Ftp.bork.embl-heidelberg.de

Linkage limits the power of natural selection
in Drosophila
Andrea J. Betancourt*† and Daven C. Presgraves†
Department of Biology, University of Rochester, Rochester, NY 14627 Edited by M. T. Clegg, University of California, Riverside, CA, and approved August 26, 2002 (received for review May 8, 2002) Population genetic theory shows that the efficacy of natural
First, we ask whether protein evolution is constrained by linkage.
selection is limited by linkage—selection at one site interferes with
Second, we ask whether the efficacy of weak selection is com- selection at linked sites. Such interference slows adaptation in
promised by a stream of strongly selected traffic at nearby sites.
asexual genomes and may explain the evolutionary advantage
of sex. Here, we test for two signatures of constraint caused
by linkage in a sexual genome, by using sequence data from 255
The Data. Coding sequences from 102 D. melanogaster and D.
Drosophila melanogaster and Drosophila simulans loci. We find
simulans genes were downloaded from GenBank, and aligned by that (i) the rate of protein adaptation is reduced in regions of low
eye using SE-AL V. 1.0. For genes with multiple transcripts of recombination, and (ii) evolution at strongly selected amino acid
different lengths, we used the longest transcript; for those with sites interferes with optimal codon usage at weakly selected,
multiple transcripts of the same length, we used an arbitrarily tightly linked synonymous sites. Together these findings suggest
selected transcript. The remaining 153 genes (for which W.
that linkage limits the rate and degree of adaptation even in
Swanson generously provided alignments) come from a D. recombining genomes.
simulans male-specific EST screen (20) and their D. melanogas- ter homologues, downloaded from Flybase (http:͞͞flybase.bio.
Natural selection is imperfect. To become fixed, beneficial indiana.edu).WeusedonlycodingESTsfromthisscreenthatwere
mutations must overcome both stochastic loss and inter- either nonredundant or that were the most rapidly evolving of a set ference from selection at linked loci. In asexual genomes, where of redundant ESTs. Although ESTs are unreplicated single-pass linkage is complete, two kinds of interference compromise sequences, none of the results in this paper are caused by a adaptation. The first, ‘‘ruby-in-the-rubbish’’ interference, occurs difference in sequencing error rate (and consequent elevated because beneficial mutations often appear on genetic back- divergence estimates) between the EST and reference quality data, grounds loaded with segregating deleterious mutations. Since as all of our results are also found within the EST data set.
deleterious mutations are, on average, probably more strongly selected than favorable ones, adaptation is mostly limited to Estimates of Divergence, Recombination, and Codon Usage. We used
those lucky few beneficial mutations that arise on unloaded maximum likelihood estimates of the rates of amino acid (dN) backgrounds (1–4). The second form of interference, ‘‘clonal’’ and synonymous (dS) site divergence using PAML (22). One interference, is caused by competition among multiple segre- anomalously high dS value (1.71) was excluded from the analysis gating beneficial mutations (1, 5–7). Because only one asexual (excluding this value does not affect the results). We used a genome can be fixed at a time, adaptive substitutions are forced likelihood ratio test of Goldman and Yang (23) to test for a to be nearly sequential. Both kinds of interference limit the rate significant excess of amino acid substitutions in those genes with dN͞dS Ͼ 1. Briefly, we used PAML to calculate the likelihood The effects of both kinds of interference can be thought of as under the null model of equal dN and dS (L0) and the alternative a reduction in the effective population size (Ne) caused by model with dN and dS free to vary (L1). The likelihood test selection at linked loci (11–13). Recombination, by alleviating statistic (Ϫ2[ln L0 Ϫ ln L1]) was then compared with the ␹2 interference between linked sites, alleviates this reduction in Ne.
distribution with 1 degree of freedom. dN͞dS Ͼ 1 is strong Consequently, genomic regions that differ in recombination rate evidence for adaptive evolution, as this test is conservative.
also differ in effective size—indeed, this is the basis of the We estimated GC-content and the frequency of optimal codon well-known correlation between recombination rate and levels usage (Fop) by using the online program CODONW (http:͞͞ of neutral polymorphism (4Ne times the neutral mutation rate) bioweb.pasteur.fr͞seqanal͞interfaces͞codonw.html). Fop is the seen in the genomes of Drosophila, humans, and others (14–18).
proportion of codons in a gene that are optimal codons, defined That variation in linkage affects levels of neutral polymorphism as those used most frequently in highly expressed genes (24).
suggests that it may also affect rates of nonneutral substitution.
