Introduction

Cultivated sunflower (Helianthus annuus L.), native to temperate North America, is one of the major global oilseed crops. Most currently used sunflower hybrids were derived from a germplasm pool intensely selected during the nineteenth century for high oil content (Baute et al. 2015). This effort narrowed the genetic base of cultivated sunflower against other biotic and/or abiotic stresses. However, the genus Helianthus comprises 14 annual and 39 perennial species (Schilling 2006; Stebbins et al. 2013), among which wild sunflower species have long been a valuable resource of genetic variability (for review, see Seiler et al. 2017). It has been reported that wild sunflower species harbor resistance (R) genes against downy mildew, rust, Verticillium wilt, Phomopsis stem canker, Phoma black stem, and diseases caused by Sclerotinia sclerotiorum (Gulya 2005; Seiler 2010). The identification of disease resistance genes from the crop wild relatives and their incorporation into cultivated sunflower have resulted in an effective strategy for providing durable sunflower disease resistance.

Sunflower downy mildew (DM), caused by pathogen Plasmopara halstedii (Farl.) Berl. et de Toni, occurs in all sunflower-growing regions throughout the world, except for Australia and New Zealand (http://www.cabi.org/isc/datasheet/41911; Constantinescu and Thines 2010). The pathogen can overwinter for many years as a thick-walled oospore in the soil. Under favorable environmental conditions, oospores produce zoosporangia, which release motile zoospores that infect germinating seedlings and cause a systemic infection that kills most infected plants (Humann et al. 2016; Markell et al. 2016). If infected seedlings survive the vegetative stages they will remain stunted and produce erect heads with few or no seeds. In wet and cool years, not only will infections reduce sunflower yield, but also affect the seed quality (Molinero-Ruiz et al. 2003). Management tools for DM include the destruction of volunteer hosts, fungicides applied as seed treatments, and planting DM-resistant hybrids; the latter being the most economically and environmentally-friendly tool (Humann et al. 2016).

The control of DM resistance in sunflower through dominant major genes, designated Pl, was first demonstrated in the early 1970s (Vrânceanu and Stoenescu 1970; Zimmer and Kinman 1972). The DM resistance in sunflower and DM virulence in pathogens are consistent with the gene-for-gene hypothesis (Flor 1955), contributing to the use of DM race-specific resistance as the main breeding strategy against downy mildew disease in sunflower. More than twenty Pl genes have currently been reported in cultivated and wild sunflowers, i.e., Pl 1 –Pl 19 , Pl 21 , and Pl Arg (Table S1, Zimmer and Kinman 1972; Miller and Gulya 1991; Mouzeyar et al. 1995; Roeckel-Drevet et al. 1996; Vear et al. 1997, 2008; Bert et al. 2001; Rahim et al. 2002; Molinero-Ruiz et al. 2003; Radwan et al. 2003; Vear et al. 2008; Mulpuri et al. 2009; de Romano et al. 2010; Bachlava et al. 2011; Liu et al. 2012; Qi et al. 2015, 2016a; Zhang et al. 2017). Among these resistance genes, Pl 13 , Pl 14 , Pl 16 , and Pl Arg are localized on linkage group (LG) 1 of the sunflower genome; Pl 1 , Pl 2 , Pl 6 , Pl 7 , and Pl 15 are localized on LG8; Pl 5 , Pl 8 and Pl 21 are localized on LG13; Pl 17 and Pl 19 are localized on LG4; and Pl 18 is localized on LG2. Most of these Pl genes were derived from wild H. annuus species, while some genes were derived from other wild Helianthus species (Vear et al. 2008). Pl 5 originated from H. tuberosus L. (Vrânceanu et al. 1981), whereas Pl 7 originated from H. praecox Englem. and Gray (Miller and Gulya 1991), and Pl 8 , Pl Arg , and Pl 18 were derived from H. argophyllus Torrey and Gray (Miller and Gulya 1988; Seiler 1991; Qi et al. 2016a).

