基于主成分和SOM聚类分析的高粱品种萌发期抗旱性鉴定与分类

采用人工气候箱内培养皿培养,PEG-6000溶液模拟干旱胁迫环境,在萌发期以80、120、150和175 g L–1PEG-6000水溶液处理31个高粱品种,旨在根据高粱品种萌发期对不同干旱胁迫程度的响应,筛选出具有抗旱能力的高粱品种并探讨高粱萌发期抗旱性鉴定的方法。通过主成分分析法(PCA)和神经网络自组织映射(SOM)聚类分析法对各高粱品种进行抗旱性综合分析与评定。PCA结果表明,相对芽长、相对根长和相对萌发抗旱指数载荷量最大,将其作为萌发期高粱抗旱性筛选的主要评价指标,并对31个高粱品种抗旱性排序。通过SOM聚类分析将31个高粱品种按抗旱性强弱分为5类,吉杂305等4个品种为高度抗旱品种,HL5等4个品种为抗旱品种,辽杂10号等8个品种为中等抗旱品种,锦杂103等7个品种为干旱敏感品种,锦杂93等8个品种为高度干旱敏感品种。研究认为,相对芽长、相对根长和相对萌发抗旱指数等可以作为高粱品种抗旱性鉴定的重要指标;SOM聚类分析可作为品种抗旱性分类的重要方法。

Identification and Classification of Sorghum Cultivars for Drought Resistance during Germination Stage Based on Principal Components Analysis and Self Organizing Map Cluster Analysis

PEG-6000 was used to simulate drought stress environmentin artificial climate chamber for studying the effects of different concentrations of PEG-6000 (80, 120, 150, and 175 g L^–1) on the response of 31 sorghum cultivars at germination stage, with the objectives of screening drought-resistant cultivars and exploring methods suitable for assessingdrought-resistance of sorghum cultivars. Through principal component analysis (PCA) and Self-Organizing-Map (SOM) cluster analysis, the drought-resistance of 31 sorghum cultivars was evaluated. PCA showed that relative bud length, relative root length and drought-resistance index during germination had the maximum weight in values, and were used as the main indices for evaluating and ranking the drought-resistant capabilities of 31 sorghum cultivars. Five groups of drought-resistance capabilities of 31 sorghum cultivars were detected based on SOM cluster analysis: four cultivars including Jiza305 were highly drought-resistant, four cultivars including HL8 were drought-resistant, eight cultivars including Liaoza 10 were medium drought-resistant, seven cultivars including Jinza 103 were drought-sensitive and eight cultivars including Jinza93 were highly drought-sensitive. In conclusion, relative bud length, relative root length and germination drought-resistance index can be used as indices for identifying drought-resistance in sorghum, and SOM cluster analysis is an important method for classifying drought-resistance.