Biparental populations based on crosses between elite breeding, conventional, or introgressed lines (e.g., Berry et al., 1995; Horn et al., 2003; Vera-Ruiz et al., 2006; Kane et al., 2013; Livaja et al., 2016) as well as landraces and wild species (e.g., Quillet et al., 1995; Kim and Rieseberg, 1999; Brouillette et al., 2007; Ma et al., 2017) have been employed in sunflower for mapping of genes, marker detection, QTL analyses and gene cloning. In addition, recombinant inbred lines (RILs) have been developed that as immortals can be maintained forever by self-propagation (e.g., Berrios et al., 1999; Tang et al., 2002; Tang et al., 2006; Poormohammad Kiani et al., 2007a; Talukder et al., 2016). However, biparental populations have three major disadvantages: (1) these populations have to be individually established for each research project requiring time and resources, (2) only two alleles per locus can be evaluated and (3) due to missing recombination events populations show low resolutions in mapping (Bernardo, 2008; Patrick and Alfonso, 2013). The use of association panels overcomes these problems. To verify the usefulness of association panels the genetic diversity between wild populations and the genetic diversity fixed in association panels were compared (Mandel et al., 2011; Filippi et al., 2015). Even though alleles present only in the wild populations were detected, the majority of the alleles were present in the investigated association panels.

The largest sunflower collection is handled at the Institute of Field and Vegetable Crops, Novi Sad, Serbia consisting of over 7,000 sunflower inbred lines developed from different genetic sources and 21 perennial and 7 annual species (447 accessions in total)¹ (Atlagic and Terzic, 2014). The next largest collection of more than 5000 cultivated and wild Helianthus accessions is held at the USDA-ARS NPGS in Ames (Marek, 2016). About half of these, 2,519 accessions, represent the world’s largest wild relatives sunflower collection, comprising 53 species – 39 perennial and 14 annual species (Seiler et al., 2017). Another large collection for sunflower (cultivated and wild) is maintained at the Vavilov Institute of Plant Industry, which consists of a total of 2,780 accessions from which 2,230 represent cultivated sunflower accessions and 550 wild sunflower accessions belonging to 24 species (19 perennials and 5 annuals) (Gavrilova et al., 2014). Some smaller numbers of 585 accessions of H. annuus are available via GRIN-CA, the Plant Gene Resources of Canada ² and additional 613 sunflower accessions of diverse origin are distributed by the IPK Gatersleben ³. These resources represent mostly uncharacterized plant material. In contrast to these, a well-defined collection of 400 open-pollinated varieties, landraces and breeding pools has been assembled by INRA to reflect the worldwide diversity present in sunflower (Mangin et al., 2017b). However, conservation of population diversity of sunflower populations represents a challenge in the maintenance process due to the self-incompatibility of wild sunflowers (Gandhi et al., 2005) and the possibility of genetic drifts occurring during the propagation of seed stocks (Mangin et al., 2017b). To study the preservation of the genetic variability, a set of 114 cultivated sunflower populations of the INRA collection were genotyped using a 384 Golden Gate SNP Assay. In conclusion, multiplication in isolation fields or use of cages is recommended to reduce loss of genetic variability in cultivated genetic resources.

These worldwide available collections of sunflower represent a valuable resource for the sunflower community. It could be of interest to include some additional accessions of these large collections to the existing association panels described below.

Association panels have to be characterized by molecular markers like SSRs or SNPs to avoid false associations due to the population structure and family relationship. The review here focusses on association panels that are online available as the prior mentioned sunflower collections. To analyze the primary gene pool of sunflower an association panel consisting of 433 cultivated accessions from North America and Europe in addition to 24 wild sunflower populations distributed over the whole of United States were characterized by 34 selected EST-SSRs chosen on the presumptive neutrality toward domestication and breeding efforts (Chapman et al., 2008; Mandel et al., 2011). USDA cultivated accessions in this panel were assigned to the following categories: HA and RHA being either non-oil or oil, landrace, open-pollinated variety (OPV), non-oil introgressed, oil introgressed, other non-oil, and other oil. The INRA accessions could only be categorized into INRA-HA and INRA-RHA as the information on oil and non-oil was not always available. Analyses using the software STRUCTURE (Pritchard et al., 2000) and Principle Coordinates (PCO) analyses (Patterson et al., 2006) did not reveal deep genetic divisions within the germplasm (Mandel et al., 2011). The cultivated and the wild populations separated into two different groups and within the cultivated accessions the restorer-oil (RHA-oil) category stayed apart from the remaining gene pool (Mandel et al., 2011). This is not unexpected due to the hybrid history of sunflower in which the maintainer and restorer pools have been kept separate on purpose to maximize heterosis (Fick and Miller, 1997). A selection of 288 accessions still covers nearly 90% of the genetic diversity available in the original larger panel. This association panel was named UGA-SAM1 and consists of 259 accessions, which are distributed by the Germplasm Resources Information Network (GRIN⁴) of the USDA National Plant Germplasm System (NPGS) and 29 accessions available through the French National Institute for Agricultural Research (INRA, France). This UGA-SAM1 population, which has been successfully employed in association studies (Mandel et al., 2013; Nambeesan et al., 2015), represents a very valuable tool for future association studies for the whole sunflower community. A minimal core set of 12 accessions (representing HA, RHA, oil and non-oil accessions as well as INRA material) capturing nearly 50% of the total allelic diversity might be ideal to build up a MAGIC (Multi-parent advanced generation intercross) population for sunflower. The MAGIC strategy is interesting for studies of multiple alleles in order to exploit higher recombination frequencies and better mapping resolution (Cavanagh et al., 2008). Development of MAGIC populations is in progress for numerous plant species (Bandillo et al., 2013) and would be interesting for sunflower as well.

