Introduction

Rotundone is the main contributor to peppery notes in grapes and wines (Wood et al., 2008). The biological function of this sesquiterpene has not yet been elucidated, but it could be involved in plant natural response, notably regarding powdery mildew (Erysiphe necator) (Geffroy et al., 2015). As any other compound from the same class, its biosynthesis and emission by the plant could be affected by environmental factors such as temperature, plant water constraints, and light (Duhl et al., 2008). Indeed, previous studies have shown that temperatures particularly above 25 °C, were detrimental (Zhang et al., 2015) whereas water supply through irrigation effectively enhances rotundone levels in grapes and wines (Geffroy et al., 2016).

The influence of light and ultraviolet (UV) radiation is not yet clearly established, with studies reporting contrasting outcomes. Indeed, investigations were generally focused on defoliation under field conditions, where it was challenging to separate the respective effects of light and temperature. Defoliation applied on both sides of the row markedly reduced rotundone concentration in Duras wines from south-west France (Geffroy et al., 2014). Contradictory outcomes were reported in the north-east USA on Noiret, where fruit exposure either had no effect or even enhanced rotundone levels (Homich et al., 2017). Similar results were observed in Fer cultivated under temperate conditions with very cool nights (Geffroy et al., 2019a), highlighting the complexity of environmental interactions.

Other research works suggest a stimulatory role of light and UV radiation, as correlations between rotundone concentration and indices of solar radiation, including mean irradiation and sunshine duration during ripening, have been described (Geffroy et al., 2019b). Although no controlled studies have specifically examined rotundone, evidence suggests that UV exposure can influence its biosynthesis. For example, UV treatment of grapevine leaves has been shown to enhance the production of the sesquiterpene (E)-nerolidol (Gil et al., 2012). From a mechanistical point of view, UV is perceived by the UVR8 photoreceptor, which activates signalling cascades involving reactive oxygen species (ROS) and defence-related phytohormones such as jasmonic acid, salicylic acid, and ethylene (Liu et al., 2015). In the case of rotundone, an oxygenated derivative of its precursor α-guaiene, it remains possible that ROS contribute directly by enhancing the oxidation reaction (Huang et al., 2014).

This field research aimed to investigate in greater depth the effects of different timings of defoliation, as well as the influence of UV exposure, either reduced by UV screens following a previously published approach (Šuklje et al., 2014), or enhanced through external UV-C treatments. Although the trial was initially planned over three consecutive seasons (2021–2023), only data from the 2022 vintage were retained for analysis due to severe climatic and phytosanitary constraints in 2021 and 2023, respectively.

Material and methods

1. Experimental site and design

The trial was conducted in 2022 using a complete randomised block design on a plot of Vitis vinifera L. Tardif (PDO Saint-Mont), an almost extinct genotype that is regaining popularity and is known for its high rotundone production (Baerenzung dit Baron et al., 2025; Geffroy et al., 2025). The 0.4 ha vineyard (lat. 43° 38' 11.0" N long. 0° 7' 25.6" W), typical of the area with 2.20 m × 1 m vine spacing, was planted in 2018. Orientation of the rows was north-west to south-east, and vines were trained with vertical shoot positioning on a single Guyot pruning system. The soil was managed by chemical weed control under the vines and by grass cover or mechanical weeding in every second inter-row. Each treatment was replicated three times on experimental units made up of 12 adjacent vines. It should be emphasised that this study was conducted as part of a three-year collaborative project with a cooperative cellar, which began in early 2021 and concluded in late 2023. For the 2021 and 2023 seasons, reliable data collection was not possible. This was due to an extreme spring frost in April 2021 and severe downy mildew (Plasmopara viticola) damage in 2023, both of which reduced the available plant material and introduced heterogeneity, preventing reliable data collection. During these two seasons, relocating the trial to an undamaged vineyard was not feasible, as other Tardif vineyards planted after 2020 were still too young for experimentation.

