Joint Linkage Mapping Of Maize Ear Traits

Maize yield is related to the production of world grains and animal products, ear traits directly relate to yield and it’s important to dissect the genetic architecture of maize ear traits. Using linkage mapping, only limited number of QTLs can be detected with low resolution and it is also time consuming to develop the segregating populations. Association mapping has high resolution, but it depends on the germplasm background that maybe lead to high false positive rate. Therefore, it will be better to construct population based on multiple parents that the population structure is clear, but this design needs specific population construction process. In this study, we combined 10 existing recombinant inbred line populations containing 1,948 lines and 14,613 unique recombination bins to map the QTLs affecting ear length, ear weight, cob weight and ear row number. According to the genetic structure, we used the mixed linear model with marker effect as random effect which can estimate the difference between parents and improve detection power. We also used eigenvalue decomposition, matrix determinant lemma and Sherman-Morrison formula to deal with the kinship matrix that reduced the load of computation. We detected 41, 74, 61 and 67 QTLs for ear length, ear weight, cob weight and ear row number, respectively. They account for 86.3% of the phenotypic variation including QTL-by-QTL interactions in average and the QTL-by-environment interactions or QTL-by-QTL interactions affect a little. Comparing the detection power with composite interval mapping model, BPP-based association mapping model and meta analysis model, we found that different models had different advantages in dealing with different types of QTL which means their results are complementary and none can detect all the loci in one model. This genetic design and analysis method provide the opportunity to explore the genetic architecture of plant quantitative traits in a more comprehensive level by combining the existing segregating populations, since this design is more open and flexible.