An experimental comparison of individual, family and combination selection.

Two selection experiments comparing the relative efficiencies of individual, family, and combination selection were conducted. The expected results for larval weight of Tribolium (h(2) = 0.20) and for pupal weight (h(2) = 0.40) were that combination selection would be a more efficient method than family selection, and that family selection would exceed individual selection. In experiment I, individual selection produced more response (P < 0.05) than did combination or family, which was not in agreement with expectation. There was confounding of inbreeding levels and random drift due to differential effective population sizes in the lines selected by different methods. Experiment II consisted of ten single-generation selection tests. An advantage of this approach is that it eleminates the inherent problems of differential inbreeding levels and differential rates of genetic drift due to unequal population sizes among the methods of selection. There were no statistically significant differences in efficiency among the three methods of selection for both traits. This was contrary to theoretical expectations but does suggest that with traits of 20% h(2) or higher, and where feasible, one may be justified in basing selection decisions on the phenotype of the individual only. Other advantages of single generation testing are that it allows more precise testing of selection theory and unbiased standard errors for estimates of realized heritability.