Rate Variation of DNA Sequence Evolution in the Drosophila Lineages

Corresponding author: Toshiyuki S. Takano, Department of Population Genetics, National Institute of Genetics, Mishima Shizuoka-ken 411-8540, Japan, totakano{at}lab.nig.ac.jp (E-mail).

Communicating editor: A. G. CLARK

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Rate constancy of DNA sequence evolution was examined for three species of Drosophila, using two samples: the published sequences of eight genes from regions of the normal recombination rates and new data of the four AS-C (ac, sc, l'sc and ase) and ci genes. The AS-C and ci genes were chosen because these genes are located in the regions of very reduced recombination in Drosophila melanogaster and their locations remain unchanged throughout the entire lineages involved, yielding less effect of ancestral polymorphism in the study of rate constancy. The synonymous substitution pattern of the three lineages was found to be erratic in both samples. The dispersion index for replacement substitution was relatively high for the per, G6pd and ac genes. A significant heterogeneity was found in the number of synonymous substitutions in the three lineages between the two samples of genes with different recombination rates. This is partly due to a lack of the lineage effect in the D. melanogaster and Drosophila simulans lineages in the AS-C and ci genes in contrast to AKASHI's observation of genes in regions of normal recombination. The higher codon bias in Drosophila yakuba as compared with D. melanogaster and D. simulans was observed in the four AS-C genes, which suggests change(s) in action of natural selection involved in codon usage on these genes. Fluctuating selection intensity may also be responsible for the observed locus-lineage interaction effects in synonymous substitution.

THE constancy of the evolutionary rate across multiple lineages was first examined by OHTA and KIMURA 1971 . Subsequently LANGLEY and FITCH 1974 and GILLESPIE 1989 have distinguished locus-lineage interaction effects from lineage effects. Episodic protein sequence evolution has been taken as evidence of positive natural selection in action (GILLESPIE 1989 ), while lineage dependent evolution may arise due to generation-time and/or population-size effects (OHTA 1995 ; AKASHI 1996 ). Formally, other biological causes, however, can be responsible for irregular molecular evolution, and this issue still remains to be solved.

The constancy of the rate has been examined using the variance-to-mean ratio of the number of substitutions among lineages (dispersion index) as a measure (OHTA and KIMURA 1971 ; GILLESPIE 1989 ). This ratio should be one under the assumption of simple Poisson process and no estimation error. When the variance in the number of substitutions among lineages is greater than the expectation, several mechanisms can be put forward to explain the presence of locus-lineage interaction: changes in mutation rate at the individual gene basis, time-heterogeneous directional selection (GILLESPIE 1991 ), and fluctuating neutral space (TAKAHATA 1987 ). Region-dependent changes in recombination rate can also result in changes in substitution rate of nearly neutral mutations in the regions involved through the background selection effects (CHARLESWORTH et al. 1993 ), and thus may lead to the inflation of the dispersion index. This effect should be distinguished from generation-time and population-size effects, which affect all the genes in the genome in the same direction. The final goal of this study is to know how often the degree and pattern of natural selection varies in time and to evaluate the action of natural selection in molecular evolution on this basis. The test of rate constancy among closely related species is the first step to estimate the frequency of changes in selective forces acting on individual genes and the relative contribution of the factors as mentioned above. In this context, knowledge of evolutionary-rate variation is important to deepen our understanding of the mechanisms of molecular evolution.

Among several methods used to examine the action of natural selection, the rate constancy test also has a couple of advantages. First, an excess of replacement substitutions compared with the number of synonymous substitutions is not necessarily required to get a significant result, meaning a higher detection power in certain situations. Second, independent analysis for replacement and synonymous substitutions can be made. The drawback of the necessity of assumptions of absolute divergence time is overcome in this study by the use of multiple-gene comparison.

Many of the previous analyses on rate-constancy have paid less attention to ancestral polymorphism, although GILLESPIE and LANGLEY 1979 and HUDSON 1983 have done analytical and simulation studies of its effect on the rate-constancy test. On the other hand, TAKAHATA 1986 has developed means of estimating the effective population size in an ancestral population using the mean and variance of the number of substitutions between two extant species among genes under constant-rate assumption. However, there seems to be considerable rate variation among loci at the silent site (SHARP and LI 1989 ; MORIYAMA and GOJOBORI 1992 ; WOLFE and SHARP 1993 ; OHTA and INA 1995 ). It is highly desirable to discriminate one effect from the other.

Recently, MORIYAMA and POWELL 1996 summarized intraspecific variation in Drosophila. The amount of variation can be as high as 0.03 in terms of nucleotide difference per silent site in Drosophila melanogaster (Amy; INOMATA et al. 1995 ) and 0.08 in Drosophila simulans (Est-6; KAROTAM et al. 1995 ), which is expected to have a big sampling variance in a sample of two sequences (NEI and TAJIMA 1981 ). On the other hand, the smallest estimate obtained in both D. melanogaster and D. simulans is 0. It is well known that in Drosophila the degree of intraspecies variation is highly dependent on the recombination rate in the region involved (BEGUN and AQUADRO 1992 ). Thus, genes with low silent variation exclusively come from regions of very reduced recombination rates such as the tip of the X chromosome and the fourth chromosome. The changes of the banding pattern of the salivary gland polytene chromosome have been studied in some detail, and the structural changes were inferred for the D. melanogaster species subgroup (LEMEUNIER and ASHBURNER 1976 ). A comparison of D. melanogaster and Drosophila yakuba chromosomes, for instance, revealed the occurrences of at least four inversions on the X chromosome. As a result, the 2B region of D. melanogaster is located far from the tip of the X chromosome in D. yakuba due to a long inversion covering 2B through 6D. Frequent chromosomal structural changes mean that the degree of intraspecies variation may vary greatly, even in relatively closely related species pairs. In the presence of high variation and frequent chromosomal structural changes, the ancestral polymorphism could seriously inflate the rate variation in the case of a low degree of between-species divergence, 0.5, per site, whereas a comparison of more distantly related species pairs decreases the accuracy of estimates of the number of substitutions. The impact of ancestral polymorphism is presented graphically in Figure 1.

View larger version (14K): In this window In a new window Download PPT slide   Figure 1. —Possible gene trees of a sample of three genes, one from each of three species. Thin lines in (b) and (c) stand for ancestral polymorphism at the time of species separation shown by arrows. (a) shows a history without ancestral-population variation. (b) and (c) depict the impact of ancestral polymorphism: (b) high variation in ancestral population of species A and B but low variation in common ancestor of three species, and the opposite pattern in (c). The mixture of these histories among genes produces variation in the number of substitutions inferred for three lineages even under the condition of constant evolutionary rate.

