A Quantitative Genetic Analysis of Nuclear-Cytoplasmic Male Sterility in Structured Populations of Silene vulgaris

Corresponding author: Douglas R. Taylor, Department of Biology, Gilmer Hall, University of Virginia, Charlottesville, VA 22903., drt3b{at}virginia.edu (E-mail)

Communicating editor: R. G. SHAW

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Gynodioecy, the coexistence of functionally female and hermaphroditic morphs within plant populations, often has a complicated genetic basis involving several cytoplasmic male-sterility factors and nuclear restorers. This complexity has made it difficult to study the genetics and evolution of gynodioecy in natural populations. We use a quantitative genetic analysis of crosses within and among populations of Silene vulgaris to partition genetic variance for sex expression into nuclear and cytoplasmic components. We also use mitochondrial markers to determine whether cytoplasmic effects on sex expression can be traced to mitochondrial variance. Cytoplasmic variation and epistatic interactions between nuclear and cytoplasmic loci accounted for a significant portion of the variation in sex expression among the crosses. Source population also accounted for a significant portion of the sex ratio variation. Crosses among populations greatly enhanced the dam (cytoplasmic) effect, indicating that most among-population variance was at cytoplasmic loci. This is supported by the large among-population variance in the frequency of mitochondrial haplotypes, which also accounted for a significant portion of the sex ratio variance in our data. We discuss the similarities between the population structure we observed at loci that influence sex expression and previous work on putatively neutral loci, as well as the implications this has for what mechanisms may create and maintain population structure at loci that are influenced by natural selection.

UNDERSTANDING adaptive evolution in structured populations is a complicated problem. Evolution proceeds according to patterns of selection and genetic variance within populations, as well as patterns of migration among genetically divergent populations. We have been using the evolution of gynodioecy in plants as a model system to study selection and the response to selection in spatially structured populations (MCCAULEY and TAYLOR 1997 ; MCCAULEY and BROCK 1998 ; TAYLOR et al. 1999A ; MCCAULEY et al. 2000 ; OLSON and MCCAULEY 2000 ).

Gynodioecious plants have a mixture of hermaphroditic and functionally female individuals (FRANK 1989 ; COUVET et al. 1990 ). A male-sterility mutation, which causes the female phenotype, can spread in populations if the loss of fitness through the male function is more than offset by an increase in fitness through the female function (LEWIS 1941 ; LLOYD 1974 ; CHARLESWORTH and GANDERS 1979 ; CHARLESWORTH 1981 ). This is especially likely to occur when the male-sterility allele is a maternally inherited cytoplasmic element [a so-called cytoplasmic male-sterility (CMS) factor], because cytoplasmic genes suffer no loss of fitness when they eliminate the male function.

The evolution and maintenance of gynodioecious mating systems have received considerable attention from evolutionary biologists because the spread of a male-sterility mutation is thought to be a first step toward the evolution of separate sexes (CHARLESWORTH 1991 ). CMS factors are usually accompanied by nuclear alleles that restore fertility, a clear example of intragenomic conflict (HURST et al. 1996 ). CMS factors are also of profound practical importance in the large-scale production of hybrid seed for agriculture (KUCK and WRICKE 1995 ; SCHNABLE and WISE 1998 ).

Molecular genetic studies of male sterility, primarily in crop species, suggest that CMS factors are usually mitochondrial mutants (SAUMITOU-LAPRADE et al. 1994 ). Often they are chimeric genes created by the relatively frequent intragenomic recombination and rearrangements that occur in plant mitochondrial genomes (SCHNABLE and WISE 1998 ). Crossing studies have shown that the genetic basis of male sterility can be extremely complicated, especially in natural systems. For example, there are several known CMS alleles in Plantago lanceolata (VAN DAMME 1983 ; DE HAAN et al. 1997A ), Plantago coronopus (KOELEWIJN and VAN DAMME 1995A ), Thymus vulgaris (BELHASSEN et al. 1991 ), Beta vulgaris (SAUMITOU-LAPRADE et al. 1993 ), and Silene vulgaris (CHARLESWORTH and LAPORTE 1998 ). Often, each CMS allele is accompanied by several restorer loci (VAN DAMME 1983 ; BELHASSEN et al. 1991 ; KOELEWIJN and VAN DAMME 1995B ; DE HAAN et al. 1997B ; CHARLESWORTH and LAPORTE 1998 ).

Multiple CMS factors are expected with cytonuclear male-sterility systems because, once populations with CMS factors have been restored, they are susceptible to invasion by new CMS factors, and so on. Recently, population genetic models have emphasized a resemblance between gynodioecy and nonequilibrium host-pathogen systems, incorporating a more complex genetic basis for male sterility and/or examining the evolutionary dynamics in spatially structured populations (FRANK 1989 ; MCCAULEY and TAYLOR 1997 ; PANNELL 1997 ; COUVET et al. 1998 ; GIGORD et al. 1998 ). Empirical studies have therefore begun to focus on the importance of population structure for the evolution of CMS systems and have demonstrated genetically based sex ratio variation among local populations that may have significant evolutionary consequences (MCCAULEY et al. 2000 ).