However, new P. halstedii races frequently emerge and can render deployed resistance genes ineffective (Markell et al. 2016). To date, at least 24 P. halstedii races have been identified in Europe and 40 P. halstedii races have been identified in North America (Gulya et al. 2011; Gascuel et al. 2015; Viranyi et al. 2015), and virulence to many of the resistance genes deployed exists widely in sunflower production regions (Markell et al. 2016). The USDA-released inbred lines carrying Pl 6 and Pl 7 (Miller and Gulya 1988, 1991) have been extensively used to develop DM-resistant hybrids in sunflower production worldwide. In 2000, virulence to the widely-used resistance gene Pl 6 gene (race 304) was first identified in France (de Labrouhe et al. 2000; Viranyi et al. 2015). Between 2000 and 2008, six more races conferring virulence to Pl 6 were identified in France, namely races 307, 314, 334, 704, 707, and 714 (Ahmed et al. 2012). The first race conferring virulence to Pl 6 in the United States (U.S.) was reported in 2009 (race 734), and by 2010, five P. halstedii races (races 314, 704, 714, 734, and 774) conferring virulence to both Pl 6 and Pl 7 were identified in the country (Gulya et al. 2010, 2011). These newly identified races of P. halstedii pose a serious threat to sunflower production, and substantial economic loss is possible when commercial lines possess only the Pl 6 or Pl 7 single DM resistance gene. In 2014–2015, Gilley et al. (2016) tested 185 P. halstedii isolates collected from the U.S. sunflower production region of North Dakota, South Dakota, Minnesota, and Nebraska against 16 sunflower lines carrying 11 downy mildew R-genes; Pl 1 , Pl 2 , Pl 5 , Pl 6 , Pl 13 , Pl 15 -Pl 18 , Pl 21 , and Pl Arg . Only lines carrying Pl Arg , Pl 15 , Pl 17 , Pl 18 genes, and the lines RHA 468 and TX 16R [with an unknown gene(s)] were resistant to all isolates tested. With few effective resistance genes remaining, expanding and diversifying the DM resistance gene pool is needed to protect cultivated sunflower with DM resistance, thereby solidifying the basis of DM resistance breeding efforts.

Helianthus argophyllus (2n = 2x = 34), a wild diploid annual species, primarily occurs in beach environments or sandy habitats in Texas, U.S. (Rogers et al. 1982; Gulya 2005). This species is an important donor of different disease resistance genes in sunflower, such as rust resistance genes R adv and R 5 (Bachlava et al. 2011; Qi et al. 2012), DM resistance genes Pl 8 , Pl Arg , and Pl 18 (Miller and Gulya 1991; Qi et al. 2016a), and Sclerotinia resistance (Qi et al. 2016b). In an investigation of the DM resistance of H. argophyllus, Gulya (2005) inoculated 13 H. argophyllus accessions with a mixture of P. halstedii races with an aggregate pathogen phenotype of race 777. Of the 13 H. argophyllus accessions, six including the accession PI 494578 were highly resistant to DM, in which more than 90% of the tested individuals showed resistance.

Genotyping-by-sequencing (GBS) was designed and performed based on next-generation sequencing (NGS) for the efficient genotyping of a large number of samples. The simultaneous discovery and genotyping of thousands of SNPs can be achieved at 96- to 384-plex levels per flow channel on NGS sequencers, such as the Illumina HiSeq instrument. The genomic regions with low copy number are targeted and enriched using methylation-sensitive restriction enzyme(s) that efficiently reduce the genome complexity (Elshire et al. 2011). As an efficient and cost-effective method, GBS technology generates an enormous number of SNP markers, making it feasible to tag gene(s) from the genome in a timely manner. Here, we report the incorporation of a new DM resistance gene, Pl 20 , from wild H. argophyllus accession PI 494578 into cultivated sunflower and the molecular mapping of this novel gene to LG8 of the sunflower genome using a genotyping-by-sequencing approach.