Optimal codons for both species were assumed to be those of D. In particular, adaptive evolution may be limited in regions of low melanogaster. Although we estimate Fop from a single sequence recombination (i.e., where Ne is reduced) or in situations of from each species, these estimates should accurately reflect extreme linkage (e.g., among sites within the same gene).
population levels of optimal codon usage (25).
Here we ask whether linkage systematically constrains adap- We estimated recombination rates by using the data and tation in the Drosophila genome. We use divergence estimates standard method of Kliman and Hey (26). For the X, second, and from 255 Drosophila melanogaster and Drosophila simulans loci.
third chromosomes, we fit least-squares polynomial curves re- These data are unique in that they include a large number of lating recombination rate to DNA content per interval on the rapidly evolving genes, many of which are candidate male cytological map (all curves have R2 Ͼ 0.989), and used equations accessory gland proteins (Acps) and thus likely targets of sexual from these curves (available upon request) to predict recombi- selection (19–21). We have reason to believe a priori not only nation rates. Estimates for X-linked loci were multiplied by 4͞3 that many of these substitutions are adaptive, but also that many to correct for the absence of recombination in Drosophila males.
of these genes may have experienced a long history of sexual selection possibly predating the D. melanogaster–D. simulans split. Such genes are especially good candidates for detecting the This paper was submitted directly (Track II) to the PNAS office.
signature of interference. We therefore use these data to test for *To whom correspondence should be addressed. E-mail: [email protected].
two kinds of limits imposed on adaptive evolution by linkage.
†A.J.B. and D.C.P. contributed equally to this work.
13616 –13620 ͉ PNAS ͉ October 15, 2002 ͉ vol. 99 ͉ no. 21
www.pnas.org͞cgi͞doi͞10.1073͞pnas.212277199 For the Y and dot-fourth chromosomes, we assume recombina- tion is absent. The mean recombination rate in this data set is c ϭ 0.0029 centimorgan (cM)͞kb, close to the global mean for the D. melanogaster genome (see supplimentary information for ref.
27). We therefore defined ‘‘low’’ recombination as c Ͻ 0.0029 and ‘‘high’’ as c Ͼ 0.0029. We used recombination rate estimates from D. melanogaster for both species, as recombination data for D. simulans are sparse. Although recombination rates in D. simulans may differ (they are, on average, Ϸ30% higher than in D. melanogaster; see ref. 28), subtle local changes in recombi- nation rate are unlikely to lead to misclassification of loci into high vs. low recombination regions. Changes in recombination rate that do result in misclassification likely contribute noise to our analysis, but this would obscure rather than create the Statistical Tests. For nonnormally distributed variables, we used
nonparametric Spearman rank correlations to test for associa- tions and permutation t tests (with null distributions generated by Ն10,000 randomization of the data) to test for differences between means. For multivariate analyses, nonnormal variables were first normalized by log-transformation. All tests are two- tailed. Means are reported with Ϯ 1 SE. All data are available online in Table 1, which is published as supporting information Results and Discussion
Does Linkage Limit Protein Evolution? We first ask whether linkage
limits protein adaptation. Rates of adaptive protein evolution, if limited by interference, should be relatively constrained in regions of low versus high recombination. We therefore test for an excess (paucity) of rapid evolution in regions of high (low) recombination. It is important to note, however, that slowly evolving genes (those mostly subject to purifying selection) should occur in all recombinational environments. A plot of dN vs. recombination should thus reveal a wedge-shaped distribu- tion, with slowly evolving genes in regions of low recombination and both slowly and rapidly evolving genes in regions of high N, the rate of amino acid substitution (a), and dS, the rate of silent substitution (b), vs. recombination rate (in cM͞kb). Black and gray circles are recombination. Fig. 1a confirms this prediction; genes in high putative Acp’s and non-Acp’s, respectively.
