ARS | Publication request: Effectiveness of Estimating Individual Herd Heritabilities Using Regression Techniques

Submitted to: Journal of Dairy Science
Publication Type: Abstract
Publication Acceptance Date: February 28, 2005
Publication Date: July 24, 2005
Citation: Dechow, C., Norman, H.D. 2005. Effectiveness of estimating individual herd heritabilities using regression techniques [abstract]. Journal of Dairy Science. 88(Suppl. 1):11.

Technical Abstract: The objectives of this study were to estimate heritabilities for individual herds using regression techniques and determine if incorporating individual herd heritability would increase accuracy of genetic evaluations. First lactation mature equivalent milk (MEM) was obtained from 64,350 first lactation cows in 45 herds with 500 or more cows. Individual herd heritability was estimated using REML (h2RML), daughter-dam regression (h2DD), and daughter-sire estimated breeding value regression (h2DS). Regression coefficients were estimated for each herd with the MIXED procedure of SAS. The model included fixed year-season and age at calving effects, either dam MEM (DM) or sire estimated breeding value (SEBV) as a random covariable, and random error. Animal models were also used to estimate heritability for each herd with ASREML. Heritability from daughter-dam regression was twice the regression of MEM on DM. Twice the regression of MEM on SEBV was multiplied by genetic variance of the US Holstein population and divided by herd phenotypic variance to estimate h2DS. Sex averaged heritability (h2SA) was the mean of h2DD and h2DS. Heritability estimates were constrained to range from 0.25 to 0.35 and were used to standardize records to a constant genetic variance across herds. Estimated breeding values (EBV) were generated with the adjusted records, which were weighted by the ratio of base error variance to herd error variance. Average individual herd heritability estimates ranged from 0.28 for h2RML to 0.31 for h2DD. Correlations of h2RML with h2DD, h2DS, and h2SA were 0.53, 0.42 and 0.61, respectively. Correlations among sire EBV generated with the adjusted records and official sire PTAM from November 2004 national genetic evaluations ranged from 0.007 to 0.012 higher than the correlation when using non-adjusted records. Correlations between yield deviations and parent averages were lower (range -0.026 to -0.002) after adjustment for heritability, except adjustments for h2DD (0.006 higher). Individual herd heritability estimates may improve accuracy of national genetic evaluations, or help identify progeny test herds with poor parent identification.