Under the Red Queen hypothesis, host–parasite coevolution selects against common host genotypes. Although this mechanism might underlie the persistence of sexual reproduction, it might also maintain high clonal diversity. Alternatively, clonal diversity might be maintained by multiple origins of parthenogens from conspecific sexuals, a feature in many animal groups. Herein, we addressed the maintenance of overall genetic diversity by coevolving parasites, as predicted by the Red Queen hypothesis. We specifically examined the contribution of parasites to host clonal diversity and the frequency of sexually reproducing individuals in natural stream populations of Potamopyrgus antipodarum snails. We also tested the alternative hypothesis that clonal diversity is maintained by the input of clones by mutation from sympatric sexuals. Clonal diversity and the frequency of sexual individuals were both positively related to infection frequency. Surprisingly, although clones are derived by mutation from sexual snails, parasites explained more of the genotypic variation among parthenogenetic subpopulations. Our findings thus highlight the importance of parasites as drivers of clonal diversity, as well as sex.

In spite of the well-known effects of directional selection and genetic drift in reducing genetic variation, natural populations are often genetically diverse (Lewontin 1974; Nevo 1978). One proposed hypothesis for the maintenance of high levels of genetic diversity relies on parasite-mediated selection against common host genotypes, also known as the Red Queen hypothesis (Haldane 1949; Jaenike 1978; Hamilton 1980; Bell 1982). Under this hypothesis, parasites adapt to infect host genotypes after they become locally common; and thus, coevolution with parasites promotes the accumulation and long-term persistence of genetic polymorphism (for experimental evidence, see Bérénos et al. 2011). For instance, polyandry in eusocial insect queens is thought to be an evolutionary response to parasite pressure as it increases the genetic diversity of colonies (Schmid-Hempel 1994; Schmid-Hempel and Crozier 1999; Brown and Schmid-Hempel 2003). The Red Queen mechanism could also contribute to the persistence of sexual reproduction in host populations where parthenogenetic lineages arise from sexual ancestors (Jaenike 1978; Hamilton 1980; Bell 1982), a phenomenon that occurs in a broad range of animals (reviewed in Simon et al. 2003). Virulent parasites that disproportionately infect common genotypes could provide a means by which sex is protected from elimination by parthenogenetic clones, even when the clones have a twofold reproductive advantage (e.g., Hamilton 1980; Hamilton et al. 1990).

Parasite-mediated selection might also favor a diversity of clonal genotypes (for examples, see Duncan and Little 2007; Duffy et al. 2008; Wolinska and Spaak 2009). As a result, in mixed populations with competing asexual and sexual conspecifics, both rare clonal genotypes and sexual genotypes may be selected for. Coexistence of sexuals and asexuals could persist provided clonal diversity is not high enough to remove the advantage of sex (Lively and Howard 1994). Clonal polymorphism might alternatively be maintained if clonal lineages continuously originate from the local sexual population (Vrijenhoek 1998; Simon et al. 2003). Therefore, the input of new clones from local sexuals and their removal, through drift or natural selection, may affect the level of clonal diversity in a population. Parasites may play no role. The extent to which parasite-driven selection underlies clonal diversity in mixed populations of clonal and sexual individuals remains largely unexplored.

We predicted that host population genetic diversity, as represented by clonal polymorphism and the frequency of sexual individuals, should increase with parasite pressure in natural, mixed populations of a New Zealand snail, Potamopyrgus antipodarum. Our prediction stems from the empirical evidence to date that parasites select against common clones of this freshwater snail (Dybdahl and Lively 1995b; Lively and Dybdahl 2000; Jokela et al. 2009; Koskella and Lively 2009). We conducted large-scale genotyping of 17 stream populations throughout the South Island of New Zealand, and related the frequency of sexual snails and clonal diversity to infection by sterilizing parasites. Such parasites are common in populations of P. antipodarum (Lively 1987; Jokela and Lively 1995; King and Lively 2009). Infection frequency naturally varies across populations, providing the opportunity to compare diversity across a range of selection pressures. Lastly, we tested an alternative to the parasite hypothesis for the promotion of high clonal diversity. We examined to what degree, if any, the frequency of sympatric sexual snails affects the diversity of clones observed.



