Validation of Diaporthe toxica resistance markers in European Lupinus angustifolius germplasm and identification of novel resistance donors for marker-assisted selection

The fungus, Diaporthe toxica, anamorph Phomopsis sp., previously classified as P. leptostromiformis, is a plant endophyte and occasional pathogen, causing Phomopsis stem blight. This disease is damaging not only to lupins but also to the animals grazing on infected plants, due to the toxic secondary metabolites called phomopsins. The aim of this work was to validate markers for resistance to Phomopsis stem blight in narrow-leafed lupins and identify novel germplasm with increased levels of resistance to the disease. Plant inoculations were performed using ten isolates of D. toxica, originating from Australia and Poland. The European core collection of L. angustifolius was evaluated both in a controlled environment and with field experiments to classify the accessions based on their resistance to the disease. Simultaneously, the accessions were assayed with disease resistance markers to identify donors of hypothetical resistance alleles. We have found that the European lupin germplasm collection preserves wild and domesticated donors of at least two resistance genes to Phomopsis stem blight, including Phr1 and PhtjR. Molecular markers PhtjM7, InDel2, and InDel10, tagging PhtjR gene, were applicable for marker-assisted selection targeting the European gene pool with an expected accuracy of 95%. None of diagnostic markers for the Phr1 locus was found useful for European breeding programs; two existing markers Ph258M1 and Ph258M2 were unreliable, due to a high percentage of false-positive results (up to 58%) and a high recombination rate between markers (~ 30%). Electronic supplementary material The online version of this article (10.1007/s13353-019-00521-y) contains supplementary material, which is available to authorized users.


Introduction
The legume Lupinus angustifolius L. (narrow-leafed lupin) belongs to the genus Lupinus (tribe of Genisteae, family Fabaceae, subfamily Faboideae). It is well known as a source of protein for food and feed, as well as being a crop that contributes to the improvement of soil structure and fertility, increasing yields of succeeding crops (Peoples et al. 2009).
Registered lupin cultivars are characterized by moderate grain yield with a high content of protein and oil, accompanied by low alkaloid content and limited fiber (Cernay et al. 2015;Lucas et al. 2015). Due to its relatively low chromosome number (2n = 40) and small genome size (2C = 1.89 pg), compared with other lupins (Naganowska et al. 2003), L. angustifolius became the species of choice for extensive molecular studies. Molecular research has been greatly facilitated by the development of bacterial artificial chromosome (BAC) libraries of the nuclear genomes for two L. angustifolius cultivars: Polish cv. Sonet (Kasprzak et al. 2006) and Australian cv. Tanjil (Gao et al. 2011). Highdensity linkage maps carrying thousands of markers, including gene-based sequence tagged sites (STS), were constructed and aligned to the draft genome sequence (Hane et al. 2017;Kamphuis et al. 2015;Nelson et al. 2006;Yang et al. 2013b). Exploitation of BAC resources for cytogenetic mapping resulted in the integration of all linkage groups with the corresponding chromosomes, as well as in the identification of several gene-rich regions (Książkiewicz et al. 2013Leśniewska et al. 2011;Przysiecka et al. 2015;Wyrwa et al. 2016). The release of reference transcriptomes for wild and domesticated lupin accessions constituted a platform for targeting particular genes of interest .
