QTL Mapping of Seedling and Adult Plant Resistance to Septoria Tritici Blotch in Winter Wheat cv. Mandub ( Triticum aestivum L.)

: Septoria tritici blotch (STB) is one of the most devastating foliar diseases of wheat worldwide. Host resistance is the most economical and safest method of controlling the disease, and information on resistance loci is crucial for effective breeding for resistance programs. In this study we used a mapping population consisting of 126 doubled-haploid lines developed from a cross between the resistant cultivar Mandub and the susceptible cultivar Begra. Three monopycnidiospore isolates of Z. tritici with diverse pathogenicity were used to test the mapping population and parents’ STB resistance at the seedling stage (under a controlled environment) and adult plant stage (polytunnel). For both types of environments, the percentage leaf area covered by necrosis (NEC) and pycnidia (PYC) was determined. A linkage map comprising 5899 DArTSNP and silicoDArT markers was used for the quantitative trait loci (QTL) analysis. The analysis showed five resistance loci on chromosomes 1B, 2B and 5B, four of which were derived from cv. Mandub. The location of QTL detected in our study on chromosomes 1B and 5B may suggest a possible identity or close linkage with Stb2 / Stb11 / StbWW and Stb1 loci, respectively. QStb.ihar-2B.4 and QStb.ihar-2B.5 detected on chromosome 2B do not co-localize with any known Stb genes. QStb.ihar-2B.4 seems to be a new resistance locus with a moderate effect (explaining 29.3% of NEC and 31.4% of PYC), conferring resistance at the seedling stage. The phenotypic variance explained by QTL detected in cv. Mandub ranged from 11.9% to 70.0%, thus proving that it is a good STB resistance source and can potentially be utilized in breeding programs.


Introduction
Septoria tritici blotch (STB), a fungal disease caused by the ascomycete fungus Zymoseptoria tritici (Desm.) (teleomorph Mycosphaerella graminicola, syn. Septoria tritici), is one of the most devastating foliar diseases of wheat, which accounts for approximately 70% of the annual usage of cereal fungicide in Europe [1,2]. Wind-dispersible ascospores on wheat debris are the most significant source of primary inoculum and contribute greatly to year-to-year disease transmission, but the asexual pycnidiospores promote polycyclic host infection and spread of the disease over a growing season [3]. Pycnidiospores are spread by rain splash, which results in STB epidemics being more severe in regions characterized by frequent rainfall and high humidity [4]. Particularly severe yield losses can occur in crops when the top leaves become infected, as flag leaf photosynthesis in wheat contributes about 30%-50% of the assimilates for grain filling [5,6].
Wheat is considered to be the most important crop in the world according to area of harvest [7]. As the risk of an epidemic can be reduced by good agricultural practices, seeking new sources of resistance to STB should be a priority, especially given the reports concerning the possible rapid development of fungicide resistance in Z. tritici populations. Lucas et al. [8] suggest that a mutation allele associated with resistance to fungicides could almost completely replace the wild type within a single season, as described for resistance to methyl benzimidazole carbamates (MBC) fungicides in the years 1984-1985. Similarly, resistance to quinone outside inhibitors (QoI) fungicides was detected in northwestern Europe in the early 2000s, and it is now widespread across the UK and northern regions of France and Germany [9,10]. A continuous decline in efficacy has also been observed for azoles, currently the most common fungicide class used in spray programs. In contrast to MBC and QoI, shifts in sensitivity to this particular fungicide class in Z. tritici populations have been gradual, but have undoubtedly accelerated over the past years [11][12][13]. Taking all that into account, breeders should not be reliant on fungicides as the only method of crop protection, but should rather seek support in the form of host varietal resistance. Therefore, resistance breeding seems to be of great importance.
In general, resistance to STB can be qualitative or quantitative. Qualitative resistance is usually an isolate-specific and near-complete resistance [14,15]. It is most likely controlled by major genes with a large effect, following the gene-for-gene concept, as was demonstrated by Brading et al. [16] for the resistance gene Stb6. To date, at least 22 major genes associated with resistance to Z. tritici have been identified in wheat [17,18]. Many of these loci contribute to STB resistance independently of the plant growth stage [14][15][16][19][20][21][22], although this type of resistance can also be effective only in seedlings [23] or only in adult plants [22]. Recent cloning of Stb6 and Stb16q showed that those genes encode a wall-associated receptor kinase (WAK)-like protein and a plasma membrane cysteine-rich receptor-like kinase, respectively [24,25]. The inheritance of qualitative resistance has been reported to be dominant, partially dominant or recessive [14,16,19,26,27].
On the other hand, quantitative, polygenic resistance, although partial and incomplete, is considered to be more durable [28]. That kind of resistance can be expressed by the length of incubation and latent period, development rate of sporulating area, maximal sporulating area, pycnidial density and sporulation capacity [29]. Most known resistance loci have been identified by QTL analysis of biparental populations [17,[30][31][32][33]. Nevertheless, in recent years, association mapping has significantly expanded the knowledge of new regions related to STB resistance and their occurrence in wheat varieties [34][35][36][37][38][39][40][41]. Many quantitative trait loci (QTL) and marker-trait associations (MTAs) of resistance against STB have been identified in wheat over the years, at least one on every chromosome, although 3BL, 6BS and 7DL arms seem to be especially involved in quantitative resistance, as shown by the number of loci identified [17]. QTL for STB resistance so far have been detected for total necrotic leaf area [33,42,43], necrotic area bearing pycnidia [44,45] or both [46][47][48], and for chlorosis [32], the disease progress has been estimated as the area under the disease progress curve (AUDPC) [42,49] and latent period [46].
The aim of the study was to detect genetic loci contributing to STB resistance in the German winter wheat variety Mandub at the seedling and adult plant stages using QTL analysis.

