A Comparison of Dairy Sire Progeny Tests Made at Special Danish Testing Stations with Tests Made in Farmer Herds

The data consisted of the first-lactation milk and butterfat records of 5,454 daughters of 305 Red Danish Milkrace sires tested at the Danish bull testing stations and the first test-year milk and fat records of 3,270 daughters of 110 of these same sires tested in farmer herds. All records were expressed in kilograms. For the test station data, the components of variance associated with differences between sires within stations and years were 229 for butterfat and 107,652 for milk. The corresponding components for field data after removing herd differences were 71 and 40,540. After expressing each daughter's record as a deviation from its contemporary herd average, the corresponding components were 53 and 32,125. The within-sire components for butterfat were 1,280, 970, and 916, respectively, for the three methods, and 541,696, 513,697, and 527,477, respectively, for milk. The heritabilities resulting from the three methods, assuming that the sire component is wholly genetic, are 0.66, 0.29, and 0.23 for milk and 0.61, 0.27, and 0.22 for butterfat, respectively. One might expect the estimates of heritability obtained from test station data to be somewhat higher than those obtained from field data, but not nearly so much higher as was found in this study. It is concluded that the estimates based on test station data are higher mainly because environmental differences at the same test station in the same year are confounded with progeny groups, thus putting much environmental variance in the sire components and increasing them more than the within-sire components. The estimates of the genetic correlation between station tests and field tests were 0.68 for milk and 0.75 for butterfat. On the other hand, the genetic correlations between independent field tests were 0.94 and 0.92 for milk and butterfat, respectively. This suggests that there is either a large interaction between sires and level of management or that the between-sire components from test station data are inflated with environmental differences. The latter cause is a more logical explanation, as the present data and previous data indicate that the interaction between sires and herds is small. For a constant number of daughters per sire of five or more, the regression of a future daughter in farmer herds on a test in farmer herds exceeds the regression of a future daughter in farmer herds on a test at the test stations. This is true for both milk and butterfat. As the number of daughters per sire