The study was conducted during three growing seasons, from 2017 to 2019, on potted red grapevines (Vitis vinifera L.) cv Cannonau and Carignano, grafted on 1103 Paulsen rootstock in 2016, at the experimental farm of the University of Sassari in Oristano, Italy (39°54′11.87″ N, 8°37′7.62″ E at 13 m.a.s.l.). The climate of the study area is typically Mediterranean. The annual precipitation is approximately 540 mm; the minimum, maximum, and average daily solar exposure are approximately 6.8, 26.1, and 15.7 MJ m⁻², respectively; and air temperatures (T _air) range from average maximum values of 30°C during the hottest month and minimum averages of 7°C during the colder month (Fernandes de Oliveira et al., 2019). The experimental blocks were conducted on 2-year-old plants (40 plants per treatment and variety) potted on 50-L containers, with the soil substrate composed of fine soil, peat, and perlite (2:1:2). During the first year, all pots were colored white, and a single shoot was maintained and trained vertically along the trellis, supported by a vertical trellis with two parallel movable wires ( Figure 1 ). In the 2nd and 3rd years, two and three vertical shoots were kept after spur pruning, and the canopy was topped frequently to keep the height of the vegetative wall within 2 m. In the spring of 2017, two blocks of plants of the two varieties were placed in a greenhouse (20-m length, 10-m width, 5-m maximum roof height, and 2.5 m at the gutter) with a metal support structure and covered with UV-blocking corrugated polycarbonate panels (Omega 76/15, 0.8 mm Suntuf Plus clear embossed, Palram Industries Ltd., Ramat-Yohanan, Israel), with 80% solar photosynthetically active radiation (PAR) transmission. Two doors located on the north and south sides, two side vents on the east side, and two circulation fans at 4 min height in the west promoted ventilation ( Figure 1 ). The Cannonau and Carignano plants were arranged in adjacent rows, spaced 2.5 m apart, with north–south orientation. The greenhouse was divided in two modules of 80 plants each, separated by a UV-shielding polycarbonate sliding door. The northern module was equipped with a UV-B lighting system (Narrowband TL 100W/01 SLV/10, Phillips Lighting Holding B.V., Amsterdam, NL), suspended on a longitudinal track of adjustable height, above each plant row (treatment +UV). The daily UV-B dose was supplied throughout the growing seasons, between 9.00 h and 11.00 h, was 4.63 kJ m⁻² d⁻¹, while the environmental daily dose at the site, measured in June in the control block, was approximately 4.05 kJ m⁻² d⁻¹, and it was approximately 0 kJ m⁻² d⁻¹ in the south greenhouse UV-blocking module (−UV treatment). A PAR lighting system with high-pressure sodium lamps (Master GreenPower CG T 400W E40 1 SL/12, Phillips Lighting Holding B.V., Amsterdam, NL) ensured that, during the early morning hours (7.00 to 9.00 h), PAR photon flux density inside the greenhouse reached similarly high values simultaneously with the control plants (control, C) exposed to direct solar radiation ( Figure 1 ). The PAR radiation flux was supplied both to the north and the south experimental block in the greenhouse. Irrigation was supplied with a surface, single-line drip irrigation system (4 l h⁻¹), in order to keep high water availability from budburst to cluster closer (BBCH 9 to 80) and mild to moderate water stress thresholds (Ojeda, 2007) during the fruit developmental stages (BBCH 80–89). Irrigation was applied two to three times per day to avoid deficient drainage problems, with a 30-min maximum duration during the hottest periods (i.e., approximately 6 L plant⁻¹ day⁻¹).

The two varieties are among the most important red grape varieties grown in Sardinia: Cannonau is widespread throughout the territory, occupying as much as 29% of cultivated areas, while Carignano is typically grown in hot, arid, and highly sunny environments and prone to water and heat stresses only in 7% of the viticulture areas (Nieddu, 2011). They have very different ampelographic and physiological characteristics (Cipriani et al., 2010; Nieddu, 2011; Fernandes de Oliveira and Nieddu, 2015b). Cannonau plants have glabrous or slightly tomentose shoots, small and glabrous leaves, and shorter internodes, with a lighter green color compared to Carignano. The latter has highly tomentose, green shoots with the dorsal face of the internodes streaked with red color. The leaf is large and wedge-shaped; the upper leaf blade is slightly tomentose, while the lower is bristly (Nieddu, 2011). Compared to Cannonau, Carignano develops less secondary leaf area (Fernandes de Oliveira and Nieddu, 2016b; Fernandes de Oliveira et al., 2019). The two varieties showed different physiological behavior under water deficit: Cannonau exhibits tighter stomatal control over transpiration and higher constancy in minimum water potentials, while Carignano shows higher stomatal conductance and progressively lower minimum water potential. The varieties present quite different phenolic profiles (Vacca et al., 2009), and lower berry anthocyanin contents can be found in Cannonau, especially during a hot and dry ripening season (Fernandes de Oliveira and Nieddu, 2016b). An important genetic distance was shown in previous studies (Lovicu, 2017). Among the group of traditional Sardinian varieties, the genetic distances between Cannonau and Carignano was estimated by agglomerative hierarchical clustering based on a comparison among 22 SSR profiles, confirming the genetic diversity between the two varieties, which had only 36% of alleles in common ( Figure 2 ) (Pelsy, 2010; Lovicu, 2017).

