QTL Analysis of Heterosis for Grain Yield in Inter-subspecific Hybrids of Rice

Abstract

For QTL analysis of heterosis in rice, two sets of RILs both derived from indica-japonica crosses, two sets of BC1F1s backcrossed to indica parents, and two sets of midparent heterosis (MPH) values were used in this study. A RIL population is composed of a set of 166 F8 RILs (MT RILs) derived by single seed descent (SSD) method from a cross between Milyang 23 (M.23, Korean indica-type rice) and Tong 88-7 (T88, a temperate japonica variety). Since heterosis can be measured in F1 plants, each RIL was backcrossed to the indica parent, Milyang 23. The 166 BC1F1 hybrids (M.23 x RILs) were obtained. The other RIL population is a set of 179 F11 RILs (DT RILs) derived by SSD method from a cross between Dasanbyeo (DS, Korean indica-type rice) and TR22183 (TR, a temperate japonica variety). The RILs were backcrossed to the indica parent, Dasanbyeo. The 179 BC1F1 hybrids (DS x RILs) were obtained. Two genetic linkage map with 235 and 162 SSR and STS markers were constructed using MT and DT RILs, respectively. The QTLs for yield and yield related traits were detected from three different phenotype data sets including the RILs, BC1F1s, and MPH data set acquired from the definition of mid-parental heterosis. The QTLs detected in BC1F1s and/or MPH data set was regarded as heterosis QTLs. For eight agronomic traits, a total of 32 and 22 heterosis QTLs were detected in MT and DT populations, respectively. Most QTLs explained moderate proportion of phenotypic variation (R2, < 20%). Eight QTLs in MT population and four QTLs in DT population contributed > 20%. These results suggest that heterosis is a polygenic inherited phenomenon. In most traits, even the individual epistatic QTLs can be explained by a small portion of phenotypic variation (R2, < 10%) the phenotypic variation explained by all the significant digenic interactions for the trait was relatively high, ranged from 4.1 to 59.0% and 6.7 to 52.4% in MT and DT population, respectively. These results suggest that epistatic interactions may play an important role in heterosis. The proportion of additive QTLs and over- /underdominant QTLs were predominant in this study. In MT population, a total of 52 QTLs affecting the eight agronomic traits in the RILs, BC1F1s, and MPH data set were identified and mapped across 12 rice chromosomes. Among 52 QTLs, 36 were revealed with additive QTLs, 6 were dominant and 10 were over- / underdominant QTLs. In DT population, a total of 45 QTLs affecting the eight agronomic traits in the RILs, BC1F1s, and MPH data set were identified and mapped across 12 rice chromosomes except chromosome 11. Among 45 QTLs, 33 QTLs appeared to be additive, 3 appeared to be dominant, and 9 appeared to be over- / underdominant. To measure the heterosis for grain yield in the indica-japonica cross the effect of low spikelet fertility is needs to be offset. Presupposition yield per plant was suggested. The py6 was a major heterosis QTL as detected with large LOD score of 11.2 in BC1F1s and 12.8 in MPH data set, not detected in RILs. This heterosis QTL simultaneously influenced other heterosis for days to heading, spikelet number per panicle, panicle length and spikelet fertility. In this locus heterozygotes showed higher values than homozygotes in presupposition yield per plant, days to heading, panicle length and spikelet number per panicle, but lower values in spikelet fertility. The path analysis revealed that spikelet number per panicle had the highest positive direct effect on grain yield per plant in MT and DT RILs, and spikelet fertility had an absolutely high positive direct effect on grain yield per plant in MT and DT BC1F1s and mid-parent heterosis. The yield component traits, panicle number per hill, spikelet number per panicle, spikelet fertility, and 1000-grain weight had a high positive direct effect on grain yield per plant. Days to heading, culm length, and panicle length had weak direct effect on grain yield per plant, meanwhile had a high indirect effect via yield component traits, especially via spikelet number per panicle. The results from QTL analysis using various inter-subspecific cross populations and midparent heterosis data set and path coefficient analysis will be beneficial to understand the genetic basis of heterosis in rice and breeding high heterosis rice plants.