2. Treatments

Four different treatments were investigated and compared to a control treatment without intervention (Control) (Figure 1):

• Early defoliation (ED): elimination of leaves in the fruit-bearing zone up to the 3rd node on the side of least exposure (i.e., north-east), on June 22, which corresponds to 22 days after mid-flowering (Eichhorn & Lorenz 32).
• Late defoliation (LD): elimination of leaves in the fruit-bearing zone up to the 3rd node on the side of least exposure (i.e., north-east), on July 26, which corresponds to 13 days before mid-veraison (Eichhorn & Lorenz 34).
• Late defoliation combined with UV-screens (LD-UV): after late defoliation performed as per the LD treatment, a UV radiation-shielding sheet filter blocking 99 % of UV-A and UV-B radiation (Altuglas, La Garenne-Colombes, France), hereafter referred to as UV screens, was installed in the bunch area on the north-east side of the row from August 8, which corresponds to mid-veraison, until harvest.
• Late defoliation combined to UV treatments (LD+UV): after late defoliation performed as for the LD treatment, four UV-C modulated light treatments were applied for 1 min on the north-east side of the row at a distance of 10 cm from the fruits on August 8 (mid-veraison), August 19, August 30, and September 12 using the Boxilumix™ patented device (Asclepios Tech, Tournefeuille, France) (Roynette & Roynette, 2024). The treatments were conducted at 11 am for practical reasons, as the experimental site was located a two-hour drive from our laboratory. With a wavelength ranging from 200 to 280 nm, UV-C is more energetic than UV-A and UV-B and is usually blocked from reaching the lower atmosphere by the ozone layer (Paulo et al., 2019). The detailed settings (i.e., wavelength, frequency, and modulation time), which cannot be divulged for confidentiality reasons, were obtained based on previous experiments conducted with grapevine cell cultures, highlighting a 10-fold increase in several sesquiterpenes (i.e., bergamotene, β-caryophyllene, and α-humulene) using the protocol proposed by Diot et al. (2023). A key advantage of the system lies in its use of modulated light-emitting diodes (LEDs) in place of conventional mercury vapour lamps. This approach significantly reduces thermal load and enables precise, real-time control of irradiance and energy delivery at ultralow intensities (e.g., modulated <100 W⋅m⁻² vs continuous ~10,000 W⋅m⁻² for mercury lamps). Such modulation in a sustainable way minimises the risk of photothermal and phytotoxic effects, thereby allowing for safer and more controlled exposure conditions in plant physiological experiments (Roynette & Roynette, 2024).

3. Phenological data and weather measurements

Phenological data were collected across the entire experimental site and from the reproductive organs on both sides of the row. The percentage of flowering was determined by estimating the percentage of opened flowers per inflorescence for 100 inflorescences. The percentage of veraison was determined by randomly collecting 100 berries from several parts of the bunch and by counting the number of soft berries. Linear regression tests were performed using Excel software (Microsoft Corp., Redmond, WA, USA) to calculate mid-flowering and mid-veraison dates.

To characterise the season accurately from a climatic point of view, daily high-resolution data provided by Météo France (Toulouse, France), the French weather service, were obtained for the experimental site using ANTILOPE (Champeaux et al., 2011) for rainfall and AROME (Brousseau et al., 2011) for air temperature (minima, maxima, and mean values) for the 2011–2025 period. The data were used to calculate the Huglin index, and cumulative rainfall between April 1 and September 30, and between January 1 and December 31 (Tonietto & Carbonneau, 2002).

As radiation-shielding sheets are likely to impact bunch microclimate, EBI 20-T1 temperature loggers (Ebro, Ingolstadt, Germany) displayed in solar radiation shields were positioned in the bunch area, on removable stakes at a height of 1 m above the ground, on the north-east side of the row, to monitor air temperature during ripening on the control, LD, and LD-UV treatments. For this latter treatment, loggers were positioned at 20 cm from UV shields. Such loggers were not used for the two other treatments, as the temperature is expected to be similar to that of LD for LD+UV, and intermediate between the control and LD for ED, as a consequence of foliage renewal for this early stage defoliation (Homich et al., 2017). The percentage of hours when the temperature exceeds 25 °C (Dh25), an indicator anti-correlated to the level of rotundone in grapes (Zhang et al., 2015), was calculated.

