基于多元统计法的黄皮果实品质评价指标的选择

Selection of indices for wampee fruit quality evaluation based on multivariate statistics

为科学地简化黄皮果实品质的评价指标,以22份不同黄皮种质为试材,采用主成分分析和系统聚类法对测得的果实单果重、果形指数、可食率、可溶性固形物含量、可滴定酸含量、总糖含量、维生素C含量等13项常规指标进行分析。主成分分析结果表明,这13项指标反映的黄皮果实品质可用3个主成分来表示(累计贡献率达84.8951%)。根据这3个主成分对13项品质指标的特征向量进行系统聚类,可将这些指标划分为5类。从这5类指标中分别选取单果重、果形指数、可滴定酸、总糖及糖酸比5个简化指标对黄皮果实品质进行聚类分析,在欧氏距离2.07时,可将供试的22份不同种质分为3类,这一结果与实际口感评价结果相吻合。

In order to simplify the indices for evaluating fruit quality without losing the information embodied in the original indices, 13 fruit quality indices including single fruit weight, fruit shape index, edible rate, total soluble solid content, titratable acid content, total sugar content, vitamin C etc. for 22 different wampee germplasm were analyzed by using multivariate statistics. Results of principal component analysis with the above 13 indices indicated that the cumulative proportion of former 3 principal components reach 84.8951%. By using systematic cluster analysis with eigen- vectors of the former 3 principal component, 5 evaluation factors for a comprehensive system were obtained, and the single fruit weight, fruit shape index, titratable acid, total sugar, fruit sugar acid ratio were selected as simplified indices. The cluster analysis of fruit quality in wampee by this 5 selected indices showed that the fruit quality from 22 different germplasm resources could be clustered into 3 groups according to their euclidean distance of 2.07, which were the same with that of their sensory evaluation.