SNP-based linkage mapping for validation of adult plant stripe rust resistance QTL in common wheat cultivar Chakwal 86

Qingdong Zeng,Jinhui Wu,Shuo Hung,Fengping Yun,Shengjie Liu,Qilin Wng,Jingmei Mu,Shizhou Yu,Li Chen,Dejun Hn,*,Zhensheng Kng,*

a State Key Laboratory of Crop Stress Biology for Arid Areas,Northwest A&F University,Yangling 712100,Shaanxi,China

b Extension Center for Agricultural Technology,Agriculture Department of Tibetan Autonomous Region,Lasa 850000,Tibet,China

Keywords:Genetic linkage Haplotype analysis QTL mapping Puccinia striiformis Single nucleotide polymorphism Triticum aestivum

A B S T R A C T Wheat crops in China are constantly challenged by stripe rust.Deployment of cultivars with diverse resistances is the best strategy to control the disease.A recombinant inbred line(RIL)population derived from a cross between the resistant cultivar Chakwal 86 and susceptible landrace Mingxian 169 was studied in multiple environments to examine the underlying genetics and to identify quantitative trait loci(QTL)for stripe rust resistance.One hundred and twenty-eight RILs were genotyped with wheat 35K SNP array and a genome-wide linkage map with 1480 polymorphic SNP markers,or bins,was constructed.Two major QTL on chromosomes 1BL and 3BS,and one minor QTL on 6BS had significant effects in reducing stripe rust severity.The QTL were validated using composite interval mapping(CIM)and inclusive composite interval mapping(ICIM).These methods explained 59.0%-74.1%of the phenotype variation in disease response.The QTL on chromosome 1BL was confirmed to be Yr29/Lr46 and the one on 3BS was the resistance allele identified in CIMMYT germplasm but was not Yr30/Sr2.The QTL on 6BS probably corresponded to previously known QTL.RILs with combined QTL were more resistant than those with single or no QTL.Kompetitive allele-specific PCR(KASP)assays for the QTL with largest effect QTL on chromosome 3BS were performed on a subset of RILs and 150 unrelated wheat lines.The QTL on 3BS with its linked KASP markers can be used in marker-assisted selection to improve stripe rust resistance in breeding programs.

1.Introduction

Common wheat(Triticum aestivum L.)is an important cereal crop and serves as a staple food source in many countries[1].However,it is frequently challenged by diseases[2],among which stripe or yellow rust(YR),caused by Puccinia striiformis Westend.f.sp.tritici Erikss.(Pst),is one of the most damaging[3,4].Stripe rust is especially damaging in regions with mild winters and cool and wet conditions during late spring and early summer.In China the disease occurs periodically in almost all winter wheat growing regions and severe epidemics have caused grain losses of several million tons in a single year[5-7].The most serious cause of epidemics is loss of effectiveness of widely deployed resistance genes following the occurrence and build-up virulent races.The most recent example of this is the emergence of race CYR34 and other races with virulence to Yr26 combined with virulence for Yr10 and several other resistance genes[8,9].

Although effective fungicide control strategies are available,these have not been totally effective due to management and financial factors[10].Resistance is a more environmentally friendly and proven means of controlling stripe rust,but it often fails after a period of effectiveness.To address this problem pathologists and breeders have been seeking sources of resistance with greater durability.Resistance to the rusts is generally described in two categories:race-specific and race non-specific resistance.The first category is often described as seedling or all-stage resistance(ASR)and the second is adult-plant resistance(APR)or partial resistance based on growth stage when resistance is expressed[11]or on the degree protection that is conferred.ASR is usually qualitative and controlled by single or a few major genes and it is highly effective throughout the entire growth cycle.Hence,this type of resistance is more attractive to farmers and is often also preferred by breeders.The problem is that it may not be durable and previously highly resistant varieties become susceptible[12,13].On the contrary,APR is expressed mostly at the post-seedling stages,and is generally controlled by a few qualitatively inherited genes.This type of resistance usually confers partial resistance,but to all races,and is enhanced when the underlying genes are combined.With increasing understanding,and a greater likelihood of durability,APR is gradually being accepted by breeders and extension workers[11,14].APR genes such as Yr18,Yr29,and Yr46 show additive effects and provide adequate levels resistance when deployed in combination[15-18].

