Published online by Cambridge University Press: 02 September 2010
The effect of increasing the interval between recordings from 1 month to 2 months was studied by analysis of milk, fat and protein yield records obtained in alternate monthly herd visits. The data comprised records on over 34 000 daughters of 40 proven and 705 young sires which were analysed by multivariate restricted maximum likelihood.
Heritability estimates of bimonthly test day records were similar to the average of the estimates for the two corresponding monthly test day records, as were genetic and phenotypic correlations between bimonthly test day and lactation records. A quartic regression removed most of the additional variance due to the wide range of days to first test and therefore the variance-covariance structure of bimonthly test day records could be predicted from that of monthly test day records.
The accuracies of prediction of breeding value for 305-day yields of milk, fat and protein using the sum of five bimonthly test day records and the sum of seven 6-weekly tests were estimated to be at least 0·97 and 0·99, respectively, of the accuracy of using 10 monthly test day records.
If milk yield were recorded daily, prediction of lactation fat or protein yield from the product of the sum of daily milk yields and the average fat or protein content would be more accurate than using yield and content records solely from test days.