Plant Material and Fungal Isolates
In the preliminary research on STB resistance in winter wheat, the German cultivar Mandub revealed a high level of resistance at the seedling and adult plant stages (Czembor P., unpublished data). This cultivar was crossed with the susceptible cultivar Begra [48], and from anther cultures of the F1 generation a set of 126 winter wheat doubled-haploid (DH) lines was produced.
For both the seedling and adult plant pathology tests, we chose three monopycnidiospore isolates of Z. tritici of diverse pathogenicity described previously by Czembor et al. [50]: IPO86036, IPO88004 and IPO92006. The isolates were grown in the dark at a constant 20 °C temperature on a YMA medium containing 4 g of yeast, 4 g of maltose, 4 g of sucrose and 30 g of agar per 1 L of distilled water [4]. After 2 to 3 days the spores were collected and stored at −80 °C until the pathology tests began. Before the inoculation, the spore suspension was prepared by adjusting the concentration to 10-15 × 10 6 spores/mL and adding a few drops of a surfactant (TWEEN 20, Sigma-Aldrich, Poznań, Poland).

Phytopathological Tests
The population's reaction to inoculation with each Z. tritici isolate was tested in one season at both the seedling and adult plant stages in the years 2011-2012 and 2018 (Table  1). For the seedling pathology tests, the seeds both from DH lines and their parent cultivars were pre-germinated and sown in eight 160 (16 × 10)-well horticultural plastic trays. Each tray contained 160 seeds with 5 seeds per line. To avoid border effects, the left and right border cells were sown with Begra seeds. The plants were grown in a growth chamber (type HBZ 1022 M, Hereaus Votsch-Nema GmbH, Netzschkau, Germany) under 19 °C/15 °C (16-h day/8-h night) conditions until the second leaves fully emerged. Then each tray was sprayed with 100 mL of aqueous Z. tritici pycnidiospore suspension and (as described by Brading et al. [16]) was subsequently put under polyethylene incubation tents lined with wet filter paper to maintain the humidity at approximately 90%. Following the inoculation, the seedlings were held in the dark at a constant temperature of 22 °C. After 48 h, however, the 16-hour photoperiod was restored. During the test, the trays were moved around the tent to avoid any potential position effect, and newly emerged leaves were trimmed for better light penetration every seven days. In the polytunnel experiment the seeds were sown in 1-meter-long rows spaced at 18 cm in three randomized blocks, yet only two of them were inoculated, while the third one served as a control, mainly to monitor any outside source of inoculum other than that used in the experiments. Adult plants with fully expanded flag leaves were inoculated twice with an interval of two days. The inoculation took place in the evening to promote infection by overnight moisture retention on the leaf surface. As in the seedling tests, the plants were inoculated by spraying with aqueous pycnidiospore suspension (100 mL/1 m 2 ). To maintain relatively high humidity during the tests, the polytunnels were equipped with a sprinkler irrigation system programmed to turn on for 20 min three times a day.
The assessment of the disease development both in the seedling as well as polytunnel experiments took place when approximately 80% of Begra's second/flag leaves became necrotic (generally 21 days after inoculation). For the necrotic area estimation, 5-8 leaves were collected per line replicate, placed on self-adhesive foil and photographed. Then, to determine the area of lesions bearing pycnidia, the same leaves were marked manually using a red marker under a magnifying glass (×3) and photographed again. The images were then analyzed with WinCam software (Regent Instruments, Inc. 2004), which allowed us to precisely measure the disease parameters, i.e., the percentage of leaf area covered both with necrosis (NEC) and pycnidia (PYC). In the polytunnel experiments, the heading dates and heights of plants were also measured.

