Heterosis Analysis Of Indica×Indica Hybrid Rice

Rice?Oryza sativa L.?is one of the most important food crops of China.How to effectivly use heterosis is the key of improving yield for rice.In present study,under North Carolina design II,115 elite indica rice varieties,including three-way wild-abortive hybrid restorer line and mini-core collection,were separately crossed with four two-way restorer lines and one three-way sterile line to generate 575 tested crosses,and combining ability,heritability and correlations of nine agronomic traits were analyzed.Except for tillers per plant and filled grains per panicle,the general combining ability?GCA?variances of most agronomic traits were extremely significant,and except for tillers per plant,the specific combining ability?SCA?variances of other agronomic traits were extremely significant.The SCA of different traits in the same combination or SCA of same trait in different combinations sharing one parent were different,indicating no consistent relationship between SCA and parental GCA.Therefore,selection of high GCA parental line combining with superior SCA combinations is the key to obtain high yield hybrid rice.Epistasis refers to two or more loci interaction between non-allelic genes,and is an important genetic basis of quantitative traits and heterosis in crops.In this study,we established a variance analysis method for detecting higher-order interactions.To quickly and accurately detect real higher-order interactions,we performed simulation experiments and found that the best threshold for significant second-order interaction selection was P<1×10^-6.We applied the method to analyze nine quantitative traits of rice hybridization?Zhenshan97×Minghui63?,and found 64 second-order interactions.Four main-effect loci reported by Li et al.?2008?also involved in the 64 second-order interactions in our research.We founded 19 loci in our second-order interactions also reported in Li et al.'s literature?2008?,had the digenic effect.However,the 78.3%gene action types of second-order interactions were different from main-effect quantitative trait loci and first-order interactions.The proposed method provides a vital important reference for dissecting the genetic basis of quantitative traits and heterosis in rice.Genomic selection is more efficient than traditional prediction methods in hybrid breeding.The present study investigated the predictive ability of genomic best linear unbiased prediction models for rice hybrids based on the North Carolina mating design II,Under North Carolina design II,112 elite indica rice varieties,were crossed with 5 male sterile line.Using 12 traits of 560?112×5?hybrids,both univariate?UV?and multivariate?MV?prediction analyses,including additive and dominance effects,were performed.Using UV models,the prediction results indicated that including dominance effects could improve the predictive ability for a few traits in rice hybrids.Most traits are stongly influenced by additive effect.In our MV model?MV-ADV?which constructed with auxiliary variates,most traits have higher predictive accuracy than that in UV models.