Combined effects of ozone and drought stress on the emission of biogenic volatile organic compounds from Quercus robur L

Drought events are expected to become more frequent with climate change. To predict the effect of plant emissions on air-quality and potential feedback effects on climate, the study of biogenic volatile organic compound emissions under stress is of great importance. Trees can often be subject to a combination of abiotic stresses, for example 15 due to drought or ozone. Even though there is a large body of knowledge on individual stress factors, the effects of combined stressors are not much explored. This study aimed to investigate changes of biogenic volatile organic compound emissions and physiological parameters in Quercus robur L. during moderate to severe drought in combination with ozone stress. Results show that isoprene emissions decreased while monoterpene and sesquiterpene emissions increased during the progression of drought. Exposing plants additionally to ozone, resulted in faster stomatal closure partially mitigating drought 20 stress effects. Evidence of this was found in enhanced green leaf volatiles in trees without ozone fumigation indicating cellular damage. In addition we observed an enhancement in Methyl Salicylate emissions in trees with ozone treatment. Individual plant stress factors are not necessarily additive and atmospheric models should implement stress feedback loops to study regional scale effects.

Few studies have analyzed the effects of plant emissions from a combination of drought and ozone stress (Vitale et al., 2008;Yuan et al., 2016). Studying Quercus ilex, Vitale et al. (2008) reported that drought stress leads to stomatal closure therefore reducing stress by ozone as it is restricted to enter the leaf. Yuan et al. (2016) found that drought increased isoprene emissions in a hybrid poplar deltoid species, but that isoprene emissions decreased under moderate drought stress combined 85 with long-term ozone fumigation.
In this work our hypothesis was that different abiotic stresses in plants are not necessarily additive and that the plant's response to drought and ozone exposure can result in an alteration of characteristic BVOC emission strengths. As a model plant we chose Quercus robur L., a widely distributed isoprene emitting oak species in Europe (Barstow and Khela, 2017).

Plant species and stress treatments
Q. robur is a broad-leaf tree species widely distributed in Europe growing in mixed and deciduous forest ranging from sea level up to 1200 m ASL (Ülker et al., 2018). According to Ellenberg (1988), Q. robur has a high tolerance to drought due to its fast-regulated transpiration rates and stomatal conductance, and low susceptibility of water embolism in the xylem (Van Hees, 1997). 95 Fourteen 2-year-old Q. robur seedlings were planted in 7 L pots in March 2019. The substrate consisted to one-thirds of soil used by the city gardeners for city trees in Vienna and two-thirds of quartz sand to improve drainage. The plants were fertilized once after planting (universal fertilizer NovaTec, Compo, Münster, Germany) and from thereon kept well-watered in a greenhouse at near ambient light and slightly increased temperature conditions (Tulln, Austria). At the beginning of July, two weeks before the measurement campaign, the trees were moved to Vienna into another close-by greenhouse. Dust 100 was removed from the leaves by showering the trees before starting the drought stress. For the biochemical assays eight trees were used as reference, four well-watered plants (C) and four well-watered plants receiving one time 100 ppb ozone for one hour (OS) inside the enclosures.
For six plants the drought stress was initiated 10 days before the VOC measurements started and was maintained by keeping the soil water content at 4-5 vol.% using a soil moisture probe (Fieldscout TDR100, 20 cm probe depth, Spectrum 105 Technologies, UK), whereas 100 % field capacity was 13.4 vol.%. Approximately 24 hours before the air gas exchange and VOC measurements, the plants were moved from the greenhouse to an indoor climate chamber (Fitotron Weiss Gallenkamp, UK) which was kept at 25°C, ~60 % rH and 300 µmol m -² s -1 PAR at mid canopy height, to adapt to constant air temperature.
To study the effect of ozone exposure of trees during increasing drought, the six trees above mentioned, were separated into 110 two groups, three trees were drought stressed and fumigated with 100 ppb O 3 (DS×OS) inside the enclosure for one hour each day after the measurement of BVOCs. The other three trees were drought stressed but not fumigated with ozone (DS).
To continuously increase the drought stress, the plants were not watered and the humidity in the climate chamber was decreased to 40 % after the first day of measurements and temperature was increased to 30°C. At the end of the experiment leaves were harvested for leaf area and enzyme analysis. Values of the enzymatic activity of C and OS were compared to 115 DS and DS ×OS to investigate the effect of ozone fumigation.

