Modified Contemporary Comparison Research Articles

Merit of Outliers for Milk Yield as Indicators of Accuracy of Genetic Evaluations of Sires

To determine whether the accuracy of the genetic evaluations of sires can be assessed by the presence of extreme daughter records, we studied herd-years with records from first-crop daughters of 217 Holstein bulls that were not sampled by artificial insemination (AI) organizations but that later entered AI. The presence of outliers for standardized milk yield was determined within herd-year. Outliers were defined as records exceeding 1.5 interquartile ranges below the 25th percentile or above the 75th percentile. Herd-years were separated into two groups based on whether or not an outlier daughter record was present for an AI bull that had initially been sampled through non-AI. Herd-years without daughter outliers from those bulls were divided into herd-years with 1) no daughter outliers from any bull, 2) only negative daughter outliers from other bulls, 3) only positive daughter outliers from other bulls, or 4) negative and positive daughter outliers from other bulls. Herd-years with daughter outliers from AI bulls initially sampled through non-AI were divided into herd-years with 1) only negative daughter outliers, 2) only positive daughter outliers, 3) positive daughter outliers from those bulls and negative daughter outliers from other bulls, or 4) both negative and positive daughter outliers. The relationship between the frequency of outlier classes and a change in the Modified Contemporary Comparison genetic evaluations (the difference between the last available second-crop evaluation and the next to the last first-crop evaluation) was examined with logistic regression. For AI bulls that were initially sampled through non-AI and having evaluations that decreased ≥386kg, 9% of herd-years had positive first-crop daughter outliers and negative daughter outliers from other bulls; 38% had no outliers. For bulls with evaluations that increased ≥194kg, comparable percentages were 2 and 53%.

Genetic Merit and Usage Patterns of Bulls from Different Sampling Programs

Modified Contemporary Comparison and January 1991 animal model evaluations were used to categorize bulls. Bulls born from 1970 through 1981 were classified as AI- or non-AI-sampled and further divided as used (returned to service) or unused after first-crop evaluation. Bulls used heavily prior to proving were grouped separately. Groupings were effective because number of herds and daughters in first-crop and most recent animal model evaluation differed significantly across the five categories of bulls. Category of use, year of birth, and interaction were significant for PD milk from first-crop evaluation and PTA milk from January 1991 animal model. Least squares means from first crop evaluation were –161, –128, and 35kg for unused AI, heavily sampled AI, and unused non-AI bulls and 129 and 213kg for AI and non-AI bulls returned to service. Least squares means for January 1991 animal model evaluations were –262, –196, and –75kg for unused AI, heavily sampled AI, and unused non-AI bulls and 87 and 84kg for AI and non-AI bulls returned to service. Although non-AI-sampled bulls declined considerably relative to AI-sampled bulls, their average final genetic merit for milk was similar to that of AI-sampled bulls born in the same year. Risk of change in evaluations of non-AI-sampled bulls should be considered when determining extent of use.

Stability of Evaluations of Bulls Sampled by Artificial Insemination and Other Organizations

Changes in evaluations during first and second-crop periods were computed from up to 8 Modified Contemporary Comparison evaluations during each period for 1232 AI- and 217 non-AI-sampled Holstein bulls; repeatability at last Modified Contemporary Comparison evaluation was ≥.90. Evaluations were numbered consecutively within period. Effect of evaluation number within testing period on the bull's milk or fat evaluation was estimated with a model, including fixed effect of bull in addition to evaluation number. For subsets of AI- and non-AI-sampled bulls born from 1970 to 1975 and from 1976 to 1981, evaluations of both types of bulls increased from the initial first-crop evaluation and then remained fairly constant during the remainder of the first-crop period. With the inclusion of second-crop daughters, evaluations decreased significantly more for non-AI- than for AI-sampled bulls born from 1970 to 1975 (milk, 108 vs. 28kg; fat, 3.5 vs. 1kg). The reduction in evaluations was larger for non-AI-sampled bulls born after 1976 (milk, 184kg; fat, 7kg) but was unchanged for AI-sampled bulls. Last second-crop evaluation minus the second to last first-crop evaluation (difference 1) was calculated. Expected standard deviation of change for each difference of each bull was calculated and used to stratify bulls into eight change classes. Percentages of bulls in each class were compared. Relatively more non-AI- than Al-sampled bulls had first-crop evaluations that decreased more than could be explained by chance alone. Results indicated that evaluations of non-AI-sampled bulls were less stable than evaluations of AI-sampled bulls and that stability of evaluations of non-AI-sampled bulls has diminished over time. It appears that daughters in the first-crop of many non-AI-sampled bulls have received some kind of preferential treatment.