Although over 300 genes or QTL for resistance to stripe rust on all of the 21 chromosomes have been reported[19-21],not all have efficient molecular markers that can be used in marker-assisted selection(MAS)in breeding programs without need for testing in disease nurseries.Identification of resistance genes or QTL with user-friendly markers is an ongoing process.Next generation sequencing(NGS)and highthroughput SNP-based genotyping technologies have revolutionized wheat research[22].Over the past few years,there have been significant advances in the reference genome assembly of hexaploid wheat cv.Chinese Spring[23,24].NGS enables efficient high throughput discovery of DNA variants in the wheat genome,and new generation markers,based on single nucleotide polymorphisms(SNP)provide high resolution of genetic diversity[25].Different genotyping platforms with high-density markers such as the Illumina Bead Array and Affymetrix Gene Chip,Kompetitive Allele-Specific PCR(KASP)have become available for both genetic studies and breeding[26].The recently designed Axiom Wheat35K SNP array,also known as the‘Wheat Breeders Array',contains 35,143 SNP probes selected from Wheat820K SNP array.This new array with a large number of genome-wide polymorphic co-dominant markers for genotyping at lower cost and less computational load has been used in some QTL studies[27,28].

In a previous study,＞1000 common wheat germplasms were screened for resistance to stripe rust in controlled greenhouse conditions and in field nurseries at Yangling(Shaanxi province)and at Tianshui(Gansu province)since 2008,and the lines with resistance to the main prevalent Chinese Pst races were identified[29,30].The Pakistan wheat cultivar Chakwal 86 with the pedigree NS-900/Anahuac-75 was released in 1986[31].Chakwal 86 is known to have combinations of genes for resistance to leaf rust(Lr10,Lr17a,and Lr27[32])and stem rust(Sr9a,http://wheatpedigree.net/sort/show/12120).However,the genetic basis of its high resistance to stripe rust was unknown.The objectives of this study were to identify QTL for APR to stripe rust in Chakwal 86 using high-density SNP marker maps,and to develop and validate KASP markers for MAS.

2.Materials and methods

2.1.Plant materials

The parental lines for this study were the susceptible winter landrace Mingxian 169(MX169)and resistant line Chakwal 86(CW86).The mapping population consisted of 128 F6:7recombinant inbred lines(RILs)developed from the cross MX169×CW86.Chakwal 86 was highly susceptible to most Pst races in seedling stage conducted under controlled conditions,whereas it was highly resistant at the adult-plant stage in the field[30].MX169,and Xiaoyan 22(XY22)were used as susceptible controls throughout the study.A diversity panel including 150 wheat cultivars and breeding lines from the Yellow and Huai River Valley Wheat Zone was used for SNP-based haplotype analysis using the Wheat35K SNP array.The ultimate goal was to validate SNP-based KASP markers for marker-assisted selection.

2.2.Field trials and phenotyping

Field trials were sown between October and early of November during two growing seasons(2015-2017).The 128 F6:7RILs and their parents were evaluated for APR to stripe rust in fields at Yangling in Shaanxi(34°17′N,108°04′E,altitude 519 m),Tianshui in Gansu (34°27′N,105°56′E,altitude 1697 m)and Jiangyou in Sichuan(31°53′N,104°47′E,altitude 571 m).The trials were conducted in randomized complete blocks with two(Yangling 2015-2016,Yangling 2016-2017,Jiangyou 2015-2016,Tianshui 2015-2016)or three(Jiangyou 2016-2017 and Tianshui 2016-2017)replicates at each location.Each plot consisted of a single 120 cm row with 30 cm between rows.Approximately 30 seeds were sown in each row.Every twentieth row was planted with the highly susceptible control Xiaoyan 22 followed by the two parents.To provide inoculum for infection,MX169 was planted around the experimental area.The RILs at Yangling experimental site were artificially inoculated by spraying a suspended mixture of two of the most prevalent races in China(CYR32 and CYR34)and liquid paraffin(1:300)onto the susceptible controls in late March as flag leaves emerged.The fields in Tianshui and Jiangyou were tested under natural infection because both locations are in hotspot regions where stripe rust infection occurs on a regular basis.Disease severity(leaf areas infected,DS)of the RIL population including the parents was scored when MX169 reached maximum level of 90%-100%(around May 15-20 at Yangling,10-15 April at Jiangyou and June 10-15 at Tianshui).Each line was scored at least twice,and the last data set(the maximum disease severity,MDS)was used for phenotypic and QTL analyses.

