Establishment and optimization of comprehensive evaluation model for seed and seed oil traits of Handeliodendron bodinieri

Abstract: Handeliodendron bodinieri was a rare and endangered endemic single species genus plant only in karst areas of southwestern China, and it was also an excellent species of rocky desertification control, furthermore, its seed oil had higher value in nutrition, health care and industry. However, due to the lack of recognition of the related traits of H. bodinieri germplasm resources, the planting and popularization of H. bodinieri was restricted to some extent. Therefore, it was necessary to scientifically evaluate seed and seed oil traits of H. bodinieri from different provenances in order to guide the selection of good provenance and breeding. To improve the accuracy of evaluation results and optimum selection evaluation method, a new comprehensive evaluation model was constructed in this study. Fourteen indices related to seed and seed oil of H. bodinieri including seed length, seed diameter, single seed weight, 1 000- grain weight of seed, seed kernel rate, seed oil content, unsaturated fatty acid compositions of seed oil and so on were selected and determined. Five provenances of Leye of Guangxi, Tianlin of Guangxi, Fengshan of Guangxi, Huanjiang of Guangxi and Dushan of Guizhou were evaluated and sorted by four single evaluation methods of principal component analysis, factor analysis, maximizing deviation method, entropy weight method. The evaluation results of four single methods above were evaluated and sorted by three combination evaluation methods of mean, Borda and Copeland. Final evaluation rank was determined after all these methods had passed the Kendall's coefficient of concordance test. Spearman rank correlation coefficient was used to analyze the correlation between the final evaluation rank and four single methods evaluation ranking, and the best single evaluation method was selected at last. The results are showed following: Although there are some differences in the ranking of provenance evaluation in four single evaluation methods, they have passed the Kendall's coefficient of concordance test; all the three combination methods evaluation rank values are the same. Final evaluation ranking of provenance from high to low is Fengshan of Guangxi, Leye of Guangxi, Dushan of Guizhou, Huanjiang of Guangxi and Tianlin of Guangxi. Spearman rank correlation coefficient between final evaluation ranking and evaluation ranking of principal component analysis, factor analysis, maximum deviation method and entropy weight method are 1.00, 0.90, 1.00 and 0.90 respectively, and all correlations have reached a significant level (P < 0.05); the weight value of 1 000 - grain weight of seed obtained by principal component analysis and entropy weight method is listed in the first place of all indices. Therefore, the comprehensive evaluation model based on four single evaluation methods and three combination evaluation methods can provide scientific and reasonable ranking results for the seed and seed oil traits of H. bodinieri in multiple provenances. From the perspective of ease, convenience and effectiveness, the entropy method is the optimal single evaluation method for seed and seed oil traits of H. bodinieri, the most important index in the evaluation is 1 000 - grain weight of seed, and seed and oil traits of H. bodinieri in Fengshan of Guangxi provenance are the best. The research results provide a theoretical basis for efficient utilization of germplasm resources of H. bodinieri, and it also provides a new idea for the comprehensive evaluation of seed and oil field.