recombination environments show both a higher mean and variance of dN values than those in low recombination environ- ments (dN,high ϭ 0.031 Ϯ 0.003; dN,low ϭ 0.019 Ϯ 0.002; t test ϭ data (24.4%), our results do not depend on some peculiarity of 2.780, P ϭ 0.007, F151,103 ϭ 2.524, P Ͻ 0.0001; a qualitatively these genes other than their rapid evolution. Not surprisingly, if similar pattern appears in a plot of dN͞dS vs. recombination rate, all candidate Acp’s are excluded from the analysis, too few rapidly evolving genes remain to detect a pattern. But within We can eliminate three alternative explanations for this Acp’s, rapid protein evolution is also largely confined to regions regional difference in evolutionary rates. First, this wedge pat- of high recombination (dN,high ϭ 0.057 Ϯ 0.007; dN,low ϭ 0.029 Ϯ tern is not caused by regional differences in mutational input (as 0.007; t test ϭ 2.369, P ϭ 0.017; F44,18 ϭ 2.884, P ϭ 0.013; dS,high might be seen, e.g., if recombination were mutagenic) as such ϭ 0.111 Ϯ 0.011; dS,low ϭ 0.101 Ϯ 0.013; t test ϭ 0.514, P ϭ 0.601; differences would produce a similar pattern for dS. The distri- F44,18 ϭ 1.800, P ϭ 0.155). We therefore conclude that rates of EVOLUTION
S is not, however, wedge-shaped; neither the means protein adaptation are constrained in the low recombination nor the variances differ between regions of high and low environments of the Drosophila genome, as predicted by popu- recombination (Fig. 1b; dS,high ϭ 0.118 Ϯ 0.006, dS,low ϭ 0.108 Ϯ 0.007; t test ϭ 1.081, P ϭ 0.282; F150,103 ϭ 1.012, P ϭ 0.947).
Second, this result is also inconsistent with most rapid evolution Does Rapid Protein Evolution Limit Weak Selection at Linked Sites?
being caused by the fixation of many slightly deleterious muta- We now turn to a second test of the effect of linkage on tions. In that case, we would expect higher rates of evolution in adaptation. We ask whether evolution at strongly selected regions of low recombination (where Ne is reduced by interfer- (amino acid) sites limits the efficacy of selection at tightly linked, ence). Other studies have noted such elevated rates of protein weakly selected (synonymous) sites. Synonymous sites are not divergence in low recombination regions, likely because, in selectively equivalent, as certain synonymous codons (‘‘optimal’’ contrast to our data set, most of the genes in these studies are or ‘‘preferred’’ codons) are used more frequently than others, relatively conserved and therefore subject mainly to purifying apparently because of selection for translational efficiency and selection (29–32). Although many of the genes in this study are accuracy (33–36). But selection coefficients acting on alternative also relatively conserved, the signal of interference on adaptive synonymous codons are on the order of 1͞Ne, i.e., approaching evolution detected here (elevated dN in high recombination the limits of selection (36). Such weakly selected sites are regions) is detectable despite the opposing signal from relaxed especially susceptible to interference from selection at linked purifying selection (elevated dN in low recombination regions).
sites (11, 37–39). Two particularly relevant studies (40, 41) have Third, although candidate Acp’s constitute a large fraction of the shown, using different models, that a series of strongly selected PNAS ͉ October 15, 2002 ͉ vol. 99 ͉ no. 21 ͉ 13617
that this is indeed the case. As the rate of amino acid substitution increases, third position GC-content decreases significantly (dN, D. melanogaster: rs ϭ Ϫ0.540, P Ͻ 0.0001; dN, D. simulans: rs ϭ Ϫ0.549, P Ͻ 0.0001; dN͞dS, D. melanogaster: rs ϭ Ϫ0.569, P Ͻ 0.0001; dN͞dS, D. simulans: rs ϭ Ϫ0.584, P Ͻ 0.0001).