Measurement of leaf gas exchange and BVOC fluxes
Throughout the increasing drought stress, tree leaf gas exchange and BVOC emissions were measured for two sets, DS and DS×OS, over a seven-day period, one in the morning and one in the afternoon alternating daily. The plants were taken out of 120 the climate chamber and kept inside custom-made plant enclosures (Appendix A Figure A1; TC-400, Vienna Scientific https://doi.org/10.5194/bg-2020-260 Preprint. Discussion started: 6 August 2020 c Author(s) 2020. CC BY 4.0 License.
Instruments GmbH, Alland, Austria) for 2-3 hours each day in order to measure their gas exchange along with key physiological parameters (soil moisture and stem water potential). The plant enclosures covered most of the plant material excluding a few leaves (about 7 on each tree) to allow determination of stem water potential (SWP). Each day, one leaf was wrapped in aluminum foil and placed in a plastic bag for equilibrating to SWP (Williams and Araujo, 2002). After darkening 125 for 30 minutes the leaf was cut off and SWP was measured by using a Scholander pressure bomb (Soil moisture Equipment Corp., Goleta, CA, USA).
The four custom-made plant enclosures (12 liters) were lined with PTFE and sealed on top with 55×60 cm PET-bags. The plant enclosures were continuously flushed with 10 l min -1 of ambient outside air that was previously passed through a cold trap to remove water and an activated carbon filter (360 m³ h -1 , PrimaKlima Trading, Radnice, CZ) to remove VOCs and O 3 . 130 This resulted in 32 % rH air and ~370 ppm CO 2 entering the enclosures (experimental conditions in Appendix A, Table A2).
The flow rate of 10 l min -1 assured that no condensation of water occurred in the tubing and enclosures, as well as resulted in a slight overpressure preventing the entry of room air into the enclosures. Three of the enclosures were used to measure the air gas exchange of the plants and the fourth enclosure was kept empty as a reference to allow continuous monitoring of the air entering the enclosures. Trees inside the enclosure were LED-irradiated with 1450 µmol m -2 s -1 at the mid-point of the 135 enclosure (Eckel Electronics, Trofaiach, Austria). Leaf temperature was monitored in each enclosure by placing a calibrated (±0.1°C) thermocouple (type k, PTFE IEC wire; Labfacility Ltd, Bognor Regis, West Sussex, UK) on the abaxial side of a mature mid-canopy leaf. An automated valve system allowed the consecutive analysis of air exiting each enclosure for 5 minutes each, leading to a 20 minutes cycle through the four enclosures. Before inserting the three trees into the enclosures, background measurements of 140 the empty enclosures were carried out. After inserting each plant into one enclosure, the plant was allowed to acclimatize for approximately two hours and the following 40-60 minutes of data was analyzed to determine plant CO 2 assimilation, transpiration and BVOC emissions rates. After the measurements, the trees of DS×OS were fumigated for one hour with 100 ppb of ozone each day. CO 2 and H 2 O mixing ratios in the air leaving the enclosures were measured using a CIRAS-3 SC PP System (Amesbury, 145 MA, USA). Ozone measurements before and after the enclosures were conducted continuously in all enclosures with an ozone monitor (six channel ozone monitor BMT 932, BMT Messtechnik, Berlin, Germany). BVOC measurements were made using a proton transfer reaction time of flight mass spectrometer (PTR-Tof-MS, PTR-TOF6000X2, IONICON Analytik GmbH, Innsbruck, Austria; Graus et al., 2010) operated at 350 V drift voltage, ion funnel settings of 1 MHz and 35V amplitude as well as 35 VDC, and 2.5 mbar drift pressure. These settings are comparable to an E/N of 100 Td in a PTR-150 TOF8000 with no ion funnel (Markus Müller, IONICON Analytic GmbH, personal communication 2019). The drift tube temperature was 100°C. Full PTR-Tof-MS mass spectra were collected with a time resolution of 1 s and up to a mass to charge ratio m/z 547 amu.
Dynamic calibrations of VOCs using a standard gas mixture (Apel Riemer Environmental Inc., Broomfield, CO, USA), containing 15 compounds with different functionality distributed over a mass range of 33-137 amu were performed daily. 155 Compound specific sensitivities varied on the order of 8-20 % for different calibration compounds, which lies within the combined calibration uncertainties; a compound specific average experiment sensitivity was applied to all data. Limits of detection were compound dependent and on the order of 40-800 pptv for a 60 s average period. Humidity dependent calibrations were performed to account for varying humidity in the enclosures depending on plant size and plant transpiration. The PTR-Tof-MS data was analyzed using the PTRTOF Data Analyzer v4 software (Müller et al., 2013) and   (Beauchamp et al., 2005;165 Giacomuzzi et al., 2016;Portillo-Estrada et al., 2017). Shikimate BVOCs were represented by benzene m/z 79.054, phenol m/z 95.050, methyl salicylate m/z 153.055 (MeSa) and eugenol m/z 165.092 (Brilli et al., 2011;Tasin et al., 2012;Maja et al., 2014;Brilli et al., 2016;Giacomuzzi et al., 2016;Portillo-Estrada et al., 2017;Yener et al., 2016;Misztal et al., 2015).
Mass flow of air W, transpiration rate E, net photosynthesis A and stomatal conductance g S were calculated accordingly where V 0 is the volume air flow, a is the leaf area, e in is the partial water vapor pressure of the air entering the enclosures, e out is the partial water vapor pressure inside the enclosure, P is the atmospheric pressure, C in concentration of CO 2 entering and C out exiting the enclosure, e leaf is the saturation vapor pressure at leaf temperature (T leaf ), r s is the stomatal resistance and r b is the boundary layer resistance to water vapor transfer, which was assumed zero according to the recommendations of the 190 manufacturer (CIRAS-3 Operation Manual V. 2-01, PP-Systems, Amesbury, MA, USA).
The ratio of the sum of carbon lost in form of BVOC (C BVOCs ) vs. the uptake of carbon from net photosynthesis (C A ) was calculated according to Pegoraro et al. (2004), with the BVOCs used to calculate C BVOCs given in Table A3.
After seven days, finishing the emission measurements, all leaves were harvested immediately, imaged with a flatbed scanner (Epson Expression 10000XL, Epson, Japan) and analyzed with the PC program WinFOLIA 2013 Pro (Regent 195 Instruments Inc., Qúebec, Canada) to determine the leave surface area. About 80 % of the leaves' fresh mass was shockfrozen and crushed in liquid nitrogen for biochemical assays (see below). About 20 % of the leaves per plant were dried for three days in a drying room at 40°C to determine dry weight to an accuracy of ±0.001 g for the calculation of enzyme activity and specific leaf area (SLA) ( Table A4). https://doi.org/10.5194/bg-2020-260 Preprint. Discussion started: 6 August 2020 c Author(s) 2020. CC BY 4.0 License.

