Genetic Dissection And Elite Allele Identification Of Seed Traits In Soybean Cultivars Released From Huanghuai Valleys And Southern China

Soybean is one of the most important crops in the world with multiple purposes. In the region of Huanghuai Valleys and Southern China,618 cultivars were released as main achievements during 1923-2005. All these cultivars are important breeding materials. It is important to map QTLs/genes of oil, protein content and 100-seed weight and dissect their elite alleles in these cultivars for further high-yield and good quality breeding in soybean. In this study,282 cultivars from Huanghuai Valleys and Southern China were selected and a two-year field experiment including three environments was carried out. The genetic variation of oil, protein content and 100-seed weight was analyzed, and association mapping of the loci for oil, protein content and 100-seed weight were done by using 150 SSR markers. The elite alleles of the constant loci for these traits as well as the allele distribution in different ecological regions were also analyzed for molecular breeding of seed traits by pyramiding these alleles. Moreover, models based on Near Infrared Reflectance Spectroscopy (NIRS) was established to rapidly investigating phenotypic data of fatty and protein content traits of soybean with different colorof seed coat. The main results are as follows:(1) There were great variation of these released cultivars with Oil and protein content ranged from 14.86-23.34% and 37.63-52.83%, and 100-seed weight ranged from 9.88 to 32.92 gram. The population is suitable for association mapping. There existed significant differences of environment factor as well as cultivar×environment according to the joint ANOVA results. The average values of the oil content in HuangHuai valleys (Ⅱecological region) sub-population, protein content in middle-southern China (Ⅳecological region) sub-population and 100-seed weight in Changjiang valley and middle-southern China (ⅢandⅣecological region) sub-population are the highest.(2) According to the Hardy-Weinberg equilibrium model, the cultivars from Huanghuai valleys and southern China were divided into three subpopulations by STRUCTURE software. Association mapping analysis was conducted by using the GLM model with the modified population structure as covariant. Under the significance level below 0.01, eleven markers linked to oil content QTLs were detected in each of the three environments, and eighteen were found to environment-specific, among them, sct₁89 and sat₃52 were not reported yet. Ten and eighteen markers associated with QTLs of protein content were detected in all three environments or only one environment respectively, and six were newly found. Eleven markers to QTL of 100-seed weight were detected in all three environments, and other thirteen detected were environment-specific, Satt191 and Satt540 were not reported previously.(3) Sixteen elite alleles for eleven loci of oil content and seventeen elite alleles for 10 loci of protein content QTL expressed in all three environments were identified respectively. Sixteen elite alleles for nine loci of 100-seed weight were also found. Analyzing genetic composition of the cultivars, some good cultivars accumulated a larger number of elite alleles. There was significant difference of the elite allele composition among Eco-regions. Some with suitable composition of the elite alleles were chosen as potential parents for further gene pyramiding.(4) Prediction models via BRUKER VECTOR 22/N Fourier transforms near infrared reflection spectroscopy (NIRS) were established to rapidly investigate oil, protein and non-saturated fatty acid content in soybean. Total 340 landraces and 25 wild soybean accessions from 24 provinces were selected as a modeling set and a partial least squares method was used. Four groups of NIRS prediction model, detailed in 12 sub-group of NIRS model were developed. Each model was evaluated according to its coefficient of determination of cross validation (Rev2) and root mean square error of cross validation (RMSECV). Model M4B was suitable for soybean with black seed coat. Model M4A was suitable to landrace and wild soybean with various color types of seed coat. By comparison with the whole grain, the soybean powder is more suitable for soybean NIRS modeling.