主成分分析法在枝花木奶果优树选择中的应用

Plus Tree Selection of Baccaurea ramiflora by Principal Component Analysis

采用5株优势木对比法,并利用主成分分析法对海南岛不同地区的野生枝花木奶果进行优树选择。变异系数分析结果表明:28个初选优树个体间的13项选优性状既存在显著差异性,又存在稳定性。总体来看,树木的健康状况和产量在不同个体间的表型变异大,而果实形态及其果肉物质含量则表型变异小,树形态的表型变异居中,具备了进行选优的丰富差异性,同时又存在稳定的特征。根据各性状的负荷值,第一主成分与总糖(0.892)、胸径树高比(0.631)存在较强的正相关,与总酸(-0.795)存在较强的负相关,且贡献率达到22.083%,明显高于第二主成分,因此选择胸径株高比、总糖以及总酸作为枝花木奶果表型因子选择的主要指标,即选优的标准。计算基于13个性状指标的候选优树的综合得分,得到28株初选优树的综合得分排序,选出优良单株共5株,依次分别为M5,M3,M1-3,M1,M5-1。

In this study,the comparative method based on 5 plus trees and principal component analysis were applied to select superior trees of wild Baccaurea ramiflora from different regions in Hainan Island.The results of coefficient of variation analysis showed that there were both significant differences and stabilities in 13 traits,for optimal selection,among 28 candidate trees to be identified.On the whole,the health status and yield of trees varied greatly between individuals,while the fruit morphology and the content of flesh material changed little,and the variation of tree form located midst that of the former two,which has rich diversities and stable characteristics.According to the load of each trait variable from the principal component analysis,the first principal component has significant positive correlation with total sugar content(0.892) and ratio of plant DBH(Diameter of breast height) to height(0.631),and has significant negative correlation with total acid content(-0.795).The contribution percentage of the first principal component is 22.083%,which is hence much more important than the second principal component.Therefore,ratio of plant DBH to height,total sugar content,and the total acid content identified merely by the first component is sufficient to be used as the main superior phenotype for optimal selection.The original values of 13 selected traits of candidate trees were standardized firstly,then the comprehensive scores based on these standardized trait values were calculated,and the order of 28 selected trees by the comprehensive scores were obtained.A total of 5 plus trees were identified in this study,which were respectively M5,M3,M1-3,M1,M5-1.