GENETIC VARIABILITY AND SIRE EVALUATION FOR
ADDITIVE AND NONADDITIVE PREWEANING GROWTH GENETIC EFFECTS IN AN ANGUS-BRAHMAN MULTIBREED POPULATION
M.A. Elzo, D.L. Wakeman and W.P. Dixon
University of Florida, Gainesville, FL 32611
SUMMARY
Additive and nonadditive direct and maternal covariances for birth and
weaning weights were estimated using records of 1581 straightbred and crossbred
calves from an Angus-Brahman multibreed herd.
Covariances were estimated using Restricted Maximum Likelihood and a Generalized
Expectation-Maximization algorithm applied to multibreed populations. Additive and nonadditive direct and maternal
genetic predictions were computed for all bulls after convergence. Estimates of genetic covariances (and their
ratios to phenotypic covariances) were within acceptable ranges. Bull additive direct and maternal genetic
predictions were more similar across bulls than nonadditive ones. Nonadditive maternal genetic predictions
were consistently larger than nonadditive direct for both birth and weaning
weights.
Keywords: Beef
cattle, crossbreeding, genetic evaluation, growth, variance components
INTRODUCTION
Although most beef in the U.S.A. is produced by crossbred cattle, prediction
of genetic values is still largely an additive genetic affair. Comparisons of sires across breeds are
currently based on a table of correction factors computed using experimental
data from the Meat Animal Research Center (Notter and Cundiff 1991). The aim of these correction factors is to
account for group additive and
nonadditive (heterosis) genetic effects. Ideally, a multibreed national sire evaluation that uses field
data, and accounts for group and random additive and nonadditive genetic
effects should be implemented. To
contribute to this goal, an experimental Angus-Brahman multibreed herd was
formed at the University of Florida in 1988.
One of the main purposes of this herd was to gather information on
reproduction, growth, and carcass traits that would help validate genetic
evaluation procedures that account for additive and nonadditive direct and
maternal genetic effects in multibreed populations. This study reports estimates of genetic parameters and sire
evaluations for additive and
nonadditive direct and maternal genetic effects for birth weight and weaning
weight in the Angus-Brahman herd using multibreed genetic evaluation
procedures.
MATERIAL AND METHODS
Data and management. Birth weight (BW) and weaning weight (WW)
records from 1581 straightbred and crossbred calves born between 1989 and 1996
in the Angus-Brahman multibreed herd of the University of Florida were used to
estimate covariance components and to predict sire genetic values. These calves were the product of a diallel
mating strategy involving 16 Angus (A), 20 Brahman (B), 11 3A :B, 10 2A 2B, 11 :A 3B, and 17 Brangus (eA dB) sires mated to 124 A, 160 B, 78 :A 3B, 128 2A 2B, 68 3A :B, and 94 Brangus dams. Cows were maintained on bahiagrass (Paspalum
notatum) pastures, with only mineral supplementation, except in winter,
when they were supplemented with bermudagrass (Cynodon dactylon) hay,
urea, and molasses. Cows were
synchronized in March, artificially inseminated twice, then exposed to a cleanup
bull for 60 days. Calves were born from
late December to March, and weaned in September and October (Odenya et al.
1992).
Covariance component estimation and genetic evaluation procedures.
Variance and covariance components were estimated by Restricted Maximum
Likelihood procedures (Harville 1977) that used a Generalized
Expectation-Maximization (GEM) algorithm (Dempster et al. 1977) applied
to multibreed populations (MREMLEM, Elzo, 1994). Computations were carried out using an in-house FORTRAN program
compiled using XL FORTRAN for AIX, and run in an IBM RS6000 workstation, model
580. To ensure that estimates of
covariance matrices were positive definite, the MREMLEM procedure computed the
Cholesky elements of each covariance matrix first, and then each Cholesky
matrix was multiplied by its transpose to obtain the matrices of covariance
estimates (Elzo, 1996). A two-trait (BW
and WW) multibreed sire-maternal grandsire model was used. The fixed effects were contemporary group,
sex-of-calf H age-of-dam H
breed-group-of-dam interaction, and group regression effects due to intra- and
interbreed additive direct, intra (as a deviation from B) and interbreed
additive maternal, and interbreed intralocus nonadditive direct and maternal. The random effects were additive direct and maternal bull effect,
intralocus interbreed bull H
breed-group-of-dam regression effect, and residual. Additive and nonadditive relationships, and heterogeneity of
covariances were accounted for in the model.
Single-trait estimates of covariances for BW and WW, and zero
covariances between BW and WW were used as priors for the two-trait (BW and WW)
run. The convergence criterion was that
the ratio of the difference between the sum of squares of the absolute values
between two successive GEM iterations relative to the sum of squares of the
covariances of the previous GEM iteration was less than 10B4 in two consecutive GEM iterations.