Meteorological data from the closest weather station (distance = 250 m) were used to assess the monthly rainfall, average daily solar radiation, and daily maximum, average, and minimum temperatures over the seasons. The weather conditions over the three seasons were compared with the climatic 30-year reference period (https://www.sar.sardegna.it/). Supplementary Table S1 reports the monthly average temperatures and accumulated precipitation during each of the three seasons of trial and the monthly average of the reference historical series. The mean values of the study years were compared with the climate series ( Supplementary Table S1 ). Canopy temperature (T _canopy), incoming global solar radiation (Rg), and PAR were continuously recorded on three plants of each treatment and variety, from the beginning of the growing season until harvest, at 10-minute intervals using temperature microprobes (MTP microprobes, GMR instruments, FI, IT) and radiation sensors coupled to dataloggers (GMR instruments, FI, IT), placed on an internode at mid-shoot height, close to a lower leaf blade, and fixed with polytetrafluoroethylene tape. The Rg and PAR sensors were calibrated using a pyranometer and a PAR sensor (Skye instruments, Ltd., Llandrindod Wells, UK) to check the accuracy on solar irradiance assessment under field conditions. The PAR and Rg sensor calibration curves and hourly average values of Rg, PAR, UV-A, and UV-B measured on a sunny day of June, outside and inside the greenhouse, using individual sensors connected to a portable SpectroSense+ datalogger (Skye Instruments, Llandrindod, Wells, UK) are presented in Supplementary Figures S1 , S2 , respectively. All sensors were placed at the middle part of the shoots in early growth stages ( Figure 1 ) and thereafter kept within 60 cm of the apex. Supplementary Figure S3A, B reports the daily UV-B supplied in the C and +UV module of the greenhouse, measured in the plants, taking into account the sensor calibration curves and the vertical fields of irradiance attenuation under UV lamp (Aphalo et al., 2012). Sunlight intensity inside and outside the greenhouse was also measured with a portable ASD spectroradiometer (FieldSpec 4 Hi-Res 350–2,500 nm, Malvern Panalytical, Westborogh, MA, USA) ( Supplementary Figure S3C ). The duration of low (18°C–22°C), mean (24°C–30°C), high (>35°C), and very high (>40°C) T _air ranges in C, +UV, and –UV environments, respectively, was computed ( Figure 3A ) based on the data recorded by an ATMOS 14 T/RH, monitoring system inside a radiation shield positioned at 2 m in height between vine rows and connected to a Em50 datalogger (Meter Group Inc., Pullman, WA, USA). The 10, 50, and 90 percentiles of exposure to different T _canopy ranges throughout the study periods were calculated to summarize the differences in thermal environment among the three treatments throughout the growing season ( Figure 3B ). Canopy temperature depression (T _air – T _canopy) was evaluated for each variety × treatment combination, expressed as average difference between T _air and T _canopy along the day ( Supplementary Figure S4 ). Phenological succession was recorded weekly in each treatment, and the accumulated thermal time for each phenophase was compared using the Normal Heat Hour model (Mariani et al., 2012).

Stem water potential was monitored in the morning (8.30 h) and at solar noon (13.00 h) on mid-canopy adult leaves during veraison (BBCH 81) using a portable pressure chamber (Pump up, PMS Instruments, USA). Prior to the measurement, the leaves were enclosed for 1 h in plastic-coated aluminum bags to allow balancing leaf water status with that of the stem. On the same day, daily leaf gas exchange was monitored at 8.30, 10.30, 13.00, and 16.00 h using an infrared gas analyzer (CIRAS−3) coupled with a Parkinson’s leaf chamber for single leaf measurements. Net assimilation rate, stomatal conductance, transpiration rate, intrinsic water use efficiency leaf temperature (T _leaf), leaf vapor pressure (e _a), and the vapor pressure deficit (VPD) between the leaf and the air were measured on fully expanded, well-exposed leaves, keeping the reference air within a CO₂ concentration of 400 µmol mol⁻¹ and 50% relative humidity. A standard photosynthetic photon flux density of 1,000 µmol m⁻² s⁻¹ was imposed during the measurements by connecting a LED light unit to the leaf chamber.

