Pure clove oil stocks and two clove oil-based experimental formulations, named Bioxeda A and Bioxeda B, were provided by Cedax Italia s.r.l. (Forlì, Italy). The two formulations contained micro-encapsulated and nano-encapsulated clove oil, respectively, both at the concentration of 20% (v/v). Two T. durum cultivars susceptible to T. laevis infection were used: Grifoni 235 and Svevo (gently provided by CREA-CI, Research Centre for Cereal and Industrial Crops). These two cultivars showed the same susceptibility to T. laevis infection in preliminary trials carried out by CREA-CI (personal communication). Naturally infected seeds of cultivar Grifoni were used for progressive sowings performed in 2019 and 2020 while, for seed treatments trials, Grifoni and Svevo healthy seeds were artificially inoculated and used in 2020 and in 2022, respectively. Tilletia laevis teliospores used for inoculations were collected from sori of Grifoni naturally infected kernels. Their vitality and virulence were evaluated before inoculation. Artificial inoculations for seed dressing tests were performed following Dumalasová and Bartoš (2008) with dry teliospores: briefly, healthy kernels were placed in plastic bags, together with the teliospores, at the dose of 1g of inoculum for 1Kg of seeds resulting in ca 3×10⁴ spores per seed. The envelope was sealed, and the contents manually shaken for 5 minutes allowing the spores to enter in contact with kernel surface, remaining attached to them. The presence of spores was verified by visually checking the kernels under a stereo microscope.

To set-up the early diagnostic assay, based on Real Time PCR, artificially inoculated wheat seedlings were sampled at the stage immediately previous the tillering (from 20 to 40 days from seedling emergence; Feekes growth stage 2.0 – beginning of tillering; Miller, 1992). DNA was extracted from 10 single plants as described by Li et al. (2017). Briefly, 0.5-1 cm of the basal shoot was cut from the seedlings, quickly sterilized in sodium hypochlorite 2% (v/v), rinsed with sterile water and homogenized in 40 µL of lysis buffer (0.5% casein in 10 mM KOH) using a micropestel. Subsequently, the sample was incubated at 95°C for 5 minutes and cooled on ice for 5 minutes, finally, it was centrifuged at 13000 g for 1 minute. The supernatant was recovered into a new tube and the obtained crude extract was used as template for amplification. A TaqMan Real Time PCR assay was carried out using the primers Tri-DL-For (5’-ATTGCCGTACTTCTCTTC-3’), Tri-DL-Rev (5’-GTAGTCTTGTGTTTGGATAATAG-3’) and the dual-labelled probe (5’Cy5-AGAGGTCGGCTCTAATCCCATCA-BHQ2-3’), targeting a specific sequence in the ITS1 region of T. laevis and other close related species (T. caries, T. controversa, T. fusca, T. bromi, T. goloskokovii). The assay was performed following Tan et al. (2009), modified as described below: the reaction mixture was carried out in 25 μL consisted of 1 × ImmoBuffer (Bioline, London), 5 mM MgCl_2, 0.2 mg/mL BSA, 200 mM of each deoxynucleotides, dATP, dTTP, dCTP and dTTP, 1U Immolase DNA polymerase (Bioline) and 200, 400 and 900 nM of each of the dual-labelled probes, the forward and the reverse primers, respectively. The reactions were performed on a CFX96 Real Time PCR system (Bio-Rad Laboratories Inc., CA, USA). The thermal cycling parameters included an initial cycle of 95°C for 10 minutes, followed by 40 cycles of 94°C for 15 seconds, 65°C for 60 seconds with the annealing temperature decreased by 1°C/cycle for 6 cycles to 60°C. To evaluate the presence of PCR inhibitory effects and to verify the amplification efficiency, serial dilutions of purified T. laevis DNA, from 10 pg to 1 fg per reaction, spiked with a fixed amount (2 μl) of plant crude extract from Tilletia-free wheat plants, were amplified.

To identify the optimal conditions (photoperiod, temperature, rainfall) for T. laevis infection, wheat seeds were artificially inoculated as previously described, and seedlings were grown both outdoor, in pot progressive sowings, and in growth chamber. The term “progressive sowing” refers to a series of repeated sowings deferred at predetermined time points, with the aim of covering the entire period of wheat sowing. For this purpose, five progressive sowings at regular intervals of about two weeks from November to January were carried out for two consecutive years. Each sowing was structured in two pots (pot A and pot B) with 50 inoculated seeds each, located outdoor close to weather station (Vantage Pro2, Davis Instrument, USA) recording temperature, rainfall, and relative humidity.