This article presents an analysis of the evolutionary rate variation among three species of Drosophila, D. melanogaster, D. simulans and D. yakuba, using the published sequences of eight genes from regions of normal recombination rates. At the same time, to exclude possible effects of ancestral polymorphism, we have taken advantage of the fact that the genes on the tip of the X chromosome and the fourth chromosome have very reduced levels of intraspecies variation and no structural changes were inferred in these regions among these three species (LEMEUNIER and ASHBURNER 1976 ). This provides a unique opportunity to study the fluctuation of the evolutionary rate, under the condition of the least effect of ancestral polymorphism. As a check on ancestral polymorphism, new data for four genes of the achaete-scute complex (AS-C) on the tip of the X chromosome and the ci gene on the fourth chromosome were also studied in this article. The results obtained suggest irregular synonymous substitution patterns in the three lineages both in the eight genes of normal recombination rate and in the four genes from the regions of reduced recombination. This study also provides evidence for the change in codon bias of the four AS-C genes between D. melanogaster and D. simulans, on the one hand, and D. yakuba, on the other hand.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Sequence sources:
The number of sequences studied, their sources and GenBank accession numbers (release 98.0) are as follows: eleven alleles of D. melanogaster (excluding Adh^FChD), six alleles of D. simulans and twelve strains of D. yakuba for the Adh from MCDONALD and KREITMAN 1991 ; six alleles of D. melanogaster and six of D. simulans for the per from KLIMAN and HEY 1993A , and a D. yakuba per sequence (X61127) from THACKERAY and KYRIACOU 1990 ; the sequence of TN329 line of D. melanogaster (L22721) from INOMATA et al. 1995 and a D. simulans (D17734) and a D. yakuba sequence (D17738) from SHIBATA and YAMAZAKI 1995 for the Amy; a sry sequence of D. melanogaster (X03121) from VINCENT et al. 1985 , and a sry sequence of D. simulans (U64718) and D. yakuba (U64719) from CACCONE et al. 1996 ; a D. melanogaster nullo sequence (X65444) from ROSE and WIESCHAUS 1992 , and a nullo sequence of D. simulans (U44733) and D. yakuba (U44732) from CACCONE et al. 1996 ; G6pd sequences of D. melanogaster (L13900), D. simulans (L13878) from EANES et al. 1993 and a D. yakuba sequence (U42750) from EANES et al. 1996 ; five strains of D. melanogaster, five of D. simulans and four of D. yakuba for the boss (AYALA and HARTL 1993 ); five strains of D. melanogaster, five of D. simulans and five of D. yakuba for the Rh3 (AYALA et al. 1993 ); a D. melanogaster (M17120) from VILLARES and CABRERA 1987 and a D. simulans (X62400) from MARTIN-CAMPOS et al. (1992) for the ac; a sc sequence of D. melanogaster (M17119) from VILLARES and CABRERA 1987 ; a l'sc sequence of D. melanogaster (X12549) from ALONSO and CABRERA 1988 ; six alleles of D. melanogaster and six of D. simulans from HILTON et al. 1994 for the ase; ten strains of D. melanogaster and nine of D. simulans for the ci (ORENIC et al. 1990 ; BERRY et al. 1991 ).

Fly stock, cloning, sequencing and cytological examination:
A strain of D. yakuba (stock no. 14021-0261.0) was provided by the National Drosophila Species Resource Center at Bowling Green, Ohio. A P1 phage clone, DS06327, covering a large portion of AS-C was gifted from the laboratory of T. YAMAZAKI (Kyushu University, Fukuoka, Japan). The genomic library of the D. yakuba strain was constructed in the DASHII vector. The preparation of DNA probe consisted of the following: The fragments of the ac, sc, and l'sc genes were amplified by PCR from DS06327 and cloned into the pCRII vector by using the TA-cloning kit (Invitrogen, NV Leek, The Netherlands). The primers used for the amplification were derived from the published sequences as follows: 5'-TGTTTTTACTTGGCTCTGATGT-3', and 5'-GTGTTATGGTTGGGTGCGACTA-3' for the ac gene from the sequence in VILLARES and CABRERA 1987 ; 5'-AGGGTTTAGGACGAAGGGACT-3', and 5'-AGAAAATAGGGCGTGGTGGTAA-3' for the sc (VILLARES and CABRERA 1987 ); and 5'-ATTCCAGCAGCAGCATTACCAG-3', and 5'-TTTTCTATCATTGTCTTCCATT-3' for the l'sc (ALONSO and CABRERA 1988 ). The primers used for the construction of the ase and ci gene probes were the same as in HILTON et al. 1994 and BERRY et al. 1991 , respectively. The fragments were amplified from the D. melanogaster strains. The labeling of DNA probes with digoxigenin-dUTP and hybridization was carried out following the protocol from Boehringer Mannheim (Mannheim, Germany). A single phage clone of D. yakuba hybridized to the probe of D. melanogaster was selected for each gene, and the product of subcloning was inserted into the pBluescriptII vector. The DNA sequences for both strands were determined on an ABI PRISM 310 Genetic Analyzer (Perkin-Elmer Applied Biosystems, Foster City, CA).

The DNA sequences of the sc and l'sc genes were also determined for a strain of D. simulans collected in the Congo, Sim-5(G20), which was established by half sib-mating for 20 generations. A 1.2-kb sc and a 0.9-kb l'sc fragment (the same primers as those in probe construction), and a 0.6-kb sc (primers: 5'-CTCAGGGCATAATACCTACTA-3' and 5'-AAACTGGTCTCGTGGTTCTTA-3') and a 0.7-kb l'sc (5'-CTGCCGAAGTCGCTGCCTCTG-3', and 5'-GCTCCCGTCAAGAACTGTTGC-3') fragment were amplified, purified through columns, and used as templates for direct sequencing. The nucleotide sequence data reported in this article will appear in the DNA Data Bank of Japan, European Molecular Biology Laboratory and GenBank nucleotide sequence databases with the following accession numbers: AB005751 and AB005797–AB005802.

In situ hybridization with digoxigenin-labeled probes was performed on the salivary gland chromosomes of D. yakuba. The D. yakuba sc, l'sc and ci clones were labeled via PCR using the primers (5'-CTCATTCTTGTGTAAGGTGTC-3', and 5'-AGGCACTTGGTTCAACTCAAC-3') for the sc, and the same primers as in the screening probe construction for the l'sc (a 0.7kb fragment) and ci (BERRY et al. 1991 ).

