Classical and mixed-model analysis of an index selection experiment for fecundity in Drosophila melanogaster

A comparison of three family indices to increase number of pupas from 3-d lay in Drosophila melanogaster is reported. The three indices were Id, based exclusively on dam's information; Ihd, based additionally on dam's full-sibs' and half-sibs' information; and Ihs, which also included the sire's full-sibs' and half-sibs' records. Three lines, D, HD, and HS, were selected, each according to one of the three indices: Id, Ihd, or Ihs. Each line consisted of 10 sires and 60 dams. Each dam contributed one male offspring and three female offspring. Seven generations of selection were performed in a two-replicate experiment. The expected advance of Ihs and Ihd over Id in the seventh generation was .38 and .15 phenotypic standard deviations, respectively. The experiment was analyzed in terms of both classical ordinary least squares (OLS) and mixed-model methodology (MMM). The phenotypic trends in the first replicate were 4.06 +/- 1.74, 1.42 +/- 1.76, and .26 +/- 1.94 pupas per generation for Lines D, HD, and HS, respectively, and in the second replicate 8.52 +/- 2.40, 6.16 +/- 3.21, and 4.21 +/- 2.60, respectively. The genetic trends in the first replicate were 2.18 +/- .61, 1.08 +/- .67, and .24 +/- .68 pupas per generation and, in the second replicate, 2.51 +/- .60, .16 +/- .61, and .05 +/- .64 for Lines D, HD, and HS, respectively. Despite theoretical expectations, the Id index was consistently better than the more complex indices. Several explanations for these results are possible: 1) smaller than expected selection differential, 2) inbreeding depression, and 3) incorrect heritability used to construct the indices.(ABSTRACT TRUNCATED AT 250 WORDS)