Despite its dynamic domestication history and high nutritional value, worldwide use of lupin for livestock feed has been hampered by high alkaloid content and the risk of lupinosis disease. The alkaloid issue has been almost entirely solved due to the discovery of three heritable factors decreasing alkaloid content (iucundus, depressus, and esculentus) (Hackbarth and Troll 1956). Incorporation of these alleles into breeding lines resulted in development of germplasm with a greatly reduced total alkaloid level, more than hundredfold lower than that of old cultivars (Kamel et al. 2016). However, lupinosis still remains a serious threat for animals grazing on lupin stubble (Cowley et al. 2014). The chemical factor causing lupinosis was revealed to be a phomopsin, a toxin produced by pathogenic fungi, Diaporthe toxica Will., Highet, Gams & Sivasith, anamorph Phomopsis sp. (Jago et al. 1982;. As the toxin is produced during the latent stage of stem infection of susceptible plants, exploitation of heritable resistance resources is a prerequisite of further lupin improvement. The fungus, formerly known as Phomopsis leptostromiformis (Kühn) Bubák, causes the lupin disease Phomopsis stem blight. Methods of screening for Phomopsis stem blight rely on observations of lesion coverage as percentage surface area on senescent stems (Cowling et al. 1987) or staining and microscopic examination of subcuticular coralloid hyphae structures of infected stems (Williamson et al. 1991). Additionally, a non-destructive glasshouse infection test has been developed, based on inoculation of lateral branches regenerating from the second main stem node, after topping the main stem above this node . Over many years, D. toxica has been a major problem in Australia, where lupin was introduced as a winter crop. Australian breeders responded with the introduction of a moderately resistant wild population line from Morocco, CPI65211A, into a cross derivative of cv. Marri and P22872, which resulted in development of elite breeding line, 75A:258 (Cowling et al. 1987). The line is still used as a reference in phytopathological assays as its Phomopsis stem blight resistance has never been broken despite wide implementation of this genotype in Australian breeding programs (Shankar et al. 1996Stefanova and Buirchell 2010;Yang et al. 2015Yang et al. , 2013aYang et al. , 2002. In Europe, D. toxica seems to be a dormant pathogen as reports of the disease date from as early as 1880 in Germany and in 1892 in Denmark (on L. angustifolius and L. luteus) but it has never caused serious problems (Fischer 1893;Lind 1913). Although it has appeared from time to time in different countries in Europe, including Poland and Russia (Lewartowska et al. 1994;Marcinkowska 2007).
Australian lupin collection revealed extensive accession-related diversity in the severity of disease symptoms developed, from very susceptible lines (Unicrop, Uniharvest, Uniwhite), through susceptible (Chittick, Danja, Geebung), somewhat resistant (Merrit, Tanjil, and Wonga), to highly resistant (75A:258) (Yang et al. 2015;. There are at least three different genetic sources of D. toxica resistance in L. angustifolius, all originating from an Australian collection. Reference germplasm resources for these genes are line 75A:258 (Phr1 gene), cultivar Merrit (Phr2) and cv. Tanjil (PhtjR). With the use of the molecular fragment length polymorphism (MFLP) technique, markers linked to the putative Phr1 resistance gene were designed, Ph258M1 and Ph258M2 (Yang et al. 2002). Next-generation sequencing of restriction site-associated DNA fragments was exploited to develop a set of single nucleotide polymorphism (SNP) markers linked to the PhtjR gene, namely PhtjM4, PhtjM5, and PhtjM7 (Yang et al. 2013a). Recently, a whole-genome resequencing approach was harnessed to develop a new set of low-cost markers tagging the PhtjR gene, including insertion/deletion PCR markers Markers InDel2 and InDel10, were found to be an effective diagnostic method on a broad range of Australian commercial cultivars (Yang et al. 2015).
Numerous undesired traits were eliminated during the history of narrow-leafed lupin domestication, including vernalization responsiveness, pod shattering, hard seed coat, bitter taste, and susceptibility to anthracnose. L. angustifolius breeding programs are most advanced in Australia; however, this process was largely based on two European genotypes (Borre and New Zealand Blue) and subsequently only occasionally improved with externally sourced germplasm (Berger et al. 2012). Only a small fraction of the available genetic and adaptive diversity was exploited during the domestication process of the species (Berger et al. 2013). Australian D. toxica-resistant cultivars cannot be exploited as highly transferable donors of resistance alleles due to significant genetic diversity constraints. The necessity for exploitation of wild populations and primitive forms to bypass the domestication bottleneck has emerged.