Sequence analysis:
The numbers of synonymous and nonsynonymous sites were calculated according to INA 1995 . The arithmetic mean of the three species was employed in the following analysis. The ratio of transitional to transversional substitution rates was obtained as that at the third iteration cycle described in INA's method 2, using all the synonymous changes among the three species involved. For 11 of the 13 genes, one sequence, randomly chosen from each species, was used to count the transitional and transversional synonymous changes at the third codon position and to estimate the number of synonymous and replacement sites when two or more sequences were available. These are listed below: a sequences of D. melanogaster and D. yakuba and c sequence of D. simulans (X57365; X57361) in MCDONALD and KREITMAN 1991 for the Adh; ME-NJ1 of D. melanogaster (L07825) and SI-CA1 of D. simulans (L07826) for the per (KLIMAN and HEY 1993A ); ME-NJ1 of D. melanogaster and SI-CA1 of D. simulans for the ase (HILTON et al. 1994 ); a D. melanogaster (X54360) and a D. simulans sequence with T at position 2338 of the ci (ORENIC et al. 1990 ; BERRY et al. 1991 ). For the remaining two genes, the transitional and transversional synonymous substitutions at the third codon position were counted for the entire group of samples, 14 strains from three species for the boss (AYALA and HARTL 1993 ) and 15 for the Rh3 (AYALA et al. 1993 ), but the subsequent calculation was made in the same manner as for the other genes by using a sequence of D. melanogaster, f of D. simulans and k of D. yakuba in AYALA and HARTL 1993 for the boss, and a sequence of D. melanogaster, f of D. simulans and p of D. yakuba in AYALA et al. 1993 for the Rh3.

The number of substitutions was estimated by KIMURA's (1980) formulae, and the means for all the combinations were given when two or more sequences were available. In addition to this estimate, the numbers of substitutions were counted for each lineage on the basis of the following parsimonious assumptions without correction. The nucleotide at the node was estimated as one(s) to require the smallest number of total substitutions. When the nucleotide at the node could not determine uniquely, the same probability was given to each evolutionary pathway except for the following cases: There were 11 codons for 13 genes, in which two or more substitutions occurred and two different pathways to require different numbers of replacement substitutions were possible. The weight factors for each pathway were calculated according to MIYATA and YASUNAGA 1980 . Based on the factors obtained, these 11 codons were classified into two groups: one with roughly equal weight for two pathways (weight factors for favored pathway range from 0.50 to 0.65) and one with biased weight (weight factors for favored pathway range from 0.85 to 0.99). Thus, equal weight (0.5) was given to two pathways for the former (six codons) and the pathways to require the smaller number of replacements were taken uniquely for the latter (5 codons).

For segregating sites within species, first the probability that the ancestral nucleotide of that species was one of the segregating nucleotides was assumed to be given by their frequencies (WATTERSON and GUESS 1977 ), and the same counting method was applied to each pathway. For example, let us assume that species A and B have nucleotides G, and that G and T are segregating in species C with frequencies of (5/6) and (1/6), respectively. In this case, the nucleotide at the node is inferred to be G irrespective of the ancestral nucleotide of species C. The probability that the ancestral nucleotide of species C is G is (5/6), in which case the number of substitutions are calculated to be (1/6) in the species C lineage. The probability that the ancestral nucleotide of species C is T is (1/6), where the number of substitutions are calculated to be (1 + 5/6). In sum, the number of substitutions in the species C lineage is estimated to be

These numbers of substitutions obtained without correction were employed for all the statistical tests except for the relative rate test (WU and LI 1985 ).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Evolutionary rate variation in eight genes from regions of normal recombination:
DNA sequences from D. melanogaster, D. simulans and D. yakuba were compared for eight genes: the Adh (cytological map position, 35B), per (3B), Amy (54A), sry (99D), nullo (6F), G6pd (18D), boss (96F) and Rh3 (92D). None of these genes are located in the regions of very reduced recombination rate such as the tip of the X and the proximal regions of the chromosomes. D. simulans is one of the most closely related species to D. melanogaster, and D. yakuba is distantly related to these two species (Species C in Figure 1). Thus, the D. yakuba lineage in this study represents the entire lineage from the common ancestor of D. melanogaster and D. simulans to D. yakuba. The main purpose of this study is to compare the substitution pattern in the three lineages among different genes and to know the degree of variation. With this aim, no information about direction of mutations in the D. yakuba lineage (either of occurrence of mutation in the lineage leading to D. yakuba or in the internal branch from the common ancestor of the three species through the common ancestor of D. melanogaster and D. simulans) does not affect the results. Table 1 and Table 2 show the estimates of the number of synonymous and replacement substitutions per site by KIMURA's two parameter method (1980) and the estimated numbers of substitutions in the three lineages on a basis of parsimony assumptions, respectively.

Nucleotide substitution pattern in the three lineages was examined for lineage effects and locus-lineage interaction effects. The lineage effect was clearly seen, both in the synonymous and replacement substitutions, as the higher evolutionary rate in the D. melanogaster in comparison with the D. simulans lineage as described in AKASHI 1995 , AKASHI 1996 . For synonymous substitution, the relative rate test (WU and LI 1985 ) showed the ratio of melanogaster-simulans difference to its standard deviation as 2.0 or more in two genes, the proximal gene of Amy and sry. And similar results were found for replacement substitution in three genes, the proximal gene of Amy, sry and G6pd. Only the replacement substitution of the G6pd showed significantly faster evolution in D. simulans than in D. melanogaster. Since the amount of within-species variation is greater in D. simulans than in D. melanogaster (MORIYAMA and POWELL 1996 ), a difference in mutation rate between the two species is an unlikely cause for the higher evolutionary rate in D. melanogaster. The most likely explanation is a lower efficiency of selection in the D. melanogaster lineage due to a smaller population size (AKASHI 1995 ) and a lower recombination rate in D. melanogaster (OHNISHI and VOELKER 1979 ; TRUE et al. 1996 ; HAMBLIN and AQUADRO 1996 ) as compared to D. simulans.

Besides the lineage-dependent pattern of evolution, the synonymous changes in the three lineages occurred irregularly among the eight genes (3 x 8 test of independence in SOKAL and ROHLF 1995 ; G with Williams's correction = 31.8 with 14 df, P < 0.005). Although the same kind of statistical test was not applied due to the small number of replacement substitutions in each lineage, the replacement substitution pattern for the G6pd and per genes showed a higher dispersion index. The faster protein sequence evolution of the G6pd gene in D. simulans has been described well in EANES et al. 1993 , EANES et al. 1996 .

