Reconstruction of Microsatellite Mutation History Reveals a Strong and Consistent Deletion Bias in Invasive Clonal Snails, Potamopyrgus antipodarum

Corresponding author: David Weetman, Department of Biological Sciences, University of Hull, Hull HU6 7RX, United Kingdom., d.w.weetman{at}hull.ac.uk (E-mail)

Communicating editor: G. B. GOLDING

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Direct observations of mutations and comparative analyses suggest that nuclear microsatellites show a tendency to expand, with reports of deletion biases limited to very long alleles or a few loci in multilocus studies. Here we investigate microsatellite evolution in clonal snails, Potamopyrgus antipodarum, since their introduction to Britain in the 19th century, using an analysis based on minimum spanning networks of multilocus microsatellite genotypes. British populations consist of a small number of highly distinct genotype groups with very few outlying genotypes, suggesting clonal lineages containing minor variation generated by mutation. Network patterns suggest that a single introduced genotype was the ancestor of all extant variation and also provide support for wholly apomictic reproduction within the most common clonal lineage (group A). Microsatellites within group A showed a strong tendency to delete repeats, with an overall bias exceeding 88%, irrespective of the exact method used to infer mutations. This highly unusual pattern of deletion bias is consistent across populations and loci and is unrelated to allele size. We suggest that for persistence of microsatellites in this clone, some change in the mutation mechanism must have occurred in relatively recent evolutionary time. Possible causes of such a change in mechanism are discussed.

MICROSATELLITES are short, tandemly repeated sequences of noncoding DNA that are widely used as markers in studies of population differentiation, relatedness, and genetic mapping (see GOLDSTEIN and SCHLOTTERER 1999 ). Mutations at microsatellites primarily involve slipped-strand mispairing events leading to the insertion or deletion of a number of repeat units during mitotic cell divisions (EISEN 1999 ). Many studies directly recording the direction of microsatellite mutations report a bias toward addition of repeats (e.g., WEBER and WONG 1993 ; AMOS et al. 1996 ; PRIMMER et al. 1996 ), with comparative, population genetic, and phylogenetic studies also suggesting that microsatellites show a tendency to expand (RUBINSZTEIN et al. 1995 ; AMOS 1999 ; COOPER et al. 1999 ; ZHU et al. 2000 ; NEFF and GROSS 2001 ). Examples of microsatellites exhibiting a deletion bias are rare and appear to be limited to a few loci in multilocus studies (DIRIENZO et al. 1998 ; UDUPA and BAUM 2001 ) or to very long alleles at a particular locus (WIERDL et al. 1997 ; ELLEGREN 2000 ; XU et al. 2000 ; HUANG et al. 2002 ). To our knowledge there are no published records of deletion biases across a range of nuclear microsatellites in any organism.

Naturally occurring mutations are difficult to observe because of their rarity, but population bottlenecks present the opportunity to study microsatellite evolution indirectly (ESTOUP and CORNUET 1999 ). Asexual (apomictic) species may provide particularly informative models, because genotypes are inherited as complete linkage groups, with mutation as the only source of variation between mothers and daughters (SUNNUCKS et al. 1996 ). Thus, data can be analyzed by network methods (reviewed by POSADA and CRANDALL 2001 ), treating multilocus genotypes as haplotypes, and such studies may be relatively free of the population genetic assumptions that burden some indirect analyses of microsatellite evolution (AMOS 1999 ). We utilized a severe bottleneck, arising from a founder event, to study microsatellite mutations in an invasive clone of the hydrobiid snail Potamopyrgus antipodarum. The species is native to New Zealand, where diploid sexual populations are thought to produce occasional triploid apomicts (DYBDAHL and LIVELY 1995 ). Following an introduction to the River Thames in the late 19th century (SMITH 1889 ), P. antipodarum spread rapidly throughout the United Kingdom and continental Europe (HUGHES 1996 ) and has recently colonized North America (ZARANKO et al. 1997 ). Sexual reproduction has never been recorded in United Kingdom populations (HUGHES 1996 ), and genetic studies using multilocus minisatellites (HAUSER et al. 1992 ) and randomly amplified polymorphic DNAs (RAPDs; JACOBSEN et al. 1996 ) have suggested that only a few distinct clones were introduced, corresponding to previously described morphotypes (WARWICK 1952 ).

The aims of our study were to use minimum spanning networks, based on extant multilocus genotypes, to reconstruct the evolutionary history of microsatellite mutations since the introduction of P. antipodarum to Britain. This was achieved by undertaking a survey of British populations to identify multilocus genotype groups and identify their correspondence to previously identified morphological-genetic clonal groups. We report that the populations screened contain three major clonal genotype groups, corresponding to the known morphotypes, and another previously unidentified clone, with evidence for either extremely rare clonal lineages or similarly rare interclonal hybridization. Within the most common clonal group, which appears to be entirely apomictic in our study populations, microsatellites have evolved via a mutational process highly biased toward deletion of repeats.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Samples and screening:
Samples of 100 P. antipodarum were collected from eight geographically isolated populations around Britain (locations are shown and described in Fig 1). DNA was extracted following homogenization of whole snails and the eight populations were screened at four microsatellite loci: Pa112, Pa143, Pa254, and Pa121 (extraction, PCR, and scoring procedures are described in WEETMAN et al. 2001 ). Three of these populations (TBR, IMF, and BMA), which consisted almost entirely of a single clonal group, were screened at three additional loci: Pa217, Pa56, and Pa132 (WEETMAN et al. 2001 ). Other populations were not scored at these three loci either because they contained a single clone with no microsatellite variability or because they included high frequencies of other clones in which the three extra loci could not be scored unambiguously.