To address limitations of current breeding programs, we decided to leverage sources of D. toxica resistance from European L. angustifolius germplasm combining markerassisted and classical approaches. Here, diagnostic procedures of Phomopsis stem blight resistance markers were optimized and the European core collection of L. angustifolius was assayed with these markers to identify hypothetical donors of resistance alleles. The resistance of selected narrow-leafed lupin lines was evaluated both in controlled environment tests and in field experiments to validate marker-trait associations as well as to identify novel germplasm with increased levels of Phomopsis stem blight resistance.

Isolates of Diaporthe toxica
A collection of ten isolates of D. toxica was established (Table 1). Five isolates were collected from L. luteus plants, with four cultivated on fields located in the Wiatrowo Breeding Station of Poznan Plant Breeders and one was growing in the wild in Western Australia. The other five isolates were obtained from L. angustifolius, all of them were collected in Australia, mostly in Western Australia (4 sites) and one isolate was obtained from L. angustifolius collected in South Eastern Australia. The cultures of D. toxica (Phomopsis) were grown on PDA medium and the sporulation was induced under NUV light in growth chambers at 20°C in the darkness.

Preliminary resistance survey
This experiment was conducted in 2007 in a phytotron (MYTRON Bio-und Solartechnik GmbH, Heiligenstadt) in controlled conditions (temperature regime 20°C day/ 15°C night, with 14-h day/10-h night photoperiod). Plants were grown in pots filled with sterilized soil substrate (Klassmann TS3 601 supplemented with the fertilizer PG Mix (Hartmann Ltd., PL). For each plant/pathogen combination, there were 10 plants treated in 3 replicates and 10 plants for mock inoculation. Lower parts of stems of 28day plants were scarified by lancet (2 cm from root neck) and inoculated with 20 μL of conidia suspension of a given isolate of D. toxica (10 6 conidia per ml). After inoculation, plants were grown in at least 80% relative humidity and a temperature regime of 22°C day/19-20°C night. High humidity (above 80%) was maintained using HADAR micro-sprinklers (NaanDanJain Irrigation, Naan, Israel). The test was performed using L. luteus Juno (Polish susceptible cultivar) and four L. angustifolius accessions: Sonet (Polish cultivar formerly used for BAC library development), breeding line 258 (Australian breeding line 75A:258 carrying putative resistance gene Phr1), Unicrop (Australian susceptible cultivar), and Tanjil (Australian cultivar carrying the putative resistance gene PhtjR). Four isolates were used for inoculation, Polish DTOX1 and DTOX4, and Australian DTOXA1 and DTOXA2 isolates ( Table 1). The number of plants with visible Phomopsis stem blight symptoms was counted and expressed as percent of infected plants. Disease severity was scored 3, 7, 14, 21, and 30 days after inoculation and evaluated using the following scale: 0-resistance, no visible disease symptoms. 1-limited susceptibility, small light brown spots on the lower part of the main stem. 2-moderate susceptibility, medium-size brown lesions with isolated pycnidia. 3-high susceptibility, large brown lesions dispersed over all parts of the stem with pycnidia. 4-very high susceptibility, the whole stem covered by brown lesions with numerous pycnidia.
Statistical analysis of variance (ANOVA) was performed.

Disease resistance evaluation in controlled conditions
Three experiments in a controlled environment (greenhouse) were performed in 2007 and 2008 (temperature regime 20°C day/15°C night). Seeds were sown in pots filled with sterilized soil. A total of 10 plants × 3 repeats for each line/ pathogen combination and 10 plants for mock inoculation were assayed. The set of 26 L. angustifolius accessions analyzed in this survey contained 18 cultivars, 7 breeding lines, and one wild population (Online Resource 1). Inoculation was done using DTOX3 isolate. After inoculation, plants were grown with a relative humidity above 80% and a temperature regime of 22°C day/19-20°C night. The scoring of disease severity was performed 30 days after inoculation, using the same method as in the preliminary D. toxica resistance survey. Inoculation procedure, humidity control, and disease severity assessments were performed using the same methods as those used in the preliminary D. toxica resistance survey.

