Calving data of North American Holsteins were analyzed using a linear sire
model and a threshold sire model to estimate fixed effects and sire predict
ed transmitting abilities (PTA) for twinning rate. The data were 1,324,678
births of 37,174 sires of cows from the National Association of Animal Bree
ders (NAAB) calving ease database. All twins were assumed to be dizygotic.
Triplets were not included in the analysis. The twinning rate was 5.02%. Th
e sire model included relationships among sires and fixed effects of herd-y
ears, season, parity, and sire groups. Different estimates of heritability
due to models and restrictions on data have been demonstrated. Heritability
estimates for the sire of the cow effect were 2.10% by the linear model an
alysis and 8.71% by the threshold model analysis. Sire PTA for twinning rat
e ranged from 1.6 to 8.0%. Mean incidence of twins increased from first to
fifth and later parities: 1.63, 5.22, 6.66, 7.19, and 7.19%, respectively.
Some changes with season were also evident. The incidence of twins was high
est in April to June, at 5.88%, and lowest in October to December, at 4.23%
. Sire group effects showed that sires born after 1990 had a higher inciden
ce of twins than sires born before 1990. The mean twinning rate for sires b
orn before 1980 was 4.55% compared with a mean of 5.58% for sires born afte
r 1991. Characteristics of individual sire evaluations are discussed. Ample
opportunity exists to change the incidence of-twins in the Holstein popula
tion. Sire selection can be used to reduce the incidence of twins and also
the increased cost of production associated with twins. From a national per
spective, a reduction in the incidence of twin births can result in a savin
g of as much as $55 million per year.