基于主成分与聚类分析的菜用大豆品质综合评价

为提高菜用大豆品种品质性状的选择效率，优化菜用大豆品质评价体系，以江苏省主栽的18 个菜用大豆品种为试材，分析其物理特性指标（荚长、荚宽、荚厚、百荚质量、百粒质量、L*、a*、b*、硬度、水分含量）和化学特性指标（VC、叶绿素、粗脂肪、淀粉、可溶性糖、可溶性蛋白、异黄酮含量，脂氧合酶（lipoxygenase，LOX）活力）。结果表明：品质特性中叶绿素和异黄酮含量、LOX活力、淀粉含量、百粒质量、百荚质量、a*在品种间变异系数较大，而L*和水分含量变异系数较小，其余指标变化均不显著。主成分分析表明18 项指标反映的菜用大豆品质可用7 个主成分来表示（累计贡献率达92.332 9%）。进一步根据聚类分析结果，筛选出百粒质量以及叶绿素、VC、可溶性糖、粗脂肪、异黄酮含量，硬度和a*这8 个品质指标代替原有的18 项指标，为品质评价指标体系的简化提供了可能。18 个菜用大豆品种中，徐豆17号综合品质最好，其次是区凡2号和新大粒1号，苏豆8号综合品质最差。

Comprehensive Evaluation of Vegetable Soybean Quality by Principal Component Analysis and Cluster Analysis

In order to improve the selection efficiency of vegetable soybean and optimize the quality evaluation system,
eighteen leading cultivars of vegetable soybean in Jiangsu province were tested for their physical indexes (pod length, pod
width, pod thick, hundred pods weight, kernel weight, L*, a*, b*, hardness, and moisture content) and chemical indexes
(VC, chlorophyll, crude fat, starch, soluble sugar, soluble protein, isoflavone, and LOX activity). The results showed
that chlorophyll content, LOX activity, isoflavone content, starch content, kernel weight, hundred-pod weight and a* had
greater coefficients of variation among species, while L* and moisture content had smaller coefficients of variation, and
the remaining parameters were not significant. Principal component analysis showed that eighteen indicators reflecting the
quality of vegetable soybean might be represented by seven principal components (cumulative contribution rate of 92.332 9%).
According to the clustering results, eight quality parameters including kernel weight, chlorophyll, VC, soluble sugar, crude
fat, isoflavone, hardness and a* selected from seven categories of principal components could substitute the original eighteen
indicators so as to simplify the evaluation indicator system. Among eighteen soybean varieties, the comprehensive quality of
Xudou No.17 was the best, closely followed by Qufan No. 2 and Xindali No. 1, while Sudou No. 8 was the worst.