N. Mielenz et al., Effectiveness of genetic evaluation with transformed data by using dam-daughter pairs Japanese quails, ARCH TIER, 43(3), 2000, pp. 299-309
In estimation of variance components with REML it has been assumed that the
data were normally distributed. Egg production traits of poultry have been
shown to exhibit markedly non-normal distributions. In this study, six tra
its from an unselected quail line were analysed. The original data were tra
nsformed using the well-established power transformation to approach normal
ity. The genetic evaluation was carried out with a multiple-trait animal mo
del, based on transformed and untransformed data, respectively. Two traits
of laying performance, of egg weight and body weight were analysed simultan
eously. To compare the efficiency of breeding values the method of simulate
d selection with biological data (dam-daughter pairs) was used. To select t
he dams with intensities between 10% and 90% we used individual records and
BLUP-breeding values, estimated with transformed and untransformed data. T
he response of such a selection was estimated using the corresponding daugh
ter records. Only for the trait laying performance up to 200 days of life w
e could indicate an advantage of the transformation. It was shown, that by
changing from one trait to multiple-trait genetic evaluation non-normality
could be compensate. For 10% intensity the selection for individual laying
performance provided an unexpected high response in comparing with the BLUP
-method.