Dissection of heterosis for yield and related traits using populations derived from introgression lines in rice

Cho Xing,Hongjun Zhng,Hui Wng,Shoo Wei,Binying Fu,Jif Xi, Zefu Li,Yongming Go,*,Guoyou Ye*

aInstitute of Crop Sciences/National Key Facility for Crop Gene Resources and Genetic Improvement,Chinese Academy of Agricultural Sciences, Beijing 100081,China

bRice Research Institute,Anhui Academy of Agricultural Sciences,Hefei 230031,China

cGenetics and Biotechnology Division,International Rice Research Institute,DAPO Box 7777,Metro Manila,Philippines

Dissection of heterosis for yield and related traits using populations derived from introgression lines in rice

Chao Xianga,1,Hongjun Zhanga,1,Hui Wangb,Shaobo Weia,Binying Fua,Jiafa Xiab, Zefu Lib,Yongming Gaoa,*,Guoyou Yec,*

aInstitute of Crop Sciences/National Key Facility for Crop Gene Resources and Genetic Improvement,Chinese Academy of Agricultural Sciences, Beijing 100081,China

bRice Research Institute,Anhui Academy of Agricultural Sciences,Hefei 230031,China

cGenetics and Biotechnology Division,International Rice Research Institute,DAPO Box 7777,Metro Manila,Philippines

ARTICLEINFO

Article history:

Received 23 February 2016

Received in revised form 6 May 2016

Accepted 6 June 2016

Available online 14 June 2016

Rice

Yield and related traits

Introgression lines

Heterosis

Quantitative trait loci

Despite the great success achieved by the exploitation of heterosis in rice,the genetic basis of heterosis is still not well understood.We adopted an advanced-backcross breeding strategy to dissect the genetic basis of heterosis for yield and eight related traits.Four testcross(TC)populations with 228 testcross F1combinations were developed by crossing 57 introgression lines with four types of widely used male sterile lines using a North Carolina II mating design.Analysis of variance indicated that the effects of testcross F1

combinations and their parents were significant or highly significant for most of the traits in both years,and all interaction effects with year were significant for most of the traits. Positive midparent heterosis(HMP)was observed for most traits in the four TC populations in the two years.The relative HMPlevels for most traits varied from highly negative to highly positive.Sixty-two dominant-effect QTL were identified for HMPof the nine traits in the four TC populations in the two years.Of these,22 QTL were also identified for the performance of testcross F1.Most dominant-effect QTL could individually explain more than 10%of the phenotypic variation.Four QTL clusters were observed including the region surrounding the RM9–RM297 region on chromosome 1,the RM110–RM279–RM8–RM5699–RM452 region on chromosome 2,the RM5463 locus on chromosome 6 and the RM1146–RM147 region on chromosome 10.The identified QTL for heterosis provide valuable information for dissecting the genetic basis of heterosis.

?2016 Crop Science Society of China and Institute of Crop Science,CAAS.Production and hosting by Elsevier B.V.This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1.Introduction

Rice(Oryza sativa L.)is the staple food of more than half the world's population[1,2].It is estimated that 40%more rice needs to be produced to feed the increased population by 2025[3].To further improve yield potential through breeding remains a challenge.Exploitation of intra-or inter-subspecific heterosis has been demonstrated to be an effective method forsignificantlyincreasing yield potentialin rice[4,5].Rice hybrids with high yield advantages overinbred cultivars have been developed and widely adopted in many countries,especially in China,where hybrid rice varieties occupy 57%of the rice-growing area[6,7].

