Intraspecific nuclear DNA variation in Drosophila

We have summarized and analyzed all available nuclear DNA sequencepolymorphism studies for three species of Drosophila, D. melanogaster (24loci), D. simulans (12 loci), and D. pseudoobscura (5 loci). Our majorfindings are: (1) The average nucleotide heterozygosity ranges from about0.4% to 2% depending upon species and function of the region, i.e., codingor noncoding. (2) Compared to D. simulans and D. pseudoobscura (which areabout equally variable), D. melanogaster displays a low degree of DNApolymorphism. (3) Noncoding introns and 3' and 5' flanking DNA shows lesspolymorphism than silent sites within coding DNA. (4) X-linked genes areless variable than autosomal genes. (5) Transition (Ts) and transversion(Tv) polymorphisms are about equally frequent in non-coding DNA and atfourfold degenerate sites in coding DNA while Ts polymorphisms outnumber Tvpolymorphisms by about 2:1 in total coding DNA. The increased Tspolymorphism in coding regions is likely due to the structure of thegenetic code: silent changes are more often Ts's than are replacementsubstitutions. (6) The proportion of replacement polymorphisms issignificantly higher in D. melanogaster than in D. simulans. (7) The levelof variation in coding DNA and the adjacent noncoding DNA is significantlycorrelated indicating regional effects, most notably recombination. (8)Surprisingly, the level of polymorphism at silent coding sites in D.melanogaster is positively correlated with degree of codon usage bias. (9)Three proposed tests of the neutral theory of DNA polymorphisms have beenperformed on the data: Tajima's test, the HKA test, and theMcDonald-Kreitman test. About half of the loci fail to conform to theexpectations of neutral theory by one of the tests. We conclude that manyvariables are affecting levels of DNA polymorphism in Drosophila, fromproperties of nucleotides to population history and, perhaps, matingstructure. No simple, all encompassing explanation satisfactorily accountsfor the data.