- Research article
- Open Access
Similar rates of protein adaptation in Drosophila miranda and D. melanogaster, two species with different current effective population sizes
© Bachtrog; licensee BioMed Central Ltd. 2008
- Received: 09 April 2008
- Accepted: 18 December 2008
- Published: 18 December 2008
Adaptive protein evolution is common in several Drosophila species investigated. Some studies point to very weak selection operating on amino-acid mutations, with average selection intensities on the order of Nes ~ 5 in D. melanogaster and D. simulans. Species with lower effective population sizes should undergo less adaptation since they generate fewer mutations and selection is ineffective on a greater proportion of beneficial mutations.
Here I study patterns of polymorphism and divergence at 91 X-linked loci in D. miranda, a species with a roughly 5-fold smaller effective population size than D. melanogaster. Surprisingly, I find a similar fraction of amino-acid mutations being driven to fixation by positive selection in D. miranda and D. melanogaster. Genes with higher rates of amino-acid evolution show lower levels of neutral diversity, a pattern predicted by recurrent adaptive protein evolution. I fit a hitchhiking model to patterns of polymorphism in D. miranda and D. melanogaster and estimate an order of magnitude higher selection coefficients for beneficial mutations in D. miranda.
This analysis suggests that effective population size may not be a major determinant in rates of protein adaptation. Instead, adaptation may not be mutation-limited, or the distribution of fitness effects for beneficial mutations might differ vastly between different species or populations. Alternative explanation such as biases in estimating the fraction of beneficial mutations or slightly deleterious mutation models are also discussed.
- Effective Population Size
- Deleterious Mutation
- Adaptive Evolution
- Selective Sweep
- Synonymous Site
Researchers have made considerable progress in recent years to quantify rates of adaptive evolution in the genome using population variability data [1–6]. Many studies aimed at detecting adaptive evolution have applied the McDonald-Kreitman (MK) test  or modifications of it, which contrasts the number of polymorphisms within a species to the number of substitutions between species at two classes of sites, a putatively neutral and a putatively selected class. In protein-coding sequences these classes are usually synonymous and replacement sites .
Several members in the Drosophila melanogaster species group show high rates of adaptive amino-acid evolution. Using the MK test and its extensions, about half (and up to 95%) of all amino-acid mutations fixed between species are inferred to be driven by positive selection [1–4]. Some uncertainty in estimates of α, the fraction of amino-acid substitutions driven to fixation by adaptive evolution in D. melanogaster and D. simulans exists among different studies, reflecting in part the choice of loci and their chromosomal location and the populations surveyed, as well as the specific methodology employed to infer adaptive evolution (i.e. see ref.  for a discussion). However, estimates of α at X-linked loci in African, presumably ancestral populations of D. melanogaster and D. simulans are very similar between the two species. In particular, roughly 60% of amino-acid substitutions at X-linked loci are inferred to be adaptive in both a Zimbabwe population of D. melanogaster and a Madagascar population of D. simulans [3, 9]. Both populations have similar levels of X-linked synonymous diversity, πs (about 3% [3, 9]), indicating that they have similar effective population sizes. Comparable estimates of pervasive positive selection have been obtained in D. americana, a member of the virilis species group  which also has similar levels of synonymous diversity (around 2% ) and thus probably a similar effective population size (Ne).
While adaptive protein evolution appears to be common in Drosophila, we know much less about the strength of selection (s) acting on beneficial amino-acid changes. Different studies and approaches have yielded very different estimates of s for amino acid mutations. In Drosophila, there is a correlation between rates of recombination and levels of nucleotide diversity , but the causes of this correlation are controversial [12, 13]. Assuming that this correlation is entirely driven by beneficial mutations, Eyre-Walker  estimates the strength of selection of fixed mutations in D. melanogaster to be 350 <Nes < 3500. Macpherson et al.  also concluded that selection is strong (s ~ 1%; i.e. Nes ~ 104-105), by fitting a genetic hitchhiking model to patterns of genome variability in D. simulans. Similarly, Li and Stephan  used a likelihood approach to estimate selection parameters from patterns of single nucleotide polymorphisms in D. melanogaster, and infer selection to be strong (s between 0.05–0.5%). In contrast, Andolfatto  inferred much weaker selection by fitting genome variability data in D. melanogaster to a recurrent sweep model, and estimates Nes ~ 40. Finally, by fitting MK-type data from D. simulans to a weak selection model, Sawyer et al.  estimate that the average scaled selection intensity of fixed mutations is only Nes ~ 5. Thus, there is little agreement among studies on the average strength of selection of fixed mutations in Drosophila. However, if selection would indeed be on the order of Nes ~ 5 , even modest changes in the effective population size among Drosophila species would have dramatic impacts on rates of adaptation between different lineages. That is because mutations are effectively neutral if Nes < 1, i.e. their fate is mostly governed by genetic drift and not selection .
Although the data are limited, there does appear to be a possible correlation between the level of adaptive evolution and population size: hominids appear to have undergone very little adaptive evolution [5, 19], compared with Drosophila, while bacteria and viruses seem to show even higher rates of adaptive divergence [20, 21]. This is to be expected given that large populations generate more mutations and selection is effective on a greater proportion of mutations . This might mean that species with small population sizes are much less able to adapt to their environment.
There are, however, some problems with the very broad-brush nature of these patterns. The most fundamental difficulty is that many different aspects of the biology of these very different species compared are confounded. A more direct approach to estimate the influence of Ne on rates of adaptation is to compare closely related species that differ in their effective population size, such as different members from the genus Drosophila.