Because Acp’s as a class show both low optimal codon usage (19) and rapid protein evolution (19, 20), it is possible that the above findings are entirely due to some special property of these genes. We can rule out this possibility, however. The correlation between dN and optimal codon usage remains even when Acp’s are excluded from the analysis (D. melanogaster: rs ϭ Ϫ0.452, P Ͻ 0.0001; D. simulans: rs ϭ Ϫ0.431, P Ͻ 0.0001). Moreover, the correlation exists, and is in fact stronger, within Acp’s (D. melanogaster: rs ϭ Ϫ0.695, P Ͻ 0.0001; D. simulans: rs ϭ Ϫ0.711, P Ͻ 0.0001). This stronger correlation probably reflects the fact that other factors known to contribute to variation in optimal codon usage—tissue specificity, gene expression level, and gene length (reviewed in refs. 34 and 35)—are partially controlled within Acp’s, as these genes share a common tissue type (male accessory glands), similar (high) expression levels, and similar Because there is reason to believe that both protein evolution and optimal codon usage are related to gene length (44, 45) and recombination rate (refs. 25 and 45, and see above), we tested the possibility that the correlation between dN and Fop is an artifact of one of these other relationships. We find that optimal codon usage is significantly correlated with gene length (D. melano- gaster: rs ϭ Ϫ0.142, P ϭ 0.0287; D. simulans: rs ϭ Ϫ0.129, P ϭ 0.0463), as seen in previous studies (42, 43), but not with recombination (D. melanogaster: rs ϭ Ϫ0.066, P Ͼ 0.05; D. simulans: rs ϭ 0.081, P Ͼ 0.05). [The previously reported correlation between optimal codon usage and recombination is weak and detected in a much larger data set than ours (26, 45).] To distinguish the effects of gene length and protein evolution on optimal codon usage, we estimated partial correlation coef- ficients. Both relationships persist, but the correlation between dN and Fop is much stronger (partial r for gene length vs. dN in Fop, the frequency of optimal codon usage, vs. dN in D. melanogaster D. melanogaster: Ϫ0.163 vs. Ϫ0.505; in D. simulans: Ϫ0.152 vs.
(a) and D. simulans (b).
Ϫ0.518; gene length and dN are log-transformed; P Ͻ 0.05 substitutions reduces the rate of substitution of linked weakly Several workers (19, 46, 47) have observed high nonoptimal favored mutations. This effect is caused by genetic hitchhiking; codon usage in rapidly evolving genes, but suggested relaxation as the frequency of hitchhiking increases, weakly selected linked of selective constraints as the cause. These relaxed constraints sites behave more neutrally and thus come to more closely reflect explanations come in two flavors. The first invokes codon- the mutational spectrum. Because there are more ways, on specific constraints. Akashi (48) has argued that selection for average, to mutate to nonoptimal codons, the net effect translational accuracy (i.e., selection against misincorporation of of hitchhiking is to increase the fixation rate of nonoptimal amino acids) should be strongest at functionally important residues. Consistent with this, he found that evolutionarily conserved residues tend to use preferred codons, which are less We test this prediction by asking whether the frequency of often mistranslated (48). This explanation alone cannot account optimal codon usage (Fop) decays as the rate of protein evolution for the results reported here, because even conserved sites in our increases. As Fig. 2 shows, Fop declines sharply as dN increases rapidly evolving genes show poor optimal codon usage. After (statistics identical for both species: rs ϭ Ϫ0.559, P Ͻ 0.0001; the expunging all divergent codons from the genes in our data, we same patterns appear using dN͞dS ratios, not shown). dS shows find that the relationship between F a weak positive correlation with F op and dN is unchanged (D. op (D. melanogaster: rs ϭ 0.172, P Ͻ 0.0061; D. simulans: r s ϭ Ϫ0.546, P Ͻ 0.0001; D. simulans: rs ϭ Ϫ0.555, s ϭ 0.224, P ϭ 0.0004), but, as partial correlation analysis shows, the relationship between Fop and dN The second relaxed constraints explanation invokes gene- is independent of dS and twice as strong (D. melanogaster: partial specific constraints, in which constraints on amino acid sites and r ϭ Ϫ0.505 vs. 0.251; D. simulans: partial r ϭ Ϫ0.491 vs. 0.180).