The evolutionary dynamics of these complex genetic systems (e.g., CMS factors and associated restorers) are difficult to evaluate in nature, especially using classical genetics (FRANK 1989 ; TAYLOR 1999 ). In natural populations, we expect the genetic basis of male sterility to vary from one population to the next (FRANK 1989 ; COUVET et al. 1990 ; GOUYON et al. 1991 ). For example, if a restorer is absent from one population, sex expression will be caused by variation at the CMS locus. In other populations, however, the CMS factor might be fixed, and variation in sex expression may be due to variation at a nuclear restorer locus. When both are polymorphic, sex expression should be influenced by an epistatic interaction between the two loci. The situation is even more complicated when there are several CMS/restorer systems involved. Even the most detailed genetic studies of natural systems, therefore, have given only a qualitative sense of the magnitude of genetic variation for male sterility in nature (e.g., VAN DAMME 1983 ; BELHASSEN et al. 1991 ; KOELEWIJN and VAN DAMME 1995A , KOELEWIJN and VAN DAMME 1995B ; DE HAAN et al. 1997A , DE HAAN et al. 1997B ; MANICACCI et al. 1997 ; CHARLESWORTH and LAPORTE 1998 ). Although it is increasingly clear that a diversity of nuclear and cytoplasmic genes generally influence sex expression in gynodioecious plants, relatively little is known about how nuclear vs. cytoplasmic loci contribute to overall variation in sex expression in nature and, especially, how this genetic variation is distributed within and among populations.

We combine a quantitative genetic approach and a molecular marker approach to characterize the nature of genetic variation for sex expression in natural populations of S. vulgaris, a gynodioecious plant. It is known that the male-sterility polymorphism is under cytonuclear control within S. vulgaris populations (CHARLESWORTH and LAPORTE 1998 ) and that there is genetically based variation in the frequency of hermaphrodites among populations (MCCAULEY et al. 2000 ). Estimates of population structure using putatively neutral loci show that the maternally inherited chloroplast genome is highly structured relative to nuclear loci, probably owing to limited seed dispersal (MCCAULEY 1998 ).

Here, we use crossing schemes to partition genetic variance for sex expression into nuclear and cytoplasmic components, both within and among populations, to test the specific hypothesis that cytoplasmic male-sterility genes are highly structured spatially, relative to nuclear restorers. Population structure at maternally inherited cytoplasmic loci is estimated by how the genetic variation assigned to dams is distributed within vs. among populations. The use of mitochondrial DNA (mtDNA) markers allows us to test this hypothesis further since mtDNA diversity must be tied to diversity in male-sterility factors in some way, provided the cause of male sterility is in fact cytoplasmic. The likely mechanisms that create population structure at neutral loci vs. the loci influencing sex expression are discussed, as are the circumstances under which population structure at loci under selection can have important evolutionary consequences.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

S. vulgaris (Caryophyllaceae) is a short-lived perennial weedy plant. It is gynodioecious, with both functionally hermaphroditic and male-sterile individuals co-occurring within populations, along with a low incidence of intersexes, or individuals bearing both hermaphroditic and male-sterile flowers (JOLLS and CHENIER 1989 ). The sex ratio of natural populations has been observed to vary from all-hermaphrodite to as much as 75% female (MCCAULEY et al. 2000 ). Crosses suggest that sex expression is under cytonuclear control (CHARLESWORTH and LAPORTE 1998 ). Individuals bear numerous flowers, which develop asynchronously over a flowering season lasting up to 10 weeks or more (D. TAYLOR, personal observation). S. vulgaris is pollinated by bees and moths and is known to be self-fertile (CHARLESWORTH 1989 ; JOLLS and CHENIER 1989 ; PETTERSSON 1992 ). Capsules bear up to 100+ seeds, which, being passively dispersed, are most likely to fall within a few meters of the maternal plant.

The crossing designs involved plants from nine populations found along roadsides and waste places in the Allegheny Mountains of Virginia. All populations were within 25 km of each other in the vicinity of Mountain Lake Biological Station (37°22'N, 80°31'W). In this region, S. vulgaris is patchily distributed and infrequent, with patches ranging in size from several isolated individuals to several hundred individuals. For our study populations, we chose spatially distinct patches of at least moderate size (30 individuals or more), which exhibited a wide range of sex ratios in our field surveys. At least five maternal families were obtained from each population by collecting one capsule containing mature seed from separate individuals in the field. To obtain sex ratio estimates for each maternal family, 20 seeds from each capsule were grown to flowering in the greenhouse (MCCAULEY et al. 2000 ). Nearly all capsules produced both female and hermaphrodite offspring among the progeny.

