The FHB-susceptible T. turgidum cv. Marco Aurelio was the cultivar of interest in the present work for both field and greenhouse experiments. This genotype, grown in Central and South Italy, is extensively used for pasta production and it is characterized by excellent protein content and high productivity, thus it is of high economic importance. The experimental fields were located in Amelia (Central Italy, 42°31′22.9′′N, 12°25′15.5′′E, Umbria Region) and Avigliano Umbro (Central Italy, 42°40′41.1′′N, 12°27′44.6′′E, Umbria Region). The T. turgidum field experiments were carried out during two consecutive years (2019 and 2020), controlled and drip-irrigated with a nutrient solution containing nitrogen, potassium, phosphorus, and small amounts of other compounds. Therefore, abiotic stresses (nutrient deficiencies and drought stress), diseases, or pests, which cause the same symptoms as FHB, were significantly minimized. On the other hand, fungicides for FHB management (tebuconazole and azoxystrobin) were not applied in order to favor the FHB natural infection. The fields were investigated periodically (one time per day) by experts and farmers to prevent other diseases, pests, or abiotic stress. An experimental plot was allocated within each field, and individuated by a 20 m grid by positioning 16 ground control points (GCPs) to be used for georeferencing. During March, April, and May 2019 and 2020 weather data (minimum, maximum and average temperature, and precipitation) were recorded daily by two meteorological stations installed at 100 m distance from each field to monitor the climate factors influencing the establishment of FHB infection. Historical weather data (from 2010 to 2018) were obtained from the Hydrographic service of Umbria Region¹ to monitor the climatic trend of the last 11 years and to compare minimum, average and maximum temperature values (°C) and average precipitations (mm) recorded during the 9-year seasonal average (2010–2018), 2019, and 2020. The historical weather data were collected from two meteorological stations located in Amelia (42°33′25.0′′N, 12°25′01.0′′E) and Avigliano Umbro (42°40′39.0′′N, 12°26′13.0′′E). The experimental design for the field experiments is illustrated in Figures 1A,B.

An unmanned DJI Matrice 600 multi-copter (DJI, China) equipped with a Zenmuse X5 RGB camera (DJI, China) and a Zenmuse XT TIR camera (DJI, China) has been used in this study (Supplementary Figure 1). DJI Matrice 600 is a hexa-copter (110 cm diagonal size) with a highly resistant carbon fiber frame, offering a 15 kg take-off weight. Its maximum transmission distance is 5 km and its maximum flight time is 40 min (Supplementary Figure 1).

The Zenmuse XT was equipped with a 9 mm f1.4 lens. Its thermal sensitivity is less than 50 milliKelvins and the camera enables measurements in the range −25°C to +135°C. The image sensor is a focal plane array (FPA) based on uncooled microbolometers with a spectral band response in the range of 7.5 to 13 μm. The camera field of view is equal to 69° (horizontally) × 56° (vertically), its resolution to 640 × 512 pixels at 30 frames per second (fps), and its spatial resolution to 1.889 milliradians (mrad). The camera captures images at an acquisition frequency of 30 Hz. The camera was radiometrically calibrated to further increase temperature measurement precision by setting external parameters during the flight planning such as air temperature and flat field correction (FFC). This is an offset calibration usually performed at power up, when the camera changes temperature, and periodically during the operation. This calibration compensates for certain errors that build up during the camera operation. During the data acquisition phase, the auto gain mode parameter was applied; the camera automatically selected the optimal gain mode according to the temperature range of the image (Deery et al., 2016; Gómez-Candón et al., 2016; Ludovisi et al., 2017) and pictures were stored as 14-bit digital raw images.

The Zenmuse X5 is a 16-megapixel RGB camera equipped with an M4/3 sensor enabling it to capture detailed images at a resolution of 4608 × 3456 pixels and an ISO range of 100–25600.