View larger version (22K): In this window In a new window Download PPT slide   Figure 1. Phenetic tree based on the neighbor-joining method depicting distance (Dxy) among the multilocus microsatellite genotypes detected in eight British populations of P. antipodarum. Proportionate frequencies of genotype groups (genotypes within dashed circles) in each population are shown in the pie charts: white, clone A; gray, clone B; striped, clone C; stippled, clone D; black, other ungrouped genotypes. Sampling locations of the populations are shown on the map, their numbers corresponding to those on the pie charts. The populations are TBR, disused reservoir by River Thames in Barnes, London; CAE, stream flowing into the sea by Caerhays Castle, Cornwall; HAA, stream below Harlech Castle, Gwynedd; BMA, Budworth Mere (lake), Cheshire; IMF, river near Fleshwick, Isle of Man; WBB, coastal pond at West Barns, near Dunbar, Lothian; DUN, stream inland from Dunstanburgh Castle; STA, agricultural drainage ditches near Stainfield, Lincolnshire.

Data analysis:
From the data with four scored microsatellites, we calculated a genetic distance measure based on the number of shared alleles per locus between two individuals, D_xy (SAMADI et al. 1999 ), and produced a pairwise distance matrix from the mean D_xy over the four loci. From this matrix we constructed a tree using the neighbor-joining method (SAITOU and NEI 1987 ), with Phylip version 3.5c (FELSENSTEIN 1993 ). In the three populations (TBR, IMF, and BMA) for which data from seven loci were available, we constructed minimum spanning networks from the multilocus microsatellite genotypes, first assuming full interconnection between the populations and, second, assuming that the populations were isolated from one another. To construct the networks, we broadly followed the guidelines of EXCOFFIER and SMOUSE 1994 . Our initial step was to determine which were the nearest neighboring genotypes for each genotype, using the criterion of the fewest number of mutations. These connection data were used to produce a first spanning network. It then proved possible to infer the oldest genotype(s) on the basis of frequency, position, and number of connections and thus to infer also the direction of flow of mutations. This allowed a substantial reduction in the number of possible spanning networks, by minimizing the number of mutational steps in pathways. For the final stage of resolution we assumed that connections between low-frequency genotypes and (very) high-frequency (older) genotypes were more likely than those between two low-frequency genotypes, provided the number of mutations were equal (SMOUSE 1998 ).

In the networks, a single mutation was defined as a change in allele state from one size to another, irrespective of the number of repeat unit changes involved. This is likely to underestimate the total number of mutations, but was appropriate because an independent analysis of the data using simulations suggests that most of the loci show a mutation process deviating substantially from a single-step mutation model (D. WEETMAN, L. HAUSER and G. R. CARVALHO, unpublished results). However, this bias is likely to be relatively small because most of the changes observed involved shifts of only one to three repeat units.

From the minimum spanning networks, we counted the number of inferred mutations representing positive and negative changes in allele size. Two estimates for the percentage of deletions were calculated; the first used only the number of inferred mutations in the network, irrespective of the number of snails that shared genotypes. The second estimate assumed that the frequency of different nonancestral genotypes was proportional to the frequency of occurrence of the mutations that produced them. Therefore, a separate mutation was assumed to have given rise to each individual. The first counting method follows parsimony in that it ignores the frequency of genotypes and is likely to underestimate the total number of historical mutations. By contrast, the second method implicitly assumes that the frequency of a genotype is proportional to its age (WATTERSON and GUESS 1977 ; EXCOFFIER and SMOUSE 1994 ) and is more likely to overestimate mutations. These two estimates were calculated to provide approximate lower and upper boundaries for the unknown true number of mutations.

One-tailed binomial probabilities were computed to test whether apparent biases observed were significantly >50% of deletions. Where significant biases were found, we also calculated a lower 95% binomial confidence limit (ZAR 1984 ) for the estimate of the percentage of deletions.

