DlNG Xiao-yu,XU Jin-song,HUANG He,QlAO Xing,SHEN Ming-zhen,CHENG Yong,ZHANG Xue-kun

Key Laboratory of Oil Crop Biology and Genetic Improvement,Ministry of Agriculture and Rural Affairs/Oil Crops Research Institute,Chinese Academy of Agricultural Sciences,Wuhan 430070,P.R.China

Abstract Soil waterlogging is a major environmental stress that suppresses the growth and productivity of rapeseed (Brassica napus L.).Natural genetic variations in waterlogging tolerance (WT) were observed but no QTL mapping has been done for WT related traits in rapeseed.In this study,QTL associated with three WT related traits including relative root length (RRL),relative hypocotyl length (RHL) and relative fresh weight (RFW) were dissected using a set of reciprocal introgression lines(ILs) derived from the cross GH01×ZS9,which showed significant difference in WT.Genotyping-by-sequencing (GBS)of the populations were performed,totally 1 468 and 1 450 binned SNPs were identified for GIL (GH01 as the recurrent parent) and ZIL (ZS9 as the recurrent parent) population,respectively.A total of 66 distinct QTLs for WT at the seedling establishment stage including 31 for RRL,17 for RHL and 18 for RFW were detected.Among the 66 QTLs,20 (29.4%)QTLs were detected in both genetic backgrounds and then they were integrated into six QTL clusters,which can be targeted in rapeseed breeding for improvement of WT through marker-assisted selection (MAS).Based on the physical positions of SNPs and the functional annotation of the Arabidopsis thaliana genome,56 genes within the six QTL cluster regions were selected as preliminary candidate genes,then the resequencing and transcriptome information about parents were applied to narrow the extent of candidate genes.Twelve genes were determined as candidates for the six QTL clusters,some of them involved in RNA/protein degradation,most of them involved in oxidation-reduction process.These findings provided genetic resources,candidate genes to address the urgent demand of improving WT in rapeseed breeding.

Keywords:Brassica napus,candidate genes,marker-assisted selection,quantitative trait loci mapping,waterlogging tolerance

1.lntroduction

Waterlogging is the condition where the roots are fully submerged,but above-ground organs are in air (Voesenek and Bailey-Serres 2015).Soil become easily waterlogged after excessive rainfall or irrigation,dryland crops commonly face waterlogging stress in the field.

Oilseed rape (Brassica napusL.) is one of the most important oil crops worldwide and provides about 50% of the vegetable oil supply in China (Yinet al.2009).About 80% of rapeseed is planted along the Yangtze River in China.In this area,rapeseed is planted in paddy fields as a rotation crop just after rice (Zouet al.2014).Direct seeding of rapeseed in soils with poor drainage increased rapidly in recent 10 years,which resulted in severe waterlogging stress.

Rapeseed is particularly sensitive to waterlogging,resulting in significant yield reduction,ranging from 6.5 to 55.5% at the seedling establishment stage,from 1.1 to 34.9% at the seedling stage and from 8.7 to 93.2% at the maturity stage (Zouet al.2014).The physiological responses of rapeseed plants to waterlogging have been extensively investigated (Zhouet al.1997;Kuet al.2009),the effects of waterlogging stress include reduction in stomatal conductance,photosynthetic rate and so on (Zhouet al.1997;Kuet al.2009).The molecular responses of rapeseed plants to waterlogging were investigated at the global gene transcription level in roots and young leaves(Leeet al.2013;Zouet al.2013,2015),waterlogging stress repressed many genes involved in photosynthetic reaction and some of the genes that function in oxidation reduction,protein degradation,abiotic stress tolerance,and others were upregulated (Leeet al.2013;Zouet al.2013,2015).However,there have been few studies on the genetic mechanisms of waterlogging tolerance (WT) in rapeseed.

