基于主成分-聚类分析构建甜瓜幼苗耐冷性综合评价体系

摘要

The present study aimed to establish a system for evaluating chilling tolerance in melon seedlings and select the chilling-tolerance germplasms.Overall,19 melon germplasms with different genetic background were used as experimental materials and their seedlings were measured for 9 morphological indices and 8 physiological indices under low temperature stress.Multiple statistics analyses (e.g.,principal component analysis,cluster analysis,and regression analysis) were used to analyze the chilling-tolerance coefficients (α value) of the indices and evaluate the chilling tolerance of the melon germplasms.Obvious variation was observed with the α value of both morphological and physiological indices of the seedlings (coefficient of variation ＞10％) under low-temperature stress,particularly for the α value of physiological indices.By principal component analysis,the primary 17 indices were changed to 7 separate comprehensive indices with a cumulative contribution rate 84.64％.The comprehensive evaluation value (D value) for each melon germplasm was obtained by calculating their membership function values and then weighting the values with the contribution rate of the 7 principal components.According to the D values,all melon germplasms were clustered into 3 groups:chilling-tolerance (1 accession),medium chilling-tolerance (11 accessions),and chilling-sensitive (7 accessions).Finally,stepwise regression analysis was used to establish a mathematic model (forecast accuracy ＞93.0％) for evaluating the chilling tolerance of melon seedlings:D=0.048+0.048POD-0.119SOD+0.097PRO+0.042CRI+0.084RDW+0.206OFW.The evaluation system established was suitable for rapid examination and prediction of chilling tolerance of different melon germplasms and may also be widely used for screening chilling-tolerance germplasms and cultivar breeding.%为了构建甜瓜(Cucumis melo)种质耐冷性的评价体系并筛选耐冷种质,以19份遗传背景差异明显的甜瓜种质为实验材料,测定低温胁迫下幼苗的9个形态指标和8个生理指标,利用主成分分析、聚类分析及回归分析等多元统计方法对各指标的耐冷系数(α值)进行综合评价.结果表明,低温胁迫下甜瓜幼苗的形态和生理指标发生了明显的变化,其α值的变异系数均大于10％,生理指标的α值变化更为明显.利用主成分分析将原有的17个指标转换为7个独立的综合指标,其累计贡献率达84.64％,以此计算各种质的隶属函数值,并以主成分的贡献率进行加权,最终获得所有种质耐冷性的综合评价值(D值).基于D值的聚类分析将所有种质按耐冷性强弱划分为3类,其中Oujin为耐冷性最强的种质,Xujin1等1 1份种质具有中等耐冷性,Qiuxiang等7份种质耐冷性较弱.通过逐步回归分析建立了甜瓜幼苗耐冷性评价的数学模型:D=0.048+0.048POD-0.119SOD+0.097PRO+0.042CRI+0.084RDW+0.206OFW.模型的预测精度大于93.0％.该耐冷性评价体系可广泛用于不同甜瓜种质耐冷性的快速鉴定和预测.