Molecular marker detection-optimization and scoring
DNA was isolated from 3-week-old leaves using the DNeasy Plant Mini Kit (Qiagen, Hilden, Germany) according to the protocol. Two biological replicates were performed. The quality and concentration of isolated DNA were evaluated by two methods: agarose gel electrophoresis followed by ethidium bromide staining and spectrophotometer measurements (NanoDrop 2000; ThermoScientific, Waltham, USA). PCR primers (Table 2) were designed for the following markers tagging resistance genes: Ph258M1 and Ph258M2 for Phr1 gene, and PhtjM7, InDel2 and InDel10 for PhtjR (Yang et al. 2015(Yang et al. , 2013a(Yang et al. , 2002. The optimization procedure involved PCR amplification using DNA isolated from reference L. angustifolius lines as a template (Tanjil, 75A:258, Unicrop) and amplicon sequencing. A range of annealing temperatures from 52 to 68°C was tested. Length polymorphisms were visualized by 1% agarose gel electrophoresis (markers Ph258M1, InDel2, and InDel10), whereas nucleotide substitution polymorphisms were detected by the cleaved amplified polymorphic sequence (CAPS) (Konieczny and Ausubel 1993) and derived CAPS (dCAPS) (Neff et al. 1998) approaches (markers Ph258M2, PhtjM4, PhtjM5, and PhtjM7). Restriction sites were identified using dCAPS Finder 2.0 (Neff et al. 2002). Restriction products were separated by 1-3% agarose gel electrophoresis, with the agarose concentration adjusted according to the size of the expected digestion products.
The screening procedure was performed using Ph258M1, Ph258M2_dCAPS, PhtjM7_dCAPS2, InDel2, and InDel10 markers. The L. angustifolius germplasm collection used for genotyping consisted of 218 lines originating from 17 countries and differing by domestication status-which ranged from wild or primitive lines (76) through mutants (5) and cross derivatives/breeding lines (65) to cultivars (74) (Online Resource 1).

Validation of markers by disease resistance assay
The validation assay was done using 49 lines. The experiment was done in controlled environment in 2017 based on the results of L. angustifolius phenotyping against Phomopsis stem blight resistance and genotyping with Ph258M1, Ph258M2, PhtjM7, InDel2, and InDel10 markers. The inoculation of plants was done using two isolates showing the highest virulence against L. angustifolius (DTOXA2 and WAC8782). Inoculation method was similar to that used in greenhouse tests performed in 2007-2008. Disease scoring was done using the scale from 1 (immune, no symptoms) to 9 (fully susceptible with all symptoms typical to Phomopsis stem blight). Markers were validated by comparing the marker genotype with resistance phenotype by binary data similarity analysis. Taking into consideration the hypothesis that D. toxica resistance is conferred by dominant alleles Yang et al. 2002Yang et al. , 2013a, heterozygote and resistant homozygote marker scores were assigned as 1 and susceptible homozygote scores as 0. The same was done for phenotype observations-resistant and moderately resistant lines were marked as 1 and susceptible were marked as 0. Simple matching (Sokal and Michener 1958) and Rogers-Tanimoto (Rogers and Tanimoto 1960) coefficients were calculated using Binary Similarity Calculator http://www. minerazzi.com/tools/similarity/binary-similarity-calculator. php. Rogers-Tanimoto is a modification of the simple matching parameter that assigns double weight to mismatching variables, therefore emphasizing false-positive and false-negative scores.

