Identification of QTLs for resistance to 10 pathotypes of Plasmodiophora brassicae in Brassica oleracea cultivar ECD11 through genotyping-by-sequencing

Key message Two major quantitative trait loci (QTLs) and five minor QTLs for 10 pathotypes were identified on chromosomes C01, C03, C04 and C08 through genotyping-by-sequencing from Brassica oleracea. Abstract Clubroot caused by Plasmodiophora brassicae is an important disease in brassica crops. Managing clubroot disease of canola on the Canadian prairie is challenging due to the continuous emergence of new pathotypes. Brassica oleracea is considered a major source of quantitative resistance to clubroot. Genotyping-by-sequencing (GBS) was performed in the parental lines; T010000DH3 (susceptible), ECD11 (resistant) and 124 BC1 plants. A total of 4769 high-quality polymorphic SNP loci were obtained and distributed on 9 chromosomes of B. oleracea. Evaluation of 124 BC1S1 lines for resistance to 10 pathotypes: 3A, 2B, 5C, 3D, 5G, 3H, 8J, 5K, 5L and 3O of P. brassicae, was carried out. Seven QTLs, 5 originating from ECD11 and 2 from T010000DH3, were detected. One major QTL designated as Rcr_C03-1 on C03 contributed 16.0–65.6% of phenotypic variation explained (PVE) for 8 pathotypes: 2B, 5C, 5G, 3H, 8J, 5K, 5L and 3O. Another major QTL designated as Rcr_C08-1 on C08 contributed 8.3 and 23.5% PVE for resistance to 8J and 5K, respectively. Five minor QTLs designated as Rcr_C01-1, Rcr_C03-2, Rcr_C03-3, Rcr_C04-1 and Rcr_C08-2 were detected on chromosomes C01, C03, C04 and C08 that contributed 8.3–23.5% PVE for 5 pathotypes each of 3A, 2B, 3D, 8J and 5K. There were 1, 10 and 4 genes encoding TIR-NBS-LRR/CC-NBS-LRR class disease resistance proteins in the Rcr_C01-1, Rcr_C03-1 and Rcr_C08-1 flanking regions. The syntenic regions of the two major QTLs Rcr_C03-1 and Rcr_C08-1 in the B. rapa genome ‘Chiifu’ were searched. Supplementary Information The online version contains supplementary material available at 10.1007/s00122-023-04483-y.