Here, I report and analyze data from 91 X-linked protein-coding genes from D. miranda, to estimate rates of adaptation at the protein level. Synonymous site diversity in this species is substantially lower than in members of the D. melanogaster species group, only about 0.4% [22–24], suggesting a 5-fold lower current effective population size for D. miranda. Thus, if the average selection intensity for beneficial mutations in D. simulans is indeed in the order of Nes ~ 5, as suggested by some studies , and the distribution of s is similar between species, a large fraction of the beneficial mutations fixed in D. simulans would behave effectively neutral in D. miranda. Previous studies based on many fewer genes scattered over different chromosomes have indeed found little evidence for positive selection operating on amino-acid mutations in D. miranda [23, 25]. Applying MK-tests to this much larger data set, I find evidence for high rates of adaptive protein evolution, similar to those reported in the D. melanogaster species group. In addition, I show that genes with high rates of protein evolution harbor lower levels of synonymous site diversity, a signature of hitchhiking effects associated with linked beneficial amino-acid substitutions. These findings are discussed in comparisons with inferences drawn from D. melanogaster, in light of the effect of population size on rates of adaptation.
Levels of polymorphism in D. miranda
Summary statistics for 91 X-linked protein-coding genes in a sample of 14 alleles of D. miranda.
Dxy pse (%)
Dxy aff (%)
Dxy ANC (%)
Here, I use synonymous sites as a neutral marker in MK tests, and to detect hitchhiking effects associated with adaptive protein evolution. In Drosophila, synonymous codons are not used randomly, but instead some codons are used preferentially over a different codon encoding for the same amino acid [28, 29]. Thus, synonymous sites are not necessarily free of selective constraints. Nevertheless, selection for codon bias is strongly reduced in D. miranda, and synonymous sites are evolving probably close to neutral in this species [25, 30]. Thus, synonymous polymorphisms are a suitable "almost neutral" marker for tracking adaptive events in D. miranda.
Adaptive protein evolution in D. miranda
Count of synonymous and replacement polymorphism in D. miranda and divergence to D. pseudoobscura.
f > 0.1
Proportion of amino-acid substitutions driven by positive selection (α)
α (95% CI)
α (95% CI)
α (95% CI)
α (95% CI)
Fay et al.
0.41 (0.19, 0.56)
0.61 (0.45, 0.74)
0.39 (-0.08, 0.69)
0.59 (0.25, 0.78)
Smith & Eyre-Walker
0.12 (-0.22, 0.38)
0.48 (0.28, 0.66)
0.32 (-0.17, 0.63)
0.54 (0.20, 0.75)
Bierne & Eyre-Walker
0.17 (-0.11, 0.36)
0.44 (0.19, 0.61)
0.41 (0.06, 0.61)
0.61 (0.28, 0.80)
Average synonymous divergence between D. miranda and D. pseudoobscura is low, roughly only 4% (Table 1). It has been suggested that low levels of divergence can lead to upwardly biased estimates of α for two reasons [8, 31]. First, advantageous mutations spread through a population much more rapidly than neutral mutations. Thus, the divergence time may have been to short for a stochastic steady state for fixations to have been reached, resulting in an upward bias in the estimation of α . The divergence time between D. miranda and D. pseudoobscura is about 7-fold larger than the population coalescence time, very similar to that observed for D. melanogaster and D. simulans [9, 34]. Thus, if low divergence leads to a bias in the estimation of α, a similar bias may be expected in the two species comparisons. Secondly, since only a single D. pseudoobscura sequence is used for outgroup comparison, some of the inferred divergence includes undetected segregating polymorphism, which could also result in an upward bias in the estimation of α . To avoid potential biases in inferring the fraction of beneficial amino-acid fixations due to low levels of divergence, I also estimated α using divergence to a more distantly related species. To this end, D. affinis was employed, a species which shows about 25% synonymous divergence to D. miranda at synonymous sites (see Table 1). Interestingly, using D. affinis as an outgroup species results in similar, although slightly lower, estimates of α (i.e. 30%–40%, see Table 3). If α is estimated along the D. miranda lineage, using a reconstructed D. miranda – D. pseudoobscura ancestral sequence, estimates of α are increased (Table 3). Thus, while α may be overestimated for short divergence times, D. pseudoobscura appears sufficiently diverged from D. miranda to result in little bias in estimating α due to low levels of sequence divergence.
Hitchhiking effects at fast evolving genes
Positively selected amino-acid mutations should leave characteristic signatures at surrounding genomic regions. One such signature is reduced neutral diversity surrounding sites that are targets of positive selection; i.e. selective sweeps [35–38]. Indeed, several recent studies have found reduced levels of synonymous diversity at genes with high rates of protein evolution in D. melanogaster [17, 39] and D. simulans .
Estimating the strength of selection (s) under a recurrent hitchhiking model
Parameters inferred for positive amino-acid mutations in D. miranda and D. melanogaster
T = (Ks-πs)/(2μ)
Ne = θ0/(3μ)
λ = (α *Ka)/(2T)
uben = λ/(4*Nes)
In the recurrent hitchhiking model used, s and α (or λ) are conflated parameters and only their product can be estimated. However, I can infer s using an independent estimate of α. I estimate α = 44% for D. miranda and α = 51% for D. melanogaster, using a maximum likelihood method . This value of α implies that s ~ 2.7 × 10-3for D. miranda and s ~ 2.3 × 10-5 for D. melanogaster. Given estimates of λ, Ne and s, we can also estimate the rate at which beneficial mutations arise per site and generation (uben = λ/(4Ne s), see Methods). The beneficial mutation rate is uben ~ 6.5 × 10-14 in D. miranda and uben ~ 4.7 × 10-12 in D. melanogaster. Thus, while the inferred rate of adaptation (i.e. the fixation rate of beneficial mutations) is similar between D. miranda and D. melanogaster (λ ~ 4.0 × 10-10 in D. miranda and λ ~ 7.0 × 10-10 in D. melanogaster, corresponding to λ 2Nes ~ 1.2 × 10-6 and λ 2Nes ~ 5.1 × 10-8, Table 4), I estimate a lower beneficial mutation rate in D. miranda but larger average effects of beneficial amino-acid mutations.