Biochemical assay 200
For the interpretation of the emissions of GLVs and Shikimate volatiles, enzymatic activities were analyzed additionally to better understand the effect of ozone fumigation during a situation of severe drought. Using foliar materials collected after the seven day period of emission measurements (see above) and stored at -80°C until analysis, peroxidase and antioxidant capacity, and phenol content (TPhe) were measured. Values from plants after seven days of increasing drought (DS×OS, DS) were compared to well-watered reference plants (C) and a well-watered set of plants that received ozone fumigation once 205 For measurements of peroxidase activities, 0.5 g plant material, 0.25 g Polyclar AT (Serva Electrophoresis, Heidelberg, Germany) and 0.25 g quartz sand (Sigma-Aldrich, Steinheim, Germany), were homogenized in a mortar with 3 ml 0.1 M potassium phosphate buffer (pH 6.0). After removal of solid compounds by centrifugation at 4°C and 10000 × g for 10 minutes, 400 µL of the supernatant were subjected to gel chromatography with Sephadex G25 medium (GE Healthcare, 210 Chicago, IL, USA) to remove low molecular weight compounds. Peroxidase activity was determined according to the Worthington Manual (1972). Briefly, the enzyme assay contained in a final volume of 1110 µL, 1095 µL buffer 0.1 M potassium phosphate buffer + 0.003 % (v/v) H 2 O 2 (pH 6.0), 5 µL enzyme preparation, and 10 µL 1 % The activity was determined by measuring the extinction at 460 nm on a DU-65 spectrophotometer (Beckman Instruments, 215 Brea, CA, USA) in intervals of 30 s for a period of 6 minutes. The activity was calculated for the linear range of ΔE min -1 , using an extinction coefficient of 1.13 × 10 4 M -1 cm -1 for oxidized o-dianisidine (Worthington manual, 1972). The protein content was determined by a modified Lowry procedure (Sandermann and Strominger, 1972) using bovine serum albumin as a standard. All measurements were performed in two technical replicates.
For the determination of the antioxidant capacity and the TPhen, the material was lyophilised and homogenized by grinding 220 to fine powder in a mortar. 0.25 g of the lyophilised powder was extracted with 3 ml distilled water for 1 hour in a cooled water bath during sonication. After centrifugation for 5 minutes at 4°C and 10000 × g, the supernatant was filtered through a Chromafil AO-20/25 polyamide filter (Roth, Karlsruhe, Germany).
The TPhen was determined as described (Wootton-Beard et al., 2011) with some modifications. Briefly, 100 µL of the aqueous solution was mixed with 6 mL distilled water and 500 µL Folin Ciocalteu Reagent (Sigma-Aldrich, Vienna, 225 Austria) (1:1 v/v with distilled water). After equilibration for 8 minutes, 1.5 ml 20 % Na 2 CO 3 (w/v) and 1.9 ml distilled water were added and the mixture was incubated at 40°C for 30 minutes. The TPhen was obtained by measuring the absorbance of the mixture at 765 nm using a freshly prepared standard curve obtained with gallic acid. The results were expressed as µg gallic acid equivalents per g sample. All measurements were performed in technical triplicates.
The in vivo antioxidant activity was determined with Saccharomyces cerevisiae ZIM 2155 as model system following the 230 procedures described in Slatnar et al. (2012), which estimates intracellular oxidation by fluorometrical measurements using the ROS-sensitive dye 2',7'-dichlorofluorescin (H 2 DCF). 100 µl of the aqueous samples were incubated with 10 mL yeast suspension at their stationary phase in phosphate buffered saline (PBS, Merck KGaA, Darmstadt, Germany) at a density of 10 8 cells/suspension at 28°C and 220 rpm for 2 h. After a centrifugation step at room temperature for 5 minutes at 14000 × g, the pellet was washed three times with 50 mM potassium phosphate buffer (pH 7.8) and was finally resuspended in 9 235 volumes of 500 µL 50 mM potassium phosphate buffer (pH 7.8) and incubated for ten minutes at 28°C and 220 rpm in the dark. After addition of 10 µL H 2 DCF (1 mM stock solution in 96 % ethanol), the mixture was incubated for further 30 minutes at 28°C and 220 rpm. The fluorescence of the yeast cell suspensions was measured at a GloMax ® Multi Microplate Reader (Promega, Walldorf, Germany) using excitation and emission wavelengths of 490 and 520 nm, respectively. Values of fluorescence intensity were measured against a blank, in which the sample was replaced with water. Data are expressed as 240 relative fluorescence intensity, where the values obtained with the blank are defined as 1. Values lower than 1 indicate a https://doi.org/10.5194/bg-2020-260 Preprint. Discussion started: 6 August 2020 c Author(s) 2020. CC BY 4.0 License. higher antioxidant activity than the blank (Slatnar et al., 2012). All measurements were performed in two technical replicates.