Introduction
Brassica species are grown as source of oil, vegetable, condiment, and fodder worldwide.The 'Triangle of U' (Morinaga 1934;Nagaharu and Nagaharu 1935) revealed relationships among the main species of Brassica genus, which makes more interesting of these species to the researchers for genomic utilization among the species.B. napus (AACC; n = 19), B. juncea (AABB; n = 18) and B. carinata (BBCC; n = 17) are amphidiploid species resulting from hybridization between pairs of the diploid species: B. rapa (AA; n = 10), B. nigra (BB; n = 8) and B. oleracea (CC; n = 9).Clubroot, caused by Plasmodiophora brassicae Woronin, a soil-borne obligate biotrophic protist, is an important disease in brassica crops worldwide.The pathogen causes club or spindle shape gall formation on roots, which hampers nutrients and water uptake for plants growth.The disease has been reported in more than 60 countries, causing about 10-15% yield loss globally (Dixon 2009).The disease not only caused yield loss but also affected the quality of the crops (Engqvist 1994;Pageau et al. 2006).Some practices such as liming and early spring seeding can reduce clubroot levels but is not economical for large-scale canola production (Voorrips et al. 1995;Gossen et al. 2012, Donald et al. 2014;Hwang et al. 2014).Utilization of clubroot resistant (CR) cultivars is the most effective way to control the disease (Peng et al. 2014;Rahman et al. 2014).However, repeated utilization of the same CR source is not a good strategy as it provides opportunity to minor pathotypes to become predominant pathotypes (Sedaghatkish et al. 2019;Cao et al. 2020).In Canada, breeding companies released CR varieties in 2009-2010 after identifying clubroot disease in canola 249 Page 2 of 15 fields in 2003 (Tewari et al. 2005).Most CR canola cultivars probably were developed by using a CR gene from the oilseed rape 'Mendel' (Fredua-Agyeman et al. 2018), which became susceptible just after 4 years in 2013 due to emergence of new virulent pathotype 5X (Strelkov et al. 2016).This incidence led to closely focus on pathogen populations in Canadian canola fields.As a result, > 30 pathotypes have been classified based on the Canadian Clubroot Differential (CCD) set (Strelkov et al. 2018;Hollman et al. 2021).In addition to the most prevalent original pathotype 3H, current predominant new pathotypes 3A, 3D (Dakouri et al. 2021) and other pathotypes are also needed to consider to find new resistant sources.Along with crop rotation, gene pyramiding from different sources is considered an effective strategy for successful control of clubroot disease caused by different pathotypes of the pathogen.
As resistance sources to clubroot in B. napus (AACC) are very limited, researchers have searched resistance in its diploid progenitor species B. rapa (AA) and B. oleracea (CC).Genes can be transferred from B. rapa (AA) and B. oleracea (CC) to B. napus either interspecific breeding or developing new re-synthesized B. napus.CR in B. rapa was usually controlled by major genes, while CR in B. oleracea was mostly reported as quantitative trait loci (QTL), polygenic control.Although polygenic resistance is considered more durable and can be race non-specific (Diederichsen et al. 2009), resistance sources in B. oleracea are very limited (Dias et al. 1993;Manzanares-Dauleux et al. 2000;Voorrips 1995;Diederichsen et al. 2009).Cabbage (B.oleracea var.capitata) cultivar 'Badger Shipper' (ECD11) showed resistance to 15 pathotypes out of 17 pathotypes in the CCD Set (Strelkov et al. 2018), which can be very useful as race non-specific CR source.
Although no CR gene has been cloned from C-genome, three CR genes CRa, Crr1 and CRb kato were cloned from A-genome, which encode toll-interleukin-1 receptor, nucleotide binding site and leucine-rich repeat (TIR-NBS-LRR) proteins (Ueno et al. 2012;Hatakeyama et al. 2013Hatakeyama et al. , 2017)).Resistant genes encoded NBS-LRR protein families can be subdivided according to their functional domain as TIR domain containing (TNL) and coiled-coil (CC) domain containing (CNL) subfamilies (McHale et al. 2006).NBS-LRR-related disease resistance is effective against obligate and hemi-biotrophic pathogens (Glazebrook et al. 2005).Although TNL protein encoding genes are reported for CR resistance, no CR gene encoding CNL protein has been reported.
In this study, a BC 1 /BC 1 S 1 mapping population was developed using resistant cabbage cultivar ECD11 crossed with clubroot-susceptible doubled haploid (DH) line T010000DH3.The objectives of the current study were: 1) to characterize the genome-wide DNA variants in the BC 1 lines through genotyping-by-sequencing (GBS); 2) to identify QTLs associated with resistance to 10 pathotypes of P. brassicae characterized with the CCD set; and 3) to identify possible candidate genes for each QTL.

Plant materials
Seeds of resistant cultivar ECD11 and susceptible DH line T010000DH3 derived from Chinese kale (B.oleracea) cultivar 'TO1434' were provided by Dr. G. R. Dixon (The University of Warwick, Wellesbourne, Warwick, UK) and Dr. I. Parkin (Saskatoon Research and Development Centre, Saskatoon, SK, Canada), respectively.ECD11 was crossed to T010000DH3 (pollen donor) to produce F 1 progeny.Backcross with T010000DH3 (recurrent parent) was performed to produce the BC 1 population.In total, 124 BC 1 plants were used for GBS and self-pollinated to produce a BC 1 S 1 population consisting of 124 lines for inoculation testing at the Saskatoon Research and Development Centre.

