：台灣蕹菜栽培品種(系)名稱甚為混雜，本研究主要在利用幼年期之營養性狀進行多變量分析，瞭解蕹菜收集品系之親緣關係。利用17個營養性狀做群集分析的結果可將41個收集品系分為四群，經主成份分析，前三個主成份共可解釋72.37％的變異量，蕹菜收集品系在三軸的排列結果與群集分析之分析結果相似。第一群(I)為子葉寬大、植株最高且直立、莖葉濃綠、葉呈長披針形，如：‘屏東尖葉’品系；第二群(II)為株高次之、莖葉淡綠、節間短、葉呈戟形或長卵形，如：‘白骨青葉99’品系；第三群(III)莖葉濃綠、葉呈卵形，如：‘竹母單選4號’品系；第四群(IV)為植株矮小、莖葉黃綠、葉呈心形，主要為‘白骨’系統群。10個‘白骨’收集品系之17個性狀經群集分析結果可明顯區分在II、IV群，前者傾向植株高、葉及株高之相對生長量小、節間短、節數多，後者傾向植株矮小、葉及株高之相對生長量大、節間長、節數少。經主成份分析，前三個主成份共可解釋88.70％ 的變異量，依主成份分析結果，選出本葉期株高、子葉長寬比、株高相對生長量及葉長相對生長量四個性狀，其群集分析結果與原來17個性狀的結果相似，顯示此四個性狀可做為性狀評估時之主要代表性狀，有利於品系分群。

The varieties (lines) of water spinach (Ipomoea aquatica Forsk.) were not clearly classified in Taiwan. The objective of this study was to identify the genetic relationship among the 41 water spinach accessions through numerical analysis of the vegetative characters in the juvenile stage. Using cluster analysis of the 17 vegetative characters, 41 accessions were divided into four main groups, and the accessions within each groups displayed diversity; besides, the first three groups clustered together before they clustered with the fourth group. The first group (I) included the accessions such as ‘Pyng Dong Da Yeh’ line with the morphology of big cotyledon (both width and length), tall and straight plant type, dark-green in color, and lanceolate leaf shape. Accessions in the second group (II) were plants shorter than those in the first group with light green color, short internodes, hastate or long ovate leaf shape, such as ‘Bair Ku Ching Yeh’ line. The plant type of the third group (III)asdark-green color, and ovate leaf shape, such as ‘Jwu Mu Dan’ line. The plants of the fourth group (IV) were shortest and yellow-green with cordate leaf shape, such as ‘Bair Ku’. The ordination pattern of the principal component analysis was similar to the grouping pattern of the cluster analysis, and the first three principal components derived the principal component analysis explained 72.37% of the observed variation. The 10 ‘Bair Ku’ accessions were distributed into two groups, II and IV. As we analyzed the 10 accessions by the principal component analysis on the basis of the 17 characters, the first three principal components derived from the principal component analysis explained 88.70% of the observed variation and four characters were selected as principle components. The four characters were shoot length, cotyledon length/cotyledon wide, relative shoot biomass and leaf length. The dendrogram derived from the four characters resembled that derived from 17 vegetative characters.