4. Berry sampling, basic fruit analysis, and yield components

On October 3, at harvest time, 400 berries were sampled for each experimental unit from the side of the row exposed to the treatments. A cooler with ice packs was used to transport the samples from the vineyard to the laboratory. Half of the berries were weighed, gently crushed, and the resulting juice was centrifuged for basic fruit analyses. Sugar concentration (°Brix) was determined with a digital hand-held Pocket refractometer PAL (Atago, Japan). Titratable acidity was measured according to the OIV method (OIV, 2009) and pH measured with a HI 3221 pH meter (Hanna Instruments, France). A Konelab Arena 20 sequential analyser (Thermo Electron Corporation, USA) associated with enzyme kits provided by several suppliers was used to determine amino acids, ammonium (Megazyme, Ireland), and malic acid (Thermo Fisher Scientific, USA). Potassium determination was carried out by flame photometry (Bio Arrow, France) according to the OIV method (OIV, 2009) and tartaric acid determination by colorimetric titration (Hill & Caputi, 1970). Anthocyanins and Total Phenolic Index (TPI) were quantified in grapes according to Cayla et al. (2002) using an Evolution 100 spectrophotometer (Thermo Electron Corporation, USA). This analytical method for phenolic determination in grapes is based on a 1 h micromaceration in a hydroalcoholic and acidic medium. δ¹³C, a parameter reflecting the level of water deficit experienced by the plant during maturation, was determined according to a previously published protocol (Coplen, 2011). All determinations were carried out in duplicate. The remaining berries were frozen at –18 °C for rotundone analysis, which was performed without technical replication.

The number of clusters per vine was determined, and crop load was measured individually with a Precia Molen C20 K balance (Precia SA, Privas, France).

5. Rotundone analysis

Rotundone analysis was conducted using the protocol proposed by Baerenzung dit Baron et al. (2025) on an UltraTRACE gas chromatograph equipped with a split/splitless injector and coupled to an ITQ 900 ion trap mass spectrometer (Thermo Scientific, Courtaboeuf, France). A TriPlus autosampler (Thermo Scientific, Courtabœuf, France) was used for sample introduction. Compound separation was achieved using an HP-5ms column (60 m × 0.25 mm ID × 0.25 µm film thickness; Agilent, Santa Clara, CA, USA). This protocol can be summarised here briefly.

The still-frozen 200-berry samples were resampled into 120 g batches, thawed, and homogenised using a GRINDOMIX GM 200 knife mill (Retsch, Eragny, France). After centrifugation, juice was collected, and solids were extracted with ethanol/water (1:1) spiked with d₅-rotundone with purity of 95 % provided by Eptes (Vevey, Switzerland) as an internal standard. The extract was shaken for 24 h, sonicated, centrifuged, combined with the juice, filtered (0.45 μm), and diluted to 200 mL with MilliQ water before solid phase extraction (SPE). SPE cartridges (SDB-L, 500 mg/6 mL) provided by Phenomenex (Torrance, CA, USA) were conditioned and loaded with the extract, washed, and eluted with n-pentane/ethyl acetate (9:1). The eluate was evaporated under nitrogen, reconstituted in ethanol, and diluted with Milli-Q water. Analysis was performed by direct immersion solid-phase microextraction coupled with gas chromatography–mass spectrometry (DI-SPME-GC-MS) using a PDMS/DVB fiber (Sigma-Aldrich, Darmstadt, Germany), with incubation at 40 °C, 60 min immersion, and desorption at 250 °C. Mass spectrometry was conducted in EI+ mode (70 eV) with full scan acquisition (120–230 amu). Rotundone and d5-rotundone were identified by comparison of retention time (respectively 65.60 ± 0.10 min and 65.42 ± 0.10 min) and mass spectra with standard data. Identification was confirmed by ion ratios (m/z 218/203 and 223/208). Calibration curves (20–980 ng/kg) were linear (y = 0.7591x + 0.0105, R² = 0.9995) with an LOQ of 20 ng/kg (S/N = 10) and LOD of 7 ng/kg (S/N = 3). Repeatability tests at 5×, 25×, and 50× LOQ yielded RSDs of 9 %, 4 %, and 3 %, while recovery ranged from 98–107 %.

