Abstract
Assessing the rate of evolution depends on our ability to detect selection at several genes simultaneously. We summarize DNA sequence variation data in three new and six previously published data sets from the left arm of the second chromosome of Drosophila melanogaster in a population from West Africa, the presumed area of origin of this species. Four loci [Acp26Aa, Fbp2, Vha68-1, and Su(H)] were previously found to deviate from a neutral mutation-drift equilibrium as a consequence of one or several selective sweeps. Polymorphism data from five loci from intervening regions (dpp, Acp26Ab, Acp29AB, GH10711, and Sos) did not show the characteristic deviation from neutrality caused by local selective sweeps. This genomic region is polymorphic for the In(2L)t inversion. Four loci located near inversion breakpoints [dpp, sos, GH10711, and Su(H)] showed significant structuring between the two arrangements or significant deviation from neutrality in the inverted class, probably as a result of a recent shift in inversion frequency. Overall, these patterns of variation suggest that the four selective events were independent. Six loci were observed with no a priori knowledge of selection, and independent selective sweeps were detected in three of them. This suggests that a large part of the D. melanogaster genome has experienced the effect of positive selection in its ancestral African range.
YEARS after the neutralist challenge to the Darwinian theory (Kimura 1968), the interpretation of molecular variation still rests on an unresolved combination of the two main mechanisms of genetic change, selection and drift. Our difficulty in integrating these mechanisms bars us from properly using genomic data for assessing the level of natural selection involved in species evolution. An illustration of this inability can be found in the disagreement among studies on the scale at which hitchhiking events must be investigated in the genome (Maynard Smith and Haigh 1974; Kaplanet al. 1989). A study of the extent of a selective sweep on flanking regions in the Drosophila genome found a significant reduction in variation over ∼20 kb (Hudsonet al. 1997). Thus the selection event affected several neighboring genes. However, a recent study of the NF-κB/IκB region in the same species concluded that the expected window of reduced polymorphism caused by selective fixation of a beneficial mutant is only 200 bases (Begun and Whitley 2000), thus suggesting that selection can potentially be independently detected at virtually any Drosophila locus. The study of a selective sweep at the Sdic locus (Nurminskyet al. 2001) detected the signature of a unique selection event over a very large region spanning two divisions (∼2%) of the Drosophila melanogaster genome. No such pattern was found in the corresponding region for the related D. simulans, thus providing a neutral standard. This implicitly assumed that a selective event modifies variation patterns over a very large region of the genome. With such large effects, independent selection events could be detected in no more than 25 fruitfly genes.
We previously surveyed D. melanogaster molecular variation in Africa, where this species is thought to have originated (Lachaiseet al. 1988). In a sample from the Lamto ecological station (Ivory Coast), the signature of selective sweep events was previously recorded at four genes: the Accessory gland-specific peptide 26Aa gene (Acp26Aa; Aguadé 1998; Fay and Wu 2000), the Fat body protein-2 gene (Fbp2; Bénassiet al. 1999), the vacuolar ATPase 68 kD gene (Vha68-1; Depauliset al. 2000), and the Suppressor of Hairless gene [Su(H); Depauliset al. 1999]. These genes are located on the left arm of the second chromosome. Given this, one may ask whether these sweeps were driven by a single selection event or were causally independent. The In(2L)t inversion occurs on the same chromosome arm. Contrary to inversions found in the third chromosome of D. pseudoobscura (Dobzhansky and Queal 1938; Aquadroet al. 1991), the In(2L)t inversion of D. melanogaster is probably recent (Andolfattoet al. 1999) and has reached a high frequency in the Lamto population (61.4%, n = 88; Bénassiet al. 1993). A parsimonious interpretation would be that selection linked to this inversion has affected allele frequency at several loci simultaneously. Using available data on DNA variation at Fbp2, Vha68-1, and Su(H), Galtier et al. (2000) could exclude a demographic interpretation as a unique cause of departure from equilibrium. However, they could not exclude a single selective sweep involving all loci. The purpose of this study is to resolve this question. We recorded DNA sequence variation at genes located at intervening positions in this linear arrangement of genes. We also examined genetic structuring between chromosome arrangements. We show that independent selective events were involved. Overall, one-half of randomly examined genes were affected by selection. This is the first estimate of the number of selective events having occurred in the genome of D. melanogaster in its ancestral range. This proportion is much higher than was previously believed.