Previously,natural genetic variations in WT were observed and quantitative trait loci (QTL) mapping has been conducted in dryland crops,including wheat (Boruet al.2001),maize (Mano and Omori2015;Zaidiet al.2015) and soybean (Rhineet al.2010;Nguyenet al.2012).In rapeseed,the waterlogging-tolerant lines yielded 2-fold more than waterlogging sensitive lines under severe waterlogging stress (Zouet al.2014),which indicates that these natural variations can be used to breed new cultivars with better adaptation to waterlogging stress to protect yield from waterlogging stress.However,despite the discovery of heritable natural variations in WT,no QTL mapping has been conducted for WT in rapeseed.

There has been no study on the relationship between WT and other traits,such as,seedling biomass,root length and so on in rapeseed.Whether the genetic improvement of WT and seedling biomass can be conducted simultaneously in rapeseed breeding was still unknown.

This research aimed to reveal the genetic mechanisms of WT and dissect the relationship between WT and seedling biomass in rapeseed.The objectives were to:1) map QTL associated with WT;2) map QTL associated with seeding biomass;3) explore the stable QTL with less relied on genetic background and identify candidate genes for these stable QTLs;and 4) explore the possibility of simultaneous improvement of WT and seeding biomass in rapeseed.

2.Materials and methods

2.1.Plant materials with different waterlogging tolerance

The reciprocal intervarietal introgression lines (IL) were developed from GH01 (waterlogging-sensitive inbred line)×ZS9 (waterlogging-tolerance cultivar) (Appendix A).According to the different recurrent parents,they were named ZIL (ZS9 is the recurrent parent) and GIL (GH01 is the recurrent parent),respectively,the ZIL population consists of 58 BC2F3families and GIL consists of 50 BC2F3families.

2.2.Experimental treatments and phenotype evaluation

Fresh seeds of parents and each family were selected and were placed on moist filter paper.Seeds were grown at 25°C (light:dark,16 h:8 h) until their radicals were about 2–5 mm.Uniform seeds were selected and divided into two groups.One group was cultured with a normal water supply as the control and the other was followed by waterlogging treatment.Thirty normal germinated seeds were treated for 12 h in a sealed tube completely filled with distilled water to avoid oxygen exchange with the air.Following removal from anaerobic chambers,seeds were sown in dishes filled with moist fine vermiculite and were cultured for 5 d.All the surviving seedlings of each family per replicate were used for measuring of root length (RL),hypocotyl length (HL) and fresh weight (FW) under control (normal growth condition)and waterlogged conditions.The other three parameters for WT were calculated as follow:

Relative root length (RRL)=Mean root length of treated seedling/Mean root length of control seedlings

Relative hypocotyl length (RHL)=Mean hypocotyl length of treated seedling/Mean hypocotyl length of control seedlings

Relative fresh weight (RFW)=Mean fresh weight of treated seedling/Mean fresh weight of control seedlings

Totally three parameters for seedling biomass (RL,HL and FW) under normal growth condition and three parameters for WT (RRL,RHL and RFW) were investigated for QTL mapping.

2.3.DNA extraction,genotyping and SNP calling

DNA was extracted from bulked young leaf tissue of several individuals per family using Charge Switch gDNA plant kit(Invitrogen;Carlsbad,CA,USA) and diluted to 100 ng μL–1.Parental DNA was extracted from two separate bulked samples each.Genomic DNA was sent to the Huazhong Agricultural University (Wuhan,China) for GBS using protocols modified from Elshireet al.(2011).Briefly,96-plex libraries were constructed with ApeKI digestion and sequenced on an Illumina HiSeq 2500 producing 100-bp pair-end reads.Sequences were aligned to theB.napusgenome (Chalhoubet al.2014) and SNPs were called using the GBS pipeline of TASSEL v3.0.173 (Glaubitzet al.2014).

