Seed phenotype and maturity groups as determinants of protein, oil, and fatty acid composition patterns in diverse soybean germplasm

Abstract

Soybean seed physical characteristics are crucial for quality assessment, but the link between these characteristics and biochemical composition across different maturity groups (MGs) remains unclear. This study examined the relationships between seed physical characteristics (color and weight) and biochemical constituents, including oil content (OC), protein content (PC), and fatty acid (FA) composition in 191 diverse soybean accessions across eight MGs (0-VII) at three locations over two years. The results indicated that black-seeded accessions demonstrated a notably higher average of PC (47.33%) and a lower average of OC (15.78%) in contrast to yellow-seeded varieties, which had an average PC and OC of 42.52% and 19.12%, respectively. In addition, larger seeds exhibited increased OC (19.15%) and OA levels (23.27%), whereas smaller seeds revealed higher concentrations of PC (44.23%), LA (55.06%), and LNA (8.53%). Multivariate analyses, including principal component analysis, clustering heatmap, and radar plot, demonstrated distinct clustering patterns, exhibiting unique compositional profiles closely linked to seed physical characteristics. Furthermore, MGs exhibited notable correlations with LNA (R² = 0.238) and OC (R² = 0.233), especially in black-seeded and large-seeded accessions. These findings elucidate the complex interaction between seed physical traits and biochemical composition, presenting significant implications for soybean breeding programs aimed at specific quality attributes.

Keywords: Glycine max; Maturity groups; Multivariate analysis; Seed coat color; Seed composition; Seed weight.

Fig. 1

The geographical distribution of 191 soybean accessions from 4 regions: China (142 accessions), USA (43 accessions), Russia (5 accessions), and Japan (1 accession). a Four geographical origins of the 191accessions plotted on a world map, with the number of accessions represented by the size of the marker. b A map includes the 3 planting locations within China

Fig. 2

Distribution of seed composition traits across different soybean seed coat colors. Boxplots show the variation in protein content, oil content, and fatty acid compositions (palmitic acid, stearic acid, oleic acid, linoleic acid, and linolenic acid) among yellow, hazel, and black soybean seeds from 191 accessions. The boxes represent the interquartile range (IQR), with the middle line indicating the median. Whiskers extend to 1.5 times the IQR, and individual points represent outliers. Statistical significance is indicated by asterisks: *p < 0.05, **p < 0.01, ***p < 0.001, ns: not significant. Horizontal brackets show the pairwise comparisons between seed coat colors

Fig. 3

Distribution of seed composition traits across different soybean seed weights. Boxplots showing the variation in oil content, protein content, and fatty acid compositions (palmitic acid, stearic acid, oleic acid, linoleic acid, and linolenic acid) among small, medium, and large soybean seeds from 191 accessions. The boxes represent the interquartile range (IQR), with the middle line indicating the median. Whiskers extend to 1.5 times the IQR, and individual points represent outliers. Statistical significance is indicated by asterisks: *p < 0.05, **p < 0.01, ***p < 0.001, ns: not significant. Horizontal brackets show pairwise comparisons between seed weight classes

Fig. 4

Principal Component Analysis (PCA) revealing the relationship between seed physical characteristics and seed biochemical compositions traits in soybean. a PCA biplot showing the distribution patterns of different seed colors (black, hazel, and yellow) in relation to seven compositional traits. b PCA biplot demonstrating the relationship between seed weight categories (large, medium, and small) and compositional traits. Ellipses represent 95% confidence intervals for each seed weight category. Vector lengths indicate the magnitude of variable contributions, while angles between vectors represent correlations between variables. Opposing vectors suggest negative correlations, while vectors pointing in similar directions indicate positive correlations among traits

Fig. 5

Multivariate analysis of seed samples based on physical and biochemical characteristics. (a) Clustering heatmap presents the relationship between seed samples based on their physical attributes (seed weight and color) and biochemical traits, including protein, oil, palmitic acid, stearic acid, oleic acid, linoleic acid, and linolenic acid. The heatmap color gradient reflects the z-scores of each trait, with red indicating higher values and blue indicating lower values. (b) Radar plots display the comparative biochemical profiles across different seed colors (top) and seed weights (bottom), showing the relative percentages of protein, oil, and fatty acid compositions (PA: palmitic acid, SA: stearic acid, OA: oleic acid, LA: linoleic acid, LNA: linolenic acid)

Fig. 6

Relationship between maturity groups and biochemical traits in 191 soybean accessions categorized by seed color. Scatter plots with trend lines show the distribution of seven traits: protein content (%), oil content (%), and fatty acid composition including palmitic acid, stearic acid, oleic acid, linoleic acid, and linolenic acid. Each point represents individual accessions colored by seed color (black, hazel, and yellow). Error bars indicate standard error of the mean. R² values and p-values for each trait indicate the strength and significance of the relationship with MGs

Fig. 7

Distribution patterns of biochemical traits across MGs in relation to seed weight categories among 191 soybean accessions. Scatter plots with trend lines illustrate the relationship between MGs and seven traits: protein content (%), oil content (%), and fatty acid composition including palmitic acid, stearic acid, oleic acid, linoleic acid, and linolenic acid. Individual points represent accessions categorized by seed weight (large, medium, and small). Error bars indicate standard error of the mean. R² values and p-values demonstrate the strength and significance of correlations between traits and MGs