基于系谱和基因组信息估计荷斯坦青年母牛体重性状遗传参数

旨在设计利用不同信息来源的模型估计荷斯坦后备牛不同月龄体重性状的遗传参数。本研究于2014—2020年测定并收集了7 122头荷斯坦牛32 338条0~12月龄体重数据，分别利用系谱信息（linear mixed model with pedigree relationship matrix，LM_A）和系谱-基因组信息构建亲缘关系矩阵（linear mixed model with genotype-pedigree joint relationship matrix，LM_H），基于母体效应动物模型估计初生重，基于是否考虑初生重作为协变量的单性状动物模型估计2~12月龄各月龄体重遗传力，并利用双性状动物模型估计初生重与其它月龄体重的遗传相关。结果显示，对于初生重，根据赤池信息量准则（Akaike information criterion，AIC），LM_H方法的拟合程度显著优于LM_A方法，但两种方法估计的遗传参数相差不大：直接遗传力分别为0.30和0.32，母体遗传力分别为0.08和0.09，个体直接遗传效应和母体遗传效应遗传相关系数分别为-0.65和-0.64；对于2~12月龄体重，LM_A和LM_H两种方法估计的校正初生重后的各月龄体重遗传力分别为0.15~0.55和0.28~0.49，未校正初生重的各月龄体重遗传力分别为0.16~0.54和0.28~0.51。初生重与2、5月龄体重之间为高遗传相关（相关系数>0.6）。5月龄后，各月龄体重与初生重的遗传相关系数随着时间间隔的增加而减小。相较于LM_A，LM_H方法更稳定，AIC值较小（即拟合优度较大），遗传参数标准误较小。综上，采用LM_H方法估计目标性状可获得更准确、更稳定的遗传参数。本研究为建立中国荷斯坦牛生长性状基因组选择体系提供了理论依据。

Estimating Genetic Parameters for Body Weights using Pedigree and Genotype-pedigree based Approaches in Holstein Heifers

The aim of this study was to estimate genetic parameters of body weight (BW) at different ages in Holstein heifers using models considering different source of information. A data set including 32 338 BW records of 0-12 months old on 7 122 female Holstein cattle measured from 2019 to 2020 was collected. Variance components and parameters were estimated using linear mixed model with pedigree relationship matrix (LM_A) and genotype-pedigree joint relationship matrix (LM_H). For calf birth weight (CBW), animal models with maternal genetic effects were applied. The single-trait animal models with or without a CBW as a covariate were used for the estimation of heritability of monthly body weight from month 2 to 12. Genetic correlations between CAW and monthly BW were estimated using bivariate animal models. The results showed that models with LM_H had better goodness of fit based on Akaike information criterion (AIC), although similar parameters for CBW were obtained through LM_A and LM_H methods. For CBW, direct heritabilities estimated by LM_A and LM_H were 0.30 and 0.32, respectively, and maternal heritabilities were 0.08 and 0.09, respectively. The correlation coefficients between individual direct genetic effect and maternal genetic effect was antagonistic for LM_A (-0.65) and LM_H (-0.64). Monthly BW traits had moderate to high heritabilities. Estimates from LM_A and LM_H with model including CBW as a covariate ranged from 0.15 to 0.55 and 0.28 to 0.49, respectively, and heritabilities from LM_A and LM_H with model without CBW ranged from 0.16 to 0.54 and 0.28 to 0.51, respectively. BW at age of month 2 and 5 were highly genetically correlated with CBW (correlation coefficient was higher than 0.6). After 5 months old, genetic correlations between CBW and BW at different months were decreased with increasing time spans. Compared with LM_A, the LM_H method was more stable, with a smaller AIC value (that is, a larger goodness of fit) and a smaller standard error of the genetic parameters. In conclusion, using the LM_H method to estimate target traits can obtain more accurate and stable genetic parameters. This study provides a theoretical basis for the establishment of a genomic selection system for growth traits in Chinese Holstein cattle.