Direct chlorophyll fluorescence of PSII was monitored in vivo on fully expanded leaves from pea size (BBCH 75) until the beginning of senescence and leaf color change (BBCH 92). After a dark adaptation period of 30 min, fluorescence transients were recorded using a Handy PEA fluorimeter (Hansatech Instruments Ltd., King’s Lynn, UK), after applying an actinic saturating red light (650 nm) intensity of 3,000 µmol m⁻² s⁻¹, in single flashes, with the signal gain at 1.0 and a 30-s duration in each replicate. Rapid fluorescence kinetics from the minimum (F ₀) to the maximum (F _m) fluorescence of dark-adapted leaves was recorded at a 10-µs time step, allowing for main and derived fluorescence variables calculation ( Table 1 ) and thus for evaluating PSII performance during the photochemical and thermal phases of the fluorescence transient (Strasser et al., 2000). The main derived variables analyzed were as follows: maximum yield of primary photochemistry of PSII, φP_o; maximum water splitting efficiency, WSE; quantum yield for electro transport, φ_eo; quantum yield for energy dissipation, φDI_o; density of the reaction centers, RC_QA; and the linear indicator of functional PSII units, 1/F ₀ − 1/F _m (Flexas et al., 2001; Strasser et al., 2007). After the direct measurements on intact leaves, PSII chlorophyll fluorescence was also monitored on leaf samples collected for lab analysis.

Based on chlorophyll fluorescence screening method (Goulas et al., 2004), leaf chlorophyll, CHL, anthocyanin, ANT, flavonol, FLA e nitrogen balance index, and NBI (Cartelat et al., 2005; Diago et al., 2016) were estimated in vivo, in field, and in laboratory adult leaf samples from the middle shoot height using Dualex scientific+ (Force-A, Paris, FR), a portable leaf clip analyzer that measures the transmittance and screening effect of polyphenol chlorophyll fluorescence (Cerovic et al., 2012). The measurements were performed during flowering (BBCH 65), pea size (BBCH 73), cluster closure (BBCH 80), veraison (BBCH 83), and in 6 days of the year (DOY), from May to October 2018, on six adult leaves per treatment and variety with five reads in each leaf blade side. The measurements were repeated in the laboratory, and the leaf area of each replicate was estimated (Fernandes de Oliveira and Nieddu, 2016a; Fernandes de Oliveira et al., 2021). Furthermore, leaf mass per area (LMA) was estimated on 5-cm-diameter leaf blade discs, dried at 60°C for 48 h, and weighed. Then, leaf relative water content (RWC) was calculated. All the leaf samples used for spectrophotometric measurements were previously washed and dried at room temperature.

Spectral reflectance was measured using a FieldSpec 4 Wide-Res spectroradiometer (range, 350–2,500 nm; ASD Inc., Boulder, CO, USA) coupled to a plant probe with leaf clip for live vegetation. Each spectral measurement was the average of 10 spectra, keeping the black standard under the leaf sample (n = 6). The measurements were preceded by a dark current measurement and a white reference measurement. Five independent measurements were taken on each sample on both adaxial and abaxial leaf blades, and the average reflectance of each leaf was then calculated. Thereafter, the reflectance spectra were normalized at the reciprocal reflection spectrum of chlorophyll a (678 nm) to highlight the different proportion of pigments as compared to chlorophyll a (Merzlyak et al., 2003; 2008). Other reflectance indexes were calculated in order to characterize specific differences among treatments and varieties related to leaf composition and structural traits, namely—the normalized difference chlorophyll index, NDCI, for chlorophyll estimation (Gitelson and Merzlyak, 1994; Mishra and Mishra, 2012); TCARI/OSAVI (Daughtry et al., 2000; Wu et al., 2008); the modified anthocyanin reflectance index, mARI (Gitelson et al., 2009); the normalized difference index for leaf area mass per area, ND_LMA; and the normalized dry matter index, NDMI (Cheng et al., 2014).

Leaf colorimetric estimation of chlorophyll, anthocyanin, and total polyphenol contents was carried out using a CARY 50 Scan UV-Vis Varian spectrophotometer (Agilent Technologies, Santa Clara, CA, USA), following the method of Lichtenthaler (1987), after the extraction of compounds from 15 discs (of 16-mm diameter) on 10 mL methanol solution. Finally, chlorophyll a (Chl a), chlorophyll b (Chl b), and carotenoid (Car) contents were derived as proposed by Lichtenthaler and Wellburn (1983). Total phenols were determined with Folin–Ciocalteu colorimetric method using a solution of 100 µL extract. The absorbance readings were taken at 750 nm using Hitachi model 100–60 spectrophotometer (Hitachi High-Technologies Corporation, Tokyo, Japan). Total phenol content was expressed in terms of catechin concentration, mg m^-2. For anthocyanin determination, the Di Stefano and Cravero (1991) methodology was applied. In addition to the fresh weight of each leaf sample, dry weight was determined after drying in an oven at 105°C for more than 24 h until a constant mass value was reached. Thereafter, the samples were pulverized for carbon, hydrogen, and nitrogen content determination using an infrared EN CHN 628 series elemental analyzer (LECO Corporation, St. Joseph, MI, USA).

