Introduction

Ostracods are not well known among evolutionary geneticists but they deserve more attention in relation to a range of problems from the maintenance of sexual reproduction to the evolution of giant spermatozoa or the genetics of metapopulations (Butlin & Menozzi, 1999). They are small crustaceans (typically 0.5–2 mm) in which the body is completed enclosed in a bivalved carapace. They occupy a wide range of aquatic and semiaquatic habitats but most are benthic. Their excellent fossil record stretches from about 400 Ma to Recent (Moore, 1961).

Non-marine ostracods are of special interest as there have, apparently, been many independent origins of asexual reproduction within the two dominant lineages (Cypridoidea and Cytheroidea), while a third (Darwinuloidea) is one of the best available examples of an ‘ancient asexual’ lineage (Fig. 1) (Chaplin et al., 1994; Judson & Normark, 1996). These lineages have evolved separately for about 350 Myr, so it should come as no surprise if they have different evolutionary strategies. The present review deals primarily with Cypridoidea (70–80% of ostracod species in nonmarine habitats) and Darwinuloidea, since no genetic data are available on nonmarine Cytheroidea.

Fig. 1
figure 1

SEM illustrations of valves and carapaces of representative species of the three superfamilies: Limnocythere inopinata (a,b), Darwinula stevensoni (c,d) and Heterocypris incongruens (e,g). L. inopinata (females, Lake Qinghai, China): (a) carapace, dorsal view. (b) Left valve, external view. Darwinula stevensoni (females, Lake Sibaya, South Africa): (c) carapace, left lateral view. (d) carapace, dorsal view. Heterocypris incongruens (e=female, Italy; f,g=male, Algeria): (e) carapace, dorsal view. (f) RV, internal view. (g) LV, internal view. scale=321 μm for a,b; 521 μm for c,d; 833 μm for e–g.

Full size image

Asexuality in ostracods exists in several forms: there are ancient asexuals that apparently have no close sexual relatives, there are lineages with sexual congeners, and there are species that have geographically restricted sexual populations and widespread asexual populations (‘geographical parthenogenesis’; Vandel, 1928; Lynch, 1984), and whose sexual and asexual lineages sometimes coexist. There are therefore rich opportunities to study the origin and fate of asexual lineages.

The biological species concept cannot be applied to asexual lineages. Species definitions are therefore based primarily on discontinuous morphologies but, where genetic data are available, they generally support these groupings. We use currently accepted taxonomy to delimit species and to distinguish intraspecific from interspecific differentiation. The term ‘asexual species’ refers to a set of all-female lineages that form a morphological and genetic cluster that is not known to include any sexually reproducing lineages.

We consider the following questions for which genetic studies of ostracods can potentially provide answers: (i) why do some taxa give rise to asexual lineages more frequently than others, (ii) why are some asexual taxa so diverse genetically, (iii) how long do asexual lineages persist, (iv) are apparently ancient asexuals, genuinely ancient, and (v) if so, how do they manage to persist for so long?

Asexual reproduction in nonmarine ostracods

Marine ostracods most commonly reproduce sexually whereas asexual reproduction is common among nonmarine species (Bell, 1982). Current data (Horne, 1998) show that no male has been recorded in 57% of 286 species of Cypridoidea, or in 28% of 50 species of Cytheroidea in Europe. However, asexual lineages are known to occur in species that also have sexual populations, so these figures underestimate the occurrence of asexuality. The major source of genetic data on asexual nonmarine ostracods is from allozyme electrophoresis. From the first work (Sywula & Lorenc, 1982; Sywula et al., 1985), the number and range of studies has grown. The results have been reviewed previously (Havel & Hebert, 1993; Chaplin et al., 1994) and Rossi et al. (1998) provide a full list. Recently, DNA based techniques have started to provide an additional perspective (Little & Hebert, 1997; Schön et al., 1998).