synonymous sites are correlated within a gene, i.e., genes evolv- The effect of strongly selected traffic on linked synonymous sites ing rapidly because of relaxed selection on amino acid compo- can be illustrated in another way. In Drosophila, AT-biased sition are likely to also have relaxed selection for optimal codon mutation pressure causes GC-content at mutational equilibrium usage (49). There are three reasons to think this explanation to approach Ϸ35% (42, 43). Coding sequences are nevertheless does not explain our results. First, the rapid evolution in our data highly GC-biased, at least partly because of constraints on codon set does not appear to be caused by relaxed purifying selection, usage as virtually all preferred codons end in C or G. If a stream as evidenced by the relationship between rates of recombination of strongly selected amino acid traffic depresses Ne at weakly and protein evolution (barring some spurious relationship be- selected linked synonymous sites, GC-content at those sites tween high recombination rate and relaxed constraint). Second, should more closely reflect the mutational spectrum. We find if the rapid evolution in our data were mostly caused by relaxed 13618 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.212277199
constraints, GC content in both synonymous and amino acid sites high recombination. There is some indication, though, that should decay in those loci with high dN͞dS because of AT-biased genomes are not luxuriant. A comparative quantitative trait mutation pressure. Although dN͞dS is negatively related to GC locus study in domesticated cereal species found that convergent content of synonymous sites (see above), no such negative traits map to homeologous genomic regions (and therefore relationship exists for amino acid sites (D. melanogaster: rs ϭ possibly to the same genes), suggesting that there may be a 0.204, P ϭ 0.0013; D. simulans: rs ϭ 0.214, P ϭ 0.0007). Third, limited number of ways to construct these traits genetically (51).
interference is the only explanation for why genes with dN͞dS Ͼ More convincingly, high resolution molecular and experimental 1 (i.e., genes whose rapid evolution is caused by positive selec- evolution studies have uncovered convergence at the DNA tion, not relaxed constraints) should show depressed codon sequence level (52–54), suggesting that selection at least some- usage: we find that mean Fop for loci with dN͞dS Ͼ 1 is times uses the same nucleotide repeatedly. If the Drosophila significantly lower than that for loci with dN͞dS Ͻ 1 (D. mela- genome is not luxuriant, then our results imply that flies are not nogaster: Fop, dN/dS Ͻ 1 ϭ 0.534 Ϯ 0.009 vs. Fop, dN/dS Ͼ 1 ϭ 0.355 Ϯ perfectly adapted because of the slower average response of 0.023, t test ϭ 6.645, P Ͻ 0.0001; D. simulans: Fop, dN/dS Ͻ 1 ϭ genes in regions of low recombination to directional selection.
0.544 Ϯ 0.009 vs. Fop, dN/dS Ͼ 1 ϭ 0.356 Ϯ 0.022, t test ϭ 7.077, P Ͻ The weakly selected silent sites of rapidly evolving genes, in 0.0001). Even when we consider only those loci with dN͞dS contrast, seem more clearly maladapted. With perfect recombi- significantly greater than 1 (a conservative standard), Fop re- nation, rapidly evolving genes could both substitute beneficial mains significantly depressed (D. melanogaster: Fop, dN/dS Ͻ 1 ϭ amino acids and maintain optimal codon usage.
0.516 Ϯ 0.009 vs. Fop, dN/dS Ͼ 1 ϭ 0.411 Ϯ 0.084, t test ϭ 2.999, P ϭ The detrimental effects of interference appear hierarchical in 0.0020; D. simulans: Fop, dN/dS Ͻ 1 ϭ 0.523 Ϯ 0.009 vs. Fop, dN/dS Ͼ 1 that linkage constrains protein adaptation, which in turn con- ϭ 0.415 Ϯ 0.077, t test ϭ 2.878, P ϭ 0.0044). Taken together, these strains codon adaptation. Formally, either clonal or ruby-in-the lines of evidence suggest that neither flavor of relaxed constraint rubbish interference can limit protein evolution. But ruby-in- hypothesis accounts for the correlation between rapid protein the-rubbish seems likely more important as the rate of delete- evolution and low optimal codon usage. Instead, it seems that rious mutation (55–57), and so the opportunity for interference interference from strongly selected traffic compromises weakly from deleterious mutations, far exceeds the rate of favorable selected codon usage at tightly linked sites.
