D.-B.Yoon ÆK.-H.Kang ÆH.-J.KimH.-G.Ju ÆS.-J.Kwon ÆJ.-P.Suh ÆO.-Y.Jeong S.-N.AhnMapping quantitative trait loci for yield components and morphological traits in an advanced backcross population between Oryza grandiglumis and the O .sativa japonica cultivar HwaseongbyeoReceived:8March 2005/Accepted:4January 2006/Published online:24January 2006ÓSpringer-Verlag 2006Abstract Introgression has been achieved from wild species Oryza grandiglumis (2n =48,CCDD,Acc.No.101154)into O.sativa subsp.japonica cv.Hwaseongb-yeo as a recurrent parent.An advanced introgression (backcross)line,HG101,produced from a single plant from BC 5F 3families resembled Hwaseongbyeo,but it showed differences from Hwaseongbyeo in several traits,including days to heading and culm length.To detect the introgressions,450microsatellite markers of known chromosomal position were used for the parental survey.Of the 450markers,51(11.3%)detected O.grandiglumis segments in HG101.To characterize the effects of alien genes introgressed into HG101,an F 2:3population (150families)from the cross Hwaseongbyeo/HG101was developed and evaluated for 13agronomic traits.Several lines outperformed Hwaseongbyeo in several traits,including days to heading.Genotypes were deter-mined for 150F 2plants using simple sequence repeat markers.Qualitative trait locus (QTL)analysis was carried out to determine the relationship between mar-ker genotype and the traits evaluated.A total of 39QTL and 1gene conferring resistance to blast isolate were identified using single-point analysis.Phenotypic variation associated with each QTL ranged from 4.2to 30.5%.For 18(46.2%)of the QTL identified in this study,the O.grandiglumis -derived alleles contributed a desir-able agronomic effect despite the overall undesirablecharacteristics of the wild phenotype.Favorable wild alleles were detected for days to heading,spikelets per panicle,and grain shape traits.Grain shape QTL for grain weight,thickness,and width identified in the F 2:3lines were further confirmed based on the F 4progeny test.The confirmed locus,tgw2for grain weight is of particular interest because of its independence from undesirable height and maturity.Several QTL controlling amylose content and grain traits have not been detected in the previous QTL studies between Oryza cultivars,indicating potentially novel alleles from O.grandiglumis .The QTL detected in this study could be a rich source of natural genetic variation underlying the evolution and breeding of rice.IntroductionThe genus Oryza has 2cultivated and 22diploid or allotetraploid wild species (Vaughan et al.2003).The wild species of Oryza are an important reservoir of ge-netic variability for agronomic traits such as biotic and abiotic stresses and for improved yield potential (Brar and Khush 1997;Xiao et al.1998;Moncada et al.2001).The cultivated species Oryza sativa and O.glaberrima share a common AA genome with closely related wild species,and these wild species can easily be crossed with O.sativa by conventional crossing methods.Valuable genes from AA genome wild species could be transferred into the cultivated rice (Xiao et al.1998;Moncada et al.2001;Nguyen et al.2003).However,it is difficult to transfer genes from wild species with other genomes into cultivated rice genotypes because of the crossability and recombination barriers.Nevertheless,several beneficial genes have been successfully transferred from distantly related species such as O.officinalis (CC),O.australi-ensis (EE),and O.minuta (BBCC)using advanced techniques of tissue culture and chromosome manipu-lation (Jena et al.1992;Ishii et al.1994;Brar and Khush 1997;Liu et al.2002;Jena et al.2003).The possibleCommunicated by Q.ZhangD.-B.Yoon ÆH.-J.Kim ÆH.-G.Ju ÆS.-N.Ahn (&)Department of Agronomy,College of Agriculture &Life Sciences,Chungnam National University,305-764Daejeon,Korea E-mail:ahnsn@cnu.ac.kr Fax:+82-42-8222631K.-H.Kang ÆJ.-P.Suh ÆO.-Y.JeongNational Institute of Crop Science,RDA,441-100Suwon,Korea S.