1816 Mol. Biol. Evol. 16(12):1816–1819. 1999  1999 by the Society for Molecular Biology and Evolution. ISSN: 0737-4038 Letter to the Editor Is the Fast/Slow Allozyme Variation at the Adh Locus of Drosophila melanogaster an Ancient Balanced Polymorphism? David J. Begun,* Andrea J. Betancourt,*² Charles H. Langley,‡ and Wolfgang Stephan² *Section of Integrative Biology and Institute for Cellular and Molecular Biology, University of Texas at Austin; ²Department of Biology, University of Rochester; and ‡Section of Evolution and Ecology, University of California at Davis Any general model of population genetics must be able to accommodate observations both of standing var- iation (polymorphism) and of evolutionary history (di- vergence). Interpretation of the connection (or lack thereof) between these two kinds of data impinges di- rectly on our ideas of evolutionary mechanism. Studies of the alcohol dehydrogenase locus (Adh) in Drosophila melanogaster and its close relatives provide what is per- haps our most detailed description of variation and evo- lution of a gene at the molecular and phenotypic levels. There is abundant evidence that two common elec- trophoretic variants in D. melanogaster populations caused by a threonine (Fast)/lysine (Slow) polymor- phism at codon 192 have different biochemical proper- ties and are currently under some form of variable se- lection (reviewed in Powell 1997). Other polymor- phisms at the locus may also be current targets of natural selection (Berry and Kreitman 1993), and epistatic ef- fects among some of these polymorphisms must be con- sidered (Stam and Laurie 1996). Molecular population genetic studies of Adh have revealed at least four inter- esting patterns. First, some fraction of the amino acid evolution in the melanogaster subgroup has been caused by natural selection (McDonald and Kreitman 1991). Second, a peak of excess silent polymorphisms is found at nucleotide sites that are closely linked to the Fast/ Slow polymorphism in D. melanogaster. The excess polymorphisms are incompatible with neutral molecular evolution; however, a model of balancing selection fits the data well (Kreitman and Hudson 1991). Third, the Fast electromorph in D. melanogaster appears to have increased in frequency in the recent past, as Fast hap- lotypes harbor little silent polymorphism and show little sequence divergence from Slow haplotypes (Kreitman 1983; Aquadro et al. 1986). Finally, correlated evolution of particular sites among different lineages suggests that epistatic selection acts on some property of Adh tran- scripts (Parsch, Tanda, and Stephan 1997). One message from these molecular analyses is that the evolutionary history of Adh is as complex as it is interesting. Despite the wealth of accumulated knowledge about Adh polymorphism and divergence, several questions re- main. One of the mysteries associated with Adh nucleotide variation in D. melanogaster is the observation by Kreit- man and Aguade ´ (1986) and Kreitman and Hudson (1991) Key words: Adh, DNA sequence variation, Drosophila melano- gaster, Zimbabwe population, balanced polymorphism. Address for correspondence and reprints: Wolfgang Stephan, De- partment of Biology, University of Rochester, Rochester, New York 14627-0211. E-mail: stephan@troi.cc.rochester.edu. that the Slow allozyme class shows ‘‘excess’’ DNA poly- morphism. This is difficult to explain if all nonneutral pat- terns of nucleotide variation in D. melanogaster are caused by balancing selection on the Fast/Slow polymorphism (Kreitman and Hudson 1991). The fact that the Fast hap- lotypes are of very recent origin (Kreitman 1983; Aquadro et al. 1986) also presents some problems for the simple hypothesis of ancient balancing selection acting on the Fast/Slow site. Despite these uncertainties and despite the caveats of Kreitman and his collaborators that some as- pects of the data are not easily reconciled with the hy- pothesis that the Fast/Slow polymorphism is an ancient balanced polymorphism (Kreitman and Aguade ´ 1986; Kreitman and Hudson 1991), the Adh data have entered the literature as a ‘‘textbook’’ example of ancient balancing selection (e.g., Li 1997; Powell 1997; Futuyma 1998). Currently available Adh sequences are primarily from non- African populations of D. melanogaster. Given the evi- dence that non-African populations are only recently de- rived from African populations (David and Capy 1988; Hale and Singh 1991; Begun and Aquadro 1993), we thought that new insights regarding the history of Adh in D. melanogaster might be gleaned from sequence data from African populations. Because these data are a new sample from nature, another advantage is that they can be used to test specific hypotheses raised from previous an- alyses (Kreitman and Hudson 1991). We sequenced part of the transcriptional unit of 10 randomly sampled Adh genes from Zimbabwe (fig. 1). These alleles were obtained from isochromosomal lines or by sequencing single copies of Adh from isofemale lines that were made hemizygous over the Df(2L)Sco7 deficiency. Lines designated by ‘‘k’’ are from Karibe; lines designated by ‘‘h’’ are from Harare. Sequencing reactions were performed directly on PCR products, run on an ABI 377 automated sequencer, and analyzed with DnaSP (Rozas and Rozas 1997). The sequences span coordinates 735–1683 (coordinates follow Kreitman 1983); they contain the entire coding region (exons 2, 3, 4), the two small introns (introns 1 and 2), and part of the larval leader sequence. The polymorphisms de- tected in this region are displayed in figure 1; the data of Kreitman (1983) and Laurie, Bridgham, and Choud- hary (1991) are also shown for comparison. As we might have expected given previous allo- zyme surveys of African populations (Singh, Hickey, and David 1982), all 10 Zimbabwe alleles belonged to the Slow electrophoretic class (i.e., they were lysine at codon 192). Estimates of  per site (Watterson 1975) at silent + noncoding sites for the Zimbabwe and Kreit- man samples were 0.0071 and 0.0079, respectively. One haplotype class, designated ZimS in subsequent discus- sion, is monomorphic and present at high frequency (4/