Argentinean germplasm also represents a valuable genetic resource due to the long history of sunflower breeding in Argentina (de Bertero, 2003; Moreno et al., 2013). Filippi et al. (2015) characterized the population structure und calculated the genetic diversity of the association mapping population (AMP-IL), representing 137 INTA (National Institute of Agricultural Technology, Argentina) accessions and 33 accessions of open-pollinated and composite populations. The plant material is maintained by The Active Germplasm Bank of INTA Manfredi (AGB-IM). Using 42 SSR markers and/or SNP markers, detected by a 384 Illumina SNP-oligo pool array, estimated and observed heterozygosity as well as clustering using STRUCTURE and Discriminant Analysis of Principle Components (DAPC) were compared for the two marker types. As in other studies (Mandel et al., 2011, 2013) the population structure was dominated by the maintainer/restorer trait (Filippi et al., 2015).

A germplasm collection of 196 Spanish confectionary sunflower accessions is maintained at the Centre of Plant Genetic Resources of the National Institute for Agricultural and Food Research and Technology (CRF-INIA)⁵. A large genetic variation was revealed regarding hundred-seed weight, kernel percentage, seed oil content, fatty acid and tocopherol composition, phytosterols and other traits (Velasco et al., 2014; Pérez Vich et al., 2017).

In addition to well characterized association panels, considerable plant genetic resources are nowadays available in sunflower (cultivated as well as wild H. annuus accessions and accessions representing other species in the genus Helianthus).

To increase the naturally available genetic variability sunflower has been mutagenized (Zambelli et al., 2015). Mutant populations have been successfully developed and used to screen for mutant phenotypes interesting for breeding purposes with regard to flowering time, dwarf habitus, oil content, high oleic trait, herbicide resistance and branching (Soldatov, 1976; Gabard and Huby, 2001; Sala et al., 2008a; Cvejic et al., 2011b; Leon et al., 2013). Recently, a TILLING (Targeted Induced Local Lesion In Genomes) population for high throughput screening of EMS (ethyl methane sulfonate)-induced mutations in sunflower was established by Sabetta et al. (2011) and used for studies of genes involved in the fatty acid biosynthesis. Optimized mutagenesis using EMS was used to develop an additional sunflower TILLING platform (Kumar et al., 2013). Phenotypic characterization of 5,000 M2 lines was performed to estimate the mutation rates and to select interesting mutants. As seed oil biosynthesis is of major importance in sunflower, TILLING of FatA and SAD genes were investigated and revealed an overall mutation rate of one mutation every 480 kb (Kumar et al., 2013). Another possibility to develop mutant populations is to apply gamma irradiation or fast neutrons (Cvejic et al., 2011a). Optimal ranges for gamma irradiation and fast neutrons were explored in comparison to EMS concentrations.

