Adaptability and phenotypic stability of soybean genotypes regarding epicotyl length using artificial neural network and non-parametric test

Abstract

Genetic improvement together with statistics has contributed to the growth of the importance of soybean in Brazil. One of the contributions has been the launching of new cultivars in the national market, which requires, in its legal procedures for registration and protection, the verification of several tests, one of them being the distinguishability test. Several studies have reported that some phenotypic characters are potential in this distinction, one of them is the length of the epicotyl. In this work, the objective was to identify soybean genotypes that present low or high average, highly stable throughout the analyzed environments and that present adaptability to different environments. Two groups of experiments were conducted in a greenhouse to measure the epicotyl length of soybean plants submitted to different environments (planting season). The data obtained were analyzed using the analysis of individual variance, analysis of joint variance, Scott-Knott test and adaptability and stability through the Artificial Neural Network and non-parametric test. It can be concluded that the genotypes that showed low average for epicotyl length, wide adaptability or poor responsiveness to environmental improvements and stable over the seasons were TMG 1175 RR (in V2), BMX Tornado RR (in V2), BG 4272 (in V2), BRS283 (in V2 and V3) and FT-Cristalina (in V2 and V3). BRSMG 752 S (in V2 and V3), TMG 4185 (in V3) and BRSGO 7560 (in V3) behaved as high medium, high stability and wide adaptability. The genotypes BRS 8381, TMG 4185, MG/BR46_Conquista, BRSMG 850 GRR, BRS Valiosa RR and BG 4277 were stable and recommended for favorable environments.