Automatic collecting of phenotypic information from plants has become a trend in breeding and smart agriculture. Targeting mature soybean plants at the harvesting stage, which are dense and overlapping, we have proposed the SPP-extractor (soybean plant phenotype extractor) algorithm to acquire phenotypic traits. First, to address the mutual occultation of pods, we augmented the standard YOLOv5s model for target detection with an additional attention mechanism. The resulting model could accurately identify pods and stems and could count the entire pod set of a plant in a single scan. Second, considering that mature branches are usually bent and covered with pods, we designed a branch recognition and measurement module combining image processing, target detection, semantic segmentation, and heuristic search. Experimental results on real plants showed that SPP-extractor achieved respective R² scores of 0.93–0.99 for four phenotypic traits, based on regression on manual measurements.

A biparental soybean population of 364 recombinant inbred lines (RILs) derived from Zhongdou 41 × ZYD 02.878 was used to identify quantitative trait loci (QTL) associated with hundred-seed weight (100-SW), pod length (PL), and pod width (PW). 100-SW, PL, and PW showed moderate correlations among one another, and 100-SW was correlated most strongly with PW (0.64–0.74). Respectively 74, 70, 75 and 19 QTL accounting for 38.7%–78.8% of total phenotypic variance were identified by inclusive composite interval mapping, restricted two-stage multi-locus genome-wide association analysis, 3 variance-component multi-locus random-SNP-effect mixed linear model analysis, and conditional genome-wide association analysis. Of these QTL, 189 were novel, and 24 were detected by multiple methods. Six loci were associated with 100-SW, PL, and PW and may be pleiotropic loci. A total of 284 candidate genes were identified in colocalizing QTL regions, including the verified gene Seed thickness 1 (ST1). Eleven genes with functions involved in pectin biosynthesis, phytohormone, ubiquitin-protein, and photosynthesis pathways were prioritized by examining single nucleotide polymorphism (SNP) variation, calculating genetic differentiation index, and inquiring gene expression. The prediction accuracies of genomic selection (GS) for 100-SW, PL, and PW based on single trait-associated markers reached 0.82, 0.76, and 0.86 respectively, but selection index (SI)-assisted GS strategy did not increase GS efficiency and inclusion of trait-associated markers as fixed effects reduced prediction accuracy. These results shed light on the genetic basis of 100-SW, PL, and PW and provide GS models for these traits with potential application in breeding programs.

The β subunit of soybean [Glycine max (L.) Merr.] seed storage protein is of great significance in sulfur-containing amino acid balance and soybean processing properties. The objective of this study was to elucidate the relationship between the β subunit and sulfur-containing amino acid composition, and the potential regulatory mechanism. The β subunit was independently accumulated in comparison with other major subunits (α/α′, acidic, basic, and A3) during seed filling, and a low level of β subunit content (BSC) was formed during the accumulation process. In low-BSC mature seeds, crude protein, oil content, and fatty acid composition were not changed, but sulfur-containing amino acids (Cys + Met) in the low-BSC seeds increased significantly (by 31.5%), suggesting that an internal regulatory mechanism within seed might be responsible for the rebalance of seed protein composition and that sulfur assimilation might be deeply involved in β subunit accumulation. Transcriptomic analysis revealed that genes involved in anabolism of cysteine, methionine, and glutathione were up-regulated but those involved in the catabolism of these compounds were down-regulated, suggesting a relationship between the elevation of methionine and glutathione and low BSC. Our study sheds light on seed composition in low BSC lines and on the potential molecular regulatory mechanism of β subunit accumulation, broadening our understanding of soybean seed protein synthesis and its regulation.