The Zenmuse X5 and XT cameras were mounted on a highly reliable gyrostabilized 3-axis gimbal (DJI, China) that automatically stabilizes them in flight. The gimbal constantly communicates with the UAV, and quickly compensates for every minor movement with a precision accuracy of 0.02°.

The 16 GCPs were used to georeference RGB images. For calibration of thermal images, four 60 × 60 cm ground reference panels (GRPs) consisting of two plastic panels covered by black vinyl tape and two white Teflon^® were positioned along the borders of each of the two study areas. A real-time kinematic (RTK) global navigation satellite system (GNSS) CS10 model (Leica Geosystems, Switzerland) with an accuracy of 1 mm was used for capturing GCPs and GRPs locations.

UAV campaigns were conducted at the following phenological stages: (i) three-quarters of inflorescence emerged (Zadoks stage 57, ZS 57) (Zadoks et al., 1974) on 6 May 2019 and 7 May 2020; (ii) anthesis half-way (ZS 65) on 17 May 2019 and 18 May 2020; and (iii) kernel watery ripe (ZS 71) on 28 May 2019 and 29 May 2020. Two flights were performed during each campaign. The flight missions have been planned using Ground Station Pro app (DJI GS Pro, China). Each flight lasted 11 min covering an area of 1 ha at a nominal speed of 4 m/s and an altitude of 20 m, during which both cameras acquired nadiral images with 90% frontal and side overlap. To ensure similar solar illumination angles flights were performed between 11:00 and 12:00 local time under stable cloudless and low-wind conditions (Figures 1D,E).

For each of the 16 GCPs, a circumference of 1 m of diameter was individuated and denominated as sampling area. During the UAV campaigns, we manually measured the spike temperature of sixteen randomly selected plants in each of the 16 sampling areas using a portable infrared thermometer (Fluke 568, Fluke Corporation, United States) with an accuracy of ± 1% or ± 1.0°C (whichever is greater), positioned at 10 cm distance from the spike. At the same time, the FHB severity was calculated by counting the number of diseased spikelets and the total number of spikelets for each of the sixteen plants. Furthermore, the flag leaves of the selected plants were sampled and dark-adapted for 1 h before measuring their photosynthetic efficiency, by quantifying F_v/F_m reflecting the potential quantum efficiency of photosystem II (Maxwell and Johnson, 2000), with a portable fluorometer (V2.00f PAM 2000, Heinz Walz GmbH, Germany). Moreover, the spikes were also sampled to distinguish FHB positive (FHB+) from FHB negative (FHB-) areas by molecular identification of Fusarium spp. (Figure 1C).

Spikes were stored in a portable fridge immediately after sampling. Afterward, different tissues (palea, lemma, glume, rachis, and kernel when present) from each collected spike were plated on Petri dishes containing potato dextrose agar (PDA) and incubated at 21°C for 72 h. The different morphotypes were subsequentially isolated and cultured on PDA for 1 week. After each morphotype filled the Petri dish, the produced mycelium was gently scraped with a glass rod. 100 mg of mycelium were grinded and DNA was extracted using 80 μL of extraction buffer composed of Tris 100 mM, EDTA 50 mM and NaCl 500 mM; after that, 32 μL of SDS 10% (w/v) were added and the samples were incubated at 65°C for 10 min. 27 μL of potassium acetate 5 M were added and the samples were placed on ice for 20 min, then centrifuged for 20 min at 13.000 rpm. The supernatant was recovered and 80 μL of cold (−20°C) isopropanol were added to each sample; then, the samples were placed on ice for 10 min and centrifuged for 5 min at 13.000 rpm. The supernatant was discarded and 150 μL of cold (−20°C) ethanol (70% v/v) were added to the samples. Samples were centrifuged for 3 min at 13.000 rpm, the supernatant was discarded and the DNA was resuspended in 20 μL of DNase and RNase-free sterile distilled water and stored at −20°C (D’Ovidio and Porceddu, 1996). Total DNA was quantified with Qubit^TM fluorometer 1.01 (Invitrogen, United States) using the Qubit^TM dsDNA BR Assay Kit (Thermo Fisher Scientific, United States) and diluted to 10 ng/μL. The molecular identification of Fusarium spp. was performed by amplifying the translational elongation factor 1-α (TEF) sequence using the primer pair EF1_F 5′-ATGGGTAAGGAGGACAAGAC-3′/EF2_R 5′-GGAAGTACCAGTGATCATGTT-3′ designed to identify the FHB complex spp. (Geiser et al., 2004). The polymerase chain reaction (PCR) was performed following the instructions of GoTaq^® Green Master Mix (Promega, United States) and prepared in a total volume of 10 μL. The amplification conditions consisted of: (i) an initial denaturation step of 2 min at 95°C; (ii) 35 cycles of 30 s denaturation at 95°C; (iii) 40 s of annealing at 53°C; (iv) 60 s of elongation at 72°C; and (v) a final elongation step of 5 min at 72°C. The amplicon unicity was visualized on 1.5% agarose gel and sequenced by Sanger sequencing (Eurofins Genomics, Germany). The resulted sequences were submitted to BLASTn² in order to identify the corresponding Fusarium spp.