Materials and methods

Plant materials and mapping population

H. argophyllus accession PI 494578 was collected at Premont, Texas, U.S. in 1984 (GRIN 2017). This accession was later characterized as highly resistant to DM with approximately 97% of tested plants showing resistance to a mixture of P. halstedii races (aggregate phenotyping of P. halstedii race 777) (Gulya 2005). The high oil Russian cultivar VNIMK 8931 (PI 262517) was a parent of HA 89 (PI 599773) susceptible to DM infection, that was released by the USDA and Texas Agricultural Experiment Station in 1971. The nuclear male sterile (NMS) HA 89 was produced through the chemically induced mutation of HA 89 using streptomycin, and this line possesses a single recessive nuclear male-sterile gene, ms9 (Jan and Rutger 1988; Chen et al. 2006). Cargill 270, a sunflower hybrid developed by the Cargill Company (Minneapolis, MN, USA), was used as susceptible control, and HA-DM1 (harboring DM R-gene Pl 18 ) (Qi et al. 2016a) was used as a resistant control in the downy mildew resistance tests of the BC1F3 of NMS HA 89 × H. argophyllus PI 494578.

The initial cross was made between NMS HA 89 and H. argophyllus PI 494578 in 2013. The open-pollinated and highly heterozygous nature of H. argophyllus made it necessary to test DM reaction of F1 hybrids. After screening against P. halstedii race 734, the resistant F1 individuals were selected, and the pollen was mixed to pollinate HA 89 and generate the BC1 generation. Similarly, DM-resistant BC1 individuals were selected to perform another round of crosses with HA 89 to generate the BC2 generation. The selected resistant BC2F1 plants were further advanced to BC2F2. Homozygous resistant introgression lines were selected from the BC2F2 population and advanced to BC2F3.

The BC1F2 population used for mapping the DM resistance gene from PI 494578 was developed from two resistant BC1F1 plants. A total of 207 BC1F2 plants were grown in the greenhouse in 2014, and 118 BC1F2:3 plants were harvested; 114 of these plants had sufficient seeds for further phenotypic analysis.

Downy mildew inoculation and phenotyping

The whole seedling immersion method according to Gulya et al. (1991a) and Qi et al. (2015) was used to test the reaction to downy mildew for sunflower seedlings. P. halstedii race 734, which overcomes the Pl 6 (residing in HA 335) and Pl 7 (residing in HA 337) resistance genes, was used to test the F1 to BC2F2 generations (Gulya et al. 2011). Further, six P. halstedii races, 314, 700, 710, 714, 734, and 774, were used to test the seedlings of the homozygous resistant BC1F3 families. The P. halstedii races 314, 700, 710, 714, and 774 were the most prevalent races observed in 2014 and 2015 in North and South Dakota, U.S. (Gilley et al. 2016). Sunflower lines Cargill 270 and HA 89 were used as susceptible controls, and HA-DM1 was used as a resistant control. A plant was considered susceptible (S) if sporulation was observed on cotyledons and true leaves and resistant (R) if no sporulation was observed.

For DM tests of the BC1F3 mapping population, 40 seeds from each of the 114 BC1F2:3 families were germinated at 22–24 °C in a germinator, and at least 30 seedlings were inoculated with isolate of P. halstedii race 734 in October, 2015. The F3 families were classified as homozygous resistant if none of the seedlings had sporulation, segregating if some seedlings (about one-quarter in a F3 family) had sporulation on the cotyledons and true leaves, and homozygous susceptible if all seedlings had sporulation on cotyledons and true leaves.

To select homozygous BC2F2 plants, 300 BC2F2 seedlings from the cross of NMS HA 89/PI 494578 were inoculated with P. halstedii race 734 on May, 2016. Ten days after inoculation, 191 resistant seedlings were identified, and 186 plants were transferred into 36-cell plastic flats (one seed per cell measuring 4.6 × 5.4 cm) for the selection of homozygous resistant plants using the SNP markers developed in the present study.

DNA extraction and genotyping-by-sequencing (GBS) analysis

Leaf tissue was collected from the parental lines, HA 89 and PI 494578, and 114 BC1F2 plants. Tissue samples were also collected from DM resistant BC2F2 plants of NMS HA 89 × PI 494578. Genomic DNA was isolated from the lyophilized tissues using the DNeasy 96 plant kit according to the manufacturer’s instructions (Qiagen, Valencia, CA, USA), and the quantity and quality of DNA were determined using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Wilmington, DE, USA).

