Unravelling the genetic architecture of soybean tofu quality traits

Abstract

Tofu is a popular soybean (Glycine max (L.) Merr.) food with a long tradition in Asia and rising popularity worldwide, including Central Europe. Due to the labour-intensive phenotyping procedures, breeding for improved tofu quality is challenging. Therefore, our objective was to unravel the genetic architecture of traits relevant for tofu production in order to assess the potential of marker-assisted selection and genomic selection in breeding for these traits. To this end, we performed QTL mapping with 188 genotypes from a biparental mapping population. The population was evaluated in a two-location field trial, and tofu was produced in the laboratory to evaluate tofu quality. We identified QTL for all investigated agronomic and quality traits, each explaining between 6.40% and 27.55% of the genotypic variation, including the most important tofu quality traits, tofu yield and tofu hardness. Both traits showed a strong negative correlation (r = -0.65), and consequently a pleiotropic QTL on chromosome 10 was found with opposite effects on tofu hardness and tofu weight, highlighting the need to balance selection for both traits. Four QTL identified for tofu hardness jointly explained 68.7% of the genotypic variation and are possible targets for QTL stacking by marker-assisted selection. To exploit also small-effect QTL, genomic selection revealed moderate to high mean prediction accuracies for all traits, ranging from 0.47 to 0.78. In conclusion, inheritance of tofu quality traits is highly quantitative, and both marker-assisted selection and genomic selection present valuable tools to advance tofu quality by soybean breeding.

Supplementary information: The online version contains supplementary material available at 10.1007/s11032-024-01529-x.

Keywords: Genomic selection; Plant breeding; QTL mapping; Soybean; Tofu traits.

Fig. 1

Trait distributions. Distribution of BLUE values across locations for the traits thousand-seed weight (TSW), protein content, oil content, soaking factor, soymilk weight, tofu weight, tofu yield, tofu hardness, tofu value, and number of hard beans. The performance of the parental genotypes is indicated by red triangles. Note, BLUE values for hard bean counts were estimated based on square root transformed values, the resulting BLUEs were transformed back for this figure

Fig. 2

Correlation among the investigated traits. a Network plot and b matrix with the correlations among the traits thousand-seed weight (TSW), protein content (PC), oil content (OC), soaking factor (SF), soymilk weight (SM), tofu weight (TW), tofu yield (TY), tofu hardness (TH), tofu value (TV), and number of hard beans (HB). Asterisk indicates correlation coefficients significantly different from zero at the 5% significance level

Fig. 3

The effect of QTL stacking. The boxplot illustrates the effect of combining favourable alleles of the four identified QTL for tofu hardness on chromosomes 1, 8, 10 and 18 in 188 genotypes. + indicates the presence of the homozygous favourable allele, -indicates the homozygous unfavourable allele, heterozygous loci and missing values. The red dotted line represents the quality threshold of 75 N desired in German tofu production

Fig. 4

The potential of genomic prediction. Genomic prediction accuracies from 1000 runs of cross-validation and their respective means are shown