6. Additional experiment in growth chamber

To obtain preliminary insights on how the time of the day (daytime vs nighttime) could affect the efficiency of the UV treatments, an additional experiment was conducted in growth chamber conditions. Twenty-five fruiting cuttings of Vitis vinifera L. Syrah, another rotundone-producing genotype were used to run this complementary experiment. Such cuttings produced using a previously published protocol (Ollat et al., 1998) were maintained from mid-veraison at 20 °C and 40 % relative humidity, supplied with four irrigations of 25 mL of water every 24 h, and subjected to a night/day cycle of 8 h/16 h with a day light intensity of 250 ± 50 µmol/m²/s. Thirteen plants were used as controls and received no treatment, while twelve plants received three UV-C treatments at the beginning of the night (mid-veraison + 2 days, mid-veraison + 12 days, mid-veraison + 22 days) using the same equipment and conditions as for the field trial. Fifty days after mid-veraison, the bunches were harvested, and the berries destemmed. For each treatment, the berries were pooled, divided into three batches of 110 g, and frozen for subsequent rotundone extraction and quantification, while the remaining fruit was used for basic fruit analyses. Due to the small quantity of material available, basic fruit analyses were limited to sugar concentration using the method previously described. However, rotundone was quantified in extracts obtained following the previously described protocol (Baerenzung dit Baron et al., 2025) by stirbar sorptive extraction (SBSE) combined with multidimensional gas chromatography-mass spectrometry (GC-GC-MS) to get a higher sensitivity (Wen et al., 2018) than the previous approach, which had a LOQ of 20 ng/kg. Indeed, lower sesquiterpene (SQT) levels and thus rotundone concentrations were expected in controlled conditions, for which grapes are less exposed to abiotic and biotic stressors that can enhance SQT biosynthesis (Schwab & Wüst, 2015). For the SBSE-GC-GC-MS analysis, a stirbar coated with 126 μL polydimethylsiloxane (20 mm length × 1.0 mm thickness; Gerstel, Mülheim an der Ruhr, Germany) was added to a 25 mL Erlenmeyer flask containing 5 mL of berry extract and 50 μL of internal standard solution. The mixture was stirred at 750 rpm for 30 min at room temperature. Following extraction, the stirbar was removed from the flask, rinsed briefly with Milli-Q water and dried with a lint-free tissue. Each stirbar was then transferred into a thermal desorption tube and placed in the autosampler tray. Thermal desorption was conducted in splitless mode at 270 °C for 7 min. The multidimensional chromatographic system comprised a first-dimension DB-5MS column (J&W Scientific, Folsom, CA, USA) coupled to a second-dimension SAPIENS-WAX MS column (Teknokroma, Barcelona, Spain) via a Deans switch (Agilent Technologies, Santa Clara, CA, USA). During the elution of rotundone and its corresponding deuterated standard from the first column, the Deans switch was programmed to divert the flow to the second column to enhance separation. A quadrupole mass detector was connected to the second column and operated using the parameters previously published.

7. Statistical processing

Statistical processing was performed using XLSTATS software (Addinsoft, Paris). Data were treated through a two-factor analysis of variance (ANOVA) (treatment × block) followed by a Fisher’s test. The interaction factor was included in the model only for variables with replicate measurements for each experimental unit (i.e., number of bunches per vine and crop load per vine). Data obtained from the additional experiment conducted under growth chamber conditions were subjected to a Student’s t-test.