Results
Germplasm resources resistant to D. toxica are preserved in the European L. angustifolius gene bank The disease resistance survey under controlled conditions revealed that most of the analyzed L. angustifolius germplasm accessions were susceptible to D. toxica during the latent phase of the infection (Table 3, Online Resource 2). Susceptibility was demonstrated by the extensive colonization of stem tissues and formation of pycnidia. Susceptible lines were characterized both by high average disease severity scores and a high percentage of plants with visible disease symptoms. However, Australian cultivar Tanjil and German cultivar Arabella as well as Polish (W-226 and WTD-1406) and Australian (83A:476) breeding lines showed high levels of resistance to the pathogen, manifested by the lack of developed disease symptoms. Australian cultivar Myaille as well as Polish cultivar Bojar and German cultivar Boruta exhibited moderate resistance, with considerably delayed colonization of stem tissues and the development of only small light brown spots, limited to the lower part of the main stems.
The field disease resistance survey included several L. angustifolius cultivars previously characterized in the controlled environment experiment as susceptible (Sonet, Karo, Zeus, Mirela, Graf, Elf, Kalif, Baron), moderately resistant (Bojar, Boruta), or resistant (Arabella), as well as seven L. luteus cultivars not yet evaluated for D. toxica susceptibility. Breeding lines were not assayed due to seed availability constraints. The results obtained for L. angustifolius-resistant and moderately resistant lines were consistent with those from the controlled environment experiment; however, some differences in the percentage of colonized plants or disease severity symptoms were observed, for example Bojar turned out to be less susceptible than Boruta. Moreover, significant differences in developed disease symptoms between field and controlled environment assessments were observed for L. angustifolius susceptible lines Kalif, Elf, and Graf, which had very low (Graf) or low (Elf and Kalif) levels of stem tissue colonization by D. toxica. All analyzed L. luteus cultivars exhibited high susceptibility to D. toxica (Table 3, Online Resource 2).
Markers tagging D. toxica resistance genes are relevant to PCR-based genotyping PCR products of markers Ph258M1, Ph258M2, PhtjM4, PhtjM5, PhtjM7, InDel2, and InDel10 (Yang et al. 2015(Yang et al. , 2013a(Yang et al. , 2002 were amplified using DNA isolated from reference lines (resistant and susceptible), sequenced, and compared with the target sequences. Amplification products with appropriate sequences were obtained for all analyzed markers. Markers Ph258M1, InDel2, and InDel10 were not sequenced as they were based on length differences of the amplified PCR products which could be directly visualized by simple agarose gel electrophoresis. Optimization was performed for certain markers: Ph258M2-anchored in a simple sequence repeat (SSR) locus; as well as PhtjM4, PhtjM5, and PhtjM7-tagging single nucleotide polymorphisms (SNPs). Ph258M2, PhtjM4, and PhtjM7 markers were converted to dCAPS markers, whereas PhtjM5 was converted to a CAPS marker (Table 2). Genotyping attempts with the use of reference lines and selected wild and domesticated accessions r e v e a l e d t h a t m a r k e r s P h t j M 4 _ d C A P S 1 , PhtjM4_dCAPS2, PhtjM5_CAPS, and PhtjM7_dCAPS1 had a very low rate of reproducibility resulting from weak  (Yang et al. 2015(Yang et al. , 2013a(Yang et al. , 2002 PCR amplification and/or very low enzyme cleavage efficiency. On average, 4-5 replicates per sample had to be performed to obtain single score. Therefore, these four markers were discarded from further analysis. Marker sequences were aligned to the most recent genome assembly (Zhou et al. 2018). Markers PhtjM7, InDel2, and InDel10 and are localized in the same region of the chromosome CP023117.1 (LG-05) and are separated by 108,610 bp and 513 bp, respectively. Markers Ph258M1 and Ph258M2 are flanking the sequence of 269,913 bp in the chromosome CP023120.1 (LG-08). Alignment details are provided in Online Resource 3.
