Although ∼1 in 10,000 animal species is capable of parthenogenetic reproduction, the evolutionary causes and consequences of such transitions remain uncertain. The microcrustacean Daphnia pulex provides a potentially powerful tool for investigating these issues because lineages that are obligately asexual in terms of female function can nevertheless transmit meiosis-suppressing genes to sexual populations via haploid sperm produced by environmentally induced males. The application of association mapping to a wide geographic collection of D. pulex clones suggests that sex-limited meiosis suppression in D. pulex has spread westward from a northeastern glacial refugium, conveyed by a dominant epistatic interaction among the products of at least four unlinked loci, with one entire chromosome being inherited through males in a nearly nonrecombining fashion. With the enormous set of genomic tools now available for D. pulex, these results set the stage for the determination of the functional underpinnings of the conversion of meiosis to a mitotic-like mode of inheritance.

SEVERAL of the central unresolved issues in evolutionary genetics concern the evolution of sexual vs. asexual reproduction. Although substantial attention has been given to the selective advantages and disadvantages associated with meiotic recombination (e.g., Peck 1994; Barton and Charlesworth 1998; Otto and Barton 2001), less clear are the molecular/cytogenetic mechanisms that lead to the conversion of an ancestrally meiotic condition to a derived form of obligate asexuality, in particular, reproduction via the use of unfertilized eggs (obligate parthenogenesis) (Suomalainen et al. 1987). A major impediment to resolving these matters is the general difficulty in performing genetic analysis with purely asexual eukaryotes. However, a few asexual species exhibit biological features that provide a potential entrée into the mechanisms underlying sexual-to-asexual transitions. Such species include asexual lineages that retain the capability of producing fully functional males (via an environmental sex-determination pathway). One such species, the aquatic microcrustacean Daphnia pulex, is the subject of the following investigation.

Nearly all members of the Cladocera are capable of both sexual and asexual reproduction, with the former being triggered by environmental induction of male and haploid resting-egg production. However, the D. pulex complex is unique in harboring numerous lineages in which females have lost the ability to engage in meiosis and instead produce diploid resting eggs by a mechanism that is genetically equivalent to mitosis (Hebert and Crease 1980). Moreover, many lineages of this type retain the ability to produce males, at least some of which are capable of meiotic production of haploid sperm (Hebert and Crease 1983; Innes and Hebert 1988). Such sex-limited meiosis suppression provides a potentially powerful mechanism for the transformation of an entire sexual population to asexuality, as males from otherwise obligately asexual lineages backcross to sexual females and in doing so convert a proportion of the hybrid progeny to obligate asexuality. Because this process could eventually assimilate much of the diversity of a sexually reproducing species into a diverse clonal assemblage, such a drive mechanism might explain the widespread geographic distribution of numerous obligately asexual lineages of D. pulex throughout eastern Canada into the midwestern United States (Hebert et al. 1988).

Under the simplest model envisioned for this system, sex-limited meiosis suppression is conferred by a dominant allele at a single locus, such that a cross between a sexual female and an adventitious male from an asexual lineage is expected to yield a 1:1 ratio of obligately asexual and cyclically parthenogenetic progeny (Richards 1973; Jaenike and Selander 1979; Hebert 1981). Crosses between the two breeding systems have led to the synthesis of novel obligate parthenogens, demonstrating that successful transmission of meiosis-suppressing factors is indeed possible, although the number of observations is too small to either confidently accept or reject the “dominant sex-limited meiosis suppressor” model (Hebert et al. 1988). Here we apply a large battery of molecular markers to a broad geographic survey of D. pulex to gain insight into the likely mechanism of meiosis suppression (monogenic vs. polygenic), the approximate ages of individual asexual lineages, and the potential source of origin of meiosis suppression in the system. The recent development of a wide array of genomic tools, including a complete genome sequence, a genetic map, and a well-characterized, high-coverage cDNA library, makes this a powerful system for understanding the genetic basis of obligate parthenogenesis.