Besides induced mutagenized populations, natural mutations that have occurred in wild sunflower populations have had significant impact on sunflower hybrid breeding, especially in the area of herbicide resistance. In the recent years intensive use of herbicides has led to the emergence of resistant wild sunflower populations. The first case was a population of common sunflower found in a soybean field in Rossville (KS, United States), in which imazethapyr that belongs the group of AHAS (acetohydroxy acid synthase) inhibitors was used over a time course of seven consecutive years for weed control. Thus, creating the first sunflower population, named ANN-PUR, resistant to one of the AHAS inhibitors (Al-Khatib et al., 1998). Resistance from this population was successfully introduced into commercial sunflower hybrids (Miller and Al-Khatib, 2000; Jocić et al., 2004). Sunflower production based on the use of this imidazolinone (IMI) resistance, which provides an efficient and easy control of post-emergence broadleaf weeds in Europe, is called Clearfield^® technology. In addition to the discovery of IMI resistant sunflowers, another population of wild sunflowers (ANN-KAN), tolerant to another AHAS herbicide group called sulfonylurea, was discovered in Kansas (United States) (Al-Khatib et al., 1999). The same tolerance was also obtained by EMS mutagenesis (Gabard and Huby, 2001). Later, more populations of wild sunflowers resistant to AHAS herbicides were found (e.g., White et al., 2002, 2003; Jacob et al., 2017). In addition, a new tolerance for imidazoline called Clearfield Plus^® was selected from an M2 population of 600,000 plants treated with EMS (Sala et al., 2008a).

Natural genetic diversity and naturally occurring or chemically/gamma-ray induced genetic variability represent a perquisite for selection in breeding. The wide range of accessions maintained and made available by the germplasm banks for the research community is an extremely valuable starting point for successful breeding programs in sunflower allowing association studies and introduction of new traits into existing commercial breeding material. However, mutagenesis can create additional new genetic variability in traits where the natural variability is not sufficient.

Developed linkage maps set a good basis for localization and mapping of simply inherited traits. Most of the downy mildew resistance genes, conferring resistance to the oomycete Plasmopara halstedii, have been found to be dominantly inherited and consequently, relatively easy to map by using molecular markers. Identification of closely linked markers also represents a good basis for map-based cloning of the genes.

Great efforts were put into examination of downy mildew resistance genes (R genes), designated as Pl genes, that are distributed throughout sunflower genome: Pl₁₃, Pl₁₄, Pl₁₆, and Pl_Arg on LG1; Pl₁, Pl₂, Pl₆, Pl₇, Pl₁₅, and Pl₂₀ on LG8; Pl₅, Pl₈, and Pl₂₁ on LG13; Pl₁₇ and Pl₁₉ on LG4; while Pl₁₈ is localized on LG2 of the sunflower SSR reference map (Mouzeyar et al., 1995; Roeckel-Drevet et al., 1996; Vear et al., 1997; Bert et al., 2001; Slabaugh et al., 2003; Yu et al., 2003; Mulpuri et al., 2009; Wieckhorst et al., 2010; Bachlava et al., 2011; Vincourt et al., 2012; Qi et al., 2015a; Qi L.L. et al., 2016; Zhang et al., 2016; Ma et al., 2017). Most of Pl genes are clustered, except for Pl_Arg and Pl₁₈.

The Pl cluster on LG8 was the first to be detected by molecular markers. Mouzeyar et al. (1995) used RAPD and RFLP markers for mapping the first downy mildew resistance gene, Pl₁, which is a part of a large Pl cluster (Pl₁, Pl₂, Pl₆, Pl₇). Of all genes in the cluster, Pl₆ gene was the most intensively examined since it conferred resistance to all the races present for a long time, except race 304. Different marker types were used for the introgression of Pl₆ into susceptible sunflower material including STS (Sequence-Tagged Sites) markers belonging to the TIR-NBS-LRR class of RGA (Resistance-Gene Analog) (Bouzidi et al., 2002) and co-dominant CAPS (Cleaved Amplified Polymorphic Sequence) markers (Table 1). The developed markers have been successfully used to introduce Pl₆ by MAS or to track the introduction of Pl₆ in backcrosses during conversion of downy mildew susceptible lines into resistant ones (Dimitrijevic et al., 2010; Jocic et al., 2010).

Two Pl genes originating from H. argophyllus, Pl₈ and Pl_Arg, were also subject of numerous studies. Pl_Arg confers resistance to all present downy mildew races and Pl₈ to 96% of all isolates collected in the north-central region of United States (Gilley et al., 2016). Radwan et al. (2004) developed STS markers for detection of the Pl₅/Pl₈ locus that were later on also explored in other sunflower genotypes (Dimitrijevic et al., 2011). Bachlava et al. (2011) developed two resistance gene candidate (RGC) markers, RGC251 and RGC15/16, closely linked to Pl₈ that belong to the group of SSCP (Single-Strand Conformational Polymorphism) markers. However, SSCPs are labor-intensive and time-consuming in MAS. A comprehensive study of Pl₈ by Qi et al. (2017) explored previously published SNP markers as well as two SSR markers (Bowers et al., 2012; Talukder et al., 2014a) to genotype a F₂ population derived from the cross HA434 × RHA340. The three closest SNP markers, NSA_000423, NSA_002220, and NSA_002251, were then investigated to check the specificity of the identified markers, concluding that NSA_000423 and NSA_002220 could serve as diagnostic markers in 87% of the tested sunflower lines when RHA 340 is used as donor for the Pl₈ gene. Validation of these markers across 548 sunflower lines proved their usefulness for MAS. However, a larger panel of sunflower lines should to be tested.