After approval for DNA quality, 50 µl of genomic DNA from each of the 114 BC1F2 and two parental lines was submitted to the Biotechnology Resource Center (BRC) Genomic Diversity Facility at Cornell University for genotyping-by-sequencing according to Elshire et al. (2011, http://www.biotech.cornell.edu/brc/genomic-diversity-facility/services). The GBS library was constructed according to Elshire et al. (2011), with some modifications. Briefly, at the 96-plex level, the genomic DNA was digested with the restriction enzyme EcoT221, a six base cutter (ATGCAT), followed by ligation with a barcode adaptor and a common Illumina sequencing adaptor. After PCR amplification, the library was sequenced on the Illumina HiSeq 2000/2500 (100 bp, single-end reads). The FASTQ files from sequencing were pipeline-analyzed with the use of TASSEL-GBS discovery pipeline version 3.0.166 (http://tassel.bitbucket.org/TasselArchived.html) (Bradbury et al. 2007) to screen the barcoded (tagged) sequence reads of good quality. These sequence reads were aligned using the Burrow–Wheelers Alignment tool (BWA) version 0.7.8-r455 (Li and Durbin 2009) to the sunflower reference genome HA412.v1.0.bronze.20140814.fasta.gz (http://sunflowergenome.org/early_access/repository/main/pseudomolecules/). The TASSEL-GBS quantitative SNP caller was used for SNP calling after conversion into a ‘TagsOnPhysicalMap’ (TOPM) file. Identical SNPs were merged using the ‘MergeDuplicateSNPs’, and SNPs monomorphic between the two parents were removed. The sequences flanking SNP sites of LG8 were extracted for further use (Table S2). The chromosomes were renamed in a manner compatible with the GBS pipeline, where S1–S17 represents linkage groups 1–17 of the sunflower genome, and S18 represents those that were not assigned to any specific linkage group. The SNPs were named using a prefix of S1–S17 followed by a number representing the physical position of the SNP on the genome.

Linkage mapping

Software JoinMap 4.1 was used to perform linkage analysis (Van Ooijen 2006). The Chi square test was used to assess goodness-of-fit to the expected 1:2:1 segregation ratio of the markers using the ‘locus genotype frequencies’ feature of the JoinMap. Linkage maps were constructed using a regression mapping algorithm and the Kosambi mapping function (Van Ooijen 2006). A minimum likelihood of odds (LOD) ≥3.0 and a maximum distance of ≤50 centimorgans (cM) were used to test linkage among markers. The SNP markers resulting from GBS analysis were, respectively, classified into 17 LGs, and markers belonging to each of the 17 LGs were separately analyzed using DM phenotypic data to detect linkage.