SNPs were filtered with TASSEL v5.2.16 (Bradburyet al.2007) to exclude those with indels,more than 90%missing data,and more than two alleles.Missing data were imputed with the FSFHap Imputation Plugin of TASSEL(Swartset al.2014) using default settings except for use of the cluster algorithm.Nucleotide data were recoded into parental genotypes with the ABH plugin of TASSEL,which also filters out SNPs where the parental genotype is missing,ambiguous or heterozygous.Additional data integrity checks were performed in the QTL package (Bromanet al.2003)of R v3.3.1 (R Core Team 2016).SNPs with a minor allele frequency less than 0.05 or more than 5% missing data were removed.Redundancy was removed by binning markers with identical segregation patterns.The sequence data were deposited into the NCBI with the Sequence Read Archive(accession number SRP151331).

2.4.Data analysis and QTL mapping

SAS was used to analyze trait variances.QTL analysis was conducted by the method of single marker analysis in Windows QTL Cartographer 2.5 (Wanget al.2007).The location of a QTL was described according to the physical position of the marker with the lowestP-value.All QTLs,were named as previously described (Shiet al.2009).

All the original detected QTLs for the three WT related traits were subjected to QTL integration.A two-round strategy of QTL meta-analysis was adopted.Firstly,the QTLs detected repeatedly in the same population were integrated mainly comparing the physical position of SNPs significantly associated with different phenotypes,in the second round,QTLs detected in different populations were integrated.When comparing the physical position of SNPs associated with QTLs detected in GIL and ZIL,respectively,the following method was applied:if marker A was detected with the lowestP-value in one population,which means a QTL (QTL1) nearby marker A was associated with trait variation,and marker A also was significantly (P<0.01)associated with trait variation in another population but the nearby one-marker B with the lowestP-value,then QTL1 was regarded as detected repeatedly in both populations,which was located between markers A and B.

3.Results

3.1.Genotyping by sequencing

Genotyping by sequencing (GBS) of the parents-ZS9 and GH01 resulted in 93 193 506 and 107 135 700 good-quality bar-coded reads,respectively.After align to the reference genome (Chalhoubet al.2014),2 689 319 SNPs were detected between ZS9 and GH01.GBS of the introgression lines (ILs) resulted in average 3 720 321 good-quality bar-coded reads.The following three prerequisites were adopted during population genotyping:1) only the SNPs both detected among the ILs and between the parents were retained for further analysis;2) the SNPs with a minor allele frequency less than 0.05 or more than 5% missing data in each population were removed;3) in order to compare the position of QTLs detected in ZIL and GIL population more conveniently,the same series of SNPs were selected for genotyping the two populations.Finally,5 485 SNPs were used to genotype the two populations.The SNPs were well distributed across the chromosomes except C9 (Appendix B).Major of the SNPs had identical segregation patterns and were reduced to 1 468 binned SNPs for GIL and 1 450 binned SNPs for ZIL.

The level of donor genome coverage was investigated in the GIL and ZIL populations,which was 97.5% for GIL and 98.4% for ZIL population,respectively.

3.2.Phenotyping

Two-way ANOVA of three traits (RL,HL and FW) across waterlogging treatments indicated that genotype of each IL (G),waterlogging treatment and genotype-waterlogging treatment interactions (G×E) had significant effects on the three traits.However,the G×E effects were ignorable because they were minor when compared to the effects of genotypes and environment.

The performance of three WT related traits (RRL,RHL and RFW) differed between ZS9 and GH01 (Table 1 and Fig.1).The average RRL,RHL and RFW of ZS9 were 0.593±0.111,0.673±0.155 and 0.794±0.063 (n=53),respectively,which were significantly greater than that of GH01 (0.222±0.061,0.484±0.112 and 0.521±0.081,n=42)(P<0.001).The both populations had the same trends in the following aspects.Firstly,for the same family,the value of RRL was significantly smaller than that of RHL and RFW,which was consistent with the fact that the roots were more sensitive to waterlogging than other parts of the plant in rapeseed (Zouet al.2014).Secondly,as shown in Table 1 and Fig.1,the both populations displayed transgressive segregation in both directions for all the three traits,whichimplied favor alleles for WT occurred not only in ZS9,the WT parent,but also in GH01,the waterlogging-sensitive parent.

