High-throughput Phenotyping of maturity, grain yield, and flowering date using UAV Multispectral and RGB Sensors

Soybean stands out as one of the most economically important crops in the world. In this sense, high-throughput phenotyping is a key process for automating the evaluation of phenotypes on a large scale using a non-destructive method. Our study aimed to apply high-throughput phenotyping using unmanned aerial vehicle (UAV) Multispectral and RGB sensors to evaluate soybean genotypes through mixed model analysis with evaluation of vegetation indices (VIs). The experiment was conducted at Embrapa during the 20/21 season in 1004 experimental plots. A total of 284 genotypes were evaluated, as part of the Value for Cultivation and Use (VCU) testing from the genetic breeding program at Embrapa Soybean. The experimental design used was a randomized block with plots containing 4 rows of 5 meters and a spacing of 0.40 meters between rows. The experimental design consisted of 56 rows and 18 columns, encompassing different maturity groups and biotechnologies in soybean breeding trials. The traits evaluated were flowering date, maturity (R8 stage), grain yield (kg ha-1), and vegetation indices NDVI (Normalized Difference Vegetation Index), GNDVI (Green Normalized Difference Vegetation Index), NDRE (Normalized Difference Red Edge Index), and NGRDI (Normalized Green Red Difference Index). Seven flights were conducted during the crop cycle. The flights were planned at a height of 50 meters, with 4 control points located in the study area. The data was subjected to mixed model analysis in a row and column spatial model using the SpATS package in R programming. The blups and heritability of the traits were analyzed. Pearson's correlation matrix analysis between the traits and the vegetation indices, especially for maturity, revealed a strong correlation with values of 0.95 for both MAT/FLY5_NDVI and MAT/FLY5_NGRDI. Regarding the indirect selection efficiency (ISE), the optimal flight day for evaluating maturity was observed at approximately 95 days after planting and varied for each trait and vegetation index. In conclusion, the selection of the best flight day depends on the trait and VIs. Our findings demonstrate that a cost-effective RGB index performs as well as the multispectral ones.