为更好地评价和利用广西建兰〔Cymbidium ensifolium (Linn.) Sw.〕种质资源，以来自广西6个中国兰花（Cymbidium spp.）主要产区的248份建兰材料为研究对象，对其9个质量性状和16个数量性状进行多样性分析、主成分分析和聚类分析，并构建核心种质。结果表明：248份建兰材料9个质量性状的Shannon-Wiener多样性指数为0.114~1.800；16个数量性状的Shannon-Wiener多样性指数为1.672~2.054，变异系数为11.44%~29.55%。主成分分析结果显示：前6个主成分的累计方差贡献率为79.093%。基于系统聚类法将248份建兰材料划分为3组。在取样比例25%下，采取聚类结合评分的取样方法，备选核心种质的极差符合率和遗传多样性保留率为100.00%，变异系数变化率为140.83%。采用优先取样和遗传距离动态聚类，并结合综合评分筛选核心种质，最终构建含62份建兰材料的核心种质。综合以上研究结果，广西248份建兰材料表型多样性丰富，采用优先取样和遗传距离动态聚类，并结合综合评分的方法构建的核心种质能够代表原始种质的遗传多样性。

To better evaluate and utilize the germplasm resources of Cymbidium ensifolium (Linn.) Sw. in Guangxi, 248 C. ensifolium accessions from six major Cymbidium spp. production areas in Guangxi were taken as research materials, diversity analysis, principal component analysis, and cluster analysis were conducted on 9 qualitative traits and 16 quantitative traits, and a core collection was constructed. The results show that the Shannon-Wiener diversity indexes of the 9 qualitative traits are 0.114-1.800 among the 248 C. ensifolium accessions; the Shannon-Wiener diversity indexes of the 16 quantitative traits are 1.672-2.054, and the coefficients of variation are 11.44%-29.55%. The principal component analysis results show that the cumulative variance contribution rate of the first six principal components is 79.093%. Based on the hierarchical clustering method, the 248 C. ensifolium accessions are divided into three groups. Under a sampling ratio of 25%, using the  screening method of cluster combined with score, the  range coincidence rate and genetic diversity retention rate of  the alternative core collection are 100.00%, and the  coefficient of variation change rate is 140.83%. Using priority sampling and genetic distance dynamic clustering, combined with comprehensive score  to screen the core collection, and a core collection containing 62 C. ensifolium accessions is ultimately constructed. Based on the above research results, the 248 C. ensifolium accessions in Guangxi exhibit rich phenotypic diversity, and the core collection constructed by  priority sampling and genetic distance dynamic clustering, combined with comprehensive score, can represent the genetic diversity of the original collection.