-J.KwonNational Institute of Agricultural Biotechnology,RDA,441-707Suwon,KoreaTheor Appl Genet (2006)112:1052–1062DOI 10.1007/s00122-006-0207-4mechanism of gene transfer was by restricted reciprocal recombination detectable only at the molecular level (Jena et al.1992).An allotetraploid CCDD genome species,O.gran-diglumis is endemic to Central and South America.This species has large panicles and has potentially useful genes for higher biomass production and resistance to diseases and insect pests(Heinrichs et al.1985;Yu et al. 1997;Vaughan et al.2003).A few studies have been conducted to detect genetic variability between O. grandiglumis and O.glumaepatula(Akimoto and Mori-shima2003),and among three CCDD genome species endemic to Central and South America at the molecular level(Jena and Kochert1991;Wang et al.1992).How-ever,attempts to transfer genes from O.grandiglumis to cultivars have been limited mainly due to incompatibil-ity between unrelated genomes(Yu et al.1997).Recent progress in plant genome analysis has made it possible to examine naturally occurring allelic variation underlying complex traits.Quantitative trait loci(QTL) analysis can provide information relevant to agricultural traits by using molecular markers to identify specific regions of the genome affecting any measurable trait (Tanksley1993).Several QTL,including yield,yield components,and morphological traits from O.sativa cultivars,have been identified in the past decade (McCouch and Doerge1995;Yano and Sasaki1997). Also,map-based cloning of QTL in tomato and rice confirm the feasibility of using QTL to uncover the molecular nature of quantitative variation(Frary et al. 2000;Yano et al.2000).This study was carried out using an introgression line,‘‘HG101’’,developed from the cross between O.sativa cv.Hwaseongbyeo and O.grandiglumis,(1)to identify the regions of the genome related to the agronomic traits of interest,(2)to identify the trait-improving alleles from O.grandiglumis and selectively transfer novel alleles into the background of Hwaseongbyeo,and(3)to determine the conservation of QTL across other rice cultivars.To our knowledge,this study represents thefirst re-port of a genetic study conducted on an interspecific cross of O.sativa/O.grandiglumis.Materials and methodsPlant materials andfield trialsInterspecific backcross progenies were developed from the cross between the elite japonica cultivar Hwaseongbyeo used as a recurrent parent and the wild species O.grandiglumis(IRGC Acc.No.101154)with the CCDD genome as a donor parent.Among them,one BC5F6introgression line(HG101)was selected.HG101 was produced from a single plant from BC5F3families, which was subsequently self-pollinated for three more generations.HG101differed from Hwaseongbyeo in several traits,including days to heading,and these differences can be attributed to the O.grandiglumis chromosome segments introgressed into HG101.To evaluate the effects of these segments on the agronomic traits,150F2:3families were developed from the cross Hwaseongbyeo/HG101.From150F3families,1family,CR1242,was selected as the starting material for detailed charac-terization and NIL development of the grain weight QTL tgw2.This family was selected because it segre-gated at the target loci and had a significantly bigger grain weight than the other lines,and a relatively few nontarget background introgressions.This family was screened for two parent type homozygotes to produce 22F4lines(6for Hwaseongbyeo and16O.glandiglumis alleles,respectively)for evaluation of traits including grain weight.Phenotypic evaluationOne hundred andfifty F3families were grown in thefield during the summer of2002and2003at Chungnam National University,Deajeon,Korea.Twenty-two F4 lines were grown during the summer of2004.Each family was represented by a single row of3030-day-old seedlings planted with15cm between plants and30cm between rows in a randomized complete block design with two replications.A total of13agronomic traits were evaluated in the150lines.Three traits(blast resistance,grain yield,and amylose content)were eval-uated in2002.The other ten traits(days to heading, culm length,panicle number,panicle length,spikelets per panicle,grain length,grain width,grain thickness, length/width ratio of grain,and1,000-grain weight)were evaluated in2002and2003,respectively.Twenty-two F4 lines were evaluated for the1,000-grain weight.One qualitative trait,resistance to blast isolate,was included.Fungal inoculation and disease evaluation of the150F2:3lines(20plants per line)to one blast iso-late,KI-313,were carried out as described in Kwon et al.