Despite the great success achieved by the exploitation of heterosis for improving rice yield[8,9],the genetic basis of the heterosis exhibited in hybrid rice is still not well characterized. Three classical genetic hypotheses(dominance,over dominance, and epistasis)have been proposed as the driving factors for heterosis[10].The validity of each hypothesis seems to depend on genetic makeups oftraits and hybrids in question,as indicated in many recent quantitative trait locus(QTL)mapping studies. For example,Xiao et al.[11]reported that dominance complementation was the major genetic basis of heterosis using the BC1F7progeny of an intersubspecific cross between 9024(indica) and LH422(japonica).Most of the identified QTL for yield and a few yield-component traits had overdominant effects in the study of Yu et al.[12].These authors[12]also found that epistasis played an important role in determining the heterosis observed in the cross Zhenshan 97×Minghui 63.Similarly,the importance of epistasis was reported for heterosis of yield component traits by Li et al.[13]using the F4progeny of a cross between the japonica cultivar Lemontand the indica cultivar Teqing.Using a population of recombinant inbred lines(RILs)derived from the same cross (Lemont×Teqing),Li et al.[14]confirmed that epistasis and overdominance were the main factors in heterosis.Luo et al.[15] conducted a large QTL mapping study using two backcross and two testcross populations by crossing RILs derived from the cross Lemont×Teqing to the two parentallines plus two testers (Zhong 413 and IR64)and further confirmed that epistasis and overdominance were important for heterosis.Similarly,additive and overdominant effects resulting from epistatic loci may have been the primary genetic basis of heterosis in Luo et al. [16].Hua et al.[17]found that overdominance at the single-locus level and all three forms of digenic interaction (additive×additive,additive×dominance,and dominance× dominance)could adequately explain the genetic basis of yield heterosis observed in an elite indica hybrid,Shanyou 63 (Zhenshan 97×Minghui 63)using an immortalized F2population produced by randomly permuted intermating of 240 RILs.The same immortalized F2population was recently genotyped by population sequencing to construct an ultra-dense bin map by Zhou et al.[8].QTL mapping based on the bin map found that overdominance/pseudo-overdominance was the most important contributor to heterosis ofyield,number of grains per panicle,and grain weight.Dominance×dominance interaction played an important role in the genetic basis of heterosis of tillers per plant and grain weight,as well as roles in yield and grain number,and single-locus dominance showed relatively small contributions for all of the traits[18].These results suggested that the relative contributions of the genetic components varied with trait and that the cumulative effects of these components may adequately explain the genetic basis of heterosis.

Introgression lines(ILs)are developed using one parent (recurrent parent,RP)as the genetic background and others as introgression parents(donors)by sequential backcrossing and selfing,and there are no differences between RP and each ILs except the introgressed loci.Consequently,a library(population) of ILs offers an accurate means ofinvestigating the genetic effects of introgression in a relatively uniform and elite background [19,20].Thus,ILs are well suited to defining the core genomic segments influencing target traits and further genetic improvement.Recently,severalstudies have used ILs to identify favorable genes/QTL for heterosis[20–22].Luo et al.[20]tested a set of 265 ILs derived from the indica cultivar Guichao 2 and Dongxiang common wild rice(O.rufipogon Griff.)and found that 71.5%of heterotic loci(HL)showed significantly positive effects,indicating that favorable HL capable of improving agronomic traits were available in O.rufipogon.Xin et al.[21]identified HL using a set of 70 ILs carrying introgressed segments of a japonica cultivar Asominori in the background of an indica cultivar IR24 and corresponding testcross F1populations.A total of 41 HL were detected on the basis of midparent heterosis values using single-point analysis.Of the HL,24 had positive effects and could be used in improving yield potential.

Recently,we developed a large number of ILs using Shuhui 527(SH527),an elite indica hybrid rice restorer line,as a recurrent parent and three high-yielding indica cultivars ZDZ057,Fuhui 838 and Teqing as donor parents.These ILs have been used in genetic analysis of various traits[22].In the present study,we analyzed the genetic effects and main features of HL associated with yield and yield-related traits.We developed four testcross populations by crossing a set of high yielding ILs and tested,in two years, four male sterile lines representing the four most commonly used types of male sterile lines in Chinese hybrid rice breeding programs.

2.Materials and methods

2.1.Plant materials

Shuhui 527(SH527)is an elite indica restorer line of hybrid rice in China.High-yielding BC2F3:4ILs were developed using SH527 as recurrent parent and ZDZ057 and Teqing as donor parent[22]. Four types ofmale sterile lines(MSL)including Xieqingzao A(XA), II-32 A(IIA),Gang 46 A(GA),and Jin 23 A(JA)were selected and crossed as females with the selected ILs and the three parental lines(Shuhui 527,ZDZ057,and Teqing)to generate hybrids based on the North Carolina II mating design.Only 57 ILs were successfully crossed with all MSL.According to the MSL,these testcross F1(TCF1)hybrids were divided into four testcross populations:TCP1(IIA/ILs),TCP2(XA/ILs),TCP3(GA/ILs),and TCP4(JA/ILs).In addition,Xieyou 527(XA×SH527),IIyou 527 (IIA×SH527),Gangyou 527(GA×SH527),and Jinyou 527(JA× SH527)were included as check combinations(CC).

