A comparison of the selection with linear programming and restricted best linear unbiased prediction for a closed herd breeding of swine

“Restricted best linear unbiased prediction” (RBLUP) with genetic gains of traits restricted by desired levels and “linear programming (LP)”, which maximize an aggregate genotype (H) under the proportinal restriction on genetic gains of traits, based on estimated breeding values (EBV) and frequency of mating were compared. And these selection methods with restricted genetic gains were compared with that for H based on multi-traits best linear unbiased prediction with an animal model (MBLUP). The traits considered were daily gain (DG) and backfat thickness (BF). Selection of the best ten from 30 boars in 1st generation of a closed herd of swine in Kumamoto was examined in alternative selection methods. The results obtained were summarised as follows: 1) The fixed effects were not estimated in RBLUP and RBLUP of breeding values were proportional in every boar. 2) Eighty percent of the boars selected by RBLUP, and 70% of the boars selected by MBLUP were selected also by LP. 3) The average of H in the boars selected by RBLUP and LP under the proportional resteiction on genetic gains of traits were respectively 55% and 71% of H in the ones selected by H based on MBLUP. From these results, it is concluded that the selection method with LP based on EBV is equivalent to RBLUP and is usuful for a closed herd in swine in case of need for restrictions on genetic gains of traits.