A total of 250 RGB images were acquired using the UAV-mounted Zenmuse X5 camera. The camera exposure mode was set to shutter priority (S) with a shutter speed fixed 1/500 s to ensure minimization of motion blur, ISO of 100 and the white balance to sunny mode.

Geometric camera calibration, orthorectification, and mosaicking of the captured images were performed using Pix4Dmapper (Pix4D, Switzerland) software, specifically designed to process UAV images using techniques rooted in both computer vision and photogrammetry to match conjugate points in overlapped images and to define their relative positions and orientations using bundle block adjustments (Bollard-Breen et al., 2015; Nishar et al., 2016; Figure 1F). The output of this step was the RGB orthomosaics of the study areas for each flight mission. No filtering process was applied to the images.

Finally, VIs from the orthomosaics were computed using quantum geographic information system (QGIS) software (version 3.4 Madeira - QGIS Development Team, Open Source Geospatial Foundation). The VIs carried out were: vegetative (VEG) (Hague et al., 2006) and green leaf index (GLI) (Louhaichi et al., 2001). VEG and GLI were calculated using the following formulas:

where R, G, and B are the reflectance of red, green and blue channels, respectively, and a is equal to 0.667.

Inside each sampling area, leaves and spikes were not separated in the images, because the upper part of the fields was uniformly composed by spikes (Figures 1A,B), since wheat canopies were very dense.

The UAV-mounted Zenmuse XT TIR camera captured 219 greyscale and georeferenced images. Thermal orthomosaics were generated for each flight, and the radiometric conversion was automatically performed using the Pix4Dmapper software (Figure 1F). The removal of bare soil pixels was not necessary, since the wheat planting were extremely dense, reducing the mixed pixel problem. FLIR Tools software (2020^© FLIR^® Systems, United States) was used to calibrate temperatures in TIR images using the temperature values of GRPs measured immediately after the UAV flights using a portable infrared thermometer (Fluke 568, Fluke Corporation, United States). Calibration checks were performed by comparing the GRPs-measured to the UAV-derived temperatures. No filtering process was applied to the images. The extraction of temperature values was performed for each of the 16 sampling areas. Inside each sampling area, leaves and spikes were not separated in the images, because the upper part of the fields was uniformly composed by spikes (Figures 1A,B), since wheat canopies were very dense.

