CORRIGENDA

J. A. Coyne , A. A. Felton , R. C. Lewontin , and R. S. Singh

In Coyne, J. A., 1976 Lack of genic similarity between two sibling species of Drosophila as revealed by varied techniques. Genetics 84: 593–607.

In our two papers, we used a statistical test of Warren Ewens to ask whether the distribution of alleles deviated significantly from that expected under the neutral theory. Our conclusion was that the distribution was too uneven as compared with the neutral prediction, but that this was not statistically significant. It has been pointed out to us by Dr. Geoffrey Watterson of Monash University that we used incorrect sample sizes in the computer program for Ewens' test, and that when the correct sample sizes are used, the differences become statistically significant. The correct value of F for D. pseudoobscura is 0.4605 with 126 and 26 degrees of freedom, while for D. persimilis, F is 0.2352 with 189 and 22 degrees of freedom. As pointed out by Singh, Lewontin and Felton, however, the significance cannot be depended upon because some of the unevenness may result from the pooling of separate population samples. In fact, 3 of the 12 populations of D. pseudoobscura are too even and 9 are too uneven, but only one of each type is significant. In D. persimilis all three populations are too uneven and all three are significantly so. Thus, the evidence for deviation from neutral theory is stronger than we had originally supposed, but not yet really convincing.