Statistical Analyses
To avoid any inconsistencies in the normality of data distribution, NEC and PYC data were transformed by logit transformation [33,41,48]. According to the central limit theorem, the height and heading date were assumed to be distributed normally and, therefore, were not transformed. The analysis of variance (ANOVA) was performed with XLSTAT software (Addinsoft, version 2016.02.28540) and broad-sense heritability (h 2 ) was calculated for both disease parameters as well as height and heading date [51]. Furthermore, to estimate the relationship between necrosis/pycnidia coverage, the plant height and the heading date (calculated from 1 January), Pearson's correlation coefficient was calculated.

Linkage and QTL Analyses
For linkage analysis, DNA both from 126 DH lines as well as their parental cultivars was extracted using the CTAB extraction method [52]. DNA samples were then subjected to genotypic analysis using the DArTseq platform by Diversity Arrays Technology, Pty. Ltd., Australia. Obtained data were used for genetic map construction also provided by DArT P/L. QTL analyses were performed using MapQTL 6.0 [53]. Eighteen data sets were used to represent different isolates, analyzed traits and experimental conditions (Table 1). For adult plant tests, the corresponding heading date and height data were selected as covariates [54]. To calculate the likelihood of odds (LOD) threshold value, 1000 permutation tests were done at the 0.05 significance level. Cofactors for multiple-QTL mapping (MQM) were selected automatically, although to initiate the program, at least one for each data set had to be chosen manually. To determine which one, interval mapping (IM) was carried out and the one with the highest LOD value was used. Genetic maps of chromosomes and the detected QTL were drafted using the MapChart 2.2 software [55]. The confidence intervals for the QTL effects were established using the one-LOD rule [56,57]; these intervals are indicated in the figures as boxes. To increase the likelihood that the interval would contain the QTL, a support interval calculated as a two-LOD rule is indicated by the lines shown in the figures.
To compare the location of QTL identified in this analysis with the positions of known STB resistance loci (Brown et al. 2015), markers associated with these loci were assigned a physical location in the wheat reference genome IWGSC RefSeq v1.0 in the EnsemblPlants database with the BLASTN algorithm [58]. The same methodology was applied for candidate gene prediction and to compare the position of the detected QTL with known reduced height (Rht), vernalization (Vrn) and photoperiod sensitivity (Ppd) genes. Furthermore, physical maps of chromosomes 1B, 2B and 5B were generated to determine possible colocalizations between QTL detected in this study and known STB resistance loci. The maps were constructed for physical locations of the silicoDArT and DArTSNP markers used for mapping.
The physical locations of detected QTL were searched for overlapping genes and proteins encoded using the BioMart tool [59] and InterPro database [60].

