Association Mapping Of QTL For Agronomic Traits In A Yangtze-Huai Soybean Breeding Germplasm Population By Different Methods

Understanding of genetic base of the target traits can promote breeding efficiency in soybean. Association analysis is one of the mapping methods based on linkage disequilibrium among genes (marker loci) to detect genes or QTL associated with the target traits, so it can be used to reveal the genetic basis of a set of elite parents for breeding practice. However, population structure and other factors can affect the detection effects in soybeans and other self-pollinated crops, and new methods for association analysis are still improved. To obtain relatively ideal results, it is necessary to compare and select a suitable method (model) of currently used software for association analysis. In the present study,401breeding lines from the crosses between four core parents and about100domestic and exotic germplasm and some of their parents were selected to establish a Yangtze-Huai soybean breeding germplasm population (YHSBGP). Five traits,100-seed weight, maturity time, flowering time, plant height and main-stem node number, were surveyed in five environments. Furthermore,189InDel markers were used to reveal population structure and simulate data to compare effects of different association analysis methods (models). According to the simulation results, the suitable method (model) was chosen to conduct association analysis for these five agronomic traits; and the trait performance of the derived progenies from crosses among parents with big seed size or early maturity was estimated to guide molecular breeding according to the effects of elite alleles. The main results are as follows:1. The ANOVA results showed that there were great variation and significant differences among the tested lines for100-seed weight, plant height, main-stem node number, flowering time and maturity time. The broad heritability values of the five traits were85.34%,71.11%,65.56%,81.95%,83.56%respectively. Total398alleles were identified by189InDel markers. The average allele number per locus was2.11, and the average of gene diversity value was0.34. The lines of YHSBGP could be classified as four subpopulations by STRUCTURE2.2. Linkage disequilibrium was complex, and existed on different chromosomes, the average distance that LD exists was4.33Mbp. Kinship values were calculated based on the InDel marker data, and92.39%of the pairwise kinship between two lines was less than0.5, indicating relatively low relationship among the lines.2. The simulation tests were conducted by GLM model, MLM_K model, MLM_Q+K model of TASSEL v3.0and stepwise regression method (SRM). The results indicated that when the heritability was low, detection power was low and false detection ratio was high for all methods (models), and the power increased when the heritability values increase. The GLM model had the highest false detection ratio, while the SRM had the lowest ratio when heritability was high. The power and detection ratios of MLM_K and MLM_Q+K models were similar. So MLM_Q+K and SRM were chosen for association analysis in YHSBGP.3. Association mapping was conducted to detect marker loci for the five traits by MLM_Q+K and SRM. A total of233times was identified to be significantly associated (-log(P)>1.30) with five agronomic traits in two year and five environments by MLM_Q+K model, and there were58times associated to100-seed weight,51times associated to maturity time,21times associated to flowering time,51times associated to plant height,45times associated to main-stem node number. A total of140times were identified to be significantly associated (-log(P)>1.30) with five agronomic traits by SRM, and there were25times associated to100-seed weight,36times associated to maturity time,28times associated to flowering time,30times associated to plant height,21times associated to main-stem node number. Combined with two methods,18loci were identified to be associated with100-seed weight,12and15loci associated with flowering time and maturity time respectively, and14and11loci associated with plant height and main-stem node number by both methods, respectively.4. Elite alleles of the associated loci for100-seed weight and maturity time were estimated, and the prediction values of the progenies from the crosses between parents with big seed size or early maturity were also estimated. The best crosses for developing big seed or earlier maturity soybean were selected.