Abstract: 【Objective】Rapeseed is the most important oil crop in China, and it is mainly grown in south of the Yangtze River, that is heavily contaminated with aluminum (Al). Al toxicity is one of the important factors that limit rapeseed production in south China. In order to reduce and avoid the harms of Al toxicity to rapeseed, breeding Al-tolerant varieties is a cost-effective and practical way. Therefore, screening and identifying of Al-tolerant genotypes in rapeseed is significant. The main purpose of this paper was to explore the methods of Al tolerance evaluation, and to screen rapeseed genotypes with Al tolerance at seedling stage. 【Method】The Al tolerance of 23 rapeseed genotypes which are cultivated in large areas of Jiangxi Province was investigated by a pot culture. Principal component analysis, subordinate function analysis, clustering analysis and step regression analysis were used to evaluate the Al tolerance of rapeseed based on the Al-tolerance coefficients of morphological and physiological traits under the condition of Al stress and control. These traits included plant height (PH), root length (RL), rhizome thickness (RT), root dry weight (RDW), shoot dry weight (SDW), soluble sugar content in leaf (SSCL), soluble sugar content in root (SSCR), praline content in leaf (PCL), praline content in root (PCR) and SPAD. 【Result】There were significantly different responses of morphological and physiological indexes among genotypes under Al stress. Correlation analysis showed that Al tolerance coefficients had large or small correlation between traits, consequently the data provided were overlapped and only one single index could not accurately evaluate the Al tolerance of each rapeseed. Principal component analysis reduced the complex interrelationships between all indexes to four independent comprehensive components. The four comprehensive components could present 86.36% information of the raw data of the traits related to Al tolerance of rapeseed. The calculate membership function values were measured on the basis of the contribution rates of four comprehensive component values, and then according to the relative importance of the comprehensive components, these values were weighted to obtain the comprehensive evaluation of Al tolerance values (D value) for different genotypes. The cluster analysis was used to divide 23 genotypes into three kinds of Al tolerance type. Three varieties Zheyou50, Deyou5 and Xiangzayou2 were Al-resistant type. Thirteen varieties such as Zhongyouza12, Fengyou730 and so on were moderately Al-resistant type. Seven varieties such as Nanyouza1, Chuangzayou5 and so on were Al-sensitive type. To screen for the identification indexes of Al tolerance at seedling stage of rapeseed, and analyze the relationship between identification indexes and Al tolerance, a mathematical model of Al tolerance evaluation was built. The stepwise regression equation was established based on D value which was assigned as the dependent variable and the indexes of Al-tolerance coefficients which were assigned as independent variables. Al tolerance predictive value and D value of 23 rapeseed varieties were significantly correlated at seedling stage. The step regression analysis showed that plant height, root length, soluble sugar content in leaf and praline content in leaf significantly influenced the Al tolerance and could be the indexes to identify the Al tolerance in rapeseed. 【Conclusion】It is an effective way to comprehensively evaluate the Al tolerance of rapeseed by principal component analysis, clustering analysis and step regression analysis which could be used to screen genotypes of Al tolerance in rapeseed. It not only could avoid the bias and the instability of single index analysis, but also reflect the relationship between different Al-tolerance related traits and Al tolerance in rapeseed.

Key words: rapeseed, aluminum tolerance, principal component analysis, subordinate function, comprehensive evaluation