GENETIC EVALUATION OF LACTATION YIELD FROM TEST DAY RECORDS ON INCOMPLETE LACTATION

Genetic prediction of heifer 305-day lactation yield from complete tes t day records or from records in progress was investigated. The accura cies of genetic indices to predict breeding value for total yield from all 10 test day records of milk, fat and protein yields were 0.71, 0. 66 and 0.67, respectively. These accuracies were slightly higher than if indices were computed to predict phenotypes and from these breeding values of 305-day records. The accuracy for a repeatability model (gi ving equal weight to each record) was not far below that of an optimal index. Inclusion of records in progress in genetic evaluation was inv estigated using a repeatability model and a phenotypic index to predic t the phenotype for complete lactation from test day records. Approxim ate expansion factors to equate the genetic variance of part records t o that of complete records and weights to give to part records in gene tic evaluation using an animal model were derived. For genetic predict ion of heifer lactation yield from test day records, a repeatability m odel giving equal weight to each record could be used without increasi ng computational facilities and could be implemented directly in curre nt genetic evaluation in the UK. Records in progress could easily be i ncorporated.