The work described herein was based on the analysis of clones collected from broad geographic surveys ranging from the midwestern United States to eastern Canada, made in 2003 to 2004 and subsequently maintained in the laboratory as clonal isolates (supplemental Table 1). In accordance with prior observations, clones were deemed to be obligately asexual if they were able to produce resting eggs in the absence of males, contrary to the situation in cyclical parthenogens, where fertilization is required before release of an egg into an ephippium (Paland et al. 2005).

DNA extraction and amplification:

In an effort to find allele-specific associations between breeding systems and molecular markers, we initially relied on genetically mapped microsatellite loci reported on in Cristescu et al. (2006) and then supplemented such analyses with additional physically mapped markers derived from scaffolds associated with the fully sequenced D. pulex genome (supplemental Table 2). Primers for genetically mapped microsatellite loci were obtained from Cristescu et al. (2006). Those for additional markers, which were added to further saturate the physical map prior to analysis (but with a few added post hoc in regions identified as highly informative in the first pass), were created using the program Primer3 (Rozen and Skaletsky 2000), following BLASTN searches of the D. pulex genome sequence on WFleaBase (http://wfleabase.org), using scaffolds with known chromosomal affinities. Locus-specific amplifications were carried out using the 12-μl polymerase chain reaction described in Cristescu et al. (2006). The touchdown thermal cycle program began with a 3-min denaturation at 95°, followed by 15 cycles of 35-sec denaturation at 95°, 15 cycles of 35-sec annealing at 56° (the temperature was reduced by 1° for each of the 14 remaining cycles), and 15 cycles of 45-sec extension at 72°. The latter step was followed by 45 cycles of 95° for 35 sec, 48° for 35 sec, and 72° for 45 sec and one final extension cycle at 72° for 10 min.

Primer sequences for mitochondrial genes were found, using the complete mitochondrial genome sequence (Crease 1999) and the program Primer3. Mitochondrial genes were PCR amplified, using the same 12-μl polymerase chain reaction noted above with a thermal cycle program beginning with 2 min of denaturation at 94° followed by 30 cycles of 30-sec denaturation at 94°, 30 cycles of 30-sec annealing at 51°–61°, 30 cycles of 2-min extension at 72°, and a final cycle of 5-min extension at 72°. Different annealing temperatures were used to optimize amplification of the five genes: 61° for Cox I, 58° for Cox II, 51° for Cox III, 61° for CytB, and 51° for ND5.

Genomic DNA was extracted using a modified cetyltrimethylammonium bromide (CTAB) extraction method as in Cristescu et al. (2006). Eight to 10 adult individuals were extracted for each sample.

Genotyping and sequencing:

Microsatellite loci were labeled using the M13(-21) fluorescent labeling system and genotyped on an ABI 3730 genotyper. The products of four loci (each labeled with a different fluorescent dye) were combined and diluted 60- to 100-fold. One microliter of this dilution was then added to 8.95 μl of water and 0.05 μl of GeneScan-500 Liz size standard. This solution was denatured for 5 min at 93° and then cooled immediately on ice. Genotyping data were analyzed using Genemapper v.4.0 software (Applied Biosystems, Foster City, CA). In the first pass of such analyses, any potential triploids were reextracted and regenotyped to eliminate the possibility of contamination.

Mitochondrial genes were sequenced using a 10-μl reaction mixture with 1 pmol primer, 0.5 μl of BigDye Terminator v3.1 (Applied Biosystems), and 1 μl (10–20 ng) of amplified PCR product. A two-step thermal cycle program was used for the sequencing reaction, beginning with a 2-min 96° denaturing step, followed by 27 cycles of 96° for 30 sec, 60° for 4 min, and then one cycle of 10° for 3 min. Sequence data were analyzed using CodonCode Aligner.

Statistical methods for sequence and microsatellite analysis follow the procedures outlined in Lynch et al. (1999) and utilize software developed by the first author.