mutation. Furthermore, low recombination regions in Drosoph- ila may suffer an additional load of deleterious mutations Concluding Remarks. We find evidence that linkage constrains—
because of the higher numbers of transposable element inser- and recombination facilitates—adaptation in Drosophila. We tions found there (58). Assuming that adaptive protein diver- have shown that (i) the rate of protein adaptation appears gence is mostly limited by linkage to deleterious mutations, the limited by interference in low recombination regions, and (ii) hierarchical effects of interference may reflect the relative strong directional selection on proteins interferes with weak magnitudes of selection coefficients in Drosophila: mean selec- selection for optimal codon usage at linked sites. That such limits on adaptation are detectable even in a recombining tion against deleterious mutations is probably stronger than that genome is surprising and has several implications. It follows, favoring beneficial amino acid mutations, which in turn is larger for example, that the limiting effects of linkage on protein than that favoring preferred codons (i.e., sd Ϸ 10Ϫ2 Ͼ sb,amino acid adaptation may be manifest in the genetic basis of phenotypic Ϸ 10Ϫ3 to 10Ϫ4 Ͼ sb,codon Ϸ 10Ϫ6; see refs. 34, 56, and 59). This evolution. As Birky and Walsh (50) point out in their classic does not, however, mean that nonoptimal codon usage has a study of the theory of linkage and selection, ‘‘recombination negligible effect on genomes. Although selection on individual enhances the rate of phenotypic evolution, to the extent that preferred codons is weak (33), the cumulative effect of many phenotypic evolution is driven by the fixation of advantageous unpreferred codons may be considerable (33, 37).
mutations.’’ We might expect, therefore, that adaptive species differences should map disproportionately to high recombi- We thank P. Andolfatto, D. Begun, J. Bollback, Y. Chen, J. Gillespie, J.
Huelsenbeck, C. Jones, Y. Kim, W. Stephan, two anonymous reviewers, and especially A. Orr for helpful comments and discussion. This work was It does not necessarily follow, however, that genes in low supported by National Institutes of Health Grant GM526738 and by recombination regions are maladapted. If genomes are luxuriant, funding from the David and Lucile Packard Foundation (to A. Orr), and so that there are many ways to adapt to new environments, by Caspari Fellowships and Messersmith Fellowships (to A.J.B. and adaptation will simply proceed via substitutions in regions of 1. Fisher, R. A. (1930) The Genetical Theory of Natural Selection (Oxford Univ.
19. Begun, D. J., Whitley, P., Todd, B. L., Waldrip-Dail, H. M. & Clark, A. G.
(2000) Genetics 156, 1879–1888.
EVOLUTION
2. Manning, J. T. & Thompson, D. J. (1984) Acta Biotheor. 33, 219–225.
20. Swanson, W. J., Clark, A. G., Waldrip-Dail, H. M., Wolfner, M. F. & Aquadro, 3. Peck, J. (1994) Genetics 137, 597–606.
C. F. (2001) Proc. Natl. Acad. Sci. USA 98, 7375–7379.
4. Orr, H. A. (2000) Genetics 155, 961–968.
21. Wolfner, M. F. (1997) Heredity 88, 85–93.
5. Muller, H. J. (1932) Am. Nat. 66, 118–138.
22. Yang, Z. (1997) Comput. Appl. Biosci. 13, 555–556.
6. Crow, J. F. & Kimura, M. (1965) Am. Nat. 99, 439–450.
23. Goldman, N. & Yang, Z. (1994) Mol. Biol. Evol. 11, 725–736.
7. Gerrish, P. J. & Lenski, R. E. (1998) Genetica 102͞103, 127–144.
24. Ikemura, T. (1981) J. Mol. Biol. 151, 389–409.
8. de Visser, J. A. G. M., Zeyl, C. W., Gerrish, P. J., Blanchard, J. L. & Lenski, 25. McVean, G. A. T & Charlesworth, B. (1999) Genet. Res. 74, 145–158.
R. L. (1999) Science 283, 404–406.
26. Kliman, R. M. & Hey, J. (1993) Mol. Biol. Evol. 10, 1239–1258.
9. Miralles, R., Gerrish, P. J., Moya, A. & Elena, S. F. (1999) Science 285, 1745–1747.
27. Marais, G., Mouchiroud, D. & Duret, L. (2001) Proc. Natl. Acad. Sci. USA 98,
10. Miralles, R., Moya, A. & Elena, S. F. (2000) J. Virol. 74, 3566–3571.
11. Hill, W. G. & Robertson, A. (1966) Genet. Res. 8, 269–294.
28. True, J. R., Mercer, J. M. & Laurie C. C. (1996) Genetics 142, 507–523.
12. Felsenstein, J. (1974) Genetics 78, 737–756.
29. Hilton, H. R., Kliman, R. M. & Hey, J. (1994) Evolution (Lawrence, Kans.) 48,
13. Lewontin, R. C. (1974) The Genetic Basis of Evolutionary Change (Columbia 30. Pal, C., Papp, B. & Hurst, L. D. (2001) Mol. Biol. Evol. 18, 2323–2326.