Analysis of Disease Parameters
The phytopathological tests resulted in 18 data sets that included the percentage of the leaf area covered with necrosis and pycnidia as well as the heading dates and height data from the polytunnel tests ( Table 2). The data associated with the disease parameters NEC and PYC showed a wide range of variation, as indicated by the coefficient of variation that ranged from 30.84% to 120.06%. The population demonstrated a relatively continuous distribution and the parent lines, Mandub and Begra, were always on the opposite sides of the scale, although they did not always show extremely high or low values for measured traits ( Figure 1). Consequently, in most data sets there are some DH lines that showed much higher or much lower phenotypic values than their parental lines, indicating transgressive segregation. The most noteworthy examples of this phenomenon were observed for the susceptible parental line Begra: T_IPO92006_PYC (24.63% of the leaf area covered by necrotic lesions bearing pycnidia when the population maximum was 74.59%), T_IPO88004_PYC (21.02%, max. 67.10%) and F_IPO88004_PYC (16.21%, max. 70.99%). On the other hand, there was only one data set where the resistant parent Mandub showed a rather high percentage of leaf area covered with necrosis in comparison to the DH progeny (T_IPO86036_NEC data set: Mandub-36.98%, minimum-3.29%) ( Table 2).

Heritabilities and Correlations
Broad-sense heritability (h 2 ), which was in the range 0.65-0.95 for disease parameters, 0.84-0.94 for heading date and 0.90-0.92 for plant height, proved that the variation observed in the DH population is indeed due to genetic factors, and therefore, the obtained phenotypic data are suitable for QTL analysis. A positive and rather strong correlation between both disease parameters was observed across all six experiments, ranging from 0.44 to 0.90. Pearson's correlation coefficient for the relationship between disease parameters and disease escape traits (plant height and heading date) was also calculated, but it was mostly weak or statistically insignificant (Table 3).

Genetic Map and QTL Analysis
As a result of DArTseq analysis, we obtained 16,854 codominant DArTSNP markers and 17,689 dominant silicoDArT markers. The linkage map for the Mandub × Begra population, provided by DArT P/L, consisted of 5899 markers of both kinds. The markers were assigned to 25 linkage groups, resulting in a map of 2666 cM total length with the shortest group being 2A_1 (5.06 cM) and the longest being 7A (192.23 cM). The average marker saturation was 236 markers per linkage group. The smallest groups (1D and 4B_1) comprised 22 markers, and the biggest (3B) comprised 631.
The QTL analysis was carried out for 18 data sets including percentage of leaf area covered by necrosis or pycnidia, height and heading date. It resulted in detection of 12 QTL associated with resistance to STB and 11 associated with plant height or heading date. Identified disease resistance QTL were located on chromosomes 1B, 2B and 5B. Co-localization of the four QTL detected on chromosome 1B indicates the presence of a single QTL (QStb.ihar-1B.2) (Table 4, Figure 2) responsible for the reaction to isolate IPO92006. The phenotypic variation explained by individual QTL was 66   None of the detected QTL were associated with resistance to STB were overlapping with QTL detected for disease escape traits (plant height or heading date). Analysis of the remaining six data sets revealed two QTL associated with plant height on chromosomes 2D and 4B (QHt.ihar-2D and QHt.ihar-4B) and two associated with heading date on chromosomes 2D and 7B (QHd.ihar-2D.2 and QHd.ihar-7B) (Table 4, Figure 2). The phenotypic variation explained by individual QTL for the plant height ranged from 17.5% to 19.6% on chromosome 2D and 27.0% to 27.2% on chromosome 4B. Individual QTL detected for heading date explained 17.3%-23.8% of the variation on chromosome 2D and 9.2%-11.2% on chromosome 7B (Table 4). QHd.ihar-7B, QHt.ihar-2D and QHd.ihar-2D.2 alleles originated from the susceptible parent Begra. QHt.ihar-4B was derived from the resistant parent Mandub.

Prediction of Candidate Genes
Identified resistance QTL spanned from 0.2 to 251 Mbp in the wheat reference genome. Analysis revealed 12 genes overlapping with those loci. For nine of them, the encoded protein domains were identified.