Sexual × asexual crosses:

In an effort to test the ability of sons of obligately asexual females to mate with sexual females and to evaluate the genetic consequences of such events, hybrid crosses were carried out by combining males from an obligately asexual clone with females from a cyclically parthenogenetic clone in a 1:5 ratio. To ensure that crosses could proceed in only one direction, the sexual clones chosen for use were usually incapable of male production, as such variants are frequent in natural populations, independent of the sexual/asexual breeding system. Crosses were performed at 20° on a 12-hr light:12-hr dark cycle and checked every 2 days for the removal of any live young. Resting eggs were collected and placed in the dark at 4° for 2 weeks, after which the eggs were removed from their ephippium and then exposed to high light at 15° with a 12-hr light:12-hr dark cycle.

To assist in the interpretation of the following results by those unfamiliar with the Daphnia system, an overview of the alternative breeding systems is provided in Figure 1.

Figure 1.—

Schematic of the breeding systems of the alternative forms of Daphnia pulex. Note that (1) cyclical parthenogens are not truly “cyclical,” as they are capable of switching to sexual reproduction at any time, and (2) not all cyclical or obligate parthenogens produce males, although males are essential for resting-egg production by cyclical parthenogens and nonessential in obligate asexuals.


Phylogenetic distribution of meiosis suppression:

Under the assumption that obligate asexuality arises either by de novo cytogenetic events within sexual lineages or by transmission of sex-limited meiosis-suppressing genes via asexually produced males, the maternally inherited mitochondrial genome is expected to provide an unbiased view of the phylogenetic background within which asexuality has arisen. Like several preceding studies (e.g., Crease et al. 1989; Paland et al. 2005), our results indicate that obligate asexuality is widely distributed throughout the entire D. pulex lineage, with individual asexual lineages more or less randomly distributed among those reproducing by cyclical parthenogenesis (Figure 2). All lineages of both types from the midwestern United States, Maine, and Ontario are contained within a single clade, with two other distinct clades appearing at the base of the tree, one from Oregon and the other from northeastern Canada (Quebec and New Brunswick). On the basis of the reproductive modes of outgroup species, the ancestral state of D. pulex is inferred to be cyclically parthenogenetic, although the northeastern Canadian lineage appears to be nearly entirely asexual, with the possibility of a single reversion to sexuality appearing in a clone from Michigan.

Figure 2.—

A phylogeny of cyclically parthenogenetic (dotted lines, regular type) vs. obligately asexual (solid lines, boldface type) lineages based on mitochondrial sequence (including both silent and replacement sites). The scale is in units of substitutions per nucleotide site, and the bootstrap support for individual nodes is given on internal branches (in percentage). Tree construction used the neighbor-joining method (Saitou and Nei 1987), as implemented in MEGA (Tamura et al. 2007). Ages of various clades are determined from silent-site divergence alone, as described in the text. The base of the tree is rooted with sequences from two clones of Daphnia melanica. Clone notation: state/province, clone number, breeding system (A, obligately asexual; S, cyclical parthenogen).

To estimate the times of origin of the various asexual clades, we employed as a molecular clock the only direct estimate of the rate of nucleotide substitutional mutation for the mitochondrial genome, 9.7 × 10−8/site/generation, derived from the nematode Caenorhabditis elegans (Denver et al. 2000). Assuming neutrality, substitutions at silent sites in protein-coding genes are expected to accumulate at this rate, providing a basis for estimating the timing of internal nodes of a phylogenetic tree on the basis of silent sites alone. Such analyses suggest that all but the New Brunswick/Quebec clade have depths in the range of 12,000–58,000 generations, whereas the northeastern Canada clade extends back ∼172,000 generations (Figure 2). These estimates may be downwardly biased, as indirect estimates of the mitochondrial silent-substitution rate derived from interspecific comparisons and assumed divergence dates based on the fossil record average 4.7 × 10−8/site/year for four groups of invertebrates (Lynch et al. 2006). Assuming 10 D. pulex generations per year, the latter rate leads to estimates ∼21 times greater than those given in Figure 2. Even at this extreme, however, the divergence times of most asexual clades are still in the range of 25,000–122,000 years. Moreover, regardless of which set of estimates is deemed most appropriate, they are both upwardly biased because the precise timing of the transition to asexuality within individual clades must postdate the base of the clades, and more intensive clonal sampling could break up apparently monophyletic lineages of asexuals. Thus, taken together, these results imply a relatively short evolutionary longevity for individual asexual lineages, while not ruling out an ancient origin for the meiosis suppressor itself.