Map saturation and conversion of SNP into PCR-based length polymorphism marker

To saturate the DM R-gene region, a total of 59 SNP markers potentially surrounding the region on LG8 were selected upon comparison with publically available genetic maps. Twelve SNPs were selected from the SNP map developed by the National Sunflower Association SNP Consortium (SNP prefixed with NSA, Table S3, Talukder et al. 2014), and 47 SNPs were selected from the published SNP map (SNP prefixed with SFW, Table S3, Bowers et al. 2012). Additionally, four SSR markers, ORS166, ORS185, ORS830, and CRT169, linked to the DM R-gene cluster in LG8 were included in the present study (Slabaugh et al. 2003). Four GBS generated SNPs flanking the novel DM R-gene from H. argophyllus PI 494578 and 59 selected SNPs were converted into PCR-based length polymorphism markers according to Long et al. (2016), with some modifications. Briefly, three primers were designed for each SNP, two-tailed forward allele-specific primers (AS-primer, F1 and F2) and a common reverse primer (Table 1). Two alternate permutations of artificial mismatches were introduced at the 3rd and 4th bases from the 3′ end in either of the AS-primers according to A/T → C, G → A, and C → T. An additional 5-base oligonucleotide (5′-ATGAC-3′) was inserted between the tail and allele-specific sequences of AS-primer F2 to generate length polymorphism between two alleles during amplification. A fluorescence labeled universal priming-element-adjustable primer (PEA-primer 5′-GCAACAGGAACCAGCTATGAC-3′) was applied in each polymerase chain reaction (PCR). The PCR amplifications were performed in a 15-μl reaction volume containing 1 × PCR buffer, 0.8 M betaine, 0.04% BSA, 2 mM MgCl2, 100 μM of each dNTP, 0.2 μM universal fluorescence-labeled PEA-primer, 0.2 μM common reverse primer, 0.04 μM each of AS-primers (F1 and F2), 1 U of Taq polymerase (Bioline, Randolph, MA, USA), and 10–20 ng of genomic DNA. The PCR amplifications were performed using a two-step protocol under the following conditions: initial denaturation of the template DNA at 94 °C for 3 min, 6 cycles at 94 °C for 20 s, and annealing at 55 °C for 30 s (1 °C decrease in each cycle) followed by another 33 cycles at 94 °C for 20 s and 60 °C for 30 s. The PCR products were diluted 40 times and visualized using an IR2 4200/4300 DNA Analyzer with denaturing polyacrylamide gel electrophoresis (LI-COR, Lincoln, NE, USA). The SSR markers were genotyped according to Qi et al. (2011).

Table 1 Primer sequences of the SNPs mapped in the present study
Full size table

Results

Introduction of downy mildew resistance from H. argophyllus into cultivated sunflower

DM resistance from wild sunflower species H. argophyllus was introgressed into cultivated oilseed sunflower through crosses and backcrosses. H. argophyllus PI 494578 was initially crossed with NMS HA 89 in January, 2013. The F1 plants were segregating for DM resistance because of the open-pollinated nature of wild species. A total of 38 F1 plants were tested for resistance against P. halstedii race 734. Twenty F1 plants were resistant, and 18 plants were susceptible. The selected resistant F1 hybrids were used as male parents in the first backcross to HA 89 to create BC1 in November, 2013. Similarly, the selected resistant BC1 plants from the DM tests were again used as male parents to backcross with the recurrent parent HA 89 in June, 2014. The BC2F1 seeds were planted in greenhouse for DM testing with race 734, and the resistant seedlings were selected and self-pollinated for BC2F2 in March, 2015 for further selection of homozygously resistant plants.

Genetic mapping of downy mildew resistance from PI 494578

A total of 118 BC1F2:3 families were harvested in 2014, and four of these plants did not germinate well in 2015. Thirty seedlings each from the 114 BC1F2:3 families were inoculated with P. halstedii race 734 in a greenhouse in October, 2015. The DM reaction data suggested that 27 families were homozygous susceptible, 57 families were segregating, and 30 families were homozygous resistant. The observed ratio of 27:57:30 did not significantly differ from the theoretical 1:2:1 ratio (χ 2 = 0.16, df = 2, P = 0.9231) according to Chi square test, suggesting that the resistance to P. halstedii race 734 in this mapping population was conferred by a single dominant gene.

A total of 118,577 SNP markers were generated from GBS, which were not evenly distributed among the 17 LGs of the sunflower genome (Table S4). A total of 18,504 (15.6%) SNPs were polymorphic between the two parents, HA 89 and H. argophyllus PI 494578. These polymorphic SNP markers were subjected to Chi square tests, and those showing significant segregation distortion from the expected 1:2:1 ratio were excluded, leaving a total of 336 SNPs, ranging from 0 SNP on LGs 6, 7, and 16 to 191 SNPs on LG8, for further linkage analysis (Table S4). The SNPs from each of the 17 LGs were subsequently analyzed using the DM phenotypic data, and the results indicated that the DM resistance from PI 494578 was associated with the SNP markers on LG8. The LG8 genetic map was constructed with 187 SNP markers encompassing 79.4 cM of genetic distance (Table S5). SNP markers were evenly distributed in LG8, except for a large marker cluster with 53 SNPs in an interval at the position of 61.9–63.9 cM. The DM R-gene derived from PI 494578 was located on the upper end of LG8 and co-segregated with marker S8_11272046. Since no other Pl gene derived from H. argophyllus has been mapped to LG8, the Pl gene introgressed from H. argophyllus PI 494578 is a novel DM resistance gene, designated Pl 20 .