Implementation of an Animal Model for Genetic Evaluation of Dairy Cattle in the United States

Procedures were developed for animal model evaluation of dairy cattle for milk, fat, and protein yields as well as evaluation for fat and protein percentages as a function of yield evaluations. Data available for January and July 1987 Modified Contemporary Comparison evaluations were used to compute animal model evaluations. For each cow, records from her first five lactations are included. Cows without a first lactation record are evaluated separately. Relationships with all known male and female relatives influence evaluations. Solutions are computed by iteration on data for fixed management and random herd-sire, permanent environment, and breeding value effects. Iteration is by second-order Jacobi for breeding value and by Gauss-Seidel for other effects. Accuracy (expressed as repeatability) is computed as in Modified Contemporary Comparison procedures except that sire repeatability includes contributions from parents and dam repeatability is adjusted for number of daughters. A genetic base is imposed by subtracting average evaluation of cows born in 1984 from all evaluations. Results include predicted transmitting ability (one-half breeding value), predicted producing ability (sum of herd-sire, permanent environmental, and breeding value effects), parent average (average transmitting ability of parents), and average effect of management groups containing a cow's records.

What Extension Workers Need to Tell Dairy Farmers

The animal model is a method of estimating breeding values of dairy animals that uses the individual animal as the basis of calculations. Previous methods (the Modified Contemporary Comparison and other sire evaluations) have used performance of progeny groups as the basis of comparisons. Evaluation of females has been secondary, using results of sire evaluations. With an animal model, cows and sires are evaluated simultaneously with each evaluation impacting evaluation of all relatives. Other differences associated with adopting an animal model include a new genetic base and automatic adjustment for genetic merit of mates. Changes in sire evaluation will likely be small for widely sampled bulls, except for a lowering of all evaluations due to use of a more recent genetic base. Substantial changes in evaluations of some individual cows are likely.

Trends in Estimated Transmitting Ability from Sire Evaluations Based on Different Lactations

Most recently calculated sire evaluations through the January 1984 Modified Contemporary Comparison for first, all, and later lactations were available on 770 Ayrshire, 3188 Guernsey, 29,665 Holstein, 3551 Jersey, and 988 Brown Swiss bulls. A subset of 6006 Holsteins entering artificial insemination service before 49 mo of age was also examined. Averages for evaluations from first, all, and later lactations by year of birth of sire were studied. Regression of evaluation on year of birth showed two distinct slopes for each breed, with slopes changing in 1968 for Jerseys, 1969 for Holsteins, 1971 for artificial insemination Holsteins, 1972 for Ayrshires and Guernseys, and 1974 for Brown Swiss. Annual change was greater in more recent years for each group of lactations and ranged from 25 kg/yr for later lactations in Ayrshires to 55 kg/yr for later lactations in all Holsteins. Linear contrasts showed significantly greater annual change in transmitting ability for later than first lactations in both groups of Holsteins, while first lactation showed greater change in Ayrshires. Differences in slopes for first vs. all, first vs. later, or all vs. later lactations were not significant in other breeds.

Relationships Between Sires and Sons for Evaluations from First, All, and Later Records

January 1984 Modified Contemporary Comparison sire evaluations for first, all, and later records were used to create sire-son pairs in five dairy breeds. Each evaluation for each bull included 10 or more daughters. Regressions of son on sire for evaluations for first, all, or later records exceeded the expectation of .5 and were most divergent for Jerseys and Brown Swiss. Regressions and coefficients of determination were generally highest when sire's evaluation based on all records predicted sons’ evaluations from first, all, or later records. Regression of son on sire for difference in evaluations (later minus first) was positive in all breeds, ranging from .08 to .10, and was significant for all breeds except Brown Swiss. Coefficients of determination were low (.01 to .02). Both regression of son on sire for difference and coefficient of determination increased with son's Repeatability, with a regression of .33 and a coefficient of determination of .09 for sons over 99% Repeatability, Regressions for 8,055 Holsteins with evaluations for sires and maternal grandsires on those ancestors were .46 to .54 for sires and .25 to .31 for maternal grandsires. Regressions of son's evaluation on pedigree index were .94 for first, 1.05 for all, and 1.12 for later lactations.