(2002).The other12traits were quantitatively inherited and evaluated based on the commonly used methods in rice genetics and breeding.Days to heading (DTH)was evaluated as the number of days from seeding until50%of the panicles of the30plants flowered.For culm length,panicle length,and panicles per plant,ten plants per line in the middle were mea-sured,and the average of the measurements was used as the phenotype of each line.Culm length(CL)was measured in centimeters from the soil surface to the neck of tallest panicle.Panicle length(PL)was mea-sured in centimeters from panicle neck to the panicle tip.Panicle number(PN)was calculated as the number of panicles per plant.Two representative panicles per plant(a total of20panicles per line)were bagged after panicle emergence to avoid seed shattering and evalu-ated for spikelets per panicle and percent seed set. Spikelets per panicle(SPP)was calculated as the aver-age number of spikelets per panicle.Grain length(GL), grain width(GW),grain thickness(GT),length/width1053ratio of grain(LW),and1,000-grain weight(TGW) were measured on50grains per plant(10plants per line).Yield per plant(YD)was measured and calculated as the average weight per plant of bulked grain har-vested from the ten plants per line.The1,000-grain weight and the yield per plant were corrected for the 12%grain moisture content.Amylose content(AC)was measured as described by Williams et al.(1958)and Juliano(1971)using a sample of100g100-mesh-sieved riceflour.Means for the two replications were calcu-lated for each trait and used in data analysis.Twenty-two F4lines from the CR1242were grown during the summer of2004.Each line was represented by a single row of3030-day-old seedlings planted with 15cm between plants and30cm between rows in a randomized complete block design with two replica-tions.The22F4lines were evaluated for traits using the aforementioned.DNA extraction and SSR genotypingDNA was extracted from fresh leaves of150F2plants. DNA extraction was performed as described in Causse et al.(1994).A total of450simple sequence repeat(SSR) markers of known chromosomal position were used to survey the parents for polymorphism(Temnykh et al. 2000,2001;McCouch et al.2002).The51markers showing polymorphism between Hwaseongbyeo and HG101were used for genotype analysis of150F2plants. SSR analysis was performed according to the method described in Panaud et al.(1996),with the following modification in the PCR profile:94°C for5min,fol-lowed by35cycles of94°C for1min,55°C for1min, and72°C for1min,and lastly5min at72°C.The linkage map with the order and distance between markers was constructed based on two previously developed SSR maps(Temnykh et al.2001;McCouch et al.2002).When the marker segregation data was analyzed,deviation from the expected Mendelian ratios of1Hwaseongbyeo homo:2hetero:1HG101homo al-leles was examined by Chi-square tests.DNA was extracted from fresh leaves of22F3plants of the CR1242population,and their SSR genotypes of the target loci were determined.QTL analysisThe chromosomal location of QTL was determined by single-point analysis(SPA).The primarily analysis using SPA was performed using QGene(Nelson1997).In SPA,QTL was declared if the phenotype was associated with a marker locus at P<0.005or with two adjacent marker loci at P<0.05.Due to limited introgressions from O.grandiglumis into HG101and limited number of polymorphic markers between the parents, Hwaseongbyeo and HG101,interval mapping analysis could not be performed.The proportion of observed phenotypic variation attributable to a particular QTL was estimated by the coefficient of determination(R2). The total phenotypic variation explained was