ESTIMATION OF HERITABILITY AND REPEATABILITY OF MILK AND MILK FAT PRODUCTION WITH SELECTION EFFECTS REMOVED

Two groups of data were used for this study. One group contained 110,084 and the other group 106,296 lactations on registered and grade Holstein cows. All records were analyzed as deviations from regressed adjusted herd-year-season averages. Estimates of the relationship between various pairs of lactations made by using five different methods gave essentially the same results. It was concluded that regressions or intraclass correlations, where all records of the pair are included, are much easier to compute and give essentially the same results as the more complicated procedures of Maximum Likelihood and construction of variances and covariances to try to remove the effects of selection. Heritability estimates obtained by using paternal half-sib analyses of variance were much lower than daughter-dam regression estimates for one group of data, but they were in good agreement in the other group of data. Estimates of heritability obtained by intrasire, intraherd regression of daughter on dam were slightly lower that use of devated records does not remove all variations due to herds from a daughter-dam regression. The derivation of a method to estimate heritability and repeatability by removing the effects of selection is presented. © 1969, American Dairy Science Association. All rights reserved.