To estimate the relative magnitude of nuclear vs. cytoplasmic contributions to genetic variance in sex expression, we compared the relative contribution of sire vs. dam to offspring sex ratio using crossing schemes. In crossing designs of this type, the expectation is that nuclear loci contribute equally to sire and dam effects on offspring phenotype, although the process of sampling may cause deviations from this expectation. Cytoplasmic loci cause an additional contribution by the dams to the offspring phenotype. In our crossing scheme, all families carry cytoplasmic male-sterility factors (see below). Within a given CMS type, dominant nuclear restorers contribute only to sire effects, because females would all be homozygous for the recessive nonrestorer. Recessive restorers contribute only to dam effects, because hermaphrodite sires would all be homozygous. Codominant restorers contribute equally to sire and dam effects. Restoration of cytoplasmic male sterility in most species, including studies in S. vulgaris (CHARLESWORTH and LAPORTE 1998 ), seems to involve a mixture of mostly dominant and some recessive alleles (GOTTSCHALK and KAUL 1974 ; VAN DAMME 1983 ; KOELEWIJN and VAN DAMME 1995B ). Data from the present study (see below) also point to restorers being more dominant than recessive. We therefore ascribe the excess dam effects in our study to cytoplasmic variation.

Attributing excess dam effects as variation in CMS factors also relies on the assumption that maternal environmental effects on sex ratio are negligible relative to variation in cytoplasmic male-sterility factors. This is justified given the strong effects CMS factors are known to have on sex expression in S. vulgaris and the minimal influence of environmental variation (CHARLESWORTH and LAPORTE 1998 ). In the populations we study, a separate series of crossing experiments performed in temporal blocks revealed environmental influences (block effects) on several aspects of plant size, but not on gender (S. EMERY, unpublished results). In addition, maternal environmental effects were minimized in our study by rearing all maternal parents from seed under the same controlled conditions in the greenhouse.

The objective was to estimate the importance of nuclear genetic effects, cytoplasmic genetic effects, and epistatic interactions between nuclear and cytoplasmic loci and to determine how these effects were distributed within vs. among populations in nature. The within-population and among-population components of genetic variation were estimated using two separate crossing schemes. First, crosses were made between plants from the same population to estimate the within-population genetic variance for sex expression. These so-called "within-population" crossing designs were then replicated for nine populations. Second, crosses were made between plants from different populations ("among-population crosses") to estimate the degree to which nuclear vs. cytoplasmic genetic variance for sex expression was distributed among populations.

Within-population crosses:
Sire and dam effects were estimated using a factorial crossing scheme from each population separately (Fig 1). Within each population, up to five families were randomly selected, and one female offspring from each was used as a dam in the crossing scheme. The same procedure was used to select hermaphrodites as sires. All individuals were selected from sibships that had both female and hermaphrodite progeny, so all individuals carried a fertility cytoplasm that is susceptible to restoration.

View larger version (26K): In this window In a new window Download PPT slide   Figure 1. Crossing design for crosses among plants from the same population (within-population crosses) and crosses among plants from different populations (among-population crosses). H, hermaphrodite sires; F, female dams. The first subscript following H or F identifies the population the plant came from, followed by a subscript representing individuals within populations. One sire and one dam from each within-population design were randomly selected to represent that population in the among-population design.

Crosses were made using hermaphrodite flowers <24 hr post-anthesis and female flowers <48 hr after the stigmas were exserted (S. vulgaris is protandrous). Seeds were allowed to mature and collected as capsules opened in the greenhouse. Seeds were placed in dry storage for 5 weeks and then sown in 2.5-cm-diameter tubular plastic pots (Conetainers, Stuewe and Sons, Corvallis, OR) containing a standard potting mixture. For each cross, two seeds were sown in each of 50 pots, but they were thinned to one seedling per pot (50 seedlings per cross) after germination. The plants were germinated on a mist bench and then transferred to benches where high intensity lamps were used to enforce long (16 hr) days. The sex of each plant, based on examination of at least five flowers per offspring, was recorded to estimate the sex ratios produced by each cross.

The proportion of hermaphrodite progeny in each cross was analyzed using logistic regression (JMP, SAS INSTITUTE 1995). Specifically, the crossing designs from all populations were analyzed simultaneously specifying population effects, sire effects (nested within population), dam effects (nested within population), and sire*dam interaction (within population). The sex ratio data were also analyzed for each population separately. Because there were often significant interaction effects, significance of the main effects was estimated using F-statistics with the chi-square of the main effect (divided by the degrees of freedom) as the numerator and the chi-square of the interaction (divided by the degrees of freedom) as the denominator (see NUNNEY 1990 ; TAYLOR et al. 1999B ).

Among-population crosses:
The among-population crossing design was similar to the within-population crosses. We randomly selected a single sire and a single dam from each within-population design to represent each of the nine populations in the among-population crossing scheme (Fig 1). The crosses were performed, the plants were grown, and the sex ratios were estimated as they were in the within-population crosses (above). The data were also analyzed the same way as the within-population crosses, except that there was no population effect included in the models.