estimated byfitting a model including all putative QTL for the respective trait simultaneously.Because the QTL de-tected with the2002data were in good agreement with those with the2003phenotypic data,the2002data were used in analysis.The effect of digenic interaction on each trait was analyzed with two-way ANOVA,using the SAS program.The QTL locations from this study were compared with those of the QTL reported in27previous QTL studies across rice cultivars and wild species(http:// ).In total,446previously reported QTL were found for the12traits,and QTL in the similar location on the same chromosome as a QTL in the present study were selected for comparisons.ResultsMorphology and DNA polymorphism of HG101 Among the BC5F5population developed,HG101was the most divergent in plant morphology and DNA polymorphism from Hwaseongbyeo(Yu et al.1997). HG101differed from Hwaseongbyeo in several traits, including culm length and days to heading(Table1; Ahn et al.2001,2003).HG101showed delayed heading and short culm length compared with Hwaseongbyeo. Also,the grain characteristics of HG101were different from those of Hwaseongbyeo.This might be responsible for the high DNA polymorphism of11.3%betweenTable1Comparison of the agronomic characteristics of the parents and HG101Acc.TraitDTH CL PL SH GL GW TGW AC Blast isolate reactionKI-313KI-409KI-1113a Oryza grandiglumis1801744517.61 1.999.9–R R R Hwaseongbyeo10881209 5.03 2.8421.520.5S S SHG10111775219 5.53 3.0826.315.5R R RDTH days to heading,CL culm length,PL panicle length,SH shattering(evaluated using a1to9scale;1=easy,9=hard),GL grain length,GW grain width,TGW1,000-grain weight,AC amylose content,R resistant,S susceptible1054Hwaseongbyeo and HG101(51·100/450=11.3%) which is much higher than the expected ratio of1.56% afterfive continuous backcrosses.The most notable feature of HG101is the absence of undesirable traits of O.grandiglumis,such as grain shattering and tall plant stature.Correlations among traitsSignificant correlations(P<0.05,P<0.01)were observed among many traits.Table2summarizes the correlation coefficients among the12traits evaluated with the F2:3family.Negative correlation was found between dth and pn,and dth and spp.Significant corre-lations were found among the traits determining plant morphology,cl,pl,and spp.Significant correlations were also detected among the traits associated with grain morphology,gl,gt,gw,and lw.For most of the correlations,the direction(+andÀ)and degree of correlation was similar with previous studies(Xiao et al. 1998;Thomson et al.2003).In agreement with previ-ous studies,yield showed positive correlations with culm length,panicle length,and spikelets per panicle. Negative correlations existed between days to heading and panicle length,1,000-grain weight and panicle length,and1,000-grain weight and panicles per plant. Trait distributionThe frequency distributions of phenotypes for each trait in the parents and F3families are shown in Fig.1.A large amount of variation was observed between the two parents for most of the traits.The150F3families showed continuous distribution for all the measured traits except for grain width and grain length/width ra-tio.In most cases,transgressive lines that fell beyond the high or low mean of the two parents were observed. Four traits,cl,pl,pn,and spp showed the most trans-gressive segregation,with the average value of the lines falling outside the means of the two parents.Genetic analysis of blast resistanceHG101was resistant to three Pyricularia grisea isolates that were collected in Korea(Ryu et al.1987).Several independent studies demonstrated that resistance to the isolate in several rice accessions is under dominant gene control and the resistance gene to an isolate KI-313was mapped using molecular markers(Ahn et al.2000;Kim et al.1995).The F1plant of the cross between HG101 and Hwaseongbyeo was resistant to three isolates, indicating that the resistance is under dominant gene control.For the KI-313isolate,segregation analysis based on disease evaluation of the150F2:3lines for the race showed a1:2:1(45:65:40)ratio(v2=3.0, 