Mt. Skorupski et al., ESTIMATES OF GENETIC-PARAMETERS FOR PRODUCTION AND REPRODUCTION TRAITS IN 3 BREEDS OF PIGS, New Zealand Journal of Agricultural Research, 39(3), 1996, pp. 387-395
Restricted Maximum Likelihood (REML) procedures for multiple trait ani
mal models were used to estimate heritabilities, genetic correlations,
and common environmental effects for average daily gain (ADG), backfa
t thickness (BF), and number of pigs born alive per litter (NEA). Data
included 5561 litter records and 38622 ADC and BF individual performa
nce records for on-farm-tested Large White, Landrace, and Duroc pigs f
ed ad libitum from three New Zealand nucleus herds recorded over the p
eriod 1980-93. A bivariate animal model for ADG and BF contained fixed
effects for herd-year-season (HYS) of test, sex, and age as a linear
covariable, as well as random litter and animal effects. The NEA model
included fixed season of farrowing and parity effects, and random ani
mal (sow) and permanent environmental effects. Repeated records for NE
A were accommodated by fitting a permanent environmental effect, uncor
related to additive genetic effects, for each sow. The estimates of he
ritability (h(2)) for ADG were 0.20, 0.18, and 0.16, and the estimates
of the litter variance in proportion to the phenotypic variance (c(2)
) were 0.11, 0.12, and 0.09 for Large White, Landrace, and Duroc breed
s, respectively. The h(2) estimates for BF were 0.44, 0.45, and 0.46 f
or Large White, Landrace, and Duroc breed, respectively, and proportio
nate c(2) estimates were 0.06 for all breeds. The phenotypic, genetic,
and litter correlations between ADG and BF ranged from 0.32 to 0.54.
The h(2) estimates for NEA were 0.13, 0.09, and 0.16, permanent enviro
nmental variance ratios (m(2)) were 0.06, 0.05, and 0.05, and repeatab
ility estimates (t) were 0.19, 0.14, and 0.21 for Large White, Landrac
e, acid Duroc breeds, respectively. Correlations found in this study b
etween ADG and BF indicate that selection to improve one trait may be
associated with unfavourable change in the other trait. Therefore, a m
ultiple trait selection procedure such as a selection index is require
d to accommodate these antagonistic associations between traits.