Results CMCR30. Directory UMM :Data Elmu:jurnal:L:Livestock Production Science:Vol67.Issue3.Jan2001:
268 B
. Heringstad et al. Livestock Production Science 67 2001 265 –272 Table 2
Summary statistics of the data set CM120FIX used for the bivariate analysis of mastitis and culling and to examine the effect of accounting for culling as a fixed effect on variance component estimates of clinical mastitis. The sampling period was from 15 days before to 120 days
after first calving, and the reference data set CM120 was defined as in Table 1
a
CM120FIX CM120
Number of records 549,995
514,028 Number of sires
2043 2043
Number of herd3year classes 253,371
246,960 Mean number of daughters per sire
269 252
Mean number of records per herd3year 2.2
2.1 Mastitis frequency
14.9 15.9
Year of calving 1978–1995
1978–1995
a
A total of 8.7 of the cows were culled before 120 days after first calving.
2
the first class and .32 months the last class, and the reasonably high h 5 0.033 and 0.034. It is also
other classes are in single months, M is the fixed interesting to note the heritability of 0.016 when the
j
effect of the jth month of calving in 12 classes, HY sampling period ended at the day of first calving.
k
is the fixed effect of the kth herd by year class, S is Heritability estimates for CM120, CM150 and
1
the random effect of the cth sire, and E is a
CM180, which correspond to the sampling periods
ijklm
random error term. used in Norway, Finland and Sweden, and Denmark,
An additional fixed effect was included in the were 0.037, 0.039 and 0.040, respectively. Heritabili-
univariate analysis of data set CM120FIX: C is the ty estimates transformed to the underlying scale were
fixed effect of culling or not before 120 days after fairly constant, between 0.084 and 0.092. Estimates
calving, in two classes 05not culled; 15culled. In of heritability from CMCR were slightly higher than
the bivariate analysis of clinical mastitis and culling, from CM on both the observable and underlying
and in the bivariate analyses of clinical mastitis from scale.
different sampling periods, the same model as in the Table 3 shows the estimated variance components
univariate analyses was assumed for both traits. for clinical mastitis CM and the mastitis variable
An additive relationship matrix containing the where the trait is defined on the basis of clinical
relationship between sires was included in the analy- mastitis and culling reason CMCR, for different
ses. Variance components for random effects were length of the sampling period. Both the sire and the
estimated with REML, using the program VCE4 residual variance components increase with increas-
Neumaier and Groeneveld, 1998. Heritability esti- ing sampling period. Sire variance increases rela-
mates were transformed from the observable 0 1 tively more than the residual variance, which results
scale to the assumed underlying scale using the in higher heritabilities. The genetic variance was
classical formula of Dempster and Lerner 1950. roughly twice as high for sampling periods up to 210
days CM210 and CMCR210 as for sampling periods ending 30 days after calving CM30 and