Effects of constant and of changing diets on responses to selection were studied. Mice were fed maize meal and dried skim milk from 3 to 5 weeks of age and were selected for live-weight gains in that period. Gains in the generation prior to selection were 17·7 g on 76% maize (24% milk), 8·0 g on 16% maize and 5·5 g on 100% maize.
The Fixed optimum line was selected on 76% maize, the Milk line on 16% maize and the Maize line on 100% maize. The diets of three other lines were reviewed at each generation, and two diets were fed on each occasion. The diets of the Shifting optimum line straddled the optimum of the previous generation; those of the Milk-impelled line at first had a little more than the optimum amount of milk, but were made more milky as normal growth resumed; and the Maize-impelled line was similarly directed towards 100% maize. Back selected milk, optimum and maize sublines were taken off the three lines on constant diets. The Control line measured unimproved rates of growth on 16%, 52%, 76% and 100% maize.
Dietary effects were not always repeatable. Controls on the optimum diet, 76% maize, were the most stable across generations and with this diet, gains generally had the lowest phenotypic variance within generations —g2. With 100% maize, this variance was slightly larger—8 g2. The diet with the least stable effect was 16% maize; as Control gains rose from 8 to 15 g, phenotypic variance fell from 21 to 6 g2. On various diets between 88% and 96% maize, the variance of the Maize-impelled line rose from 6 to 34 g2 without commensurate changes in its mean gain.
Responses of generally about 5 g above controls were made in five generations of selection. The Maize line was exceptional in making no forward response. Heritability of growth between forward and back selected sublines was highest on good diets (0·49 on 76% maize) and lower on poor diets (0·20 on 16% maize and 0·28 on 100% maize).
The Fixed and Shifting optimum lines do not as yet differ on comparable diets; nor the Milk and Milk-impelled lines on 16% maize. The Maize-impelled line has not yet been tested on 100% maize and cannot be compared with the Maize line. Family-by-diet interactions were smaller than consistent family differences. It is concluded that genotypic improvements may be effective over a wide range of maize-milk diets.