Ultra High Density Markers Reveal The Genetic Basis Of Grain Yield Heterosis In An "Immortalized F₂" Maize Population

Utilization of heterosis is an effective means to increase crop yield, strengthen resistence, and improve quality substancially. Heterosis has been extensively applicated in maize and plant breeders have been tremendously successful at increasing maize grain yield since the adoption of single hybrid in the early20th century. Extensive researches about the genetic basis of heterosis have been conducted for more than a century, but the genetic and molecular basis of the phenomenon remains conjectural. Application of gemonic technology provides novel methods for the basis research of heterosis utilization. Based on the previous studies of dissection of yield heterosis for an elite maize hybrid using a simple sequence repeat (SSR) linkage map and an immortalized population, the present study made futher efforts to research yield heterosis mechanism through construction of high-density single nucleotide polymorphism (SNP) and advanced statistical methods.In this study, the material scources were an "immortalized F2" and recombinant inbred line populations (RILs) derived from an elite hybrid Yuyu22, and the objective traits were grain yield and yield related traits. SNPs cosegregating in two contiguous block borders were lumped as a bin. A high density linkage map consisting of3184bins was used to assess (1) the additive and additive-by-additive effects determined using recombinant inbred lines;(2) the dominance and dominance-by-dominance effects from mid-parent heterosis dataset; and (3) all types of genetic effects in an "immortalized F2" population. Major results from this study were summarized as follows.1. We constructed a ultra-high-density bin map, which could identify more QTLs with higher resolution than a SSR linkage map. The average of genetic distance between flanking markers on the bin map was0.84cM, which was much shorter than9.41cM on the SSR map. The high saturation of markers on the genome indicated the bin map with high density relative to SSR map. First,63QTLs were identified with the bin map for all traits in IF2population that was more than49QTLs identified using the low density map. Second, the bin map improved the resolution of QTL mapping more than12-fold (from an average of10.82cM to an average of0.88cM). Third, the QTLs already detected using the low density SSR map had higher LOD scores when the high density bin map was used. Forth, the total phenotypic variation explained (PVE) by the QTLs detected using the high density bin map was larger than the value using the low density SSR map. The PVE increased from33%-57%to41%-71%across grain yield and yield component traits.2. A total of11QTLs that affected grain yield were mapped at the given value of the LOD score and each accounted for an average phenotypic variance of4.9%. Five QTLs with|d/a|≤1exhibited dominance, whereas the remaining six QTLs exhibited overdominance. Sixteen QTLs that were identified for ear length accounted for average3.5%phenotypic variance each. Six of these QTLs had|d/a|>1, displaying overdominance, the others were of dominance type. The21QTLs controlling100-kernel weight included16dominant QTL, and5overdominant QTL. All of them distributed on the whole genome except chromosome4, each QTL accounted for an average phenotypic variance of3.8%. A total of15QTL were identified for row number, each accounting for an average of4.3%phenotypic variance. Only two of the15QTLs were of the overdominance type. The phenotypic variance explained by all the loci were40.8%,50.3%,68.8%and70.6%for grain yield, ear length, row number and100-kernel weight.3. Dominance degree(d/a) was calculated in two ways:both additive and dominance genetic effects (a and d) were estimated from IF2population; additive effect (a) was estimated from RILs and dominant effect (d) from MPH dataset. The number of dominance bins in both IF2and RIL/MPH datasets on the whole genome accounted for58%and55%for grain yield,59%and70%for ear length,68%and85%for100-kernel weight,75%and85%for row number. When the bins that did not exceed the given LOD score threshold were excluded,81%bins for grain yield and>90%bins for yield component traits significantly associated with traits were of the dominance type for all traits. The bin map showed that, among all traits, dominance was more important to heterosis than other genetic effects. The importance of overdominance or pseudo-overdominance was proportional to the amount of heterosis. In addition, epistasis contributed to heterosis as well.4. We simulated phenotypes and genotypes in IF2population based on four genetic architectures (VD/VA=0.5,1,2,3). The QTL mapping results manifested that92.2%-96.7%true positive QTL in MPH dataset were also detected in the IF2populations, in which68.9-90.5%QTL were of overdominance. Although the QTL for performance of grain yield did not overlapped with heterotic loci in the previous study, the simulations at present study indicated heterotic loci by mid-parental heterosis were not independent, but part of QTL controlling grain yield in immortalized F2population..