Multi-environment models using marker-based kinship information for both additive and dominance effects can accurately predict hybrid performance in different environments. Sorghum is an important hybrid crop that is grown extensively in many subtropical and tropical regions including Northern NSW and Queensland in Australia. The highly varying weather patterns in the Australian summer months mean that sorghum hybrids exhibit a great deal of variation in yield between locations. To ultimately enable prediction of the outcome of crossing parental lines, both additive effects on yield performance and dominance interaction effects need to be characterised. This paper demonstrates that fitting a linear mixed model that includes both types of effects calculated using genetic markers in relationship matrices improves predictions. Genotype by environment interactions was investigated by comparing FA1 (single-factor analytic) and FA2 (two-factor analytic) structures. The G×E causes a change in hybrid rankings between trials with a difference of up to 25% of the hybrids in the top 10% of each trial. The prediction accuracies increased with the addition of the dominance term (over and above that achieved with an additive effect alone) by an average of 15% and a maximum of 60%. The percentage of dominance of the total genetic variance varied between trials with the trials with higher broad-sense heritability having the greater percentage of dominance. The inclusion of dominance in the factor analytic models improves the accuracy of the additive effects. Breeders selecting high yielding parents for crossing need to be aware of effects due to environment and dominance.

中文翻译：

---

使用加性和显性基因组关系对高粱育种试验进行多环境分析。

使用基于标记的亲属关系信息的多环境模型可以准确预测不同环境中的混合性能。高粱是一种重要的杂交作物，广泛种植在许多亚热带和热带地区，包括澳大利亚新南威尔士州北部和昆士兰州。澳大利亚夏季月份高度变化的天气模式意味着高粱杂交种在不同地区的产量差异很大。为了最终能够预测杂交亲本系的结果，需要表征对产量性能的加性效应和优势相互作用效应。本文表明，拟合一个线性混合模型，该模型包括使用关系矩阵中的遗传标记计算的两种类型的效应，可以改进预测。通过比较 FA1（单因素分析）和 FA2（双因素分析）结构来研究环境相互作用的基因型。G×E 会导致试验之间的杂交排名发生变化，每次试验的前 10% 的杂交中差异高达 25%。随着优势项的加入（超过仅通过加性效应实现的），预测准确度平均提高了 15%，最大提高了 60%。总遗传变异的显性百分比在试验之间有所不同，具有较高广义遗传力的试验具有更大的显性百分比。在因子分析模型中包含优势提高了加性效应的准确性。选择高产亲本进行杂交的育种者需要了解环境和优势造成的影响。