At least two genetic sources of D. toxica resistance exist in the European L. angustifolius core collection The European Lupin GenBank collection of 218 L. angustifolius accessions was screened with two markers tagging the Phr1 gene (Ph258M1 and Ph258M2_dCAPS) and three markers linked to PhtjR gene (PhtjM7_dCAPS2, InDel2, and InDel10). Amplification products of expected sizes were obtained for 99.6% of samples. Additional alleles were very rare and occurred only in two lines: AN-80154a (marker Ph258M1) and W-226 (InDel2). These alleles were encoded as BR^homozygotes for marker validation procedure. Amplification consistently failed for Yorrel (PhtjM7_dCAPS2) and Badajoz-1 (InDel2). No single accession carrying R bands for all Phr1 and PhtjR gene markers has been identified in the core collection. Therefore, it might be assumed that the analyzed germplasm array did not carry any line containing both D. toxica resistance genes in combination.
To enable comparison of marker genotypes with D. toxica resistance phenotype, a phytopathological assay in controlled environment was performed, involving 49 lines showing diverse combinations of marker scores. From 17 lines carrying both homozygous R alleles of Ph258M1 and Ph258M2 markers, only one (75A:258) was resistant in this experiment. From 13 lines analyzed, having at least two homozygous R alleles for PhtjM7, InDel2, and InDel10 markers, the resistance (or moderate resistance) was confirmed only for four accessions (Population B-551/79, Population 22695, Wonga and Tanjil). Moderate resistance was revealed also for Myallie and Boruta. These two lines carry S homozygous alleles for all tested markers but showed moderate resistance in one or two previous tests, respectively. On the contrary, the resistance of WTD-1406, W-226, and 83A:476, inferred from the initial controlled environment test, was not confirmed in this marker validation assay. The same phenomenon was observed for Arabella and Bojar which were revealed to be resistant in both previous tests. It should be noted that this test was highly selective, because some Australian lines reported in other studies Yang et al. 2002) as moderately resistant (Yorrel, Gunguru and Merrit) were revealed to be susceptible with an average disease scores from 3.6 to 3.8. Yet comparatively, reference lines 75A:258 and Wonga, carrying Phr1 and PhtjR genes, were classified as resistant with scores 1.7 and 1.8, respectively.
The set of PhtjM7, InDel2, and InDel10 markers is applicable for molecular selection of PhtjR gene Marker genotype scores were compared with the results of this resistance survey as well as those published elsewhere. None of the markers displayed 100% consistency with Phr1 and PhtjR genotypes inferred from phenotypic observations or published pedigree relationships among certain narrow-leafed lupin cultivars (Cowling 1999;Shankar et al. 2002;Stefanova and Buirchell 2010;Yang et al. 2002Yang et al. , 2015. Markers Ph258M1 and Ph258M2 generated higher number of false-positive results than markers PhtjM7, InDel2, and InDel10, namely 21 and 30 vs 6, 9, and 8. Low values of simple matching (0.48 and 0.38) and Rogers-Tanimoto (0.32 and 0.21) coefficients highlighted the negligible applicability of markers Ph258M1 and Ph258M2 for molecular breeding. Markers PhtjM7, InDel2, and InDel10 revealed to have higher values of both parameters, i.e., 0.63-0.68 for Rogers-Tanimoto and 0.77-0.81 for simple matching. All these three markers used together constitute relatively versatile selection tool, applicable for wide range of crosses (4% falsepositive and 0% false-negative scores) ( Table 4). The correlation between marker scores and phenotype observations enabled us to conclude that resistance of Population B-551/79 and Population 22695 lines is putatively conferred by the PhtjR gene (see Online Resource 4).

