Uso de imagens de baixo custo em programas de melhoramento visando seleção e recomendação de linhagens de feijão-caupi com elevado potencial produtivo

Abstract

The development of new cowpea cultivars through genetic improvement helps reduce the need to expand new agricultural areas, something driven by population growth and changes in diet. Image analysis in the selection of new genotypes is a low-cost alternative that helps maximize productivity, contributing to more efficient and sustainable agriculture aimed at promoting food security in less developed regions. This research was carried out in the state of Tocantins, northern Brazil, and studied several parameters using RGB images to evaluate the productive performance of cowpea, identify the best phase to weigh productivity and select lines in a genetic improvement program. In chapter I, an experiment was carried out in pots with three cowpea cultivars (BRS Guariba, BRS Nova Era and Pingo de Ouro). The image analyses were compared with traditional phenotyping techniques and an electronic chlorophyll meter. Where Tukey at 5% was used to compare means and analyze correlation of variables. The cultivar Pingo de Ouro had the highest grain productivity per pot. The image analyses showed results compatible with productivity indices, and the image analysis elements that best correlated with yield were Hue, GA, and GGA. Chapters II and III were conducted during the 2022 and 2023 agricultural years, where 9 cowpea lines and a commercial control were analyzed in 4 replicates, using a drone with an RGB camera. In the second chapter, the images were captured at 5 development stages (V3, V7, R5, R7, and R9), and analyzed with the Breedpix 2.0 software for the Hue, GA, and GGA parameters. In addition, plant height, thousand-seed weight, and productivity were evaluated. The analysis of variance performed by the F test and the Tukey test at 5% showed statistical difference in several parameters. The R5 phase was the most suitable for evaluating cowpea yield through RGB images, providing a more accurate analysis. In the third chapter, a single flight was carried out at the R5 phase to capture the RGB images. Nitrogen content of leaves, plant height, stem diameter, number and size of pods, number of grains per pod, weight of a thousand seeds and productivity were also analyzed. Statistical analysis indicated differences in plant height, number of pods per plant, weight of a thousand seeds, saturation and v* and b* parameters analyzed through images. The Hue and GGA parameters were strongly associated with productivity, highlighting the lines FC 602, FC 2006, FC 1103 and FC 2309 as the most promising for cultivation in the state of Tocantins.