We found only one case in which two alleles at the same locus changed simultaneously between an inferred mutant individual and its most probable progenitor, and for this ambiguity we recorded each mutational alternative. In a few instances within the networks, more than one alternative pathway of approximately equal probability to a genotype was found, creating loops in the network. Loops may reflect homoplasies or be the result of recombination (CRANDALL and TEMPLETON 1993 ; POSADA and CRANDALL 2001 ) so the genotypes involved were examined to determine whether the loop could be resolved by the assumption of a single recombination event. In cases of loops, the percentage of deletions was calculated once for each mutational pathway involved in each loop, creating alternative mutational scenarios. In fact, the particular alternative scenario assumed had no effect on the estimate of the percentage of deletions, until the data were subdivided into individual loci within populations, because each set of alternatives involved the same number of additions and deletions. Thus, estimates of the percentage of deletions for the maximum and minimum number of possible mutations are given only at the stage when specific loci are examined. Finally the relationship between the size of progenitor alleles (in repeat units) and the sign of the mutational change (addition or deletion) was tested by logistic regression (in SPSS version 9), using pooled data from all populations and loci to provide suitable statistical power for the test.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Genotypic composition of the sampled populations:
Thirty-four multilocus genotypes (based on four loci) were found in the British populations surveyed. Thirty of these clustered into four widely divergent groups, corresponding to four putative clonal lineages (Fig 1). Clonal lineage A was by far the most common and widespread of these groups, comprising 67% of all the snails screened, and was the only lineage present in four of the populations (TBR, IMF, DUN, and STA) and the most common in a further two (BMA and HAA). All locations dominated by clone A are freshwater sites, whereas the other two, WBB (dominated by clonal group B) and CAE (a mixture of three major clonal groups) are coastal, brackish-water locations (see Fig 1). Outside of these groupings, the four other rare genotypes identified comprised only 0.6% of the total sample. All were highly divergent from both the major groups and one another (Fig 1).

Microsatellite variation within clonal group A:
Fig 2 shows a minimum spanning network for multilocus genotypes from three of the populations (BMA, IMF, and TBR) based on seven microsatellite loci (genotypes are listed in Table 1). Two very high-frequency genotypes are central within the network and are separated by a difference (of a single step) at only one allele, with low-frequency genotypes radiating from each. This link between the two dominant genotypes represents the only parsimonious connection between the group of genotypes found in the BMA population and those in the IMF and TBR populations. The lack of interconnection between BMA and IMF confirmed our a priori expectation that the populations would be isolated, and so networks are also shown for each of the populations treated independently (Fig 3). Two outlying genotypes can be identified (31 and 34 in Fig 2 and Fig 3) by the high number of mutations separating these from their nearest neighbor. Genotype 34 falls outside the clone A lineage group (Fig 1), and although this is not the case for genotype 31, both are excluded from further analysis of patterns of variation within clone A. Of the loops involved in the networks in Fig 2 and Fig 3, only that leading to genotype 29 could conceivably have been the result of a recombination event between the two most probable progenitors, genotypes 10 and 15.

View larger version (24K): In this window In a new window Download PPT slide   Figure 2. Network of multilocus microsatellite genotypes (based on seven loci) found in three of the populations surveyed (TBR, IMF, and BMA), assuming fully interconnected populations. Genotypes are shown as circles whose area is proportional to their frequency of occurrence. In most cases the length of connections is proportional to the number of inferred mutational changes between genotypes, but bars are also provided to show the number of changes. Deletions are shown as thick bars; additions are shown as thin bars. The central dashed line illustrates the division between the genotypes found in TBR/IMF and BMA, the only overlap occurring with central genotypes 1 and 7. Pathways of approximately equal probability are shown as more than one connection to a genotype. Genotype numbers correspond to those in Table 1.

View larger version (24K): In this window In a new window Download PPT slide   Figure 3. Networks constructed from multilocus microsatellite genotypes assuming separate, unconnected populations. See Fig 2 legend for further explanation.

Deletion bias within clone A:
The highest-frequency genotypes in the networks shown in Fig 2 and Fig 3 fulfilled a number of criteria expected for candidate ancestral genotypes (see DISCUSSION). On the basis of this inference, mutations were assumed to flow from these central nodes (genotypes 1 and 7, respectively in Fig 2 and Fig 3) outward toward the tips of the network.

An obvious feature of the mutational pathways inferred is the relatively high number of events representing a reduction in allele size. From the network that assumes full interconnection between populations (Fig 2), the estimated percentage of deletions was 88.1% (N = 42, P << 0.001, 95% lower confidence limit = 76.7%), and a consistently high bias was found across populations (Table 2) when these are treated independently (Fig 3), irrespective of whether or not the estimate incorporated genotype frequencies (Table 2). When data from the population networks (Fig 3) were subdivided to examine the bias for specific loci, a similar general pattern emerged (Table 3). No loci showed a significant percentage of additions and, depending on how mutations were counted, either three or five of the seven loci showed a significant percentage of deletions (Table 3). There was no relationship between allele size (in repeat units) and the direction of mutations (logistic regression, P > 0.7). All the shifts in allele size inferred from the minimum spanning networks involved changes of a maximum of four repeat units, with most involving changes of three repeat units or less (see Table 1).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Clonal composition of British populations:
Microsatellite variation among the sampled populations shows a pattern consistent with the existence of a small number of highly divergent clonal lineages within which minor variation has been generated by mutation since the introduction of P. antipodarum to Britain in the late 19th century. The genetic distance between clonal groups is too great to allow for the possibility that one clone has given rise to another by mutation since invasion of Britain. Within groups, variation is minor and variant genotypes tend to be specific to particular populations (see Table 1) suggesting independent origins by mutation. These patterns support the hypothesis that only a few founder clones were introduced to Britain (HAUSER et al. 1992 ). In previous genetic studies using multilocus minisatellites (HAUSER et al. 1992 ) and RAPDs (JACOBSEN et al. 1996 ), the morphotypes of clones A, B, and C (WARWICK 1952 ) were found to correspond to distinct genotypes. In this study we did not attempt to match every individual morphotype to a microsatellite genotype. However, morphological examinations of shells suggested that morphotypes of clones A, B, and C were present in roughly the frequencies within populations revealed by subsequent microsatellite analysis. Thus, morphological identification of clones, multilocus minisatellites, RAPDs, and multilocus microsatellites all appear to produce concordant results.