Further inferences from divergence of nuclear markers:

One of the central goals of this study is to identify markers associated with obligate asexuality. However, a concern with any attempt to identify marker–phenotype associations over a broad geographic range is the possibility of spurious correlations resulting from geographic divergence of phenotypes, in this case of alternative breeding systems. This appears not to be a significant issue in this study, as chromosomewide estimates of Gst (a measure of population subdivision, on a 0–1 scale; Nei 1987), contrasting all obligate asexuals with all cyclical parthenogens, are all <0.08 for nuclear microsatellite loci after a few very strong associations (below) are eliminated from the analysis. The average Gst is just 0.023 (SE = 0.004) across the entire genome (Figure 3). In other words, with the exception of a few candidate regions discussed below, only ∼2% of the total variation at microsatellite loci in D. pulex is associated with breeding-system differences. Thus, while prior Gst estimates for populations within breeding systems are large for this species, typically on the order of 0.3–0.5 for a region the size of a few U.S. states (Lynch et al. 1999), the additional subdivision between breeding systems is negligible.

Figure 3.—

Estimates of the degree of average marker subdivision between the aggregated sets of sexual and asexual lineages.

Once locked into a regimen of obligate asexuality, a diploid locus is expected to experience an increased level of allelic divergence owing to the accumulation of nonsegregating mutations, unless gene conversion is of overwhelming importance. Consistent with this expectation, the average microsatellite heterozygosity in asexual derivatives of D. pulex exceeds that in sexual clones on 11 of 12 chromosomes, although not greatly so, with an average elevation of ×1.21 (0.04) or, in absolute terms, of 0.092 (0.020) (Figure 4). The divergence of allele sizes within individuals is also greater in asexuals than in sexuals on 11 of 12 chromosomes, but again only to a small extent, with an average of 0.97 (0.30) additional nucleotides separating any two alleles in clones of the former (Figure 4). Such differences are most notable on chromosome IX.

Figure 4.—

Estimates of the average degrees of marker heterozygosity and allelic-size divergence for each chromosome with respect to the alternative breeding systems. The dashed line in the top denotes the level of heterozygosity expected in a purely asexual lineage under mutation–gene conversion equilibrium.

These low levels of microsatellite divergence both within asexual lineages and across the two breeding systems are qualitatively consistent with our conclusion that individual asexual lineages are relatively young, but also yield to more quantitative perspective. Given an initial level of heterozygosity H0, and per-generation rates of mutation and gene conversion, u and c, respectively, the expected heterozygosity within an asexual line of descent can be described asMathwhich generalizes toMathwhere H* = 2u/(c + 2u) is the expected level of heterozygosity at conversion–mutation equilibrium. Although this expression ignores the possibility of loss of heterozygosity by back mutations, this assumption appears reasonable for moderate timescales given that most microsatellite mutations in Daphnia involve multiple-step changes and that almost all single-step changes involve single-repeat insertions to the exclusion of deletions (Seyfert et al. 2008). Results from long-term mutation-accumulation experiments indicate that, for the loci involved in this study, average u = 7.1 (2.6) × 10−5/allele/generation (Seyfert et al. 2008) and that the average rate of loss of heterozygosity by gene conversion (c) = 1.2 (0.6) × 10−4/allele/generation (Omilian et al. 2006).

Under this model, the expected equilibrium level of microsatellite heterozygosity within an obligately asexual clone is 0.523, whereas the observed average is 0.508 excluding the highly divergent chromosome IX (discussed below). Assuming that newly arisen asexual lineages have an average level of heterozygosity equal to that in the sexual subdivision of the species (0.431), the time to achieve the observed average level in asexual lineages is just 4487 generations. Because there is a slight level of divergence among sexuals and asexuals, as reflected in an average Gst of 0.023, a newly arisen obligate asexual resulting from a backcross to a sexual individual is expected to have a level of heterozygosity elevated above that in the sexual species by ∼0.011, but this reduces the prior estimate only to 4090 generations.