2.2.Phenotypic evaluation

The experiment was conducted in the growing seasons of 2008 and 2009(May–September)at the experimental station of Anhui Academy of Agricultural Sciences,Hefei,China.The 62 parental lines(four maintainer lines,57 ILs,and SH527),four TC populations,and four CC were planted in a randomized complete block design with two replicates.Thirty-day-old seedlings were transplanted into four-row plots(10 plants per row)with a spacing of 26.4 cm within rows and 16.5 cm between rows.

Table 1–Analysis of variance of grain yield and related traits measured in two years.

Five plants per plot were randomly sampled for trait evaluation as described by Zhang et al.[22].The nine traits evaluated included heading date(HD,d),plant height(PH, cm),panicle length(PL,cm),panicle number(PN),filled grain number per panicle(GN),spikelet number per panicle(SN), spikeletfertility(FT,%),1000-grain weight(GW,g)and grain yield per plant(GY,g).The trait measurements of the maintainers were used to represent those of their corresponding MSL.

2.3.Genotyping

Large-scale genotyping using SSR markers was described by Zhang etal.[22].In theirwork,a totalof128 and 144 polymorphic markers were identified for the SH527/ZDZ057 and SH527/ Teqing populations,respectively.From these polymorphic SSR markers,65 and 73 markers evenly distributed throughout the whole genome and covering all polymorphic genomic regions ofthe two populations were chosen for genotyping.Given that 32 markers were common to the two ILs populations,106 markers were used in the present study.These markers were used to genotype the four MSL.The genotype of each TCF1was deduced from its parents.The TCF1genotype was assigned as missing if either of the parental genotypes was heterozygous.

Table 2–Means and ranges of grain yield and related traits of ILs and testcross populations measured in two years.

2.4.Data analysis

Midparent heterosis(HMP),relative midparent heterosis(RHMP), and comparative heterosis(HC)of each testcross combination were calculated according to the following formulae:

HMP＝ F1–MP，RHMP＝ F1–MP（ ）/MP×100，HC＝F1–CC，

where F1is the performance of the F1hybrid,MP is the mean of the two parents,and CC is the trait value of the common checks.

2.5.QTL mapping

To identify QTL underlying heterosis for yield and its related traits,the four testcross populations were used as mapping populations.The HMPvalues were used as heterosis phenotypic values to identify QTL for HMP.QTL analysis was also performed for the testcross F1performance.Single-marker analysis was conducted using SAS PROCGLM(SAS Institute,2010).Aputative QTL was assigned at the significance level of P≤0.005.The genetic effects of the testcross F1were defined as follows:d= HMP=[F1?(IL+MSL)/2],where F1is the performance value of testcross F1,IL is the performance value of introgression lines, and MSL is the performance value of MSL measured using their corresponding maintainer lines.

3.Results

3.1.Performances and variance analysis of heterosis for yield and related traits

Analysis of variance indicated that the effects of ILs(male), maintainer lines(female),and ILs×Maintainers were significant or highly significant in the two years(Table 1)for all traits but PN (ILs)and GN(ILs×Maintainers)and SN(ILs×Maintainers). The effects of year and all interaction effects with year were significant for all traits but PN(Year)and PL,PN,GN,SN,and GY(ILs×Year)and GY(ILs×Maintainers×Year).

The means and ranges of yield and eight related traits measured are shown in Table 2.Wide variations were observed for all the nine traits in the ILs and TC populations in the two years.For the traits of PH,PL,SN,and GY,the TC populations had higher average trait values than the IL populations in both years.PN of the TCP3and GN of the TCP2population were lower than those of the IL populations in 2008.For FT,the TC populations showed lower values than the IL populations in the two years.For HD,three TC populations including TCP1,TCP2,and TCP3showed higher values than the IL populations in 2008 but lower values in 2009.TCP4showed lower values than the IL population in both years.For GW,two TC populations,TCP2and TCP3,showed higher values than the IL population in 2008 but lower values in 2009.TCP1and TCP4showed lower GW than the IL population in the two years.