14. Begun, D. J. & Aquadro, C. F. (1992) Nature 356, 519–520.
31. Comeron, J. & Kreitman, M. (2000) Genetics 156, 1175–1190.
15. Nachman, M. (1997) Genetics 147, 1303–1316.
32. Bachtrog, D. & Charlesworth, B. (2002) Nature 416, 323–326.
16. Stephan, W. & Langley, C. H. (1998) Genetics 150, 1585–1593.
33. Powell, J. R. & Moriyama, E. N. (1997) Proc. Natl. Acad. Sci. USA 94,
17. Nachman, M. W., Bauer, V. L., Crowell, S. L. & Aquadro, C. F. (1998) Genetics 150, 1133–1141.
34. Akashi, H. & Eyre-Walker, A. (1998) Curr. Biol. 8, 688–693.
18. Dvorak, J., Luo, M.-C. & Yang, Z.-L. (1998) Genetics 148, 423–434.
35. Kreitman, M. & Antezana, M. (2000) in Evolutionary Genetics: From Molecules PNAS ͉ October 15, 2002 ͉ vol. 99 ͉ no. 21 ͉ 13619
to Morphology, eds. Singh, R. S. & Krimbas, C. B. (Cambridge Univ. Press, New 50. Birky, C. W. & Walsh, J. B. (1988) Proc. Natl. Acad. Sci. USA 85, 6414–6418.
51. Paterson, A. H., Yann-Rong, L., Zhikang, L., Schertz, K. F., Doebley, J. F., 36. Akashi, H. (1995) Genetics 139, 1067–1076.
Pinson, S. R. M., Liu, S.-C., Stansel, J. W. & Irvine, J. E. (1995) Science 269,
37. Barton, N. H. (1995) Genetics 140, 821–841.
38. Stephan, W., Charlesworth, B. & McVean, G. (1999) Genet. Res. 73, 133–146.
52. Stewart, C.-B. & Wilson, A. C. (1987) Cold Spring Harbor Symp. Quant. Biol. 39. McVean, G. A. T. & Charlesworth, B. (2000) Genetics 155, 929–944.
52, 891–899.
40. Barton, N. H. (1994) Genet. Res. 64, 199–208.
53. ffrench-Constant, R. H. (1994) Insect Biochem. Mol. Biol. 24, 335–345.
41. Gillespie, J. H. (2001) Evolution (Lawrence, Kans.) 55, 2161–2169.
54. Wichman, H. A., Badgett, M. R., Scott, L. A., Boulianne, C. M. & Bull, J. J.
42. Moriyama, E. N. & Hartl, D. L. (1993) Genetics 134, 847–858.
(1999) Science 285, 422–424.
43. Petrov, D. A. & Hartl, D. L. (1999) Proc. Natl. Acad. Sci. USA 96, 1475–1479.
55. Mukai, T., Chigusa, S. I., Mettler, L. E. & Crow, J. F. (1972) Genetics 72,
44. Marais, G. & Duret, L. (2001) J. Mol. Evol. 52, 275–280.
45. Comeron, J. M., Kreitman, M. & Aguade, M. (1999) Genetics 151, 239–249.
56. Keightley, P. D. & Eyre-Walker, A. (1999) Genetics 153, 515–523.
46. Schmidt, P. S., Duvernell, D. D. & Eanes, W. F. (2000) Proc. Natl. Acad. Sci. 57. Lynch, M., Blanchard, J., Houle, D., Kibota, T., Schultz, S., Vassilieva, L. & USA 97, 10861–10865.
Willis, J. (1999) Evolution (Lawrence, Kans.) 53, 645–663.
47. Schmid, K. J. & Aquadro, C. F. (2001) Genetics 159, 589–598.
58. Charlesworth, B. (1996) Genet. Res. 68, 131–150.
48. Akashi, H. (1994) Genetics 136, 927–935.
59. Gillespie, J. H. (1991) The Causes of Molecular Evolution (Oxford Univ. Press, 49. Comeron, J. M. & Kreitman, M. (1998) Genetics 150, 767–775.
13620 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.212277199

Source: ftp://ftp.bork.embl-heidelberg.de/pub/users/lercher/Ka/Betancourt_Presgraves2002PNAS.pdf