D. toxica resistance genes in lupins
Resistance to Phomopsis stem blight in the narrow-leafed lupin is a complex trait and at least three genes are involved, originating from different germplasm resources, namely 75A:258, Merrit, and Tanjil. The highly resistant line 75A:258 was selected from a cross between cv. Marri and a wild line collected in Morocco (P22872) . Moderately resistant cultivar Merrit was derived from a cross between cv. Illyarrie and a wild line from Spain (P22750) (Gladstones 1992). Resistant cultivar Wonga was derived from a cross between Gungurru and 75A54-5-8 (Stefanova and Buirchell 2010). The studies based on F 1 -F 3 generations of crosses Unicrop × 75A:258 and Merrit × Unicrop revealed that both accessions carry independently segregating resistance alleles, named Phr1 (75A:258) and Phr2 (Merrit) . Phr1 was revealed to be fully dominant, whereas Phr2 appeared to be incompletely dominant. Phenotyping studies based on crosses between Tanjil and other cultivars, including the susceptible Unicrop, revealed that Tanjil and Wonga possess a single-dominant D. toxica resistance gene which is different from Phr1 and Phr2, this gene was named PhtjR (Yang et al. 2013a). Genetic analysis of L. angustifolius material from European breeding programs (Belarus, Russia, and Poland) with reference Australian lines Merrit and Gungurru also showed that resistance to D. toxica is determined by a single dominant gene, which was named Rpl1 (Kuptsov et al. 2006). Several single dominant genes also control resistance against anthracnose in the narrow-leafed lupin germplasm, namely Lanr1 in cv. Tanjil, AnMan in cv. Mandelup, and LanrBo in line Bo7212 (Fischer et al. 2015;Yang et al. 2004Yang et al. , 2008. In contrast, D. toxica resistance in another Old World lupin crop, white lupin (Lupinus albus L.), is under polygenic control as revealed by quantitative trait loci mapping in recombinant inbred line population derived from a cross between the susceptible Ukrainian cultivar Kiev Mutant and the resistant Ethiopian primitive P27174 accession (Cowley et al. 2014;Vipin et al. 2013).

Tools for marker-assisted selection
At the time of their development, markers Ph258M1 and Ph258M2 were considered as useful for marker-assisted selection, with, however, a limited range of target crosses due to presence of false-positive scores in domesticated Australian germplasm, including cv. Merrit. This phenomenon resulted from the large distance between these markers and the Phr1 gene, estimated to be 7.8 and 5.7 cM, respectively (Yang et al. 2002). Our study supported this observation providing evidence for both the occurrence of false-positive scores (accounting for 40-58% of analyzed lines) and the high rate of recombination between the markers (30% analyzed of lines, of which 63% were wild and 37% domesticated), see Online Resource 4. Better markers have not yet been developed because both existing L. angustifolius mapping populations are monomorphic for Phr1 gene and could not be exploited to solve this issue (Boersma et al. 2005;Yang et al. 2013b).
Harnessing of next-generation sequencing techniques based on whole-genome resequencing of reference resistant and susceptible lines may help to overcome this barrier in the near future and facilitate generation of high-quality, cost-effective markers, as it was the case for L. angustifolius anthracnose resistance (Yang et al. 2015). Markers InDel2 and InDel10 originated from scaffold 84773 carrying the putative PhtjR resistance gene and were considered to be truly diagnostic since the marker genotypes were consistent with Phomopsis stem blight resistance phenotypes on all Australian cultivars analyzed (Yang et al. 2015). Our study showed that they are also diagnostic on a wide range of accessions from the European germplasm collection, including cultivars and breeding lines. We identified only three incidences of recombination events between InDel2 and InDel10 markers, which all occurred only in wild populations (Online Resource 4). Additionally, three heterozygote scores were found for InDel10, also only in wild populations. Marker PhtjM7, located roughly 1 cM from the target PhtjR gene, was originally recommended as applicable for markerassisted selection in narrow-leafed lupin breeding due to the lack of recombination between the marker and the gene in the set of 26 Australian cultivars; expected genotyping accuracy was estimated to be approximately 99% (Yang et al. 2013a). However, we identified the presence of recombination between PhtjM7 and InDel2 or InDel10 markers in as many as 30 lines, including 22 wild and 8 domesticated lines. Nevertheless, this marker had slightly higher genotype to phenotype similarity coefficients than InDel2 or InDel10 in the validation assay. All three markers used together provided9 5% confidence on selection of the desirable PhtjR allele.
