Artificial intelligence in the selection of common bean genotypes with high phenotypic stability

Area of Science:

• Agricultural Science
• Plant Breeding
• Computational Biology

Background:

• Artificial neural networks (ANNs) are increasingly applied in plant breeding.
• Investigating genotype x environment interactions is crucial for crop improvement.
• ANNs offer potential for analyzing complex phenotypic data.

Purpose of the Study:

• To utilize ANNs for selecting common bean genotypes with superior phenotypic adaptability and stability.
• To compare ANN results with the established Eberhart and Russell method for adaptability and stability analysis.

Main Methods:

• Conducted six trials with 13 common bean genotypes from 2002-2006.
• Employed a randomized block design with three replicates.
• Analyzed grain yield data using variance analysis, Eberhart and Russell, and ANN methods.

Main Results:

• High concordance was observed between ANNs and the Eberhart and Russell method in discriminating phenotypic adaptability.
• ANNs demonstrated effectiveness in identifying stable and adaptable common bean genotypes.
• Genotypes Aporé, Rudá, and CNFv 8025 showed promise for specific environmental conditions.

Conclusions:

• Artificial neural networks are a viable tool for common bean breeding programs.
• ANNs provide consistent results with traditional methods for assessing genotype adaptability.
• Recommended genotypes offer improved grain yield and stability across diverse environments.