Polygenic mutation in Drosophila melanogaster: non-linear divergence among unselected strains.

A highly inbred strain of Drosophila melanogaster was subdivided into 20 replicate sublines that were maintained independently with 10 pairs of randomly sampled parents per generation for 180 generations. The variance between lines in abdominal and sternopleural bristle number increased little after 100 generations, in contrast to the neutral expectation of a linear increase; and the covariances of line means between different generations declined with increasing number of generations apart, in contrast to the neutral expectation of constant covariance. Thus, under a neutral model, the estimates of mutational variance were lower than for previous estimates from the first 100 generations of subline divergence. An autoregressive model was fitted to the variance of line means that indicated strong natural selection. There is no single unequivocal explanation for the results. Possible and nonexclusive alternatives include stabilizing selection on bristle number and deleterious effects on fitness of bristle mutations. The inferred strengths of selection on both traits are too high for stabilizing selection alone, and the between-line variance did not continue to increase sufficiently for pleiotropy alone to account for the observations. A third potential explanation that does not invoke selection is duplicate epistasis between mutations affecting bristle number.