Table 1 Performance of reciprocal introgression lines (ILs) of GH01 and ZS9 and parents for three waterlogging tolerance (WT)related traits and three seeding biomass related traits1)

Fig.1 Three waterlogging tolerance (WT) related traits and three seeding biomass related traits variation among introgression lines (ILs).Frequency distributions of relative root length (A),relative hypocotyl length (B),relative fresh weight (C),root length(D),hypocotyl length (E),and fresh weight (F).GIL,introgression lines with GH01 as recurrent parent;ZIL,introgression lines with ZS9 as recurrent parent.The phenotype for the same trait in different populations is shown in the same picture.

The average RL and HL of GH01 were (4.472±1.617)and (2.383±0.332) cm,respectively,which were significantly greater than that of ZS09 ((3.914±1.392) and (2.034±0.325) cm,respectively) (P<0.01),but for FW,the difference between GH01 ((46.112±5.612) mg) and ZS9 ((44.373±5.172) mg)was not significant (Table 1).All the three biomass related traits showed continuous distribution in the two populations(Fig.1).

Correlation coefficients among the six traits were calculated in the two populations,respectively (Appendix C).Among the three WT related traits,RRL showed significant positive correlations with RHL and RFW (P<0.01),RHL also showed significant positive correlations with RFW(P<0.01).Among the three biomass related traits,HL showed significant positive correlations with RFW (P<0.01).No significant correlation was observed between the three waterlogging-related traits and the three seedling biomass related traits.

3.3.QTL associated with WT traits

A total of 66 distinct QTLs for three WT traits,including 31 for RRL,17 for RHL and 18 for RFW were detected in the reciprocal ILs (Table 2;Fig.2).

For RRL,15 and 16 QTLs were identified in GIL and ZIL,respectively (Table 2;Fig.2).Among the 15 QTLs detected in GIL,the ZS9 alleles at all loci,except forqRRLA2.a,qRRLA4.a,qRRLA7-1.a,andqRRLC5.a,increased RRL by an average of 0.039 (ranging from 0.032 to 0.044).Among the 16 QTLs detected in ZIL,the ZS9 alleles at all loci,exceptqRRLC1-1.bandqRRLC2.b,increased RRL by an average of 0.043 (ranging from 0.036 to 0.053).

For RHL,eight and nine QTLs were identified in GIL and ZIL,respectively (Table 2;Fig.2).Among the eight QTLs detected in GIL,the ZS9 alleles at all loci,exceptqRHLA4.a,qRHLA9.aandqRHLA10.a,increased RHL by an average of 0.093 (ranging from 0.089 to 0.103).Among the nine QTLs detected in ZIL,the ZS9 alleles at all loci,exceptqRHLA1.b,qRHLA9.bandqRHLC1.b,increased RHL by an average of 0.057 (ranging from 0.047 to 0.068).

For RFW,nine and nine QTLs were identified in GIL and ZIL,respectively (Table 2;Fig.2).Among the nine QTLs detected in GIL,the ZS9 alleles at all loci,exceptqRFWA6-2.aandqRHLC4.a,increased RFW by an average of 0.051 (ranging from 0.042 to 0.057).Among the nine QTLs detected in ZIL,the ZS9 alleles at all loci,exceptqRFWC1.b,increased RFW by an average of 0.040 (ranging from 0.034 to 0.045).

After integration of the above QTLs detected in the same population,the results showed that three QTL clusters namedqWTA6.a,qWTA7.aandqWTC1.awere detected in GIL,account for 25% of the total.And eight QTL clusters namedqWTA3.b,qWTA7.b,qWTC1.b,qWTC3.b,qWTC6.b,qWTC7-1.b,qWTC7-2.b,andqWTC7-3.bwere detected in ZIL,account for 62.5% of the total (Table 2).