Results and discussion

At the experimental site, the average bioclimatic data for the 2011–2025 period (Figure 2) were 2,108 ± 161 for the Huglin index, 399 ± 104 mm for cumulative rainfall from April 1 to September 30, and 899 ± 256 mm for annual cumulative rainfall (January 1 to December 31). According to the Huglin index, 2022 was the warmest season recorded since 2011. It was also the third driest year for both the April–September and the full-year periods.

Data loggers were not able to detect modifications in mean hourly bunch zone air temperature among the three monitored treatments (Figure 3). Similarly, for Dh25, no statistically significant differences were observed at P < 0.05 between the control (33.08 % ± 0.83), LD (32.81 % ± 0.56), and LD-UV (33.08 % ± 0.49) treatments. This finding is particularly important, as it demonstrates that UV suppression can be assessed without a concomitant increase in bunch zone air temperature. The mean hourly temperature data also illustrate the particularly warm conditions experienced during maturation on the experimental site, with mean temperature exceeding 30 °C between 1 pm and 5 pm. Although, to our knowledge, no studies have been conducted in the southwest of France to determine Dh25 using data loggers, research from Australia did not report values for Syrah above 15 % (Zhang et al., 2015), indicating that low rotundone concentrations are expected in grapes collected at our experimental site.

Sunburn damage was not observed on the side of the row subjected to the treatments. The number of bunches and crop load per vine did not differ among treatments (Table 1), which illustrates the homogeneity of our experimental vineyard. Indeed, although early defoliation at flowering or before flowering has been shown to impact berry set and yield through source limitation, no impact of the studied techniques, which were applied much later, was expected on yield parameters (Verdenal et al., 2019).

Among the measured parameters, pH and the content of potassium were the only variables significantly impacted by the treatments. For ED, LD, and LD+UV, lower berry potassium content was observed, which could result from the reduced leaf area limiting the translocation of assimilates and minerals from vegetative organs to the fruits (Mpelasoka et al., 2003). Logically, this resulted in a decrease in pH for ED and LD+UV treatments, as potassium is known to influence the acid balance in the must (Mpelasoka et al., 2003). For unknown reasons, this decrease was not observed for LD. It must be outlined that the potassium content was not affected for LD-UV in comparison to the control, which suggests that UV exclusion mitigated stress and stabilised berry metabolism (Mpelasoka et al., 2003), allowing potassium levels to be maintained despite reduced leaf area. No effect of leaf removal on berry sugar concentration was observed, likely because the removed leaves were older and photosynthetically less active, or due to compensatory mechanisms, as reported in previous studies (Geffroy et al., 2019a; Verdenal et al., 2019). It is worth noting that a lower level of water constraint, as reflected by lower δ¹³C values, could have been expected for the defoliated treatments due to reduced plant transpiration, but δ¹³C values showed no significant variation (Table 1).

Surprisingly, despite previous studies reporting in most cases higher phenolic content, notably anthocyanins in grapes or wines from defoliated vines (Verdenal et al., 2019), the treatments had no significant effect on anthocyanins or TPI, although a trend toward an increase was observed with greater UV exposure (Table 1). It cannot be discarded that the very high temperature experienced on the experimental site may have disrupted anthocyanin accumulation, whose biosynthesis is negatively affected by temperature above 30 °C in berry skin (Yamane et al., 2006). It should also be noted that berry weight was not affected by the treatments, indicating that variations in skin-to-juice ratio are not likely to influence rotundone concentration in berries, considering that this compound is mainly found in the grape exocarp skin (Caputi et al., 2011). This allows for a direct comparison of the effects of the treatments on rotundone biosynthesis.

As expected under such extreme climatic conditions, rotundone concentrations were low (Figure 4) compared to previous studies conducted on high rotundone producers for which levels up to 3,450 ng/kg on Noiret and 6,130 ng/kg on Vespolina have been observed (Geffroy et al., 2020). Data are scarcer for Tardif, with a maximal reported level of about 600 ng/kg in grapes (Baerenzung dit Baron et al., 2025). Nevertheless, the treatments had a significant effect on rotundone concentration (P = 0.027). Its content was similar to that of the control for ED (Figure 3), whereas for LD, it was significantly greater (+33 %). These findings align with previous studies highlighting the advantages of late defoliation at veraison or treatments that sustain fruit-zone sunlight exposure, compared to early defoliation, where regrowth of leaves may negate the initial increase in cluster exposure (Homich et al., 2017).