We used variance components of the population, sire, dam, and sire*dam interaction effects to estimate the magnitude of genetic variance within and among populations. To calculate variance components, we scored the hermaphrodite and female offspring with a phenotype of "1" and "0," respectively, and analyzed the within- and among-population data using the SAS VARCOMP procedure (SAS INSTITUTE 1985). This is essentially the same method used by WEIR 1990 to partition components of genetic variation when computing F-statistics. Since the significance tests in this analysis violate the assumption of normality, we report only statistical significance from the logistic regressions.

To compare the variance components within vs. among populations, we used a jacknife procedure. Specifically, each dam was sequentially omitted to calculate pseudovalues for each of the variance components. The pseudovalues were then used to construct 95% confidence intervals for each variance component (SOKAL and ROHLF 1995 ).

mtDNA haplotype analysis:
CMS factors are typically mitochondrial mutants. Since the entire mitochondria is co-inherited, we expect there to be linkage disequilibrium between mitochondrial DNA variation and CMS factors and that mitochondrial markers will be associated with CMS variation in our crossing data. Mitochondrial variation may be responsible for the sex ratio variation among dams, and among-population variation in the frequency of mitochondrial variants could be responsible for some of the population effects on sex ratio. If these were true, we would expect the mtDNA haplotypes to account for much of the variation within populations in the original model.

OLSON and MCCAULEY 2000 have developed restriction fragment length polymorphism (RFLP) markers to discriminate among several mitochondrial haplotypes in S. vulgaris. Briefly, DNA was extracted using a DNeasy kit for plants (QIAGEN, Valencia, CA), digested with NdeI and HindIII, and electrophoresed on 0.7% agarose in TBE. Digestions were transferred to Hybond N+ via capillary blots. An 1.5-kb portion of the cytochrome oxidase I (coxI) gene was PCR amplified from S. vulgaris using primers coxIf82 (BOWE et al. 2000 ) and coxIr1.6k (CHO et al. 1998 ). The coxI PCR product was labeled with radioactive dCTP using a random primer labeling kit (Boehringer Mannheim, Indianapolis) and hybridized to the total genomic blots. The coxI gene in S. vulgaris is monomorphic for an NdeI restriction site, and thus the labeled coxI probe hybridizes to two regions from the genomic DNA corresponding to the regions flanking the NdeI site. These mitochondrial RFLPs exhibited strict maternal inheritance in greenhouse crosses (OLSON and MCCAULEY 2000 ).

The mtDNA haplotypes were determined for all of the dams in our crossing schemes and incorporated into an additional statistical analysis. The effect of mtDNA haplotype on offspring sex ratio was analyzed for the within-population crosses using logistic regression. In the within-population crosses, mtDNA haplotype was confounded to some extent with the population effect; thus, we removed the population effect from the model and included dam effects nested within the mitochondrial haplotype.

The population structure of the RFLP mitochondrial haplotypes was estimated by calculating F_ST for haploid data (using methods outlined in MCCAULEY 1998 ). Briefly, each mitochondrial haplotype was treated as an allele. F_ST was calculated separately for each allele and then was combined across alleles to provide a summary F_ST statistic for the mitochondrial haplotypes.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The within-population crosses produced a wide range of sex ratio variation (Fig 2A). The average within-population cross produced 56% hermaphrodites, although there were distinct classes of crosses that produced all-hermaphrodite or all-female progeny. Crosses that produced all-hermaphrodite progeny were far more common than crosses that produced all-female progeny.

View larger version (17K): In this window In a new window Download PPT slide   Figure 2. Distribution of sex ratios (proportion of hermaphrodites) produced from (a) within-population crosses and (b) among-population crosses.

The within-population crosses showed significant variation in the frequency of hermaphrodites (Table 1) as well as highly significant variation in the sex ratios produced by different dams within populations. No overall sire effect was detected (Table 1), but the sire*dam interaction explained a large portion of the sex ratio variance (Table 1 and Table 3).

The among-population crosses produced similar sex ratios, on average 59% hermaphrodite, but a greater proportion of crosses produced all-female or all-hermaphrodite progeny (Fig 2). The general results were similar to the within-population crosses; the dam effect and the sire*dam interaction were significant, and the sire effect was negligible (Table 2).

The variance components showed that the dam effect was nearly four times stronger in the among-population crosses than it was in the within-population crosses (Table 3). The difference in dam effects in the two crossing schemes was statistically significant using a jacknife procedure; i.e., the 95% confidence interval for dam effect in the within-population crosses (-0.0164 to 0.2133) did not overlap with the confidence interval for the dam effect in the among-population crosses (0.2591–0.8906). The sire effects and the sire*dam interaction effects were not significantly different between the two crossing schemes.