In each treatment and variety, six adult leaves were collected from the outer leaf wall of the canopy, sealed in plastic bags, and transported to the laboratory for morphological and anatomical analyses. The leaves were washed, cleaned, scanned, and processed with Image J software (https://imagej.net/) to measure the main leaf vein length. Leaf swelling spots were delimited in each leaf image, using freehand selections, to calculate the percentual leaf swelling area in each leaf area. Then, leaf cross-sections were collected, aside from the main vein, with a razor blade. The sections were colored using two differential stains applied successively, iodine green and carmine alum, mounted on glass slides, then washed and covered with adhesive tape, and dehydrated with increasing concentrations of ethyl alcohol solutions (50%, 75%, 90%, and 100%). The images of the leaf cross-section were captured with an Infinity digital camera (Lumenera Corporation, Ottawa, ON, CA) mounted on a Zeiss Axiophot optical microscope (Zeiss Microscopy, Oberkochen, DE). Leaf thickness, mesophyll thickness, and percentage of mesophyll cross sectional area occupied by palisade and spongy mesophyll were observed and measured with the ×20 objective. Stomata and intercellular chamber were also observed in leaf cross-section images.

In order to observe further morphological differences in leaf tissues, namely, organelle structure, metabolite deposition, and overall cellular activity, leaf blade cross-sections (3 × 3 mm) were excised from leaves in each variety and treatment, fixed with 2.5% glutaraldehyde in 0.2 M sodium phosphate buffer (pH 7.2), and post-fixed with 1% osmium tetroxide in the same buffer for 1 h. The samples were then dehydrated in graded acetone series and embedded in Agar100 epoxy resin; ultra-thin sections were cut on an ultramicrotome RMC MT-7 using a diamond knife and stained with uranyl acetate and lead citrate. The sections were then examined in a Zeiss EM109 transmission electron microscope operating at 80 kV (Zeiss Microscopy, Oberkochen, DE).

Two-way analysis of variance (ANOVA) and least significant difference (LSD) test were performed on all data, using the SPSS 25.0 software (SPSS Inc., Chicago, IL, USA), in order to compare means and detect significant differences among treatments and cultivar at the 95% confidence level as well as to highlight the statistically different interaction effect among the two main factors leading to plant susceptibility and adaptive responses to light and heat stresses.

The most relevant weather records during the 3 years of study concerned the systematic increases in maximum temperature during the summer months, from 2017 to 2019 (the anomalies ranging from +3.2°C to +5°C), and in 2018 the maximum temperatures were above the climatic 30-year average from April until July (+3.2°C and 2.8°C). As far as the light microclimate is concerned, the solar irradiance transmitted inside the greenhouse, measured in June during a clear day, was significantly different in both UV-A, visible, and infrared wavelengths (350 to 1,000 nm) compared to the outdoor environment ( Supplementary Figure S3C ). For what concerns the visible wavelengths, the solar irradiance transmitted inside the greenhouse at midday was similar to that recorded outside by midmorning and approximately 11% lower than that measured outside in the early afternoon and 12% higher than that measured in the greenhouse at midmorning. The average irradiance in near-infrared wavelengths (760–1,000 nm) was doubled inside the greenhouse compared to that which occur naturally (about double in 960–1,000-nm wavelength; 0.31 vs. 0.15 kW m⁻², respectively). The Rg intercepted by the canopies inside the greenhouse was reduced by ca. 24% due to the foliage filtering effect ( Supplementary Figure S3A ). On average, UV-B radiation in the inner canopy (at ca. 60 cm) reached similar percent values in +UV plants as those intercepted by the control canopies exposed to direct solar radiation. Finally, the daily average canopy Rg and UV-B intercepted and the daily average accumulated Rg and UV-B in control and greenhouse plants are shown in Supplementary Figure S3B . Inside the greenhouse, the +UV and −UV canopies intercepted ca. 1,539 W m⁻² of Rg per day, i.e., approximately 76% of the Rg measured in control canopies during the summer.

