A spelt diversity panel including 80 genotypes of winter spelt was established within the FP7 project HealthyMinorCereals¹. Seeds of spelt genotypes were originally sourced from the germplasm collections of AGES (Linz, Austria), Agroscope (Changins, Switzerland), Crop Research Institute (Prague, Czechia), Institute of Agrobotany (Tápiószele, Hungary), IPK (Gatersleben, Germany), LSA Hohenheim (Stuttgart, Germany), and NordGen (Alnarp, Sweden). The diversity panel contains landraces, old and obsolete varieties, as well as modern varieties and breeding lines, which were classified into three groups in order to investigate the effect of breeding: landraces and old varieties from (i) Switzerland and Germany; (ii) modern varieties (post 1970) from Switzerland, Germany, and Belgium; and (iii) plant genetic resources and genebank accessions of uncertain breeding status (PGR). The Czech variety “Rubiota” (released 1981), Austrian “Ebners Rotkorn” (released 1999), and Hungarian “Öko 10” (released 2000) were treated as landraces/old varieties, as they were selected from landrace accessions.

Field experiments with artificial inoculation of the HMC spelt diversity panel were conducted in Austria, Czechia, and Switzerland from 2016 to 2018, resulting in nine test environments. Artificial inoculation was performed at full flowering of the spikes (BBCH 65) by spraying a spore suspension. FHB symptoms were scored several times after inoculation by visual evaluation of the percentage of diseased spikelets of whole plots. Plant height was measured from the ground to the top of the spike.

In Austria, the field trials were carried out in Tulln (48°19′05″N, 16°04′10″E). The field experiments were performed similar as described in Buerstmayr and Buerstmayr (2016). In brief, single rows of the spelt genotypes were sown in two replications with semi-dwarf wheat rows in between in order to prevent severe lodging. Inoculations were carried out first at BBCH 65 by spraying the highly aggressive F. culmorum isolate IFA-91015 at a concentration of 2.5 × 10⁴ conidia per ml in tap water at a rate of 100 ml m^–2. Inoculations were repeated four times in 2 day intervals in order to inoculate each plot at least once at maximum flowering. After each inoculation, the crop stand was mist irrigated for 21 h with a total of 4–5 mm water. FHB severity scoring (% diseased spikelets) started on day 10 after anthesis and was repeated four times in 4 day intervals. Anther retention was recorded in the 2017 trial, as described by Buerstmayr and Buerstmayr (2016), on five randomly selected ears per plot. After harvest, 20 randomly selected ears were weighted, and the reduction in ear weight compared to 20 ears from control plots was calculated.

In Czechia, the field experiments were established in Prague-Ruzyně (50°05′07″N, 14°18′10″E). The spelt genotypes were sown manually as spikelets in hill plots with three replications. At the flowering period, bunches of 10 spikes were inoculated with the highly pathogenic isolate B of F. culmorum (Šíp et al., 2002) at a concentration of 0.8 × 10⁷ conidia per ml and covered with a polythene bag for 24 h to provide moisture. If necessary, the trial was irrigated from time to time to provoke disease development. FHB symptoms were evaluated three times (i.e., 14, 21, and 28 days after inoculation) on a 1–9 scale, as described (Chrpová et al., 2013).

In Switzerland, the trial was sown under organic conditions in Feldbach (47°14′17″N, 8°47′20″E). The genotypes were sown in 50 cm-long double rows with two replications. The inoculum was a conidial suspension of F. culmorum spores. Spores were kindly provided by Agroscope, Changins. Inoculation started at the beginning of flowering and was repeated three times in 2016 and 2017 and two times in 2018 in a 2 day interval. Scoring of visual symptoms (1–9 scale) was started 10 days after the last inoculation and repeated three times in a 7 day interval.

Dehulled seed samples from all sites were analyzed for deoxynivalenol (DON) content by the RIDASCREEN^® FAST DON enzyme immunoassay (R-Biopharm AG, Darmstadt, Germany). Additionally, the DON content of the glumes was also determined in the Czech samples from 2017 and 2018. A representative grain sample was ground by a Yellowline IKA A10 Analytical Grinder Universal Mill (IKA^®-Werke GmbH & Co. KG, Staufen, Germany) and thoroughly mixed. Afterward, 100 ml of distilled water were added to 5 g ground sample and shaken for 3 min in a horizontal rotary shaker at 200 rpm. After filtration, 50 μl of the filtrate were used for the ELISA test together with standards according to the manufacturer’s instructions. The absorption of the final solution was measured at 450 nm using a Sunrise^TM spectrophotometer. Data were processed with RIDASOFT^®Win.NET software (R-Biopharm AG).

