| 摘要: |
| 为构建蝴蝶兰品质综合评价体系及分级标准，2010—2012年对收集保存的213份蝴蝶兰资源主要品质性状进行观察统计，应用变异分析、主成分分析、层次分析、聚类分析和判别分析建立综合评价体系。 18项品质性状的平均变异系数为34.28%，花梗侧分枝数的变异系数最大，为128.45%，类型的变异系数最小，仅为9.71%。运用主成分分析将18项评价指标压缩成10个综合因子，并确定株幅、叶长、叶宽、花梗数、花梗侧分枝数、花梗长、花梗粗、花朵长度、花朵宽度、花瓣厚、花朵数、花朵间距、花色、感病指数、花香、类型等16项指标为蝴蝶兰品质评价指标。利用概率分布和层次分析建立了品质指标的分级标准及评分标准；通过K-均值聚类分析和判别分析建立了蝴蝶兰4个品质判别函数；并对原始样品进行了分类，其中90.1%已正确分类，88.3%在交叉验证中正确分类，因此建立的蝴蝶兰品质判别函数正确率较高；筛选出优级品种70个，占32.86%；良级品种19个，占8.92%；中级品种77个，占36.15%，差级品种47个，占22.07%。以上所采用的方法可从不同层面对蝴蝶兰资源进行评价分析，得出量化结果、评价标准和判别函数，全面、客观地评价蝴蝶兰品种的优劣，体现评价体系和分级标准的有效性。 |
| 关键词: 蝴蝶兰 品质性状 主成分分析 层次分析 聚类分析 判别分析 综合评价 |
| DOI：10.11841/j.issn.1077-4333.2017.08.11 |
| 投稿时间：2016-09-06 |
| 基金项目:广东省省级科技计划项目（2015B020231005，2016B070701014，2014A020208063，2016A090924006）；广东省科技基础条件建设项目（2013B060400032、2008A060207012）；东莞市科技计划项目（2012108101021） |
|
| Establishment of comprehensive evaluation system on quality traits in Phalaenopsis |
|
CHEN Heming, LV Fubing, LI Zuo, XIAO Wenfang, ZHU Genfa
|
| (Environmental Horticulture Research Institute/Guangdong Provincial Key Laboratory of Ornamental Plant Germplasm Innovation and Utilization, Guangdong Academy of Agricultural Sciences, Guangzhou 510640, China) |
| Abstract: |
| A comprehensive evaluation system and grading system were established in order to provide technological support for screening,cross breeding,development and utilization of Phalaenopsis germplasm resources.Main traits of 213 varieties were observed and counted in 2010—2012.A system was established by means of variation analysis,principal component analysis,analytical hierarchy process,cluster analysis and discriminant analysis.The results showed that:The average coefficient of variation of 18 quality traits was 34.28%,the pedicel lateral branch number was the largest and the coefficient of variation was 128.45%,and the smallest coefficient of variation of type was 9.71%.The 18 evaluation variables could be compressed into ten factors which determined 16 quality evaluation indicators e.g.plant width,leaf length,leaf width,peduncle number,pedicel lateral branch number,pedicel length,pedicel thick,flower length,flower wide,petals thickness,flower number,flower space,flower color,disease index,flower fragrance,type.Probability distribution and analyticalhierarchy process were used to grade the quality indicators.Four quality discriminant functions were established by K-means clustering analysis and discriminant analysis,and 90.1% of original samples had been classified correctly,88.3% of 90.1% original samples were correct classification in cross validation,resulting in high discriminant function of quality has high accuracy;Two hundred and thirteen varieties of Phalaenopsis were clustered into 4 grades by K-means cluster and there were 70 varieties graded excellent(32.86%),19 graded good(8.92%),77graded moderate(36.15%)and 47 graded poor(22.07%).The above methods analyzed and evaluated germplasm resources from different views,and gained the quantitative results,evaluation standard and discriminant function which could comprehensively and objectively evaluate the quality of Phalaenopsis,and reflect the effectiveness of the evaluation system and classification standard. |
| Key words: Phalaenopsis quality traits principal component analysis analytical hierarchy process cluster analysis discriminant analysis comprehensive evaluation |