Unlike, the Pl₈ gene, Pl_Arg is not clustered. Several authors identified and developed different types of markers (SSRs, SNPs, RGCs) for MAS (Dußle et al., 2004; Wieckhorst et al., 2010; Imerovski et al., 2014b), some of these were also validated across a panel of sunflower lines. ORS716 was identified as the most useful marker in MAS (Table 1). Recently, Qi et al. (2017) combined available genomic data for the population obtained from the cross HA 89 × RHA 464 by use of SNP markers (Pegadaraju et al., 2013; Talukder et al., 2014a) with the phenotypic evaluation for resistance. The two nearest SNP markers (NSA_007595 and NSA_001835) narrowed the Pl_Arg locus down to an area of 2.83 Mb. The nine identified SNP markers represent valuable diagnostic tools for introgression of Pl_Arg into most genetic backgrounds in sunflower.

Other markers for use in MAS for downy mildew resistance include the identification of the tightly linked SSR marker ORS1008 to Pl₁₃ gene (Mulpuri et al., 2009), development of RGC markers tightly linked to Pl₁₄ gene (Bachlava et al., 2011) and the identification of one dominant co-segregating SSR marker (ORS1008) and one co-dominant tightly linked (EST)-SSR (HT636) to Pl₁₆. Interestingly, HT636 and ORS1008 were reported to be linked to both, Pl₁₃ and Pl₁₆, indicating that these genes are in close vicinity to each other (Liu et al., 2012a). Qi et al. (2015a) used SSRs to place Pl₁₇ onto LG4 and then used SNPs identified by the National Sunflower SNP Consortium (Talukder et al., 2014a) and by Bowers et al. (2012) to saturate the region surrounding the Pl₁₇ gene. The authors identified SNP SFW04052 and ORS963 as the closest flanking markers linked to Pl₁₇. A year later, Qi L.L. et al. (2016) used the same methodology to map Pl₁₈ to LG2 and found two SSRs and 10 SNPs flanking the Pl₁₈ gene. Pl₁₈ represents the first gene mapped to LG2. In 2017, two new Pl genes, Pl₁₉ and Pl₂₀, were reported and mapped to LG4 and LG8, respectively (Ma et al., 2017; Zhang et al., 2017). Two SSRs and two SNPs were mapped in close vicinity to Pl₁₉, while four SNP markers (SFW02745, SFW09076, S8_11272025, and S8_11272046) co-segregated with Pl₂₀. All markers can be used in MAS and most importantly in pyramiding Pl genes in order to achieve long lasting resistance toward downy mildew. The development of SNP markers is of special interest because of the large number of markers generated that increase the likelihood to have markers available for any cross combination.

Infections of sunflower plants with Puccinia helianthi Schwein lead to the rust disease. This fungus, which is mostly spread in North America, Argentina, South Africa, and Australia, can cause significant damage and yield reduction in infected fields. Genetic control of the disease can be effective; however, due to fast emergence of new races either by sexual or asexual reproduction, resistance achieved is short-termed. Consequently, a significant effort has been made into discovering rust resistance genes and the introduction into commercial lines and hybrids with a final goal of pyramiding several resistance genes in order to achieve long-term resistance. Most of the rust resistance genes (R genes), described so far, are monogenic dominant. R genes are located on different LGs of the sunflower genome with the majority being located on the LG13 [R₄, R_u6, R₁₁, R_adv, R_13a (R_HAR6), and R_13b] (Bachlava et al., 2011; Qi et al., 2011b, 2012b; Gong et al., 2013b; Bulos et al., 2014).

First molecular studies were conducted on discovering markers for R₁ and R_adv genes by use of RAPD and SCAR markers (Lawson et al., 1996, 1998). While R₁ gene was the first rust resistance gene present in a large number of sunflower lines, R_adv is present in the line P2 owned by Pioneer Hi-Bred Australia (Lawson et al., 1998; Qi et al., 2011a). R_adv is also present in the USDA line RHA 340, which Bachlava et al. (2011) used for mapping of the gene. Lawson et al. (1998) developed the SCAR marker SCT06₉₅₀ linked to R₁ gene, which proved to be useful for detection of R₁ in different genetic backgrounds, except for the sunflower line MC29, which carries the R₂ and R₁₀ genes. For mapping of R₂, Qi et al. (2015c) used a different MC29 line, called MC29 (USDA) as it was cultivated in the USDA-ARS Sunflower Research Unit, Fargo, North Dakota, which differs in term of resistance to NA race 6 in comparison to the MC29 line used by Lawson et al. (1998). Qi et al. (2015c) reported two SNP markers, NSA_002316 and SFW01272, flanking the R₂ gene on LG14. Since, the closest marker, SFW01272, can only to a certain extent be used to detect the R₂ gene across different genetic backgrounds; the authors recommend the use of two flanking SNP markers in order to minimize selection of false positives in MAS.