Potamopyrgus antipodarum is a common freshwater snail found in lakes and streams throughout New Zealand. Individual snails are either triploid parthenogenetic females or diploid dioecious sexuals (Wallace 1992). Populations can be mixed, with both clones and sexual individuals, or strictly clonal (Lively 1987; Dybdahl and Lively 1995a). Based on mitochondrial haplotypes, parthenogenetic snail lineages are thought to be reasonably young (70,000 to 150,000 years) and derived from sexual ancestors (Dybdahl and Lively 1995a; Neiman et al. 2005, 2010). The rate at which they are produced remains speculative, as for most species (Neiman et al. 2009). Several lines of evidence suggest that the parthenogenetic clones are much younger than the mitochondrial haplotype analysis suggests. First, lakes (Dybdahl and Lively 1995a) and streams (this study) have unique parthenogenetic genotypes that are found in local populations, not widespread across biogeographic regions. Second, many of the best described clonal populations occupy lakes that are postglacial (<13-kyr old) and still present highly specific local clonal structure between ecological habitat zones, not much older than few thousand years (Jokela and Lively 1995; Fox et al. 1996). Third, as a rule, the parthenogens share the same isozyme alleles as local sexual populations indicating a local origin (Dybdahl and Lively 1995a).

Potamopyrgus antipodarum is a first intermediate host for several trematode parasite species (Winterbourn 1974; Jokela and Lively 1995). Trematodes use gastropods as first intermediate hosts and vertebrates as final hosts wherein they reproduce sexually. Eggs are released with the final host's faeces. The eggs are ingested by gastropods or hatch into free-swimming miracidia and penetrate gastropod tissue. Once in the gonads, trematodes develop parthenogenetically and sterilize the host.


Snail populations in 17 independent streams across the South Island of New Zealand were sampled in summer (January–February) of 2008/2009. The streams were not interconnected and ranged from 5 km to hundreds of kilometers apart. One collection was made from each stream by washing snails from rocks into kick nets, or by pushing the nets through aquatic vegetation. All collected snails were stored live in plastic containers at the Edward Percival Field Station in Kaikoura, New Zealand. Water in the containers was replaced every other day, and the snails were fed Spirulina algae ad libitum. Once transported to Indiana University, a random subsample of mature individuals (>2.5 mm) per stream were sexed and examined for infection under a dissecting microscope. The infection frequency of a parasite species was estimated as the proportion of infected individuals. Immediately after dissection, snail heads were snap-frozen and stored individually at −80°C for electrophoresis.

To differentiate individual snails based on ploidy (i.e., reproductive mode) and to identify individual clonal genotypes, we performed cellulose acetate electrophoresis (Dybdahl and Lively 1995a). We used seven polymorphic enzyme loci (Dybdahl and Lively 1995a): peptidase (Pep-D), isozymes of isocitrate dehydrogenase (Idh-1 and Idh-2), 6-phosphogluconate dehydrogenase (6pgd), isozymes of phosphoglucomutase (Pgm-1 and Pgm-2), and mannose-6-phosphate isomerase (Mpi). We used Phosphate (for 6pgd and Mpi), Tris-Maleate (for Pep-D and Idh), and Tris-EDTA-Borate (for Pgm) as running and soaking buffers. Run times were 35–90 min at 150–220 V. Cellulose acetate equipment and gels were supplied by Helena Labs (Beaumont, TX).

Multilocus genotypes of P. antipodarum individuals from each stream were analyzed. Asymmetric banding intensities of heterozygotes at two loci (Pep-D and 6pgd) can be used to reliably distinguish triploid from diploid snails (Dybdahl and Lively 1995a; Jokela et al. 2009). Triploid individuals with different allozyme genotypes were considered to be unique clones. We assigned individuals with incomplete genotype profiles (i.e., alleles were unidentifiable at one or more loci) to a single clonal group with matching resolvable loci. If more than one clonal group matched those loci, individuals with incomplete allozyme profiles were removed from further analyses. A complete genotype for some individuals could not be read due to low-quality samples or laboratory error. A total of 1052 snails from 17 populations had resolvable allozyme profiles.

Snails from different clonal lineages, as identified by allozyme neutral loci, are unlikely to be monomorphic for genes associated with host–parasite interactions. Previous studies of P. antipodarum snails have shown that clonal genotypes vary in susceptibility, and resistant clones can become susceptible to infection after they become locally common (Dybdahl and Lively 1995b; Jokela et al. 2009; Koskella and Lively 2009). Thus, a diversity of snail genotypes may be equated with genetic diversity for resistance in the population.