Heterosis for the nine measured traits in the four TC populations is described in Table 3.Positive midparent heterosis (HMP)in the TC populations in the two years was observed for all traits but PN(TCP1and TCP4in 2008,TCP1and TCP3in 2009), FT(TCP3and TCP4in 2008,TCP1and TCP4in 2009),and GW(TCP4in 2009).Grain yield showed the highest relative midparent heterosis(an average of 52.3%),followed by GN(32.0%),GW (25.0%),PH(16.3%),SN(11.6%),HD(10.2%),PL(9.7%),PN(7.3%),and FT(1.8%)of the four TC populations over the two years.The relative HMPlevels for most traits varied from highly negative to highly positive in the TCF1over the two years.

Table 4–Correlations between mean trait values of ILs and TCF1,and between mean trait values of testcross F1and HMPfor grain yield and related traits in two years.

Table 5–QTL for heterosis of yield and related traits identified in four testcross populations in two years.

Table 5(continued)

3.2.Correlation between traits in ILs and TC populations

The correlations between the phenotypic values of the individual introgression line and those of its TCF1were not significant for most traits(Table 4).In 2008,significant positive correlations were observed for PL and GW in all four TC populations and for GYin TCP2and TCP3(Table 4).In 2009,significant correlation was observed only for SN in TCP1and TCP2(Table 4).Significant positive correlations were observed between the phenotypic values of the TCF1and HMPfor all the nine traits in 2008(Table 4). In 2009,the correlation was not significant for PH,PL,and SN in TCP4(Table 4).

3.3.QTL for heterosis

Phenotypic values of midparent heterosis were used to identify QTL for heterosis.A total of 62 QTL were detected for the nine traits in the four TC populations under two years(Table 5). Most of these QTL individually explained more than 10%of the phenotypic variation.

Eight QTL were identified for HD.Five of them were located on chromosome 2 and the other three on chromosomes 1,7, and 11,respectively.The phenotypic variations explained by individual QTL ranged from 12.9%to 29.3%.Two QTL,qHHD2a and qHHD2b,were identified in two of the TC populations, while the other six QTL were identified in only one of the TC populations.QTL qHHD2a and qHHD2b identified in TCP2in 2009 exerted effects in opposite directions.Similarly,QTL qHHD2b and qHHD2c identified in TCP2in 2008 and exerted effects in opposite directions.

Ten QTL were identified for PH,of which five were located on chromosome 2,two on chromosome 6,and the remaining three on chromosomes 4,7,and 9.The phenotypic variation explained by each QTL ranged from 12.1%to 27.7%.Three QTL, qHPH2b,qHPH2e,and qHPH6,were identified in three,two and two of the TC populations,respectively.The other seven QTL were identified in only one of the populations.QTL qHPH2a and qHPH2b were identified in TCP2in 2009 and showed opposite effects.Similarly,QTL qHPH2b and qHPH2e identified in TCP4in 2008 and showed opposite effects.

Only one QTL for PL was identified in TCP3in 2008.It was located on chromosome 3 and explained 15.2%of the phenotypic variation.

Four QTL associated with PN were mapped to chromosomes 4,5,7,and 9.The phenotypic variations explained by each QTL ranged from10.7%to 22.1%.qHPN4 and qHPN9 were identified in TCP2in 2008 and 2009,respectively.qHPN5 and qHPN7 were identified in TCP4and TCP1in 2009,respectively.

Eleven QTL associated with GN were mapped to chromosomes 1,2,3,8,9,10,and 11.Chromosomes 1,2,10,and 11 were assigned two QTL each.The phenotypic variation explained by the QTL ranged from 10.8%to 25.0%.Six QTL,qHGN1b,qHGN2a, qHGN2b,qHGN3,qHGN8,and qHGN11b,were identified in TCP1in 2008.qHGN9 and qHGN11a were detected in 2008 in TCP3and TCP4,respectively.qHGN10a and qHGN10b were identified in TCP4in 2008.qHGN1a was identified in TCP4in 2009.