Changes of Evaluation for Natural-Service Sampled Bulls Brought into Artificial Insemination

United States Department of Agriculture Sire Summary files were used to select bulls with a published Modified Contemporary Comparison sire evaluation prior to entry into artificial insemination. Canadian bulls were not included. Of the Brown Swiss, Guernsey, Holstein, and Jersey bulls that entered artificial insemination since 1974, 28 Brown Swiss, 19 Guernseys, 298 Holsteins, and 45 Jerseys (4 to 12%) had natural service evaluations. Of these bulls, 10 Guernseys, 154 Holsteins, and 28 Jerseys had an increase of Repeatability of 30% or more after entering artificial insemination. Evaluations for bulls just prior to their entering artificial insemination were compared with their most recent evaluations through July 1983. Average Repeatabilities for evaluations of bulls just prior to their entering artificial insemination were 32% for Guernseys, 32% for Holsteins, and 31% for Jerseys and 88, 90, and 87% for their most recent evaluations. Average Predicted Differences prior to entering artificial insemination were 305kg for Guernseys, 394kg for Holsteins, and 452kg for Jerseys for milk and 11, 10, and 16kg for fat. Most recent average Predicted Differences were 268, 380, and 532kg for milk and 7, 10, and 18kg for fat. Predicted Differences of natural service bulls were not overevaluated regardless of number of herds for sampling or year of entry into artificial insemination. Artificial insemination organizations can continue to acquire bulls with natural service evaluations calculated with the Modified Contemporary Comparison and be confident that these Predicted Differences are not overestimated.

Increasing Pedigree Contribution to Dairy Sire Evaluation

Methods to improve contributions of ancestor information to sire evaluations computed by the Modified Contemporary Comparison were studied. Pedigree index (one-half evaluation of sire plus one-fourth evaluation of maternal grandsire) was used to indicate ancestor merit with adjustment for type of sampling program and birth date. For Holsteins, effect of sampling through artificial insemination was 35kg for milk yield; effect of birth date (trend of daughter merit not accounted for by pedigree index or type of sampling) was 7.3kg/yr for milk yield and .36kg/yr for fat yield. Trends were larger for milk and fat yields of Ayrshires, Brown Swiss, Guernseys, and Jerseys except for fat yield of Ayrshires, which had no significant trend. Correlations of daughter merit with pedigree index and with present group average (average daughter merit of bulls with similar pedigree indexes for milk yield) were computed. For milk yield, correlations of daughter merit with pedigree index and with group average were nearly the same; however, for fat yield, pedigree index had a higher correlation with daughter merit than did group average. Accounting for residual trend resulted in further increase of correlation. Replacing one-fourth evaluation of maternal grandsire with one-half Cow Index of dam also improved correlations slightly. Evaluations can be modified so that pedigree index adjusted by birth date replaces group average in representing ancestor merit. Such evaluations better predict performance of future daughters than do evaluations with grouping by pedigree index for milk yield, particularly for evaluations for fat yield. Dam's Cow Index provides little additional information.

Estimation of Genetic Trends and Correlations for Jersey Cattle

Genetic correlations were estimated between sire evaluations for 14 type traits, 4 production traits, and stayability to 36, 48, 60, and 72 mo of age from type information and 469,460 lactation records on daughters of 3789 Jersey sires.Most correlated genetically with stayability were production traits with estimates that ranged from .44 to 1.01. Of the stayability measures, stayability at 48 mo of age had the highest correlation with production and other stayability measures.Modified Contemporary Comparison and Northeast Artificial Insemination Sire Comparison sire evaluations on daughter records had correlations of .77 and .71 for milk and fat. The Northeast Artificial Insemination Sire Comparison was correlated with stayability whereas the Modified Contemporary Comparison was correlated with type evaluations.Estimates of yearly genetic means indicate substantial improvement in recent years for production and stayability. Animals freshening from 1969 to 1977 have improved 290kg of milk, 10.5kg of fat, and 1.3, 4.0, 4.5, and 3.1 units of stayability for 36-, 48-, 60-, and 72-mo stayability because of improved sires only. This is an annual improvement due to sires of 36kg of milk, 1.3kg of fat, and .2, .6, .8, and .6 units of stayability.

