Effects of Errors in Genetic Parameter Estimates on Accuracy of Selection Index for Relative Desired Genetic Changes in Pigs
Access this Article
Search this Article
The effects of errors in estimates of genetic parameters on genetic response to selection in pigs were investigated when the selection was based on selection index for relative desired genetic changes (DC-index). The effects of errors on gains based on economic selection index (Smith-Hazel index: SH-index) were also investigated to compare with DC-index. Information on individual phenotype of candidate for selection (SIC) and on family (SIF) of the candidate was used in selection index. Average daily gain (DG), backfat thickness (BF) and litter size at birth (LS) were considered as traits for breeding objective and selection criteria. The respective heritabilities of 0.3, 0.5 and 0.1 for DG, BF and LS, with genetic and phenotypic correlations between DG and BF of 0.2, and no correlations between LS and DG and BF were assumed. A distinction is drawn between the response [E(_??_H<sub>0</sub>)] possible with use of the optimum index based on exact parameters and that achieved [E(_??_ H<sub>1</sub>)] with an index which uses the estimates of the genetic parameters in SH-index. We also draw a distinction between the response [E(_??_<sub>g0i</sub>)] possible with use of the optimum index and that achieved [E(_??_<sub>g1i</sub>)] of each i<sup>th</sup> trait with an index which uses the estimates of the parameters in DC-index. The efficiencies of selection in SH-index and DC-index were defined as RH[=E(_??_H<sub>1</sub>)/E(_??_H<sub>0</sub>)] and RD[the minimum value of E(_??_<sub>g1i</sub>)/E(_??_<sub>g0i</sub>) in all selected traits], respectively. Errors in estimates of genetic correlation had smaller effect for R<sub>H</sub> and R<sub>D</sub> than errors in estimates of heritability per unit. Expected loss in efficiency in DC-index was more sensitive than SH-index to incorrect estimates of heritability or genetic correlation. Errors in phenotypic correlation scarcely affect both R<sub>H</sub> and R<sub>D</sub>. For response to selection, the effect of error in estimate of heritability for LS was the largest in three traits. If there is no confirmation of accurate estimates of genetic parameters (e. g. genetic parameters in literature are used in selection programs) and if absolute of relative desired change for LS per genetic standard deviation unit is larger than that for both DG and BF, a higher heritability for LS should be used to construct selection index.
- Nihon Chikusan Gakkaiho
Nihon Chikusan Gakkaiho 71(4), 363-370, 2000-07-25
Japanese Society of Animal Science