0.10<P<0.50)for resistant,segregating,and suscepti-ble lines,thereby confirming that a single dominant gene governs the resistance to KI-313in HG101.Linkage analysis showed that the gene,Pi-18(Ahn et al.2000) conferring resistance to blast isolate KI-313was linked to RM144at the distal end of the long arm of chro-mosome11(Fig.2).The linkage relationship of the genes conferring resistance to two isolates remains to be clarified.Marker segregationThe F2plants were genotyped with51markers that detected O.grandiglumis alleles in HG101(Fig.2).Sig-nificant segregation distortion was observed in this study for5(10.2%)marker loci at a significance level of P<0.05.Of these,three marker loci were skewed to-wards the Hwaseongbyeo alleles and two loci were skewed towards the O.grandiglumis alleles(Fig.2). QTL identificationSignificant QTL were identified for12agronomic traits as summarized in Table3.A total of39QTL and1 gene controlling blast resistance were identified usingTable2Significant correlation coefficients(r)among12traits in the F2:3familyTrait DTH CL PL PN AC SPP GT TGW GL GW LW CLÀ0.09PLÀ0.210.29PNÀ0.340.320.11ACÀ0.100.040.10–SPPÀ0.400.260.260.040.06GT0.07À0.14À0.26À0.280.03À0.14TGW0.12À0.10À0.18À0.370.06À0.020.82GLÀ0.03À0.120.07À0.26À0.06À0.020.500.55GW0.02À0.15À0.29À0.350.070.010.880.800.49LWÀ0.050.080.370.20À0.13À0.03À0.63À0.500.17À0.77YDÀ0.030.220.200.030.020.33À0.07À0.02À0.13À0.08À0.02 DTH days to heading,CL culm length,PL panicle length,PN panicle number,AC amylose content,SPP spikelets per panicle,GT grain thickness,TGW1,000-grain weight,YD yield,GL grain length,GW grain width,LW grain length/width ratior=0.159at P=0.05;r=0.208at P=0.011055SPA,and 1to 6QTL were detected for each trait.The phenotypic variation explained by each QTL (R 2)ranged from 4.2to 30.5%.The effect of digenic interaction on the 12traits analyzed with two-way ANOVA,using the SAS program,was not significant at the level of P <0.01.Most of the QTL detected in this study were located in the same or adjacent regions (Fig.2).Days to heading (dth)Four QTL were identified for days to heading in the F 2:3lines.The phenotypic varia-tion explained by each QTL ranged from 4.9to 19.5%.Collectively,these four QTL explained 30.8%of the phenotypic variation.The Hwaseongbyeo alleles con-tributed earliness at three loci,dth3.1,dth3.2,and dth4,and the O.grandiglumis allele contributed earliness at dth6.Culm length (cl)Six QTL,cl1.1,cl1.2,cl4.1,cl4.2,cl11,and cl12were mapped to four chromosomes.At four loci,cl1.1,cl1.2,cl4.1,and cl4.2,the wild alleles decreased culm length,and at the other two loci,the wild alleles increased culm length.Collectively,these six QTL explained 50.6%of the phenotypic variation.The cl1.2is likely to correspond to sd-1,a semi-dwarf gene which has been previously located (Cho et al.1994).Panicle length (pl)One QTL on chromosome 6was significantly associated with panicle length.The wild allele increased panicle length at pl6.Panicle number (pn)One QTL on chromosome 11was detected for panicle number.The O.grandiglumis allele increased panicle number at pn11.Fig.1Frequency distribution of 12agronomic traits in the 150F 3lines (P 1Hwayeongbyeo,P 2HG101).The vertical axis of each figure represents the number of F 2:3lines1056Spikelets per panicle (spp)Five QTL located on chromosomes 2,3,4,6,and 11were identified for spikelets per panicle.The wild allele increased spikelets per panicle at spp6.At the other four loci,the Hwaseongbyeo alleles increased spikelets per panicle.Amylose content (ac)Three QTL located on chromo-somes 3,5,and 7were identified for amylose content.They were clustered in the same regions as QTL associ-ated with the grain trait QTL except for ac5.These three QTL accounted for 11.5%of the phenotypic variation.Grain length (gl)Two QTL located on chromosomes 6and 11were detected for grain length.The phenotypic variation explained by each QTL ranged from 10.2to 15.3%.The wild allele increased grain length at gl6but decreased at gl11.Grain width (gw)Five QTL located on chromosomes 2,3,6,8,and 11were detected for grain