For SN,nine QTL were mapped to chromosomes 1,2,3,5,7, and 11.Chromosomes 1,3,and 7 were assigned two QTL each. The phenotypic variation explained by individual QTL ranged from 11.9%to 27.5%.Four QTL,qHSN1a,qHSN1b,qHSN2,and qHSN3b,were identified in TCP1in 2008.qHSN11 was detected in TCP2in 2008.qHSN5 and qHSN7a were identified in TCP3in 2009 and 2008,respectively.qHSN3a and qHSN7b were identified in TCP4in 2008.

Six QTL for FT were mapped to chromosomes 2,6,and 10. Three QTL were on chromosome 2 and two were on chromosome 10.The phenotypic variations explained by individual QTL ranged from 9.9%to 34.7%.QTL qHFT2c was identified in TCP3and TCP4in 2008.qHFT2b and qHFT6 were identified in TCP1in 2008.qHFT2a was detected in TCP2in 2009.qHFT10a and qHFT10b were identified in TCP4in 2008.

Four QTLfor GWwere mapped to chromosomes 1,2,3,and 6. The phenotypic variation explained by individual QTL rangedfrom 12.2%to 15.1%.qHGW1 was identified in TCP2in 2008. qHGW2 was identified in TCP1in 2008.qHGW3 and qHGW6 were identified in TCP1in 2009.

Table 6–QTL for grain yield and related traits identified using the performance of testcross F1in two years.

Table 6(continued)

Nine QTL associated with GY were mapped to chromosomes 1,2,3,4,6,and 10.Chromosome 2 had three QTL and chromosome 6 had two QTL.The phenotypic variation explained by individual QTL ranged from 8.9%to 23.3%.Two QTL,qHGY2a and qHGY3,were identified in TCP1in 2008.Three QTL,qHGY1, qHGY2b,and qHGY4,were identified in TCP2in 2008.Three QTL, qHGY2c,qHGY6a,and qHGY6b,were identified in TCP3in 2009. qHGY10 was identified in TCP4in 2008.

A few chromosomal regions had multiple QTL for the same and/or different traits.They were the region surrounding the RM9–RM297 region in chromosome 1,the RM110–RM279–RM8–RM5699–RM452 region in chromosome 2,the RM5463 locus on chromosome 6,and the RM1146–RM147 region in chromosome 10.

Table 7–The six testcross F1combinations with highest comparative heterosis for grain yield.

4.Discussion

In this study,a total of 62 QTL underlying heterosis for yield and its related traits were detected in four testcross populations across two years.Of these,22 QTL were common to the testcross F1and HMPdata sets(Tables 5,6),which was consistent with the correlation results between TCF1and HMPfor the nine traits.Forty QTL were detected using only HMP,especially for FT and GY (Tables 5 and 6).The QTL for TCF1performance found in the presentstudy,namely RM297 for PH,RM213 for GN,RM213 for SN, and RM3 for GW(Table 5),were detected as well by Zhang et al.[22]using the IL populations.Weak but positive correlations between parental and hybrid performance for GY and yield-related traits were observed(Table 4),suggesting their control by different sets of genes and suggesting the potential to increase the yield of hybrids by increasing the yield of parental lines[23].Most QTL explained more than 10%of the phenotypic variation individually in this study.However,the variation explained by a single QTL for heterosis was less than 10%in most of the heterosis QTL mapping studies in rice [11,13,24,25].The discrepancies among those studies may result from differences in materials,methods and threshold for identifying QTL[26].Of the 62 QTL for heterosis,only one was identified in multiple years,suggesting that QTL underlying heterosis are influenced by environment and QTL-by-environment interaction effects,as reported in previous studies[26,27]. Six of the 62 QTL for heterosis were detectable in only one of the four populations in the present study,a finding consistent with the finding that MSL effects were significant for most of the traits.SeveralQTL for some traits could be identified in different TC populations,including qHHD2a and qHHD2b for HD,qHPH2b, qHPH2e,and qHPH6 for PH,and qHFT2c for FT.The effect directions of these QTL in different TC populations was the same in some cases(qHHD2a and qHFT2c),and different in the other cases(qHHD2b,qHPH2b,qHPH2e,and qHPH6).A similar phenomenon has been observed in previous studies[25–27].The allele difference,the presence of multiple alleles,and dominance and epistasis effects may account for the variation in magnitude and direction of the heterosis effect[18,27].