High rates of protein adaptation in D. miranda
If the rate of beneficial mutations and their selective effects are constant across species, larger populations are expected to show higher rates of adaptation, both because they generate more mutations and selection is effective on a greater proportion of mutations [18, 46]. Kimura and Ohta  noticed that rates of protein evolution are rather similar across taxa with different population sizes, and argued against models of adaptive substitutions to explain protein evolution. Here, I compare rates of protein evolution in two Drosophila species that differ about 5-fold in their levels of neutral variability, but find little difference in rates of adaptive protein divergence between them.
Specifically, two independent types of evidence suggest that D. melanogaster and D. miranda both show high rates of adaptive protein evolution. First, application of the MK test suggest that a similar fraction of amino-acid mutations is driven to fixation in both species groups, and my estimate of λ, the rate of selective sweeps, is very similar between the two species. Second, both species show a correlation between rates of protein evolution and synonymous site diversity, as expected under recurrent adaptive protein evolution [17, 39]. Fitting a recurrent hitchhiking model to polymorphism data suggests a higher value of αs in D. miranda than in D. melanogaster. Since α is estimated to be very similar between species, s is inferred to be larger in D. miranda.
What would we expect α, the rate of adaptive protein divergence, to be in D. miranda using selection parameters inferred from D. melanogaster but accounting for differences in their synonymous site diversity? The rate of adaptive divergence Kben can be calculated as Kben = 2N*uben *2s*2T. Using estimates of uben and s from D. melanogaster (Table 4), we would expect the rate of adaptive amino-acid divergence in D. miranda to be 0.107% per site. This means that only 18% of total amino acid divergence was driven by positive selection, i.e. α = 0.18. This value of α lies below the confidence limits we estimate for α in D. miranda, using MK approaches (Table 3). Thus, assuming that selection parameters are similar between the species, D. miranda displays higher rates of protein evolution than would be expected given its level of diversity. Below, I discuss several implications of this finding and possible explanations.
Long-term vs. short-term differences in the effective population size
Levels of synonymous diversity are lower in D. miranda relative to D. melanogaster (see Table 4), suggesting a smaller effective population size in D. miranda. However, levels of diversity only contain information about the recent effective population size of a species, and it is possible that both species had similarly sized populations for much of their evolutionary history. Indeed, there is some evidence that D. miranda has experienced a reduction in its effective population size given reduced levels of diversity compared to its closest relatives, reduced selection to maintain codon bias, or mulitlocus patterns of diversity [22, 23, 30]. Thus, much of the adaptive evolution detected in D. miranda could reflect selection in a larger, ancestral population, or selection in D. pseudoobscura (which based on levels of synonymous polymorphism, is thought to have a larger population size ) since divergence also includes fixations along the D. pseudoobscura lineage. However, similar estimates of α are inferred, regardless of whether D. pseudoobscura or D. affinis is used as the outgroup species, or whether the reconstructed ancestral sequence of D. miranda and D. pseudoobscura is used (Table 3). While estimates of α might be biased upwards if divergence between species is low [8, 31], a large fraction of amino-acid mutations along the D. miranda are clearly driven by adaptive evolution. Furthermore, if D. miranda has experienced a recent reduction in its effective population size, the estimate of α is likely to be biased downwards, due to segregating deleterious mutations (see below). Similar arguments of a reduced population size in D. melanogaster compared to its close relative to D. simulans have been invoked to account for lower levels of diversity at autosomal regions, decreased selection for codon bias and patterns of polymorphisms in D. melanogaster [48–50]. However, X-linked diversity appears similar between the two species [9, 34], and estimates of adaptive protein evolution are comparable [8, 9, 31]. Thus, the exact dynamics of the long-term effective population size is unclear in both species, but both may have undergone a recent contraction in size.
While inferences of adaptive evolution based on the MK test are sensitive to assumptions about the long-term effective population size of a species, the correlation between synonymous diversity and rate of protein evolution provides an independent conformation of high rates of adaptive protein evolution in both species. This observation is less affected by fluctuations in the long-term effective population size, since reduced diversity in fast evolving proteins reflects the action of very recent selection (i.e. in the order of the population coalescence time). Interestingly, the rate of selective sweeps inferred from the observed correlation between πs and Ka is very similar for the two species. Thus, despite uncertainties about the long-term effective population size in both D. melanogaster and D. miranda, several independent lines of evidence suggest similar and high rates of protein evolution in both species. It will be of great interest to study a diverse set of species that differ in their inferred effective population sizes, to better study the influence of population size on rates of adaptive evolution.
Different distribution of fitness effects for beneficial mutations among species?
The expectation of less adaptation in smaller populations relies on the assumption that the mutation rate and the strength of selection for beneficial mutations are constant among species investigated. However, I infer that both the beneficial mutation rate and the strength of selection are different between D. melanogaster and D. miranda. But how plausible is such a scenario? In a species with a larger effective population size, a smaller fraction of mutations is effectively neutral, and one might expect such a species to be better adapted since more advantageous mutations will fix . This could change the distribution of fitness effects because, as a species adapts, its fitness is expected to move closer to an optimum [51, 52]. Thus, one could intuit that better adapted species have fitness distributions where beneficial mutations are weaker on average, but it may be harder to explain why the rate of adaptive mutations should be higher in that case.
There is some experimental evidence that the distribution of fitness effects is different between species that have different effective population sizes. Silander et al.  performed a mutation-accumulation experiment where bacteriophage populations were passaged through different population sizes. As expected, the small population size lines had lower fitness because they had accumulated more deleterious mutations. Surprisingly, about 15% of the mutations in the small population size lines were adaptive, in contrast to almost none in the large population size lines. The mean effect of mutations, however, did not seem to differ between lines. In comparisons between D. miranda and D. melanogaster, however, I infer that mutation rates are actually lower in the species with the lower population size, while fitness effects for beneficial mutations appear much larger.
