A Strategy and Its Application to Acquire Superior Progenies from a Soybean Bi-Parental Cross Based on Prediction By a Computer Simulation

Important agronomic traits such as yield and nutrition content are known to be controlled quantitatively, and their effective improvement is never easy. In this study, we proposed a three-step strategy for the improvement of important agronomic traits: (i) identification of the genomic region associated with a target trait, (ii) computer simulation of crossing patterns to acquire superior progenies by bi-parental crosses, and (iii) integration of desirable genotypes by actual crosses. This strategy was validated for increasing the seed protein content (SPC) of soybean by using recombinant inbred lines derived from bi-parental cross. Individual 194 RILs were evaluated at four locations and genotyped using 513 SNP markers. Twenty-nine markers were close to the QTL and significantly associated with SPC. Prediction models were constructed using the phenotype and marker genotype of the RILs. Marker genotypes of F₁ and F₂ were simulated in all possible crosses, and the genotypes were used to predict the SPC of each F₂. Based on the simulations of 18,721 possible cross combinations, we selected a combination to obtain F₂ with predicted SPC of more than 1 % higher than that of the highest RIL. Actual crosses were conducted, and the obtained F₂ plant genotypes were similar to the simulated genotypes. Some F₂ plants with desirable genotypes showed high SPC, and the stability of increasing SPC was confirmed from their progenies. These results suggest that the proposed strategy is effective for obtaining reliably superior progenies according to marker genotypes, thereby contributing to efficient improvement of a target quantitative trait.