explained 27.6% of the variation in clonal diversity (binary logistic regression χ2 = 3.929, df = 1, P = 0.047).

Table 1. Regression analysis showing the variation of clonal diversity explained by infection prevalence and snail density. Habitat geography, but not the presence/absence of fish and habitat type, improved regression models with snail density as an independent variable, although they remained nonsignificant (> 0.18 for all categorical variables).
Dependent variableIndependent variable df F P R 2
Clonal richness (Stoddart index)Infection prevalence217.1520.0150.263
Effective number of genotypesInfection prevalence215.2240.0330.207
Clonal richness (Stoddart index)Snail density200.5650.460.029
Effective number of genotypesSnail density200.5940.450.030
Figure 3.

Relationship between infection prevalence and (a) clonal richness (Stoddart index), and (b) effective number of host genotypes.

Figure 4.

Relationship between snail density and (a) clonal richness (Stoddart index), and (b) effective number of host genotypes.

Finally, we compared male frequency between 1999 and 2011 across eight sampling sites (Ben-Ami & Heller, 2005). We found that male frequency in 2011 was significantly correlated with male frequency 12 years earlier (Pearson = 0.55, = 0.035).

Discussion

The goal of this study was to determine whether parasite-mediated selection promotes clonal diversity. This goal stems from a pivotal prediction of the Red Queen hypothesis that generating genetically variable progeny is the key advantage of sex vs. asex, which may in return overcome the many disadvantages of sexual reproduction. Put differently, genetic variation between resistant and susceptible host clones may be driven by negative frequency-dependent selection. We indeed found that infection prevalence is significantly correlated with the observed variation in clonal diversity of M. tuberculata populations. Consistent with previous work (Heller & Farstey, 1990; Ben-Ami & Heller, 2005, 2007, 2008), we found no relationship between infection prevalence and male frequency. Furthermore, snail density could neither explain the variation in host clonal diversity nor explain the variation in male frequency. Taken together, our study suggests that parasite-mediated selection promotes clonal diversity.

In a study with another snail–trematode system, King et al. (2011) genotyped 1,052 snails from 17 populations using seven allozyme markers that were identical to the ones used in the present study. They reported that no clonal host genotype was shared between populations, whereas in our study, 11.0% clonal genotypes were shared among 2–4 populations (most of the time not the same populations) up to 350 km apart (Fig. 5a). In fact, linking all populations with one or more shared clonal genotypes resulted in a web of 17 potentially interconnected populations (of 22 sampled populations, Fig. 5b). This surprising finding is probably due to differences in the dispersal ecology of the two host–parasite systems. Dispersal of the definitive host (ducks) is mostly local in P. antipodarum (C. Lively, personal communication), but appears to be carried out by migrating birds in M. tuberculata. Israel is located on a very narrow and highly frequented bird migration route that connects African wintering grounds and Eurasian breeding grounds (Moreau, 1972; Cox, 1985). The interconnectivity of M. tuberculata populations could be explained by host gene flow between snail populations via migrant birds (on their feet or plumage, Rees, 1965). Clonal diversity may therefore be promoted by host migration.

Figure 5.

(a) Distribution of shared host genotypes by maximal distance among populations, (b) map of Israel showing the interconnectivity among Melanoides tuberculata populations based on shared clonal host genotypes. Nonconnected populations are shown as open circles.

King et al. (2011) also reported that male frequency was positively related to infection frequency. The selection imposed by parasites in King's study was on average two-fold lower (8.1 ± 1.6% infected snails per population, with 11 of 17 populations with ≤ 5% infected snails; King et al., 2011; Table 1) than in our study. Therefore, if parasitism had played a role in driving sexual reproduction in M. tuberculata, these effects should have been detected in our study (particularly if we assume that parasite-mediated selection can turn the eight predominantly asexual populations with ≤ 3% males into truly mixed populations). In addition, the distribution of infection in our study was wide and skewed, with 11 parasitized populations (35.1 ± 8.2% infected snails per infected population) and 11 parasite-free. It has been suggested that sexual reproduction should be positively correlated with infection prevalence only if the range of prevalence is wide, with very low and high prevalence (i.e. ‘if the variance in risk among populations is large, sexual reproduction should be positively associated with the prevalence of infection’; Lively, 2001). Yet our results (Fig. 1b) and a survey conducted twelve years earlier (Ben-Ami & Heller, 2005; Fig. 3) show a wide and skewed distribution of infection, with no apparent correlation between sexuality and parasitism, despite the correlation between male frequency in 1999 and 2011.