Bias in the estimation procedure to infer α?
While this model could account for the observed differences in rates and effects of beneficial amino-acid mutations, it would also predict that D. melanogaster shows much higher rates of adaptive protein evolution than D. miranda. However, my estimate of α is very similar between the two species. This discrepancy could potentially be explained by a bias in estimating α from MK tables, which assumes that all amino-acid polymorphism segregating in the population are neutral . However, if the fraction of beneficial amino-acid mutations is high and selection is weak, a substantial proportion of the amino-acid polymorphism segregating in the population might actually represent beneficial mutations. For example, Sawyer et al.  estimate that about half of all amino acids segregating in a population sample from D. simulans are beneficial. Including these mutations into MK tables will downward bias our estimate of α. Thus, if only very few amino-acid mutations are neutral in D. melanogaster (i.e. the majority is either deleterious or slightly beneficial), then a large fraction of amino-acid mutations that segregate in the population will be beneficial (once removing the low-frequency class), resulting in an underestimate of α using the MK framework employed here. In contrast, a larger fraction of amino-acid mutations will be effectively neutral in D. miranda, due to its smaller current population size, and thus a much smaller proportion of the observed amino acid polymorphisms are beneficial. This will cause the estimate of α to be much closer to its true value in D. miranda, while α might be substantially underestimated in D. melanogaster.
Consistent with a substantial fraction of amino-acid mutations segregating in D. melanogaster being beneficial, amino-acid polymorphism is less reduced relative to synonymous sites in D. melanogaster compared to D. miranda (πa/πs = 0.08 in D. melanogaster vs. πa/πs = 0.07 in D. miranda). Purifying selection against deleterious amino-acid mutations would be more effective in a larger species, predicting the opposite patterns, while the πa/πs ratio should be similar among species if amino-acid polymorphisms are neutral. In fact, if half the amino-acid polymorphisms are removed from D. melanogaster (i.e. they are assumed to be beneficial instead of neutral), the maximum likelihood estimate of α becomes closer to 0.8 in that species. In this case, we would conclude that D. miranda shows reduced rates of adaptation, in accordance with expectations based on its smaller current effective population size. While this model could qualitatively account for the observed patterns with regards to population size and adaptive evolution, it remains to be seen whether it can also quantitatively do so.
The influence of deleterious mutations
In the above considerations I assume that both the excess divergence at amino acid mutations relative to polymorphisms as well as the correlation between neutral diversity and protein evolution are entirely driven by beneficial mutations. However, for both interpretations, weakly deleterious mutation models pose a potential problem.
Slightly deleterious nonsynonymous mutations can yield biased estimates of α using MK methods, and the direction of this bias depends on the demography of the population. As mentioned before, if the population size has been relatively stable, the estimate of α is likely an underestimate, because slightly deleterious amino-acid mutations tend to contribute relatively more to polymorphism than they do to divergence when compared with neutral mutations. However, slightly deleterious mutations can lead to an overestimate of α if population sizes have expanded, because mutations that might have been fixed in the past, when the population size was small, no longer segregate as polymorphisms. Even fairly modest increases in population size can create artifactual evidence of adaptive evolution . Thus, it is possible that D. miranda has undergone a recent population size expansion (i.e. its current Ne is larger than the ancestral one) and that α is therefore overestimated. However, as mentioned above, if anything D. miranda appears to have gone through a population size decrease based on several consistent patterns of polymorphisms and divergence [22, 23, 30]. This suggests that α has not been overestimated in D. miranda because of a population size increase.
I invoke recurrent hitchhiking to explain the observed correlation between Ka and πs in D. miranda; recurrent beneficial amino acid mutations reduce variation more frequently at faster evolving genes, causing a reduction in codon bias and an excess of low frequency variants. It is, however, formally also possible to explain the observed correlation between Ka and πs solely by deleterious mutations models. The removal of recurrent deleterious mutations from the population reduces the local effective population size, causing regions with higher deleterious mutation rates (or lower recombination rates) to harbour reduced levels of neutral diversity; i.e. background selection [54, 55]. A reduction in Ne, in turn, predicts an acceleration in the rate of accumulation of very weakly deleterious mutations [32, 56], including possible unpreferred codons or slightly deleterious amino acid mutations. Weak background selection can also distort the allele-frequency spectrum of neutral mutations towards rare variants [56, 57]. Thus, selection against weakly deleterious mutations can by itself account for reduced levels of neutral diversity associated with increased rates of accumulation of slightly deleterious amino-acid mutations [32, 56], without invoking any beneficial mutations.
In D. miranda, the correlation between Ka and πs is stronger than in D. melanogaster. This would suggest a larger fraction of slightly deleterious amino-acid mutations in the roam of weak background selection at some genes in D. miranda, causing a stronger reduction in Ne and πs and higher rates of accumulation of deleterious amino-acid mutations. Again, a distribution of fitness effects against deleterious amino-acid mutations may in principle explain the pattern, if a larger fraction of deleterious amino-acid mutations in D. melanogaster is selected against efficiently. That is, only few amino-acid mutations with a relatively small range of negative selection coefficients cause weak background selection in D. melanogaster, while the majority of deleterious amino-acid mutations are selected against effectively. In contrast, more slightly deleterious amino-acid mutations segregate in D. miranda, resulting in a larger variance in the amount of weak background selection among genes. It is also possible that there is more heterogeneity in rates of recombination among the loci sampled in D. miranda, since physical and genetic map positions are not known for the loci investigated. Recombination rate variation may result in heterogeneous levels of background selection along a chromosome, and thus cause the local Ne (and the efficacy of selection) to vary among regions. Indeed, there is evidence for recombination rate heterogeneity in D. pseudoobscura, where mild "recombination hotspots" were discovered . Unfortunately, no genetic map exists in D. miranda, preventing us from directly investigating the effect of recombination rate heterogeneity on levels of polymorphisms in this species. However, as Andolfatto  points out, there are several empirical arguments against weak background selection causing the observed correlation between Ka and πs, such as no general elevation of Ka in regions of the Drosophila genome with reduced recombination rates [42, 59], and no effect of gene density on rates of protein evolution .