Evaluation of plants for resistance to clubroot
Strains of P. brassicae collected from canola fields in Alberta were provided by Dr. S.E.Strelkov at the University of Alberta, Canada.The inoculum preparation method used in the study was as described by Suwabe et al. (2003).Seedlings of a susceptible cultivar were inoculated and maintained under controlled conditions.Clubbed roots were harvested from infected plants after 5-6 weeks and stored at − 20 °C.Fresh inoculum of each pathotype was prepared after softening about 5 g of frozen club in a small amount of distilled water for 1-2 h.The material was homogenized in a blender for 2 min and strained through 2-3 layers of nylon mesh cloth.The resulting spore suspension was diluted with deionized water to produce a final concentration of 1 × 10 7 resting spores mL −1 .Plants were tested for resistance to 10 pathotypes of P. brassicae (strain F.3-14 for pathotype 3A, F.183-14 for 2B, F.175-14 for 5C, F.1-14 for 3D, CDCS for 5G, P. 41-14 for 3H, F.12-15 for 8J, F.10-15 for 5K, CDCN#2 for 5L and F.381-16 for 3O).The inoculation method used in the current study was as described by Yu et al. (2021).Seeds of the BC 1 S 1 population were sown into Sunshine #3 soilless mix (Sun Gro Horticulture Canada Ltd.; Seba Beach, AB) with Osmocote (Everris NA Inc.; Dublin, OH, USA) in 32-pots inserts held by trays (The HC Companies; Twinsburg, OH, USA).Adequate amount of water was added to each tray to soak the soilless mix overnight.Similar experimental design as described by Yu et al. (2017) was used in this study.Ten to twelve seeds were sown in each pot/line, which were inoculated with 15 ml of inoculum (1 × 10 7 spores/ml) after seven days of sowing.The tray holding pot was covered with a dome for two weeks.The inoculated plants were grown in a growth chamber set at 22/18 °C day/night temperature with a 16 h photoperiod.The pots were always kept moist by adding adequate amount of water whenever necessary.Leaves of the plants were pruned every alternate days after two weeks of seeding.Watering was stopped at 2 days before ratting clubroot.The B. rapa var.pekinensis 'Granaat' ECD05 (susceptible to all 10 pathotypes used in this study) and the parental lines (ECD11 and T010000DH3) were included as controls.
Six weeks after inoculation, each plant was rated for clubroot symptoms using a 0-3 scale (Kuginuki et al. 1999), where 0 = no symptoms, 1 = a few small clubs, 2 = moderate clubbing, and 3 = severe clubbing (Fig. S1).A disease severity index (DSI) was calculated for each line using the method of Horiuchi and Hori (1980): Pearson correlation coefficient was calculated with disease severity indexes (DSIs) of 124 BC 1 S 1 populations.
The F 1 (T010000DH3 × ECD11) also included in the inoculation test for all the ten pathotypes in this study.Reciprocal F 1 (ECD11 × T010000DH3) were tested with only 3 pathotypes 3A, 2B and 3D as only few seeds from the cross were obtained, so it was not possible to test all pathotypes.The experiment was repeated twice.Each of the repetitions provided a similar result in most cases.For those lines with inconsistent results, the higher DSI among the two repetitions of the assessment was considered to be more accurate and was used to characterize the resistance response of the line.

GBS of the parental lines and 124 BC 1 plants
Young leaves of the 124 BC 1 plants and two replications of parental lines were collected and freeze-dried in a Freezone 6 dryer (Labconco Corp, Kansas City, MO) for 48 h and grinded using the Mixture Mills 300 (Retsch Inc., Newtown, PA).DNA was extracted using DNeasy 96 Plant Kit (Qiagen, Toronto, ON, Canada) following DNeasy Plant Handbook from QIAGEN, quantified using a NanoVue Plus spectrophotometer (GE Healthcare, Piscataway, NJ), diluted to 10 ng/μl and kept at −20 °C until subsequent use for sequencing and genotyping.In total, 128 DNA samples were prepared for sequencing.GBS library generation, PstI/MspI for Illumina with size selection were done in Institute of integrative biology and systems (Université Laval, Québec, Canada).Illumina library QC and Illumina HiSeq2500 PE125 sequencing were performed at Genome Quebec (Montreal, Canada).Two replications of parental lines were performed to increase the sequencing depth, to provide a more accurate call of the genotype at each SNP site in the BC 1 population.