FHB scoring values of the last evaluation date from each test site were converted to the 1–9 scale as described by Chrpová et al. (2013). Normal distribution of data was tested by the Shapiro–Wilk test using procedure CAPABILITY of SAS 9.4 (SAS Institute, Inc., Cary, NC). Heritabilities were calculated according to the concept of “operative heritability” (Strube, 1967). The variance components therefore were calculated by procedure GLIMMIX to account for non-normally distributed data (Stroup, 2015). Procedure GLIMMIX was also used for generalized linear mixed model (GLMM) analysis of variance with genotypes’ status (breeding periods) as fixed effect and environment (i.e., location × year) and its interaction with genotypes as random effects. Procedure CORR was applied to calculate Pearson product-moment and Spearman’s rank correlations (for non-normally distributed traits and correlations between field replications) and procedure TEMPLATE to create the statistical graphs. Genstat 20th Ed. (VSNi, Hemel Hempstead, United Kingdom) was used to calculate a superiority index P_i for each individual genotype according to Lin and Binns (1988) in a modified way using the genotype with the lowest FHB score and DON concentration, respectively, as “best performing” genotype. P_i is defined as the distance mean square between a genotype’s response and each environment’s best response averaged over all environments. The P_i values for FHB and DON were further used to calculate a Fusarium rating as devised by Tamburic-Ilincic et al. (2011) for the Ontario winter wheat performance trials.

The genotypic variation in the FHB score across environments is displayed in Figure 1. It is striking that the variation and FHB response were similar in Austria and Czechia, with highest FHB scores observed in 2018 followed by 2016 and 2017, whereas in Switzerland, the highest FHB scores were recorded in 2017 followed by 2018 and 2016 (Supplementary Table 1). Variations in the FHB scores were observed not only between years and countries but also between field replications. Spearman’s rank correlations were significant (p < 0.0001) for all years in Austria (2016: ρ = 0.77; 2017: ρ = 0.59; 2018: ρ = 0.42) and Czechia in 2016 (ρ = 0.54–0.57), however, not significant for Switzerland and Czechia in 2016 and 2017, respectively. In the other environments, the correlation between field replications was significant (p < 0.05) but weak (ρ = 0.21–0.34).

Heritability for FHB score was high with h²_o = 0.77. Considering the single countries, the highest heritability was observed for Austria (h²_o = 0.79), where the genotypic variance component was 1.4 times the variance due to the residual error. Heritabilities for Czechia (h²_o = 0.54) and Switzerland (h²_o = 0.43) were medium. In these cases, the variance component for the residual error was four to five times higher than the variance explained by the genotypes.

For DON contamination of the grains, genetic variation was negligible in Austria 2017 and Switzerland 2016 (Supplementary Figure 1). Heritability was with h²_o = 0.61 lower than for FHB and did not change if the two before mentioned environments were excluded from analysis. Similarly, to FHB score, the heritability was highest for the Austrian trials (h²_o = 0.67) and significantly lower for Switzerland and Czechia (h²_o = 0.36 and 0.29, respectively). Similarly, to the visual scoring of FHB symptoms, the correlations between the single field replications vary between countries and years. For Austria, the correlations are significant (p < 0.0001) for each year (2016: ρ = 0.79; 2017: ρ = 0.72; 2018: ρ = 0.46); for Czechia (ρ = 0.61–0.75) and Switzerland (ρ = 0.58), similar correlations were observed in 2016, whereas the field replications were not correlated in Czechia 2017 (ρ = 0.15–0.27) and weak (ρ = 0.35–0.47) but significant (p < 0.01) in the other environments.

Across all test environments, modern spelt varieties showed a significantly higher susceptibility to FHB, a higher, but not significantly different, contamination of the grains with DON, and a significantly reduced plant height compared to old landraces/varieties and plant genetic resources (Table 1). The Swiss landrace “Muri Rotkorn” showed very low disease severity in all nine environments, resulting in a low superiority index P_i = 0.18. Considering a cutoff of 10%, the other genotypes with a low P_i were “Toess 6D” (P_i = 0.24), “Fuggers Babenhauser Zuchtvesen” (P_i = 0.28), “Sofia 1” (P_i = 0.45), “Gugg 4H” (P_i = 0.48), “Roter Schlegeldinkel” (P_i = 0.57), “Riniken Weißkorn” (P_i = 0.71), and “Strickhof” (P_i = 0.75). Hence, all genotypes with low disease severity were from the group of old landraces/varieties and plant genetic resources, respectively. The best modern variety was not ranked before 25 (i.e., “Samir”). In contrary, among the 10% most susceptible genotypes with the highest P_i values, six out of eight were modern varieties, i.e., “Alkor” (P_i = 5.31), “Schwabenspelz” (P_i = 5.46), “Titan” (P_i = 5.6), “Badenkrone” (P_i = 5.96), “Zollernspelz” (P_i = 6.02), and “Cosmos” (P_i = 6.09). “Thürig Rotkorn Nr. 4” (P_i = 5.84) and “Waggershauser Hohenheimer Weißer Kolben” (P_i = 6.33) were the only old varieties within the group of highly susceptible spelt varieties (Supplementary Table 2).