width.The phe-notypic variation explained by each QTL ranged from 4.9to 7.7%and the Hwayeongbyeo alleles were associated with greater grain width in all cases except for gw2.Grain thickness (gt)Four QTL were detected for grain thickness.The phenotypic variation explained by the QTL ranged from 4.2to 17.2%.The wild allele in-creased grain thickness at two loci,gt2and gt7and de-creased at the other two loci.1,000-grain weight (tgw)Two QTL were detected for grain weight.The phenotypic variation explained by the QTL ranged from 5.0to 13.4%.The wild allele in-creased grain weight at the tgw2locus but decreased at the tgw11locus.Fig.2A molecular map was developed using SSR markers polymorphic between Hwaseongbyeo and Oryza grandiglumis ;SSR markers in bold mark the O.grandiglumis -specific genomic regions introgressed into HG101.Quantitative trait loci (QTL )detected by single-point analysis were represented at the right of the chromosomes.The QTL detected in the same interval are boxed .The markers showing segregation distortion are indicated in italics (in favor of Hwayeongbyeo),and underlined (in favor of O.grandiglumis )at the significance level of P <0.05.QTL with the desirable allele from O.grandiglumis are underlined .Also,graphical genotype of CR1242is denoted on the chromosomes.Black ,gray ,and blank regions mark the O.grandiglumis -specific,heterozygous for Hwaseongbyeo and O.grandiglumis alleles,and the Hwaseongbyeo genomic regions in CR1242,respectively1057Grain length/width ratio(lw)Five QTL located on chromosomes2,3,6,8,and11were identified for the length/width ratio of the grain,and the phenotypic variation explained by the QTL ranged from8.0to 13.4%.The wild allele increased the ratio at three loci, lw3,lw6,and lw8but decreased at the other loci.Yield per plant(yd)One QTL on chromosome2was detected for grain yield,which explained11.5%of the phenotypic variation.The wild allele decreased grain yield at yd2.Confirmation of QTL for grain shape traitsCR1242,the F3source population for the characteriza-tion of the QTL including tgw2,had large seeds and contained O.grandiglumis introgression across the region marked by RM290and RM550(Fig.2).CR1242also contained16additional background introgressions located on7of the12chromosomes(Fig.2).In22F4 lines derived from CR1242,the difference in1,000-grain weight between lines having an O.grandiglumis intro-gression in the target region and those havingTable3Quantitative trait loci(QTL)for12traits based on single-point analysis in an F3populationTrait QTL Marker Chr.P R2(%)MeanH/H a H/G a G/G a Allele effect b Days to heading dth3.1RM14330.0050 5.3111113113 1.0dth3.2RM16–RM41130.0153 4.9111112114 1.5dth4RM25540.0047 6.7111112114 1.5dth6RM53960.000019.5115113110À2.5Total30.8Culm length cl1.1RM5–RM48810.00099.1868281À2.5cl1.2RM52910.000030.5878377À5.0cl4.1RM335–RM51840.0107 6.0848381À1.5cl4.2RM25540.00268.1858380À2.5cl11RM229–RM21110.0100 4.7808384 2.0cl12RM235120.0040 4.6818385 2.0Total48.7Panicle length pl6RM539–RM12160.00218.219.219.720.00.4 Panicle no.pn11RM224-RM144110.0235 5.61111120.5 Spikelets/panicle spp2RM47520.0037 4.4125122115À5.0spp3RM14330.0040 6.7128118118À5.0spp4RM25540.0032 4.8127120117À5.0spp6RM539–RM12160.0089 6.31131211277.0spp11RM224–RM144110.00387.4121127115À3.0Total15.5Amylose content ac3RM16–RM41130.0496 5.418.618.117.5À0.6ac5RM50750.0047 6.618.618.017.5À0.6ac7RM478–RM42970.00317.718.718.117.6À0.6Total11.5Grain length gl6RM539–RM12160.022410.2 5.01 5.12 5.150.07gl11RM229–RM21110.000015.3 5.18 5.11 5.00À0.09Total18.0Grain width gw2RM290–RM55020.0072 6.7 2.92 2.92 3.020.05gw3RM16–RM41130.0257 4.9 2.96 2.95 2.87À0.05gw6RM50–RM12160.0068 6.7 2.99 2.92 2.89À0.05gw8RM502–RM26480.0105 6.1 2.99 2.92 2.90À0.05gw11RM224–RM144110.00427.7 2.99 2.92 2.89À0.05Total21.3Grain thickness gt2RM290–RM55020.000017.2 1.99 2.08 2.110.11gt6RM50–RM12160.0295 4.8 2.10 2.06 2.05À0.03gt7RM429-RM47870.0253 4.2 2.04 2.08 2.100.03gt11RM224–RM144110.0218 4.8 2.10 2.07 2.05À0.03Total21.01000-grain weight tgw2RM290–RM55020.000013.422.123.824.3 1.1tgw11RM224–RM144110.0243 5.024.023.722.9À0.55Total15.2Grain