Discussion
QTL analysis of the Mandub × Begra mapping population revealed five QTL conferring resistance to STB. Generated maps had a satisfactory resolution, but unfortunately, the order of the markers in the genetic map used in this study was not always reflected by the physical map. Those results may be due to the limited population size or resolution of mapping. Nevertheless, differences between parental genomes and the reference genome may have generated those markers' inconsistency. Tremendous genetic diversity was observed within different genotypes of certain species. A recent analysis of the wheat pan-genome obtained from 18 cultivars revealed substantial rearrangements, present-absent variations of the genes and repeated elements. In general, the full sequencing of 18 wheat genotypes resulted in a 3.3% increase of the wheat reference genome size [61]. Similar results were obtained from sequencing of japonica and indica rice accessions, which showed increases of 4% and 6%, respectively, in genome size [62]. Structural reengagements were identified within Chinese Spring and Hope (Sr2) wheat cultivars in a ca. 900 kbp segment spanning the Sr2 resistance locus [63]. Comparative analysis of two 2D chromosomes obtained from two wheat genotypes revealed InDels hundreds of kbp long, which cover 0.3% of the chromosome [64]. Therefore, the assumptions of co-localization need to be taken with caution. Supposed positions of both QTL detected in this study as well as known resistance loci are marked on generated physical maps, which can be found in Supplementary Figures S1 (chromosome 1B), S2 (chromosome 2B) and S3 (chromosome 5B).

Detected QTL and Known Resistance Loci
QStb.ihar-1B.2 was detected on chromosome 1B (Figure 2, Figure S1). On the same chromosome, three major STB resistance genes were identified previously: Stb2, Stb11 and StbWW. However, it has been suggested that these genes are either identical, allelic or at least closely linked [22,65,66]. By assigning physical locations to the markers used for mapping those resistance loci, it was found that QStb.ihar-1B.2 co-localizes with those resistance genes ( Figure S1). It spans 1.3 Mbp, from 48.2 to 49.6 Mbp, while Stb2 and Stb11 span 38.6-42.3 Mbp and StbWW spans 38.6-212.5 Mbp. That and the fact that QStb.ihar-1B.2 explained such a large part of the observed variation suggests that it might be the same resistance locus as the major resistance genes mapped on chromosome 1B previously.
Several quantitative resistance loci were detected on this chromosome as well [32,33,43,44,[46][47][48]. QStb.ihar-1B (localized at 229.5-258.8 Mbp) identified by Radecka-Janusik et al. [48] in cv. Liwilla expressed resistance to the same single Z. tritici isolate as cv. Mandub, i.e., IPO92006, and such resistance was detected in both the seedling and adult plant tests in both studies. Despite slightly different localizations ( Figure S1) it seems that cv. Liwilla and cv. Mandub may carry the same STB resistance locus on chromosome 1B. Discrepancy in QTL localizations could occur due to different sizes of the mapping populations used in this study and the study by Radecka-Janusik et al. [48]-the Liwilla × Begra population used to map QStb.ihar-1B comprised only 74 lines, which could have had a negative impact on the mapping precision.
A few more resistance loci were mapped in the same region as QStb.ihar-1B.2. 1BS detected by Tabib Ghaffary et al. [46] explained nearly 70% of the variation in the seedling stage. QStb.teagasc-1B.1 [33], MQTL2 [43], QStb.lsa_fb-1B [44] and QStb.1B.b [47], unlike QStb.ihar-1B.2 and 1BS, had only minor to moderate effects on the phenotypic variation Therefore, considering the rather large portion of the variation explained by the QTL detected on chromosome 1B in this study and the physical locations of the resistance loci, it is most likely that cv. Mandub carries the Stb2/Stb11/StbWW gene, which may have spread across global wheat breeding because of the movement of elite breeding lines from CIMMYT [17].
One QTL for resistance to STB was identified on chromosome 5B, QStb.ihar-5B ( Figure 2, Figure S3). To date, one major resistance gene has been mapped on chromosome 5B, i.e., Stb1, which was mapped in cv. Bulgaria 88 by Adhikari et al. [20]. It is located at 418.8-528.3 Mbp; therefore, both loci may be linked, but it is not likely that they are the same locus, since QStb.ihar-5B explained only 11.9% of the phenotypic variation. A number of QTL have been mapped to this region as well ( Figure S3). 5BL [67] and QTL10 [43] may be located at the same locus as QStb.ihar-5B.
QTL identified in this study were co-localized with QTL previously identified and published. Nevertheless, due to various mapping populations, different marker types and distinct Z. tritici isolates used in STB resistance studies over the years, it is difficult to clearly define the identity of co-localizing loci. Since the value of the explained phenotypic variation strongly depends on the genetic background, any similarities or discrepancies within them should not be treated as a reliable proof of distinction between known resistance loci and those described in this study. However, high levels of the variation explained by a QTL might be an indicator of a major resistance gene.