Alignment of GBS short reads into B. oleracea reference genome sequence and SNP calling
The reference genome sequence of DH line 'D134' (cabbage; B. oleracea var.capitata) generated by third-generation sequencing technology (Lv et al. 2020) and downloaded from https:// db.cngb.org/ searc h/?q= CNP00 00469 was used for alignment and data analysis.Short reads from each of the 124 BC 1 samples, combined two replicates of parental lines ECD11 and T010000DH3, were aligned to the reference genome sequence.GBS-SNP-CROP pipeline v3.0 (Melo et al. 2016) was used to get SNP panel using Linux platform of AAFC Biocluster, following steps were performed: (1) parse the raw reads, ( 2 (3) demultiplex, (4) align with BWA-MEM and process with samtools, (5) parse mpileup outputs and produce the variant discovery matrix, (6) filter variants and call genotypes, and (7) create input files for downstream analyses.The trimming of the reads was performed using trimmomatic-0.39(Bolger et al. 2014) 2007).KNNimp imputation was done to recover missing data.Data were filtered with minimum count 5, minimum frequency 0.1 (10%), maximum frequency 1.0 (100%).SNP alleles from the susceptible parent (T010000DH3) were scored as 'a', resistant parent (ECD11) were scored as 'b' and heterozygous allele as 'h'.As the parent T010000DH3 is a DH line, all SNP sites should theoretically be homozygous (a).Excel sort function was used to filter out allele 'h' from the DH line and SNP with higher missing sites.

Construction of linkage map and QTL mapping
To remove redundant markers, another two-step filtering was performed with JoinMap and the BIN function with ICI-Mapping.The high-quality polymorphic SNPs were further analyzed using JoinMap 4.1 (Van Ooijen 2011).Maximum likelihood in Kosambi's model with a minimum logarithm of the odds (LOD) values of 10 was used to determine marker orders and positions in the genetic map.Only SNP sites that could be assigned into the 9 chromosomes of the C-genome sequence at LOD scores of 10.0 were kept.The set of filtered SNP sites obtained from JoinMap4.1 was used for binning of redundant markers, construction of linkage map, and mapping of QTLs for resistance to clubroot using the QTL IciMapping Inclusive Composite Interval Mapping (ICIM) method (Meng et al. 2015).Mapchart 2.1 (Voorrips 2002) was used to draw a linkage map based on the genetic location determined with QTL IciMapping.The LOD score threshold was set using a 1,000-permutation test with a Type I error of 0.05 for QTL declaration.The QTL effects were estimated as phenotypic variation explained (PVE) and additive (Add) values by each QTL.

Identification of potential candidate genes in the flanking region of QTL mapping
Coding sequences (CDS) of the genes in the identified QTL flanking regions were extracted from the reference genome sequence of B. oleracea 'D134' to perform gene annotation with Blast2GO (Conesa et al. 2005).Genes related to disease resistance and defense responses were identified from Blast2GO description.CDS of the disease resistance genes identified from Blast2GO were used for blast search at www. arabi dopsis.org to find Arabidopsis homolog genes and the class of disease resistance proteins.

Search for A-genome syntenic regions of identified C-genome QTLs
The B. rapa reference genome sequence v3.0 (

Alignment of GBS short reads into the Brassica oleracea reference genome sequence of cabbage DH line 'D134'
In total 980.0 million (M) short reads were recovered from Illumina HiSeq2500 pair end sequencing from two replicates of parental lines and 124 BC 1 lines (Supplementary Table S2).About 17.0 M and 15.4 M short reads were obtained from the parental lines ECD11 and T010000DH3, from there 16.1 M (94.7%) and 14.2 M (92.4%) reads were aligned into the B. oleracea reference genome sequence.A total 947.6 M short read sequence from 124 BC 1 lines were identified ranging 0.9-13.9M/line, average 7.6 M/line.The average number of reads aligned into the reference genome sequence from each line was 6.9 M, ranging 0.8-12.6M and 84.4-93.9%(Supplementary Table S2).

Variants calling and polymorphic SNP selection
Total 8092 SNPs from 124 BC 1 plants were assigned to nine chromosomes of the reference genome sequence.The average number of SNP/chromosome was 899, ranging 646-1129 of the reference genome sequence 'D134'(Supplementary Table S3).The number of SNPs identified in the population was strongly correlated (R 2 = 0.84) with length of chromosomes in the reference genome sequence of 'D134' (Supplementary Table S3).Total 7405 and 8011 SNPs from the parental lines ECD11 and T010000DH3 were assigned to the 9 chromosomes (Supplementary Table S3).The number of SNPs identified in the parental lines ECD11 and T010000DH3 was strongly correlated (R 2 = 0.82, 0.84) with length of chromosomes in the reference genome sequence (Supplementary Table S3).Total 4,769 polymorphic SNP sites aligned to 9 chromosomes of 'D134' (Supplementary Table S3) and the number of polymorphic SNPs identified in the population were also strongly correlated (R 2 = 0.71) with length of chromosomes (Supplementary Table S3).Total number SNPs and polymorphic SNPs were also positively correlated (R 2 = 0.65) (Supplementary Table S3).