The validity of some QTL underlying heterosis for grain yield and related traits in the present study is supported by previous studies.qHPH2d,qHGN2b,qHSN2,and qHGY2c were colocalized with the previously reported QTL,D-hd2[26].qHGN11b for filled grain number per panicle was colocalized with hsp11[21].qHGW1 for 1000-grain weight was colocalized with QGw1[16]and hgw1 [21].qHGY2a for grain yield mapped to the same chromosomal region as the previously reported QTL QYp2[16].Specifically,the regions surrounding markers RM110 and RM279 on chromosome 2,where QTL for several traits were detected,coincided with the location of qGY2-1,a yield-enhancing QTL mapped by Li et al.[28] and cloned by Zha etal.[29].The location of qHGN11b near RM254 on chromosome 11 coincided with that of hsp11,a heterotic locus that increased the number of spikelets per panicle[16,21].In addition,a few chromosomal regions harboring multiple QTL for the same or different traits were observed.They were the RM9–RM297 region on chromosome 1,the RM110–RM279–RM8–RM5699–RM452 region on chromosome 2,the RM5463 locus on chromosome 6,and the RM1146–RM147 region on chromosome 10.Similar concentrated distributions of QTL have also been observed in many other studies[16,21,26,28,30].The results observed in this study suggested that particular attention should be paid to such markers and QTLs,which are viable candidates for marker-assisted improvement of rice yield potential in future studies.

In this study,we analyzed the genetic effects and main features of HL associated with yield and yield-related traits in testcross F1combinations derived from a set of ILs.Advanced back cross progeny such as ILs have been widely used in QTL mapping studies of crops[31].ILs offer a precise estimate of genetic effects of introgression in a relatively uniform and elite lineage background[20,21].They thus allow reliable definition of core genomic segments for target traits and further genetic improvement.The present study shows that ILs derived from multiple donors test-crossed with elite MSL could facilitate the identification and transfer of useful QTL for heterosis.About 45% (28 of 62)of QTL associated with midparent heterosis showed significantly positive heterotic effects(P≤0.005)on yield-related traits.In particular,the QTL for GY,FT,and GN showed positive heterosis effects.For all nine traits evaluated,we identified QTL with alleles of the donor parents contributing to improved hybrid performance when transferred to the recurrent parent SH527.For instance,the ZDZ057 alleles of qHPN7 and qHPH6 detected in the testcross F1combinations II-32 A/WD-36 and Jin23A/WD-25, Teqing alleles of qHGY4 and qHPN4 detected in the testcross F1combination Xieqingzao A/WD-110,and Teqing alleles of qHSN7b detected in Jin23A/WD-126 all showed significantly higher comparative heterosis of GY over two years(Table 7).These results suggested that partially substituting a segment of chromosome ofan elite parentalline(such as SH527)by choosing donors could lead to the development of new breeding lines with markedly increased heterosis.

Acknowledgments

This study was funded by the National High Technology Research and Development Program of China(No.2014AA10A604)and the Shenzhen Municipal Peacock Plan for introducing high-level overseas talents.

[1]G.S.Khush,What it will take to feed 5.0 billion rice consumers in 2030,Plant Mol.Biol.59(2005)1–6.

[2]Q.F.Zhang,R.Wing,Genome studies and molecular genetics: understanding the functional genome based on the rice model,Curr.Opin.Plant Biol.16(2013)129–132.

[3]Economic and social development department,FAO,summary report.World agriculture:towards 2015/2030,http://www.fao. org/docrep/004/y3557e/y3557e00.htm(accessed Jan.3,2016).

[4]S.B.Peng,G.S.Khush,P.Virk,Q.Y.Tang,Y.B.Zou,Progress in ideotype breeding to increase rice yield potential,Field Crops Res. 108(2008)32–38.

[5]X.J.Wu,Prospects of developing hybrid rice with super high yield,Agron.J.101(2009)688–695.