Allozyme data can test the inference of asexual reproduction based on the absence of males because asexual populations consist of limited numbers of multilocus genotypes, departing significantly from both Hardy–Weinberg and linkage equilibrium. So far, all studies have found the expected differences between populations with and without males (Chaplin & Ayre, 1989; Havel & Hebert, 1989; Havel et al., 1990a, 1990b; Chaplin, 1992; Little & Hebert, 1994). This observation allows inferences to be extended to other species on the basis of sex ratios without genetic studies. In some cases, electrophoretic examination of laboratory colony material has also confirmed that the mode of reproduction is apomictic (Havel & Hebert, 1989 for Cypridopsis vidua; Rossi & Menozzi, 1990 for Heterocypris incongruens; Chaplin, 1992 for Candonocypris novaezelandiae, see noteFootnote 1). No exceptions, due to mutation or recombination, have been observed (except perhaps one, Chaplin, 1992). The total number of alleles×generations studied is greater than 104 (≲103 pairs in the case of recombination), so these data indicate that mutation rates are not unusually high and that automixis or somatic recombination are rare, if they occur at all. On this basis, all unisexual nonmarine ostracods can be assumed to reproduce apomictically, although the extrapolation is relatively insecure for the distantly related, ancient asexual darwinulids, and this is generally consistent with the clonal genotypes observed in the field.

Allozyme data define multilocus genotypes within apomictic populations. These genotypes are termed ‘clones’, although they are likely to be assemblages of related lineages that could be distinguished if additional loci were studied. Mitochondrial DNA does distinguish clones within such multilocus genotypes in Heterocypris incongruens (Chaplin & Hebert, 1997) and in Eucypris virens (I. Schön, pers. obs.). RAPDs also distinguish more clones (Rossi et al., 1998).

Some apparently apomictic ostracod populations or species contain occasional rare males. They may well be dysfunctional, especially in older asexual lineages, since mutations are likely to accumulate in pathways necessary for exclusively male functions. However, rare sexual reproduction can be important in preventing mutational deterioration of otherwise apomictic lineages (Hurst & Peck, 1996). Little & Hebert (1996) claim that a single potential male observed in 150 years of sampling Darwinula stevensoni challenges its status as an ancient asexual (a claim refuted by Rossetti & Martens, 1996, and Schön et al., 1996). Genetic data can resolve this debate.

Two other features of ostracods should be mentioned that are relevant to interpretation of the allozyme data. First, many ostracods have multiple sex chromosomes. An extreme example is Heterocypris incongruens where females have 2n=8A+12X and males have 2n=8A+6X+1Y (Dietz, 1958). This means that sex linkage is common for allozyme loci in sexual species (e.g. Turgeon & Hebert, 1994). Second, allozyme data, combined with direct chromosome counts and optical densitometry measurements of DNA content per nucleus (Turgeon & Hebert, 1994), have demonstrated high frequencies of polyploidy in asexual clones but not in sexual populations. The pattern observed in Heterocypris species (Turgeon & Hebert, 1994) is apparently widespread: a sexual species H. glaucus has 2n=22 (female) while asexual H. incongruens have very variable chromosome numbers with modes at 20–22, 29–30 and 43. These presumably represent diploid, triploid and tetraploid groups but with variability in chromosome number associated with multiple sex chromosomes. In this study, seven out of 10 clones of H. incongruens were polyploid. Similarly, nine out of 11 clones in Bradleystrandesia fuscata were polyploid (Turgeon & Hebert, 1995).

Clonal diversity

High clonal diversity is expected, and observed, in cyclical parthenogens (Hebert et al., 1988) and in taxa where new parthenogenetic lineages are continuously generated by interspecific hybridization (Avise et al., 1992; Vrijenhoek, 1994). However, it was not expected in obligately apomictic taxa. The initial indications of high numbers of clones in ostracods were therefore surprising (Chaplin & Ayre, 1989; Havel & Hebert, 1989) although there is actually no general reason to expect asexual taxa to be less genetically diverse than sexual taxa (Kondrashov, 1993). In fact, there is a wide range of clonal diversity both among species and among populations within species.

