Cowpea is one of the most significant food and nutrient sources worldwide, with importance in three primary market sectors: dry grains, seeds, and the expanding green-grain sector. This study aimed to identify phenotypic patterns for selection in Vigna unguiculata (L.) Walp genotypes for green-grain production using similarity networks and identify its morpho-agronomic variables with greater discrimination ability. The rainfed experiment was conducted in the experimental area of the horticultural sector at the Plant Science Department of the Agricultural Sciences Center of the Federal University of Ceará, Brazil, with 42 treatments. Three seeds were sown per hole, and the plants were thinned to two plants per hole, 15 days after sowing. Characterization was performed based on quantitative and qualitative variables, and the data were subjected to multivariate analysis of variance based on an augmented block design. The conjugate distance matrix for the variables was graphically represented using similarity networks to identify phenotypic patterns. The results indicated that genotypes CE-164, 207, 999, 994, 1002, and 1007 should not be used in breeding programs for green-pod production since they show genetic similarity within commercial cultivars. The variables of days to fruiting, green-pod length, green-pod width, green-pod thickness, and green-grain thickness contribute to genetic divergence and have high heritability values. Crosses between cowpea genotypes CE-165, 244, 22, 96, and 98 can yield gains in green-grain production in advanced generations.

Cowpea is a nutritious species cultivated worldwide whose high genetic variability can be exploited in breeding programs. The present study aimed to identify phenotypically divergent genotypes of Vigna unguiculata (L.) Walp., from this perspective concerning agronomic and physiological variables aiming at green grain production. An experiment was conducted in Fortaleza -CE, to analyse 13 variables in 44 genotypes from the Active Germplasm Bank of the Federal University of Ceará. The experiment was set up in an augmented block design with four additional controls (commercial cultivars). The genetic and environmental variance components were estimated by restricted maximum likelihood, after which the intraclass correlation coeffi cient (ICC) was calculated. Principal components analysis, Tocher's clustering and correlation analysis were used in the study. The variables most infl uenced by genetic variance were production per plant, green pod length, green grain length, grain mass per pod, and total yield, with ICC > 0.5. Principal components analysis signalled the physiological variables gs, ci, and ci.ca and the agronomic variables of green pod length, green grain mass, number of grains per pod, and earliness of production as important for differentiating between cowpea genotypes. CE genotypes 02, 151, 165, 172, 189, 244, 986, and 1002 show genetic variability, and their use is recommended in cowpea breeding programs aimed at green-grain production.

Image analysis combined with machine learning models can be an excellent tool for classification of fava (Phaseolus lunatus L.) genotypes and is a low-cost system. Fava is grown by family farmers, mainly, in the Northeast and South regions of Brazil, presenting economic and social importance. Evaluations to gather information on qualitative and quantitative characters of seeds enable the description and distinction of genotypes, allowing the evaluation of variability of plant species, which is essential in breeding programs. The use of image analysis is a fast and economic tool for obtaining large quantity of information. Machine learning techniques have been developed and implemented in the agricultural sector due to technological advances and increasing use of artificial intelligence, which enables the automatization of several processes. In this context, the objective of this work was to evaluate different machine learning models to classify fava genotypes, using data obtained through image analysis. Images of fava seeds were captured using a table scanner (HP Scanjet 2004), set to true color mode, arranged upside down inside of an aluminum box fully closed during the capture of the images for an adequate illumination and prevention of environmental noises. The K-Nearest Neighbor, Naive Bayes, Linear Discriminant Analysis, Support Vector Machine, Gradient Boosting, Bootstrap Aggregating, Classification and Regression Trees, Random Forest, and C50 models were used for the study. Linear Discriminant Analysis was the model that presented the highest efficiency for classifying the genotypes, with an accuracy of 90%.