From each stream population sampled, we calculated several descriptors: infection frequency (frequency of snails infected with the most common local parasite species), frequency of sexuals (i.e., frequency of diploid snails), and two measures of clonal polymorphism (i.e., clonal richness and clonal diversity). The “most common” local parasite species was defined as the species having the highest frequency of infection in the sampled population. Microphallus sp. was the most common parasite in five populations, Coitocaecum anispidis in three samples, Notocotylus gippyensis in three populations, as well as Stegodexamene anguillae, Telogaster opisthorchis, and a furcocercous cercariae (F1) in one sample each. Three populations were infected with two parasite species at equal frequency, so only the frequency of one parasite was included. The frequency of infection with the most common parasite was highly correlated with infection by all parasite species (Pearson correlation: r = 0.932, P < 0.001). However, in a previous study examining various measures of parasite-mediated selection, the frequency of the most common local parasite species was found to be the most appropriate (King and Lively 2009). Therefore, we used the infection frequency of the common parasite for all analyses.

Clonal richness was calculated using the Stoddart's index, G0 (Stoddart and Taylor 1988). G0 is a measure of genotypic richness ranging from 1 (single genotype) to N (where N is the sample size, i.e., each individual belongs to a different multi-locus genotype). The upper limit of G0 is a function of sample size when G0∼ N. To get equal sample sizes, we randomly selected a subsample of 50 individuals from each observed sample, with replacement. After 1000 bootstraps, we recorded the mean G0 and the standard deviations. We also calculated Shannon–Wiener diversity index as a second measure of clonal genotypic polymorphism. This measure incorporates evenness, which describes the distribution of clones within a sample. We used Matlab 7.0.4 (MathWorks, Natick, MA) for randomizations and calculations.

We used linear regression to determine the relationship between the independent variable, total infection frequency (including all sexual and asexual snails), and the dependent variable, the frequency of sexual snails. The infection frequency of clonal snails alone was highly correlated with the total infection frequency of all snails, regardless of reproductive mode (r = 0.983, P < 0.001); however, we conducted separate regression analyses with both independent variables. To ensure that sexual snails were not more susceptible to parasites than clonal snails, we conducted a paired t-test comparing infection frequencies in clones and sexual snails in streams with >8% sexual frequency (i.e., seven streams). This cut-off was chosen to ensure we compared streams with an adequate sample size of sexual snails.

We determined the relative contributions of parasites and the local sexual subpopulation to clonal polymorphism. We ran several regression analyses with Stoddart's G0 richness measure or Shannon–Wiener diversity index as the dependent variable. We conducted a multiple regression with infection frequency and the frequency of sexual individuals as independent variables together, and then conducted two linear regressions, including only infection frequency and sexual frequency. As above, separate regression analyses were performed for total infection frequency of all snails and of only clonal snails. Only the mean values of clonal richness and diversity per sample that were generated with bootstrapping were used.

From the series of regressions, we used a variance partitioning approach to evaluate how much of the variation in clonal polymorphism could be attributed to the individual effects of parasites and sex, as well as to their “combination” (explained in Legendre and Legendre 1998). The size of the combination component of variance depends on the strength of the linear relationship between sex and infection. It is essentially the percent variance explained by infection, but confounded by the frequency of sexual snails. Because the partitioned variance values are generated from subtracting the R2-values from regression analyses, they do not have standard errors or tests of significance associated with them. Therefore, once we determined the percent variation explained by infection alone, we conducted a test of the one-tailed significance of the Pearson product-moment correlation coefficient (√R2).

Any variation of the two measures of clonal polymorphism among populations was not due to variation of the number of sampled diploid sexuals (richness: Pearson correlation r = 0.467, P = 0.059; diversity: r = 0.397, P = 0.114) or triploid clones (richness: r = 0.179, P = 0.492; diversity: r = 0.194, P = 0.456).


Stream snail populations consisted of both sexual and asexual snails, as well as a wide range of clonal genotypes (Table 1). Across populations, the total frequency of infection (including both sexual and asexual snails) explained 37% of the variation of the frequency of sexual snails (Fig. 1A; Table 2), and infection of clonal snails alone explained 26% of the variation of sex (Table 2). Infection frequency among triploid clones was greater than for diploid sexuals, but the difference was not significant among streams with >8% frequency of diploid, sexual snails (t6= 0.797, P = 0.456; Fig. 2). This suggests that the relationship between sex and infection was not driven by differential susceptibility of sexual snails.