Greenhouse experiments were conducted in a glasshouse located in Viterbo (Central Italy, 42°25′35.8′′N, 12°04′49.3′′E, Lazio Region). The surface of the T. turgidum kernels was sterilized with sodium hypochlorite (0.5% v/v) for 20 min and rinsed twice with sterile distilled water for 5 min. The kernels were germinated in the dark on a paper, soaked in sterile distilled water for 15 days at 4°C to break dormancy, followed by 2 days at room temperature. The seedlings were transferred to 40 × 20 cm pots, filled with TYPical Brill soil (Brill, Germany) and grown at 16–20°C up to the boot stage (ZS 51), 20–24°C during heading and anthesis (ZS 53-69), and 24 –29°C up to maturity (ZS 71-99). The plants were fertilized using ammonium nitrate in the following proportions and at the following stages: 20% at sowing (ZS 00), 40% at tillering (ZS 20), and 40% at heading (ZS 49) (Francesconi et al., 2019). The highly virulent and mycotoxin-producing isolate of F. graminearum wild type (WT) 3824 (Tomassini et al., 2009) was cultured at 21°C on synthetic nutrient-poor agar (SNA) to obtain macroconidia (Mandalà et al., 2019). After 10 days on SNA, the conidia were scraped with a glass rod after pipetting 1 mL of sterile distilled water onto the surface of a Petri dish. The conidial suspension was recovered, and the concentration was adjusted to 1 × 10⁵ conidia/mL using a Thoma chamber (0.100 mm depth and 0.0025 mm²). The inoculum was prepared in sterile distilled water supplemented with 0.05% (v/v) of Tween-20. Plants were subjected to three different treatments: (i) drought stress treatment where plants did not receive water from ZS 51 to 65 (Zadoks et al., 1974; Tambussi et al., 2000); (ii) artificial inoculation treatment where spikes were uniformly spray-inoculated at ZS 65 with a suspension of 1 × 10⁵ conidia/mL; and (iii) mock treatment where spikes were uniformly sprayed with a suspension of Tween-20 0.05% resuspended in sterile distilled water at ZS 65. The spikes were covered with clear plastic bags for 24 h to maintain high humidity levels (>80%). Spikes subjected to drought and mock treatments were sampled after removing the plastic bags, while the inoculated spikes were sampled 24, 48 and 72 hours post inoculation (hpi) to investigate an early response to F. graminearum. Collected spikes were immediately stored in liquid nitrogen at −80°C until the extraction of RNA. FHB severity (%) was monitored in the greenhouse by counting the number of bleached spikelets and the total number of spikelets for each spike from 3 to 21 days post inoculation (dpi). In addition, spike temperature and photosynthetic efficiency were recorded at 24, 48, and 72 hpi. The experimental design for the greenhouse trial is illustrated in Figure 1G. Data were obtained from three independent experiments, each experiment consisting of 20 spikes for each treatment.

Wheat spikes were ground with mortar and pestle in liquid nitrogen until a fine powder was obtained. The RNA was extracted from 100 mg of powder following the instructions provided by InviTrap^® Spin Plant RNA Mini Kit (Stratec Molecular GmbH, Germany), resuspended in RNase-free sterile distilled water, immediately poured onto ice and quantified with Qubit^TM fluorometer 1.01 (Invitrogen, United States) using the Qubit^TM RNA BR Assay Kit (Thermo Fisher Scientific, United States). To confirm the total quantity and integrity of the RNA, 5 μL of the extracted RNA sample was subjected to a 10-min thermal shock at −80°C, followed by 5 min at 65°C and run on 1.5% denaturing agarose gel. The synthesis of cDNA was performed using 500 ng of RNA following the instructions provided by Xpert cDNA Synthesis Supermix with a gDNA eraser (GriSP Research Solutions, Portugal) in a final volume of 20 μL. To ensure that the synthesis of the cDNA and the elimination of the gDNA had succeeded, a reverse transcription PCR (RT-PCR) of T. aestivum Actin (TaACT) (containing an intron in the amplified sequence) was performed following the instructions provided by GoTaq^® Green Master Mix (Promega, United States) in a total volume of 10 μL. The amplification conditions consisted of: (i) an initial denaturation step of 2 min at 95°C; (ii) 35 cycles of 30 s denaturation at 95°C; (iii) 40 s of annealing at 60°C; (iv) 30 s of elongation at 72°C; and (v) a final elongation step of 5 min at 72°C. The amplification run included a no-template control (NTC) and a genomic DNA (gDNA) control. The amplicons were visualized on 1.5% agarose gel.