More complicated models of selection
In a series of papers, Gillespie examines the role of population size in population genetical models of molecular evolution, using extensive computer simulations [60, 61]. Gillespie has defined three domains based on their rates of substitutions, which he terms the Ohta domain (the rate of substitution decreases with increasing population size), the Kimura domain (the rate of substitution remains close to the mutation rate) and the Darwin domain (the rate of substitution increases with increasing population size). Not surprisingly, he finds that nearly neutral deleterious mutation models all fall within the Ohta domain where the rate of evolution is inversely proportional to population size. He also finds that the normal-shift model (a model of positive selection) appropriately falls within the Darwin domain, but that the rate of substitution does not linearly increase with populations size and is substantially reduced relative to the expectation under the independence among sites model. Surprisingly, Gillespie also finds that the fluctuating selection, neutral, and overdominance model all lead to the Kimura domain, where the rate of molecular evolution is independent of the population size. In addition, adaptive evolution can actually cause the rate of substitution of deleterious alleles at a linked locus to increase with increasing population size [60, 62]. Thus, the associations between Ne and rates of evolution might be complex, and intuitions drawn from very simple models of positive and negative selection might not apply to natural populations. Future work will be needed to assess how more complicated selection models, including selection from standing variation or mutation selection balance [63, 64], and interactions between beneficial and deleterious mutations will influence rates of molecular evolution and variation.
Adaptive protein evolution is common in several Drosophila species investigated, but little is known about underlying selection coefficients of beneficial amino acid mutations. If average selection intensities are very weak, as suggested by some studies, even modest changes in the effective population size between species may have drastic impacts on rates of adaptation between lineages. Here, I estimate that a similar fraction (~50%) of amino-acid mutations is being driven to fixation by positive selection in two species of Drosophila that differ roughly 5-fold in their effective population sizes. Genes with higher rates of amino-acid evolution show lower levels of neutral diversity in both species, a pattern predicted by recurrent adaptive protein evolution. Thus, while adaptive amino-acid evolution is common in the genus Drosophila, modest changes in population size appear to have little influence on protein adaptation.
Survey of coding regions
A total of 91 X-linked coding regions were surveyed in this study with a sample size of 14 D. miranda alleles. The following strains of D. miranda were used in the analysis, with their population origin given in parenthesis: 0101.3, 0101.4, 0101.5, 0101.7 (Port Coquitlam, BC, Canada); 0101.9, MA28, MA32, MA03.1, MA03.2, MA03.3, MA03.4, MA03.5, MA03.6 (Mather, CA, USA); SP138, SP235, SP295 (Spray, OR, USA); MSH22, MSH38 (Mt. St. Helena, CA, USA).
The D. pseudoobscura genome sequence was used to provide estimates of divergence . For a subset of 51 loci, the orthologous D. affinis allele was surveyed (strain 14012-0141.01 or 14012-0141.02). All genes were selected randomly with respect to gene function. Information about the specific loci surveyed and primers used can be found in Additional File 1.
Each ~1200 base pair region was PCR-amplified from genomic DNA extracted from single male flies, and primers and nucleotides were removed using Exonuclease I and Shrimp Alkaline Phosphatase. Cleaned products were sequenced on both strands with the original PCR primers and internal sequencing primers, using Big-Dye (Version 3, Applied Biosystems) and run on an ABI 3730 capillary sequencer. Sequence traces were edited using Sequencher (Gene Codes) software and multiple sequence alignments were generated using MUSCLE http://www.drive5.com/muscle/ with protein-alignment-assisted adjustments to preserve reading frames. Sequences have been deposited in Genbank under accession numbers (XXXX-XXXX).
For comparison with D. melanogaster, I used a published dataset of 12 individuals sequenced from a Zimbabwe, Africa population and their divergence to D. simulans .
Polymorphism and divergence analysis
The estimated number of synonymous sites, nonsynonymous sites, average pairwise diversity (π), average pairwise divergence to D. pseudoobscura (D xy ), as well as counts of the number of polymorphisms (S) and the summary of the frequency distribution of polymorphism frequencies, Tajima's D (Tajima 1989), were calculated using a library of Perl scripts ("Polymorphorama") written by the author and P. Andolfatto. The number of nonsynonymous and synonymous sites were estimated using the method of . Average pairwise diversity (π) and divergence (D xy ) estimates were corrected for multiple hits using a Jukes-Cantor correction . Multiply hit sites were included in all analyses but insertion-deletion polymorphisms and polymorphic sites overlapping alignment gaps were excluded. For lineage-specific estimates of divergence, I reconstructed a D. miranda – D. pseudoobscura ancestor sequence (ANC) by maximum likelihood, using D. affinis as an outgroup, as implemented in the codeml program of PAML. Locus-specific estimates of levels of diversity and divergence can be found in Additional File 1. To estimate codon bias selection at each gene, the frequency of optimal codons (as inferred from D. pseudoobscura, see ) is used.
The fraction of amino acid mutations driven by positive selection, α, is estimated using three slightly different approaches based on the MK test , as implemented in the DoFEv2 software package (kindly provided by A. Eyre-Walker). The number of divergent sites were corrected for multiple hits using a Jukes-Cantor correction. I also exclude singleton mutations, to minimize the downward bias in inferring α due to segregating deleterious amino acid mutations.