One possible explanation for the observed rate-irregularity is gene-dependent fluctuation of mutation rate. If the rate of mutation varies among lineages, then there will be a correlation between synonymous and replacement substitution rates. However, statistically significant heterogeneities were found in the numbers of synonymous and replacement substitutions for the per (a larger ratio of replacement to synonymous substitutions in the D. yakuba than the other two lineages: Fisher's exact test, P = 0.0006), and Rh3 (a larger ratio of replacement to synonymous substitutions in the D. melanogaster lineage as compared to the other lineages: Fisher's exact test, P = 0.017). Since any one of the three lineages could show a significant deviation from the other two lineages, we may take the significance level of 0.017 (= 1 - 095^1/3) instead of 0.05. There is still a significant heterogeneity between the two types of substitutions in these two genes. Thus, it is suggested that fluctuation of mutation rate alone cannot explain the whole picture of the observed substitution patterns.

Cytological examination of AS-C genes and ci gene in D. yakuba:
LEMEUNIER and ASHBURNER 1976 compared the banding pattern of the salivary chromosomes among six sibling species of D. melanogaster. The findings showed that the fourth chromosomes of D. melanogaster and D. yakuba were generally similar to each other, but not enough so to be conclusive. The identity of the tip of the X chromosomes of the three species is also ambiguous (HORTON 1939 ). Then, in situ hybridization was performed on the polytene chromosomes of D. yakuba with the D. yakuba clones. Figure 2 shows that the l'sc and ci genes are located on the tip of the X and the fourth chromosome in D. yakuba, respectively, as in D. melanogaster.

View larger version (86K): In this window In a new window Download PPT slide   Figure 2. —Localization of the l'sc and ci genes on the polytene chromosome of D. yakuba. In situ hybridization using the D. yakuba l'sc and ci clones as probes to salivary gland chromosomes indicated that the l'sc (A) and ci (B) genes are located on the tip of the X and the fourth chromosome, respectively, just as in D. melanogaster. The sc gene is also localized on the tip of the X chromosome (data not shown).

Evolutionary rate variation in four genes of achaete-scute complex and ci gene sampled from regions of reduced recombination rates:
The four AS-C genes, ac, sc, l'sc and ase, and the ci gene were cloned from a genomic library of a D. yakuba strain, and their nucleotide sequences were determined. The length of sequence determined was 878 bp for the ac, 1314 bp for the sc, 924 bp for the l'sc, 1564 bp for the ase gene, all of which include the entire putative coding sequences. Because only partial sequences of the D. simulans ase were available, the corresponding coding regions of the ase genes were employed in the analysis. A 1293-bp fragment of the ci gene was sequenced, and this region contained the whole of the coding regions surveyed in BERRY et al. 1991 . In addition, the entire coding regions of the sc and l'sc of D. simulans were amplified and their sequences were determined for the analysis.

A few of the insertions/deletions in the coding regions among the three species were observed as follows: a two-amino-acid insertion in the D. yakuba sc sequence and a one-amino-acid deletion in the D. melanogaster sc; a one-amino-acid deletion in the D. yakuba l'sc; a two-amino-acid insertion in the D. melanogaster ase; one- and nine-amino-acid insertions and a two-amino-acid deletion in the D. yakuba ase; two two-amino-acid insertions in the D. yakuba ci; no length variation in the ac. A Gln-rich sequence was found in the ase gene of D. melanogaster [(Gln)₅-Val-Gln], which is the same as in the ac and sc genes (VILLARES and CABRERA 1987 ). There was, however, a nine-amino-acid insertion (duplication) of the Gln-rich sequence in D. yakuba, resulting in the sequence (Gln)₁₂-Pro-Gln-Val-Gln. Besides these gaps, there may be a difference in the translation start site of the sc transcripts as shown in Figure 3. These gaps, including the sequence anterior to the conserved ATG codon in the sc sequence, were excluded in the following analyses.

View larger version (39K): In this window In a new window Download PPT slide   Figure 3. —Substitution in the first initiation codon of the sc transcript in D. yakuba. Upper and lower sequences show the D. melanogaster and D. yakuba sequences, respectively. * indicates the ATG codon in D. melanogaster as well as in D. simulans, but AAG in the D. yakuba sequence. # is the conserved ATG codon for all the three species. Gaps are shown by hyphens. The sequence and position of D. melanogaster refer to the sequence in VILLARES and CABRERA 1987 .

The summary of DNA variation among the three species is given in Table 1 and Table 3. The ratio of replacement to synonymous substitutions for the pooled data varied significantly among the three lineages (G with Williams's correction = 11.6, P < 0.005). This is partly due to the accelerated protein sequence evolution in the D. melanogaster lineage (AKASHI 1996 ) but no acceleration for synonymous substitution. No lineage effect in synonymous substitution provides an interesting contrast to the pattern observed for the eight genes from the regions of normal recombination, in which the greater number of substitutions in the D. melanogaster lineage as compared to those of the D. simulans lineage was observed in synonymous substitution as well as in replacement substitution (Table 2). The reduced effectiveness of selection in regions of reduced recombination is a plausible cause for no effect of the population sizes in the two species.

When we examined the pattern of substitution in the three lineages, there was a significant heterogeneity among the five genes for synonymous substitution as in the genes with normal recombination (G with Williams's correction = 19.6, P < 0.025). This result suggests that the presence of locus-lineage interaction for synonymous substitution rate does not simply come from effects of ancestral polymorphism. For replacement substitution, only the ac gene showed a significantly different substitution pattern from the other four genes (G with Williams's correction = 4.2, P < 0.05 for the numbers of substitutions between D. melanogaster + D. simulans and D. yakuba). Lower frequency of statistical significance for replacement substitution patterns is at least partly due to the small number of substitutions in each lineage, and it should not be simply compared to synonymous substitution.

The ratio of replacement to synonymous substitutions was significantly higher in the D. melanogaster for the ase (corrected G = 6.1, P < 0.025, for D. melanogaster vs. D. simulans + D. yakuba), and higher in the D. melanogaster and simulans lineages for the l'sc (corrected G = 4.9, P < 0.05, for D. melanogaster + D. simulans vs. D. yakuba). These results again suggest that the observed erratic substitution patterns are not entirely because of gene-dependent fluctuation of mutation rate.