Saturation mapping of the Pl 20 region

A total of 59 SNP and four SSR markers most likely in the Pl 20 gene region on LG8 were screened for polymorphisms between the two parents, HA 89 and PI 494578 (Table S3). Fifteen SNPs and three SSRs were polymorphic, respectively. In addition, we converted four GBS SNPs, S8_7763647, S8_11272025, S8_11272046, and S8_100385559, flanking Pl 20 into PCR-based length polymorphism markers (Table 1). All the tested markers produced co-dominant polymorphic amplicons between the parents, except for one NSA SNP marker, NSA_005795, which produced a dominant amplicon in HA 89 rather than in PI 494578. The BC1F2 population derived from HA 89 and PI 494578 was used for genotyping with these polymorphic markers and the PCR-based GBS SNPs. Eight SNPs were mapped to a region surrounding the Pl 20 gene region, whereas none of the SSR markers was mapped in this population. The segregation of PCR-based GBS SNPs in the population was consistent with the GBS genotyping data, except for one BC1F2 individual (14-207a-18) differing between PCR-based and GBS genotype for marker S8_100385559 (Fig. 1). As expected, the Pl 20 gene co-segregated with four markers, SFW02745, SFW09076, S8_11272025, and S8_11272046, and was placed in an interval of 1.8 cM, with 0.1 cM distance from marker SFW04358, and 1.7 cM distance from S8_100385559 (Fig. 2).

Fig. 1
figure 1

The segregation of PCR-based and GBS genotyping data of SNP markers S8_11272046 and S8_100385559 in the BC1F2 population and parents HA 89 and PI 494578. (Asterisk) The genotype of the PCR-based SNP was different from that of GBS data

Full size image
Fig. 2
figure 2

Part of genetic map of sunflower linkage group 8, showing the Pl 20 position on the map

Full size image

Physical location of Pl 20 and DM R-gene cluster on LG8

The Pl 20 gene is the sixth DM R-gene mapped on LG8, close to the DM R-gene cluster on this LG, which includes Pl 1 , Pl 2 , Pl 6 , Pl 7 , and Pl 15 (de Romano et al. 2010; Gascuel et al. 2015). A BLAST search against HA412.v1.1.bronze.20141015 database revealed that SSR marker ORS166 is tightly linked to the DM R-gene cluster, and the TIR3F1R2 marker within candidate Pl 6 gene was upstream of Pl 20 (Table 2, Slabaugh et al. 2003; de Romano et al. 2010). Although ORS166 did not map to the region surrounding the Pl 20 gene in the present study, this locus was physically close to Pl 20 . ORS166 and TIR3F1R2 are located at 10,687,506 and 10,939,101 bp on LG8, respectively, whereas the Pl 20 gene is located in the region from 11,271,845 to 13,781,094 bp on LG8, proximal to the DM gene cluster (Table 2). The genetic position of GBS SNP_100385559 was confirmed when it was converted to a PCR-based marker in the population; however, the physical position of this SNP conflicted with its genetic position (Fig. 1; Table 2).

Table 2 Genetic and physical positions of molecular markers flanking the Pl 20 gene and markers linked to the Pl 6 /Pl 7 cluster in linkage group 8 (LG8); the physical length of LG8 is 192,129,815 bp
Full size table

DM resistance spectrum of Pl 20 in cultivated sunflower background

Downy mildew resistance in a homozygous resistant BC1F2:3 line, 14-207-58, along with the recurrent parent HA 89, as well as susceptible (Cargill 270) and resistant (HA-DM1) checks was further evaluated for resistance against additional P. halstedii races, 314, 700, 710, 714, and 774 and race 734. These selected P. halstedii races represent the most predominant and virulent races currently identified in North America (Gulya et al. 2011; Gilley et al. 2016). As expected, HA 89 and Cargill 270 were susceptible to all six races tested, and abundant white sporulation was observed on the leaf surfaces of the seedlings. In contrast, the homozygous resistant line 14-207-58 carrying the Pl 20 gene and HA-DM1 harboring the Pl 18 gene were immune to all P. halstedii races tested, with no sporulation on cotyledons or true leaves (Table 3).

