摘要
An analysis of a selection experiment was used to assess the impact of various animal model struc- tures on REML estimates of variance components. The analyses were carried out based on 162 d body mass （BM） of 1 287 animals from 21 paternal half-sib groups of Fenneropenaeus chinensis. Estimated breeding values （EBV） of BM of all individuals were estimated using eight statistical models （A, AB, ABC, ABDC, ABMFC, ABMDC, ABFDC and ABMFDC） and BLUP （best linear unbiased prediction）. These models were designed involving factors such as sex, spawn date as fixed effects, maternal genetic effects, full-sib family effects as random effects, mean BM of families at tagging and age at recording （covariate）. The results demonstrate the importance of correct interpretation of effects in the data set, particularly those that can influence resemblance between relatives. The data structure and the particular model that was applied markedly influenced the magnitude of variance component estimates. Models based on few effects obtained upward biased estimates of additive genetic variance. The accuracy of genetic parameters and breeding value es- timated by ABFDC model was higher than other models. The results imply that additive genetic direct value, full-sib family effects, and covariance effects besides sex and spawn date as fixed effects were very important for estimating genetic parameters and breeding value of body mass. This model had a heritability estimate of 162 d BM of 0.44. The comparison of the efficiency of selection based on breeding values or phenotypic value revealed great difference： average breeding value of the best 24 families selected by the 162 d BM breeding value and phenotype were 0.577 g and 0.366 g, respectively, representing a 36.57% higher efficiency in the former. In conclusion, selection based on breeding value was more effective than selection based on phenotypic value. Our results indicate that effects influencing the magnitude of estimates should be taken into account when estim
An analysis of a selection experiment was used to assess the impact of various animal model struc- tures on REML estimates of variance components. The analyses were carried out based on 162 d body mass （BM） of 1 287 animals from 21 paternal half-sib groups of Fenneropenaeus chinensis. Estimated breeding values （EBV） of BM of all individuals were estimated using eight statistical models （A, AB, ABC, ABDC, ABMFC, ABMDC, ABFDC and ABMFDC） and BLUP （best linear unbiased prediction）. These models were designed involving factors such as sex, spawn date as fixed effects, maternal genetic effects, full-sib family effects as random effects, mean BM of families at tagging and age at recording （covariate）. The results demonstrate the importance of correct interpretation of effects in the data set, particularly those that can influence resemblance between relatives. The data structure and the particular model that was applied markedly influenced the magnitude of variance component estimates. Models based on few effects obtained upward biased estimates of additive genetic variance. The accuracy of genetic parameters and breeding value es- timated by ABFDC model was higher than other models. The results imply that additive genetic direct value, full-sib family effects, and covariance effects besides sex and spawn date as fixed effects were very important for estimating genetic parameters and breeding value of body mass. This model had a heritability estimate of 162 d BM of 0.44. The comparison of the efficiency of selection based on breeding values or phenotypic value revealed great difference： average breeding value of the best 24 families selected by the 162 d BM breeding value and phenotype were 0.577 g and 0.366 g, respectively, representing a 36.57% higher efficiency in the former. In conclusion, selection based on breeding value was more effective than selection based on phenotypic value. Our results indicate that effects influencing the magnitude of estimates should be taken into account when estim
基金
The General Program of the National Natural Science Foundation of China under contract No.30871919
the National High Technology Research and Development Program of China (863 Program) under contract No.2006AA10A406