The thermal environment along growth stages was characterized by a percent duration of low T _air (18°C–22°C) inside the greenhouse of 33%−35% during June, July, and August and of 40% in September, while in the control environment such temperatures were recorded for a 35% duration in July and August and approximately 40%−45% in June and September ( Figure 3A ). In the greenhouse, T _air ranges of 24°C to 30°C did not exceed 18−20% duration, while in the outdoor control such temperatures were recorded for as much as 30%−35% season duration. The most important differences between environments regarded the permanence of high (>35°C) and very high T _air (>40°C). Overall, the thermal trends in both greenhouse modules, +UV and –UV, were similar throughout the experimental trial. As far as plant thermal microclimate is concerned, Figure 3B shows the T _canopy trends from May to August 2018 and from April to October 2019, expressed in terms of 90th, 50th, and 10th percentiles. The T _canopy of C plants, in both varieties, was always lower than that of the plants under +UV and –UV treatments, except for the beginning of August 2018 when a prolonged extreme heat wave was recorded, resulting in a sharp increase in heat stress conditions even in the control potted plants. Throughout the duration of the trial, the Cannonau plants subjected to the –UV treatment reached higher T _canopy than the Carignano plants. In 2019, the differences in T _canopy among control and greenhouse treatments became more pronounced; the plants in the greenhouse recorded significantly higher T _canopy than that under control conditions in both varieties. Though there was no strong upward trend in temperature along the season, the T _canopy inside the greenhouse was 10°C higher than those outside for 90% of the temperature regime and 3°C−5°C higher in the greenhouse than in the control for 50% of the recorded T _canopy and 1°C−2°C higher in the greenhouse than in the control in 10% of these records. Compared to the previous season, the maximum differences in T _canopy between the control and the two treatments in the greenhouse nearly doubled in 2019. The differences between +UV and –UV treatments inside were also not evident in Carignano, while they were quite different in Cannonau for which the –UV plants had warmer canopies than that under +UV treatment. Regarding average daily thermohygrometric conditions, the greenhouse and control environments were markedly different. On average, the maximum T _air values recorded between 12.00 and 16.00 h were approximately 30°C, and air RH was 50%, while in the greenhouse average T _air was 40°C with a RH of 40%. The overnight T _air values were similar in all treatments, with the lower values reaching 20°C and with approximately 80% of RH ( Supplementary Figure S4 ). Thus, +UV and –UV treatments were characterized by the persistence of high T _air, large day–night ranges of T _air (± 20°C), and RH (± 40%).

The duration of each phenological stage was not significantly different among the three environments ( Figure 4A ). Nevertheless, the year effect on the duration of phenological stages was quite pronounced, particularly on the duration of flowering and berry growth stages. In 2017, no phenological advances from bud break until veraison were observed inside the greenhouse as treatment differentiation began shortly before bud break, yet the low chilling requirements of grapevines to overcome endormancy, together with the increased accumulated heat units for vegetative growth restart inside the greenhouse in 2018, resulted in a cumulative advance of 8 days in bud break and successive growth stages as compared to control plants ( Figure 4A ). The influence of interannual meteorological differences on the phenological stage occurrence and duration was higher than that of the meso and microclimatic differences among treatments. These included a delay of 16 days in bud break 2019 and a 24-day delay in flowering compared to 2018 ( Figure 4B ). Moreover, flowering duration was much shorter, nearly half, in all treatments in 2019. In fact, flowering duration was the most sensitive stage to inter-annual weather variations, and higher uniformity of successive fruit growth interphases was observed (only 5- and 2-day difference in pea size date in control and greenhouse, respectively). A marked effect of T _air in the occurrence of veraison was evidenced, with a 7-day earlier onset in 2019 compared to 2018. The Normal Heat Hours model enabled an accurate classification of the local thermal resources for grapevine development and illustrated the prolonged warming effect of the greenhouse on phenological successions, particularly during summer.

As expected, plant growth was stimulated in the protected environment, with intense shoot and leaf area development ( Supplementary Figure S5A ) that was kept at a height lower than 2.5 m with frequent topping operations. For this reason, up to the third year of the experimental trial, the vigor of control shoots at pruning was not statistically different from that of the greenhouse plants in Cannonau, yet the control plants of Carignan showed significantly lower vigor than greenhouse plants ( Supplementary Figure S5B ). Similarly, an increasing trend in average main leaf area development in Carignano plants subjected to C, –UV, and +UV was recorded, with an average value of 104, 207, and 249 cm² in Carignano compared to 48, 59, and 58 cm² in Cannonau, respectively. The average main leaf size was significantly larger (approximately double) in the –UV and +UV treatments (133 and 154 cm², respectively) than in C (approximately 75 cm²). From cluster closure (BBCH 80) onward, the differences between the two greenhouse treatments became significant, with the +UV plants showing twice as much single leaf area as the –UV and approximately four times the leaf area of control leaves. Conversely to Carignano, the differences in average main leaf size were not statistically significant in Cannonau, and this variety showed less development of single main leaf area from flowering to harvest (BBCH 74–89).