The time to obtain 50% of seedling emergence (T₅₀) was calculated according to Farooq et al. (2005). For each sowing the emerged plants were randomly sampled from both pots. Half of them were used for molecular assay and half for phenotypic evaluation of the presence/absence of the disease symptoms in plants grown to maturity (Feekes growth stage 11.4; Miller, 1992). In the molecular assay a plant was considered infected if produced a positive signal in the Real Time PCR, whereas in the phenotypic evaluation if at least one sorum was present. The disease incidence was calculated by the formula: (Number of infected plants/Total number of analyzed plants) × 100.

From sowing to sampling date, meteorological data (average temperature, minimum temperature, maximum temperature) were recorded every hour; for each day, the average, lowest and highest temperature values were calculated and collected for each day. Days in which the temperature falls within one of these ranges were summed: low-range included temperatures between –5°C and 5°C (range 1), mid-range included temperatures between 5°C and 15°C (range 2) and high-range included temperatures above 15°C (range 3).

To better understand the ideal pedo-climatic conditions that allow fungal infection, Svevo wheat seeds were artificially inoculated as reported above and planted in pots in a growth chamber (KW Apparecchi Scientifici, Siena, Italy). Pedo-climatic conditions were set according to Zscheile jr (1966) and Swinburne (1963): low soil temperatures after sowing (5-10°C), seed vernalization, short photoperiod early in ontogeny, soil humidity between field capacity and drying point and deep sowing at 4-7 cm. Two main sowing conditions were chosen: 1) after seed vernalization; 2) direct sowing, without seed vernalization. For each condition 60 seeds were sowed at 5 cm depth in the soil, in two different pots (30 seeds/pot) kept at 5-10°C up to third-leaf stage. For all the four pots, a 9-hour-light photoperiod was set after seedling emergence. Seed vernalization was carried out before sowing by allowing the seeds to germinate at 27°C for 48 hours (blotter test) and then leaving them at 4°C for 2 weeks. For T. laevis in planta detection, 25 seedlings per pot were harvested at the third-leaf stage and analysis was performed as reported in 2.2 paragraph.

The treatments were applied to inoculated seeds by seed submersion or seed spraying, as detailed in Table 1 . Seed submersion was carried out by soaking the seeds in the clove oil emulsion (with the addition of 0.05% pinolene as emulsifier) or in the diluted formulations with gentle agitation for 10 minutes; after that they were completely dried on a sterile blotting paper at room temperature in a laminar flow cabinet. Spray treatments were performed with two consecutive applications of the diluted formulations or essential clove oil emulsion with the addition of 0.05% pinolene using a Rotostat (Marline General Engineers Limited, England) seed dressing/coating machine. Pinolene acted as a film-forming compound for improving oil distribution and persistence. Copper sulphate (15.2%) Cutril^® (CU) (Serbios s.r.l., Badia Polesine, RO, Italy) was applied by spray in the second season as a further control treatment allowed in organic agriculture (later coded as CU). Sterile water was applied by submersion and used as control to evaluate the washing effect of the seed submersion treatments without active compounds (mechanical effect). The seeds inoculated with T. laevis spores as described in 2.1 but not treated (NT), were used as positive control. Not inoculated and not treated seeds were used as negative control. Phytotoxicity of the different treatments was evaluated as the effect on seed germination and seedling emergence by counting the number of established seedlings at the beginning of the secondary tillers arising (Feekes growth stage 2.0; Miller, 1992). The whole experiment was carried out for two growing-crop-productive seasons with cultivar Grifoni in 2020 and Svevo in 2022. For each treatment, three replicated pots, each consisting of 30 and 45 seeds for season 2020 and 2022, respectively, were sowed and placed outdoor with a completely randomized design (CRD). After 14 days from sowing, the number of germinated plants was recorded. Plant sampling was performed at the stage preceding the tillering (from 20 to 40 days from seedling emergence; Feekes growth stage 2.0 – beginning of tillering; Miller, 1992) by cutting plants at the culm base and collecting from each pot at least 5 bulks of 5 plants each; the bulks were then processed for DNA extraction following the protocol described in 2.5 paragraph.

