Artificial neural networks to identify semi-prostrate cowpea genotypes with high phenotypic adaptability and stability

Abstract:

The objective of this work was to verify the agreement between artificial neural networks (ANNs) and the Eberhart & Russel method in identifying cowpea (Vigna unguiculata) genotypes with high phenotypic adaptability and stability. The experimental design was in a randomized complete block with four replicates. The treatments consisted of 18 experimental lines and two cowpea cultivars. Four value for cultivation and use trials were conducted in the municipalities of Aquidauana, Chapadão do Sul, and Dourados, in the state of Mato Grosso do Sul, Brazil. Grain yield data were subjected to individual and joint variance analyses. Then, the data were subjected to adaptability and stability analyses through the methods of Eberhart & Russell and ANNs. There was a high agreement between the methods evaluated for discrimination of the phenotypic adaptability of semi-prostrate cowpea genotypes, indicating that ANNs can be used in breeding programs. In both evaluated methods, the BRS Xiquexique, TE97-304G-12, and MNC99-542F-5 genotypes are recommended for harsh, general, and favorable environments, respectively, for having grain yield above the overall average of environments and high phenotypic stability.

Index terms :
Vigna unguiculata; artificial intelligence; genotypes x environments interaction.