Estimating recurrent hitchhiking parameters
and equation 2 allows levels of neutral diversity to be directly related to the extent of adaptive amino acid divergence at each locus (i.e. α Ka).
Average recombination rates in D. pseudoobscura are higher than those observed in D. melanogaster; average recombination rates across the X chromosome in D. melanogaster equal 3.3 cM/Mb , while the average recombination rate across the X chromosome in D. pseudoobscura is 8.3 cM/Mb . Estimating and Ne = θ/3μ require an estimate of the neutral mutation rate, μ. I assume that = 5.8 × 10-9 per generation, the estimated average rate for single nucleotide mutations from D. melanogaster mutation-accumulation lines . With these parameter estimated and πs and Ka for 92 loci, a least-squares method is used to find the values of θ and αs that minimize the sum of the squared deviations between observed πs and E(π) predicted by the model, using the R statistical package.
I am grateful to Kevin Thornton and Peter Andolfatto for valuable discussions, and to Kevin Thornton and Stephen Wright for comments on the manuscript. I thank Emily Hom for technical assistance. This research is funded by NIH Grant GM076007 to D.B.
- Smith NG, Eyre-Walker A: Adaptive protein evolution in Drosophila. Nature. 2002, 415 (6875): 1022-1024. 10.1038/4151022a.View ArticlePubMedGoogle Scholar
- Fay JC, Wyckoff GJ, Wu CI: Testing the neutral theory of molecular evolution with genomic data from Drosophila. Nature. 2002, 415 (6875): 1024-1026. 10.1038/4151024a.View ArticlePubMedGoogle Scholar
- Andolfatto P: Adaptive evolution of non-coding DNA in Drosophila. Nature. 2005Google Scholar
- Sawyer S, Kulathinal R, Bustamante C, Hartl D: Bayesian analysis suggests that most amino acid replacements in Drosophila are driven by positive selection. J Mol Evol. 2003, 57 (Suppl 1): S154-164. 10.1007/s00239-003-0022-3.View ArticlePubMedGoogle Scholar
- Bustamante C, Fledel-Alon A, Williamson S, Nielsen R, Hubisz M, Glanowski S, Tanenbaum D, White T, Sninsky J, Hernandez R, et al: Natural selection on protein-coding genes in the human genome. Nature. 2005, 437 (7062): 1153-1157. 10.1038/nature04240.View ArticlePubMedGoogle Scholar
- Bustamante C, Nielsen R, Sawyer S, Olsen K, Purugganan M, Hartl D: The cost of inbreeding in Arabidopsis. Nature. 2002, 416 (6880): 531-534. 10.1038/416531a.View ArticlePubMedGoogle Scholar
- McDonald JH, Kreitman M: Adaptive protein evolution at the Adh locus in Drosophila. Nature. 1991, 351 (6328): 652-654. 10.1038/351652a0.View ArticlePubMedGoogle Scholar
- Welch J: Estimating the genomewide rate of adaptive protein evolution in Drosophila. Genetics. 2006, 173 (2): 821-837. 10.1534/genetics.106.056911.PubMed CentralView ArticlePubMedGoogle Scholar
- Haddrill P, Bachtrog D, Andolfatto P: Positive and negative selection on noncoding DNA in Drosophila simulans. Mol Biol Evol. 2008.Google Scholar
- Maside X, Charlesworth B: Patterns of molecular variation and evolution in Drosophila americana and its relatives. Genetics. 2007, 176 (4): 2293-2305. 10.1534/genetics.107.071191.PubMed CentralView ArticlePubMedGoogle Scholar
- Begun DJ, Aquadro CF: Levels of naturally occurring DNA polymorphism correlate with recombination rates in Drosophila melanogaster. Nature. 1992, 356 (6369): 519-520. 10.1038/356519a0.View ArticlePubMedGoogle Scholar
- Charlesworth B: Background selection and patterns of genetic diversity in Drosophila melanogaster. Genet Res. 1996, 68 (2): 131-149.View ArticlePubMedGoogle Scholar
- Wiehe TH, Stephan W: Analysis of a genetic hitchhiking model, and its application to DNA polymorphism data from Drosophila melanogaster. Mol Biol Evol. 1993, 10 (4): 842-854.PubMedGoogle Scholar
- Eyre-Walker A: The genomic rate of adaptive evolution. Trends Ecol Evol. 2006Google Scholar
- Macpherson J, Sella G, Davis J, Petrov D: On the genomewide spatial correspondence between adaptation and neutral diversity in Drosophila. Genetics. 2007.Google Scholar
- Li H, Stephan W: Inferring the demographic history and rate of adaptive substitution in Drosophila. PLoS Genet. 2006, 2 (10): e166-10.1371/journal.pgen.0020166.PubMed CentralView ArticlePubMedGoogle Scholar
- Andolfatto P: Hitchhiking effects of recurrent beneficial amino acid substitutions in the Drosophila melangoaster genome. Genome Res. 2007.Google Scholar
- Kimura M: The Neutral Theory of Molecular Evolution. 1983, Cambridge: Cambridge University PressView ArticleGoogle Scholar
- Nielsen R, Bustamante C, Clark A, Glanowski S, Sackton T, Hubisz M, Fledel-Alon A, Tanenbaum D, Civello D, White T, et al: A scan for positively selected genes in the genomes of humans and chimpanzees. PLoS Biol. 2005, 3 (6): e170-10.1371/journal.pbio.0030170.PubMed CentralView ArticlePubMedGoogle Scholar