From bud break to the onset of ripening (BBCH 9–80), the plants were kept under high soil water availability, and stem water potential did not fall below –0.6 MPa. Irrigation supplies kept plants under mild water stress, thus not limiting photosynthetic performance and efficiency ( Figure 5 ), even during the afternoon, although T _leaf exceeded 35°C and T _air remained very high for a long time ( Figure 3A ). The pattern of net assimilation rate over a hot day of veraison evidenced differences among varieties in response to the different T _air and light microclimate among treatments. During the morning, net photosynthesis reached higher values inside the greenhouse, and significantly lower values were recorded in Cannonau control plants in early morning but not in Carignano, and by midmorning the differences among treatments and varieties were significant. These results can be ascribed to the PAR supplied in early morning and the higher photosynthetic performance of leaves under high diffuse PAR environments. By midday and in the afternoon, net photosynthesis decreased, and the differences between treatments and varieties were further attenuated. During the morning, Carignano plants showed similar stomatal conductance and transpiration rates among treatments, while Cannonau exhibited a higher morning stomatal control over transpiration only in C plants. In C plants, stomatal conductance varied little over the day in the two varieties, with only a slight increase during the morning of Carignano (ca. 300 mmol m⁻² s⁻¹) as compared to Cannonau (less than 200 mmol m⁻² s⁻¹). Inside the greenhouse, stomatal conductance increased significantly but varied among similar values in both varieties, from approximately 400 mmol m⁻² s⁻¹ in the morning to an afternoon maximum that exceeded 1,600 and 1,200 mmol m⁻² s⁻¹ in –UV plants Carignano and Cannonau, respectively ( Figure 5 ). These resulted in a sharp reduction of intrinsic water use efficiency, compared to C plants, that reached more than twice the values of greenhouse plants. However, Carignano +UV leaves were able to sustain significantly higher water use efficiency in the afternoon compared to the other plants inside the greenhouse. The transpiration rate over the day followed an increasing trend inside the greenhouse, with higher average values in the afternoon, though significantly lower in Cannonau +UV leaves compared to –UV.

The effects of the treatments on the primary photochemistry performances of dark-adapted leaf samples are presented in Table 1 . At fruit set (BBCH 73), main differences regarded 1/F₀ − 1/F _m and RC_QA, the latter being higher in Carignano leaves subjected to +UV and in Cannonau –UV leaves. As the season proceeded, with increased high T _air and continuous exposure to the light treatments, the differences became more pronounced. Carignano C and Cannonau +UV plants exhibited higher QP₀, with less investment in energy dissipation, and also higher PSII functionality, RC_QA, Qe₀, and WSE. Greenhouse treatments, particularly −UV, resulted in a significant reduction in the functional units of PSII, RC_QA, and WSE measured at mid-ripening (BBCH 89). However, in Cannonau, the +UV treatment resulted in increased 1/F₀ − 1/F _m, WSE, and RC_QA compared to C. After ripening, the differences concerning leaf photochemical efficiency flattened, and as senescence began (BBCH 92) these differences disappeared.

The comparison of CHL, ANT, FLA, and NBI estimates with the relative leaf pigment and nitrogen content indicates a high correlation between estimated data and leaf compound quantification in the lab ( Figure 6 ; Supplementary Tables S2–S4 ). The correlation patterns were different between varieties. However, the most pronounced differences were observed among treatments. As far as the nitrogen balance is concerned, data indicate a higher concentration on greenhouse leaves, with higher nitrogen contents per unit of leaf dry weight in +UV and −UV than in C leaves. Moreover, the average chlorophyll contents were significantly higher in the greenhouse than in the control plants and in Carignan rather than Cannonau, although the average values of CHl b/Chl a ratio for the whole season were not statistically different among treatments and varieties ( Supplementary Table S3 ). Among the leaf compositional differences, water content and fresh weight were also significantly higher in Cannonau compared to Carignano and particularly in C compared to −UV and +UV. The −UV plants had lower turgid leaf weight compared to C and +UV in both varieties. Leaf dry matter was significantly higher in Carignano, and the leaf mass per single leaf was significantly higher in Carignano than in Cannonau, being also significantly lower in C and higher in +UV plants. The treatment effect was more pronounced in Carignano.

The patterns of CHL, ANT, FLA, and NBI highlighted the effect of greenhouse in increasing and keeping high CHL values throughout the season in both varieties ( Figure 7A ). This effect was maximized in +UV plants at fruit set (BBCH 73). Along the season, fluctuations in CHL with T _leaf were observed in all inverse treatments and varieties, except for the measurements taken by leaf senescence (BBCH 92). The ANT measured on grapevine leaves varied among very low values over the season. Higher values were recorded by fruit set (BBCH 73) and during leaf senescence (BBCH 92) in both C and +UV plants. During three central dates, this index varied among similarly low values in every treatment and variety. In addition, FLA and NBI suggest significantly higher flavonol accumulation +UV leaf tissues, in both varieties though with large fluctuation both in FLA and NBI over the season. Figure 7B shows the dynamics of total chlorophyll, anthocyanin, phenols, and leaf nitrogen contents corresponding to the CHL, ANT, FLA, and NBI values shown in Figure 7A , as estimated based on the correlation patterns in Figure 6 . When comparing the data measured in the field with that of leaf samples in the lab, the differences observed with optical estimation were amplified ( Figure 7B ). The chlorophyll and polyphenol estimates were markedly higher in +UV and smaller in −UV, and also the amplitude of variations along the season was higher. Anthocyanin estimates remained stably low in C and +UV, in Cannonau, while an important and consistent increase in estimated anthocyanins was observed in Carignano, in +UV, and even more markedly in −UV. In addition to the treatments, a different varietal response to the presence/absence of UV-B light was observed, characterized by increasing increment in chlorophyll estimates in Cannonau +UV along the season, even during the central period, for which the daily duration of high T _air was maximum. In contrast, Carignano +UV leaves showed a decrease in estimated chlorophyll during the same period.