Although the standing levels of heterozygosity for most chromosomes in the sexual species are somewhat smaller than the equilibrium expectations given above (Figure 4), this is at least in part because random genetic drift among segregating alleles constitutes an additional mechanism of loss of variation from sexual populations (which does not apply to the heterozygosity within a nonsegregating asexual lineage). An average additional loss of heterozygosity of 6.8 × 10−5/locus/generation is sufficient to explain the observed levels of variation within sexual individuals, which in turn implies average effective population sizes of ∼7400 clones.

Association mapping of meiosis-suppressing loci:

Given the results noted above, any marker exhibiting a Gst significantly >0.08 can be viewed confidently as residing in a candidate-gene region. Relying on this conservative criterion, several potential regions for the determinants of meiosis suppression are apparent, including small sections of chromosomes V, VIII, and X (Figure 5). Most notable, however, is chromosome IX, which exhibits highly significant associations over its full length, with only three exceptions appearing near the tips. For the most significant loci on all of these chromosomes, one highly informative marker allele is present in all to nearly all obligate asexuals and absent to nearly absent from all sexual members of the species (Table 1).

Figure 5.—

Chromosomal distribution of marker significance associated with obligate asexuality vs. cyclical parthenogenesis. Bars on the left denote regions containing markers with asexual–sexual Gst values significant at the level P < 0.001 and extend to the midpoints with the first flanking markers on both sides deemed as insignificant. The statistic D is a more conservative measure of significant phenotypic association, the number of standard errors by which the asexual–sexual measure of Gst exceeds the upper bound on the baseline value (0.08).

View this table:

Highly diagnostic microsatellite markers, with the fractional representation among sampled alleles given for each of the two reproductive systems

For each of the diagnostic marker alleles, a 200-kb window around the marker was searched for highly conserved proteins with known involvement in reproduction, using gene predictions and their imputed functions derived from the full-genome sequence of this species. For each marker, we found putative homologs of proteins involved in a variety of processes, including meiosis, gametogenesis, DNA repair, and sister-chromatid cohesion (supplemental Table 3). For example, contained within the significant region on chromosome V is an ortholog of Cdc6, a master regulator of DNA replication known to have a critical role in meiotic spindle formation in the mouse (Anger et al. 2005). Putative orthologs of a number of proteins with characterized meiotic functions, including polo kinase, pelota, and Esco1, are located on chromosome IX.

Residing within the significant region on chromosome VIII are two very relevant proteins, Brca1 and Rec8. Extensive studies have revealed the involvement of Brca1 in DNA repair and in the silencing of unsynapsed meiotic chromosomes (Turner et al. 2004), and equally intriguing is the role of Rec8 in sister-chromatid cohesion during meiosis. Rec8 is conserved from yeast to mammals, where it replaces the mitotic Rad21 in the multiprotein cohesin complex (Revenkova and Jessberger 2005). The D. pulex genome harbors three copies of Rec8, including two highly similar copies in a head-to-head arrangement, whereas only one copy is found in other well-studied eukaryotes (Revenkova and Jessberger 2005).

This unusual situation prompted us to begin a molecular analysis of the Rec8 loci in cyclical and obligate parthenogens. Application of PCR primers flanking the highly similar duplicated regions in the two Rec8 copies on scaffold 77 revealed a striking length polymorphism ∼650 bp upstream of one of the copies. In every cyclical parthenogen tested (n = 38), a band of 1.4 kb was amplified in this region, while all obligate asexuals tested (n = 42) revealed the 1.4-kb band as well as an additional 2.7-kb band. Preliminary sequence analysis of a small number of genotypes demonstrates that the polymorphism results from a 1.3-kb insertion, and work is ongoing to characterize the expression differences associated with this polymorphism.