2.3.Statistical analysis of phenotypic data

The MDS data from each environment were subjected to analysis of variance(ANOVA)and subsequent QTL mapping.ANOVA and computation of correlation coefficients were performed using the AOV function in QTL IciMapping(Version 4.1)software with the default parameters.The information in the ANOVA table was used to calculate broad sense heritabilityof stripe rust resistance:whereis(MSf-MSfe)/re,is(MSfe-MSe)/r andis MSe;=genetic variance,=genotype×environment interaction variance,=error variance,MSf=mean square of genotypes,MSfe=mean square of genotype×environment interaction,MSe=mean square of error,r=number of replications,and e=number of environments.

Average phenotypic values for RILs in each environment were used for analysis.The genetic effects from six environments were evaluated using the R package lme4 of BLUP(best linear unbiased prediction),where lines,environment,line×environment interaction and replicates nested in environments were all treated as random[33].

2.4.Genotyping of RIL population

Genomic DNA was extracted from fresh leaves of F6-derived each line using the SDS method[34].Each RIL(N=128 genotypes)along with the parents were genotyped by the 35K iSelect SNP array at Capital Bio Corporation(Beijing,http://www.capitalbio.com/).Allele calling was carried out using Affymetrix Genotyping Console(GTC)software,following the Axiom Best Practices Genotyping Workflow(http://media.affymetrix.com/support/downloads/manuals/axiom_genotyping_solution_analysis_guide.pdf).SNP filtering criteria were as follows:monomorphic and poor quality SNP loci with＞10%missing values,ambiguous SNP calling,or minor allele frequencies＜0.05 were excluded from further analysis.The closest SNP markers to the QTL peak were converted to KASP markers and the primers sequences list in Table S1.The procedure of conversion and KASP assays followed Ramirez-Gonzalez et al.[35]and Wu et al.[36].

2.5.Linkage map construction and QTL analysis

Genotypic markers were tested for segregation distortion(deviation expected 1:1 ratios)by Chi-squared tests.Markers with P＜0.001 were removed before constructing the genetic maps.Redundant markers were binned based on their pattern of segregation in the RIL population using the BIN function in IciMapping 4.1[37,38].Markers sharing segregation with at least one other were retained,and one marker was chosen to represent each bin(the marker with the least missing data,or a random marker when percentages of missing data were equal).Groups were ordered using the Kosambi mapping function in the “MAP”tool in IciMapping 4.1 and a LOD score≥3 after preliminary analysis using LOD scores ranging from 5 to 12.Finally MapChart 2.2 was used to draw the genetic map.For redundant loci showing co-segregation among the RILs,only one is shown in the genetic map.The identity positions of linkage groups were determined based on 35K integrated maps[27].

The average values of RILs in each environment and BLUP values were used for QTL detection.QTL analysis was performed by composite interval mapping(CIM)in Windows QTL Cartographer and inclusive composite interval mapping(ICIM)method in IciMapping 4.1 software.When the LOD score was greater than the calculated threshold value(LOD=2.5-4.0),the corresponding QTL was declared significant.Normally,there are minor differences in the peaks of LOD contours for a single QTL across different environments.The overlapping regions detected both programs were considered to be confidence intervals.QTL effects were estimated as the proportion of phenotypic variance explained(PVE)by the QTL.

2.6.Development KASP markers for marker-based selection

According to the deduced region of a QTL in the integrated maps,the 35K SNP genotype data of 150 diverse wheat accessions and a subset of RILs were used to track haplotypes for specific genomic segments linked to the major QTL QYrcw.nwafu-3BS.The Dendextend package in R(v3.3.2)was used for hierarchical clustering based on genotypic and phenotypic kinship.Data were visualized using the Heatmap function in the ComplexHeatmap package in R[39].Specific SNPs linked to the target QTL were then developed as KASP markers use for marker-assisted selection.

3.Results

3.1.Stripe rust scores in the field

In all environments,each RIL line and two parents showed significant genetic variation in APR.MX169 and CW86 gave averaged DS scores of 100%and 10%,respectively.DS of the RILs ranged from 0 to 100%in each environment indicating that the data were continuously distributed and that the responses were polygenically inherited(Fig.1,Table 1,Table S2).Pearson's correlation coefficients among the six environments ranged from 0.56 to 0.88(P＜0.01)(Table 2).P values in the ANOVA of RIL scores showed significant phenotypic variation in DS among lines,environments and line×environment interactions;there was no significant variation among replications within experiments.Thus lines were the main source of phenotypic variation.Heritabilities were high for all data sets(0.86-0.91)(Table 1).The results showed that the APR was consistent across environments,and that QTL for APR had a very large effect in reducing disease severity.