Twelve mitochondrial haplotypes were detected within the set of females used as dams in our study. Mitochondrial haplotype diversity had a large among-population component (F_ST = 0.42). The mean sex ratio for mitochondrial haplotypes ranged from 19 to 90% hermaphrodite. The mitochondrial haplotype of the dams explained a significant amount of the sex ratio variation in our within-population crossing schemes (Table 4). This accounted for roughly three times the phenotypic variation that could be explained by the dam effect (mtDNA haplotype accounted for 13.3% of the sex ratio variance vs. 4.6% for the dam effect; the full model included sire and dam effects, nested within mtDNA haplotype, plus all possible interactions). However, the dam effect remained significant even with the mtDNA effect included in the model, indicating that there was significant sex ratio variation produced among dams that had the same mitochondrial haplotype.

The nature of the sire, dam, and mtDNA effects in different populations can be seen when the populations are analyzed separately (Table 5). First, there are relatively few significant main effects, which is probably due to the limited statistical power within any single population, especially since the significance of the main effects was tested over significant interaction effects. Second, the genetic basis of the sex ratio variation was different from one population to the next, with some populations having sire effects, some having dam effects, and some having neither. Sire*dam interactions were significant in all populations. The proportion of the variation explained by sires was generally negligible in populations that were polymorphic for mtDNA haplotype, but sires accounted for as much or more sex ratio variance as the dams in populations that were monomorphic for mtDNA haplotype (Table 5).

A posteriori, much of this mtDNA haplotype diversity was captured in the sample of dams used for the among-population crossing scheme. Two haplotypes (2 and 3) were represented in two dams each, but the remaining five dams all had unique mtDNA genotypes (haplotype numbers 5, 7, 10, 11, and 15; see Table 5).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Previous work by CHARLESWORTH and LAPORTE 1998 has shown that the genetic basis of sex expression in S. vulgaris is cytonuclear and complicated by the likelihood that multiple CMS factors and multiple restorers coexist within populations. Our results allow us to estimate the relative contributions of CMS diversity, nuclear genetic variation, and cytonuclear interactions to variable sex expression and to partition these effects into within- and among-population components. In interpreting our results it must be remembered that in each cross a female functioned as the ovule donor, so all offspring resulting from such crosses should carry a CMS factor. Further, all females were selected from large sibships that included at least one hermaphrodite, so all CMS factors present in our study are theoretically susceptible to restoration.

The distribution of sex ratios among the crosses is inconclusive with respect to any specific model for the genetic basis of sex expression (e.g., number of CMS factors, number of restorer loci, etc.), but they do suggest that the restorer alleles, on average, tend toward dominance. Consider the simplest genetic model of one restorer locus with two alleles. If the restorer allele is recessive, a female dam should never produce 100% hermaphrodite broods because she must be at least heterozygous for the dominant nonrestorer. With a dominant restorer, female parents are always homozygous recessive and the sex ratios should be 100 or 50% hermaphrodite, depending on the genotype of the hermaphrodite sire at the restorer locus. Since all female parents carry a CMS factor, the abundance of all-hermaphrodite sex ratios suggests a tendency for restorer(s) to be more dominant than recessive. Dominant restoration is also suggested by the trend for the sire effects to be as large, or larger, in populations that were monomorphic for mtDNA haplotype. In our design, dominant nuclear genes contribute primarily to the sire effect in the absence of CMS variation (see MATERIALS AND METHODS). Finally, a separate crossing study has shown that the proportion of females in S. vulgaris is greater among progeny that are the products of selfing (S. EMERY, unpublished data), suggesting the female phenotype is caused by recessive alleles. Our results are therefore consistent with numerous other studies that have found dominant, or mostly dominant, restoration of male fertility (GOTTSCHALK and KAUL 1974 ; VAN DAMME 1983 ; KOELEWIJN and VAN DAMME 1995B ; CHARLESWORTH and LAPORTE 1998 ; but see DE HAAN et al. 1997B ).

The strong dam effects present in both our within- and among-population crosses suggest that there are multiple CMS factors present in our study populations and that these CMS factors differ in their likelihood of restoration. We can attribute the excess dam effects as CMS variation on the basis of several observations. First, we know that sex expression in S. vulgaris is cytonuclear (CHARLESWORTH and LAPORTE 1998 ). Second, the excess dam effects are not caused by predominantly recessive nuclear restorers. Third, a significant amount of the among-dam sex ratio variation can be explained by among-dam variation in mitochondrial DNA haplotype. Although this statistical relationship does not allow us to identify specific CMS factors, it does suggest that the sex ratio variation among dams (and the associated mtDNA haplotype diversity) reflects underlying variation in CMS factors.

The large sire*dam interaction was expected since nuclear restorers influence sex expression only in certain CMS backgrounds and vice versa. The interaction also indicates that, although there is no main effect of sire, there is nuclear genetic variation at one or more restorer loci. Given the presence of nuclear variation, one might wonder why dam, but not sire, effects were detectable. One possibility is simply that the coancestry among maternal half sibs is 1.0 for the maternally inherited cytoplasmic genome, whereas the coefficient of coancestry among paternal half sibs is 1/8 for the biparentally inherited nuclear genome.