—Map of studied loci on the left arm of chromosome 2. Inversion In(2L)t (shaded arrow).
MATERIALS AND METHODS
Data acquisition: The isolation of 84 chromosome 2 isogenic lines from an Ivory Coast sample from the Lamto ecological station and the extraction of genomic DNA were previously described (Bénassiet al. 1993). Sequence polymorphism was recorded in the same random subsample of 20 lines as for other genes (Bénassiet al. 1993; Depauliset al. 1999). This sample includes 11 lines with standard chromosomes and 9 lines with In(2L)t chromosomes. A line of D. simulans was used as an outgroup.
Three loci were amplified according to standard polymerase chain reaction techniques and sequenced (primer sequences are given as supplemental data in Table S1, available at http://www.genetics.org/supplemental/). The following were sequenced: a fragment of the decapentaplegic (dpp) gene consisting of the last 441 bp of intron 2 and the first 656 bp of exon 3; a fragment of the Son of sevenless (Sos) gene sequence (Simonet al. 1991) consisting of the last 397 bp of exon 5, the 69 nucleotides of intron 5, the 634 bp of exon 6, and the first 49 bp of intron 6; and a fragment of the GH10711 putative gene consisting of 188 bp of exon 4, 794 bp of intron 4, and the first 159 bp of exon 5. They were aligned using ClustalX software (Thompsonet al. 1997) and checked manually.
Polymorphism at these loci was compared to sequence polymorphism data previously obtained from the same sample for Su(H) (Depauliset al. 1999) and for Vha68-1 (Depauliset al. 2000). It was also compared to other sequence polymorphism data obtained from different subsamples of our Ivory Coast sample: Fbp2 (n = 10; Bénassiet al. 1999), Acp26Aa and Acp26Ab (n = 24; Aguadé 1998), and Acp29AB (n = 14; Aguadé 1999). Figure 1 shows the position of these genes on the chromosomal map. Their genetic properties are summarized in Table 1. The whole area covers 46.5 cM of genetic map.
Data treatment: The effective number of synonymous and nonsynonymous sites, descriptive population statistics, and some statistical tests were performed using the DnaSP3.53 program (Rozas and Rozas 1999). Phylogenies were obtained using MEGA 2.1 (Kumaret al. 1994). The fixation index FST was computed according to Hudson et al. (1992). Recombination rates were computed from the comparisons of adjusted coefficients of exchange (Kindahl 1994) and the DNA content of chromosomal bands (Sorsa 1988).The multiple Hudson-Kreitman-Aguadé (HKA) test was run using a program obtained from Jody Hey (http://lifesci.rutgers.edu/~heylab/). Neutral expectations for haplotype diversity, haplotype number (Depaulis and Veuille 1998; Depauliset al. 2001), and Fay and Wu's (2000) H-statistic were obtained by generating coalescents following standard procedures (Hudson 1993). Coalescent simulations using the observed number of segregating sites were run with and without recombination. Recombination was then set at the value of 4Ner = 0.008 events/bp. This value is much lower than its estimated value in any of these genes and resulted in conservative tests. The program used for simulations (Allelix) can be obtained from S. Mousset (http://www.snv.jussieu.fr/mousset/).
RESULTS
Sequence alignments for newly sequenced loci are shown in Figure 2. Sequence alignments for Vha68-1 and Su(H) loci obtained from the same sample are shown in the supplemental data Table S2 at http://www.genetics.org/supplemental/. Summary statistics of gene polymorphism are shown in Table 2.
Sequence polymorphism at dpp, GH10711, and Sos: Sequence polymorphism data from 20 D. melanogaster and 1 D. simulans chromosomes were collected at dpp, GH10711, and Sos. In D. melanogaster at dpp, a total of 30 polymorphic sites (corresponding to 31 mutations) and five length polymorphisms were identified over the 1082 bp examined (excluding sites with alignment gaps) and formed 15 haplotypes. At GH10711, 60 polymorphic sites (61 mutations) and six length polymorphisms were identified over the 1074 bp examined and formed 14 haplotypes. At Sos, 23 polymorphic sites (23 mutations) were identified over the 1146 bp examined and formed 13 haplotypes. A minimum number of four recombination events between informative sites was inferred using the four-gamete rule (Hudson and Kaplan 1985) at both Sos and dpp, whereas at least nine events were inferred at GH10711. Replacement variants were usually found to occur at low frequency, except at GH10711 where two replacement variants occurred, respectively, in 4 and 14 of the 20 sequenced lines.