Identification of rare genotypes:
A previously unidentified clone, which we have called clone D, was found as the most common genotype in the CAE population from Cornwall. Although we did not identify this as a distinct morphotype, there has been anecdotal description of a fourth potential morphological strain (T. WARWICK, personal communication). This additional clonal genotype appears to be uncommon and to have a restricted distribution.

Four other genotypes, represented by a total of just five individuals, were identified. These may constitute highly mutated individuals originating from the major clones, extremely rare clones, or they may be hybrids between genotypes from within clonal groups. With the possible exception of the individual from the very large and relatively variable BMA population, it is unlikely that rather distantly related genotypes such as these could be mutant descendants of the four major clonal lineages. A study using allozymes to investigate the genetics of P. antipodarum in Britain reported evidence for multiple clones, which did not correspond to morphological types (FOLTZ et al. 1984 ). Although this contradicts all genetic data from DNA markers (above) and was probably questionable because of technical considerations (HAUSER et al. 1992 ; HUGHES 1996 ), we cannot exclude the possibility that other rare clones do exist within Britain. Very rare hybridization events (through sexual reproduction) between clones might occur and have been recorded in other primarily asexual snails (e.g., SAMADI et al. 1999 ), but if this is the case here, based on the allele composition of the very rare genotypes (data not shown), it does not appear that they could have arisen from recombination between any pair of genotypes recorded within the clonal groups.

Clonal dynamics:
Clonal lineage A is dominant throughout freshwater habitats, with B almost entirely restricted to brackish locales and C found only in the population from Wales (HAA). This description of clonal frequencies, habitat types, and geographical locations corresponds closely to those noted >50 years ago (reported in WARWICK 1952 , WARWICK 1969 ), on the basis of morphological descriptions of types A, B, and C. This suggests rather stable clonal dynamics, which may be related to the small number of clonal genotypes, apparent lack of conspecific sexual competitors, and an absence of parasites, thought to balance advantages of apomictic reproduction in New Zealand populations (LIVELY 1987 ). These factors are also likely to explain, at least in part, the remarkable success of P. antipodarum as an invasive species. Following the introduction to Britain, P. antipodarum has spread throughout much of continental Europe, which appears to be dominated by the same clonal groups A and B as Britain (WARWICK 1969 ; JACOBSEN et al. 1996 ).

Criteria for inclusion in the study of within-lineage microsatellite evolution:
The microsatellites used in this study were isolated from individuals of lineage A and produced unambiguous triploid microsatellite patterns at all seven loci only when amplifying DNA from this clone. For this reason and because of the numerical dominance of this clonal group in our samples, we concentrated our study of within-clonal-lineage microsatellite evolution on group A individuals. This strategy also greatly limited the potential for null alleles, which are of particular concern in studies of polyploid species. Moreover, we did not consider individuals from two small populations (DUN and STA) because, although of group A, they appeared to be invariant from the main clone A genotype, at least as determined by the four loci at which they were screened (genotype with frequency of 503 in Fig 1). Group A individuals from the HAA and CAE populations were also excluded because the potential for occurrence of any interclonal hybridization would be relatively higher where there was a mixture of clones. Thus, we concentrated the analysis on three populations, TBR, IMF, and BMA.

