REGRESSION MODEL

1
Relationship of Herd-Heritability with Sire
Misidenti?cation and Entry Into a Proven Sire
Lineup C. D. Dechow1, H. D. Norman2, and N. R.
Zwald3 1The Pennsylvania State University,
University Park, 2Animal Improvement Programs
Laboratory, Beltsville, MD, 3Alta Genetics, Inc.,
Watertown, WI.
T66
REGRESSION MODEL yijklm BDi bjAj HYSk
bdstateDl bsstateSl bsd(SSDm) bdherdFm
bsherdGm eijklm, y ME milk yield, ME fat
yield, ME protein yield, SCS, fat or protein
BD the fixed effect of breed bj
coefficient for fixed regression on age nested
within parity (A) HYS fixed effect of
herd-year-season bdstate coefficient for fixed
regression on dam record nested within state
(D) bsstate coefficient for fixed regression
on sire PTA nested within state (S) bsd
coefficient for fixed regression on the
interaction between sire PTA and herd
standard deviation (SD) bdherd coefficient for
random regression on dam record nested within
herd (F) bsherd coefficient for random
regression on sire PTA nested within herd (G) e
random residual bdherd and bsherd were
assumed to be correlated

• Figure 1. Correlations between principal
  components for herd heritability and sire
  misidentification rate for 230 herds
• Figure 2. Relationship between herd
  misidentification rate and a standardized
  principal component for all herd heritability
  measures for 230 herds

ABSTRACT The objectives of this study were to
estimate individual herd heritabilities for all
herds in a large dataset and to estimate the
relationship of individual herd heritability with
sire misidenti?cation rate. Milk, fat and protein
yield and somatic cell score (SCS) were extracted
from the national dairy database. Paternity
verification results from DNA marker analysis
were provided by Alta Genetics, Inc. for 160
herds and from Accelerated Genetics for 75 herds.
The number of cows tested per herd ranged from 3
to 274. Herd heritability was calculated with
daughter-dam regression and daughter sire
predicted transmitting ability (PTA) regression
using 7,084,953 records from 20,920 herds. Herd
heritabilities were estimated with regression
models in ASREML that included fixed breed, age
within parity, herd-year-season of calving, dam
records nested within state, and sire PTA within
state random regression coef?cients were dam
records and sire PTA within herd. Average
daughter-dam herd heritability estimates ranged
from 0.26 (SCS) to 0.73 (protein percent),
whereas average daughter-sire herd heritability
ranged from 0.11 for SCS to 0.42 for protein
percent. Correlations between herd heritability
and sire misidenti?cation rate ranged from -0.28
to -0.43. The correlation between a principal
component for all measures of herd heritability
and sire misidenti?cation rate was -0.50. Higher
herd heritabilities were associated with lower
sire misidentification rates and individual herd
heritability estimates could be used to identify
progeny test herds that are candidates for parent
verification with DNA marker analysis. INTRODUCTI
ON - Individual herd heritabilities can be
generated with daughter dam and daughter
sire PTA regression (Dechow Norman, 2007) -
Misidentification reduces heritability estimates
(Van Vleck, 1970) - Individual herd
heritabilities could help identify herds with
poor sire identification OBJECTIVES - Estimat
e individual herd heritability for all herds in
the national database - Determine the
relationship between individual herd heritability
and sire misidentification rate DATA
EDITS - 7,084,953 records for six
traits - ME Milk, ME Fat, ME protein, SCS, Fat
, Protein - Lactations 1 through
5 - Calving between August 2000 and August
2005 - Identified sire with 50 reliability
for PTAM METHODS - Dam records were pre-
adjusted for age, parity and herd-year-season of
calving - Herd phenotypic standard deviation
for each trait was determined - Herd
heritability was generated for all herds
simultaneously with a regression model - Three
principal components of heritability were
generated 1. First principal component of
daughter-sire heritability estimates 2. Fi
rst principal component of daughter-dam
heritability estimates 3. First principal
component of all heritability estimates - Herita
bility and principal components were merged with
DNA parent verification results for 230
herds - Herd heritability and misidentification
rate were used to develop a misidentification
rate prediction formula - Misidentification was
predicted for all 20,920 herds
DERIVATION OF HERD HERITABILITY
ESTIMATES Daughter-Dam Heritability 2
regression of daughter performance on dam
performance 2 bdstate
bdherd Daughter-Sire Heritability Daughter
sire PTA regression used to estimate
genetic variance (bsstate
bsd(SDm) bsherd)SDUS2 / R
SDm2 SDUS genetic
standard deviation assumed in US genetic
evaluations R Average sire PTA reliability
for the herd
RESULTS Table 1. Average daughter-dam and
daughter-sire heritability estimates for 20,920
herds Table 2.
Correlation of herd misidentification rate with
daughter-dam and daughter-sire
heritability for 230 herds1 1All
correlations significant at Plt0.001