Fig.1-Distribution of mean disease severities for the 128 RILs from Mingxian 169×Chakwal 86 evaluated at Yangling,Tianshui and Jiangyou in 2015-2016(A)and 2016-2017(B).Values for the parents are indicated by arrows.

3.2.Genetic linkage maps

A total of 128 RILs were genotyped with the 35K SNP array.Among the 35,143 SNP markers,4078 were polymorphic between MX169 and CW86;368 were removed as they had＞10%missing data or showed segregation distortion.The remaining 3710 SNPs fell into 1480 bins with 2230 SNPs being redundant.Then 1480 SNPs markers or bins were used for linkage groups construction and the final linkage mapincluded 34 linkage groups(LGs)representing all 21 chromosomes based on the 35K integrated maps[27].The entire linkage map covered 4105.91 cM,with an average marker/bin interval of 2.77 cM(Table 3,Table S3).4BS/6AS and 4BL/6AL were translocation lines contributed by MX169 and that was confirmed in our previous study(Table 3)[39].Only linkage groups with significant stripe rust resistance QTL are shown.

Table 1-Summary of adult responses to stripe rust in 128 Mingxian 169×Chakwal 86 recombinant inbred lines(RIL)during the 2015-2016,2016-2017 cropping seasons at Yangling,Tianshui,and Jiangyou.

3.3.QTL analysis of stripe rust resistance

The QTL that were detected using DS data each environment and the mean across all six environments were considered to be stable.Stable QTL were identified on chromosomes 1BL,3BS,and 6BS and designated as QYrcw.nwafu-1BL,QYrcw.nwafu-3BS,and QYrcw.nwafu-6BS,respectively.All detected QTL were derived from the resistant parent CW86 using both CIM and ICIM.QYrcw.nwafu-1BL explained average values of 12.2%and 13.0%of the phenotypic variation for CIM and ICIM,respectively(Table 4,Fig.2-A,B).This QTL was roughly confined to a 22.04 cM region and the overlapping confidence interval was a 2 cM region flanked by SNP markers AX-95026093 and AX-94701609(Fig.2-C).The largest effect QTL QYrcw.nwafu-3BS,located in a 2.4 cM overlapping confidence interval flanked by SNP markers AX-94437233 and AX-95240191,explained an average 43.1%(ICIM)and 44.0%(CIM)of the phenotypic variation across environments(Table 4,Fig.2-D-F).The MDS ranged from 0 to 60%,the mean ranged from 18.8%to 37.5%of RIL carrying 3B QTL,while the MDS ranged from 10%to 100%,the mean ranged from 55.6%to 73.9%of RIL lacking 3B QTL(Fig.3-A,Table S2).The minor QTL QYrcw.nwafu-6BS,explaining an average 6.3%of the phenotypic variation,was mapped between AX-94413014 and AX-94557244 spanning a 12.81 cM interval.An unstable QTL named QYrcw.nwafu-7BS,accounting for＜5%of the phenotypic variation was disregarded(Table 4).

Table 2-Correlation coefficients(r)of mean disease severities(DS)for the Mingxian 169×Chakwal 86 RIL across six environments.

There were significant additive effects for stripe rust resistance in some RIL lines.When the three stable QTL or the two major ones were combined,the average disease severities were 20.3%and 30.2%,respectively,lower than those with only one or other QTL,or none,and approaching the level of the resistant parent(Fig.3-B,Table S2).This was supported by significant additive effects(P＜0.01)among the QTL obtained using the BIP function in QTL IciMapping 4.1.

3.4.Evaluation of linked markers in wheat genotypes

Fourteen SNP from the 35K SNP array were located in the region AX-94569370 and AX-94775611(12.82-16.12 cM)for QYrcw.nwafu-3BS(Table S3).Three different haplotype groups were identified in the 158 wheat accessions(Fig.4).Accessions within the haplotype of Chakwal 86 and its derived-RILs displayed adult plant resistance in the field tests suggesting that they possibly contained the same haplotype.A quarter of the wheataccessions contained the allele from MX 169 and they showed intermediate or high susceptibility.The remaining haplotypes were notassociated with phenotypic values in other groups indicating that these accessions may not carry the QYrcw.nwafu-3BS allele.We found that the SNP alleles combinations could properly differentiate the target group from other groups in the panel such as AX-94679411(A/G)and AX-94948985(G/C)(Fig.4,Table S4).