The most reliable markers for genotype selection are those anchored in functional mutations of genes conferring particular traits, as was established for L. angustifolius vernalization responsiveness locus Ku and corresponding large deletion in the promoter region of Flowering locus T homolog, LanFTc1 gene . Nevertheless, deciphering the molecular background underlying the Ku locus took more than decade of extensive research involving various techniques, from BAC library screening via DNA hybridization, restriction siteassociated physical and linkage mapping, fluorescent in situ hybridization of DNA probes in metaphase chromosomes, Sanger, 454 and Hi-seq sequencing to gene expression profiling Nelson et al. 2017Nelson et al. , 2006. Although a similar BAC library approach was applied to L. angustifolius Phomopsis stem blight resistance markers PhtjM2 (linked with PhtjR) and Ph258M2 (Phr1), it did not yield resistance gene identification or the development of tightly linked diagnostic markers (Książkiewicz et al. 2013. Those studies showed convincingly that MFLP-derived probes are not suitable for positional cloning of particular genes because they hybridize to numerous loci dispersed in the genome, localized both in repetitive and gene-rich regions. The falling price of sequencing and the increase in the size of sequence databases has reduced the cost of obtaining useful sequence information for analysis (Muir et al. 2016). Accelerating progress in next-generation sequencing technology and annotation has considerably shortened the time and money required for identifying genes controlling agronomic traits, as was demonstrated for LpPg1 stem rust resistance locus in perennial ryegrass (Lolium perenne) where the NBS-LRR gene was revealed through massive analysis of cDNA ends (MACE) (Bojahr et al. 2016). The MACE approach has also been implemented in L. angustifolius applied research, aimed at providing ready-to-use technology for gene-based monitoring of key agronomic traits including, among others, resistance to pathogenic fungi (Książkiewicz et al. unpublished). Both identified resistant wild lines (Population B-551/79 and Population 22695) originate from Spain. The geographical localization of these putative resistance donors in warm Mediterranean climates coincides with the ecological niche of the pathogen. Thus, in controlled conditions, the maximum infection efficiency was observed after a dew period of 48-72 h with temperatures within a range of 15-25°C with 20°C being the optimum . In Australian field conditions, infections and subsequent lupinosis appearance were determined to be correlated with rainfall patterns (Cowling and Wood 1989;Petterson and Wood 1986). Phomopsis stem blight has been considered by European lupin breeders as an unimportant disease, constituting a threat to white lupin only in wet years (Święcicki and Święcicki 1995). However, the advance of global warming may expand the D. toxica climatic optimum in to the north, reaching major European areas of lupin cultivation and forcing adaptation of breeding strategies to develop crop cultivars adapted to new threats. Unfortunately, much of the natural genetic diversity in narrow-leafed lupin has been lost during the domestication process (Berger et al. 2012). Incorporation of alleles from wild germplasm to widen the genetic diversity of the domesticated pool is currently emerging as an important need of the narrowleafed lupin breeding community (Berger et al. 2013). This process may be facilitated by using PhtjR molecular markers validated in this study for European lupin germplasm.

Conclusions
1. The European lupin germplasm collection preserves wild and domesticated donors of at least two Phomopsis stem blight resistance genes, Phr1 and PhtjR. 2. The set of molecular markers PhtjM7, InDel2, and InDel10, tagging the PhtjR gene, can be used in markerassisted selection targeting the European gene pool with an expected confidence about 95%. 3. By now, no reliable diagnostic marker for the Phr1 applicable for European breeding programs has been found; the two existing markers showed high percentage of false-positive results and high recombination rates between markers.