[6]G.S.Khush,Green revolution:the way forward,Nat.Rev.2 (2001)815–822.

[7]L.P.Yuan,Development of hybrid rice to ensure food security, Rice Sci.21(2014)1–2.

[8]G.Zhou,Y.Chen,W.Yao,C.J.Zhang,W.B.Xie,J.P.Hua,Y.Z. Xing,J.H.Xiao,Q.F.Zhang,Genetic composition of yield heterosis in an elite rice hybrid,Proc.Natl.Acad.Sci.U.S.A. 109(2012)15847–15852.

[9]X.H.Huang,S.H.Yang,J.Y.Gong,Y.Zhao,Q.Feng,H.Gong,W.J. Li,Q.L.Zhan,B.Y.Cheng,J.H.Xia,N.Chen,Z.N.Hao,K.Y.Liu,C.R. Zhu,T.Huang,Q.Zhao,L.Zhang,D.L.Fan,C.C.Zhou,Y.Q.Lu,Q.L. Weng,Z.X.Wang,J.Y.Li,B.Han,Genomic analysis of hybrid rice varieties reveals numerous superior alleles that contribute to heterosis,Nat.Commun.6(2015)1–9.

[10]Z.J.Chen,Genomic and epigenetic insights into the molecular bases of heterosis,Nat.Rev.Genet.14(2013)471–482.

[11]J.H.Xiao,J.M.Li,L.P.Yuan,S.D.Tanksley,Dominance is the major genetic basis of heterosis in rice as revealed by QTL analysis using molecular markers,Genetics 140(1995)745–754.

[12]S.B.Yu,J.X.Li,G.C.Xu,Y.F.Tan,Y.J.Gao,X.H.Li,Q.F.Zhang, M.A.Saghai Maroof,Importance of epistasis as the genetic basis of heterosis in an elite rice hybrid,Proc.Natl.Acad.Sci. U.S.A.94(1997)9226–9231.

[13]Z.K.Li,S.R.Pinson,W.D.Park,A.H.Paterson,J.W.Stansel, Epistasis for three grain yield components in rice(Oryza sativa L.),Genetics 145(1997)453–465.

[14]Z.K.Li,L.J.Luo,H.W.Mei,D.L.Wang,Q.Y.Shu,R.Tabien,D.B. Zhong,C.S.Ying,J.W.Stanse,G.S.Khush,A.H.Paterson, Overdominant epistatic loci are the primary genetic basis of inbreeding depression and heterosis in rice:I.Biomass and grain yield,Genetics 158(2001)1737–1753.

[15]L.J.Luo,Z.K.Li,H.W.Mei,D.L.Wang,Q.Y.Shu,R.Tabien,D.B. Zhong,C.S.Ying,J.W.Stanse,G.S.Khush,A.H.Paterson, Overdominant epistatic loci are the primary genetic basis of inbreeding depression and heterosis in rice:II.Grain yield components,Genetics 158(2001)1755–1771.

[16]X.J.Luo,Y.C.Fu,P.J.Zhang,S.Wu,F.Tian,J.Y.Liu,Z.F.Zhu,J.S. Yang,C.Q.Sun,Additive and over-dominant-effects resulting from epistatic loci are the primary genetic basis of heterosis in rice,J.Integr.Plant Biol.51(2009)393–408.

[17]J.P.Hua,Y.Z.Xing,W.R.Wu,C.G.Xu,X.L.Sun,S.B.Yu,Q.F.Zhang, Single-locus heterotic effects and dominance by dominance interactions can adequately explain the genetic basis of heterosis in an elite rice hybrid,Proc.Natl.Acad.Sci.U.S.A.100 (2003)2574–2579.

[18]S.D.Tanksley,J.C.Nelson,Advanced backcross QTL analysis: a method for the simultaneous discovery and transfer of valuable QTLs from unadapted germplasm into elite breeding lines,Theor.Appl.Genet.92(1996)191–203.

[19]Z.K.Li,B.Y.Fu,Y.M.Gao,J.L.Xu,J.Ali,H.R.Lafitte,Y.Z.Jiang, J.D.Rey,C.H.Vijayakumar,R.Maghirang,T.Q.Zheng,L.H. Zhu,Genome-wide introgression lines and their use in genetic and molecular dissection of complex phenotypes in rice(Oryza sativa L.),Plant Mol.Biol.59(2005)33–52.