Table 1.  Summary of genotype and infection data for individual stream populations.
Pop.Sample sizeNo. sexual snailsNo. clonal snailsNo. clonal genotypesCommon parasite sp. (Total infection frequency)
  1. 1Two parasite species were equally common.

 161 45728Microphallus (.05)
 262 65633Telogaster (.21)
 361 25937Microphallus (.05)
 455 45123Furcocercous cercariae (.04)
 560 65427Microphallus/Coitocaecum1 (.02)
 655 45117Notocotylus (.04)
 774 66847Coitocaecum (.20)
 859 75241Coitocaecum (.17)
 958 15730Notocotylus (.05)
1070 26821Microphallus (.01)
1166 85829Stegodexamene (.11)
1266 56136Microphallus (.05)
1357 55226Microphallus (.09)
1465 46127Microphallus/Telogaster1 (.05)
1561 16034Notocotylus (.03)
1656 15519Notocotylus/Coitocaecum1 (.02)
1766204639Coitocaecum (.18)
Figure 1.

Relationships between (A) infection frequency in all snails (total) and frequency of sexual snails, (B) infection frequency and Stoddart's G0 clonal richness, (C) infection frequency and Shannon–Wiener clonal diversity, (D) frequency of sexual snails and clonal richness, and (E) frequency of sexual snails and clonal diversity for 17 stream populations of Potamopyrgus antipodarum. Best-fit regression line for mean values is shown for (A), (B), and (C) where significant linear relationships were detected. Vertical bars represent ±1 SD generated by 1000 bootstraps.

Table 2.  Regression analysis showing the variation of frequency of sexual snails explained by infection frequency. Results for tests are shown separately for (i) total infection frequency of all snails and (ii) infection frequency of only clonal snails as independent variables.
TestIndependent variablesStandardized ßtPR2
(i)Total infection frequency0.6082.9690.0100.370
(ii)Clonal infection frequency0.5092.2900.0370.259
Figure 2.

Infection in sexual (open circles) and clonal (black circles) Potamopyrgus antipodarum snails relative to their frequency in the population. Data are presented for seven streams across the South Island of New Zealand with >8% frequency of sexual snails. Lines connect pairs of sexuals and clones in each stream. Stream population numbers correspond to those in Table 1.

No populations were monoclonal, and no clonal genotype was shared between populations. Parasitic infection was positively and significantly associated with both measures of clonal polymorphism, and infection explained more of the variation of clonal polymorphism than the local frequency of sexual snails (Fig. 1B,C; Tables 3 and 4). After partitioning the variance (Legendre and Legendre 1998), the independent effects of infection significantly explained 21–27% of the variation of clonal richness (total infection frequency, R2= 0.266, t15= 2.332, P = 0.017; clonal infection frequency, R2= 0.212, t15= 2.009, P = 0.031) and 21–26% in diversity (total infection frequency, R2= 0.258, t15= 2.758, P = 0.007; clonal infection frequency, R2= 0.207, t15= 1.979, P = 0.033). The local frequency of sexual snails alone explained 0–1.4% of the variation, and was not significantly associated with either measure of clonal polymorphism (Fig. 1D,E; Tables 3 and 4). The combination of infection and sex explained 15–17% of the variation of richness and 12–13% of diversity.

