Recent Technology of Genetic Selection Indices in Sheep Breeding

In developed countries, selection indices is usually used in the genetic improvement of sheep. One of the most critical assiunptions used in constnicting multiple trait selection indices is that parameters such as genetic and phenotypic variances and covariances are known . without error. In practise, however, only estimates of these parame-ters are used, and these are subject to sampling error. The objective of the present paper is to calculate genetic selection indices in sheep breeding using the "bending" technique develoved by Gunawan (1986). In the final model, 11 production traits are used in the breeding objective and 10 characters in tho index. The results show that the accuracy of selection ranged fi'om about 0.62 to 0.80. The efficiencies of selection indices including all traits relative to single trait selection are also calculated. The results show that selection indices are always more cfficient than single trait selection. The efiiciency ranged from about 102% to 176%. The highest relative efiiciency is for traits with low heritability (h2=0.l4-0.20), with efficiencies of 176% and 139% respectively. For traits with moderate heritability (117-=0.2l-0.40) the efficiencies ranged from 115% to 132%, and for traits with high heritability (h2>0.40) efficiencies are between 102% and 107%. It is concluded that a selection index is very efficient if the trait has a low heritability but also relatively high correlations with characters of high heritability in the index.