Genetic predictions. After reaching convergence, the solutions to
the mixed model equations were computed one more time to obtain additive and
nonadditive direct and maternal expected progeny differences (EPD) for
bulls. Bull EPD for additive direct and
maternal genetic effects were computed as a weighted sum of their respective
direct and maternal additive intrabreed group, additive interbreed group, and
additive random genetic effects.
Similarly, nonadditive direct and maternal bull EPD were computed as the
sum of their direct and maternal nonadditive intrabreed intralocus group and
random genetic effects, both weighted for the probability of occurrence of
interbreed intralocus interactions in a specific mating type.
RESULTS AND DISCUSSION
Covariance component estimates. The MREMLEM estimates of the
intra- and interbreed additive, and interbreed nonadditive genetic covariances
are shown in Table 1, whereas Table 2 contains the estimates of intra- and
interbreed environmental covariances.
Convergence was achieved in 38.4
min after 17 iterations.
Table 1. Estimates
of additive and nonadditive genetic covariances for BW and WW in an
Angus-Brahman multibreed herd
|
|
Genetic covariances (kg2)
|
|
Trait pairA
|
Additive
intrabreed B
|
Additive intrabreed A
|
Additive interbreed AB
|
Nonadditive interbreed A/B
|
|
BWD, BWD
BWD, WWD
BWD, BWM
BWD, WWM
WWD, WWD
WWD, BWM
WWD, WWM
BWM, BWM
BWM, WWM
WWM, WWM
|
5.87
6.14
.02
-1.05
137.16
.62
-37.20
5.03
1.92
108.17
|
7.58
9.13
-.36
3.17
228.38
3.55
-38.36
5.53
2.16
149.00
|
1.60
5.89
-5.87
-17.43
21.81
-20.79
-58.57
28.42
90.50
720.94
|
5.49
4.55
.09
-.12
139.82
1.44
3.96
6.03
3.84
156.82
|
|
A D = direct; M
= maternal.
|
Table 2. Estimates
of environmental covariances for BW and WW in an Angus-Brahman multibreed herd
|
|
Environmental covariances (kg2)
|
|
Trait pair
|
Intrabreed A
|
Intrabreed B
|
Interbreed AB
|
|
BW, BW
BW, WW
WW, WW
|
16.86
21.95
344.03
|
18.97
22.90
392.51
|
8.38
10.58
15.13
|
Estimates of covariances were within acceptable ranges, except for the
additive interbreed maternal variance for WW, which appeared to have been
grossly overestimated. Additive
intrabreed covariances were less different for BW than for WW. Nonadditive genetic covariances were as
important as additive intrabreed covariances for both traits. Environmental
intrabreed covariances were similar for A and B. Also, as with additive covariances, interbreed environmental
covariances were smaller than intrabreed ones. Intrabreed heritabilities of BWD, WWD, BWM, and WWM, were .21,
.25, .18, and .20 for A, and .24, .31, .17, and .24 for B. Similarly, the ratios of nonadditive
interbreed variances to phenotypic variances(interactibilities) for BWD, WWD,
BWM, and WWM were .16, .18, .17, and .20.
Estimates of covariances, heritabilities, and interactibilities can be
computed for any crossbred group using the intrabreed and interbreed covariance
estimates of Tables 1 and 2. For example, the heritabilities for a backcross
to A are .18, .21, .32, and .37, and its
interactibilities are .08, .09, .09, and .11.
Genetic predictions. Genetic predictions for BW and WW additive
and nonadditive direct and maternal genetic effects showed similar patterns
across bulls. To illustrate these
patterns, Figure 1 shows the direct and maternal additive and nonadditive (when
mated to F2 dams) bull EPD for WW. Bulls were sorted by their additive EPD, within
breed groups. Thus, A bulls appear
first in Figure 1, then :A 3B, 2A 2B, 3A :B and Brangus, and lastly B bulls.
Brahman bulls had, in general,
higher expected progeny differences (EPD) for both direct and maternal
effects. The bulls with the smallest
maternal EPD were group 2 (:A 3B).
Nonadditive EPD were similar across bulls, and they were usually in the
same direction as additive EPD.
REFERENCES
Dempster, A.P., Laird, N.M. and Rubin, D.B.
(1977) J. Royal Stat. Soc. Ser. B 38:1-38.
Notter, D.R. and Cundiff, L.V. (1991) J.
Anim. Sci. 69:4763-4776.
Odenya, W.O., Elzo, M.A., Manrique, C.,
McDowell, L.R. and Wakeman, D.L. (1992) J. Anim. Sci. 70:2065-2071.
Elzo, M.A. (1994) J. Anim. Sci. 72:3055-3065.
Elzo, M.A. (1996) J. Anim. Sci. 74:317-328.
Harville, D.A. (1977) J. Am. Stat. Assoc. 72:320-340.