- Charlesworth J, Eyre-Walker A: The rate of adaptive evolution in enteric bacteria. Mol Biol Evol. 2006, 23 (7): 1348-1356. 10.1093/molbev/msk025.View ArticlePubMedGoogle Scholar
- Nielsen R, Yang Z: Estimating the distribution of selection coefficients from phylogenetic data with applications to mitochondrial and viral DNA. Mol Biol Evol. 2003, 20 (8): 1231-1239. 10.1093/molbev/msg147.View ArticlePubMedGoogle Scholar
- Yi S, Bachtrog D, Charlesworth B: A survey of chromosomal and nucleotide sequence variation in Drosophila miranda. Genetics. 2003, 164 (4): 1369-1381.PubMed CentralPubMedGoogle Scholar
- Bachtrog D, Andolfatto P: Selection, recombination and demographic history in Drosophila miranda. Genetics. 2006, 174 (4): 2045-2059. 10.1534/genetics.106.062760.PubMed CentralView ArticlePubMedGoogle Scholar
- Bartolome C, Maside X, Yi S, Grant A, Charlesworth B: Patterns of Selection on Synonymous and Nonsynonymous Variants in Drosophila miranda. Genetics. 2005, 169 (3): 1495-1507. 10.1534/genetics.104.033068.PubMed CentralView ArticlePubMedGoogle Scholar
- Bartolomé C, Charlesworth B: Evolution of amino-acid sequences and codon usage on the Drosophila miranda neo-sex chromosomes. Genetics. 2006, 174 (4): 2033-2044. 10.1534/genetics.106.064113.PubMed CentralView ArticlePubMedGoogle Scholar
- Tajima F: Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics. 1989, 123 (3): 585-595.PubMed CentralPubMedGoogle Scholar
- Li W: Molecular evolution. 1997, Sunderland Massachusetts: Sinauer Associates, Inc., PublishersGoogle Scholar
- Akashi H, Schaeffer SW: Natural selection and the frequency distributions of "silent" DNA polymorphism in Drosophila. Genetics. 1997, 146 (1): 295-307.PubMed CentralPubMedGoogle Scholar
- Bulmer M: The selection-mutation-drift theory of synonymous codon usage. Genetics. 1991, 129 (3): 897-907.PubMed CentralPubMedGoogle Scholar
- Bachtrog D: Reduced selection for codon usage bias in Drosophila miranda. J Mol Evol. 2007, 64 (5): 586-590. 10.1007/s00239-006-0257-x.View ArticlePubMedGoogle Scholar
- Bierne N, Eyre-Walker A: The genomic rate of adaptive amino acid substitution in Drosophila. Mol Biol Evol. 2004, 21 (7): 1350-1360. 10.1093/molbev/msh134.View ArticlePubMedGoogle Scholar
- Charlesworth B: The effect of background selection against deleterious mutations on weakly selected, linked variants. Genet Res. 1994, 63 (3): 213-227.View ArticlePubMedGoogle Scholar
- Eyre-Walker A: Changing Effective Population Size and the McDonald-Kreitman Test. Genetics. 2002, 162 (4): 2017-2024.PubMed CentralPubMedGoogle Scholar
- Begun D, Holloway A, Stevens K, Hillier L, Poh Y, Hahn M, Nista P, Jones C, Kern A, Dewey C, et al: Population genomics: whole-genome analysis of polymorphism and divergence in Drosophila simulans. PLoS Biol. 2007, 5 (11): e310-10.1371/journal.pbio.0050310.PubMed CentralView ArticlePubMedGoogle Scholar
- Maynard Smith J, Haigh J: The hitch-hiking effect of a favourable gene. Genet Res. 1974, 23: 23-35.View ArticleGoogle Scholar
- Kaplan N, Hudson R, Langley C: The "hitchhiking effect" revisited. Genetics. 1989, 123 (4): 887-899.PubMed CentralPubMedGoogle Scholar
- Kim Y, Stephan W: Detecting a local signature of genetic hitchhiking along a recombining chromosome. Genetics. 2002, 160 (2): 765-777.PubMed CentralPubMedGoogle Scholar
- Przeworski M: The signature of positive selection at randomly chosen loci. Genetics. 2002, 160 (3): 1179-1189.PubMed CentralPubMedGoogle Scholar
- Shapiro J, Huang W, Zhang C, Hubisz M, Lu J, Turissini D, Fang S, Wang H, Hudson R, Nielsen R, et al: Adaptive genic evolution in the Drosophila genomes. Proc Natl Acad Sci USA. 2007, 104 (7): 2271-2276. 10.1073/pnas.0610385104.PubMed CentralView ArticlePubMedGoogle Scholar
- Braverman J, Hudson R, Kaplan N, Langley C, Stephan W: The hitchhiking effect on the site frequency spectrum of DNA polymorphisms. Genetics. 1995, 140 (2): 783-796.PubMed CentralPubMedGoogle Scholar
- Kim Y: Effect of strong directional selection on weakly selected mutations at linked sites: implication for synonymous codon usage. Mol Biol Evol. 2004, 21 (2): 286-294. 10.1093/molbev/msh020.View ArticlePubMedGoogle Scholar
- Betancourt AJ, Presgraves DC: Linkage limits the power of natural selection in Drosophila. Proc Natl Acad Sci USA. 2002, 99 (21): 13616-13620. 10.1073/pnas.212277199.PubMed CentralView ArticlePubMedGoogle Scholar
- Bierne N, Eyre-Walker A: Variation in synonymous codon use and DNA polymorphism within the Drosophila genome. J Evol Biol. 2006, 19 (1): 1-11. 10.1111/j.1420-9101.2005.00996.x.View ArticlePubMedGoogle Scholar
- Marais G, Domazet-Loso T, Tautz D, Charlesworth B: Correlated evolution of synonymous and nonsynonymous sites in Drosophila. J Mol Evol. 2004, 59 (6): 771-779. 10.1007/s00239-004-2671-2.View ArticlePubMedGoogle Scholar
- Stoletzki N, Eyre-Walker A: Synonymous codon usage in Escherichia coli: selection for translational accuracy. Mol Biol Evol. 2007, 24 (2): 374-381. 10.1093/molbev/msl166.View ArticlePubMedGoogle Scholar
