Modified contemporary comparisons for first and second lactations in the same and different herds

Separate sire evaluations by Modified Contemporary Comparison for first and second lactations were calculated for 200 widely used Holstein bulls from a file containing only first or second records of all daughters and from two independent subsets of the same data. Sire evaluations based on both first and second lactations also were computed. Correlations between first lactation evaluations in one independent data set and second lactation evaluations in the other were .87 and .84. First lactation evaluations accounted for only about 75% of the variation in second record evaluations, suggesting a need for some form of later record sire evaluation, at least for the selection of sires of sons. Within sire, daughter variation for first records was not useful in predicting difference between first and second record sire evaluations. Second lactation evaluations increased relative to firsts as culling increased, suggesting culling bias in Modified Contemporary Comparisons for later records. An adjustment based on an empirical regression of later records on firsts produced second evaluations with means and standard deviations nearly identical to those of first records. Correlations between first and adjusted second evaluations were .87 for both independent data sets. Difference between adjusted second and first evaluations were correlated .15 with culling, compared to .25 for unadjusted seconds. Important sire differences in first and second lactation evaluations were detected in all data sets. Such differences were unaltered by adjustment for selection.