Journal of Dairy Science, Vol 74, Issue 7 2309-2316, Copyright
© 1991 by American Dairy Science Association
Comparison of genetic evaluations from animal model and modified
contemporary comparison
H. D. Norman, R. L. Powell and G. R. Wiggans
Animal Improvement Programs Laboratory, Agricultural Research Service,
USDA, Beltsville, MD 20705-2350.
Comparisons were made between characteristics of Modified Contemporary
Comparison and animal model evaluations with data available for
January 1989 USDA-DHIA genetic evaluations. The animal model system's
requirement that cows have a valid first lactation record resulted in
a decrease in cows and daughters included. New flexible comparison
groups were slightly larger for small herds and much smaller for
large herds, which resulted in overall smaller and more uniform-sized
comparison groups. Determining the optimal method of defining
management groups was not undertaken. Correlations between bull
evaluations from the two procedures ranged from .92 to .95 across
breeds. Increases in reliability over repeatability were substantial
for bulls with limited daughter information and small for widely used
bulls. Correlations between evaluations for cows born in 1985 were
.92 to .96, whereas those for cows born in 1980 (old enough to have
daughters affecting animal model evaluations) were lower (.90 to .93),
as expected. Reliabilities for cows were .02 to .05 higher than
repeatabilities. Cows with more daughters increased more in evaluation
and accuracy between the two procedures and were genetically
superior. Bulls and cows with more prior information, cows with
higher past evaluations, and Holstein bulls with higher past
evaluations tended to have larger increases in PTA. Genetic trend
estimates were different for the animal model, which resulted in
changes in evaluations of various magnitudes depending on breed, sex,
and birth year of animal.