Modified Contemporary Comparison Sire Evaluations from First, All, and Later Lactations

Modified Contemporary Comparisons are calculated by the United States Department of Agriculture from first and later lactations, results of which form the published progeny evaluations for all lactations. Characteristics of such evaluations on 200 Holstein sires, each with more than 500 daughters in each data set in the Summer 1981 run, were studied. Predicted Differences for milk averaged 245, 286, and 292kg for first, all, and later lactation evaluations, with standard deviations of 271, 253, and 254kg. The average difference (later minus first) was 47kg (standard deviation, 149kg). Average absolute difference was 123kg (standard deviation, 97kg). Correlations were .95 between first and all, .84 between first and later, and .97 between all and later. The correlation between progeny test for first lactation and difference was –.38 (–.28 for absolute difference). Percent of first lactations culled was correlated –.49, –.47, and –.37 for first, all, and later evaluations. Important variables in prediction of later lactations were first, percent of firsts culled, and sire's age, with first alone producing a coefficient of determination of .71. Sire's age was correlated –.11, –.19, and –.23 with first, all, and later evaluations indicating possible differences in genetic trend for first and later progeny tests for these sires. Fat yield behaved similarly to milk except that correlations between progeny tests and sire's age or culling were less than for milk. Ranking of bulls for daughter milk yield in later lactation changed from first lactation.

Prediction of Son's Modified Contemporary Comparison from Pedigree Information1,2

Son's modified contemporary comparison Predicted Differences and modified contemporary deviations based on daughters’ first lactations only and on all lactations were predicted for milk and fat yield and fat percent from pedigree information. Pedigree records included Predicted Differences and modified contemporary deviations for male relatives and modified contemporary deviations and Cow Index on first available records for female relatives. For a son's dam, however, first two and first three available records also were used. Models were chosen by a stepwise procedure with stopping when there was a change in the multiple correlation coefficient squared of less than 1%. Minimum repeatabilities of 40 and 70% for sons were required also to form two data sets.Models for milk and fat included sire's and maternal grandsire's Predicted Differences and dam's Cow Index. Models for fat also included the modified contemporary deviation for the maternal grandsire. For fat percent, sire's Predicted Difference or modified contemporary deviation and dam's Cow Index were important. Multiple correlation coefficients ranged from .32 for fat percent and .31 for milk to .24 for fat. Two-variable models suggested that for milk and fat, maternal grandsire was more important than dam whereas dam was more important for fat percent. The dam's Cow Index on first available record was more important for milk and fat, but dam's Cow Index from first three available records was most important for fat percent. Results were similar for predicting son's modified contemporary deviations. Closer prediction of son's Predicted Differences to real Predicted Differences for yield of milk and fat could be accomplished by placing slightly more emphasis on the maternal grandsire, and for fat percent, by putting more emphasis on dam's later records.

Trends in Genetic Merit of Dairy Sires at Initial Proving

Initial Modified Contemporary Comparison sire summaries from 1974 through 1978 were studied for 16,791 dairy sires of six breeds. These sires were grouped into five classes. Sires sampled through artificial insemination units were in one group; natural service sires further classified by number of herds (1, 2 to 3, 4 to 6, and 7+) in the initial (first published) proof were in the other four groups. Initial Repeatabilities of the progeny test for sires in these five classes were about 40, 18, 23, 29, and 35%.Genetic merit (pedigree indexes and Predicted Differences) was increasing over the 5 yr. Regressions of either measure of merit on year sampled averaged 19kg of milk and .6kg of fat. Bulls sampled through artificial insemination organizations were superior to natural service bulls in pedigree index and initial Predicted Difference. Genetic merit of widely sampled natural service sires showed no consistent advantage over natural service sires sampled in single herds.The initial sire summaries were compared with the most recent summary through 1978. There was a slight tendency for later summaries to average higher for all classes of sires. Sires sampled by artificial insemination organizations were more likely to increase in Predicted Difference.

Modified Contemporary Comparison as a Predictor for Future Daughters

Modified Contemporary Comparisons of sires from United States Department of Agriculture files were searched to locate bulls summarized in October, 1974, with substantial additional information in a subsequent sire summary. The summary of October, 1974, and the latest subsequent summary through January, 1976, on 702 Holstein bulls were used to develop a measure of daughter performance from information added between summaries on each sire. Predicted Differences of milk, fat, and 72 other variables from the October, 1974, summary were independent variables for prediction of performance of additional daughters for milk and fat. The best predictive equation for performance of additional daughters for milk included Predicted Difference for milk, fraction culled, and Repeatability times contemporary fathers’ Modified Contemporary Deviation for milk, Predicted Difference for milk alone accounted for 94.3% as much of the predictable variation as did the entire predictive equation. Analysis of fat production of additional daughters provided similar evidence for the predictive value of Predicted Difference for fat. Analysis of sire summaries of first lactation, including data through May, 1976, verified the results. Predicted Difference for milk and fat from multi-herd proofs appear to be sufficiently accurate predictors of performance of future daughters to serve as sole criterion of selection for production in breeding programs.