Table 3 Downy mildew resistance tests of homozygous BC1F3 family (14-207-58) with six P. halstedii races
Full size table

Selection of homozygous resistant introgression lines from BC2F2 populations

Three hundred BC2F2 seedlings from NMS HA 89/PI 494578 were inoculated with P. halstedii race 734. One hundred and ninety-one resistant seedlings were identified, and 186 R-plants were transferred to 36-cell plastic flats for the selection of homozygous resistant plants. Three SNP markers, SFW04358, S8_11272025, and S8_100385559, linked to the Pl 20 gene, were used to genotype the selected resistant BC2F2 individuals, together with their parents, HA 89 and PI 494578. A total of 62 plants were identified as homozygous resistant, showing identical PI 494578 marker alleles from the 186 plants tested, and 32 of these plants were advanced to the BC2F3 generation.

Discussion

Many resistance genes in plants are organized into gene clusters along the chromosomes, widely observed in sunflower Pl genes (Islam and Shepherd 1991; Jones et al. 1993; Song et al. 1997; Meyers et al. 1998; Huang et al. 2003). Two (Pl 17 and Pl 19 ), three (Pl 5 , Pl 8 and Pl 21 ), three (Pl 13 , Pl 14 , and Pl 16 ), and five (Pl 1 , Pl 2 , Pl 6 , Pl 7 and Pl 15 ) Pl genes were, respectively, clustered in LGs 4, 13, 1, and 8 of the sunflower genome, as reported in previous studies (Table S1). Among these genes, Pl 1 , Pl 2 , Pl 6 , Pl 7 , and Pl 15 and the rust resistance gene R 1 are clustered in the upper end of LG8, which is one of the largest gene clusters in sunflower. In the present study, we also mapped a novel DM R-gene, Pl 20 , to this cluster. The SSR marker, ORS166, which has been linked to the Pl 15 and Pl 1 –Pl 2 –Pl 6 –Pl 7 cluster, was not mapped in the Pl 20 population, but rather mapped to a physically close location with the Pl 20 gene (Table 2, Slabaugh et al. 2003; de Romano et al. 2010).

Although Pl 20 was mapped to the DM R-gene cluster in LG8, this gene has a different origin and DM resistance specificity from other DM R-genes in this cluster. The origin of Pl 1 and Pl 2 can be traced to a Canadian line 953 involving wild H. annuus (Zimmer and Kinman 1972; Vear et al. 2008). Pl 6 and Pl 7 originated from wild H. annuus and H. praecox, respectively (Miller and Gulya 1988, 1991), and Pl 15 was identified from an Argentinian restorer inbred line RNID (de Romano et al. 2010), whereas Pl 20 was derived from H. argophyllus. Additionally, Pl 1 and Pl 2 are no longer effective against downy mildew infection, and Pl 6 and Pl 7 are susceptible to virulent P. halstedii races, whereas Pl 20 is resistant to all P. halstedii races tested in the present study, which are the predominant and virulent races currently identified in North America (Gilley et al. 2016). Altogether, Pl 20 is a novel and effective DM resistance gene located on the DM R-gene cluster on LG8.

Linkage group 8 was reported to harbor the largest nucleotide-binding site and leucine-rich repeats (NBS-LRR) in sunflower (Radwan et al. 2008). A total of 54 NBS-LRR loci were discovered on LG8, and three subclusters were detected in the 36-cM region. The Pl gene cluster in LG8 was associated with a TIR (toll-interleukin receptor)-NBS-LRR subcluster (Bouzidi et al. 2006; Radwan et al. 2008; Franchel et al. 2013) within the first and largest subcluster comprising TIR members exclusively in LG8.