Rossi et al. (1998) review data for clonal diversity in 42 studies of 29 species. The number of clones observed ranges from 1 to 211 but it is difficult to compare these figures directly. In a multiple regression analysis, 40% of the variation is explained by the numbers of populations, individuals, and polymorphic loci studied, with number of loci having the largest effect (b=5.81±1.66, P<0.001), contrary to the observation of Havel & Hebert (1993). However, several species depart strongly from this regression: Eucypris virens (211 clones), Bradleystrandesia reticulata (185), Cypridopsis vidua (77), and Heterocypris incongruens (68 clones in Europe) have many more clones than expected, while Candonocypris novaezelandiae (26 clones), Bradleystrandesia fuscata (11), Cytherissa lacustris (6), Darwinula stevensoni (7) and Heterocypris incongruens (10 clones in North America) have fewer clones than expected.

One can ask two, partly distinct questions: how is clonal diversity generated and how is it maintained? A subsidiary problem is what causes the range of diversities among species. Three mechanisms for the generation of diversity have been considered for ostracods: (i) accumulation of genetic differences among clones derived from a common parthenogenetic ancestor; (ii) multiple origins of asexuality from a sexual ancestor; or (iii) input of genetic variation through hybridization. In terms of maintenance, there are essentially two possibilities: either clones are differentially adapted and diversity is maintained by selection, or they are ecologically indistinguishable and standing diversity is a balance between input of variation and stochastic extinction. This second alternative may be considered equivalent to the idea of a ‘general purpose genotype’ (Lynch, 1984): a single well-adapted genotype that has accumulated neutral variation since the spread of the most recent advantageous mutation. These possibilities will be examined in turn, starting with the origin of clones and then moving to the question of maintenance.

Origins of clones

The simplest explanation for clonal diversity in an apomictic parthenogen is the accumulation of mutations during descent from a single origin. High clonal diversity would imply a long history of asexuality.

Turgeon & Hebert (1995) constructed a phenogram, based on allozyme genetic distances, for six species of Bradleystrandesia, three sexual and three asexual. The analysis implies separate origins for each of the three presumed asexual species but, within each species, is consistent with accumulation of differences among clones following a single origin. A single origin of asexuality is implied by cladograms based on mitochondrial DNA RFLPs for a sample of clones of Heterocypris incongruens (Chaplin & Hebert, 1997). Mitochondrial DNA and allozyme clonal definitions and implied relationships were fully compatible, as expected for apomictic lineages that share a common ancestor (with the possible exception of one triploid clone which was highly heterozygous).

There can be little doubt that the transition from sexual to asexual reproduction has happened many times in the history of nonmarine ostracods. The wide taxonomic distribution of asexual lineages provides clear evidence on the large scale (Chaplin et al., 1994). Within species, a study of Eucypris virens (Schön et al. in prep.), using mitochondrial cytochrome oxidase I and nuclear ITS1 sequence data, has included conspecific sexual and asexual lineages. In this case, a minimum of 3 (and up to 8) origins of asexuality are implied within the species. Sexual populations are known to exist in Heterocypris incongruens (Fryer, 1997) and may well also exist for the three ‘asexual’ Bradleystrandesia species and simply not yet have been discovered. Their inclusion in phylogenetic analyses may well increase the inferred number of independent intraspecific origins of asexuality as in Eucypris.