Different reflectance spectra, normalized at the reciprocal reflectance spectrum of chlorophyll a, were observed between treatments in the two varieties, and a different evolution of these spectra was recorded during flowering (BBCH 65), pea size (BBCH 75) and cluster closure (BBCH 80) ( Figure 8 ). During flowering, Carignan plants in the greenhouse exhibited higher values of spectral reflectance in most of the radiant spectrum, including the green (530−590 nm) and especially the red (705−745 nm) regions of the visible wavelengths, the near-infrared (760−900 nm), and short-wave infrared, SWIR (1,570−1,650 nm) regions. In Cannonau, C and −UV plants exhibited much lower reflectance values, compared to Carignano, while the +UV treatment resulted in higher reflectance values similar to those recorded in Carignano +UV. At pea size stage, the spectral reflectance in the red and infrared was much higher in all treatments, being significantly higher in Carignano −UV and Cannonau C. At this stage, the significance of differences between treatments in the SWIR region was reduced in both varieties, and Cannonau C and −UV leaves had similar reflectance values in the green region, which were higher than in +UV. At veraison, data variability increased markedly, and differences between treatments in leaf reflectance lost statistical significance over the entire radiant spectrum. However, Cannonau leaves kept the trend of higher reflectance in the green region and lower reflectance in the red and infrared. Similarly, analyzing the values of the vegetation indices ( Supplementary Table S5 ), it can be noted that, on the first measurement date, the varietal differences in the estimated contents of photosynthetic pigments with NDCI were statistically significant, higher in Carignano regardless of treatment. The UV-B treatment reduced significantly the RWC only in Carignano leaves ( Supplementary Table S3 ). In addition, TCARI/OSAVI and mARI indicated a lower relative abundance of chlorophylls and anthocyanins in Cannonau and a different leaf structure compared to Carignano. Moreover, the ND_LMA showed significant differences due to treatments, i.e., higher mass per leaf area in the +UV plants and slightly lower leaf dry matter in the −UV plants. At pea size (BBCH 73), the varietal differences were intensified, and the positive effect of treatments on NDCI increased in C and +UV compared to −UV, while TCARI/OSAVI increased in −UV plants, indicating chlorosis. Reflectance differences among treatments due to leaf structural characteristics were detected with ND_LMA, the highest values being measured on C treatment and on Cannonau plants. At veraison (BBCH 83), the main differences in the vegetation indices were due to treatments, and TCARI/OSAVI was markedly higher in −UV plants of Cannonau and lower in the Carignano +UV and Cannonau C.

Beyond varietal ampelographic differences among Carignano and Cannonau leaves, the latter being much smaller, pale, and glabrous, inside the greenhouse, Cannonau plants and more conspicuously in −UV plants, the occurrence of leaf blade spot depigmentation was observed from early stages of shoot development (BBCH 14–15) and in all years of trial. The spots then evolved into depigmented, macroscopic swellings scattered on the leaf blade that expanded as the season advanced and T _air increased ( Figure 9 ). This abnormality was most evident when the duration of very high temperatures reached or exceeded 10%, particularly in the −UV treatment. When the spot swelling symptoms were more pronounced on the leaf blade, the optical measurements of CHL, ANT, FLA, and NBI indicated absence of signal in many replications, indicating the transparency for the light flux of the leaf section sampled. These sensor’s reading failure indicated that chlorophyll excitation light pulses in the reference wavelengths and chlorophyll fluorescence light pulses for flavonoid estimation generated 100% transmission and zero fluorescence, respectively.

The areas of swelling, delimited and measured in the laboratory, on six adult leaves taken at veraison (BBCH 83) covered approximately 15.6% of leaf blades in Cannonau −UV and expanded further during ripening to ca. 25.6% of the leaf area by the end of ripening (BBCH 89) ( Supplementary Figure S6 ). It was also possible to observe and delimit depigmentation and swelling on Cannonau +UV samples. However, in this case, the area affected was much circumscribed and reduced, covering approximately 2.3% and 4.2% of the single leaf area at veraison and at ripening, respectively. The C plants did not show such leaf blade deformations, and in Carignano this phenomenon was not seen. The aggravation of symptoms in Cannonau inside the greenhouse, as T _air increased up to 40°C, has led us to hypothesize that anatomical or ultrastructural differences within the leaf tissues caused by high T _canopy determined this phenomenon.