Then we compared the position of QTLs detected in GIL and ZIL populations,the results indicated that six QTLs namedqWTA7-1,qWTA7-2,qWTC1,qWTC6,qWTC7-1,andqWTC7-2,were simultaneously detected in both genetic backgrounds (Table 2).A QTL cluster namedqWTA6.awas only detected in GIL,four QTL clusters namedqWTA3.b,qWTC1.b,qWTC3.b,andqWTC7-3.bwere only detected in ZIL.

3.4.Candidate genes mining for WT

The proximity of candidate genes was inferred based on functional annotation of theArabidopsis thalianagenome,the gene expression pattern ofA.thalianahomologous genes and the physical positions of SNPs/candidate genes on reference-sequenced genome ofB.napus(Chalhoubet al.2014).We selected 56 genes from the six QTLsdetected in both genetic backgrounds as the preliminary candidate genes.

Table 2 The list of 66 QTLs for three waterlogging tolerance (WT) related traits detected in the reciprocal introgression lines (ILs)1)

Table 2 (Continued from preceding page)

Fig.2 Chromosomal locations of detected QTLs.Positions of bins used as genetic markers are indicated by horizontal lines spaced proportionally according to their physical position on the reference genome (Chalhoub et al. 2014),the detailed positions (Mb) of the QTLs associated with wild type detected in both populations were showed.The symbols of figures represent the positions of the QTL peak.RL,root length;HL,hypocotyl length;FW,fresh weight;RRL,relative root length;RHL,relative hypocotyl;RFW,relative fresh weight.Upward arrows indicate GH01 as the donor of the increasing allele;downward arrows indicate ZS9 as the donor of the increasing allele.GIL,introgression lines with GH01 as recurrent parent;ZIL,introgression lines with ZS9 as recurrent parent.

We tried to narrow the extent of candidate genes by using the resequencing and transcriptome information about parents.Firstly,the polymorphisms in the preliminary candidate genes were annotated according to the gene structure in theB.napusreference genome (Chalhoubet al.2014),the results showed that there was no polymorphism predicted to induce premature stop,result in frame shift,or change splicing junctions;10.4% of the polymorphisms predicted to induce amino acid exchange;53.2% of the polymorphisms were located at the 5´ flanking sequences (≤2 kb from the first ATG) or the 3´ untranslated regions of genes.The results implied the difference in gene transcription level may underlie a considerable amount of QTLs in this study.Secondly,taken the gene structure in theDarmor-bzhgenome as reference (Chalhoubet al.2014),the transcriptomes of parents were reanalyzed (Zouet al.2013,2015),the results indicated that 12 preliminary candidate genes showed different expression in roots between ZS9 and GH01 under waterlogging stress at the seedling establishment stage,among which,the three genes includingBnaA07g11370D(BnA07CAT3,the decrease of expression in ZS9 was significantly higher than that in GH01 after waterlogging treatment),BnaA07g12260D(BnA07SKS17,the decrease of expression in ZS9 was significantly higher than that in GH01 after waterlogging treatment) andBnaA07g13150D(BnA07OTLD1,the decrease of expression in ZS9 was significantly higher than that in GH01 after waterlogging treatment) were determined as candidates forqWTA7-1.To test the accuracy of parental resequencing results,we amplified the genome sequence ofBnaA07g11370DandBnaA07g12260Dfrom ZS9 and GH01,and we found that most SNPs and small insertion/deletion (InDel),which mainly located on the promoter region,were accurate in resequencing data (Appendix D).The two genes includingBnaA07g29460D(peroxidase superfamily protein,the increase of expression in ZS9 was significantly higher than that in GH01 after waterlogging treatment) andBnaA07g31960D(flavodoxin family protein,the decrease of expression in ZS9 was significantly higher than that in GH01 after waterlogging treatment) were determined as candidates forqWTA7-2,BnaC01g09980D(BnC01RPN1B,the decrease of expression in ZS9 was significantly higher than that in GH01 after waterlogging treatment) was determined as candidate forqWTC1,the two genes includingBnaC06g35970D(flavodoxin family protein,the decrease of expression in ZS9 was significantly higher than that in GH01 after waterlogging treatment) andBnaC06g36330D(BnC06XRN3,the decrease of expression in ZS9 was significantly higher than that in GH01 after waterlogging treatment) were determined as candidates forqWTC6,the two genes includingBnaC07g30500D(GPIanchored adhesin-like protein,the decrease of expression in ZS9 was significantly higher than that in GH01 after waterlogging treatment) andBnaC07g32280D(BnC07FER,the decrease of expression in ZS9 was significantly higher than that in GH01 after waterlogging treatment) were determined as candidates forqWTC7-1,the two genes includingBnaC07g38970D(BnC07AR1,the decrease of expression in ZS9 was significantly higher than that in GH01 after waterlogging treatment) andBnaC07g40720D(BnC07CBL3,the decrease of expression in ZS9 was significantly higher than that in GH01 after waterlogging treatment) were determined as candidates forqWTC7-2.Some of them involved in protein/RNA degradation,such as,BnA07OTLD1,BnC01RPN1B,andBnC06XRN3,most of them involved in oxidation-reduction process,such as,BnA07CAT3,BnA07SKS17,BnaA07g29460D,BnaA07g31960D,BnaC06g35970D,andBnC07AR1.