In asexual vertebrates, clonal diversity is generated by repeated origins of asexual lineages from hybridization between sexual species (Avise et al., 1992; Vrijenhoek, 1994). This leaves a characteristic signature of very high heterozygosity, with diploid asexual lineages containing alleles typical of both parental species, that has not been observed in ostracods. However, there is growing evidence for hybridization between asexual females and males of the same, or closely related species. Fertilization of the unreduced diploid egg by a haploid sperm results in a triploid, asexual offspring with a novel genotype, thus contributing to clonal diversity. The first suggestion of this mechanism came from the observation that the most clonally diverse ostracod species have sexual congeners (Havel & Hebert, 1989, 1993; Havel et al., 1990a). Furthermore, some asexual polyploid lineages contain alleles typical of related species (e.g. Heterocypris incongruens, Turgeon & Hebert, 1994; Bradleystrandesia reticulata, Turgeon & Hebert, 1995). The clonally diverse species Heterocypris incongruens and Eucypris virens are geographical parthenogens, with localized sexual populations in the southern parts of their ranges and widespread asexual populations (Bauer, 1940; Baltanas, 1996) while Cypridopsis vidua and Bradleystrandesia reticulata are in genera that include sexual species. Although clonally diverse asexual species do not always have currently known distributions that overlap with sexual species (Havel et al., 1990a), this may be due to limited sampling, or their ranges may have overlapped in the past. Some species with low clonal diversity occur in genera with sexual species suggesting that close sexual relatives may be a necessary, but are not a sufficient, condition for high clonal diversity (Havel & Hebert, 1993).

Further evidence for hybridization between sexual and asexual lineages comes from comparison of mitochondrial and nuclear DNA phylogenies in Eucypris virens (Schön et al. in prep.). In an apomictic lineage, all phylogenetic trees, including mitochondrial trees, should be congruent (as in Heterocypris incongruens, Chaplin & Hebert, 1997). However, in Eucypris virens a small group of clones is placed differently in COI and ITS1 based phylogenies, which are otherwise congruent. The most likely explanation is, again, hybridization between a male (in this case conspecific but nevertheless quite divergent genetically) and an asexual female. However, this would generate offspring with the mtDNA of the asexual lineage and the nuclear alleles of both lineages. Since direct sequencing gives clean ITS1 sequence, there is no evidence that these clones are heterozygous or indeed that they are polyploid, although polyploids are known in E. virens (Tetart 1978; Rossi et al., 1998: 14% of individuals screened). The phylogeny suggests that the hybridization happened a long time ago. Perhaps, there has been sufficient time for the ITS repeats to be homogenised: homogenization of rDNA copies on the same and on different chromosomes continues in obligately asexual Daphnia (Crease & Lynch 1991). Alternatively, the incongruence of the phylogenies may indicate that some of the currently asexual lineages arose from sexual ancestors much more recently than is implied by their genetic divergence from the extant sexual populations.

Maintenance of clonal diversity

In the absence of ecological differentiation between clones, clonal diversity should be a function of rates of origin and loss of clones, the latter being determined by population size. Each of these parameters is hard to estimate. Nevertheless, it should be possible to test for departures from the null hypothesis of ecological equivalence of clones, for example by comparing rank-abundance distributions within populations with expectations from ecological theory, or by comparing clonal diversity across water bodies of different sizes. Low clonal diversity in ponds on Caribbean islands relative to the North American mainland was interpreted by Little & Hebert (1994) to reflect bottlenecks during colonization of the islands. Rossi et al. (1998) observe distinctly different rank-abundance patterns in three species of ostracods but systematic comparisons and tests of hypotheses have yet to be conducted.

Correlations between clonal diversity and parameters expected to reflect environmental variability have been considered. Chaplin & Ayre (1989) found no convincing association between clonal diversity and permanence of water bodies in Candonocypris novaezelandiae. Similarly, Little & Hebert (1994) found no association between diversity and latitude.

Some studies suggest that clones are not ecologically equivalent. Rossi and coworkers (Rossi & Menozzi 1993; Rossi et al., 1993) have shown that two clones of Heterocypris incongruens in Italy, one abundant in winter (W) and one in summer (S) in rice fields, have different life history characteristics, responses to temperature and responses to photoperiod. These differences are entirely compatible with the annual cycle in their relative frequencies. However, other, rarer clones exist with them in the rice fields giving a total of 5 clones in one case and 19 in the other while nearby pond populations typically have a single clone. The W and S clones are among the most divergent observed (Nei's I=0.46): their long-term coexistence, and hence divergence, may have been permitted by ecological differentiation. Similar, but less complete data on ecological differentiation among clones are now available for Eucypris virens (Otero et al. 1996) and Limnocythere inopinata (Yin 1997). Even in the clonally conservative ancient asexual, Darwinula stevensoni, there is evidence of an association of some clones with riverine rather than lacustrine environments (Rossi et al., 1998).