Figure 10 shows that at veraison (BBCH 83) the e_a in C plants varied among medium-low values during the day, much lower than the corresponding saturating vapor pressure, e_s ( Figure 10A ). Conversely, inside the greenhouse, the e_a values within the leaf mesophyll were higher and closer to the saturating pressure values ( Figures 10B, C ). In the −UV plants, e_s reached extremely high values and the e_a in the leaves was very close to those maxima ( Figure 10C ). Therefore, it is likely that the e_a within mesophyll interspaces and in almost empty cell protoplasm, in the depigmented areas of the leaves most affected by high temperatures, caused increasing swelling, leading to the deformation of the membranes. Besides tissue compression, the collapse of several mesophyll cells has certainly heightened this effect in Cannonau plants. The very rapid regression of the symptom of swelling as the T _air decreased and the leaves that were detached from the plant support the hypothesis that e_a at high temperatures caused the swelling of the depigmented and hollow leaf areas.

To shed more light on the causes of such leaf shape alteration affecting Cannonau −UV and, albeit to a lesser extent in Cannonau +UV, the anatomical differences between the two varieties and the three treatments were further investigated in the laboratory, on samples collected at veraison (BBCH 83) and mid-ripening stages (BBCH 85), by observing cross-sections of leaf blade on photonic microscope ( Figures 11 , 12 ; Supplementary Figure S7 ) and also the cell ultrastructure of cross-sections through leaf blades on a transmission electron microscope ( Figures 13 and Cameron et al., 2022). Both varieties had a single layer of palisade parenchyma and spongy parenchyma. The main anatomical differences between treatments regarded leaf thickness, proportions of epidermis, spongy and palisade mesophylls to the whole leaf thickness, and, finally, the size of intercellular spaces versus thickening of lacunose parenchyma. In Carignano C, parenchyma thickness was higher than in the greenhouse treatments. However, the overall proportion of these tissues to the whole leaf thickness was lower because the epidermis was also much thicker ( Figures 11A , 12A ; Supplementary Table S6 ). In the +UV treatment, spongy parenchyma thickness was much lower, and in Cannonau −UV a marked increase in thickness and densification of the spongy mesophyll cells adjacent to the adaxial epidermis was observed ( Figure 12F ). The differences between varieties under −UV treatment mainly concerned unevenness of leaf section and high thickness and cell density of spongy mesophyll in the cross sections of Cannonau leaves compared to Carignano ( Figures 11 , 12 ; Supplementary Figure S7 ). This was due to both the formation of swelling areas on the leaf blade and the compression of the tissues in the adjacent areas under high temperature and leaf e_a in Cannonau −UV plants.

Also ultrastructural differences were observed by transmission electron microscopy on leaf blade cross sections ( Figures 13 , 14 ). The electron micrographs of Cannonau C specimens showed a number of large plastoglobules within the thylakoid membrane, along with large central vacuoles water filled ( Figure 13A ). There were multiple starch grains and different electron-dense material lining the organelles in the cytoplasm. Cannonau –UV leaf sections showed cells with cytoplasm vacuolation, and electron-dense flocculent content with fluid aspect associated with chloroplast or nucleus envelopes, or connected with groups of small, electron-empty vesicle in the central vacuole ( Figure 13B ). Numerous plastoglobules were observed in the thylakoid membranes. As compared to −UV, the +UV leaf cells showed similar vacuolation of the cytoplasm and membrane dilation, but less organelle disturbance ( Figure 13C ), although chloroplast swollen and central hyaline vacuole expansion were evident. Plastoglobules, massive electron-dense deposits, probably of phenolic compounds, were observable, either as vesicular deposits aside the organelles in the spongy mesophyll cells, or fully filling the cell central space, or forming large series of ultra-dense droplets inside the central vacuole the mesophyll.

Leaf ultrastructure in Carignano C plants also showed extensive vacuolization with both large central vacuoles and vesicles ( Figure 14A ). Swollen chloroplasts with a number of plastoglobules of great size were observed and electro-dense deposits were visible. The −UV Carignano leaves cells showed secretory deposition and large cell vacuolation ( Figure 14B ). The micrographs of Carignano +UV leaves showed quite intense cell membrane activity ( Figure 14C ). Besides to the thick cell wall, swollen chloroplasts and large plastoglobules, also convoluted invaginations and folding of the cell wall and plasma membrane were observed. These structures formed a complex net in the inner cell walls, enveloping numerous starch grains and electro-dense deposits. The +UV Carignano leaves also presented a number of raphide and druses crystals.