Identification of an ancestral introduced genotype and apomixis within clonal lineage A:
The minimum spanning networks based on multilocus microsatellite genotypes showed the general form of star phylogenies, with a high-frequency genotype in the center linked to several much lower-frequency genotypes and a general decrease in frequency toward the tips of the network. Star phylogenies are characteristic of recent expansion from a small number of ancestral haplotypes (AVISE 2000 ), which can be inferred from (1) their high frequency, (2) their interior position in the network, (3) their high number of connections to other haplotypes, and (4) their relatively wide geographical distribution (WATTERSON and GUESS 1977 ; DONNELLY and TAVARE 1986 ; CRANDALL and TEMPLETON 1993 ; EXCOFFIER and SMOUSE 1994 ; CRANDALL 1996 ; AVISE 2000 ; POSADA and CRANDALL 2001 ). These diagnostics all apply to the minimum spanning networks produced from the multilocus microsatellite genotypes in the three populations studied. In the network assuming full interconnection between the populations there are two high-frequency genotypes, separated by a single mutation (see Fig 2), one dominant in the TBR and IMF populations and the other in the BMA population. To determine whether one of these was likely to be a single common ancestor we genotyped a small number (N = 5 per population) of clonal group A individuals from the geographically dispersed STA, CAE, and HAA populations at a later date (DNA from DUN individuals had unfortunately degraded). All 15 of these extra samples from STA, CAE, and HAA were of genotype 1, the most common in TBR and IMF. Thus, in addition to its frequency, position, and number of connections in the networks, because of this widespread distribution, this is likely to be the single ancestral genotype for clone A that was introduced to Britain. The most common genotype in BMA (genotype 7) may be the result of a mutation very early in the colonization process. This suggestion is supported by the number of connections to other genotypes originating from genotype 7, which is only slightly lower than those from genotype 1.

Minimum spanning networks have been more commonly used in analyses of mitochondrial DNA (AVISE 2000 ) but have also been applied to studies of microsatellites on the human Y chromosome (e.g., KARAFET et al. 1999 ) in which nonrecombinant haplotypes are transmitted intact except for mutation. Our networks, obtained from triploid data, correspond closely to the form of minimum spanning networks constructed from haploid data (see examples for mtDNA star phylogenies in AVISE 2000 ), with their central high-frequency genotypes and relatively few genotypes involved in loops. We found only one genotype that could have potentially arisen via recombination, in that this would resolve a loop in the networks (CRANDALL and TEMPLETON 1993 ; POSADA and CRANDALL 2001 ). However, because the genotype involved was only two mutational steps away from the core of the network and this was an isolated possible incidence of recombination, it is difficult to argue that recombination is a more probable explanation than homoplasy. We consider it more likely that reproduction within these populations is entirely apomictic, suggesting concordance between microsatellite data and the lack of evidence of sexual reproduction in P. antipodarum in Britain (HUGHES 1996 ).

Microsatellites show a strong and consistent bias toward deletion of repeats:
For interconnected populations and pooled loci we estimated that >88% of the mutations inferred from the minimum spanning networks were deletions of repeats. Analysis based on unconnected populations suggested estimates of 75 or 80% for TBR, 79 or 81% for IMF, and 95 or 97% for BMA, respectively, depending on whether the genotypes or their frequencies were counted as mutations. Although the estimate for TBR is based on very limited data, the consistency across populations suggests a deletion bias approaching or exceeding 80%. It is conceivable that estimates based on pooled data might be biased by the effect of a single locus, notably Pa217, which showed not only 100% deletions, but also the highest mutation rate (D. WEETMAN, L. HAUSER and G. R. CARVALHO, unpublished results). However, 100% of deletions were found for two of the other loci, albeit based on fewer mutations, and a bias of between 87.5 and 92% (depending on the method of estimation) was found for the second most variable locus Pa254. Indeed, only one of the seven loci showed <50% of deletions and this was from only three inferred mutations. Therefore, we suggest that a strong bias toward deletions may be a common feature of the microsatellites of clone A of P. antipodarum in Britain. To our knowledge, this is the first example of a consistent deletion bias across a sample of nuclear microsatellite loci in any organism.

Presence of microsatellites with a strong deletion bias suggests a recent change in mutation mechanism:
The existence of a strong deletion bias raises the question of how microsatellites could have persisted in this clone of P. antipodarum. Two models of microsatellite evolution could provide an explanation, if the observed deletion bias is caused by a nonrandom sample of loci. First, a bias toward deletions for long microsatellite alleles, with an insertion bias for shorter alleles, could preserve a stable overall allele distribution (XU et al. 2000 ). Here, however, this is inapplicable because the alleles of most of our loci were of only short to moderate length (see Table 3) and also we found no relationship between allele size and the direction of change. Second, constraints on the total length of linked loci might lead to insertion biases for some loci, coupled with deletions for others (DERMITZAKIS et al. 1998 ). This possibility is difficult to rule out, but it seems unlikely since none of the seven loci had a significant tendency to expand, though other studies have found significant expansion biases to be common. Therefore, to explain the existence of both the extreme directional bias and the presence of mutable loci in P. antipodarum it would seem that some factor affecting the direction of mutations has changed in relatively recent evolutionary time.

Possible causes of a change in mutation mechanism leading to a deletion bias:
Could loss of meiosis caused by conversion from diploid sexuality to triploid apomixis have altered the mutation process? Although some DNA slippage-related mutations might occur just prior to meiosis via gene conversion type events (AMOS 1999 ), microsatellite mutations occur primarily during mitosis (reviewed by SIA et al. 1997A ; EISEN 1999 ), so lack of meiosis per se is probably an unlikely explanation. Moreover, all the inferred mutations involved changes of a maximum of four repeat units, which would suggest that they have arisen wholly by slippage, and not by some unusual mutation mechanism. CHENUIL et al. 1997 reported that tetraploid sexual barbels (Barbus sp.) appear to have fewer and shorter dinucleotide microsatellites than three other diploid cyprinid species examined, suggesting that perhaps polyploidy, rather than asexuality, may be associated with reduced microsatellite size caused by a mutation bias toward deletions. A change in ploidy level might be a mechanistic cause of a change in mutation process; for example, the extra polymerase or mismatch-repair (MMR) proteins encoded by one or more extra sets of chromosomes might disrupt the MMR protein complexes formed and so affect their function. Currently, the age of the original genotype of clone A of P. antipodarum is unknown but an explanation based on change in ploidy would rely on a relatively recent conversion to triploidy to be consistent with the continued existence of microsatellites.