Helianthus argophyllus is a wild annual species primarily located in the U.S. Among the 94 H. argophyllus accessions deposited in the Germplasm Resources Information Network (GRIN), 78 accessions were located in the U.S., 11 accessions were located in Australia, two accessions were located in former Soviet Union, two accessions were located in Mozambique, and one accession was of unknown origin (http://www.ars-grin.gov/cgi-bin/npgs/html/tax_search.pl). The state of Texas is the main reservoir containing H. argophyllus in the U.S., including a total of 75 accessions, with only one accession observed in Florida (PI 468651) which is the donor of the Pl Arg gene and two accessions observed in North Carolina (PI 664729 and PI 664730). Unlike the clustered DM R-genes from wild H. annuus, the DM R-genes from H. argophyllus are widely distributed across the sunflower genome. The four Pl genes, Pl Arg , Pl 8 , Pl 18 , and Pl 20 , discovered in H. argophyllus, were separately located on LGs, 1, 13, 2, and 8, and all but Pl Arg were originally collected from Texas. It remains unknown how the host-pathogen co-evolved to develop diversified DM R-genes in H. argophyllus in a narrowed geographic area (Qi et al. 2016a).

GBS holds a number of interesting points that can benefit both scientists and breeding communities, such as time-, labor-, and cost-efficiency. An enormous number of SNPs can be produced in a relatively short time compared to other marker systems that function as an effective tool in such applications as population genetics, quantitative trait locus discovery, genome-wide association study, and high-resolution mapping. However, there are some inevitable and inherent limitations when conducting GBS analysis, particularly when dealing with the F2 population. Theoretically, 50% of the SNP sites called are expected in a heterozygous condition in the F2 population. Higher sequencing coverage, more accurate calling, and more sophisticated bioinformatics are required to call heterozygotes (Hyma et al. 2015). In the present study, one of the mapping parents, H. argophyllus PI 494578, is a wild sunflower species with an open-pollinated nature, making this species highly heterozygous between alleles, further complicating heterozygote calling. In the mapping population in the present study, a high degree of segregation distortion was observed, in which only 0.26% of SNPs fit the expected Mendelian ratio (Table S4). The number of F2 individuals carrying heterozygous genotypes was significantly reduced, with a corresponding excess of HA 89 homozygous genotypes. In some LGs, all the generated SNPs showed segregation distortion, such as those on LGs 6, 7, and 16; however, LG8 harbored 56.8% (191/336) of SNPs with no segregation distortion (Table S4). After GBS analysis of the localized Pl 20 gene on LG8, we further ‘post-processed’ and validated the linkage, i.e., converting the GBS-generated SNPs into PCR-based length polymorphism. The PCR-based genotyping data for closely linked GBS markers S8_11272025, S8_11272046, and S8_100385559 precisely match those from the GBS analysis for each of the 114 BC1F2 individuals, except of one individual (Fig. 2), further confirming the linkage relationship of Pl 20 with these markers.

Downy mildew resistance from wild sunflower species can be introgressed into cultivated sunflowers to provide resistance to many/or all currently known races of the pathogen. In the present study, a new DM R-gene, Pl 20 , was introgressed into cultivated sunflowers, which could further be used for R-gene pyramiding. It has previously been reported that P. halstedii pathogens co-evolve with host resistance genes, leading to the ineffectiveness of existing downy mildew R-genes (Gascuel et al. 2015; Viranyi et al. 2015; Gilley et al. 2016). The strategies typically utilized in sunflower production are to first continuously search for new resistance, and second, conduct gene pyramiding in which multiple resistance genes are combined into a single genotype. Reaction to P. halstedii infection showed that the Pl 20 gene is effectively resistant to the most predominant and virulent races currently identified in North America. This novel R-gene has not yet been incorporated into any commercial lines and provides a very high level of resistance (immunity) to the disease. An ongoing R-gene pyramiding project is underway to provide multiple resistances to cultivated sunflower against downy mildew infection.

Author contribution statement

Conceived and designed the experiments: LLQ, GJM. Performed the experiments: GJM, LLQ, SGM. Analyzed data: GJM, QJS, LLQ. Wrote the paper: GJM, LLQ, Commented on the manuscript before submission: SGM, QJS.