Predição fenotípica no melhoramento da soja para diferentes regiões do Brasil

Abstract

Among the various socioeconomic cultures cultivated in Brazil, soybean deserves prominence, being cultivated in various environments, thanks to the efforts of genetic improvement programs for continuous development of new cultivars. There are numerous possibilities of crosses, from the available germplasm, thus generating thousands of progenies that need to be thoroughly tested, making the breeding process very time consuming and costly. In the present work it was studied the correlation between the prediction of crosses with the real productive potential of conventional soybean progenies (Breeding Value (BV)), in order to verify the effectiveness of the use of this methodology in obtaining progenies with high productive potential for the macro-regions soybeans (MRS) and edaphoclimatic regions (RECs) of Brazil. We selected a group of 481 conventional genotypes as parents and, from them, generated the BV and proceeded to predict crosses using the procedure of mixed models with maximum likelihood - better unbiased linear estimator (REML BLUP). In 2019, the predicted crosses were performed and the F1 populations were sown in Porto Nacional-TO to obtain the F2 seeds which, after harvest, were sent to the five breeding programs of a private company based in Passo Fundo-RS, Cambé-PR, Rio Verde-GO, Lucas do Rio Verde-MT and Porto Nacional-TO. In the 2019/2020 harvest the F2 segregating populations were sown in experimental tests to obtain F3 seeds, thus, the selective process of the best progenies, although still with a high degree of heterozygosity, occurred in the adaptation region itself. In all, 1868 progenies were selected which, in the 2020/2021 harvest, were sown in F3 trials in the respective breeding programs of each MRS, evaluating the grain yield (YG). In the 2021/2022 harvest, the progenies were sown in F4 trials in 50 environments in the RECs distributed in Brazil, where they were again evaluated for YG and phenotype for characterization. To obtain the results of the effectiveness of the prediction of crosses in relation to the productive potential of the progenies in the experimental tests of F3 and F4, the correlation analysis was used. The prediction of crossings resulted in very weak to moderate correlation between values of YG BV vs YG F3 and YG BV vs YG F4 indicating, in the prediction model used in this study, little efficiency in the preliminary identification of crosses with high potential to generate agronomically superior progenies for the soybean regions of Brazil.