Further molecular studies of R genes include identification of molecular markers closely linked to R₄, R_adv, P_u6, R₁₁, R_13a (R_HAR6), and R_13b genes that are located on LG13. Qi et al. (2011b) identified two markers flanking R₄ gene (ORS581 and ZVG61) in the cross HA 89 × HA-R3, which were later also reported to be linked to rust resistance genes R_13a (R_HAR6) and R_13b located on the lower end of the LG13 (Bulos et al., 2013a; Gong et al., 2013b; Qi et al., 2015b) (Table 1). Further on, Gong et al. (2013b) saturated the region flanking the genes by analysis of RGC markers that were present in vicinity of downy mildew resistance gene Pl₈, which was also mapped in the lower end of LG13. Another R gene that mapped in vicinity of Pl₈ and fertility restorer gene Rf₁ was R_adv. A completely co-segregating SCAR marker (Lawson et al., 1998) as well as RGC and SSR markers tightly linked to R_adv were identified (Bachlava et al., 2011) (Table 1). Recently, Bulos et al. (2014) mapped P_u6 gene and identified closely linked SSRs to this gene in the sunflower line P386 on lower end of LG13. However, these markers are too far away to be useful in MAS (Table 1). P_u6 and R₄ map 6.25 cM apart from each other. Qi et al. (2012b) examined the R₁₁ gene and mapped it 1.6 cM from fertility restoration gene Rf₅ also on the lower end of LG13, hypothesizing the presence of a great rust R-gene cluster of R_adv/R₁₁/R₄. SSR marker ORS45 was the closest to R₁₁ gene and was mapped 1 cM proximal to the gene, while ORS728 was shown to be a common marker for R₁₁ and Rf₅ genes. The results allow the conclusion that the lower end of LG13 harbors the second largest cluster of NBS-LRR encoding genes: rust resistance and downy mildew resistance genes. Based on SSR and RGC markers used in this area, Gong et al. (2013b) proposed that this big cluster could be sub-divided into two clusters. R_adv and R₁₁ form sub-cluster I, while R₄, R_13a/b, Pl₅, Pl₈ form subcluster II. Pl₂₁ that was also positioned on LG13 mapped 8 cM proximal to Pl₅/Pl₈ (Radwan et al., 2004; Vincourt et al., 2012).

Other rust resistance genes investigated by use of molecular markers include analysis of R₅. This is to date the only R gene discovered on LG2. Qi et al. (2012a, 2015b) identified two SSR and two SNP markers flanking the gene, with the closest being 0.6 cM away (Table 1). On LG11 two rust resistance genes have been mapped so far: R₁₂ and R₁₄. Both genes were positioned in the middle of LG11, and were discovered in wild sunflower accessions, however, they have different origin: R₁₂ from PI413047 and R₁₄ from PI413038 (Gong et al., 2013a; Zhang et al., 2016). Both genes were mapped between the markers ORS1227 and ZVG53 (ORS1227 with 3.3 and 1.6 cM and ZVG53 with 9.6 and 6.9 cM from R₁₂ and R₁₄, respectively). Talukder et al. (2014b) performed fine mapping of the R₁₂ gene region by using SNP markers. Five SNP markers (NSA_000064, NSA_008884, NSA_004155, NSA_003320, and NSA_003426) were linked with 0.83 cM to the gene, but only two markers (NSA_003426 and NSA_004155) proved to have diagnostic quality for R₁₂ (Table 1). The nearest SNP marker to R₁₄ was NSA_000064, which was mapped with 0.7 cM from the gene in the F₂ mapping population obtained from the cross HA 343 × PH3 (Zhang et al., 2016). However, this marker amplified the same banding pattern in RHA 464 (R₁₂) and PH3 (R₁₄). Zhang et al. (2016) identified thirteen SSR/InDel and two SNP markers that amplified different profiles between the two donors of R₁₂ and R₁₄ indicating polymorphisms between these regions.