- Fisher RA: The Genetical Theory of Natural Selection. 1930, Oxford: Oxford University PressView ArticleGoogle Scholar
- Kimura M, Ota T: Protein polymorphism as a phase of molecular evolution. Nature. 1971, 229 (5285): 467-469. 10.1038/229467a0.View ArticlePubMedGoogle Scholar
- Andolfatto P: Contrasting patterns of X-linked and autosomal nucleotide variation in Drosophila melanogaster and Drosophila simulans. Mol Biol Evol. 2001, 18 (3): 279-290.View ArticlePubMedGoogle Scholar
- Akashi H: Molecular evolution between Drosophila melanogaster and D. simulans: reduced codon bias, faster rates of amino acid substitution, and larger proteins in D. melanogaster. Genetics. 1996, 144 (3): 1297-1307.PubMed CentralPubMedGoogle Scholar
- Akashi H: Inferring weak selection from patterns of polymorphism and divergence at "silent" sites in Drosophila DNA. Genetics. 1995, 139 (2): 1067-1076.PubMed CentralPubMedGoogle Scholar
- Orr HA: Adaptation and the cost of complexity. Evolution. 2000, 54 (1): 13-20. 10.1111/j.0014-3820.2000.tb00002.x.View ArticlePubMedGoogle Scholar
- Orr H: The genetic theory of adaptation: a brief history. Nat Rev Genet. 2005, 6 (2): 119-127. 10.1038/nrg1523.View ArticlePubMedGoogle Scholar
- Silander O, Tenaillon O, Chao L: Understanding the evolutionary fate of finite populations: the dynamics of mutational effects. PLoS Biol. 2007, 5 (4): e94-10.1371/journal.pbio.0050094.PubMed CentralView ArticlePubMedGoogle Scholar
- Charlesworth B, Morgan MT, Charlesworth D: The effect of deleterious mutations on neutral molecular variation. Genetics. 1993, 134 (4): 1289-1303.PubMed CentralPubMedGoogle Scholar
- Hudson RR, Kaplan NL: Deleterious background selection with recombination. Genetics. 1995, 141 (4): 1605-1617.PubMed CentralPubMedGoogle Scholar
- Gordo I, Navarro A, Charlesworth B: Muller's Ratchet and the Pattern of Variation at a Neutral Locus. Genetics. 2002, 161 (2): 835-848.PubMed CentralPubMedGoogle Scholar
- Charlesworth D, Charlesworth B, Morgan MT: The pattern of neutral molecular variation under the background selection model. Genetics. 1995, 141 (4): 1619-1632.PubMed CentralPubMedGoogle Scholar
- Cirulli ET, Kliman RM, Noor MA: Fine-scale crossover rate heterogeneity in Drosophila pseudoobscura. J Mol Evol. 2007, 64 (1): 129-135. 10.1007/s00239-006-0142-7.View ArticlePubMedGoogle Scholar
- Haddrill P, Halligan D, Tomaras D, Charlesworth B: Reduced efficacy of selection in regions of the Drosophila genome that lack crossing over. Genome Biol. 2007, 8 (2): R18-10.1186/gb-2007-8-2-r18.PubMed CentralView ArticlePubMedGoogle Scholar
- Gillespie JH: The role of population size in molecular evolution. Theor Popul Biol. 1999, 55 (2): 145-156. 10.1006/tpbi.1998.1391.View ArticlePubMedGoogle Scholar
- Gillespie JH: Is the population size of a species relevant to its evolution?. Evolution. 2001, 55 (11): 2161-2169.View ArticlePubMedGoogle Scholar
- Bachtrog D, Gordo I: Adaptive evolution of asexual populations under Muller's ratchet. Evolution. 2004, 58 (7): 1403-1413.View ArticlePubMedGoogle Scholar
- Przeworski M, Coop G, Wall J: The signature of positive selection on standing genetic variation. Evolution. 2005, 59 (11): 2312-2323. 10.1554/05-273.1.View ArticlePubMedGoogle Scholar
- Orr HA, Betancourt AJ: Haldane's sieve and adaptation from the standing genetic variation. Genetics. 2001, 157 (2): 875-884.PubMed CentralPubMedGoogle Scholar
- Richards S, Liu Y, Bettencourt B, Hradecky P, Letovsky S, Nielsen R, Thornton K, Hubisz M, Chen R, Meisel R, et al: Comparative genome sequencing of Drosophila pseudoobscura: chromosomal, gene, and cis-element evolution. Genome Res. 2005, 15 (1): 1-18. 10.1101/gr.3059305.PubMed CentralView ArticlePubMedGoogle Scholar
- Nei M, Gojobori T: Simple methods for estimating the numbers of synonymous and nonsynonymous nucleotide substitutions. Mol Biol Evol. 1986, 3 (5): 418-426.PubMedGoogle Scholar
- Jukes TH, Cantor CR: Evolution of protein molecules. Mammalian Protein Metabolism. Edited by: Munro HN. 1969, New York: Academic Press, 21-123.View ArticleGoogle Scholar
- True J, Mercer J, Laurie C: Differences in crossover frequency and distribution among three sibling species of Drosophila. Genetics. 1996, 142 (2): 507-523.PubMed CentralPubMedGoogle Scholar
- Ortiz-Barrientos D, Chang A, Noor M: A recombinational portrait of the Drosophila pseudoobscura genome. Genet Res. 2006, 87 (1): 23-31. 10.1017/S0016672306007932.View ArticlePubMedGoogle Scholar
- Haag-Liautard C, Dorris M, Maside X, Macaskill S, Halligan D, Charlesworth B, Keightley P: Direct estimation of per nucleotide and genomic deleterious mutation rates in Drosophila. Nature. 2007, 445 (7123): 82-85. 10.1038/nature05388.View ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.