Table 3.  Regression analysis showing variation of clonal richness (Stoddart's G0) explained by infection frequency and the frequency of sexual snails. Three regressions were performed with (i) both infection frequency and frequency of sexuals as independent variables, as well as (ii) infection frequency and (iii) frequency of sexuals alone. Results are shown separately for total infection frequency and infection frequency of only clonal snails. Variation was partitioned as follows: sex alone, R2(i)–R2(ii); infection alone, R2(ii)–R2(iii); sex+infection, R2(ii)+R2(iii)–R2(i) (Legendre and Legendre 1998).
TestIndependent variablesStandardized ßtPR2
(i)Total infection frequency+0.6512.589 0.0210.443
 Frequency of sexual snails0.0250.042 0.924 
(ii)Total infection frequency0.6653.452 0.0040.443
(iii)Frequency of sexual snails0.4201.794 0.0930.177
% variation explained by sex alone, 0.0%; infection alone, 26.6%; sex+infection, 17.7%
(i)Clonal infection frequency+0.5532.304 0.0370.403
 Frequency of sexual snails0.1390.579 0.572 
(ii)Clonal infection frequency0.6233.089<0.0010.389
(iii)Frequency of sexual snails0.4201.794 0.0930.177
% variation explained by sex alone, 1.4%; infection alone, 21.2%; sex+infection, 16.3%
Table 4.  Regression analysis showing variation of clonal diversity (Shannon–Wiener) explained by infection frequency and the frequency of sexual snails. Three regressions were performed with (i) both infection frequency and frequency of sexuals as independent variables, as well as (ii) infection frequency and (iii) frequency of sexuals alone. Results are shown separately for total infection frequency and infection frequency of only clonal snails. Variation was partitioned as follows: sex alone, R2(i)–R2(ii); infection alone, R2(ii)–R2(iii); sex+infection, R2(ii)+R2(iii)–R2(i) (Legendre and Legendre 1998).
TestIndependent variables Standardized ßtPR2
(i)Total infection frequency 0.641 2.4310.0290.387
 Frequency of sexual snails−0.033−0.1350.902 
(ii)Total infection frequency 0.621 3.0700.0080.386
(iii)Frequency of sexual snails 0.357 1.4810.1590.128
% variation explained by sex alone, 0.1%; infection alone, 25.8%; sex+infection, 12.7%
(i)Clonal infection frequency 0.536 2.1230.0520.340
 Frequency of sexual snails 0.085 0.3350.742 
(ii)Clonal infection frequency 0.579 2.7470.0150.335
(iii)Frequency of sexual snails 0.357 1.4810.1590.128
% variation explained by sex alone, 0.5%; infection alone, 20.7%; sex+infection, 12.3%


Host–parasite coevolution is thought to maintain substantial genetic diversity within species (Haldane 1949). Given past work on Potamopyrgus snails showing that parasites select against common clones in the laboratory (Koskella and Lively 2009) and in nature (Dybdahl and Lively 1998; Jokela et al. 2009), we predicted that infection frequency would be a significant predictor of overall genotypic diversity in natural populations. Consistent with previous work (e.g., Lively and Jokela 2002; King and Lively 2009), we found that the frequency of sexual individuals increased with the prevalence of infection (Fig. 1A). We also found that diversity among clonal lineages increased with infection prevalence (Fig. 1B,C). These results suggest that the level of clonal polymorphism and the maintenance of sex depend on local parasite pressure.

Although obligate sexual snails are presumably the source of new clones in P. antipodarum (Dybdahl and Lively 1995a; Neiman et al. 2005, 2010), the local frequency of sexual individuals did not significantly contribute to the observed variation in clonal diversity (Tables 3 and 4). This result is similar to observations of aphid populations containing both obligate and cyclical parthenogens (Vorburger 2006). A high rate of clonal lineages arising from sympatric sexual ancestors could potentially increase diversity (Vrijenhoek 1998; Simon et al. 2003), or perhaps cryptic sex may contribute (D'Souza and Michiels 2006), but it seems that infection-mediated rare advantage drives the pattern of clonal diversity in these snail populations.

The possibility that parasites promote clonal diversity poses a new problem for the Red Queen hypothesis. As the diversity among clonal lineages increases, the advantage to cross-fertilization in the sexual population should be reduced. In simulation models, clonal diversity can increase under parasite-mediated selection until the sexual population is eliminated by a diverse assemblage of clonal genotypes (Lively and Howard 1994). Hence it would seem that some other mechanism may be required to eliminate clones at least as fast as they are generated to explain the long-term persistence of sexual reproduction. It has been suggested the Muller's Ratchet could operate to eliminate clones in the long term, and that parasite-driven bottlenecks of clonal populations might increase the rate of mutation accumulation in clones (Lively and Howard 1994; Howard and Lively 1998). The rate of origin and extinction of clones in Potamopyrgus is presently unknown, but the distribution of “ancient” asexual lineages in this snail suggests that they are restricted to populations where parasites are rare (Neiman et al. 2005).

In summary, parasite-mediated selection against common host genotypes should, under theory, contribute to the maintenance of genetic diversity and sexual reproduction (Haldane 1949; Hamilton 1980; Bell 1982). Our study supports this ecological hypothesis, and finds further that parasites might also explain high levels of clonal diversity.

Associate Editor: M. Wayne


We thank J. van Berkel for providing laboratory facilities in New Zealand and K. Klappert for help with collections in the field. S. R. Hall provided helpful advice on statistical analyses. Funding was provided by NSF (DEB-0640639 to CML and JJ), Swiss NSF (31003A-129961 to JJ), and NSERC Canada (KCK).