Mapping Four-Seed Pods in Soybeans through a Quasi-Poisson Mixed Model.

Due to its unique seed composition, soybean cultivation is of major importance for the food security in the world. The low heritability has limited the genetic improvement of soybean yield. Trait dissection is proposed to overcome this limitation, splitting yield into more heritable traits, such as seed size, seed per pod, pods per plant, and the number of four-seed pods. A population of 198 recombinant inbreed lines (RILs) derived from two crosses (IA3023xLG05-4317 and IA3023x LG94-1906) was phenotyped for four-seed pods and total number of pods in West Lafayette IN, 2012. RILs were genotyped with 5305 SNP markers. After testing for goodness of fit and marker ability of carrying a gene (threshold 0.8), a subset of 2252 SNPs were deployed in this study. In genome-wide association mapping, quasi-Poisson model is adequate for over-dispersed counting traits such as four-seed pods, which allows offsetting four-seed pods by the total number of pods, here represented by ψ_i. The population structure was controlled by a polygenic term δ_i, which represents the fitted genomic enhanced breeding value (GEBV) of the i^th genotype. For each marker, the LOD score was obtained by testing the log quasi-likelihood of the full model log(E{λ_ij })=μ+log(ψ_i)+δ_i+β_j against the model without the j^th marker parameter, β_j. A significance threshold was generated by 1000 permutations. Genetic regions associated to the number of four-seed pods were found in multiple chromosomes and a major effect QTL in chromosome 20, known to be associated to the Ln gene complex. Results of this study can be employed for marker-assisted selection aiming the improvement of yield components and therefore, soybean yield.