Genetic properties of studied loci
Neutrality tests: HKA tests (Hudsonet al. 1987) were run between all pairs of loci (Table 3). One locus, Vha68-1, consistently showed significant deviation from neutrality with all other loci, except Acp26Aa and Acp29AB. All other pairwise comparisons, except the one between Acp26Ab and Acp29AB, were nonsignificant. A multilocus HKA test, including all loci, did not show significant deviation from neutrality in the nine-loci sample (result not shown).
The H- and K-haplotype tests, as proposed by Depaulis and Veuille (1998), when run with a no recombination assumption (Table 4), were significant for Su(H) as previously shown (Depauliset al. 1999). Adding recombination made the H-test (based on haplotype diversity) also significant for Vha68-1. These tests were also performed separately for each chromosomal class. Significant deviation from neutrality was found using the H- and K-tests for the inverted and the standard class at the Su(H) locus. The H-test was significant for inverted chromosomes at the GH10711 locus, and the K-test was also significant for these chromosomes when using a conservative recombination rate. However, inverted chromosomes are not a random subsample and deviation from neutrality may result from sampling data within an allelic class (Innan and Tajima 1997, 1999). Hudson et al.'s (1994) haplotype test showed significant deviation from neutrality at Fbp2, as previously shown (Bénassiet al. 1999). This test uses a sliding window approach, which was inappropriate for the other loci, due to their short fragment length (see discussion).
Fay and Wu's (2000) test was not significant for any loci when run without recombination. However, it became significant with recombination at Acp26Aa and Sos (Table 5). Significant non-neutrality had previously been found at Acp26Aa, using a mixed African population sample from Ivory Coast and Malawi (Fay and Wu 2000), whereas our study includes only the Ivory Coast subset for consistency with our other data. The P values indicated marginally significant non-neutrality (P < 0.06) for three other loci: dpp, Vha68-1, and Su(H). This test was run independently on each chromosomal class and showed significant departure from neutrality at the Acp26Aa locus for standard lines and at Su(H) for In(2L)t lines.
Tajima's D (Tajima 1989) and Fu and Li's D (Fu and Li 1993) neutrality tests were always nonsignificant (results not shown). However, high positive values were found for the two genes that were significant for haplotype tests (Fu and Li's D = 0.682 for Fbp2 and D = 1.181 for Su(H)).
Genetic structuring between chromosomal arrangements: Three loci [dpp, Sos, and Su(H)] showed significant genetic structuring between chromosome arrangements (Table 6). These genes are the closest to inversion breakpoints, excluding Vha68-1, which shows too little variation for homogeneity tests to be applied. Genetic exchange between In(2L)t and standard lines was detected at six loci using Betrán et al.'s (1997) procedure.
Phylogenetic analysis of linkage with the inversion: The association of haplotypes with chromosome arrangements is illustrated in Figure 3 using neighbor-joining trees. Since recombination events or gene conversion was found to have occurred between In(2L)t and standard lines in these genes (see above), internal branches do not represent true descent relationships and, similarly, bootstrap values should be interpreted with caution. Except where gene conversion was found, the inverted chromosomes of dpp and GH10711 were clustered within one or two subsections of the tree. This pattern was less obvious at Sos. The result for dpp strongly suggests that a single recombination event occurred after the divergence between standard and In(2L)t, and this is responsible for the origin of a new family of standard haplotypes. This is compatible with indications that the age of the In(2L)t inversion is much less than the coalescence time of autosomes in highly recombining regions (Andolfattoet al. 1999). This is also compatible with previous observations showing an unusual haplotype structure of this gene in an American population, even though no inversions were present (Richteret al. 1997).
Summary statistics of sequence polymorphism data
—Alignment of polymorphic sites. The ancestral state (when available) inferred from D. melanogaster/D. simulans comparison was used as reference. –, 1-bp deletion; *, deletion >1 bp; i, intronic mutation; s, silent mutation; r, replacement mutation. Gene conversion tracts between inverted and standard chromosomes are underlined.
DISCUSSION
As D. melanogaster probably originated in Africa (Lachaiseet al. 1988), the genes considered here provide an opportunity to examine the effect of natural selection on DNA variation in an ancestral population living in its natural environment. Under these circumstances, it is more likely that this population is at equilibrium for demographic and genetic factors. Nine genes were examined in a population in which the In(2Lt) inversion shows a high frequency. The signature of selection was observed in four of these loci [Acp26Aa, Fbp2, Vha68-1, and Su(H)].
