BLUP法による増体率および飼料利用性に関する黒毛和種種雄牛の育種価推定

Sire breeding values which are expressed as the expected progeny difference (EPD) were estimated from records of growth performance and feed utilizing ability, collected on bulls (135 and 174 head), which were performance-tested at two testing stations, Tottori and Miyazaki, respectively from 1971 to 1976. They were the progeny of 14 and 23 sires, respectively. Prior to the estimation, the records were adjusted using correction factors for testing period, season and initial age. The Mixed Model Solution method, HENDERSON'S (1973) so-called BLUP, was used, including the inverse of the nunmerator relationship matrix, calculated by the technique of HENDERSON (1976), which accounts for genetic relationships among sires. The ratio of the error variance to the sire variance component, assumed as the population parameter, was obtained from a separate analysis using the data collected at all the public stations during the same period. Finally, the estimated breeding values were compared with either sire evaluation by BLUP ignoring the relationships among sires or evaluation by regular least-squares analysis. The results are as follows; 1) Average coefficients of relationship among sires in Tottori and Miyazaki were 14.0% and 3.0%, respectively. The former represented a highly related sub-population, while the latter represented an almost unrelated one. 2) The top three sires in Tottori and Miyazaki, based on EPD of average daily gain, were Shoei 1, Kikko and Yoshitoku, and The Eighteenth Akashi, Maetani and Hatsuei, respectively. 3) The rank correlations between average daily gain and feed efficiency, between average daily gain and final weight, between TDN conversion ratio and DCP conversion ratio were positively high (0.57-0.96). However, the correlation between average daily gain and ratio of roughage intake to total feed intake appeared to be negative (-0.25 and -0.14 in Tottori and Miyazaki, respectively). 4) The degree of association among the evaluating methods based on rank and simple correlation was very high (0.87-0.98) in Miyazaki with low relationships among sires. On the other hand, in Tottori with the high relationships rank and simple correlations for BLUP with and without the inverse numerator relationship matrix were lower, 0.73 and 0.81, respectively, and those for the BLUP with the inverse numerator relationship matrix and the least-squares method were the lowest, 0.56 and 0.54, respectively. 5) Variance of error of prediction by the method using the relationships among sires was not so different from that by the method ignoring the relationships. However, the variance of error by the regular least-squares analysis was twice as large as those by the BLUP method.