Asexual × sexual crosses:

We were able to produce 51 hybrid progeny through several different crosses involving males produced by otherwise obligate asexuals and females from cyclically parthenogenetic lineages. Because a large fraction of the hybrids were the products of unreduced male gametes, these numbers are far too small for reliable analyses of segregation ratios, but the resultant observations do resolve several issues. First, all progeny resulting from these crosses inherited male-specific markers, ruling out the possibility of gynogenesis. Second, there is no evidence that any marker strongly associated with obligate asexuality is subject to a strong drive process. In all informative cases, both alternative alleles within males were observed to be transmitted into at least some progeny. This tentatively suggests that the asexual-specific markers do indeed reflect their adjacency to factors causally involved in meiosis suppression, rather than simply being residents of chromosomal segments that happen to have elevated delivery rates to successful sperm (e.g., chromosomes experiencing male-specific meiotic drive). Third, although the data are limited, the diagnostic markers associated with chromosome IX (as mapped in sexual crosses) appear to be strongly linked on the same chromosome, rather than being distributed over both members of homologous pairs, and are therefore largely heritable as a unit, albeit with some recombination (Table 2). Prior work indicates that chromosome IX recombines normally in cyclical parthenogens, as demonstrated by our ability to discriminate map positions (Cristescu et al. 2006). Fourth, as noted above, some males experience aberrant gametogenesis, as reflected in the presence of allelic pairs of unambiguous paternal markers in hybrid progeny, indicative of triploidy for one or more chromosomes. Of the 2 progeny from the cross Tre1 ♂ × Pa32 ♀, one had a genotype consistent with complete diploidy, whereas the other was consistent with complete triploidy; and all 32 of the progeny from the Sed2 ♂ × Pa33 ♀ cross had unbalanced genotypes (i.e., an apparent combination of disomy and trisomy across different loci).

View this table:

Genotypes and ploidy levels of sperm contributing to progeny of crosses between males derived from obligate asexuals and females derived from cyclical parthenogens

One caveat in these analyses is that the assignment of chromosomal locations is based on a prior mapping exercise derived entirely from a sexual × sexual cross. In principle, the asexuals might have experienced some form(s) of chromosomal rearrangement that placed all of the strong markers that we identified into a single linkage group, in which case our inference of at least four independent factors associated with meiosis suppression would be in error. Although the data for informative markers off of chromosome IX in these crosses are limited, they are sufficient to demonstrate that they are unlinked to chromosome IX (Table 2). Moreover, our observations on chromosome IX itself reveal eight apparent crossover events in 15 meiotic products (Table 2), arguing against the presence of a strong crossover-suppressing rearrangement on this chromosome in asexuals. Thus, we tentatively conclude that the asexuals have not experienced major chromosomal rearrangements.


Our results are qualitatively consistent with the hypothesis that the geographical spread of a meiosis-suppressing mechanism in D. pulex initiated from a northeastern glacial refugium, proceeding westward in a stepping-stone fashion by gradual backcrossing and conversion of sexual populations, but with populations in the far western United States remaining isolated from such events to this day (Paland et al. 2005). Consistent with such an origin, the northeastern asexual clade identified in the mitochondrial phylogeny exhibits an average heterozygosity at microsatellite loci of 0.564 (0.036), which is substantially greater than that for the entire collection of asexuals (0.508) and not significantly different from the expectation under long-term mutation–gene conversion equilibrium (0.523). Also consistent with the contagious spread of obligate asexuality by backcrossing to sexual lineages is the fact that all of the highly diagnostic microsatellite markers for asexuality (Table 1) are generally present as unique, single alleles within asexual isolates, with the aggregate pool of nondiagnostic alleles at the same loci being consistent with those segregating among sexual individuals.

Although the precise nature of the mechanism of meiosis suppression in D. pulex is not yet clear, two conclusions now appear possible. First, as postulated by previous authors (Hebert 1981; Hebert and Crease 1983; Innes and Hebert 1988), meiosis suppression in D. pulex is largely sex limited. Although males produced by some asexual lineages generate chromosomally aberrant gametes, others appear to be meiotically competent, producing only diploid progeny when backcrossed to the sexual species. Second, the fact that highly significant marker–breeding system associations reside on four chromosomes implies that the capacity for meiosis suppression may be conferred by the joint action of multiple loci, rather than by the single dominant mutation envisioned by Hebert (1981).