Rossi et al. (1998) have noted a common pattern in several clonally diverse species: there are a few common widespread clones and many rare local clones. This may reflect differences in colonizing ability, as suggested by Chaplin & Ayre (1989), or it may reflect fitness differences among clones. Widespread clones could possess a ‘general purpose genotype’ (Lynch, 1984) allowing them to exploit a range of environments while the rarer clones are adapted to geographically restricted or atypical environments.

The real difficulty here is that allozymes, or other markers, are likely to distinguish large numbers of multilocus genotypes many of which differ from related lineages only through accumulation of neutral or weakly selected alleles. The number of ecologically distinct lineages may be much less than the number of clones detectable in a sample with rare local clones being recent, ecologically equivalent, derivatives of more divergent, ecologically distinct and widespread clones. This may be especially true where clonal diversity results from divergence of lineages derived from a common asexual ancestor. Where lineages are independently derived from sexual progenitors, there may be a closer link between multilocus genotypes and adaptive characteristics, as in the clones of poeciliid fishes derived by hybridization (Vrijenhoek, 1994), because each new asexual lineage independently samples the variation present in the sexual population.

Longevity of clonal lineages

Griffiths & Butlin (1995) have used data on the occurrence of sexual and asexual ostracod species in 34 Holocene fossil sequences to compare the persistence and abundance of lineages of different reproductive mode against a background of environmental change on a time scale of thousands of years. They observed that sexual species had higher and more stable abundance, and persisted for longer than asexual species, on average. This is consistent with the argument that sexual species are better able to respond to environmental change. It is difficult to relate these data, which necessarily consider groups of clones gathered together as morphospecies, to the persistence of individual clones. However, one might expect that a clonally diverse asexual lineage, in which the clones were ecologically distinct, would persist in a changeable environment even if some clones went extinct.

Chaplin & Hebert (1997) estimated a maximum mtDNA sequence divergence of about 2.5% among clones of Heterocypris incongruens and suggested a calibration of divergence rate of 0.5% per million years. This gives an age of greater than 5 Myr for these asexual lineages (the species as a whole must be older still). In the study of Eucypris virens noted above (Schön et al. in prep.), sequence divergence is considerable: the two sexual populations studied show divergences of up to 19.6% for COI and 4.3% for ITS1 while the greatest differences within asexual lineages reach 24.3% for COI and 10.3% for ITS1. As with Heterocypris, the implication is that these asexual lineages are surprisingly old, estimated to be greater than 4 Myr for E. virens, although calibration is problematic in both cases.

These studies reveal a problem: if one accepts the possibility either of multiple origins of apomixis or of sexual lineages that are extinct or undiscovered, then the time of origin of asexuality cannot be inferred from phylogenetic analysis. Chaplin & Hebert (1997) suggested that some of the clonal diversity of Heterocypris incongruens might result from multiple origins of apomictic lineages from divergent sexual species that are now extinct. In the light of the Eucypris data, multiple origins from divergent conspecific sexual populations are more likely, and these populations may simply be undiscovered rather than extinct. For Eucypris virens itself, it is possible that genetically differentiated sexual populations were widespread in the past, whereas they are now restricted to small pockets in southern Europe and North Africa (Baltanas, 1996), and gave rise to independent asexual populations.