Comparison of the four AS-C and ci genes with the remaining eight genes:
A comparison was made of the pooled number of the substitutions in the three lineages (Table 4). There was a significant heterogeneity between the AS-C and ci genes, and the remaining eight genes for synonymous substitution, but not for replacement substitution. One reason for the heterogeneity in synonymous substitution can be attributed to a lack of difference between the D. melanogaster and D. simulans lineages of the four AS-C and ci genes. This is most likely due to the low efficiency of natural selection in regions of reduced recombination, irrespective of the population size difference between the two species. Indeed, the codon bias for these genes is generally very low for the D. melanogaster genes as described in KLIMAN and HEY 1993B . A significant heterogeneity between the two samples of genes still existed for a D. melanogaster + D. simulans vs. D. yakuba comparison (corrected G = 13.7, P < 0.001). As mentioned in the Introduction, ancestral polymorphism is one of the possible causes. The current findings may be explained by the higher variation in an ancestral population of D. melanogaster and D. simulans than that in an ancestral population of the three species (Tree b in Figure 1). This, however, seems unlikely because replacement substitution showed the opposite pattern, that is, a higher ratio of the number of substitutions in the D. yakuba lineage as compared to those of the D. melanogaster and D. simulans lineages for genes with higher recombination rates than the four AS-C and ci genes. The present finding may also be explained if the following conditions exist: Natural selection acts on synonymous changes and the degree of preference for codons varies among species, depending on genes.

AKASHI 1995 classified codons into two types, "major (preferred)" and "nonmajor (unpreferred)." Major codons are defined as codons which appear in higher frequency in higher-biased genes. Synonymous substitutions are subsequently classified into preferred substitutions which mean substitutions from a nonmajor codon to a major codon, unpreferred substitutions from a major codon to a nonmajor codon, and the others as explained in AKASHI 1995 . The numbers of these substitutions in this study are given in Table 5. Because we cannot determine the direction of the substitutions in the D. yakuba lineage without any outgroup species, the numbers of homologous codons at which D. yakuba encodes a major codon and the common ancestor of D. melanogaster and D. simulans encodes a nonmajor codon (y_maj/ms_non) and the codons in the opposite configuration (y_non/ms_maj) for the substitutions in the D. yakuba lineage are given in Table 5. An excess of preferred substitutions over unpreferred substitutions (or the number of y_maj/ms_non over that of y_non/ms_maj) was observed in the D. yakuba lineage for the genes from the regions of reduced recombination, although in the ci gene the two types of substitutions occurred almost the same number of times (corrected G = 17.4, P < 0.001, for 19.9 preferred and 36.3 unpreferred substitutions in the D. melanogaster and D. simulans lineages, and 89.2 y_maj/ms_non and 41.2 y_non/ms_maj in D. yakuba). The y_maj/ms_non substitution in Table 5 refers to the occurrence of preferred codon substitution in the lineage leading to D. yakuba or unpreferred codon substitution in the internal branch from the common ancestor of the three species through the common ancestor of D. melanogaster and D. simulans. In any case, the important finding is that the change of codon bias depends on genes. Indeed, a significant heterogeneity in the numbers of y_maj/ms_non and y_non/ms_maj in the D. yakuba lineage existed between the genes from different recombination rates (corrected G = 19.0, P < 0.001), but neither in the D. melanogaster nor in the D. simulans lineage. The higher codon bias for the four AS-C genes in D. yakuba as compared to D. melanogaster/D. simulans suggests change(s) in selection intensity on codon usage for the four AS-C genes during the evolutionary course of these three species. The higher number of y_maj/ms_non substitutions is one of the causes of a significant heterogeneity in the substitution pattern between the genes with reduced and normal recombination rates. This may also have made a contribution to the erratic synonymous substitution pattern among the three lineages. In fact, there was a significant heterogeneity in the numbers of y_maj/ms_non and y_non/ms_maj for the eight genes in regions of normal recombination (corrected G = 25.8, P < 0.001).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

While the dispersion index has been used as a parameter to test rate constancy in the previous studies (GILLESPIE 1989 ; OHTA 1995 ), there are a few criticisms of this index. BULMER 1989 pointed out that the corrections for multiple hits significantly inflate the index even in the distance of 0.25. He developed new test statistics, assuming a star phylogeny. This assumption does not hold for the three species studied here. For this reason, the number of substitutions inferred under the parsimony assumptions were employed for the statistical tests in this study, although the mean number of substitutions per site in each lineage was small even for the synonymous ones: 0.05 in the D. melanogaster lineage, 0.03 in D. simulans and 0.18 in D. yakuba. GOLDMAN 1994 criticized the statistical significance of the dispersion index in GILLESPIE's study (1989). Gene-dependent construction of phylogenies with estimated branch length in GOLDMAN's study is problematic as mentioned in NIELSEN 1997 . In fact, GOLDMAN assumed the branching order of {[(goat – rabbit) – human] – mouse} for the -hemoglobin gene, and [(cow – mouse) – human] for the cytochrome oxidase 2 gene. The difference in phylogenies assumed among genes should be tested as well. High degrees of ancestral polymorphism would be required to explain the difference in the two genes. Although more careful examination of effects of sampling designs and assumptions is needed, the index can be a useful measure of rate variation in certain conditions.

The principal findings of this study were (1) erratic synonymous substitution patterns among the three lineages and (2) lineage effect in protein sequence evolution. The first finding was obtained independently from the published sequences of the eight genes from the regions of normal recombination and the five genes located in regions of greatly reduced recombination rates in D. melanogaster. Consequently, the erratic synonymous substitution patterns cannot be attributed to the effects of ancestral polymorphism. One possible cause is fluctuation of the mutation rate, but the high occurrence of significant heterogeneity of synonymous and replacement substitution patterns does not support this hypothesis. An alternative, and more plausible, explanation is the changes in degree of selective constraint involved in codon usage. Indeed, higher codon bias was observed in the four AS-C genes of D. yakuba as compared to D. melanogaster and D. simulans.