The significant population effect in the within-population crosses, combined with the large increase in dam effect in the among-population crosses, demonstrates that there is genetic structure for the genes influencing sex expression. The fact that the dam, but not the interaction, effect is inflated in among-population crosses suggests that this is largely due to genetic structure of the cytoplasmic, rather than the nuclear, genome. The population structure of the cytoplasmic genome is also evident from the high F_ST value (0.42) generated by the distribution of our mitochondrial markers within and among populations. Note that a previous study of S. vulgaris comparing cpDNA structure to the structure evident from nuclear encoded allozymes found the cpDNA structure (F_ST = 0.62) to be approximately three times greater than that seen with allozymes (MCCAULEY 1998 ).

Although there is strong evidence that there is spatial structure at the loci that influence sex expression in S. vulgaris, it is less clear what mechanisms create and maintain that population structure. We see two possible mechanisms that could account for the population structure of CMS factors and at non-neutral loci in general. First, population structure may arise as a result of recurrent selection within relatively persistent populations. Here, local populations may reach a stable equilibrium sex ratio, but that equilibrium may vary among populations because of environmental variation (such as variation in the composition of the pollinator community) or variation in the genetic composition of populations that influences the response to selection. A more complex case might be that populations follow one or more evolutionary trajectories determined by a dynamic equilibrium, with populations differing in sex ratio because they differ in their respective positions along that trajectory (GOUYON et al. 1991 ). Finally, local populations may not be at equilibrium sex ratio (owing to the constant introduction of new genetic variants by mutation and low level migration) yet still display among-population sex ratio variation as a consequence (FRANK 1989 , FRANK 1997 ; COUVET et al. 1990 ; GOUYON et al. 1991 ). A second structuring mechanism could be the extinction and recolonization of demes. In this case, natural selection within demes is less important in generating population structure because the demes themselves are ephemeral, but population structure could be frequently regenerated by chance founder events (e.g., WILSON 1980 ; MCCAULEY and TAYLOR 1997 ).

The two contrasting views of sex ratio evolution in structured populations outlined here depend largely on the degree of persistence of local populations. They focus either on the coevolution of CMS factors and restorers within temporally stable local populations—recognizing that individual populations could diverge in their respective sex ratios for a variety of reasons—or on the possibility that selection within demes can diminish population structure but that extinction and recolonization ensure that natural selection within demes is ephemeral. Although we have no specific evidence for either scenario in S. vulgaris, there are two reasons to suggest that extinction and colonization are important. First, S. vulgaris is patchily distributed in nature, and those patches appear to be short lived. We have observed several extinctions and colonizations within our study area over a 5-yr period. Second, there is a close resemblance between how quantitative genetic variance for sex expression is distributed within and among populations and the distribution of genetic variance in mtDNA markers (this study), cpDNA markers, and allozymes (MCCAULEY 1998 ). It is not surprising that chloroplast DNA markers and mitochondrial variants that influence sex expression are similarly structured, because the chloroplast and mitochondrial genomes of S. vulgaris are in strong linkage disequilibrium with each other (OLSON and MCCAULEY 2000 ). More to the point is that for molecular markers and for the loci that influence sex expression, cytoplasmic genes showed far greater among-population genetic variance than nuclear genes, which suggests that similar mechanisms may be involved in creating population structure in the two classes of loci. For example, maternally inherited loci may all be highly structured relative to nuclear loci because their dispersal via seed limits their gene flow or because they have a smaller effective population size (due to haploidy), both of which make them particularly susceptible to stochastic processes such as drift during founder events.

The study of gynodioecy provides an opportunity for the investigation of the more general question of adaptive evolution in structured populations. The first step in understanding evolution within demes is understanding the genetic basis of trait variation. However, predicting the response to selection can be difficult. For example, in S. vulgaris, evolutionary dynamics are complicated by epistatic interactions between nuclear and cytoplasmic genes that influence sex expression. Populations may also differ in their genetic basis for traits under selection, and the overall dynamics are further complicated by the joint effects of selection within demes and gene flow among demes. These effects can be evaluated only by combining genetic studies of spatially distributed populations with ecological studies of the persistence of demes and patterns of gene flow among them. These are the sorts of general challenges that need to be confronted to understand adaptive evolution in structured populations.

	ACKNOWLEDGMENTS

We thank Sheri Church, Natalia Nikoalenko, Chris Richards, and especially Meagan Saur for technical assistance. We also thank Ruth Shaw and three anonymous reviewers for their comments on the manuscript. This work was supported by a National Science Foundation grant (DEB-9707372) to D.E.M. and D.R.T. and by a National Science Foundation/Sloan Postdoctoral Fellowship in Molecular Evolution to M.S.O. (DBI-9750033).

