Application of multidimensional methods to separate varieties on their response to environment factors
Автор: Kharitonov E.M., Goncharova Yu.K., Ochkas N.A., Sheleg V.A., Bolyanova S.V.
Журнал: Сельскохозяйственная биология @agrobiology
Рубрика: Анализ и отбор генотипов
Статья в выпуске: 1 т.52, 2017 года.
Бесплатный доступ
Till now, the areas under rice crops are mostly occupied with the limited number of varieties. For enriching genetic biodiversity, it is necessary to improve selection of unique rice genotypes, and provide ecologically-based location of each variety. Now the efficiency of breeding is decreasing because of incomplete characterization of potentially donor genotypes. Presently, the domestic standards for competitive state trail do not cover a detailed study of the samples, since the developed varieties are tested at a single level of mineral nutrients with no estimation of a response to stressful influences and yield production sustainability. That leads to rejection of those highly productive samples for which such conditions are not optimal. In the present work we firstly summarized methods to comprehensively characterize adaptive plasticity of rice plants under contrast conditions (i.e. different dates for planting, various levels of mineral nutrition and stressors). In a multifactorial experiment with 19 combinations of the factors tested, we investigated yield variability in 24 Russian rice ( Oryza sativa L.) varieties. The samples were planted on April 15, May 15, or June 15 and grown at optimum (N120P60K60) and excess (N240P120K120) fertilizer rates, in thin and dense crops (200 or 300 plants per square meter, respectively), under artificial salinization (0.35 % NaCl added to the soil at tillering). The data were processed using cluster and discriminant analysis. The multidimensional statistical methods allow us to clasterize the varieties into four groups with the closest characteristics as influenced by the full set of studied factors, and then to allocate distinct factors for the most precise discrimination between the samples. A standard cultivation was found to be less effective for developing plant plasticity. It is more correct to compare samples when the conditions are favorable for plant performance and productivity potential. Stresses, in combination with favorable factors, contribute to an increase in trait variability and dispersion, resulting in more accurate dividing varieties into groups. In our case study, with the use of «step-by-step analysis back» module we reduced the number of discriminating factors to two ones adequate for 100 % reliable allocation of typical representatives of the groups. High mineral levels and water deficit were enough to truly classify 88 % of the samples. This is sufficient in genetic research where it is necessary to select the most typical representatives. Samples of the groups 1 and 3 have been classified correctly, and only three varieties of the group 2 have got to another cluster. The discriminant analysis also shows distance of each variety from the center of the group. Samples with the minimum distance are the most typical representatives which can be used as genetic sources of desired traits, as contrast parental forms in hybridization, or involved in marker-assisted selection and GTL mapping. Early planting, dense crops, high fertilizer rates, and lack of water were the factors which mostly influenced on the clear separation of the samples into clusters according to how the varieties responded to external environment. The virtual «ideal variety» (a model) and Kurchanka variety were grouped in the same cluster, and the varieties from the group 1 were close to the «ideal variety» on the response to environment. Despite high yield production, the dispersion in the group 3 which includes Kurchanka and the model variety was 3 times as much as in other groups. Therefore, stability of the varieties was lower in this cluster (group 3) as compared to the first and the second clusters (groups 1 and 2).
Rice, oryza sativa, multidimensional methods, estimation of breeding material, claster analysis, discriminant analysis
Короткий адрес: https://sciup.org/142214006
IDR: 142214006 | DOI: 10.15389/agrobiology.2017.1.152rus