One of the latest efforts in saturation mapping of R genes was published by Qi et al. (2015b) who used previously developed SFW and NSA SNP markers in order to saturate the regions surrounding R₄, R₅, R_13a, and R_13b genes and succeeded in identifying markers that are under 1 cM distant from all analyzed genes thus raising the efficiency of introduction of rust resistance in to susceptible material (Table 1). The authors used previously developed SSR markers and newly developed SNP markers for identification of homozygous “double-resistant” F₂ individuals in a population obtained from a cross combination between a BC₃F₂ plant harboring R₅ and HA-R6 bearing R_13a. The F₄ progeny obtained from chosen plants showed improved resistance toward races 336 and 777 in comparison to lines that possess only one resistance gene. Qi et al. (2015c) also performed marker-assisted pyramiding of R₂ and R_13a in confectionary sunflower by use of SSR and SNP markers. Further pyramiding of R genes could lead to long-term improvements in sunflower rust resistance. The process of converting susceptible into resistant forms can be greatly facilitated and accelerated by use of the reported molecular markers.

Another constraint in sunflower production is broomrape (Orobanche cumana), a parasitic flowering plant, that can cause significant yield loss of up to 100%. Most of the genes that confer resistance to broomrape were found to be monogenic dominant for broomrape races A to E and G (Vranceanu et al., 1980; Velasco et al., 2012), while resistance to race F was either inherited by a monogenic dominant gene (Pacureanu-Joita et al., 1998; Pérez-Vich et al., 2004) or by two recessive genes (Rodríguez-Ojeda et al., 2001) depending on the genetic background. Broomrape resistance genes are denoted as Or genes. Imerovski et al. (2014a) reported a single recessive resistance gene in the sunflower line HA-267 that carried a resistance gene higher than Or₆. The majority of molecular analyses were conducted in investigating and creating different types of molecular markers for detection of Or₅ that conveys resistance to broomrape race E or lower (Lu et al., 2000; Tang et al., 2003a) (Table 1). The efficiency of RAPD and SSR primers in MAS for Or₅ were tested by Iuoras et al. (2004), however, none of the primers proved to be efficient or accurate enough. Imerovski et al. (2013) identified SSR markers associated with Or₂, Or₄, and Or₆ genes that could be used in converting broomrape susceptible sunflower genotypes into resistant ones. However, O. cumana populations belonging to race F have shown different aggressiveness (Molinero-Ruiz et al., 2009). Imerovski et al. (2016) mapped newly identified broomrape resistant gene conferring resistance to broomrape races overcoming race F from sunflower inbred line AB-VL-8 on LG3. The authors named the gene Or_ab-vl-8, which was shown to be recessive and ORS683 mapped 1.5 cM from the gene. Further molecular analysis are needed in order to develop co-segregating markers for some of the Or genes. In addition, finding novel resistance sources is essential since broomrape races are emerging at a high speed. Recent work of Louarn et al. (2016) involved using 586, 985 SNPs from SUNRISE project⁸ on GeneTitan^® (Affymetrix) for identification of QTL for resistance to broomrape races F and G. The authors identified 17 QTL spread throughout 9 LGs. Among them was a stable QTL on LG13 that controlled the number of broomrape emergence that explained 15–30% of the phenotypic variability. This QTL was marked as the one that could be the most rapidly used. A molecular characterization of O. cumana populations in Europe using RAPD-PCR identified four groups (Molinero-Ruiz et al., 2014). These markers might be useful as molecular tools to detect first broomrape appearances in fields that had been free of virulent races (Molinero-Ruiz et al., 2014).

Different tolerances against herbicides inhibiting the large, catalytic subunit of the acetohydroxyacid synthase (AHASL) have become a very necessary tool in sunflower hybrid production and cultivation as these facilitate the application of either imidazolinones (IMIs) or sulfonylureas (SUs) against broadleaf weeds (Sala et al., 2012). It also allows a race independent control of broomrape (Skoric and Pacureanu, 2010). Three AHASL genes were isolated from sunflower: AHASL1 located on LG9, AHASL2 on LG6 and AHASL3 on LG2 (Kolkman et al., 2004). Only mutations in AHASL1 seem to be involved in the herbicide tolerance in sunflower. Four different mutated alleles have been explored for commercial use in sunflower hybrid breeding: Imisun/Clearfield^®, Clearfield Plus^®, Sures and ExpressSun^® (Sala et al., 2012). Point mutations from C-T in codon 205 (Ahasl1-1) and in codon 197 (Ahasl1-2) (adopting the Arabidopsis nomenclature) confer moderate tolerance to IMIs and high tolerance to SUs, respectively. The allele Ahasl1-3 is characterized by a G-A mutation in codon 122 and results in high levels of IMI tolerance (Sala et al., 2008b). The broadest range of herbicide tolerance is shown by allele Ahasl1-4, which has a G-T mutation in codon 574 (Sala and Bulos, 2012). A first SNP marker based on the C-T change in codon 205 proved to be very useful as it cosegregated with partially dominant herbicide tolerance for the Imisun/Clearfield^® system (Kolkman et al., 2004), even though an additional non-target gene is required for the tolerance (Bruniard and Miller, 2001; Miller and Al-Khatib, 2002). One SSR marker exploiting the differences in the (ACC) repeats present in the AHASL gene allows the differentiation between the wild type Ahasl1 allele and alleles Ahasl1-1 and Ahasl1-2 (Sures and ExpressSun^®) (Kolkman et al., 2004; Bulos et al., 2013b). A CAPS marker developed by Bulos et al. (2013b) uses the A-T exchange to detect the Ahasl1-3 allele (Clearfield Plus^®) by digesting the PCR product with the restriction enzyme BmgBI. The markers can now help to select for herbicide tolerance. Nevertheless, the development of efficient screening tests for herbicide tolerance is crucial (e.g., Breccia et al., 2011; Vega et al., 2012).