In comparison with LD, the rotundone concentration observed for LD-UV was significantly lower (–33 %) (Figure 4), indicating that, given the absence of previously described temperature differences between the two treatments, UV-A and UV-B exclusion negatively impacts rotundone biosynthesis. In contrast, the UV-C treatments (LD+UV) did not lead to an increase in rotundone.

As previously discussed, the mechanisms underlying the effects of UV on rotundone remain unclear but may involve either a direct stimulation of its biosynthetic pathway or an overproduction of reactive oxygen species (ROS), which could oxidise α-guaiene, the precursor of rotundone, into rotundone (Huang et al., 2014). The UV-C treatments were applied on our experimental site during daytime, while it has been proven that at night, UV-C induces higher ROS accumulation mainly because antioxidant defences are reduced, photorepair is suppressed, and metabolism is slower, allowing ROS to persist longer (Urban et al., 2016). It cannot be excluded that a stimulating effect would have been observed if the treatments were conducted at night.

In the same way, during this warm and dry vintage, it can be hypothesised that the excessive amount of UV received by the vines, particularly for the LD+UV treatment, may have had detrimental effects such as extreme oxidative stress and potential damage to plant tissue (Matsuura et al., 2013).

Indeed, under controlled growth chamber conditions, the effect induced by the UV-C treatment differed from that observed under field conditions. Although the UV-C treatments induced some light stress symptoms on leaves surrounding the bunches, there was no significant difference in sugar concentration between the two treatments (16.6 °Brix ± 1.9 for the control vs 15.7 °Brix ± 0.8 for the UV treatments, P = 0.477). However, berries treated with the UV-C device exhibited a significantly higher rotundone content (176.2 ng/kg ± 105.4) compared with the control (7.1 ng/kg ± 3.5) at P < 0.05.

Although derived from a limited number of fruiting cuttings and pending validation under field conditions, these preliminary results suggest a positive effect of UV-C modulated light treatments on rotundone biosynthesis when applied nocturnally.

Conclusion

This study enabled the investigation of the impact of two timings of defoliation performed on the side of least exposure on rotundone in Vitis vinifera L. Tardif. Although the early defoliation treatment had no impact on rotundone, the late treatment appears to be a suitable practice for winegrowers to enhance rotundone biosynthesis. It should be noted that different results might have been obtained if regrowth had been removed and the fruiting area had remained exposed during the whole season. In the same way, it can be acknowledged that analyses were conducted only on grapes collected from the treated row. The observed effect was moderate and would likely be even smaller under industrial conditions, where grapes are harvested from both sides of the row.

Excluding UV-A and UV-B exposure led to a reduction in rotundone concentration, confirming the role of this abiotic factor in rotundone biosynthesis. The UV-C treatments during daytime, under field trial conditions, did not increase rotundone levels, whereas a similar treatment applied at night on a limited number of fruiting cuttings produced in the growth chamber resulted in a substantial rotundone increase. This discrepancy, which may be linked to decreased photorepair when UV-C treatments are applied at night, highlights the need for further investigation to elucidate the underlying mechanisms, to validate the observed effects across a broader range of grapevines and environmental conditions before UV-C can be considered a viable strategy to enhance rotundone in grapes.

Acknowledgements

The authors would like to thank the Occitanie Region for funding the Pepper your Wine project through the Recherche et Société(s) program and for the doctoral thesis grant of Thomas Baerenzung dit Baron. We are also grateful to Maria Angélica Usta and Marie Homs from École d’Ingénieurs de Purpan, to Laure Gontier from IFV, as well as to Alain Deloire, formerly of Institut Agro Montpellier, and Nathalie Ollat from EGFV, for their technical support and practical advice.