利用数量性状构建粳稻核心种质的方法比较

以250份粳稻为研究材料，根据7个数量性状的基因型预测值，研究粳稻核心种质构建的方法。采用2种遗传距离（欧氏距离和马氏距离）,8种聚类方法（最短距离法、最长距离法、中间距离法、重心法、类平均法、可变类平均法、可变法、离差平方和法），3种取样方法（随机取样法、偏离度取样法和优先取样法），在25%的取样比例下构建48个核心种质，以筛选出的最佳构建策略进一步比较6种不同取样比例（10%、15%、20%、25%、30%和35%）的构建效果以确定最适宜的取样比例。结果表明，在粳稻核心种质构建中，马氏距离优于欧氏距离。欧氏距离优先取样法下最短距离法构建的核心种质最优。马氏距离偏离度取样法构建的核心种质能较多保存原群体遗传变异。10%是最适宜的取样比例。

Study on Method of Constructing Core Collection of Japonica Based on Quantitative Traits

The predicted genotype value of seven quantitative traits from 250 Japonica was used to study the method of constructing Japonica core collection. A total of 48 Japonica core collections were constructed at 25% sampling proportion by two genetic distances, including Euclidean distance and Mahalanobis distance, combined with eight clustering methods including nearest neighbour method, furthest neighbour method, median method, centroid method, group average method, flexible group average methods, flexible method and sum of squares method, and with three sampling methods including random sampling, deviation sampling and preferred sampling. Using the best constructing method, the optimum sampling proportion was filtrated under six different sampling proportions(10%, 15%, 20%, 25%, 30% and 35%). The results showed that Mahalanobis distance was much better than Euclidean distance in constructing Japonica core collection. Euclidean distance and nearest neighbour method combined with preferred sampling is the most suitable method for constructing core collection. The construction of the core collection by Mahalanobis distance combined with deviation sampling can preserve more genetic variation of the original population. Ten percent was the optimum sampling proportion.