length/width Ratio lw2RM290–RM55020.000013.4 1.79 1.73 1.71À0.04lw3RM411–RM1630.00068.9 1.71 1.74 1.780.04lw6RM50–RM53960.00158.6 1.71 1.75 1.770.03lw8RM502–RM26480.00248.0 1.71 1.75 1.760.03lw11RM202–RM536110.00248.9 1.75 1.75 1.70À0.03Total32.1Yield yd2RM47520.000111.5590583529À30.5a H/H,H/G,G/G;Hwaseongbyeo homozygotes,Hwaseongbyeo/HG101heterozygotes,and HG101homozygotes,respectivelyb The additive effect is the one associated with substitution of a‘Hwaseongbyeo’allele by the corresponding O.grandiglumis allele 1058Hwaseongbyeo DNA was as large as3.7g(Table4).And the differences in grain thickness,grain width,grain length/width ratio,panicle length,and grain yield/plant between F4lines having an O.grandiglumis introgression in the target region and those having Hwaseongbyeo alleles were significant.However,the differences in traits including days to heading between F4lines having an O.grandiglumis introgression in the target region and those having Hwaseongbyeo alleles were not significant. These results confirmed primary QTL observations based on F3lines and demonstrated that transgressive segre-gants could be identified in subsequent generation.To determine the size of the introgression segment andfine map tgw2,ten additional SSR markers polymorphic be-tween Hwaseongbyeo and O.grandiglumis were mapped in the22F4lines(Fig.3).Six lines were found to befixed for Hwaseongbyeo alleles in the introgressed region defined by markers RM3390-RM13104.The other16 lines homozygous for O.grandiglumis alleles in the target region were homozygous for Hwaseongbyeo at RM7288 and RM13104indicating that recombinations occurred between RM1358and RM7288,and between RM550and RM13104during the backcross breeding program.The precise boundary of the centromere on chromosome2 was resolved to be between13,570,041and13,857,135bp with a total amount of201,035bp by using CentO satellite DNA(IRGSP2005).On the basis of available sequence information(/),the introgressed segment does not appear to harbor the cen-tromere because the SSR marker RM13104is located above the centromere.As indicated in Table4,there were significant differences in TGW between the lines having an O.grandiglumis introgression in the target region and those having Hwaseongbyeo DNA.This allowed us to conclude that tgw2was located in the interval RM7288–RM13104,a region of approximately4.4Mb in size.It should be noted that RM7288and RM13104representedthe outside borders of the introgression.DiscussionOne of the objectives of this study was to identify novel beneficial alleles from O.grandiglumis for yield and yield components.We identified O.grandiglumis alleles related to yield components,including loci for days to heading.As was evidenced for culm length in this study,of particular interest is the demonstration that some of the novel genetic variation observed in the advanced backcross progenies of the interspecific cross involves positive transgressive variation.The fact thatTable4Comparison of means (±SD)for11traits of two genotypic classes at the target loci(RM1358and RM550)in F4linesNumbers in parentheses denote the number of lines examined H/H homozygous Hwaseongbyeo,G/G homo-zygous O.grandiglumisSignificant at*P£0.05,**P£0.01,***P£0.001Trait Genotype F R2 H/H(6)G/G(16)MeanDTH107±0.7108±1.4108±1.20.73 3.5 CL94±2.191±4.292±4.1 4.1517.2 PL23.5±0.522.3±0.722.6±0.813.8**40.8 PN12±2.311±0.611±1.4 4.2117.4 SPP134±17.8143±9.6141±13.1 2.4811.0 TGW22.9±1.226.6±1.025.6±1.226.6***57.1 GL 6.91±0.08 6.91±0.11 6.91±0.100.0030.0 GT 2.22±0.05 2.40±0.05 2.35±0.0945.7***69.6 GW 3.34±0.12 3.68±0.06 3.59±0.1770.46***77.9 LW 2.07±0.06 1.88±0.04 1.93±0.1036.31***64.5 YD21.4±1.524.9±2.824.0±3.07.55*27.4Fig.3Physical map of chromosome2showing introgression fromO.grandiglumis.Gray regions indicate introgressed segments,andthe black circle indicates approximate position of centromere withthe tgw2QTL.Genotype analysis of the F4lines indicated thatrecombinations occurred between RM1358and RM7288,andbetween RM550and RM13104during the backcross breedingprogram between Hwaseongbyeo and O.grandiglumis1059。