4.Discussion

The occurrence and spread of Yr26-virulent races beginning with CYR34 that appeared around 2008 has become a major concern in China.These races are also virulent to Yr10 andvarious combinations of other genes present in local varieties[8,9,40].CW86 was chosen for this study because it has retained a high resistance level over many years[30].Using phenotypic data and genetic maps the APR to stripe rust in CW86 was largely attributed to two major and one minor QTL with additive effects that were consistently identified by CIM and ICIM analysis across six environments.

Table 3-Single-nucleotide polymorphism(SNP)markers statistics of the distribution and density on 21 wheat chromosomes derived from cross between Mingxian 169 and Chakwal 86.

Table 4-Summary of stripe rust resistance QTL detected in the MX169×CW86 RIL population across six environments using ICIM and CIM.

QYrcw.nwafu-1BL,located on chromosome 1BL and flanked by markers AX-95026093 and AX-94701609 appeared to be Yr29,which has been reported in several mapping populations[19,41].This pleiotropic gene Yr29/Lr46/Pm39 is known to be present in many wheat lines,including the CIMMYT variety Pavon 76[16,42-45].Based on the consensus 35K and 90K map[27,46]the above markers flanking QYrcw.nwafu-1BL are adjoined SSR markers linked to Yr29.Further analysis revealed that CW86 and Pavon 76 shared the same cleaved amplified polymorphic sequence(CAPS)marker csLV46G22 that is tightly linked to Yr29.This marker is also associated with the leaf tip necrosis(LTN)gene,Ltn2[47,48]and leaf tip necrosis was observed in Chakwal 86 in the field in the present study.In previous studies,Yr29 explained 7%-65%of phenotypic variation for stripe rust response depending on environmental conditions and genetic backgrounds[15,41].In the present study,QYrcw.nwafu-1BL explained 11.3%-19.2%of the variation stripe rust severity.However,the effect of Yr29 alone is not sufficient for crop protection and it must combine with other resistance genes to reach the level of CW86.

Chromosome 3BS is a well-known resistance gene-rich region that includes numerous resistance genes or QTL to multiple diseases,including the pleiotropic gene Sr2/Yr30 and the Fusarium head blight resistance gene/QTL Fhb1[49].The major effect QTL QYrcw.nwafu-3BS detected near the terminal of chromosome 3BS,was flanked by markers AX-94437233 and AX-95240191.Based on the consensus genetic map[46],these two markers are near to SSR marker Xgwm533,which has shown close linkage to Sr2/Yr30.In previous studies,many QTL for APR to stripe rust were also mapped in the region 0-16.5 cM and most of them were likely Sr2/Yr30[50-54].Several other stripe rust resistance gene/QTL reported near Xgwm533 are clearly not Yr30,such as Yr57 and Yr58.Yr57 confers all-stage resistance carried to Australian winter accession AUS91463[55]and Yr58(QYr.sun-3BS)in Sydney University accession W195 confers APR[56].In the present study,the Sr2 marker csSr2 was not present in CW86 and MX169(Fig.S1).In addition,CW86 did not have dark glumes characteristic of pseudo-black chaff at the late grain-fill growth stage.Yr30[57],Sr2[58],Lr27[59]and pseudo-black chaff[60]are all closely linked to Xgwm533.Although many studies have indicated that Sr2/Yr30 is present in the CIMMYT wheat lines,the QTL on 3BS in CW86 is different from Sr2/Yr30.When we performed molecular detection of Yr58 using its linked markers sun533 and sun474[56],CW86 did not have the resistance alleles related to Yr58.All of above results seems that QYrcw.nwafu-3BS is a distinct locus but further studies are required to dissect the chromosomal region and confirm their genetic relationships.QYrcw.nwafu-3BS had an effect on APR resistance as seen by DS values in CW86,which were significantly lower than those of the susceptible parent and may have high potential value for practical breeding.

Fig.2-Graphical display of positions of QTL for stripe rust resistance across all environments.Overlapping confidence intervals of QTL are based on IciMapping 4.1(A,D)and Windows QTL Cartographer 2.5(B,E).Genetic linkage maps of QYrcw.nwafu-1BL and QYrcw.nwafu-3BS(C,F).Overlapping confidence intervals are shown in red,and markers surrounding the QTL are in bold underlined font.

Fig.3-Effects of single QTL and their combinations on average stripe rust severities of Mingxian 169×Chakwal 86 RILs from(A)Yangling,Tianshui,Jiangyou,and combined environments(B).Box plots(boxes are quartiles,continuous lines are medians,crosses are means,whiskers include others,that are not outliers.One outlier occurred for 1B+3B)for disease severity associated with the two identified QTL(1B and 3B)and their combination.