The importance of parasite-mediated selection as a driving force of genetic diversity has long been hypothesized by Haldane (1949). There is also ample empirical evidence for the link between parasitism and genetic diversity (see Bérénos et al., 2011 and references therein). However, few studies examined the influence of parasitism on clonal diversity, even though parasitism has been suggested as a source of clonal diversity in several organisms (e.g. aphids: Fuller et al., 1999; mites: Weeks & Hoffmann, 2008; dinoflagellates: Richlen et al., 2012). In one such study, Duncan & Little (2007) found that during an epidemic of the bacterial pathogen Pasteuria ramosa, clonal diversity of its crustacean host Daphnia magna declined as parasite prevalence increased. But as the epidemic abated, host clonal diversity increased once again to pre-epidemic levels. It should be noted that in their study, Duncan & Little (2007) looked for temporal variation in a single host population, whereas in the present study, we assessed spatial variation among 22 populations. In another study of 42 lake populations of the Daphnia galeata × hyalina × cucullata species complex, Wolinska & Spaak (2009) did not find evidence of parasite-mediated increase in host genotype richness (=number of clones within a population). Therefore, our study provides novel evidence for the role of parasitism in maintaining clonal diversity.

Tight host specificity, high virulence and high levels of infection can result in rapid host–parasite co-evolution, which is supported by numerous studies showing parasite local adaptation (reviewed in Kawecki & Ebert, 2004; Greischar & Koskella, 2007). Parasites with an indirect life cycle face the additional challenges of adapting to different host species (e.g. trade-offs in infectivity and virulence between intermediate and definitive hosts, Davies et al., 2001). When two or more hosts are required, as is the case with all the trematode species explored in the present study, or when parasite dispersal is affected by patterns of bird migration, parasite generation times may be considerably longer and selection may become attenuated. The residence period of Centrocestus (most common parasite) and Philophthalmus in the motile definitive host is unknown. Nevertheless, despite the challenges faced by indirectly transmitted trematodes, parasite-mediated selection appears to regulate clonal diversity in M. tuberculata.

The use of total infection rate (by different parasite species) as an estimate of the risk of infection is frequently used in similar studies of snail–trematode co-evolution (King & Lively, 2009; King et al., 2011). Red Queen dynamics operate with cycling of different genotypes of the same parasite species (Clarke, 1976; Jaenike, 1978; Bell, 1982; Nee, 1989). This requires that resistance comes from alleles at the same host locus, which is a rather restrictive assumption for different parasite species. For Red Queen dynamics to work with multiple parasite species, they must be regulated by the same ecological factor. Hence, the pooling of parasite infections (per site) may mask expected Red Queen correlations. Nonetheless, in our study, only one parasite species was sufficiently common in any single site, thereby suggesting that selection exerted by the remaining rare parasite species on the host is negligible.

The origin of new clones in M. tuberculata is unknown, and in particular it is unknown whether asexual lineages arise from local sexual ancestors (reviewed in Simon et al., 2003). Preliminary work suggests that > 3× ploidy might occur in natural populations of M. tuberculata (F. Ben-Ami and J. Jokela, unpublished data). The source of natural autopolyploids is usually attributed to (i) the union of two unreduced gametes produced by the same diploid female or (ii) the union of gametes from a triploid female with gametes from a male with haploid or diploid sperm (Ramsey & Schemske, 1998; Husband, 2004; Neiman et al., 2011). The existence of M. tuberculata populations consisting of only diploids (2n = 22 in Egypt; Yaseen, 1995), only polyploids (2n = 90–92 in Spain; Baršiene et al., 1996) or both (2n = 32 for diploids and 2n = 90–94 for polyploids in India; Jacob, 1958) suggests that both of the abovementioned scenarios cannot be ruled out. Additionally, the observed differences in chromosome number or size in M. tuberculata may generate genetic variation to asexual lineages (Castagnone-Sereno, 2006). Elucidating the determinants of ploidy level and genome size variation in M. tuberculata would further our understanding of how parasites may promote clonal diversity. These steps are crucial because ploidy level and genome size variation affect fitness in many organisms, including resistance to parasites (King et al., 2012).

Conclusions

The main finding of this study is that parasite-mediated selection coupled with host dispersal ecology can promote clonal diversity in M. tuberculata populations. The existence of only two truly mixed populations as well as eight additional populations with very few males (≤ 3%) – of a total of 22 sampled populations – suggests that clonal diversity may erode the advantage of sexual reproduction in M. tuberculata.

Acknowledgments

We thank D. Ebert, A. Gardner, C. Lively and an anonymous reviewer for helpful comments on this manuscript, and P. Meirmans for assistance with GENODIVE. Specimens for this study were collected under permit 2011/38060 from the Israeli Authority for Nature Reserves and National Parks. This research was supported in part by grant #2011011 from the United States-Israel Binational Science Foundation (BSF) to FBA, and by a Swiss National Science Foundation grant to JJ.

Ancillary