SELECTION OF PRODUCTIVE HYBRID OIL FLAX PLANTS BY METHODS OF MULTIVARIATE ANALYSIS

Use of computer technology allows to quickly analyze and use subject, technological, analytical and other information. Biometric statistics in plant breeding is aimed at optimizing (increasing efficiency, reliability, acceleration and cheapening) the process of breeding varieties of agricultural crops. Therefore, creation and study of new varieties of oil flax requires widespread introduction of modern computer information technologies that provide information support of the breeding process at all its stages. Methods of multi-criteria mathematical statistics - factor and cluster analyses - were used in the studies for a comprehensive assessment of hybrid populations of oil flax by productivity elements (plant height, technical length, inflorescence length, number of pods per plant, number of seeds per plant, number of seeds in a box, weight of 100 seeds, and oil content in seeds). Effectiveness of selection of hybrids of the third cycle of breeding has been evaluated, and also the distinctive features of hybrid combinations in a number of generations have been established. As a result of selection and technological cycle of the analysis, 31 highly productive hybrids (or 6.9%) were identified for further reproduction. Despite the high level of the breeding differential determined in hybrid combinations during the F-2-F-3 generation change, their response to traits based selection according to "number of seeds in a box" and "weight of 100 seeds" was weak, and selection by the number of boxes and seeds from the plant turned out to be ineffective. The selection method used makes it possible to cull low-yielding plants that have fallen into the worst groups of clusters. Culling by the method of multidimensional analysis should be used in later generations (fourth-fifth cycle of selection) as homozygosity of traits is established.