Relationships Between Bulls’ Pedigree Indexes and Daughter Performance in the Modified Contemporary Comparison

The United States Department of Agriculture Modified Contemporary Comparison method calculating bull Predicted Difference is influenced directly by the merit of the bull's male ancestors. In this study, the appropriateness of that influenced and the accuracy of pedigree indexes on bulls were investigated. The complete United States Department of Agriculture-Dairy Herd Improvement Association sires files were examined. Sire pedigree indexes calculated from Predicted Differences resulting from the Fall 1974 Modified Contemporary Comparisons were nearly as accurate as expected. Regressions of average daughter performance (Modified Contemporary Deviation) on pedigree index was near the expected 1.0. Correlations were reasonably close to the approximated expected values. Corresponding relationships between Predicted Difference and pedigree index gave similar results.

Bias of Sire Evaluations for Natural Service Sires and Age of Sire

In the Herdmate Comparison natural service sires often were overevaluated relative to sires in artificial insemination, and younger sires were underevaluated relative to older sires. The Modified Contemporary Comparison was designed to remove these biases. These sire evaluation methods were compared for their evaluation of artificial insemination versus natural service sires. In addition, the Modified Contemporary Comparison and Herdmate Comparison were compared with the Northeast Artificial Insemination Sire Comparison for their evaluation of sires of different ages. Sires from natural service were evaluated lower on the Modified Contemporary Comparison than on the Herdmate Comparison. Younger sires were evaluated higher on the Modified Contemporary Comparison than on the Herdmate Comparison. In both cases the Modified Contemporary Comparison values were intermediate to those of the Herdmate Comparison and Northeast Artificial Insemination Sire Comparison. The Modified Contemporary Comparison appears to remove a substantial portion of the biases favoring natural service and older sires in the Herdmate Comparison. Both the Modified Contemporary Comparison and the Northeast Artificial Insemination Sire Comparison show positive genetic trend.

Effects of Selection Bias on Sire Evaluation Procedures

Abstract Problems associated with selected records in methods of sire evaluation have been examined, and multiplicative factors for adjustment have been developed to incorporate later lactation records in sire evaluation procedures where a first record is missing. All Holstein records in the Dairy Herd Improvement Association files at Beltsville, Maryland, with freshening dates between 1960 and 1973 were used to quantify the amount of selection bias in later records that do not have first-lactation information available. Differences for lactation groupings show the amount of bias from selection in these records. The model included an overall mean, effects due to the herd-year-season of freshening, and opportunity groups that were formed depending on the number of later lactations for individual cows. These multiplicative factors can be used to adjust approximately 47% of cows in the files that do not have first-lactation information. These records can be incorporated, if needed to increase accuracy or increase comparisons, in the Best Linear Unbiased Prediction method or the Modified Contemporary Comparison method of sire evaluation.

Plasma hormone levels in 6-methyl-17-acetoxyprogesterone and estradiol benzoate treated heifers: Hansel, W., R. J. Schechter, P. V. Malven, K. R. Simmons, D. L. Black, A. J. Hackett & R. R. Saatman: J. Anim. Sci., [formula omitted](4): 671–681, 1975. Cornell University, Ithaca, New York, NY 14853, USA

United States Department of Agriculture Sire Summary files were used to select bulls with a published Modified Contemporary Comparison sire evaluation prior to entry into artificial insemination. Canadian bulls were not included. Of the Brown Swiss, Guernsey, Holstein, and Jersey bulls that entered artificial insemination since 1974, 28 Brown Swiss, 19 Guernseys, 298 Holsteins, and 45 Jerseys (4 to 12%) had natural service evaluations. Of these bulls, 10 Guernseys, 154 Holsteins, and 28 Jerseys had an increase of Repeatability of 30% or more after entering artificial insemination. Evaluations for bulls just prior to their entering artificial insemination were compared with their most recent evaluations through July 1983. Average Repeatabilities for evaluations of bulls just prior to their entering artificial insemination were 32% for Guernseys, 32% for Holsteins, and 31% for Jerseys and 88, 90, and 87% for their most recent evaluations. Average Predicted Differences prior to entering artificial insemination were 305 kg for Guernseys, 394 kg for Holsteins, and 452 kg for Jerseys for milk and 11, 10, and 16 kg for fat. Most recent average Predicted Differences were 268, 380, and 532 kg for milk and 7, 10, and 18 kg for fat. Predicted Differences of natural service bulls were not overevaluated regardless of number of herds for sampling or year of entry into artificial insemination. Artificial insemination organizations can continue to acquire bulls with natural service evaluations calculated with the Modified Contemporary Comparison and be confident that these Predicted Differences are not overestimated.