Although these lineages are old relative to most asexual vertebrate lineages (Avise et al., 1992) and predate the postglacial expansion which might have seemed a likely time for expansion of asexual lineages, they remain young relative to ‘ancient asexuals’, including darwinulid ostracods (Butlin & Griffiths, 1993). Judson & Normark (1996) consider darwinulid ostracods among the three best supported examples of ancient asexuals, along with bdelloid rotifers and the brine shrimp, Artemia salina. The claim for Artemia salina of 30 Myr of apomictic reproduction, based on molecular data (Perez et al., 1994), is uncertain for a lineage that most likely arose from hybridization between sexual species. In contrast, the fossil record provides evidence for the age of the darwinulid lineage which has persisted for at least 100 Myr without sex (Rossetti & Martens, 1996). Sexual dimorphism of ostracod valves enables this inference, which cannot be achieved in other ancient asexual taxa (Butlin & Griffiths, 1993; Griffiths & Butlin, 1995; Judson & Normark, 1996).

Mitochondrial COI sequence divergence is substantial both within Darwinula stevensoni (3.8%) and between species of Darwinulidae (22.2–27.7%), suggesting ancient divergence as expected. However, for the nuclear ITS1 sequence the pattern is different: between species divergence is substantial (15.4–21.0%) but there is no sequence variation within D. stevensoni whatsoever (Schön et al., 1998). One possible explanation for these observations is a low mutation rate in the nuclear genome of D. stevensoni, while an alternative is some special behaviour of the tandemly repeated rDNA sequences. The possibility that a recent selective sweep has homogenised the ITS1 sequences over the large geographical range of this study (South Africa to Finland) is contradicted by the mtDNA variability observed within D. stevensoni. In a fully apomictic lineage, a selective sweep would influence the mitochondrial genome along with the nuclear genome.

An alternative approach to the estimation of the age of apomictic lineages and, indeed, to the demonstration of long-term apomixis, has been suggested by Matthew Meselson (http://golgi.harvard.edu/meselson). In a fully apomictic lineage, not only will all loci tend to become heterozygous as mutations accumulate but also the two alleles at any one locus will become progressively more divergent. The common ancestral allele for the two alleles within an individual must have been in the original progenitor sexual population, whereas alleles in different individuals, or in different clones derived from the same transition to asexuality, share a more recent common ancestor. Meselson is testing this prediction in the most celebrated of ancient asexuals, the bdelloid rotifers (see web site and Judson & Normark, 1996). In ostracods, the only relevant data available are the ITS1 sequences for Darwinula stevensoni. Here there is no evidence for allelic divergence within individuals (Schön et al., 1998). Either darwinulids have some cryptic form of sex (for which there is no evidence; Rossetti & Martens, 1996; Schön et al., 1998), or the mechanisms of concerted evolution in tandemly repeated loci, such as gene conversion or unequal cross-overs, persist in asexual lineages, or they have unusually efficient mechanisms of DNA repair (Schön & Martens, 1998).

The possibility that some form of recombination persists in asexual lineages, raised by the Darwinula data, may be more generally important. Either somatic recombination or gene conversion, or possibly occasional automixis, could occur in unisexual populations. However, these processes have not been observed in the laboratory cultures of ostracods used to demonstrate apomixis (cited above) although they may be too rare to detect in these samples. Turgeon & Hebert (1994) suggested that a totally homozygous clone (for the loci screened) of Heterocypris incongruens might result from automixis. Polyploidy might promote somatic recombination (Little & Hebert, 1997) and a semblance of chromosome pairing has been observed in the asexual Bradleystrandesia fuscata (Tetart 1975). More generally, all of the allozyme surveys of ostracods reveal clones with different homozygous genotypes at the same locus. Apomictic reproduction, with a single origin, cannot explain this observation unless the same allele has been derived more than once by mutation. While this might sometimes occur, given the limited resolution of allozyme electrophoresis, it is unlikely to explain the widespread occurrence of the pattern. Null alleles could also generate apparent homozygotes but are generally detectable from the relative intensities of staining or the absence of bands (e.g. two banded phenotypes for dimeric enzymes). Occasional somatic recombination or automixis seem more likely explanations (and have been suggested for asexual brine shrimp, Browne, 1992), although separate origins from sexual populations could also be responsible. Superimposing allozyme data onto DNA sequence based phylogenies of clones might separate these possibilities but the mtDNA phylogeny of Chaplin & Hebert (1997) is, unfortunately, not sufficiently resolved.