A deletion bias might also be favored by selection for reduced genome size. Indeed, the only published report of a deletion bias that is consistent across microsatellites is in mitochondrial microsatellites of yeast, Saccharomyces cerevisiae (the nuclear microsatellites showed a very strong insertion bias; SIA et al. 2000 ). Mitochondrial genomes are thought to be under selection to reduce their size (SELOSSE et al. 2001 ), and for opportunistic organisms, because of the expected inverse correlation between genome size and development rate (CHARLESWORTH et al. 1996 ; HUGHES 1999 ), selection would be predicted to act against the accumulation of large amounts of "junk" DNA (CHARLESWORTH et al. 1996 ). For P. antipodarum clones invading Britain, successful persistence and spread may hinge on fast development, providing the ability to rapidly increase numbers following founder events or other population bottlenecks. Selection acting on individual microsatellite mutations seems unlikely because of their marginal effect on total genome size, but variants for both polymerase- (KOKOSKA et al. 2000 ) and MMR proteins (SIA et al. 1997B ), which cause preferential deletion of repeat units, have been found in S. cerevisiae and could have a more substantial effect in the genome upon which selection can act. Such a variant might have become fixed by chance in clonal group A of P. antipodarum, but because of the consistency of the deletion bias across populations, either the variant must have been introduced or it must have occurred very early in the colonization process. MMR variants that raise the mutation rate in clones of Escherichia coli adapting to new environments ("mutators") offer a substantial, but transitory, advantage (DE VISSER et al. 1999 ). However, a polymerase or MMR variant that changes the indel ratio, rather than rate, of mutations provides the potential to regulate repetitive DNA, and so could be under long-term positive selection in an asexual colonist such as P. antipodarum.

	FOOTNOTES

² Present address: School of Aquatic and Fishery Sciences, University of Washington, Seattle, WA 98195-5020.

	ACKNOWLEDGMENTS

We are grateful to Bill Amos, Brian Charlesworth, Africa Gómez, David Lunt, and Craig Primmer for comments on an earlier version of this manuscript and to Cock van Oosterhout for helpful discussions. The manuscript was much improved following comments on a previous version from two anonymous referees. This study was supported by N.E.R.C. (UK) grant GR9/04307.

Manuscript received May 9, 2002; Accepted for publication July 15, 2002.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AMOS, W., S. J. SAWCER, R. W. FEAKES, and D. C. RUBINSZTEIN, 1996  Microsatellites show mutational bias and heterozygote instability. Nat. Genet. 13:390-391.[Medline]

AMOS, W., 1999 A comparative approach to the study of microsatellite evolution, pp. 66–79 in Microsatellites: Evolution and Applications, edited by D. B. GOLDSTEIN and C. SCHLÖTTERER. Oxford University Press, Oxford.

AVISE, J. C., 2000 Phylogeography: The History and Formation of Species. Harvard University Press, Cambridge, MA.

CHARLESWORTH, B., P. SNIEGOWSKI, and W. STEPHAN, 1996  The evolutionary dynamics of repetitive DNA in eukaryotes. Nature 371:215-220.

CHENUIL, A., E. DESMARAIS, L. POYAUD, and P. BERREBI, 1997  Does polyploidy lead to fewer and shorter microsatellite in Barbus (Teleostei: Cyprinidae)? Mol. Ecol. 6:169-178.[Medline]

COOPER, G., N. J. BURROUGHS, D. A. RAND, D. C. RUBINSZTEIN, and W. AMOS, 1999  Markov chain Monte Carlo analysis of human Y-chromosome microsatellites provides evidence of biased mutation. Proc. Natl. Acad. Sci. USA 96:11916-11921.[Abstract/Free Full Text]

CRANDALL, K. A., 1996  Multiple interspecies transmission of human and simian T-cell leukemia/lymphoma virus type I sequences. Mol. Biol. Evol. 13:115-131.[Abstract]

CRANDALL, K. A. and A. R. TEMPLETON, 1993  Empirical tests of some predictions from coalescent theory with applications to intraspecific phylogeny reconstruction. Genetics 134:959-969.[Abstract]

DERMITZAKIS, E. T., A. G. CLARK, C. BATARGIAS, A. MAGOULAS, and E. ZOUROS, 1998  Negative covariance suggests mutation bias in a two-locus microsatellite system in the fish Sparus aurata.. Genetics 150:1567-1575.[Abstract/Free Full Text]

DE VISSER, J. A. G. M., C. W. ZEYL, P. J. GERRISH, J. L. BLANCHARD, and R. E. LENSKI, 1999  Diminishing returns from mutation supply rate in asexual populations. Science 283:404-406.[Abstract/Free Full Text]

DIRIENZO, A., P. DONNELLY, C. TOOMAJIAN, B. SISK, and A. HILL et al., 1998  Heterogeneity of microsatellite mutations within and between loci, and implications for human demographies. Genetics 148:1269-1284.[Abstract/Free Full Text]

DONNELLY, P. and S. TAVARÉ, 1986  The ages of alleles and a coalescent. Adv. Appl. Probab. 18:1-19.

