水稻品种区域试验的AMMI模型分析

在农作物品种区域试验中，品种在各地点的表现通常是不一致的，也就是说品种的基因型和环境（Ｇ×Ｅ）存在着交互作用。常用的Ｇ×Ｅ交互作用分析方法是相对于环境指数的线性回归方程，但这一方法约束性较强，解释Ｇ×Ｅ交互作用较少。加性主效应和乘积交互作用模型（简称ＡＭＭＩ模型）在常规的基因型和环境的加性模型中加入了乘积形式的交互作用，能更多地解释Ｇ×Ｅ交互作用。实例分析表明ＡＭＭＩ模型是品种区域试验结果分析更为有效的方法。

The Use of AMMI Model to Analyze Regional Rice Trial Data

In regional trial, genotypes usually perform differently, implying the existsitence of the interaction between genotype and environmemt. Linear models is the classical method to analyze the interaction by regressing each variety on environmental index. Because of linearity restriction this method sometimes does not capture enough information about interaction. The additive main effects and multiplicative interaction (AMMI) model is a prevailing new method to analyze the interaction by adding multiplicative interaction terms in the classical additive models, which can gain deeper insight about the interaction. One set of data generated from south China Rice regional trial in 1996 is used an example to show effectiveness of AMMI model in analyzing region trial data.