For mycotoxin contamination, the picture is similar. The 10% genotypes with the lowest mycotoxin contents in the grains were only old landraces and varieties, i.e., “Fuggers Babenhauser Zuchtvesen” (P_i = 9.5), “Farnsburger Rotkorn Nr. 6” (P_i = 10.4), “Gugg 4E” (P_i = 13.8), “Ebners Rotkorn” (P_i = 20.4), “Gugg 9F” (P_i = 27.9), “Toess 6D” (P_i = 32.7), “Roter Schlegeldinkel” (P_i = 35.2), and “Ostro” (P_i = 36.2), whereas the highest DON concentration was observed mainly in modern varieties, i.e., “Badenkrone” (P_i = 1,805), “Titan” (P_i = 1,885), “Cosmos” (P_i = 2,090), “Alkor” (P_i = 3,736), and “Goldir” (P_i = 4,021). In this case, however, also a few old landraces showed high DON contents, i.e., “Zuzgen Nr. 15A” (P_i = 1,817), “Schnotwiler Weißkorn Nr. 35” (P_i = 2,214), and “Thürig Rotkorn Nr. 4” (P_i = 2,282) (Supplementary Table 2).

In Table 2, results of the most popular spelt varieties in Austria, Czechia, Germany, and Switzerland during the investigation period are listed. “Ostro” and “Ebners Rotkorn,” the most widely grown varieties in Switzerland and Austria, respectively, showed relatively low levels of disease severity and mycotoxin contamination, resulting in rankings on the top of the tested germplasm. Both varieties were among the tallest ones and are genetically related, having their origin in old Swiss landraces. “Oberkulmer,” another old Swiss landrace variety and check for traditional spelt quality, and “Rubiota” and “Franckenkorn,” popular varieties in Czechia and Germany, respectively, showed medium FHB resistance and DON contamination. Especially striking is that “Rubiota” performed significantly inferior to the top-ranked old German landrace variety “Fuggers Babenhauser Zuchtvesen” [P_(FHB) = 0.36; P_(DON) = 9.5] from which “Rubiota” was selected. On the other hand, “Franckenkorn” showed a similar medium resistance level as the abovementioned varieties but having a significantly reduced plant height. High disease severity and mycotoxin contamination were observed for “Titan” and “Zollernspelz,” two varieties released in the early 2000s. Due to its reduced plant height and therefore increased lodging tolerance, “Zollernspelz” is the most popular variety in Germany and is also widely grown in Austria and Czechia.

The Fusarium rating based on the ranks of the superiority indices for both FHB and DON summarizes the results obviously. Only 16 genotypes were rated as “resistant” (Supplementary Table 2), and all of them belong to landrace varieties and/or genetic resources (Figure 2) including “Ostro” and “Ebners Rotkorn.” The “moderately susceptible” group includes 25 genotypes, six of them are modern varieties; however, only two of them, i.e., “Franckenkorn” and “Zürcher Oberländer Rotkorn,” are still marketed. This group includes also “Oberkulmer Rotkorn” and several genotypes classified as genetic resources with a significantly reduced plant height, e.g., “Black Forest” and “H57-7.” In the “susceptible” group, which includes 26 genotypes, almost all modern varieties from the Belgium breeding program are present besides many landrace varieties, among them “Rubiota.” Finally, the “highly susceptible” group includes 13 genotypes, landrace and modern varieties, among them also “Zollernspelz.”

Correlations between the main traits were all significant (p < 0.05). The closest relationship was observed between FHB scores and DON content of grains (r = 0.66) (Figure 3A), whereas the correlations between plant height and FHB scores (Figure 3B) and/or grain DON content were weak (r = −0.35 and r = −0.28, respectively). Considering correlations only within the three test countries, higher correlation coefficients were observed between FHB scores and grain DON content for Austria (r = 0.80) and between FHB scores and plant height for Czechia (r = −0.43) (Table 3).

In two environments, i.e., CZ17 and CZ18, also the mycotoxin content of the hulls was determined, which was, on average, three to five times higher than in the grain (Supplementary Table 1). The correlation between the DON content in the grain and in the hulls was r = 0.56 (p < 0.001) (Figure 3C). In one environment, i.e., AT17, anther retention (number of retained anthers in the floret after anthesis) was determined, and this trait showed a correlation of ρ = 0.56 to disease severity (Figure 3D) and ρ = 0.53 to grain DON content. Reduction in ear weight, which was determined only in Austria, was significantly (p < 0.001) correlated to both FHB scores (r = 0.65) and grain DON content (r = 0.63).