Selection for and against a canalized phenotype inDrosophila melanogaster

Abstract

Selection for and against the canalized phenotype in scutellar bristles was attempted in two selection lines and a randomly selected line was used as control. The selection lines were the Decanalization line (D) and the Canalization line (N). The D line was maintained by matings of scute males (scwbl) with three scutellars with wild-type females (scwbl/yw) with five bristles, in the N line scute males with four bristles were mated with wild-type females also with four bristles, while in the C line males and females of the above genotypes were selected at random. The lines were established from a sample of flies taken from a line selected for high scutellar numbers.After eighteen generations of selection the C line was characterized by a regression of mean bristle number without appreciable change in variance. Relative to the N line, the D population showed a lower proportion of flies having four scutellars, a higher variance in bristle numbers, and a higher proportion of four-bristle scute flies having abnormal patterns.Two alternative hypotheses were advanced to account for the results of this experiment. The first postulated a relative change in the widths of the four-bristle canalization zones in the selection lines, while the second suggested a relative change in frequencies of specific modifier genes for scutellars in scute and in wild-type genotypes of the lines. The evidence favours the latter hypothesis.