In principle, at least one of the markers for asexuality could simply be associated with male production, as such a capacity is essential for the spread of a meiosis suppressor into the sexual segment of a species. Only a fraction of D. pulex are capable of male production, and although the genetic details remain to be worked out, the limited crosses that have been made between male- and nonmale-producing clones appear to be qualitatively consistent with a single-locus model in which individuals with male/female function are either MM or Mm and those incapable of male production are mm, with a dominant allele (or tight linkage group) M conferring the ability to produce males (Innes and Dunbrack 1993; Innes 1997; Tessier and Cáceres 2004). However, none of the significant markers that we identified in obligate asexuals were associated with the ability to produce males (revealed by application of male-inducing methyl farnesoate; Olmstead and Leblanc 2002). In addition, our broad geographic survey provides little support for the hypothesis that obligate asexuals tend to lose the ability to produce males (Innes et al. 2000), as 59% (n = 39) of the sexual clones and 52% (n = 40) of the asexual clones in our survey produce males. However, the possibility that newly arisen asexual lineages have a frequency of male production in excess of 52% cannot be ruled out.

Thus, our results support the idea that as many as four epistatically interacting factors, likely operating together in a dominant fashion, yield a system of sex-limited meiosis suppression in D. pulex. Assuming these factors are located on different chromosomes, independently assorting as they do in cyclical parthenogens, just one-sixteenth of the male gametes produced by asexual lineages would be expected to convey the capacity for meiosis suppression, in contrast to the 50% expected under the single-factor hypothesis. Consistent with this low expected yield, of the 31 hybrid progeny that we have been able to produce, only 2 have produced viable resting eggs, and both such clones were triploid (from the cross Sed2 ♂ × Pa33 ♀; Table 2), having inherited the full suite of genes from the asexual parent (none of the 15 diploid hybrids have initiated the production of resting eggs, and hence their reproductive status remains unknown). Innes and Hebert (1988) report a similarly low success rate for asexual–sexual crosses, but were able to produce 10 putatively diploid hybrid clones capable of viable resting egg production, of which 4 were obligately asexual and 6 were sexual. Although the latter ratio is close to the 1:1 expectation under the single-factor model, it would also be consistent with a four-factor model if successful resting egg production (either haploid or diploid) requires that a hybrid procure a full complement of the four genes of a specific mating type from the male (asexual) parent.

Because a polar body is extruded during the process of cell division during even normal parthenogenetic reproduction in Daphnia, it is highly likely that the transition to obligate asexuality in Daphnia does not simply involve a switch to mitosis (Weismann 1886; Schrader 1926; Lumer 1937; Zaffagnini and Sabelli 1972), and this simple observation helps reveal why the emergence of a system endowed with the capacity for the contagious spread of asexuality may require several cytological modifications. First, the production of a diploid, genetically clonal egg without fertilization requires a normal meiosis followed by the fusion of nonsister nuclei or a single equational division (as in mitosis). Thus, as there is no evidence for segregation or recombination in obligately asexual D. pulex (beyond background mitotic levels; Omilian et al. 2006), it is likely that sister kinetochores are oppositely oriented (as in mitosis) rather than co-oriented (as in normal meiosis), and that an equational division takes place.

Second, because all sister-chromatid cohesion must be lost at or before anaphase I if a single equational division is to occur, the switch to diploid, asexual resting egg production must require a modification of the mode of cohesion from its meiotic form. Thus, it is notable that two copies of the meiosis-specific cohesin, Rec8, are localized near one marker for meiosis suppression, while a putative ortholog of Polo kinase, a protein with known roles in mitotic and meiotic kinetochore function (Kamieniecki et al. 2005; Qi et al. 2006), is found on chromosome IX. In the fission yeast Schizosaccharomyces pombe, a knockout of the Rec8 locus results in the separation of sister chromatids during meiosis I, thereby eliminating the reductional phase of chromosome segregation (Watanabe and Nurse 1999), in accordance with the scenario outlined above. Such an alteration would result in a heterozygosity-maintaining mechanism of diploid egg production like that seen in obligately asexual Daphnia.

Third, an alteration that enables development to initiate spontaneously, without fertilization, is required. In principle, diploid-egg production may be sufficient to trigger downstream development, as cyclically parthenogenetic Daphnia are already predisposed to the immediate development of diploid eggs during the clonal phase of propagation.