Discussion

None of the ideas discussed so far really answers the first of the questions that we posed at the outset: why do some taxa give rise to asexual lineages more frequently than others? Specifically, why is asexuality so widespread in nonmarine ostracods? It seems unlikely that this is solely a response to environmental conditions, especially since there is no strong association between reproductive mode and type of water body (Chaplin & Ayre, 1989; Chaplin et al., 1994). The alternative is that some genetic feature of ostracods predisposes them to make the transition from sexual to asexual reproduction. Chaplin (1992; Chaplin et al., 1994, 1998) has suggested that there is a link between multiple sex chromosomes, skewed sex ratios, and the origin of asexual reproduction. Many sexually reproducing ostracods, including marine species, have female biased sex ratios (reviewed by Chaplin et al., 1994). These may be partly explained by greater longevity of females but could also indicate the presence of segregation distorters on the X chromosomes. This possibility is supported by skewed sex ratios, and greater than chance variation in ratios, among the offspring of single females of Candonocypris novaezelandiae (Chaplin, 1992). Whatever the cause of skewed sex ratios, they result in sperm limitation for females and may therefore provide an additional impetus towards parthenogenetic reproduction. This may be especially true where populations are highly structured since males may be absent by chance from some demes. Incidentally, female biased sex ratios may also explain another striking feature of ostracods, the giant sperm of males (Wingstrand, 1988; Butlin & Menozzi, 1999). Where there is an excess of females, sperm competition will be rare and males may invest in few large sperm rather than many small sperm, as in fish (Stockley et al., 1997). Low sperm numbers may make sperm limitation more severe. Of course, some skewed sex ratios in bisexual populations are due to the co-occurrence of sexual and asexual lineages. These cases must be accounted for, using genetic data, before conclusions can be drawn about other causes of departure from expectation.

All three explanations suggested above appear to contribute to the generation of large numbers of clones in some asexual ostracod lineages. First, at least some lineages are old so that there has been sufficient time for variability to accumulate through mutation. Second, there are multiple origins of asexuality, even within species, and third, hybridization with sexual conspecifics or congeners seems likely to introduce genetic variability into some clonal populations. Somatic recombination, gene conversion or automixis might also introduce genetic variation to asexual lineages but the evidence available to date is indirect. It is particularly important to seek ways of detecting these processes because they can be important to the persistence of asexual lineages, even if they occur at low frequency (Hurst & Peck, 1996).

Ecological differentiation, both spatial and temporal, definitely contributes to the maintenance of clonal diversity in some cases but even so can explain only a fraction of the observed variation. Systematically conducted comparisons are needed to determine how widespread clonal differentiation is across species.

There is some conflict over the age of asexual lineages. The small amount of DNA-based information so far available suggests that they are surprisingly persistent. This conflicts with the observation that asexual lineages are less persistent than sexual species in Holocene core sequences (Griffiths & Butlin, 1995). The resolution to this conflict may lie in metapopulation structure: local extinctions of asexual populations are common but the high vagility of desiccation-resistant stages allows rapid recolonization and thus persistence both of the species as a whole and of at least some of its clonal lineages (Baltanas, 1996). The widespread occurrence of common clones (Rossi et al., 1998) and the rapid colonization of new habitat patches (Fryer, 1997) provide support for this interpretation but, again, direct studies of metapopulation structure are needed, taking into account the existence of a ‘bank’ of desiccation resistant eggs in sediments (Hairston & de Stasio, 1988).

Whether ancient asexuals are genuinely ancient apomicts, and how they manage to defy the Red Queen and Muller's Ratchet (the last two of our initial set of questions), we still cannot say. Somatic recombination or automixis, if they really do occur in otherwise apomictic lineages, or improved DNA repair (Schön & Martens, 1998), may provide part of the solution. The recent start made in the study of molecular evolution in darwinulid ostracods (Schön et al., 1998) suggests that some answers may soon be forthcoming.