Manuscript received August 30, 2000; Accepted for publication March 2, 2001.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BELHASSEN, E., B. DOMMEE, A. ATLAN, P.-H. GOUYON, and D. POMENTE et al., 1991  Complex determination of male sterility in Thymus vulgaris: genetic and molecular analysis. Theor. Appl. Genet. 82:137-143.

BOWE, L. M., G. COAT, and C. W. DEPAMPHILIS, 2000  Phylogeny of seed plants based on all three plant genomic compartments: extant gymnosperms are monophyletic and Gnetales' closest relatives are conifers. Proc. Natl. Acad. Sci. USA 97:4092-4097[Abstract/Free Full Text].

CHARLESWORTH, B., 1991  Evolution of sex chromosomes. Science 251:1030-1033[Abstract/Free Full Text].

CHARLESWORTH, D., 1981  A further study of the problem of the maintenance of females in gynodioecious species. Heredity 46:27-39.

CHARLESWORTH, D., 1989  The population biology of gynodioecy in Silene vulgaris (Abstr.). Am. J. Bot. 76(Suppl.):74.

CHARLESWORTH, D. and F. R. GANDERS, 1979  The population genetics of gynodioecy with cytoplasmic-genic male-sterility. Heredity 43:213-218.

CHARLESWORTH, D. and V. LAPORTE, 1998  The male-sterility polymorphism of Silene vulgaris: analysis of genetic data from two populations and comparison with Thymus vulgaris.. Genetics 150:1267-1282[Abstract/Free Full Text].

CHO, Y., Y.-L. QIU, P. KUHLMAN, and J. D. PALMER, 1998  Explosive invasion of plant mitochondria by a group I intron. Proc. Natl. Acad. Sci. USA 95:14244-14249[Abstract/Free Full Text].

COUVET, D., A. ATLAN, E. BELHASSEN, C. GLIDDON, and P.-H. GOUYON et al., 1990  Coevolution between two symbionts: the case of cytoplasmic male sterility in higher plants. Oxf. Surv. Evol. Biol. 7:225-250.

COUVET, D., O. RONCE, and C. GLIDDON, 1998  The maintenance of nucleocytoplasmic polymorphism in a metapopulation: the case of gynodioecy. Am. Nat. 152:59-70.

DE HAAN, A. A., A. C. MATEMAN, P. J. VAN DIJK, and J. M. M. VAN DAMME, 1997a  New CMS types in Plantago lanceolata and their relatedness. Theor. Appl. Genet. 94:539-548.

DE HAAN, A. A., H. P. KOELEWIJN, M. P. J. HUNDSCHEID, and J. M. M. VAN DAMME, 1997b  The dynamics of gynodioecy in Plantago lanceolata L. II. Mode of action and frequencies of restorer alleles. Genetics 147:1317-1328[Abstract].

FRANK, S. A., 1989  The evolutionary dynamics of cytoplasmic male sterility. Am. Nat. 133:345-376.

FRANK, S. A., 1997 Spatial processes in host-parasite genetics, pp. 325–352 in Metapopulation Biology: Ecology, Genetics, and Evolution, edited by I. HANSKI and M. E. GILPIN. Academic Press, San Diego.

GIGORD, L., C. LAVIGNE, J. A. SHYKOFF, and A. ATLAN, 1998  No evidence for local adaptation between cytoplasmic male sterility and nuclear restorer genes in the gynodioecious species Thymus vulgaris L. Heredity 81:156-163.

GOTTSCHALK, W. and M. L. H. KAUL, 1974  The genetic control of microsporogenesis in higher plants. Nucleus 17:133-166.

GOUYON, P.-H., F. VICHOT, and J. VAN DAMME, 1991  Nuclear-cytoplasmic male sterility: single point equilibria versus limit cycles. Am. Nat. 137:498-514.

HURST, L. D., A. ATLAN, and B. O. BENGTSSON, 1996  Genetic conflicts. Q. Rev. Biol. 71:317-364[Medline].

JOLLS, C. L. and T. C. CHENIER, 1989  Gynodioecy in Silene vulgaris (Caryophyllaceae): progeny success, experimental design, and maternal effects. Am. J. Bot. 76:1360-1367.

KOELEWIJN, H. P. and J. M. M. VAN DAMME, 1995a  Genetics of male sterility in gynodioecious Plantago coronopus. I. Cytoplasmic variation. Genetics 139:1749-1758[Abstract].

KOELEWIJN, H. P. and J. M. M. VAN DAMME, 1995b  Genetics of male sterility in gynodioecious Plantago coronopus. II. Nuclear genetic variation. Genetics 139:1759-1775[Abstract].

KÜCK, U., and G. WRICKE (Editors), 1995 Advances in Plant Breeding 18: Genetic Mechanisms for Hybrid Breeding. Blackwell Wissenschafts-Verlag, Berlin.

LEWIS, D., 1941  Male sterility in natural populations of hermaphroditic plants. New Phytol. 40:56-63.

