中文摘要：

利用NTSYS软件对柚类33个品种65个形态学性状进行了品种Q型聚类及其性状R型聚类，采用Mantel相关性检测优化出29个性状为核心性状。结果表明，29个性状对33个柚类品种的聚类，可分成6个组群；29个性状聚类也可分成6组。主成分分析，前9个主成分累积贡献率达75％，根据前9个主成分与性状间的相关性，选出影响力较大18个性状，综合了柚类特有的果实因子、花器官因子、枝条因子、叶片因子，说明这些因子是柚类品种重要特征。试验结果将有利于今后更好地利用形态学指标评定柚类品种及根据分类结果更好地利用这些丰富育种材料的遗传背景。

英文摘要：

Q cluster analysis and R cluster analysis were used to examine the 65 characteristics of 33 pomelo cultivars based on the NTSYS 2.10e software, and 29 core characteristics were optimized from 65 characteristics with Mantel correspondence test. The results showed that, 33 pomelo cuhivars and 29 core characteristics all were divided into six groups with UPGMA analysis. The additive contributing rate of the first nine principal components （PC） amounts to 75%. According to the relativity between nine PC and characters, 18 major discriminating characteristics were selected, synthesizing pomelo fruit factors, flower factors, blanches and leaves factors. The results are beneficial for us to use various characters to evaluate pomelo cultivars and to use various target varieties to widen the genetic background of pomelo.