Tamanho ótimo de parcela experimental e seleção de descritores morfoagronômicos em Ricinus communis L

Abstract

The aim of this work was to estimate the optimum size of experimental plots for the evaluation of morphoagronomic descriptors in castor bean, as well as to select the descriptors that best characterize the species. The experiments were carried out in an experimental area belonging to the Nucleus of Genetic Improvement and Biotechnology (NBIO), linked to the Agricultural, Environmental and Biological Sciences Center (CCAAB) of the Universidade Federal do Recôncavo da Bahia (UFRB), in Cruz das Almas - BA. For the determination of the optimal plot size, the modified maximum curvature method and linear model of plateau response were used and the following characteristics were evaluated: flowering, primary racemic insertion, plant height, number of seeds per plant, seed weight per plant and productivity. And for the selection of the morphoagronomic descriptors, 18 quantitative and 22 qualitative descriptors were evaluated. The average maximum plot size by the modified maximum curvature method is 4.75 basic units or approximately 5ub (15 m²). And by the linear model of plateau response is 12.95ub (38.84 m²) or approximately 13ub (39 m²). The modified maximum curvature method presented better adjustment quality in relation to the data, by the coefficient of determination. 13 quantitative descriptors were selected in the simultaneous analysis of the two years of cultivation by the methodology proposed by Jolliffe and 11 qualitative descriptors by the level of entropy. The discarding of 28% of the quantitative descriptors did not cause considerable loss of information, since the redundant descriptors are correlated to the remaining descriptors, with possibility of cost reduction and better dynamics in the management and characterization of the culture. Of the 22 qualitative descriptors evaluated, 59% were relevant in the discrimination of the study population.