Construction of linkage map with comparison previously published map
A total of 1087 SNPs were filtered out in JoinMap and 2268 SNPs were filtered out by BIN function in ICIMapping.A genetic map consisting of 1,256.5 cM (Fig. S2) were constructed from the 1,414 non-redundant polymorphic SNP sites (Supplementary Table S3).The average number of SNPs mapped per chromosome was 157.1, ranging 126-185.The length of each chromosome ranged from 117.6 to 193.4 cM with an average length of 139 cM.The SNP interval of each chromosome ranged from 0.7 to 1.2 cM, with an average of 0.9 cM (Supplementary Table S3 and Table S4).

Identification of potential candidates genes for the QTLs
Genes in the flanking region of the seven QTLs were identified using CDS of the reference genome sequence of 'D134' by Blast2GO (Supplementary Table S5-8) and searched for candidate genes that encoded disease resistance proteins and defense-related genes (Table 3, Supplementary Table S9).
Rcr_C01-1 derived from T010000DH3 was identified in the flanking region of 7.49-8.39Mb of chromosome C01 for resistance to pathotype 2B.One hundred eleven genes were identified in the flanking region of Rcr_C01-1, where   3, Supplementary Table S9).
Rcr_C03-3 derived from ECD11 was identified in chromosome C03 in the flanking region of 35.22-36.28Mb, for resistance to pathotype 3A.Eighty-eight genes were identified in the flanking region of Rcr_C03-2, where no gene encoded disease resistance proteins or function related to plant defense response (Supplementary Table S6).
Rcr_C04-1 derived from T010000DH3 was identified in chromosome C04 in the flanking region of 51.28-52.14Mb for resistance to pathotype 3D.Seventy-three genes were identified in the flanking region of Rcr_C04-1, where no gene encoded disease resistance protein or function related to plant defense response (Supplementary Table S7).
Rcr_C08-2 derived from T010000DH3 was identified in chromosome C08 in the flanking region of 28.50-29.07Mb for resistance to pathotype 5K.Thirty-six genes were identified in the flanking region of Rcr_C08-2, where no gene encoded disease resistance protein or function related to plant defense response (Supplementary Table S8).
Polymorphic sequence variants identified in the QTLs flanking region between the two parents of chromosome C01, C03, C04 and C08 were searched.Out of 1139 genes in the flanking regions only 52 genes produced polymorphic variants (Supplementary Table S10).
Search for the syntenic regions of the two significant QTLs of C03 and C08 in the B. rapa 'Chiifu' reference genome sequence B. rapa is a major source of CR genes or QTLs for clubroot disease resistance in Brassica species.In this study, the BC 1 population was developed from B. oleracea with seven QTLs identified.More importantly, two QTLs Rcr_ C03-1 and Rcr_C08-1 were identified for resistance to 8 and 2 pathotypes, respectively.B. oleracea chromosomes C03 and C08 have higher synteny with B. rapa chromosomes A03 and A08, respectively (Perumal et al. 2020).Many CR genes have been identified in chromosomes A03 and A08 (Yu et al. 2021;Rahaman et al. 2022).Two QTLs Rcr_C03-1 and Rcr_C08-1 with TNL or CNL genes identified in this study were compared with A03 and A08 of B. rapa using B. rapa reference genome sequence 'Chiifu' v3.0 (Zhang et al. 2018).