Supplementary Table 1 shows the list of target genes, their functions, and the corresponding primer pairs used to perform RT quantitative PCR (RT-qPCR) (Francesconi and Balestra, 2020). Briefly, the primer pair for T. aestivum glyceraldeyde-3-phosphate dehydrogenase (TaGAPDH) amplification is from Jarošová and Kundu (2010), for T. aestivum pathogenesis related protein 1 (TaPR1) from Lu et al. (2006), for TaACT from Tundo et al. (2016), and T. aestivumβ-tubulin2 (TaTUB) and T. aestivum ferredoxin-NADP(H)-oxidoreductase (TaFNR) from Tenea et al. (2011). The remaining primers are from Francesconi and Balestra (2020). The amplification efficiency (E) of RT-qPCR was determined for each primer pair as follows: five 1:10 serial dilutions (1:1-1:10000) were obtained for each cDNA and amplified in four replicates. E and coefficient of determination (R²) values were calculated by means of the slope of the standard curve obtained by plotting the fluorescence versus the serial dilution concentrations using the equation (Bustin et al., 2009)

Reference genes with closest E values to target genes, highest R², and lowest variability were selected for the quantification cycles (Cq). The relative expression levels of target genes were calculated on the basis of the Cq values of four independent biological replicates, each with four technical replicates, for each plant treatment by applying the equation (Bustin et al., 2009)

using TaACT, TaTUB, and TaFNR as reference genes and the mock treatment to normalize the relative expression levels. Relative expression levels of TaPR1 and TaGAPDH were quantified as internal control of the progression of the infection (Muthukrishnan et al., 2001) and changes in photosynthesis (Zhang et al., 2013). The RT-qPCR was performed following the instructions provided by Rotor-Gene Q (Qiagen, Germany) and Xpert Fast SYBR (uni) MasterMix (GRiSP Research Solutions, Portugal), in a final volume of 10 μL. The amplification conditions consisted of: (i) an initial denaturation step of 3 min at 95°C; (ii) 40 cycles of 5 s denaturation at 95°C; (iii) 30 s of annealing at 60°C; and (iv) 20 s of elongation at 72°C. A final melt cycle (70–99°C) was performed to confirm the unicity of the amplicons. NTC controls were included and the amplification was considered negative when a value of Cq ≥ 38 was detected (Bustin et al., 2009).

Spike temperature and photosynthetic efficiency were measured for each plant treatment as performed during the field trial, described in section “Ground Measurements.” Photosynthetic-related parameters were measured for three independent replicates, each consisting of 20 individual spikes, for every treatment.

One-way analysis of variance (ANOVA) was performed to analyze FHB severity, ground-measurements (temperature and photosynthetic efficiency) of FHB+ and FHB- spikes during the UAV-campaigns, UAV-based VEG, GLI and temperature of FHB+ and FHB- sampling areas during the UAV-campaigns, gene expression values, and temperature and photosynthetic efficiency of drought stressed and inoculated plants during the greenhouse experiments. One level of significance (p < 0.01) was computed to assess the significance of the F values. A pairwise analysis was carried out using Tukey’s honest significant difference (HSD) test at 0.99 confidence level. Statistical analyses were performed using XLSTAT 2020.4 software (Addinsoft, France). Principal component analysis (PCA) was carried out to classify spike temperature, photosynthetic efficiency and VEG or GLI or UAV-based temperature values coming from FHB+ or FHB- areas and gene expression values, spike temperature and photosynthetic efficiency coming from drought stressed or F. graminearum inoculated plants. Heatmap was carried out by computing the z-score of the relative gene expression values. PCA and heatmap were computed by using ClustVis software (Tartu, Estonia)³ (Metsalu and Vilo, 2015).