Several oil properties have been characterized as quantitative traits, however, some traits such as oleic acid content (OAC) could, to a certain extent, be considered a semi-qualitative trait since OAC is dependent not only on the environment, but also on the genetic background of the receiver line (Ferfuia et al., 2015; Regitano Neto et al., 2016). A partial duplication of the FAD2-1 allele caused by chemical mutation leads to an increase in OAC by silencing the FAD2-1 gene encoding FAD2 (oleoyl-phosphatidylcholine desaturase) (Lacombe et al., 2002; Schuppert et al., 2006). This enzyme catalyzes the synthesis of linoleic acid from oleic acid and by silencing its activity oleic acid is accumulated. Soldatov (1976) created the Pervenets cultivar with elevated OAC, which has become the main source of elevated OAC in sunflower breeding programs worldwide due to the beneficial properties of high oleic sunflower oil (Allman-Farinelli et al., 2005; Vannozzi, 2006). Inheritance of the OAC trait has been a subject of numerous studies and different results were reported from a single dominant gene to several genes influencing OAC (Urie, 1984; Lacombe et al., 2004; Joksimovic et al., 2006; Bervillé, 2010; Ferfuia and Vannozzi, 2015; Premnath et al., 2016; Dimitrijevic et al., 2017). Gene/genes involved in inheritance of OAC have been denoted as Ol genes. Different markers were employed in mapping and detecting the mutation (Ol mutation) in sunflower. The two RAPD markers, F15-690 and AC10-765, were linked with 7.0 and 7.2 cM to Ol₁ gene, respectively (Dehmer and Friedt, 1998). Later on, the Ol₁-FAD2-1 locus was placed onto LG14 (Pérez-Vich et al., 2002; Schuppert et al., 2006). One major QTL identified by Pérez-Vich et al. (2002) explained 84.5% of the variation in the OAC. Schuppert et al. (2006) provided dominant INDEL markers for tracking the Ol mutation in addition to identifying 49 SNPs and five INDELs in the 3′-region of FAD2-1. Three years later, a co-dominant SSR marker tightly linked to the Ol mutation and dominant markers specific for the mutation were published (Lacombe et al., 2009). Recently, Premnath et al. (2016) identified in addition to the QTL on LG14, two additional QTL for OAC on LG8 and LG9. The two markers HO_Fsp_b for the QTL on LG14 (Schuppert et al., 2006) and ORS762 for the QTL on LG8 explained about 60% of the phenotypic variation in OAC. Several of the markers have been used for validation across numerous sunflower lines (Nagarathna et al., 2011; Singchai et al., 2013; Bilgen, 2016; Dimitrijevic et al., 2016). Dimitrijevic et al. (2017) reported marker F4-R1 created by Schuppert et al. (2006) as the most efficient in MAS for OAC.