Discussion
Two major QTLs, Rcr_C03-1 and Rcr_C08-1, and five  minor QTLs, Rcr_C01-1, Rcr_C03-2, Rcr_C03-3, Rcr_C04-1 and Rcr_C08-2, were detected from the current study on four chromosomes C01, C03, C04 and C08 against 10 pathotypes; 3A, 2B, 5C, 3D, 5G, 3H, 8J, 5K, 5L and 3O.Previous QTLs against different pathotypes were identified from different B. oleracea resistant sources on those chromosomes of C01, C03, C04 and C08 (Voorrips et al. 1997;Moriguchi et al. 1999;Rocherieux et al. 2004;Nagaoka et al. 2010;Lee et al. 2016;Peng et al. 2018).Farid et al. 2020 also identified QTLs on C03, C04 and C08 with two pathotypes, 3A and 5X-LG2 through the association mapping.In this study, ten pathotypes were used for identification of QTLs in B. oleracea through the bi-parental mapping method.As there was no common reference genome sequence and molecular markers that can be used for the identification of QTLs, it was not possible to find relationships of the QTLs identified in this study with those previously identified by others.
Due to the rapid emergence of new pathotypes and the A-genome CR resistance breakdown in 1st generation CR cultivars, it becomes an urgency to utilize polygenic resistance of B. oleracea.Until now, most of the CR genes identified in Canada from A-genome of B. rapa (Yu et al. 2017(Yu et al. , 2016(Yu et al. , 2021;;Chu et al. 2014;Huang et al. 2017;Gao et al. 2014;Karim et al. 2020;Rahaman et al. 2022) andB. napus (Fredua-Agyeman et al. 2016;Hasan et al. 2016;Zhang et al. 2016).One gene on chromosome A03 was used for developing the 1st generation CR cultivars in Canada.Recently, a report has been published on CR resistance of C-genome for resistance to pathotype 3 and 5X-LG2 (Dakouri et al. 2018).The study was based on RNA-seq.Compared to RNA-seq, QTL mapping from BC 1 /BC 1 S 1 lines could be used for identifying CR loci for multiple pathotypes effectively (Yu et al. 2017).Therefore, identification of CR resistance from C-genome with more pathotypes became a priority to expand B. napus CR resistance from C-genome.
In this study, we performed bi-parental QTL mapping with polymorphic SNPs through GBS from 124 BC 1 lines using a latest reference genome sequence 'D134'.Depending on different morphotypes in B. oleracea species, five reference assemblies were published (Parkin et al. 2014;Liu et al. 2014;Sun et al. 2019;Belser et al. 2018;Lv et al. 2020) as variability between different morphotypes is high.Two types of DNA sequencing technologies were used: short-read technology such as next-generation sequencing (NGS) and long-read technology such as Oxford Nanopore Technology (ONT) and Pacific Biosciences (PacBio).Two B. oleracea genome sequences available based on NGS technology: the TO1000 (kale-like; B. oleracea var.alboglabra) assembly (Parkin et al. 2014) and the 02-12 (cabbage; B. oleracea var.capitata) assembly (Liu et al. 2014), but their errors and gaps make them difficult to use for many studies (Lee et al. 2016;Liu et al. 2016Liu et al. , 2017;;Zhang et al. 2015).Another three B. oleracea genomes recently published based on combination of short and long-read technologies; the C-8 (cauliflower; B. oleracea L. var.botrytis) assembly (Sun et al. 2019), the HDEM (Broccoli; B. oleracea L. var.italica) assembly (Belser et al. 2018) and the 'D134' (Cabbage; B. oleracea var.capitata) assembly (Lv et al. 2020).Long-read technology with using efficient algorithms can provide highquality assemblies in chromosome-level (Jiao et al. 2017;Belser et al. 2018;Michael et al. 2018).As the CR parental line ECD11 is a cabbage cultivar, we used 'D134' reference genome sequence (Lv et al. 2020) as the template for SNP calling in this study.
