Fenotipagem de alto rendimento: protocolo para seleção de genótipos de milho com imagens aéreas RGB

Abstract

Maize (Zea mays L.) is one of the most well-established crops in the world, as it has a wide range of uses and adaptability for cultivation. At the same time, unmanned aerial vehicles are playing an important role in crop improvement, allowing the use of high-quality images to select genotypes, optimizing the time and costs of obtaining cultivars. The aim was to study the reliability of RGB aerial images in identifying genotypes with productive potential during the vegetative stage. Four genotypes were evaluated in a randomized block design (RBL) with four replications, at three row spacings (0.60, 0.70 and 0.80 m), in a strip scheme. Seven flights were made over the experiment, using two heights (60 and 80 m). 29 vegetation indices for the RGB sensor were extracted from the image processing and used to differentiate the genotypes in terms of grain yield at plot level. A nested model was fitted to predict the temporal BLUPs and the best indices were selected based on the estimated repeatability. Manual phenotyping showed significant differences between genotypes and spacings. The vegetation indices with the highest repeatability indicated different patterns during the vegetative phase, making it possible to characterize the flight date that establishes the difference in grain yield during the vegetative stages. The optimum flight date is 43 days after planting (DAP) for the two heights evaluated. The MRCC, R-B and RCC vegetation indices are recommended for use in image phenotyping.