DYBDAHL, M. and C. M. LIVELY, 1995  Diverse, endemic and polyphyletic clones in mixed populations of a freshwater snail (Potamopyrgus antipodarum). J. Evol. Biol. 8:385-398.

EISEN, J. A., 1999 Mechanistic basis for microsatellite instability, pp. 34–48 in Microsatellites: Evolution and Applications, edited by D. B. GOLDSTEIN and C. SCHLÖTTERER. Oxford University Press, Oxford.

ELLEGREN, H., 2000  Heterogeneous mutation processes in human microsatellite DNA sequences. Nat. Genet. 24:400-402.[Medline]

ESTOUP, A., and J.-M. CORNUET, 1999 Microsatellite evolution: inferences from population data, pp. 49–65 in Microsatellites: Evolution and Applications, edited by D. B. GOLDSTEIN and C. SCHLÖTTERER. Oxford University Press, Oxford.

EXCOFFIER, L. and P. E. SMOUSE, 1994  Using allele frequencies and geographic subdivision to reconstruct gene trees within species: molecular variance parsimony. Genetics 136:343-359.[Abstract]

FELSENSTEIN, J., 1993 PHYLIP (Phylogeny Inference Package), version 3.5c. Department of Genetics, University of Washington, Seattle.

FOLTZ, D. W., H. OCHMAN, J. S. JONES, and R. K. SELANDER, 1984  Genetic heterogeneity within and among morphological types of the parthenogenetic snail, Potamopyrgus jenkinsi (Smith 1889). J. Molluscan Stud. 50:242-245.

GOLDSTEIN, D. B., and C. SCHLÖTTERER, 1999 Microsatellites: Evolution and Applications. Oxford University Press, Oxford.

HAUSER, L., G. R. CARVALHO, R. N. HUGHES, and R. E. CARTER, 1992  Clonal structure of the introduced freshwater snail Potamopyrgus antipodarum (Prosobranchia: Hydrobiidae), as revealed by DNA fingerprinting. Proc. R. Soc. Lond. Ser. B 249:19-25.[Medline]

HUANG, Q. Y., F. H. XU, H. SHEN, H. Y. DENG, and Y. J. LIU et al., 2002  Mutation patterns at dinucleotide microsatellite loci in humans. Am. J. Hum. Genet. 70:625-634.[Medline]

HUGHES, A. L., 1999 Adaptive Evolution of Genes and Genomes. Oxford University Press, New York.

HUGHES, R. N., 1996  Evolutionary ecology of parthenogenetic strains of the prosobranch snail, Potamopyrgus antipodarum (Gray) = P. jenkinsi (Smith). Malacol. Rev. Suppl. 6:101-113.

JACOBSEN, R., V. E. FORBES, and O. SKOVGAARD, 1996  Genetic population structure of the prosobranch snail Potamopyrgus antipodarum (Gray) in Denmark using PCR-RAPD fingerprints. Proc. R. Soc. Lond. Ser. B 263:1065-1070.[Medline]

KARAFET, T., L. P. OSIPOVA, O. L. POSUKH, V. WIEBE and M. F. HAMMER, 1999 Y chromosome microsatellite haplotypes and the history of Samoyed-speaking populations in northwest Siberia, pp. 249–265 in Microsatellites: Evolution and Applications, edited by D. B. GOLDSTEIN and C. SCHLÖTTERER. Oxford University Press, Oxford.

KOKOSKA, R. J., L. STEFANOVIC, J. DEMAI, and T. D. PETES, 2000  Increased rates of genomic deletions generated by mutations in the yeast gene encoding DNA polymerase delta or by decreases in the cellular levels of DNA polymerase delta. Mol. Cell. Biol. 20:7490-7504.[Abstract/Free Full Text]

LIVELY, C. M., 1987  Evidence from a New Zealand snail for the maintenance of sex by parasitism. Nature 328:519-521.