Registered data were compared to historical data (2010–2018) (Figure 2A), showing that May 2020 was particularly hotter than 2010–2019 (the recorded average temperatures were 20°C, 14°C, and 18°C in May 2020, May 2019 and May 2010–2018, respectively), while May 2019 was characterized by high daily average rainfall (7 mm) compared to May 2020 (2 mm) and the 9-year seasonal average (5 mm). These conditions favored the naturally occurring FHB in the fields of interest. In fact, in 2019, FHB severity (Figure 2B) reached 59% and 92% at ZS 65 and 71, respectively, indicating that the wet season (mean relative humidity was > 70%) was particularly favorable for the FHB to spread. Although April and May 2020 were not characterized by frequent rains (average precipitation of 2 mm), low precipitations favored a moderate FHB infection with a severity of 27% and 68% at ZS 65 and 71, respectively.

Figure 3 illustrates the results obtained by ground-based measurements during the UAV campaigns. The molecular identification of FHB was performed by amplifying the TEF sequence from a bulk sample obtained from the samples collected in each sampling area, producing an amplicon of 700 bp. At ZS 57, all the sampling areas resulted FHB- in both 2019 and 2020; at ZS 65, 10 of the 16 sampling areas were FHB+ in 2019, while 6 sampling areas were FHB+ in 2020; at ZS 71, all the sampling areas were FHB+ in both 2019 and 2020 (Figure 3A). In 2019, thirty Fusarium morphotypes were isolated: ten morphotypes were F. graminearum, eleven were F. poae, eight were F. avenaceum, and one was F. proliferatum. In 2020, twenty-four morphotypes were identified: ten were F. graminearum, ten were F. poae and four were F. avenaceum. The isolated morphotypes and the data resulted from the BLASTn analyses are listed in Supplementary Table 2. Ground-based spike temperature values recorded in 2019 and 2020 revealed that FHB+ spikes had a higher temperature than the FHB- ones at ZS 65 (Figures 3B,D). On the other hand, photosynthetic efficiency had an inverse relationship with the FHB severity: variable fluorescence/maximum fluorescence (F_v/F_m) demonstrated to be lower in FHB+ than in FHB- at ZS 65 (Figures 3C,E).

Recorded weather data indicated that during the two UAV-campaigns conducted at ZS 65, the average daily air temperature and humidity values were 15°C (2019), 21°C (2020), and 67% (2019) and 72% (2020), respectively. Supplementary Figure 2 shows the 16 sampling areas resulting from RGB (A and C) and thermal images (B and D) in Amelia (Supplementary Figures 2A,B) and Avigliano Umbro (Supplementary Figures 2C,D). PCA (Figure 4) demonstrated that VEG (A), GLI (B), and UAV-based temperatures (C) distinguished between FHB+ and FHB- plants at ZS 65 for both the 2019 and 2020 campaigns. Moreover, box-plots indicate that VEG (D), GLI (E), and UAV-based temperature values significantly differed between FHB+ and FHB− areas during the two UAV-campaigns (2019 and 2020). The average VEG values (D) recorded in FHB− areas were 1.46 and 1.42, while in FHB+ areas were 1.00 and 1.02 in 2019 and 2020, respectively. The average GLI values (E) recorded in FHB− areas were 0.21 and 0.13, while in FHB+ were 0.09 and 0.10 in 2019 and 2020, respectively. The average UAV-based temperature values (F) recorded in FHB− areas were 18.32°C and 18.94°C, while in FHB+ were 21.06°C and 20.94°C in 2019 and 2020, respectively.

The progress of FHB severity was monitored in F. graminearum-inoculated plants in the greenhouse from 3 to 21 dpi (Figure 5A). The severity gradually increased reaching values close to 100% at 17 dpi, confirming the susceptibility of T. turgidum cv. Marco Aurelio. Figure 5B shows the FHB symptoms at 21 dpi. While no symptoms were observed on the mock treatment, few non-necrotic bleached spikelets appeared on drought-stressed plants.