3.5.QTL associated with seedling biomass traits

For RL,8 and 14 QTLs were identified in GIL and ZIL,respectively (Fig.2).Among the eight QTLs detected in GIL,the ZS9 alleles at four loci increased RL.Among the 14 QTLs detected in ZIL,the ZS9 alleles at six loci increased RL.

For HL,11 and 7 QTLs were identified in GIL and ZIL,respectively (Fig.2).Among the 11 QTLs detected in GIL,the ZS9 alleles at five loci increased HL.Among the seven QTLs detected in ZIL,the ZS9 alleles at three loci increased HL.

For FW,nine and nine QTLs were identified in GIL and ZIL,respectively (Fig.2).Among the nine QTLs detected in GIL,the ZS9 alleles at four loci increased FW.Among the nine QTLs detected in ZIL,the ZS9 alleles at four loci increased FW.

3.6.Comparison between QTLs associated with WT and QTLs associated with seedling biomass

To explore the possibility of simultaneous improvement of WT and seedling biomass,the comparison between location of QTLs associated with WT and QTLs associated with biomass were carried out.Five pairs QTLs (qRRLA2.aandqFWA2.a,qRRLA6-1.aandqFWA6.a,qRRLA7-2.aandqRLA7.a,qRHLA10.aandqHLA10.a,qRRLC2.aandqRLC2-2.a) were mapped in the same or adjacent position in GIL,account for 20% of the total.Three pairs QTLs(qRRLA7-3.bandqHLA7.b,qWTC3.bandqFWC3-2.b,qWTC7-2.bandqRLC7.b) were mapped in the same or adjacent position in ZIL,account for 14.1% of the total.

4.Discussion

4.1.QTLs and candidate genes associated with WT were valuable in rapeseed WT improvement

There was significant correlation between the three waterlogging-related traits (RRL,RHL and RFW) at the seedling establishment stage and the final yield loss in the field under waterlogging stress in rapeseed (Zouet al.2014).Therefore,dissecting the genetic mechanism of WT at the seedling establishment stage made significance to breed new cultivars with better adaptation to waterlogging stress to protect yield from waterlogging stress.According to the rapeseed breeding history,problem of waterlogging did not attract a lot of attention before,and breeders have not focused on WT in breeding program (Chenget al.2010).As far as we know,this was the first study to dissect the genetic mechanism of WT using QTL mapping in rapeseed.Introgression of the QTLs detected in both genetic backgrounds (discussed in the section below) into elite lines has been initiated to develop new waterlogging tolerant cultivars with help from marker-assisted selection(MAS) in the rapeseed breeding program.