Development of reliable tools for detection of cytoplasmic male sterility (cms) and restorer of fertility (Rf) genes would significantly improve and accelerate the process of developing sunflower hybrids. In sunflower, CMS PET1 originating from an interspecific hybridization of H. petiolaris with H. annuus (Leclercq, 1969) is the only CMS cytoplasm worldwide used for hybrid breeding. Male sterility is caused by the co-transcription of the atpA gene with the new CMS-specific orfH522 leading to the expression of a 16-kDa-protein (Horn et al., 1991; Köhler et al., 1991). Fertility restoration suppresses the co-transcription anther-specific (Monéger et al., 1994). In sunflower, the restorer genes for the PET1 cytoplasm represent the best characterized due to the commercial use of this cytoplasm in sunflower hybrid breeding. The restorer gene Rf₁, which was originally discovered by Kinman (1970) in the line T66006-2-1-B, has since then been integrated into a number of USDA/ARS RHA lines like RHA 271, RHA 272, RHA 273, and others (Korell et al., 1992; Serieys, 2005). A second major dominant restorer gene Rf₂ was discovered in a test cross between T66006-2-1-B and MZ01398. However, this Rf₂ gene seems to be ubiquitously present in almost all cultivated sunflower lines, along with maintainer lines of CMS PET1 (Serieys, 2005). Only Rf₁ is responsible for restoring male fertility in sunflower hybrids (Leclercq, 1984). RAPD markers in combination with AFLP markers were very useful for mapping of the restorer gene Rf₁ (Horn et al., 2003), which was first positioned on LG6 of the RFLP sunflower map (Gentzbittel et al., 1995). Two RAPD markers OPK13_454 and OPY10_740, which mapped 0.8 and 2.0 cM from Rf₁, respectively, were converted into more reliable, easier to handle SCAR markers HRG01 and HRG02 (Horn et al., 2003). A recent study of these SCAR markers for breeding practice proved that HRG01 is more efficient for Rf₁ detection in perennial species, whereas HRG02 gave better results for annual species (Markin et al., 2017). In addition a multiplex TaqMan assay was established that allowed the detection of HRG01 and orfH522 at the same time (Markin et al., 2017). Using the SSR markers ORS1030, Rf₁ had been mapped to LG13 (Kusterer et al., 2005) of the sunflower reference map (Tang et al., 2003b). In addition, a CAPS marker H13, which mapped 7.7 cM from Rf₁ gene, was developed from the RAPD marker OPH13_337 by digesting the PCR product with HinfI (Kusterer et al., 2005). The tight linkage between CAPS H13 and Rf₁ was confirmed in Xenia hybrid combination (Port et al., 2013). An additional SSR marker ORS511 and a TRAP marker K11F05Sa12-160 were mapped to the Rf₁ gene with distances of 3.7 and 0.4 cM, respectively (Yue et al., 2010). A fertility restorer gene Rf₃, which could be shown to be different from Rf₁ and Rf₂, was identified in the confectionery restorer line RHA 280 (Jan and Vick, 2007). Rf₃ could be linked with eight markers to LG7, including five known SSR markers (ORS328, ORS331, ORS928, ORS966, and ORS1092) and three new SSR markers HT-619-1, HT619-2, and HT1013 derived from expressed sequence tags (Liu et al., 2012b). SSR ORS328, which mapped 0.7 cM distant from Rf₃, represents so far the closest co-dominant marker to the gene (Liu et al., 2012b). Another restorer gene Rf₃ in RHA 340 has also been mapped to LG7 (Abratti et al., 2008). Rf ANN-1742, a restorer line derived from wild H. annuus showed resistance to rust (Qi et al., 2012b). The new rust resistance gene R₁₁ mapped with 1.6 cM closely to a restorer gene on the lower end of LG13. The SSR marker ORS728 was mapped 1.3 cM proximal from this restorer gene and 0.3 cM distal to R₁₁. Marker analyses using HRG01, HRG02, STS115, and ORS728 indicated that this restorer gene, now called Rf₅, might not be allelic to Rf₁ (Qi et al., 2012b).

So far, 72 new CMS sources have been described for sunflower (Serieys, 2005). However, only for very few of these CMS sources markers have been detected linked to the corresponding restorer genes (Horn et al., 2016). Feng and Jan (2008) tagged an additional restorer gene Rf₄ with molecular markers and assigned it to LG3 of the sunflower general reference map (Tang et al., 2003b). Rf₄ is restoring male fertility to a newly identified CMS cytoplasm GIG2. Schnabel et al. (2008) identified AFLP markers that mapped in close vicinity of the restorer gene Rf_PEF1, which represent a major restorer gene for the PEF1 CMS cytoplasm, another potentially interesting CMS source for commercial sunflower hybrid breeding. In addition, markers were developed that allowed the distinction between the PET1 cytoplasm and the PEF1 cytoplasm. For CMS 514A, a H. tuberosus based male sterile cytoplasm, the restorer gene Rf₆ was located on LG3 with eight markers (Liu et al., 2013). Two SSR markers, ORS13 and ORS1114, mapped as close as 1.6 cM to Rf₆. GISH showed Rf₆ to be present on a small translocation introgressed from H. angustifolius.

Further analyses are needed in order to develop more tightly linked molecular markers to Rf genes to locate them on the genetic map and to get an insight on the fertility restoration mechanisms in sunflower. In other species, most of the so far cloned restorer of fertility genes belong to the pentatricopeptide repeat gene family (PPR), however, also other types of restorer genes have been identified (Horn et al., 2014).