Experimental comparison of selection methods to improve a non-linear trait in Tribolium

Summary Five lines of Tribolium castaneum (flour beetle) were selected in each of three, replicates for biomass, using quadratic (Q) and linear (L) selection indices, direct selection for biomass (B) and indirect selection for each component trait, i.e. pupal weight (P) and family size (F), respectively, for five generations. The highest response in biomass was obtained in the Q line using an exact quadratic index in which the genetic value for each trait was based on information from both traits. The expected results were that Q selection would not exceed L selection while the L method would be superior with respect to direct selection for biomass (B line); P selection would be better and F selection worse than B selection. Selection was effective for pupal weight in the Q, L, B and P lines, but was not effective for family size, although the Q line increased slightly. The lines had no linear response for family size. Selection for biomass was influenced more by family size than by pupal weight. It is advantageous to include terms such as squares and products in the selection index, even though the response to selection may have been non-linear because of the low heritability of family size and an intermediate optimum for this component trait.