[20]X.J.Luo,S.Wu,F.Tian,X.Y.Xin,X.J.Zha,X.X.Dong,Y.C.Fu, X.K.Wang,J.S.Yang,C.Q.Sun,Identification of heterotic loci associated with yield-related traits in Chinese common wild rice(Oryza rufipogon Griff.),Plant Sci.181(2011)14–22.

[21]X.Y.Xin,W.X.Wang,J.S.Yang,X.J.Luo,Genetic analysis of heterotic loci detected in a cross between indica and japonica rice(Oryza sativa L.),Breed.Sci.61(2011)380–388.

[22]H.J.Zhang,H.Wang,Y.L.Qian,J.F.Xia,Z.F.Li,Y.Y.Shi,L.H. Zhu,J.Ali,Y.M.Gao,Z.K.Li,Simultaneous improvement and genetic dissection of grain yield and its related traits in a backbone parent of hybrid rice(Oryza sativa L.)using selective introgression,Mol.Breed.31(2013)181–194.

[23]Y.Z.Xing,Y.F.Tan,J.P.Hua,X.L.Sun,C.G.Xu,Q.F.Zhang, Characterization of the main effects,epistatic effects and their environmental interactions of QTLs on the genetic basis of yield traits in rice,Theor.Appl.Genet.105(2002) 248–257.

[24]L.Z.Li,K.Y.Lu,Z.M.Chen,T.M.Mu,Z.L.Hu,X.Q.Li,Dominance, overdominance and epistasis condition the heterosis in two heterotic rice hybrids,Genetics 180(2008)1725–1742.

[25]A.Q.You,X.G.Lu,H.J.Jin,X.Ren,K.Liu,G.C.Yang,H.Y.Yang,L.L. Zhu,G.C.He,Identification of quantitative trait loci across recombinant inbred lines and testcross populations for traits of agronomic importance in rice,Genetics 172(2006) 1287–1300.

[26]Z.Qu,L.Z.Li,J.Y.Luo,P.Wang,S.B.Yu,T.M.Mou,X.F.Zheng, Z.L.Hu,QTL mapping of combining ability and heterosis of agronomic traits in rice backcross recombinant inbred lines and hybrid crosses,PLoS One 7(2012),e28463.

[27]C.Y.Yu,L.Jiang,Y.H.Xiao,H.Q.Zhai,J.M.Wan,Combining ability of yield component traits for India chromosome substituted segements in japonica hybirds,Acta Agron.Sin.34 (2008)1308–1316(in Chinese with English abstract).

[28]D.J.Li,C.Q.Sun,Y.C.Fu,L.Cheng,Z.F.Zhu,L.Chen,H.W.Cai,X.K. Wang,Identification and mapping of genes for improving yield from Chinese common wild rice(O.rufipogon Griff.)using advanced backcross QTL analysis,Chin.Sci.Bull.47(2002) 1533–1537.

[29]X.J.Zha,X.J.Luo,X.Y.Qian,G.M.He,M.F.Yang,Y.Li,J.S.Yang, Over-expression of the rice LRK1 gene improves quantitative yield components,Plant Biotechnol.J.7(2009)611–620.

[30]S.L.Ahuja,L.S.Dhayal,Combining ability estimates for yield and fiber quality traits in 4×13 line×tester crosses of Gossypium hirsutum,Euphytica 153(2007)87–98.

[31]D.Zamir,Improving plant breeding with exotic genetic libraries,Nat.Rev.Genet.2(2001)983–989.

*Corresponding authors.

E-mail addresses:irriygao@126.com(Y.Gao),g.ye@irri.org(G.Ye).

Peer review under responsibility of Crop Science Society of China and Institute of Crop Science,CAAS.1These authors contributed equally to this work.

http://dx.doi.org/10.1016/j.cj.2016.05.001

2214-5141/?2016 Crop Science Society of China and Institute of Crop Science,CAAS.Production and hosting by Elsevier B.V.This is an open access article under the CC BY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).