基于主成分和聚类分析的山东省区试小麦品种（系）品质的综合评价

摘要

Based on the principal component analysis and cluster analysis, we analyzed and comprehensively evaluated the wheat quality of 297 varieties participated in the regional test of Shandong province in 2008-2009 and 2009-2010. Three principal components were extracted for evaluating the overall wheat quality. The first principal component was protein quality factor (gluten index, sedimentation value and formation time, setting time). The second principal component was milling factor (hardness index, flour yield, water absorption, whiteness). The third principal component was protein quantitative factor (moisture content and grain protein content ). The cumulative variance contribution rates of the three principal components were 77%and 82%, respectively. The contribution rate of the first principal component factors were 34.453%and 36.291%in the two years, which indicated gluten index, sedimentation value and formation time, setting time were the main factors affecting the wheat quality. From the evaluation results of 96 varieties in 2009-2010 based on principal component analysis, we found Tainong7058, Tainong05428, Taishan4173, Shannong71 and so on had high quality score, outstanding comprehensive quality traits. While integrated with contribution rates of principal components, the eigenvalues size of different indicators and maneuverability, we proposed that gluten index, sedimentation value and hardness index could evaluate the wheat quality indirectly in the early breeding program. R-type analysis clustered 10 traits into four categories (flour whiteness into a separate category), in which indicators of three traits coincided with indicators of three components in principal components. Q-type analysis clustered 96 varieties in 2009-2010 based on principal component analysis, the indicators of 6 varieties in the group-Ⅲ were high, which agreed with the results of principal component analysis. That further validated the principal component analysis could be used for comprehensive evaluation of wheat varieties (lines) quality.%利用主成分和聚类分析方法，对2008~2009年度和2009~2010年度参加山东省区试的297个品种（系）的小麦品质进行了分析和综合评价。结果表明，评价小麦整体的品质指标，可以提取三个主成分，第1主成分为蛋白质量因子（含面筋指数、沉淀值、形成时间、稳定时间），第2主成分为磨粉因子（含硬度指数、出粉率、吸水量、白度），第3主成分为蛋白数量因子（含湿面筋含量和籽粒蛋白质含量）。三个主成分的累计方差贡献率两年分别为77％和82％，其中第1主成分的贡献率两年分别高达34.453％和36.291％，说明面筋指数、沉淀值、形成时间和稳定时间是影响小麦品质的主要因素。利用主成分分析评价小麦的综合品质，2009~2010年度96个样品中的泰农7058、05428、泰山4173、山农71等品种（系）得分较高，说明品质表现突出。同时综合各主成分的贡献率、不同指标的特征值大小和可操作性，提出在育种早期面筋指数、沉淀值和硬度指数可间接评价小麦品质。用R型聚类将10个品质性状聚为四类，其中三类性状（面粉白度另成一类）所包含的指标和主成分分析的三个主成分所包含的指标基本吻合。在主成分分析的基础上，对2009-2010年度96个样品进行了Q型聚类，其中第Ⅲ类群包括的6个品种（系）各类指标较高，结果和主成分综合评分中得出的品质较好的品种（系）结果一致，进一步验证了主成分分析可以用于小麦品种（系）品质的综合评价。