Heritability Estimates for First Lactation Milk Yield of Registered and Nonregistered Holstein Cows

Heritabilities of milk yield and fat test were estimated from first lactation records of 196,672 Holstein daughterdam pairs separately for registered and nonregistered cows for four milk yield groups and by years of freshening of the daughter. The model included herdseason and sire of the daughter. Heritability estimates were similar for registered and nonregistered cows averaging about .33 for milk yield and .62 for fat test. Estimates of heritability increased slightly over time for both milk yield and fat test. Heritability of fat test was similar for all groups. Heritability estimates for milk yield, however, averaged about .25 in low, about .35 in middle, and about .40 in high groups. I N T R O D U C T I O N Heritability estimates for milk yield generally have been larger at higher production than at lower production (3, 7, 8, 12). Estimates from daughter on dam regression usually are larger than from paternal half-sib correlation (7, 8, 12), probably partially due to reduction in the sire component of variance resulting from selection of sires and dams of bulls entering sire proving programs (2, 9). Therefore, estimates from daughter on dam regression may be more representative of genetic variation than estimates from sire components of variance unless nonadditive genetic or cytoplasmic effects are important (1, 11) or unless treatment of daughters is based on dams' performance. One criterion to consider when deciding whether to use only grade daughters to evaluate bulls for return to service is the heritability associated with grade (nonregistered) and registered cows. Similarly, any differences in heritability may be Received November 13, 1985. a criterion to use when choosing grade and registered cows as dams of bulls (10). The purpose of this study was to compare heritability estimates from daughter on dam regression for grade and registered cows at different milk yield groups and by year of freshening of the daughter. M A T E R I A L S A N D METHODS The data set has been described by Mirande and Van Vleck (7) and consists of first lactation, 2×, 305-d, mature equivalent (ME) records from the Northeast Dairy Records Processing Laboratory of 667,913 artificially sired Holstein cows. From these 196,672 daughter and dam matches were made on a within herd basis. A dam could be matched with more than one daughter, but each daughter and dam match would be included in a different yearly analysis. Each herd was assigned to one of four production groups for each year of the data from the rolling herd average for milk yield as of May of the year of freshening of the daughter. The rolling herd average is based on actual milk yield from all cows in the herd including first and later lactations. For each year the average and standard deviation of the rolling herd average was calculated from the complete data set. The production groups were assigned corresponding to three dividing points: the average minus one standard deviation, the average, and the average plus one standard deviation. These dividing points are given by Mirande (6). Heritability estimates were obtained from twice the residual covariance between daughter and dam records divided by the residual variance of dam records. Twice the residual covariance was estimated by subtracting the residual variances from both daughter and dam records from the residual variance obtained from analysis of the sum of daughter and dam records. The model used for both daughter and dam records was the herd-season (December through April and May through November) of freshening of the daughter and the sire of the 1986 J Dairy Sci 69:1652-1655 1652 PRODUCTION TECHNICAL NOTE 1653 daughter. This procedure allowed the same computer program used to estimate variance components to be used to estimate covariance components. The sire effect on the dam's record is a dummy effect included for computing convenience. The model assumes that dams having a daughter freshening, for example in herd-season 1, all freshen in the same herdseason. Therefore, the estimate of the residual variance for dam records may be slightly overestimated (4). Daughters were required to freshen at least 18 mo after their dams. Logarithmic and square root transformations of milk yield were also analyzed. Heritabilities are not reported, however, because they followed the same pattern by year and production group as for untransformed records, although the transformations resulted in slightly smaller heritabili ty estimates with estimates for the square root transformation being intermediate to heritabili ty estimates from untransformed and log transformed records (12). Estimates over years were averaged by weighting by the reciprocal of the squared standard error of the estimates. Examination of heritabili ty estimates plot ted by year led to averaging the estimates for milk yield for two 5-yr periods. Estimates summarized are for years 1973 through 1982. Earlier years had small degrees of freedom for residual variances, in most cases less than 100 for each product ion group by registry status combination.