PREDICTION OF SIRE TRANSMITTING ABILITIES FOR HERD LIFE FROM TRANSMITTING ABILITIES FOR LINEAR TYPE TRAITS

Use of sire linear type trait transmitting abilities as indirect predictors of herd life transmitting abilities was investigated. Data consisted of 53,830 grade Holstein daughters of 617 sires evaluated during first lactation for 18 type traits and having the opportunity to survive to 72 mo. Two measures of herd life were used; true herd life was not adjusted for yield, and functional herd life was linearly adjusted for the cow's last lactation yield relative to that of herdmates. (Co)variances among the 20 traits were estimated via a multiple-trait REML algorithm. Heritabilities for linear traits ranged from .08 to .41. Heritability estimates for both herd life traits were .03. The genetic correlation between the two measures of herd life was .84. Weights to predict herd life from linear traits, estimated as a function of the sire (co)variance matrix among herd life measures and linear traits, were different for true and for functional herd life. Predictions of herd life from the weights are equivalent to those from a multiple-trait BLUP model with herd life as a correlated but unobserved trait. For die estimated parameters, indirect prediction of herd life from type had a maximum reliability (squared correlation of predicted and true transmitting ability) of .56, but it was more reliable than direct prediction with 75 or fewer effective progeny. Additionally, linear type data are available earlier than herd life data.