The estimated dispersion index for the replacement substitutions was generally low, and only the per, G6pd and ac genes showed an indication of erratic evolution among the three lineages. In contrast to this finding, previous reports on three orders of mammals show generally higher dispersion indices for replacement substitutions than for synonymous ones (GILLESPIE 1989 ; OHTA 1995 ). It is of great interest to analyze possible causes for the difference in the rate variation between the two studies, one of Drosophila and the other of mammals, in detail. Effects of the differences in the analytical designs between the two studies, however, should be taken into consideration at first. The smaller number of genes examined in this study, as compared with 49 genes in OHTA 1995 , could produce a larger correlation between the weight factors and the number of substitutions at individual genes. Thus, the relatively low dispersion indices in the Drosophila genes could be contributed by the sample size. The level of divergence in this study is also lower than in the mammalian studies: the mean numbers of substitutions per site in all the lineages in the present study and in OHTA 1995 were 0.27 and 0.67 for the synonymous substitutions and 0.03 and 0.16 for the replacement substitutions, respectively. OHTA 1995 found a significant correlation between the dispersion index and the number of substitutions, which could be due to the correction bias (BULMER 1989 ) and/or other reasons (OHTA 1995 ). Thus, the difference in the levels of divergence is a possible explanation for the difference in dispersion index between the two studies. In addition, branching patterns may also affect the levels of the dispersion index. The difference among the weight factors for the three lineages is smaller in OHTA 1995 than in this study. The three weight factors for the replacement substitutions are 1.3, 1.0 and 0.8 in OHTA 1995 , while those in this study were 1.7, 0.9 and 0.4 for the five genes from the regions of reduced recombination (Table 3). Bias in the weight factors might reduce the power of detection of rate variation as compared with a star phylogeny with the same levels of total divergence.

Besides these technical biases due to the different research designs employed in each study, the higher levels of overdispersion for the replacement substitutions in the mammals as compared to Drosophila may reflect differences in the degree of functional redundancy due to gene duplications. Accelerated protein sequence evolution has been observed during the limited periods following gene duplications in a number of genes (LI 1985 ; OHTA 1994 ; IWABE et al. 1996 ), although whether the accelerated rates are due to relaxation of selective constraints or accumulation of advantageous mutations is a controversial issue. Furthermore, a significant fraction of erratic evolution observed in the mammals comes from multiplicated genes such as the growth hormone and prolactin genes (GILLESPIE 1989 ; OHTA 1995 ). It is certainly true that there is a big difference in the number of genes between mammals and Drosophila (LEWIN 1997 ). Advances in genome projects for many different taxa will help measure and compare the degree of redundancy or gene duplication in genomes. Keeping in mind the possible biases in comparisons, further study is needed to understand the relationship between the degree of redundancy and frequency of episodes in molecular evolution.

Another factor which may affect relative degrees of rate variation for the synonymous and replacement substitutions is the efficiency of natural selection. As mentioned above, the higher rate variation in the Drosophila genes could be due to a combination of weak selection with fluctuating selection intensities and fluctuating population sizes. Lower selection efficiency on the synonymous substitutions due to smaller population sizes of the mammals and thus lower rate variation could partly explain the relatively higher dispersion index for the replacement substitutions.

AKASHI 1995 , AKASHI 1996 discovered the lineage effect, that is, the accelerated protein sequence evolution and the accumulation of unpreferred synonymous changes in the D. melanogaster lineage as compared to the D. simulans lineage. The present study showed no acceleration in the synonymous substitution rate in D. melanogaster for the five genes from the regions with reduced recombination, while a significantly higher accumulation of synonymous substitutions was observed for the eight genes from the regions with variable recombination rates in the D. melanogaster lineage as compared to the D. simulans lineage. The absence of lineage effect in the four AS-C and ci genes can be explained by less efficient natural selection in the regions of very reduced recombination rates. On the other hand, there still exists a significantly greater accumulation of replacement substitutions in the five genes in the regions of low recombination in the D. melanogaster lineage as compared to the D. simulans lineage (G = 6.7 with Williams's correction, P < 0.01). This suggests that a significant fraction of replacement changes has greater selection coefficients than synonymous ones, but that it may be fixed in smaller populations such as D. melanogaster.

The four AS-C genes, the proneural ac, sc and l'sc genes, and the neural precursor gene ase, encode the basic helix-loop-helix (bHLH) class of transcription regulators. Although these proteins are not completely interchangeable in several aspects such as the numerator function of the sc (sisterless b) in sex determination, a great deal of functional redundancy has been reported among these genes in the proneural function promoting neural development (e.g., PARKHURST et al. 1993 ; BALCELLS et al. 1988 ; JIMENEZ and CAMPOS-ORTEGA 1990 ; PARRAS et al. 1996 ). Based on their functions in neural development and sex determination, the level of expression of these genes is considered to be very important in the course of development. Comparison of the numbers of "preferred" and "unpreferred" codon changes revealed a significant difference between the D. melanogaster-simulans lineage and the D. yakuba lineage, suggesting change(s) in selection intensity. Natural selection intensity varies due to changes in the required level of expression. Efficiency of selection can also vary according to the recombination rate of the region involved, although the AS-C genes are cytologically located in the tip region of the X chromosome in all three species. Investigation of intraspecific variation of these genes and of other genes in this region in D. yakuba will help determine which is a better explanation for the variation in natural selection intensity.

Nonrandomness of distribution of intra- or interspecific variation along DNA sequences is often studied by window analysis. This analysis, coupled with knowledge of protein structure and its function, can be a powerful tool to detect the action of natural selection. Nonrandomness, however, does not necessarily come from variation in mutation rate or degree of selective constraint along the sequence alone. If we can infer which lineages have mutations (based on the parsimonious assumption or some other methods), nonrandom distribution of substitutions may be further divided into lineage-dependent/lineage-specific clustering and lineage-independent clustering. The former refers to a clustering of substitutions along sequences in individual lineages, and the occurrence of lineage-dependent clustering may be due to epistatic interaction among sites including compensatory mutations. The pattern observed in the G6pd gene is a good example of the lineage-specific clustering of substitutions. As mentioned above, EANES et al. 1993 , EANES et al. 1996 showed the accelerated protein sequence evolution in the D. simulans lineage (15 replacement substitutions in D. simulans and 6 in D. melanogaster). Many of the replacement changes in D. simulans are found in the very narrow regions in exons 2 and 3, but the substitutions in the D. melanogaster lineage are evenly distributed over the entire region examined. The ac and per genes showed relatively higher rates of protein sequence evolution in the D. yakuba lineage. In this context, it would be intriguing to test for presence of lineage-specific clustering of the substitutions for these genes using outgroup sequences. When we have a large enough number of substitutions in each of the lineages compared, computer simulation analysis will provide a mean of statistical analysis of lineage-dependent clustering effects.