Prediction of Candidate Genes
Among the genes annotated on identified QTL, there were regions coding proteins or protein domains with previously described function in the resistance process. The plant LTP, non-specific lipid-transfer protein, which was identified within QStb.ihar-1B.2, plays a role in extracellular lipid apposition and protective layer formation, like cutin and suberin [68]. Alpha-amylase inhibitors (bifunctional inhibitors) have been shown to have antifungal effects due to nutrient supply suppression and mycelium growth inhibition [69,70]. QStb.ihar-2B. 3 and QStb.ihar-5B loci spanned over genes engaged in glycosyl groups' manipulation. Glycosylation is a widespread modification of molecules that leads to alteration of protein properties, activity or target location. Glycosylation of metabolites and hormones occurs during biotic and abiotic stress responses. Glycosyltransferases are involved in conversion of the DON Fusarium graminearum toxin into non-toxic DON-3-glucoside in wheat [71]. UDP-glycosyltransferases are involved in the aluminum stress response in flax [72] and the defense process against Sclerotinia sclerotiorum and Botrytis cinerea in rapeseed (Brassica napus L.) [73]. The kinase activity, which domains indicated within QStb.ihar-2B.4 (jointly with Ca2+ capturing NAF/FISL motif) and QStb.ihar-5B (with leucine-rich repeat motif) showed, was distinguished within nine classes of R genes' action [74], and the leucine-rich repeat domain plays a crucial role in pathogen recognition. This overview of QTL overlapping genes suggests a list of promising candidate genes for further investigations.

QTLs for Disease-Escape Traits
Plant height and heading date are considered to be disease-escape mechanisms dependent on the probability of contact between the pathogen and the host [75].
QTL analysis revealed two QTL associated with plant height: QHt.ihar-2D on chromosome 2D and QHt.ihar-4B on 4B. They did not co-localize with any QTL for resistance to STB detected in this study. However, QHt.ihar-2D was mapped in close proximity to the region of a reduced height gene, Rht8 [76]. QHt.ihar-4B, on the other hand, was mapped near the semi-dwarfing gene Rht-B1 [77], Therefore, it is possible that cv. Mandub and cv. Begra carry those genes and it is their expression that we observed.
Two QTL for heading date QHd.ihar-2D. 2 and QHd.ihar-7B were also detected in this study. Chromosome 2D is known to carry a photoperiod response gene-Ppd-D1 (34.0 Mbp)-which is considered to be the key locus determining the photoperiodic sensitivity of hexaploid wheat [78]. However, the physical location of QHd.ihar-2D.2 suggests that those are different loci, as the QTL is located at 629.3-642.4 Mbp. QHd.iahr-7B (6.8-10.9 Mbp) on chromosome 7B was mapped in close proximity to the region carrying the vernalization gene Vrn-B3 (12.0 Mbp) and the photoperiod response gene Ppd-B2 (26.8 Mbp).
Pearson's correlation coefficients for the relationship between disease parameters and disease-escape traits measured in this study turned out to be weak or statistically insignificant. It is most likely because the disease-escape mechanism is considered to be dependent on the probability of contact between the pathogen and the host, and the experimental conditions (inoculation technique, humidity) are set to ensure that contact and, therefore, to produce an appropriate resistance response across the whole population studied.

Conclusions
In conclusion, our study revealed four QTL for resistance to STB in cv. Mandub. Those QTL confer resistance to all three different Z. tritici isolates tested. QStb.ihar-2B.4 is a potentially new resistance locus that was identified on chromosome 2B, conferring resistance to STB at the seedling stage and explaining a moderate portion of the phenotypic variation (29.3% for NEC and 31.4% for PYC). Furthermore, QStb.ihar-1B.2 (possibly Stb2/Stb11/StbWW) detected on chromosome 1B was a strong resistance locus, conferring resistance in both the seedling and adult plants and explaining up to 70% of the phenotypic variation. Therefore, cv. Mandub proved to be a good source of STB resistance and can potentially be used for resistance breeding.