Lande R, Price T
Department of Ecology, University of Chicago, Illinois 60637.
Genetics. 1989 Aug;122(4):915-22. doi: 10.1093/genetics/122.4.915.
Additive genetic variances and covariances of quantitative characters are necessary to predict the evolutionary response of the mean phenotype vector in a population to natural or artificial selection. Standard formulas for estimating these parameters, from the resemblance between relatives in one or two characters at a time, are biased by natural selection on the parents and by maternal effects. We show how these biases can be removed using a multivariate analysis of offspring-parent regressions. A dynamic model of maternal effects demonstrates that, in addition to the phenotypic variance-covariance matrix of the characters, sufficient parameters for predicting the response of the mean phenotype vector to weak selection are the additive genetic variance-covariance matrix and a set of causal coefficients for maternal effects. These can be simultaneously estimated from offspring-parent regressions alone, in some cases just from the daughter-mother regressions, if all of the important selected and maternal characters have been measured and included in the analysis.
数量性状的加性遗传方差和协方差对于预测群体中平均表型向量对自然选择或人工选择的进化响应是必要的。通过一次分析一个或两个性状中亲属间的相似性来估计这些参数的标准公式，会受到对亲本的自然选择和母体效应的影响而产生偏差。我们展示了如何使用后代-亲本回归的多变量分析来消除这些偏差。一个母体效应的动态模型表明，除了性状的表型方差-协方差矩阵外，预测平均表型向量对弱选择的响应所需的足够参数是加性遗传方差-协方差矩阵和一组母体效应的因果系数。如果所有重要的选择性状和母体性状都已被测量并纳入分析，那么这些参数可以仅从后代-亲本回归中同时估计出来，在某些情况下仅从女儿-母亲回归中估计出来。