LLOYD, D. G., 1974  Theoretical sex ratios of dioecious and gynodioecious angiosperms. Heredity 32:11-31.

MANICACCI, D., A. ATLAN, and D. COUVET, 1997  Spatial structure of nuclear factors in sex determination in the gynodioecious Thymus vulgaris L. J. Evol. Biol. 10:889-907.

MCCAULEY, D. E., 1998  The genetic structure of a gynodioecious plant: nuclear and cytoplasmic genes. Evolution 52:255-260.

MCCAULEY, D. E. and M. T. BROCK, 1998  Frequency-dependent fitness in Silene vulgaris, a gynodioecious plant. Evolution 52:30-36.

MCCAULEY, D. E. and D. R. TAYLOR, 1997  Local population structure and sex ratio: evolution in gynodioecious plants. Am. Nat. 150:406-419.

MCCAULEY, D. E., M. S. OLSON, S. N. EMERY, and D. R. TAYLOR, 2000  Population structure influences sex ratio evolution in a gynodioecious plant. Am. Nat. 155:814-819[Medline].

NUNNEY, L., 1990  Drosophila on oranges: colonization, competition, and coexistence. Ecology 71:1904-1915.

OLSON, M. S. and D. E. MCCAULEY, 2000  Linkage disequilibrium and phylogenetic congruence between chloroplast and mitochondrial haplotypes in Silene vulgaris. Proc. R. Soc. Lond. Ser. B 267:1-8[Medline].

PANNELL, J., 1997  The maintenance of gynodioecy and androdioecy in a metapopulation. Evolution 51:10-20.

PETTERSSON, M. W., 1992  Advantages of being a specialist female in nondioecious Silene vulgaris S. L. (Caryophyllaceae). Am. J. Bot. 79:1389-1395.

SAS INSTITUTE, 1985 SAS Users Guide: Statistics. Version 5. SAS Institute, Cary, NC.

SAS INSTITUTE, 1995 JMP Statistics and Graphics Guide. Version 3.1. SAS Institute, Cary, NC.

SAUMITOU-LAPRADE, P., G. J. A. ROUWENDAL, J. CUGEN, F. A. KRENS, and G. MICHAELIS, 1993  Different CMS sources found in Beta vulgaris ssp maritima: mitochondrial variability in wild populations revealed by a rapid screening procedure. Theor. Appl. Genet. 85:529-535.

SAUMITOU-LAPRADE, P., J. CUGEN, and P. VERNET, 1994  Cytoplasmic male sterility in plants: molecular evidence and the nucleocytoplasmic conflict. Trends Ecol. Evol. 9:431-435.

SCHNABLE, P. S. and R. P. WISE, 1998  The molecular basis of cytoplasmic male sterility and fertility restoration. Trends Plant Sci. 3:175-180.

SOKAL, R. R., and F. J. ROHLF, 1995 Biometry. Ed. 3. Freeman, New York.

TAYLOR, D. R., 1999  Genetics of sex ratio variation among natural populations of a dioecious plant. Evolution 53:55-62.

TAYLOR, D. R., D. MCCAULEY, and S. TRIMBLE, 1999a  Colonization success of females and hermaphrodites in the gynodioecious plant, Silene vulgaris. Evolution 53:745-751.

TAYLOR, D. R., M. J. SAUR, and E. ADAMS, 1999b  Variation in pollen performance and its consequences for sex ratio evolution in a dioecious plant. Evolution 53:1028-1036.

VAN DAMME, J. M. M., 1983  Gynodioecy in Plantago lanceolata L. II. Inheritance of three male sterility types. Heredity 50:253-273.

WEIR, B. S., 1990 Genetic Data Analysis. Sinauer Associates, Sunderland, MA.

WILSON, D. S., 1980 The Natural Selection of Populations and Communities. Benjamin/Cummings, Menlo Park, CA.

This article has been cited by other articles:

				M{a} L. Buide Disentangling the causes of intrainflorescence variation in floral traits and fecundity in the hermaphrodite Silene acutifolia Am. J. Botany, April 1, 2008; 95(4): 490 - 497. [Abstract] [Full Text] [PDF]

				M. F. Bailey and L. F. Delph Sex-Ratio Evolution in Nuclear-Cytoplasmic Gynodioecy When Restoration Is a Threshold Trait Genetics, August 1, 2007; 176(4): 2465 - 2476. [Abstract] [Full Text] [PDF]

				M. E. Welch, M. Z. Darnell, and D. E. McCauley Variable Populations Within Variable Populations: Quantifying Mitochondrial Heteroplasmy in Natural Populations of the Gynodioecious Plant Silene vulgaris Genetics, October 1, 2006; 174(2): 829 - 837. [Abstract] [Full Text] [PDF]

				P. K. Ingvarsson and D. R. Taylor Genealogical evidence for epidemics of selfish genes PNAS, August 20, 2002; 99(17): 11265 - 11269. [Abstract] [Full Text] [PDF]