Figure 6 shows the temperature of spikes (Figure 6A) and the photosynthetic efficiency (Figure 6B) after the three treatments. Compared to the mock treatment, the temperature increased and the photosynthetic efficiency decreased in drought-stressed and inoculated plants, confirming a perturbation of the photosynthetic activity. In fact, the more the infection progressed, the higher the differences in spike temperature and photosynthetic efficiency between the inoculated and mock plants: inoculated spikes reached a temperature of 17.89, 18.48, and 18.67°C while mock spikes measured 17.12, 16.44, and 16.19°C, at 24, 48 and 72 hpi, respectively; the photosynthetic efficiency measured 0.719, 0.695 and 0.618 F_v/F_m for the inoculated spikes and 0.794, 0.793 and 0.801 F_v/F_m for the mock at 24, 48 and 72 hpi, respectively. Notably, F. graminearum infection perturbated the photosynthetic parameters more than drought stress, highlighting that, by using these measures, it is possible to distinguish between drought-stressed and FHB−infected plants.

E, R², and the stability of the reference genes were calculated to validate the RT-qPCR results. E ranged from 0.9652 to 1.2741 and R² from 0.9651 to 0.9954. Standard errors (SE) among the Cq values of the three reference genes ranged from 0.198 to 0.369 indicating their stable expression under the three different treatments.

Since metrics derived from TIR and RGB images allowed the detection of infected spikes at ZS 65, the greenhouse experiments were designed to investigate the differential stomatal regulation response in proximity of the same phenological stage.

Moreover, drought-stressed plants were studied to observe their gene expression differences with plants under F. graminearum inoculation. Figure 7A shows a heatmap of the relative expression values of plant genes under drought stress and F. graminearum inoculation at 24, 48, and 72 hpi. Supplementary Table 3 provides the relative expression values, SE, and the HSD test computed at 0.99 confidence level. Under terminal drought stress, T. aestivum allene oxide synthase (TaAOS), T. aestivum terpene synthase (TaKSL), T. aestivum mitogen-activated protein kinases (TaMAPK), T. aestivum calcium dependent protein kinase (TaCDPK), T. aestivum phosphatase (TaABI), T. aestivum MYB domain transcription factor (TaPIMP), T. aestivum NADPH oxidase (TaRBOH), and T. aestivum zeaxanthin epoxidase (TaZEP) were slightly up-regulated showing expression values ranging from 1.254-fold to 1.892-fold. Among the different time-points of F. graminearum inoculation (24, 48, and 72 hpi), the relative expression values of TaAOS, T. aestivum abscisic acid (ABA) aldehyde oxidase (TaAAO), T. aestivum ABA receptor (TaREC), T. aestivumβ-1,3-glucanase (TaBG), TaMAPK, TaCDPK, T. aestivum epoxycarotenoid dioxygenase (TaNCED), TaRBOH, and TaZEP gradually increased from 24 to 72 hpi, while T. aestivum hydroperoxide lyase (TaHPL) and T. aestivum cytochrome P450 (TaCYP450) were down-regulated. Notably, the expression patterns of TaKSL, TaAAO, TaREC, TaBG, TaCYP450, TaNCED, and TaZEP were different between drought-stressed and F. graminearum-inoculated plants. In particular, in the inoculation treatment, at 72 hpi TaREC, TaBG, TaNCED, and TaZEP were strongly up-regulated (5.729, 5.143, 4.988, and 4.256-fold change, respectively) while TaPR1 and TaGAPDH were gradually up-regulated from 24 to 72 hpi, indicating that the F. graminearum infection perturbated the innate immunity and physiological photosynthesis of the plants. Figure 7B represents the PCA obtained by computing relative gene expression values, spike temperature and photosynthetic efficiency values from drought-stressed and F. graminearum inoculated plants. PCA demonstrated that these data distinguished between hydric stress and F. graminearum inoculation.