Responses of selection for milk production and secondary traits were predicted in open nucleus schemes using a deterministic model. Secondary traits considered were: traits recorded during lactation (e.g. mastitis resistance; calving ease); traits recorded in the nucleus only (e.g. food intake); traits recorded early in life (e.g. growth rate); and traits recorded late in life (e.g. longevity). Also, genotype × environment interactions between nucleus and commercial herds and predictors of merit in juveniles were considered.
Extension of the breeding goal to include an uncorrelated secondary trait, which was recorded at each lactation, had the same heritability as milk production (assumed throughout to be 0·25) and half its economic value, increased total economic gain by a factor of 0·12. This increase was only 0·04, if the heritability of the secondary trait was 0·1. The situation for traits of low heritability was not improved by progeny testing of young bulls due to the short optimized generation intervals. Gain increased only by a factor of 0·04, if the economic value was 0·25.
Including a secondary trait of heritability 0·25 and a genetic correlation with yield of 0·5 in the index, only increased economic response rates by a factor of 0·04. However, when the genetic correlation was –0·5 the benefits were greater with increases of 0·09, 0·10 and 0·22 for heritabilities of 0·05, 0·10 and 0·25, respectively. Hence, including traits with low heritability but with strong negative correlations with yield, which might apply to fertility and disease resistance, increased rates of gain moderately.
If an uncorrelated secondary trait was recorded in the nucleus only, e.g. food intake, and had half the economic value of milk production, total gains increased by a factor of 0·10. Hence, recording of secondary traits can be restricted to the nucleus with only minor loss of gain. The extra economic benefit was greatest from secondary traits measured early in life compared with late in life, e.g. longevity, with benefits increased by factors of 0·24 and 0·06, respectively.
Open nucleus schemes are robust in the presence of genotype × environment interactions between nucleus and commercial herds, if the breeding value estimation method accounts for these interactions, juvenile indicator traits of milk production may increase rates of gain by a factor of 0·11, if the heritability of the indicator trait is 0·25 and the correlation with milk production is 0·5.