Pairwise HKA tests between loci
Detectability of selective sweep events: Detecting selective sweeps depends on the significance of neutrality tests. Given the depletion of nucleotide polymorphism after a complete sweep, variation is scarce and evidence of selection is difficult to obtain from tests such as Tajima's (1989) or Fu and Li's (1993) that use a single distribution of variation. Power is presumably greater for tests such as the HKA test (Hudsonet al. 1987) that use another locus as a reference. In partial selective sweeps, recombination has occurred between the selected and the surveyed loci during the selective phase, resulting in the preservation of a large amount of variation in recombining chromosomes. This leaves aberrant haplotype patterns (Kaplanet al. 1989), thus leading to increased power. Interestingly, most selective sweep events detected in Drosophila concern partial sweeps [e.g.,at Sod, White, Est-6, Acp70A, Fbp2, Su(H), and In(2L)t; Kirby and Stephan 1995; Cirera and Aguadé 1997; Andolfattoet al. 1999; Balakirevet al. 1999; Bénassiet al. 1999; Depauliset al. 1999]. This may indicate that selective sweeps affect polymorphism at many loci in addition to the actual targets of selection or that selective sweeps are more readily detected in partial sweeps. The haplotype tests that reveal partial selective sweeps are designed to detect linkage disequilibrium immediately after the sweep and are thus efficient only for recent events.
H- and K-haplotype tests
Using computer simulations Fay and Wu (2000; Figure 3) estimated that their test has no power to detect selective sweeps when the c/s ratio is >0.1 (where c is the rate of recombination between the observed and the selected loci and s is the haploid selection coefficient). The smallest physical distance between loci showing evidence of selection in our sample is 2 × 106 nucleotides (Table 1). The lowest recombination rate at studied loci is r = 6 × 10–9 event/bp. Assuming this value over the whole region, a favored gene occurring between two observed loci should have a selective advantage of ≥0.06 for the selective sweep to be detected at one of these loci with at least a 5% probability. The power of the other haplotype tests is probably not very different. It is therefore highly improbable that the partial selective sweeps observed at Acp26Aa, Fbp2, Vha68-1, and Su(H) result from the same selective event. The only two possibilities are either a shift in the frequency of the In(2L)t inversion, which extends over a large region, or population structuring.
Neutrality test based on derived mutations
Population structure was previously found in African D. melanogaster (Michalakis and Veuille 1996). Local genetic drift combined with a low level of migration could produce the observed pattern of variation. Population structure, however, increases the length of internal branches of a coalescent and thus should lead to positive Tajima's D values. This is not observed at the studied loci since four of the nine Tajima's D are negative (Table 2).
Below we examine whether a recent increase of In(2L)t frequency, as suggested by Andolfatto et al. (1999), could have caused a general deviation from neutral expectation. We then try to identify independent selective events and to assess whether selection was suspected before the study or found at random.
Linkage with the In(2L)t inversion: The frequency of In(2L)t varies widely in Africa (Veuilleet al. 1998). This may result from selection or from drift. The absence of nucleotide variation at its breakpoints suggests that the inversion is very young (Andolfattoet al. 1999). Old world populations of D. melanogaster are genetically differentiated for silent variation in Adh slow haplotypes, which are in linkage disequilibrium with this inversion (Bénassi and Veuille 1995). No such structuring was observed for microsatellites from other regions of the second chromosome (Michalakis and Veuille 1996). These observations suggest that the very high frequency of this inversion in Ivory Coast results from selection. In this study, genetic structuring between inverted and standard chromosomes was found for loci close to In(2L)t breakpoints [dpp, Sos, and Su(H)].
Genetic structuring between chromosomal arrangements
In a comparison of nucleotide diversity between In(2L)t and standard, inverted chromosomes were less polymorphic in eight out of nine loci (Table 6). This result was significant in a sign test (P = 0.02). The trend (nonsignificant) is the same for Watterson's estimator of nucleotide diversity, which should be equal to nucleotide diversity under a neutral model, since only seven out of nine loci are less polymorphic within the inverted than within the standard chromosomal class (P = 0.11). However, this trend suggests that the In(2L)t chromosomal class is less polymorphic. This is consistent with a recent increase of In(2L)t frequency (Andolfattoet al. 1999). The significant lack of haplotype diversity for In(2L)t at GH10711 (Table 4) could also result from a recent shift in In(2L)t frequency. However, these results should be interpreted with caution since a chromosomal class isolated from the rest of the population does not constitute a random sample of chromosomes, and applying neutrality tests on this subsample may lead to invalid conclusions (Innan and Tajima 1997, 1999).