Fourth, as noted above, if a meiosis-suppressing mechanism is to spread via backcrossing to a sexual species, obligately asexual lineages must retain the capacity for male production and such males must be capable of the meiotic production of functional haploid sperm. Because sex-specific meiosis genes are known to exist in Drosophila (Baker 1975; Bopp et al. 1999; Tomkiel et al. 2001; Hirai et al. 2004; Thomas et al. 2005), and sex-specific differences exist in meiotic progression and dysfunction in mammals (Morelli and Cohen 2005), the restriction of a meiosis-suppressing gene to one sex may entail no special requirements.

Even with as many as four independently segregating factors critical to meiosis suppression, the spread of obligate asexuality into a sexual population by repeated backcrossing can be extremely rapid, provided any immediate fitness costs to asexuality are small relative to the fraction of hybrid progeny converted to asexuality. Assuming random mating of males produced by asexuals with females of the sexual population, the change in frequency of asexuals over a generation of mating isMathwhere β is the fraction of hybrid progeny converted to obligate asexuality, the solution of which isMathfor small β. Under this model, starting from an undetectable frequency of asexuals, essentially complete loss of sexual reproduction is expected to occur within 50 sexual generations under a one- or a two-locus model (β = 0.5 and 0.25, respectively) and within 300 generations under a four-locus model (β = 0.0625) (Figure 6). Thus, because Daphnia populations typically experience at least one bout of sexual reproduction per year, even under a four-factor mechanism, complete displacement/assimilation of a sexual population by a single asexual invader can be accomplished in just a few centuries.

Figure 6.—

Dynamic takeover of a sexual population by the invasion of a male-producing obligate asexual, starting with an initial asexual frequency of 0.0001. n denotes the number of freely segregating loci containing dominant alleles essential to sex-limited meiosis suppression, with obligate asexuality being conferred only on individuals harboring such alleles at all n loci.

Finally, although our results suggest that individual lineages of obligately asexual D. pulex typically survive for no more than a few thousand years, a likely consequence of deleterious-mutation accumulation in nonrecombining genomes (Lynch et al. 1993; Paland and Lynch 2006), this does not rule out a much deeper evolutionary origin of the meiosis-suppressing mechanism itself, which may survive in the long run via periodic transfers into fresh sexual backgrounds with low mutational loads. Under this hypothesis, the hallmarks of deleterious-mutation accumulation should be enriched in the close chromosomal vicinity of the meiosis-suppressing factors, which will effectively represent nearly completely nonrecombining backgrounds (except during rare phases of transmission through males). Asexual chromosome IX, the contents of which are largely inherited together by asexuals, appears to be an especially likely candidate for such a condition.

Aside from this study, very little work has been done on the genetic basis of parthenogenesis in animals. However, observations on various plant species are consistent with the idea that the transition to parthenogenesis (apomixis) may require only a small number of key mutations. As in D. pulex, meiosis suppression in plants is typically female specific, with males producing functional haploid sperm (Grimanelli et al. 2001; van Dijk and Bakx-Schotman 2004; Catanach et al. 2006; Noyes et al. 2007; Ravi et al. 2008). In addition, female meiosis suppression in plants is frequently associated with a single dominant factor (which does not rule out the presence of a tight linkage group of several loci), and the ability to initiate development in the absence of fertilization is conferred by one additional factor at most (again, possibly a tight cluster of genes). This consistent picture of an oligogenic basis to obligate parthenogenesis, across both animals and land plants, combined with the rapidly emerging field of functional genomics, bodes well for the future elucidation of the molecular mechanisms responsible for the conversion of sexual populations to effectively clonal lineages.


We thank J. Colbourne, J. Dudycha, S. Paland, and S. Schaack for providing many of the clones included in this study and T. Crease, M. Cristescu, D. Innes, and M. Zolan for helpful comments. We are also grateful to Alice Prickett for the artwork. This project was supported by a National Science Foundation Frontiers in Integrative Biological Research grant to M.L. and colleagues.


  • Communicating editor: D. Charlesworth

  • Received November 16, 2007.
  • Accepted June 10, 2008.


View Abstract