Most MAS experiments concern introgression of QTLs for improvement of single trait in the genetic background in which they were originally detected (Ahmadiet al.2001;Berlooet al.2001).The use of several recipient parents is essential to assess the effect of a QTL detected in one population when transferred to unrelated genetic backgrounds.When several recipient parents were used,the consistency of the QTL effects in the different genetic backgrounds became less obvious (Seboltet al.2000;Yousef and Juvik 2002).So consistency of QTLs’ effects in different genetic backgrounds is a major concern for MAS breeding of complex traits.Among the 68 QTLs associated with WT in this study,only 20 (29.4%) were detected in both populations,clearly indicating that most QTLs detected in one background were not identified in the other.This was in agreement with data showing that only 21% of the QTLs for branch numbers and spikelet number per panicle in reciprocal ILs in rice (Meiet al.2006).Furthermore,in agreement with that in rice (Meiet al.2006),the results of our study also showed that the direction of additive effects was consistent for QTLs detected in both populations,indicating that the genetic background can influence QTL s’ effect without affecting the direction of gene effect.In the present study,six QTLs,includingqWTA7-1,qWTA7-2,qWTC1,qWTC6,qWTC7-1,andqWTC7-2were consistently detected in the two genetic backgrounds,so they could be applied to improve WT in MAS breeding.

It was difficult to determine the candidate genes for QTLs in linkage mapping in rapeseed 10 years ago,but it has been possible to narrow QTLs to dozens of genes in recent years (Shenet al.2018).Many researches on the molecular mechanism of waterlogging stress inA.thaliana(Voesenek and Bailey-Serres 2013),rice (Lasanthi-Kudahettigeet al.2007),cotton (Christiansonet al.2010) and rapeseed (Leeet al.2013;Zouet al.2013,2015) have been done,which provided important references to us.On the other hand,more convenient access to omics data has paved the way for us to identify candidate genes in linkage mapping.In our study,12 genes were determined as candidates for 6 QTLs,and validation of their function has been in progress.Another notable thing was that most of the polymorphisms between parents were located at the 5´ flanking sequences or the 3´ untranslated regions of candidate genes,which implied the difference in gene transcription level may underlie a considerable amount of QTLs associated with WT.

4.2.Simultaneous improvement of WT and seedling biomass can be achieved in rapeseed breeding

In rapeseed,some QTLs associated with seedling biomass traits appeared to co-localize with that associated with seed yield was reported by Basunandaet al.(2010),and the alleles from same parents increased performance of both traits,indicating that greater seedling biomass may be a potential indicator for higher seed yield.Therefore,it made sense to explore QTLs associated with seedling biomass for high yield breeding program under normal condition.Further,simultaneous improvement of WT and seedling biomass would be the ideal situation,which would be valuable to breed new cultivars with high yield under normal condition and reduced yield loss under waterlogging stress.An in-depth study of this issue hasn’t yet been reported in rapeseed.In the present study,five(20%) pairs QTLs (qRRLA2.aandqFWA2.a,qRRLA6-1.aandqFWA6.a,qRRLA7-2.aandqRLA7.a,qRHLA10.aandqHLA10.a,qRRLC2.aandqRLC2-2.a) were mapped in the same or adjacent position in GIL.Three (14.1%) pairs QTLs (qRRLA7-3.bandqHLA7.b,qWTC3.bandqFWC3-2.b,qWTC7-2.bandqRLC7.b) were mapped in the same or adjacent position in ZIL.Most of the QTLs associated with WT didn’t appear to co-localize with that associated with seedling biomass,indicating that simultaneous improvement of WT and seedling biomass can be achieved in breeding program.

5.Conclusion

As far as we know,it was the first report about dissection of QTLs associated with WT related traits using a set of reciprocal ILs in rapeseed.Our findings provided genetic resources,candidate genes to address the urgent demand of improving WT in rapeseed breeding.

Acknowledgements

This research was funded by the National Key Research and Development Program of China (2017YFD0101700) and the National Natural Science Foundation of China (31301361 and 31171589).

Appendicesassociated with this paper can be available on http://www.ChinaAgriSci.com/V2/En/appendix.htm