1. Effectiveness of genetic evaluations in predicting daughter performance in individual herds
H. D. Norman 1, J. R. Wright 1*, C. D. Dechow 2 and R. C. Goodling, Jr. 2
1 Animal Improvement Programs Laboratory, Agricultural Research Service, USDA, Beltsville, MD
2 Pennsylvania State University, University Park, PA
Abstr.
W90
2010
INTRODUCTION
Response to genetic selection has been well documented on a population basis
Dairy producers are more likely to appreciate its value if it is demonstrated in their own herd
Prediction on an individual herd level should be a useful extension tool
RESULTS
Means and standard deviations (SD) of individual herd regression coefficients (b) for Holstein and Jersey herds
RESULTS (cont.)
RESULTS (cont.)
Means and standard deviations of individual Holstein herd regression coefficients (b) by number of records per herd
Comparison of distribution of regression coefficients (b) from predicting 1st parity ME Milk from Parent BV Milk for Holstein herds with high or low levels of usable id and production
Trait Holstein Jersey
ME Milk ± ± 0.42
ME Fat ± ± 0.39
ME Protein ± ± 0.43
ME SCS ± ± 0.69
Days open ± ± 1.12
PL ± ± 0.80
Means
Trait >1000
ME Milk
ME Fat
ME Protein
ME SCS
Days Open PL
Standard Deviations
ME Milk
ME Fat
ME Protein
ME SCS
Days Open PL
Herds , , ,
Percent of herds
OBJECTIVE
To illustrate genetic progress can be shown on an individual herd basis.
Distribution of herd regression coefficients (b) from predicting 1st parity ME Milk from Parent BV Milk
DATA
First lactation cows calving
Required to have sire PTA and dam PTA
Cow’s phenotypic record was matched with sire PTA and dam PTA from evaluation prior to cow’s first calving
Herds were required to have a minimum of 50 cows meeting above criteria
8,257 Holstein herds; 441 Jersey herds
Regression Coefficient (b)
Percent of herds
Standard deviations of the regression coefficients decreased as number of records per herd increased
CONCLUSIONS
Standard deviations were large but response to genetic selection on a within-herd basis can be demonstrated
Increasing number of records per herd did not improve predictive ability, but did increase the uniformity of the observed response
Response was greater in higher production herds
Response was also greater in herds with more complete identification
Illustration of individual herd response should increase confidence in national genetic evaluations in well managed herds
Regression Coefficient (b)
Means of individual Holstein herd regression coefficients (b) by herd production level (kg)
Means of individual Holstein herd regression coefficients (b) by percent of herd with usable identification (id)
METHODS
SAS GLM for each herd using the model:
Phenotypic trait=
Parent Breeding Value (BV)
(Sire PTA + Dam PTA) of same trait
+ Cow’s calving year
+ Error
Traits predicted:
1st lactation Mature Equivalent (ME) Milk, Fat, Protein, and Somatic Cell Score (SCS)
1st lactation Days open (DO) (using Parent BV Daughter Pregnancy Rate * -4)
Productive life (PL)
10, , ,000-
Trait <10, , , ,999 ≥13,000
ME Milk
ME Fat
ME Protein
ME SCS
Days Open PL
Herds: , , , ,143
Trait <
ME Milk
ME Fat
ME Protein
ME SCS
Days Open PL
Herds , , , ,489
Standard deviations of the regression coefficients (not shown) had little change across groups for milk, fat and protein (0.42 to 0.48) but decreased as production increased for SCS, Days open, and PL
Standard deviations of the regression coefficients (not shown) increased slightly as percent usable herd identification increased