For seed dressing evaluation, DNA was extracted from plants sampled in the pre-tillering stage (about one month after sowing). From each seedling, a fragment of 4-5 cm was taken from the culm base. Bulks of 5 seedlings each were homogenized in liquid nitrogen; 150-200 mg of powder was lysed in 700 µL of lysis buffer (2% CTAB; 0.2 M Tris HCl pH 7.5; 0.05 M EDTA pH 8; 2 M NaCl) and incubated at 65°C for 10 min. After adding 70 µL of NaAc (3 M, pH 5.2), the samples were kept in ice for 5 minutes and centrifuged at 13.000 g per 10 minutes at 4°C; the supernatant was recovered and treated with RNAse. DNA was purified with 1 volume of chloroform:isoamyl alcohol 24:1 (Sigma-Aldrich, USA) and precipitated with ½ volume of isopropanol. The pellet was resuspended in 200 µL of milliQ water. Finally, DNA was quantified by a Qubit dsDNA HS assay on Qubit 2 fluorometer (Life Technologies™, Invitrogen, USA). A duplex Real Time PCR was set up, using a 18S universal target for fungi and plants (Ioos et al., 2009) as endogenous calibrator to normalize the T. laevis qPCR signal to the plant qPCR signal. The reaction was carried out in 20 µL volume, containing SensiMix™ II Probe (Bioline, Taunton, Massachusetts, USA), 0.2 mg/mL BSA, 900 and 400 nM of Tri-DL-For and Tri-DL-Rev primers respectively, 200 nM T. laevis probe, 100 nM each 18S primers (18S uni-F, 5’-GCAAGGCTGAAACTTAAAGGAA-3’;18S uni-R, 5’- CCACCACCCATAGAATCAAGA-3’) and 18S probe (5’-JOE-ACGGAAGGGCACCACCAGGAGT-BHQ1-3’) and 10 ng of plant DNA. The thermocycling conditions used were the same applied for the rapid molecular assay. The pathogen biomass was quantified as the ratio of the amplification of T. laevis ITS1 relative to the 18S target, which was calculated as 2^−ΔCt (Livak and Schmittgen, 2001). To verify the correlation between the number of infected plants and the relative quantification values, 5 bulks with known amounts of infected plants were analyzed.

Tilletia infection in progressive sowing assay was evaluated by molecular and phenotypical analysis. The correlation coefficient between the two methods was calculated by Pearson correlation analysis using R software v4.2.2 (R Core Team, 2022) and the package ggpubr v0.5.0 (Kassambara, 2022). A homogeneity test, by a chi-square analysis, was carried out to compare the infection frequencies between the two pots (A and B) both for molecular and phenotypical results.

The meteorological data analysis was performed using a Principal Component Analysis (PCA) biplot, to determine the correlation between temperature and infection. The PCA was performed with R software v4.2.2 (R Core Team, 2022) by using the package factoextra v1.0.7 (Kassambara and Mundt, 2020).

The seed treatments data obtained by Real Time PCR were normalized as described in 2.5 paragraph. Data were then subjected to the analysis of variance (ANOVA) and LSD (least significant difference) test for multiple comparison at p = 0.05. To display and compare the results obtained in the two seasons, relative quantification data were balanced to the mean value of the NT control and the proportion to it of the other treatment mean values was calculated: (T/NT) × 100 with T= relative quantification data mean of the treated samples, NT= relative quantification data mean of the not treated samples. The correlation coefficient between the 2020 and 2022 seasons was calculated by Pearson correlation as described above.

For oil phytotoxicity evaluation, the average percentages of established seedlings of 2020 and 2022 seasons were combined after homogeneity evaluation by a chi-square test (data not shown). They were arcsin transformed before ANOVA, followed by Duncan’s test (p=0.01), however, mean values were back transformed into percentages for making data reading easier.

A Real Time PCR-based method for the early diagnosis of common bunt in wheat seedling has been developed. The qPCR conditions set up in this work showed no inhibition using the crude extract DNA and no differences in the amplification performance between the two extraction methods: the amplification standard curves for both extracts showed an optimal efficiency, a high coefficient of determination (R²>0.99) and the same sensitivity, detecting up to 10 fg of target DNA ( Supplementary Figure 1 ).

The progressive sowings were evaluated together with infection percentages obtained from molecular and phenotypic assay. The disease incidence values in pot A and pot B were combined after homogeneity evaluation by a chi-square test ( Supplementary Table 1 ). In the season 2019/20, the first three sowings (01-03), gave back no bunt infection both in molecular and phenotypic analysis. In the following sowings the infection increased, and the highest values were obtained in the fourth sowing (82% for molecular and 90% for phenotypic analysis), then decreasing by the fifth sowing (22% for molecular and 2.7% for phenotypic analysis). In the season 2020/2021 T. laevis infection was recorded for all the sowings; only by phenotypic analysis for sowing n. 07, and only by molecular analysis for sowing n. 10. The highest infection values were found in sowing n. 06 (20% from molecular and 53% from phenotypic analysis) and n. 08 (48% from molecular and 41.4% from phenotypic analysis) ( Table 2 ).

Despite some minor differences between the two different analyses, the Pearson correlation coefficient showed a strong correlation between them, with a coefficient R of 0.9 ( Figure 2 ).

With the aim of understanding the optimal in vivo conditions for T. laevis infection, the meteorological conditions during the growth period from sowing to the second true leaf stage (about 30 days) were analyzed: this period represents the limit phase within which the seedlings are considered susceptible to infection to Tilletia spp. (Gaudet et al., 2007).