NEFF, B. D. and M. R. GROSS, 2001  Microsatellite evolution in vertebrates: inference from AC dinucleotide repeats. Evolution 55:1717-1733.[Medline]

POSADA, D. and K. A. CRANDALL, 2001  Intraspecific gene genealogies: trees grafting into networks. Trends Ecol. Evol. 16:37-45.[Medline]

PRIMMER, C. R., H. ELLEGREN, N. SAINO, and A. P. MØLLER, 1996  Directional evolution in germline microsatellite mutations. Nat. Genet. 13:391-393.[Medline]

RUBINSZTEIN, D. C., W. AMOS, J. LEGGO, S. GOODBURN, and R. L. MARGOLIS et al., 1995  Microsatellite evolution—evidence for directionality and variation in rate between species. Nat. Genet. 10:337-343.[Medline]

SAITOU, N. and M. NEI, 1987  The neighbour-joining method: a new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4:406-425.[Abstract]

SAMADI, S., J. MAVÁREZ, J. P. POINTIER, B. DELAY, and P. JARNE, 1999  Microsatellite and morphological analysis of population structure in the parthenogenetic freshwater snail Melanoides tuberculata: insights into the creation of clonal variability. Mol. Ecol. 8:1141-1153.[Medline]

SELOSSE, M.-A., B. ALBERT, and B. GODELLE, 2001  Reducing the genome size of organelles favours gene transfer to the nucleus. Trends Ecol. Evol. 16:135-141.[Medline]

SIA, E. A., S. JINKS-ROBERTSON, and T. D. PETES, 1997a  Genetic control of microsatellite stability. Mutat. Res. 383:61-70.[Medline]

SIA, E. A., R. J. KOKOSKA, M. DOMINSKA, P. GREENWELL, and T. D. PETES, 1997b  Microsatellite instability in yeast: dependence on repeat unit size and DNA mismatch repair genes. Mol. Cell. Biol. 17:2851-2858.[Abstract]

SIA, E. A., C. A. BUTLER, M. DOMINSKA, P. GREENWELL, and T. D. FOX et al., 2000  Analysis of microsatellite mutations in the mitochondrial DNA of Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. USA 97:250-255.[Abstract/Free Full Text]

SMITH, E. A., 1889  Notes on British Hydrobiidae with a description of a supposed new species. J. Conchol. 6:142-146.

SMOUSE, P. E., 1998  To tree or not to tree. Mol. Ecol. 7:399-412.

SUNNUCKS, P., P. R. ENGLAND, A. C. TAYLOR, and D. F. HALES, 1996  Microsatellite and chromosomal evolution of parthenogenetic Sitobion aphids in Australia. Genetics 144:747-756.[Abstract]

UDUPA, S. M. and M. BAUM, 2001  High mutation rate and mutational bias at (TAA)_n microsatellite loci in chickpea (Cicer arietinum L.). Mol. Genet. Genomics 265:1097-1103.[Medline]

WARWICK, T., 1952  Strains in the mollusc Potamopyrgus jenkinsi (Smith). Nature 169:551-552.

WARWICK, T., 1969  Systematics of the genus Potamopyrgus (Hydrobiidae) in Europe and causation of the keel in this clone. Proc.Third Eur. Malacol. Congr. 9:301-302.

WATTERSON, G. A. and H. A. GUESS, 1977  Is the most frequent allele the oldest? Theor. Popul. Biol. 11:141-160.[Medline]

WEBER, J. L. and C. WONG, 1993  Mutation of human short tandem repeats. Hum. Mol. Genet. 2:1123-1128.[Abstract/Free Full Text]

WEETMAN, D., L. HAUSER, and G. R. CARVALHO, 2001  Isolation and characterization of di- and trinucleotide microsatellites in the freshwater snail Potamopyrgus antipodarum. Mol. Ecol. Notes 1:185-187.

WIERDL, M., M. DOMINSKA, and T. D. PETES, 1997  Microsatellite instability in yeast: dependence on the length of the microsatellite. Genetics 146:769-779.[Abstract]

XU, X., M. PENG, Z. FANG, and X. XU, 2000  The direction of microsatellite mutations is dependent upon allele length. Nat. Genet. 24:396-399.[Medline]

ZAR, J. H., 1984 Biostatistical Analysis. Prentice-Hall International, London.

ZARANKO, D. T., D. G. FARARA, and F. G. THOMPSON, 1997  Another exotic mollusc in the Laurentian Great Lakes: the New Zealand native Potamopyrgus antipodaum (Gray 1843) (Gastropoda, Hydrobiidae). Can. J. Fish. Aquat. Sci. 54:809-814.

ZHU, Y., D. C. QUELLER, and J. E. STRASSMANN, 2000  A phylogenetic perspective on sequence evolution in microsatellite loci. J. Mol. Evol. 50:324-338.[Medline]

This article has been cited by other articles:

				M. GuhaMajumdar, E. Dawson-Baglien, and B. B. Sears Creation of a Chloroplast Microsatellite Reporter for Detection of Replication Slippage in Chlamydomonas reinhardtii Eukaryot. Cell, April 1, 2008; 7(4): 639 - 646. [Abstract] [Full Text] [PDF]

				O. Paun and E. Horandl Evolution of Hypervariable Microsatellites in Apomictic Polyploid Lineages of Ranunculus carpaticola: Directional Bias at Dinucleotide Loci Genetics, September 1, 2006; 174(1): 387 - 398. [Abstract] [Full Text] [PDF]