遗传 ›› 1995, Vol. 17 ›› Issue (6): 21-23.
张胜利; 杨秀文; 靳明武; 陈廷槐 ZHANG Sheng-Li;YANG Xiu-Wen;JIN Ming-Wu;CHEN Ting-Huai
摘要: 利用模拟方法比较了5种方差组分估计方法(ANOVA,Henderson-Ⅲ,MLMT,R EMLMT和MIVQUE)对遗传参数的估计效果。结果表明:REMLMT法在各种情况下均能得到较好的参数估值,估计的准确度相对较高;ANOVA方法在小群体和不均衡资料情况下估计效果最差。同时,群体含量和结构能影响各种方法的估计效果, 群体含量较小或资料来源的公畜数太少,将导致遗传参数的估计误差增大,准确性降低。
Abstract:Five methods for estimating components of (Co) variance,including ANOVA,Henderson-III,MLMT,REMLMT and MIVQUE were compared using computer simulation.The results showed that REMLMT could obtain more accurate estimates for all data sets,while the estimates obtained by ANOVA always had greater deviations from the true values,especial for small and well-unbalanced data sets.Also,the effects of estimation were dependent on structure of population size,accuracy of estimates would be decreased when number of sires was too few or data contained small number of animals.