Components of heterosis for growth traits and litter size in line crosses of mice after long-term selection

Strains selected for 54 generations for large 8-week weight (N8, S8), for small weight (N6, S6), for high 3–5-week gain and low 8-week weight (N1) as well as unselected randomly mated control strains (N9, S9) were crossed, and F ₁, F ₂ and both back crosses created. N strains are derived from a NMRI base, S strains from a synthetic four-way cross among inbred lines. Heterosis for 8-week weight (8-ww) was some 6%, but 30% for 3–5-week gain and it was on average negative (22%) for 5–8-week gain. When epistatic effects were taken into account, it appeared that heterosis for 8-ww was largely due to beneficial effects of non-parental gene combinations, i.e. additive × additive epistatic effects were negative. These non-parental combinations neutralized the mostly negative dominance effects. The latter were mostly positive for early gain but in all 9 line crosses negative for late gain. The digenic effects of the non-parental origin were not large enough to balance the negative dominance effects for late gain, so that heterosis was negative. It appears therefore that early and late gain are affected by different sets of genes. Maternal additive effects on 8-ww and early gain are smaller than direct additive effects but mostly of similar sign. For late gain there appears to be no connection with direct additive effects. Direct litter heterosis was almost 50% but much of this was due to positively acting recombinational gene combinations, while direct litter dominance was mostly negative. Maternal additive effects on litter size differed depending on whether they were estimated from dams with crossbred or with purebred litter. Heterosis was small in crosses between control strains. Heterosis caused by recombinant gene pairs should be amenable to improvement by selection.