There are numerous genes/QTL for stripe rust resistance on chromosome 6BS[35].Based on an integrated genetic map[46]most of them are concentrated in the interval 29.2-49.0 cM,and include QYr.tam-6BS in TAM 111[61],QYr.caas-6BS.2 in Naxos[62],QYr.caas-6BS in Bainong 64[63],QYrste.wgp-6BS.2 in Stephens[64],QYr.sun-6BS in Janz[65],QYr-6B in Oligoculm[66],Yr36 in RSL65[67],QYr.wsu-6B.1[21],QYr.ucw-6B(Yr78)in PI 519805[68],QYr.wgp-6B.1 in Stephens[64],QYr.caas-6BS in Pingyuan 50[69],and QYr.nwafu-6BS in Qinnong 142[28].QYrcw.nwafu-6BS was also located in this region and explained＜10%the phenotypic variation,similar to that reported for minor effect QTL QYr.tam-6B and QYr.caas-6BS.Further studies are required to dissect the chromosomal region and confirm the genetic relationships among the stripe rust resistance genes or QTL on 6BS.

The minor QTL QYrcw.nwafu-7BS was environmentally dependent and explained＜5%of the phenotypic variation.These QTL probably have limited practical value,at least individually.They need to be further examined for any potential role in gene pyramiding.In this study,the QTL on 6BS enhanced the level of resistance conferred by QYrcs.nwafu-1BL and QYrcs.nwafu-3BS;RILs combining two or more QTL(1BL+6BS or 3BS+6BS)showed higher resistance than those with just one QTL(1BL or 3BS)(Table S2).

MAS provides an orientated method to detecting and tracking target genes in breeding programs.Benefited from standardized high-throughput genotyping platforms development and lower and lower cost,the application of MAS will increase.Recent developments in genome sequencing enables generation of many markers based on SNP or InDels that will enable high resolution of genetic diversity.Wheat SNP arrays provide an extremely abundant avenue to the study of natural variation in germplasm resources and permits to obtain the markers tightly linked to object genes/QTL[70].This will make it possible to stack favorable haplotypes associated with QYrcw.nwafu-3BS in wheat breeding following development of breeder-friendly KASP markers.These codominant markers allow selection of homozygous resistant lines in the early generations of breeding programs.A potential problem with QYrcs.nwafu-3BS is firstly that pseudo-black chaff can sometimes be excessively expressed making such genotypes unattractive,if not yield limiting,and secondly that the widely used Fhb1 will be closely linked in repulsion with it.A concerted effort should be made to obtain recombinants with these genes in coupling as reported in North American work[71].

In this study,we identified four QTL for stripe rust resistance in a RIL population derived from MX169×CW86.Three of them,all in CW86,were detected in six environments over two years and were validated using comparative methods(CIM and ICIM).Analysis of QTL interaction indicated additive effects of these QTL on rust resistance.Significant additive effects of Yr29/Lr46 and QYrcw.nwafu-3BS with other genes were reported previously.In addition,we developed KASP markers for QYrcw.nwafu-3BS with the largest effect by haplotype analysis in a subset of the RILs and 150 unrelated wheat lines.These KASP markers can be used for highthroughput marker-assisted selection of QYrcw.nwafu-3BS in wheat breeding programs to enhance APR to stripe rust.

Fig.4-Haplotypes of 158 wheat accessions in the stable QYrcw.nwafu-3BS interval.The lines were grouped into haplotypes by two-dimensional hierarchical clustering analysis.The group showed as a color tree along the top X-axis and a marking order(cm)along the right Y-axis.The genotyping data according to the 35K SNP array.The same SNP as the Chakwal 86 allele is blue,while the same SNP as the Mingxian 169 allele is red.The heterozygous or failure calling allele is white.The name and genetic location of each 35K SNP marker was based on the genetic linkage map.The average disease severity of each line is displayed as the box plot along the bottom X-axis(further details are provided in Table S4).

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2018.12.002.

Acknowledgments

The authors are grateful to Prof.R.A.McIntosh,Plant Breeding Institute,University of Sydney,forreview ofthis manuscript.This study was financially supported by the National Science Foundation for Young Scientists of China(31701421),the National Key Research and Development Program of China(2016YFE0108600),the China Agriculture Research System(CARS-3-1-11),the Genetically Modified Organisms Breeding Major Project(2016ZX08002001),and the Key Projectof Science and Technology of Tibetan Autonomous Region,China(XZ201702NB15).