A recent shift in In(2L)t frequency may have skewed the frequency spectrum of mutations at loci where significant structuring is observed with respect to the inversion. This may explain the low P values obtained from Fay and Wu's test for dpp, Vha68-1, Sos, and Su(H) (Table 5), which are always associated with significant FST values (Table 6). At Vha68-1 and Su(H), however, there is independent evidence of a selective sweep, suggesting that patterns of variation in each of these genes were altered by different events: a selective sweep and a shift in the inversion frequency. A single selective sweep occurs at a given time and extends over a given area. Considering the time parameter, Galtier et al. (2000) showed that a single bottleneck could not have caused the deviation from neutrality observed at Fbp2, Vha68-1, and Su(H). Polymorphism data from the three newly sequenced intervening loci (dpp, GH10711, and Sos) enable us to partition the deviations from neutrality previously detected at other loci [Acp26Aa, Fbp2,Vha68-1, and Su(H)] into effects from different selective events.
Selective sweep at Su(H): Sequence variation in this gene was substantial, and yet was distributed among few haplotypes, thus departing significantly from neutral equilibrium in a haplotype test (Depauliset al. 1999). This departure from neutrality was also observed in each chromosomal class (Table 4), suggesting that the shift in inversion frequency was not responsible for the observed pattern of variation at this locus. There were few singletons (5 out of 44 polymorphisms), suggesting that the sweep is recent and that few mutations have accumulated since. This polymorphic pattern makes it unlikely that the observed 1-kb DNA fragment was the focus of the sweep. More likely, selection occurred in a nearby region, within or outside Su(H), and some haplotypes were rescued from the sweep through recombination. Linkage disequilibrium with In(2L)t at Su(H) was first noticed during a microsatellite survey (Depauliset al. 1999). Sequence variation then showed that two phenomena were superimposed; there was structuring with the inversion, but there was also a selective sweep within each chromosomal arrangement. This means that this gene was taken at random with respect to the selective sweep, but not with respect to linkage disequilibrium with the inversion.
Selective sweep at Vha68-1: Variation in this gene was very low and significantly departed from neutrality in a HKA test (Hudsonet al. 1987) when other genes were used as a reference. An HKA test performed against the GH10711 locus led to significant results both within In(2L)t (χ2 = 8.441, P = 0.004) and within the standard class (χ2 = 5.474, P = 0.019), suggesting that the selective sweep was not driven by the inversion shift only. Moreover, the 11 variants found in this gene were not very frequent; 8 of them are singletons and the other 3 are rare variants. A selective sweep possibly occurred across this gene or very close to it, and observed polymorphisms probably result from mutations occurring afterward. In the standard arrangement, this gene is located between two neutral loci (GH10711 and Sos). This suggests that the selective sweep at this locus is independent of those found elsewhere. Departure from neutrality in this gene was initially unexpected when first observed, as the authors' purpose was to study population structuring at a gene close to In(2L)t (Depauliset al. 2000).
—Neighbor-joining trees of haplotypes at dpp, GH10711, and Sos. Bootstrap percentages were calculated among 5000 replicates; sequences for which genetic exchange between In(2L)t and standard lines was identified are underlined.
Selective sweep at Fbp2: Sequence variation in this gene deviated from equilibrium using a haplotype test based on the frequency of the major haplotype (Hudsonet al. 1994; Kirby and Stephan 1995). This gene lies within the In(2L)t inversion. Of 10 sequences, 5 showed the same haplotype over the entire coding unit, 3 of them belonged to the In(2L)t class, and 2 belonged to the standard class. The Fbp2 gene probably underwent hitchhiking due to selection at a nearby locus (Bénassiet al. 1999). The driving gene of this selective sweep probably lies close to Fbp2 within the inversion. Nucleotide variation was first investigated at Fbp2 to examine amino acid variation as this gene encodes a protein that is methionine rich (Ratet al. 1991; Meghlaoui and Veuille 1997). There was no segregating methionine in the data set and no methionine was fixed in the D. melanogaster lineage since its speciation with D. simulans. However, inspection of the silent sites showed evidence of hitchhiking due to an undetermined selection event. After this study, we changed our sampling method and recorded polymorphism >1 kb from intronic regions in 20 lines rather than 2 kb from coding regions in 10 lines. This change aimed to use a priori tests like H- and K-tests rather than a posteriori tests that use a sliding window approach. Despite these changes in the sampling design, Fbp2 significantly departed from neutrality. Therefore, it is valid to compare Fbp2 results with those of other loci, even though more satisfying tests were applied to them.
