Multi-locus selection and the structure of variation at the white gene of Drosophila melanogaster

We surveyed sequence variation and divergence for the entire 5972-bp transcriptional unit of the white gene in 15 lines of Drosophila melanogaster and one line of D. simulans. We found a very high degree of haplotypic structuring for the polymorphisms in the 3' half of the gene, as opposed to the polymorphisms in the 5' half. To determine the evolutionary mechanisms responsible for this pattern we sequenced a 1612-bp segment of the white gene from an additional 33 lines of D. melanogaster from a European and a North American population. This 1612-bp segment encompasses an 834-bp region of the white gene in which the polymorphisms form high frequency haplotypes that cannot be explained by a neutral equilibrium model of molecular evolution. The small number of recombinants in the 834-bp region suggests epistatic selection as the cause of the haplotypic structuring, while an investigation of nucleotide diversity supports a directional selection hypothesis. A multi-locus selection model that combines features from both hypotheses and takes the recent history of D. melanogaster into account may be the best explanation for these data.