Association between unmanned aerial vehicle high‐throughput canopy phenotyping and soybean yield

Abstract

Identifying agronomic traits correlated to grain yield can be very useful for soybean [Glycine max (L.) Merr.] breeding, especially if these traits can be measured through unmanned aerial vehicle high-throughput phenotyping rather than through manual measurements. The objective of the present study was to assess the association between canopy coverage and soybean grain yield through different statistical methodologies. A panel with 97 soybean genotypes was evaluated in two field experiments conducted in Paraná State, Brazil. Canopy coverage was determined by using an RGB camera coupled to a drone. Images taken during flights at phenological stages V3-V4, V5-V6, V7-V8, and V9-R1 were used to calculate canopy coverage based on the green pixel ratio in each experimental unit. There were significant genotype × environment interactions in all evaluated traits. Selective accuracy values (0.73-0.96) revealed indirect yield selection efficiency based on canopy coverage. High genetic correlation estimates (0.76) were observed between grain yield and canopy coverage at flowering in one of the assessed environments. These results were confirmed through genetic correlation coefficient decomposition in direct and indirect effects and of gain estimates presenting indirect selection. Thus, canopy coverage data remotely collected using drones to soybean indirect selection for grain yield can be a promising strategy to accelerate genetic gains in soybean breeding programs.

Crop HTP Technologies: Applications and Prospects

“…Casagrande et al [16] used a drone-mounted RGB camera to capture field images and calculated crown coverage for soybeans during the V3-R1 stage based on the ratio of green pixels for each experimental unit. The results showed that, as the soybean growth stage progressed, soybean yield was positively correlated with canopy photosynthesis, with a correlation of 0.76 at the V9-R1 stage.…”

High-Throughput Field Phenotyping to Advance Precision Agriculture and Enhance Genetic Gain, Volume II

“…There was a significant positive correlation between panicle number and yield at the maturity stage, and it could be recognized and counted directly on a certain regional scale in most monocotyledons (Jun et al, 2021;Wanli et al, 2021). However, in many dicotyledons, researchers can predict yields using only a large number of other indicators or measure the yield by picking and laying out fruits at maturity (Casagrande et al, 2022) or by picking and spreading out fruits at maturity (Li et al, 2021;Xiaobin et al, 2022), which greatly increases labor costs and is not beneficial to the development of high-throughput phenotypes.…”

High-throughput Phenotyping of Soybean Biomass: Conventional Trait Estimation and Novel Latent Feature Extraction Using UAV Remote Sensing and Deep Learning Models

“…In soybean breeding programs, traits related to aboveground biomass, which is crucial for defining the canopy structure, are key selection targets [34]. Previous studies have focused on developing HTP techniques to estimate traits such as DW, canopy coverage rate, canopy volume, and PH [15, 16, 35]. Despite this progress, a gap remains in the HTP techniques available for capturing the phenotype of specific traits that determine canopy structure, such as NN, NB, and the extent of spread.…”