樱桃番茄核心种质资源构建策略

为了有效确定樱桃番茄核心资源,促进樱桃番茄种质资源的创新利用,本研究采用混合线性模型对219份樱桃番茄种质的叶片长、叶片宽、第一花序节位、果实纵径、果实横径、果肉厚、单果重7个性状的基因型值进行预测;基于基因型预测值,按照30%的抽样率,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、重心法、中间距离法、类平均法、可变法、可变类平均法和离差平方和法）和3种抽样方法（优先抽样法、随机抽样法和偏离度抽样法）构建樱桃番茄核心种质资源,采用均值、方差、极差和变异系数评价不同方法构建樱桃番茄核心种质的优劣。欧氏距离和马氏距离法构建的樱桃番茄核心种质7个性状的方差均高于原群体,采用欧氏距离构建的核心种质7个性状的均值与原群体没有显著性差异,采用马氏距离构建的核心种质4个性状的均值与原群体没有显著性差异,3个性状的均值与原群体存在显著性差异,表明欧氏距离法优于马氏距离法。8种聚类方法构建的樱桃番茄核心种质7个性状的方差和变异系数均高于原群体,采用最短距离法构建的樱桃番茄核心种质具有较大的遗传变异,明显优于其他方法。采用3种抽样方法构建的樱桃番茄核心种质7个性状的变异系数和方差均有所提高,偏离度抽样法优于优先抽样和随机抽样法。基于欧氏距离、最短距离聚类和偏离度抽样法获取的65份樱桃番茄核心资源能够代表原群体的遗传多样性。

Construction Strategy of Core Germplasm of Cherry Tomato

In order to effectively identify the core germplasm and promote the innovative utilization of cherry tomato germplasm, the genotype values of seven traits (leaf length, leaf width, first inflorescence node, fruit length, fruit width, flesh thickness and weight per fruit) of 219 cherry tomato germplasms were predicted by a mixed linear model. Cherry tomato core germplasms were constructed based on genotype values by adopting two genetic distance (Euclidean distance and Mahalanobis distance), eight clustering methods (Single linkage, Complete linkage, Centroid method, Median method, Unweighted pair-group average, Flexible-beta method, Weighted pair-group average and Ward’s method) and three sampling methods (Preferred sampling, Random sampling and Deviation sampling) with 30% sampling proportion. The core germplasms constructed by different methods were evaluated based on mean, variance, range and variation coefficient. The variance of seven traits of cherry tomato core germplasm constructed by Euclidean distance and Mahalanobis distance were higher than that of the original population. By Euclidean distance, the means of seven traits had no significant difference between the core germplasms and the original population. By Mahalanobis distance, except four traits, the means of the other three traits were significantly different between the core germplasms and the original population. The result indicated that Euclidean distance was better than Mahalanobis distance. The variance and variation coefficient of seven traits of the core germplasms constructed by eight clustering methods were higher than that of the original population. The cherry tomato core germplasm constructed by the single linkage method had relatively larger genetic variation, which was better than the other clustering methods. The variance and variation coefficient of seven traits of the core germplasms constructed by three sampling methods were higher than that of the original population. The deviation sampling was superior to the preferred sampling and random sampling method. A total of 65 cherry tomato core germplasms were selected based on Euclidean distance, single linkage clustering and deviation sampling, which could represent the genetic diversity of the original population.