Avaliações quantitativas em linhagens endogâmicas recombinantes de arroz (RILs) com ênfase em caracteres de qualidade de grãos

Abstract

Rice is one of the cereals with the greatest social and economical importance, responsible for being the staple food for more than half of the world population, mainly in developing countries. Breeding programs search for new technologies and strategies with the aim to attend to growing demands of food needs, taking into account productive performance combined with the traits of culinary interest. Due to the narrowing of the crops genetic base, finding superior genotypes that meet the need of farmers and consumers becomes increasingly complicated. Thus, hybridization, especially between different subspecies of rice is one of the ways found for the expansion of genetic variability. In this regard, in order to assist making-decision in breeding programs, such as the selection of superior genotypes for agronomic and culinary traits, and the prediction of genetic distance, the statistical support through the use of biometric models has been shown as a precise alternative. The present study aimed to identify the phenotypic and genotypic linear and canonical correlations, to select promising families and characterize the genetic distance of a set of recombinant inbred lines of paddy rice. Reciprocal crosses between BRS Querência (Oryza sativa sp. indica) and Nipponbare (Oryza sativa sp. japonica), the generation advancing up to the F7 generation and the field assessments, along with the parents, were carried out at the experimental field of Embrapa Clima Temperado, Capão do Leão – RS. The experimental design was the incomplete blocks with intercalary controls, arranged in four replications. From the study of the present segregating population, with the aid of canonical, phenotypic and genotypic correlations it is possible to verify that groups of grains quality and agronomic traits are not independent, and there is a tendency for the amylose content and broken grain traits to be associated to grain yield and/or its components. The families F5 and F105, through the BLUP, reveal as the potential candidates for improved cultivars or to integrate cross breeding blocks. According to multivariate analysis and neural networks the families F17 and F128, as well as F108 and the parent BRS Querência, are the most similar genotypes to each other. On the other hand, the most genetically distant from most of the others are the families F17, F60, F108 and F128.