Estratificação ambiental, estabilidade e adaptabilidade produtiva de genótipos de soja a partir de ensaios multiambientais

Abstract

ABSTRACT The soybean (Glicine max L. Merril) is the main agricultural commoditie produced and exported by Brazil, being the second major crop producer. Because of this, the breeding programs has focused in developing cultivars with high grain yield, production stability and broad adaptability. The process of selection and positioning of superior genotypes involves a complex experimentation net and a high coast in the process, with the main goal to evaluate the productive potential of new lines that are at the final level of development and refine the technical knowledge about pre commercial products. In this sense, the objective of this study was to establish a consistent process of data summarization of multienvironment experiments aiming to identify redundant locations at the experimentation net, and to establish environment strata were the interaction G x E (Genotype x Environment) is of low magnitude and identify genotypes with productive adaptability and stability for the established strata. The experiments was conducted at 2015/2016 growing season, were was evaluated 36 genotypes (29 inbred lines + seven commercial cultivars) at 27 locations and 2017/2018 growing season, being evaluated 30 genotypes (22 inbred lines + eight commercial cultivars) at 17 locations. The experiments were conducted in randomized block design with three replications and the variable analyzed was the grain yield in kg ha-1. The collected data were submitted firstly to individual variance analysis aiming to confirm the experimental precision in all the locations. After that, the environment strata was realized from two methodologies. The first was conducted from the grouping of the environments by the UPGMA method (Unweighted Pair-Group Method Using Arithmetic Average) being the dendrogram construction from the environments with similar G x E interaction. At the second was conducted the joint variance analysis with AMMI (Additive Main effects and Multiplicative Interaction analysis) decomposition. From the environment strata by AMMI analysis were identified the genotypes with major stability and more adapted to the different strata based on prediction of the genotypes performance from the mixed models methodology REML/BLUP. The results for the 2015/2016 growing season showed for the UPGMA grouping methodology the formation of two groups (with two and four locations, respectively) remaining the rest of the environments (21) isolated. On the other hand, the results for the AMMI methodology showed that the 27 locations were reduced to six environment strata. From this stratification were identified the most promising genotypes, with the better adaptation and stability, highlighting the line L4 that was ranked between the better’s genotypes at the strata II, III and IV. The UPGMA grouping at 2017/2018 growing season also evidenced the formation of two groups with two locations grouped in each one. In contrast, the second methodology showed the reduction of the 17 locations in six environment strata from which was possible to predict the most promising genotypes. Positive feature was observed to the line L19 ranked in first place in two strata, and being among the best genotypes in the others. From the results obtained for the both evaluation growing seasons, it was possible to confirm that the grouping of environments via UPGMA was relevant in the identification of redundant locations. On the other hand, the AMMI methodology demonstrated greater efficiency in the reduction of the test locations in consistent environmental strata. In addition, the grain yield prediction via REML / BLUP enabled the identification of superior genetic genotypes and associated with AMMI analysis the exploration of broad and specific adaptation.