The number of SNPs site per chromosome is often correlated with chromosome size (Yu et al. 2016).In this study, the number of SNPs and polymorphic SNPs identified in the population were strongly correlated (R 2 = 0.84, 0.71) of the reference genome sequence (Table S3).The total numbers of SNPs shows a better correlation with the chromosome length than the number of polymorphic SNPs, suggesting that polymorphic SNPs may be clustered.However, this may not be an issue for filtering, construction of genetic map and subsequent QTL mapping as BIN function was performed to remove redundant SNP loci using IciMapping.Although polymorphic SNPs seem more clustered but still total number SNPs and polymorphic SNPs were also positively correlated (R 2 = 0.65).Genetic linkage map is pre-requisite for QTL analysis and genetic mapping.Linkage map length could depend on population type and size, reference genome sequence size as well as number of polymorphic variants used for mapping.In this study, we constructed the linkage map with polymorphic variants of 124 BC 1 lines using the latest reference genome sequence 'D134'.It was 1,256.5 cM in length, longer than previously published B. oleracea linkage maps; 879.9 cM (Lee et al. 2016), 913.5 cM (Haque et al. 2017) and 1028.0 cM (Peng et al. 2018).
The R parental line ECD11 was moderate to highly resistant to all 10 pathotypes, while S parental line T010000DH3 was highly susceptible for all pathotypes.The F 1 plants from the interspecific crosses T010000DH3 × ECD11 were moderate to highly susceptible to all pathotypes and BC 1 S 1 population showed continuous distribution of %DSI (Fig. 1), which indicate that CR resistance of these 10 pathotypes is controlled by QTLs.Similar quantitative resistance was reported in B. oleracea for clubroot (Figdore et al. 1993;Nagaoka et al. 2010) as well black rot (Tonu et al. 2013) diseases.The frequency distribution in the BC 1 S 1 population Page 11 of 15 249 for resistance to some pathotypes also showed skewed tendency, similar skewed phenotype also reported for clubroot quantitative resistance in B. oleracea (Figdore et al. 1993;Peng et al. 2018;Ce et al. 2021).These results indicated that ECD11 clubroot resistance is controlled by quantitative manner with multiple genes effect.
A QTL that can be consistently detected with a PVE of > 10% of trait value can be designated as the main effect QTL or major QTL (Wang et al. 2019).In this study, one QTL, Rcr_C03-1 on C03 which contributed 16.0-65.6%phenotypic variation for 8 pathotypes; 2B, 5C, 5G, 3H, 8J, 5K, 5L and 3O (Table 2).Another QTL, Rcr_C08-1 were identified with 8.3 and 23.5% PVE for resistance to 8J and 5K, respectively (Table 2).As these two QTLs were identified for resistance to more than one pathotypes with PVEs of > 10% in the most cases.They can consider as major QTLs in this study.These two QTLs in Rcr_C03-1 and Rcr_C08-1 were identified with higher LOD value, PVE and Add values, indicating that they might have significant contribution for CR resistance in ECD11.Previously, one major QTL (Pb-Bo1) was detected on C01 for five isolates (Pathotypes 1, 2, 4 & 7; according to the classification by Some et al. 1996) and explained 20.7-80.7% of the phenotypic variation (Rocherieux et al. 2004).Five minor QTLs Rcr_C01-1, Rcr_C03-2, Rcr_C03-3, Rcr_C04-1 and Rcr_C08-2 were detected on chromosomes C01, C03, C04 and C08 which contributed 8.3-18.7%PVE for only one pathotype each (Table 2).A previous study reported that a single involvement of the major CR gene, or accumulation of CR genes in the minor CR-QTL, is not enough to confer sufficient resistance (Tomita et al. 2013).Further investigations for understanding individual or cumulative effect of identified major QTLs in ECD11 are needed.The values of Add for the five QTLs (Rcr_C01-1, Rcr_C03-1, Rcr_C03-2, Rcr_C03-3, Rcr_C08-1) on C01, C03 and C08 were positive, while for 2 QTLs (Rcr_C04-1 and Rcr_C08-2) on C04 and C08 were negative, indicating that the five resistant loci were derived from the resistant parent ECD11 and two resistant loci were derived from the susceptible parent T010000DH3.The presence of QTL for resistance to clubroot in T010000DH3 needs further confirmation.Previously, a resistant locus, PbBo(GC)1 identified from susceptible broccoli parent account for 9% of the phenotypic variation (Nagaoka et al. 2010).

Table 2
QTL position, interval and phenotypic variation explained (PVE) for resistance of 10 pathotypes of Plasmodiophora brassicae in B. oleracea cultivar ECD11 using Brassica oleracea reference genome sequence of cabbage DH line 'D134' QTL quantitative trait loci; LOD logarithm of the odds; PVE phenotypic variation explained; Add additive; CI confidence interval