Analysis of QTL-allele system conferring drought tolerance at seedling stage in a nested association mapping population of soybean [Glycine max (L.) Merr.] using a novel GWAS procedure.

RTM-GWAS identified 111 DT QTLs, 262 alleles with high proportion of QEI and genetic variation accounting for 88.55-95.92% PV in NAM, from which QTL-allele matrices were established and candidate genes annotated. Drought tolerance (DT) is one of the major challenges for world soybean production. A nested association mapping (NAM) population with 403 lines comprising two recombinant inbred line (RIL) populations: M8206 × TongShan and ZhengYang × M8206 was tested for DT using polyethylene-glycol (PEG) treatment under spring and summer environments. The population was sequenced using restriction-site-associated DNA sequencing (RAD-seq) filtered with minor allele frequency (MAF) ≥ 0.01, 55,936 single nucleotide polymorphisms (SNPs) were obtained and organized into 6137 SNP linkage disequilibrium blocks (SNPLDBs). The restricted two-stage multi-locus genome-wide association studies (RTM-GWAS) identified 73 and 38 QTLs with 174 and 88 alleles contributed main effect 40.43 and 26.11% to phenotypic variance (PV) and QTL-environment interaction (QEI) effect 24.64 and 10.35% to PV for relative root length (RRL) and relative shoot length (RSL), respectively. The DT traits were characterized with high proportion of QEI variation (37.52-41.65%), plus genetic variation (46.90-58.40%) in a total of 88.55-95.92% PV. The identified QTLs-alleles were organized into main-effect and QEI-effect QTL-allele matrices, showing the genetic and QEI architecture of the three parents/NAM population. From the matrices, the possible best genotype was predicted to have a weighted average value over two indicators (WAV) of 1.873, while the top ten optimal crosses among RILs with 95th percentile WAV 1.098-1.132, transgressive over the parents (0.651-0.773) but much less than 1.873, implying further pyramiding potential. From the matrices, 134 candidate genes were annotated involved in nine biological processes. The present results provide a novel way for molecular breeding in QTL-allele-based genomic selection for optimal cross selection.