We now come to understand that, besides gene duplication and protein sequence variation, changes in tissue or developmental specificity and in expression levels have played crucial roles in the morphological and developmental evolution of organisms. It is certainly a necessary task to survey the kind of changes that have occurred during the course of evolution. Further DNA sequence analysis may detect even small changes in the action of natural selection, for instance, suggested by the codon usage change in the four AS-C genes with the redundant function. This information, in turn, may provide insights into evolutionary changes in expression levels and changes responsible for morphological evolution. Interspecific variation may also be useful as a source to study genetic interaction among sites in one molecule.

	ACKNOWLEDGMENTS

I am grateful to TOMOKO OHTA, MASARU IIZUKA and YASUO INA for their suggestion and encouragement. I thank SHIGEO HAYASHI for permission to use sequencing facilities in his laboratory. I also thank YUMIKO YAMASHITA for her technical assistance with this project and LEAH GILNER and two anonymous reviewers for improving the manuscript. This work was supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Science, Sports and Culture of Japan.

Manuscript received September 19, 1997; Accepted for publication February 4, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AKASHI, H., 1995  Inferring weak selection from patterns of polymorphism and divergence at "silent" sites in Drosophila DNA. Genetics 139:1067-1076[Abstract].

AKASHI, H., 1996  Molecular evolution between Drosophila melanogaster and D. simulans: reduced codon bias, faster rates of amino acid substitution, and larger proteins in D. melanogaster.. Genetics 144:1297-1307[Abstract].

ALONSO, M. C. and C. V. CABRERA, 1988  The achaete-scute gene complex of Drosophila melanogaster comprises four homologous genes. EMBO J. 7:2585-2591[Medline].

AYALA, F. J. and D. L. HARTL, 1993  Molecular drift of the bride of sevenless (boss) gene in Drosophila.. Mol. Biol. Evol. 10:1030-1040[Abstract].

AYALA, F. J., B. S. W. CHANG, and D. L. HARTL, 1993  Molecular evolution of the Rh3 gene in Drosophila.. Genetica 92:23-32[Medline].

BALCELLS, L., J. MODOLELL, and M. RUIZ-GÓMEZ, 1988  A unitary basis for different Hairy-wing mutations of Drosophila melanogaster.. EMBO J. 7:3899-3906[Medline].

BEGUN, D. J. and C. F. AQUADRO, 1992  Levels of naturally occurring DNA polymorphism correlate with recombination rates in D. melanogaster.. Nature 356:519-520[Medline].

BERRY, A. J., J. W. AJIOKA, and M. KREITMAN, 1991  Lack of polymorphism on the Drosophila fourth chromosome resulting from selection. Genetics 129:1111-1117[Abstract].

BULMER, M., 1989  Estimating the variability of substitution rates. Genetics 123:615-619[Abstract/Free Full Text].

CACCONE, A., E. N. MORIYAMA, J. M. GLEASON, L. NIGRO, and J. R. POWELL, 1996  A molecular phylogeny for the Drosophila melanogaster subgroup and the problem of polymorphism data. Mol. Biol. Evol. 13:1224-1232[Abstract].

CHARLESWORTH, B., M. T. MORGAN, and D. CHARLESWORTH, 1993  The effect of deleterious mutations on neutral molecular variation. Genetics 134:1289-1303[Abstract].

EANES, W. F., M. KIRCHNER, and J. YOON, 1993  Evidence for adaptive evolution of the G6pd gene in the Drosophila melanogaster and Drosophila simulans lineages. Proc. Natl. Acad. Sci. USA 90:7475-7479[Abstract/Free Full Text].

EANES, W. F., M. KIRCHNER, J. YOON, C. H. BIERMANN, and I. WANG et al., 1996  Historical selection, amino acid polymorphism and lineage-specific divergence at the G6pd locus in Drosophila melanogaster and Drosophila simulans.. Genetics 144:1027-1041[Abstract].

GILLESPIE, J. H., 1989  Lineage effects and the index of dispersion of molecular evolution. Mol. Biol. Evol. 6:636-647[Abstract].

GILLESPIE, J. H., 1991 The Causes of Molecular Evolution. Oxford University Press, Oxford.

GILLESPIE, J. H. and C. H. LANGLEY, 1979  Are evolutionary rates really variable? J. Mol. Evol. 13:27-34[Medline].

GOLDMAN, N., 1994  Variance to mean ratio, R(t), for Poisson processes on phylogenetic trees. Mol. Phylogenet. Evol. 3:230-239[Medline].

HAMBLIN, M. T. and C. F. AQUADRO, 1996  High nucleotide sequence variation in a region of low recombination in Drosophila simulans is consistent with the background selection model. Mol. Biol. Evol. 13:1133-1140[Abstract].

HILTON, H., R. M. KLIMAN, and J. HEY, 1994  Using hitchhiking genes to study adaptation and divergence during speciation within the Drosophila melanogaster species complex. Evolution 48:1900-1913.

HORTON, I. H., 1939  A comparison of the salivary gland chromosomes of Drosophila melanogaster and D. simulans. Genetics 24:234-243[Free Full Text].

HUDSON, R. R., 1983  Testing the constant-rate neutral allele model with protein sequence data. Evolution 37:203-217.

INA, Y., 1995  New methods for estimating the numbers of synonymous and nonsynonymous substitutions. J. Mol. Evol. 40:190-226[Medline].

INOMATA, N., H. SHIBATA, E. OKUYAMA, and T. YAMAZAKI, 1995  Evolutionary relationships and sequence variation of -amylase variants encoded by duplicated genes in the Amy locus of Drosophila melanogaster.. Genetics 141:237-244[Abstract].

IWABE, N., K. KUMA, and T. MIYATA, 1996  Evolution of gene families and relationship with organismal evolution: rapid divergence of tissue-specific genes in the early evolution of chordates. Mol. Biol. Evol. 13:483-493[Abstract].

JIMÉNEZ, F. and J. A. CAMPOS-ORTEGA, 1990  Defective neuroblast commitment in mutants of the achaete-scute complex and adjacent genes of D. melanogaster. Neuron 5:81-89[Medline].

KAROTAM, J., T. M. BOYCE, and J. G. OAKESHOTT, 1995  Nucleotide variation at the hypervariable Esterase 6 isozyme locus of Drosophila simulans.. Mol. Biol. Evol. 12:113-122[Abstract].

KIMURA, M., 1980  A simple method for estimating evolutionary rates of base substitutions through comparative studies of nucleotide sequences. J. Mol. Evol. 16:111-120[Medline].

KLIMAN, R. M. and J. HEY, 1993a  DNA sequence variation at the period locus within and among species of the Drosophila melanogaster complex. Genetics 133:375-387[Abstract].