To assess and discriminate if specific temperature ranges play a role in the infection process, data were analyzed by using the PCA biplot (Data in Supplementary Table 2 ). This method geometrically analyzes information in multivariate data. When the variable positions are close, there is a high correlation, moreover, the acute angle between each pair of variables indicates a positive correlation, while the obtuse angle indicates a negative correlation (Jolliffe and Cadima, 2016).

The disease incidence data, from both molecular and phenotypical analyses (MA and PA), T50 and all the variables indicating the number of the days with lower temperature (AVG.T1, LT1 and TH2) were well negatively correlated with the variables indicating the number of the days with higher temperature (AVG.T3, LT2 and HT3) ( Figure 3 ). The negative correlations between the number of the days with higher and lower temperature, together with the distribution of the sowings point on the PC1, suggest a significant role of the temperature in the infection process. As shown in Figure 3 , the sowings dates were well separated on the PCA axis 1 (59.5% of variance) in two groups: presence or absence of infection. Moreover, the greater the days with low temperatures, the higher the molecular and phenotypical infection rates. In addition, T₅₀ had a positive correlation with higher number of days at lower temperatures and higher infection rates. The percentages reported on the axis referred to the variability explained by the principal component. There was no correlation between all variables and number of days with an average temperature ranging from 5 to 15°C (AVG.T2); no day was recorded with the maximum temperature below 5°C (TH1) ( Figure 3 ).

Wheat is a winter cereal and T. laevis needs low temperatures to infect this crop, so we tested the in vivo fungal infection capability in the typical winter temperature range of 5-10°C. The main goal of the experiment performed in growth chamber was to achieve a high level of seedling infection also in controlled conditions, in order to be able to replicate the method for seed dressing and sowing tests. In both seed vernalization and direct sowing, T. laevis infection rates were comparable (approx. 70%), suggesting that seed vernalization did not affect the early infection. The infection rates obtained were in agreement with those reported in literature (Wilcoxson and Saari, 1996).

Duplex qPCR showed both very good efficiency and coefficient of determination (R²>0.99), and a limit of detection (LOD) of 10 fg of T. laevis DNA per reaction ( Figure 4 ). Relative quantification, showing the amount of target amplicon (T. laevis) normalized to the total DNA (18S), correlated efficiently with the amounts of target DNA. Normalized amplifications corresponded to an exponentially increasing trend of about an order of magnitude ( Figure 5A ), and the best fit curve, correlating the data of normalized expression signal to the known amounts of T. laevis DNA, showed a good linear correlation (R² = 0.99) ( Figure 5B ). Furthermore, the presence of the 18S endogenous calibrator guaranteed greater reliability in the quantification, representing an internal amplification control, and allowing the precise quantification of total DNA used as template in the reaction.

The correlation between relative quantification and disease incidence was verified by analyzing bulks with a known number of infected plants (R² = 0.93; Figure 6 ). To compare the infection success between treatment assays in seasons 2020 and 2022, the equation resulting from the interpolation line of the aforesaid data was used to obtain an approximate percentage of infected plants in the untreated control tests. In 2020 the average infection of NT plants resulted to be about 50%, versus 20% in 2022.

Tilletia laevis quantification in seedlings deriving from dressed seeds was performed by TaqMan Real Time PCR. Data were analyzed considering the two growing seasons separately ( Figures 7 , 8 ). The seed treatments were generally able to control bunt infection. In season 2020, significant differences were observed between the treatments performed by submersion (BioxA, BioxB and CO) and NT control. Among the spray treatments, only Bioxeda A gave results significatively different from the NT control, while the other two (Clove oil spray and Bioxeda B spray), together with water treatment, were comparable to NT control. Water treatment has produced a reduction of infection too, probably due to the water washing effect on the teliospores present on the kernel surface ( Figure 7 ). Negative control (not treated and not inoculated) always gave negative results (data not shown).

In season 2022, the Bioxeda B submersion treatment showed the best results as no T. laevis infection was recorded, as well as with the chemical treatment (CU) used as control. The other treatments showed a lower infection rate compared to NT control, suggesting a protective (or just washing) effect against the pathogen, however no significant differences were found ( Figure 8 ). In this season, expression data of the whole experiment, including the NT control, were lower than in season 2020, and no correlation was found between the average infection rates of the tested oil treatments of 2020 and 2022 seasons (R² = 0.09; Supplementary Figure 2 ).

The phytotoxicity of the different treatments used for seed dressing was evaluated as effect on seed germination and seedling emergence, assessing the number of established seedlings before the secondary tillers arising. Data collected from the two seasons were merged and analyzed together: significant differences were found only for clove oil treatment by immersion (CO), compared to NT control ( Table 3 ). This treatment showed a toxic effect on wheat seeds significantly reducing germinability and seedling emergence. A significant correlation was found from the analysis of 2020 and 2022 phytotoxicity data (R² = 0.94; Supplementary Figure 3 ), indicating the reproducibility in the different seasons.