Selective sweep at Acp26Aa: In this study, we confirmed that sequence variation in this gene deviated from neutral equilibrium using Fay and Wu's test (Fay and Wu 2000) for our population sample. This test remained significant when run on standard chromosomes, showing that the shift in In(2L)t frequency is unlikely to be the cause of departure from neutrality. Fay and Wu (2000) previously showed this result to be associated with the absence of polymorphism >200 bp in this gene (Aguadé 1999). As Drosophila Acp genes encode sex peptides that are thought to contribute to sperm competition, hypotheses on selection clearly motivated population genetics studies in these genes (Aguadéet al. 1992; Aguadé 1998, 1999), meaning that Acp26Aa was not chosen at random in the Drosophila genome.
Proportion of selective events in a random sample of genes: In this study, selection was shown to have independently affected variation patterns at four genetic loci [Acp26Aa, Fbp2, Vha68-1, and Su(H)]. The contrasting patterns of polymorphism observed between loci in our African sample make alternative demographic explanations, such as bottleneck, founder effects, or population structure unlikely, although these events may have increased the variance over loci. In addition, a selective sweep was probably transmitted to several genes through a change in In(2L)t frequency. An obvious consequence of this shift in inversion frequency is the lower level of polymorphism in inverted chromosomes, as previously observed by Andolfatto et al. (1999) at the proximal breakpoint of this inversion. As the four regions lying proximally and distally to breakpoints cosegregate as a single unit, an inversion can also be considered a “locus.”
However, this sample of loci was not taken at random, since selection was already suspected for Acp genes, and since inversion polymorphisms are considered potential targets for selection in many population genetics studies. Sequence variation was blindly examined at only six loci. Three loci [Fbp2, Vha68-1, and Su(H)] out of the six deviated from neutrality. Some of these deviations may be due to a demographic event. For instance, we would expect about three genes out of six to depart significantly from neutrality using a test with 50% power to detect an event. Although this proportion is imprecise, it suggests that “footprints” of positive selection are present in a substantial proportion of genes.
D. melanogaster is a reference organism that has been used in many evolutionary studies. Most of these studies were conducted in “derived” populations that may have been affected by adaptation to new environments and to founding events. Such a high proportion of non-neutral loci has been previously recorded in a sample of 20 genes from highly recombining regions in D. melanogaster (Moriyama and Powell 1996). However, this initial review considered data from studies carried out on a nonrandom sample of genes. Furthermore, all of these elementary studies consisted of data for derived populations of D. melanogaster. The present study is the first to consider sequence polymorphism in several genes from the same African population of D. melanogaster, thus providing a rough idea of selection pressure in this organism in its original habitat. The shift in In(2L)t frequency was observed both at neutral loci (dpp, Sos, GH10711) and at Su(H). In dpp, a family of haplotypes recombined away from the inversion, thus generating a singular haplotype structure of the sample. It is probably the same pattern that was observed in a New Jersey population (Richteret al. 1997). However, the low frequency of this inversion in North America did not allow the authors to make any definitive conclusions. Our observations show that haplotype patterns can have a complex historical origin and that populations are likely to be by no means simple. The impact of selection upon Drosophila genome variation in the recent history of this species appears very substantial. Chromosomal inversions could play an important role in shaping variation along chromosomes. It would be of interest to survey molecular polymorphism from regions where no common inversions occur in the range of the species.
Acknowledgments
We thank two anonymous reviewers for valuable comments and suggestions, Michèle Huet for technical assistance, and John Ewen and Karen McCoy for help in preparing the manuscript. This research was financially supported by the Centre National de la Recherche Scientifique Groupe de Recherche 1928.
Footnotes
-
Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accession nos. AF459524–AF459586.
-
Communicating editor: M. Aguadé
- Received May 25, 2002.
- Accepted October 31, 2002.
- Copyright © 2003 by the Genetics Society of America