KLIMAN, R. M. and J. HEY, 1993b  Reduced natural selection associated with low recombination in Drosophila melanogaster.. Mol. Biol. Evol. 10:1239-1258[Abstract].

LANGLEY, C. H. and W. M. FITCH, 1974  An examination of the constancy of the rate of molecular evolution. J. Mol. Evol. 3:161-177[Medline].

LEMEUNIER, F. and M. ASHBURNER, 1976  Relationships within the melanogaster species subgroup of the genus Drosophila (Sophophora). II. Phylogenetic relationships between six species based upon polytene chromosome banding sequences. Proc. R. Soc. Lond. 193B:275-294[Medline].

LEWIN, B., 1997 Genes, VI. Oxford University Press, Oxford.

LI, W.-H., 1985 Accelerated evolution following gene duplication and its implication for the neutralist-selectionist controversy, pp. 333–352 in Population Genetics and Molecular Evolution, edited by T. OHTA and K. AOKI. Japan Scientific Societies Press, Tokyo and Springer-Verlag, Berlin.

MARTÍN-CAMPOS, J. M., J. M. COMERÓN, N. MIYASHITA, and M. AGUADÉ, 1992  Intraspecific and interspecific variation at the y-ac-sc region of Drosophila simulans and Drosophila melanogaster.. Genetics 130:805-816[Abstract].

MCDONALD, J. H. and M. KREITMAN, 1991  Adaptive protein evolution at the Adh locus in Drosophila.. Nature 351:652-654[Medline].

MIYATA, T. and T. YASUNAGA, 1980  Molecular evolution of mRNA: A method for estimating evolutionary rates of synonymous and amino acid substitutions from homologous nucleotide sequences and its application. J. Mol. Evol. 16:23-36[Medline].

MORIYAMA, E. N. and T. GOJOBORI, 1992  Rates of synonymous substitution and base composition of nuclear genes in Drosophila. Genetics 130:855-864[Abstract].

MORIYAMA, E. N. and J. R. POWELL, 1996  Intraspecific nuclear DNA variation in Drosophila.. Mol. Biol. Evol. 13:261-277[Abstract].

NEI, M. and F. TAJIMA, 1981  DNA polymorphism detectable by restriction endonucleases. Genetics 97:145-163[Abstract/Free Full Text].

NIELSEN, R., 1997  Robustness of the estimator of the index of dispersion for DNA sequences. Mol. Phylogenet. Evol. 7:346-351[Medline].

OHNISHI, S. and R. A. VOELKER, 1979  Comparative studies of allozyme loci in Drosophila simulans and D. melanogaster. II. Gene arrangement on the third chromosome. Jpn. J. Genet. 54:203-209.

OHTA, T., 1994  Further examples of evolution by gene duplication revealed through DNA sequence comparisons. Genetics 138:1331-1337[Abstract].

OHTA, T., 1995  Synonymous and nonsynonymous substitutions in mammalian genes and the nearly neutral theory. J. Mol. Evol. 40:56-63[Medline].

OHTA, T. and Y. INA, 1995  Variation in synonymous substitution rates among mammalian genes and the correlation between synonymous and nonsynonymous divergences. J. Mol. Evol. 41:717-720[Medline].

OHTA, T. and M. KIMURA, 1971  On the constancy of the evolutionary rate of cistrons. J. Mol. Evol. 1:18-25[Medline].

ORENIC, T. V., D. C. SLUSARSKI, K. L. KROLL, and R. A. HOLMGREN, 1990  Cloning and characterization of the segment polarity gene cubitus interruptus Dominant of Drosophila.. Genes Dev. 4:1053-1067[Abstract/Free Full Text].

PARKHURST, S. M., H. D. LIPSHITZ, and D. ISH-HOROWICZ, 1993  achaete-scute feminizing activities and Drosophila sex determination. Development 117:737-749[Abstract].

PARRAS, C., L. A. GARCIA-ALONSO, I. RODRÍGUEZ, and F. JIMÉNEZ, 1996  Control of neural precursor specification by proneural proteins in the CNS of Drosophila.. EMBO J. 15:6394-6399[Medline].

ROSE, L. S. and E. WIESCHAUS, 1992  The Drosophila cellularization gene nullo produces a blastoderm-specific transcript whose levels respond to the nucleocytoplasmic ratio. Genes Dev. 6:1255-1268[Abstract/Free Full Text].

SHARP, P. M. and W.-H. LI, 1989  On the rate of DNA sequence evolution in Drosophila.. J. Mol. Evol. 28:398-402[Medline].

SHIBATA, H. and T. YAMAZAKI, 1995  Molecular evolution of the duplicated Amy locus in the Drosophila melanogaster species subgroup: concerted evolution only in the coding region and an excess of nonsynonymous substitutions in speciation. Genetics 141:223-236[Abstract].

SOKAL, R. R., and F. J. ROHLF, 1995 Biometry. W. H. Freeman and Company, New York.

TAKAHATA, N., 1986  An attempt to estimate the effective size of the ancestral species common to two extant species from which homologous genes are sequenced. Genet. Res. 48:187-190[Medline].

TAKAHATA, N., 1987  On the overdispersed molecular clock. Genetics 116:169-179[Abstract/Free Full Text].

THACKERAY, J. R. and C. P. KYRIACOU, 1990  Molecular evolution in the Drosophila yakuba period locus. J. Mol. Evol. 31:389-401[Medline].

TRUE, J. R., J. M. MERCER, and C. C. LAURIE, 1996  Differences in crossover frequency and distribution among three sibling species of Drosophila. Genetics 142:507-523[Abstract].

VILLARES, R. and C. V. CABRERA, 1987  The achaete-scute gene complex of D. melanogaster: conserved domains in a subset of genes required for neurogenesis and their homology to myc.. Cell 50:415-424[Medline].

VINCENT, A., H. V. COLOT, and M. ROSBASH, 1985  Sequence and structure of the serendipity locus of Drosophila melanogaster: a densely transcribed region including a blastoderm-specific gene. J. Mol. Biol. 186:149-166[Medline].

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

WOLFE, K. H. and P. M. SHARP, 1993  Mammalian gene evolution: nucleotide sequence divergence between mouse and rat. J. Mol. Evol. 37:441-456[Medline].

WU, C.-I. and W.-H. LI, 1985  Evidence for higher rates of nucleotide substitution in rodents than in man. Proc. Natl. Acad. Sci. USA 82:1741-1745[Abstract/Free Full Text]