Statistical analyses
Emission rates, physiological parameters, means and standard deviation were calculated with Matlab (MATLAB and 245 Statistics Toolbox Release 2017a, The MatWorks, Inc., Natick, MA, United States). All leaf gas exchange and BVOC flux measurements collected over the seven-day period for the set DS and DS×OS were aggregated into four ranges of SWP (R1: 0.00 to -1.40 MPa; R2: -1.45 to -2.85 MPa; R3: -2.90 to -4.30 MPa, R4: -4.35 to -6.00 MPa) to perform statistical analysis using the Wilcoxon rank sum test. To test for significant differences in the biochemical markers a one-way ANOVA test was used. For both tests p-values below 0.05 were considered significant. 250

Stomatal closure and net photosynthesis
SWP was measured daily and used as a drought stress indicator to study the evolution of Q. robur under continuously increasing drought condition. All six trees began the experiment with a high to moderate mean SWP of -0.9 MPa and reached low values in the order of -5.5 MPa after seven days of continuously increasing drought stress. Mean and standard 255 deviation of stomatal conductance (g S ), net photosynthesis (A), leaf temperature (T leaf ) and SWP as well as notes for statistically significant differences are summarized in Table 1 for the four drought stress ranges defined in 2.4. The mean stomatal conductance (g S ) of DS×OS was 20.2 mmol m -2 s -1 in R1 and decreased to 6.8 mmol m -2 s -1 in R2 (Table 1). For DS it was 42.4 mmol m -2 s -1 in R1 and decreased to 6.6 mmol m -2 s -1 in R2. For both sets the reduction of g S and SWP between R1 and R4 was significant. R1, shown in Fig. 1(a), includes values of trees fumigated with ozone (DS×OS) from the first and 260 the second day of analysis because SWP did not change much. Differently, R1 includes only measurements of the first day for DS. This shows that trees of DS×OS closed their stoma quickly at higher stem water potential after the first ozone fumigation session, and confirms what was reported in other studies that moderate ozone concentrations can induce partially closed stomata (Khatamian et al., 1973;Farage et al., 1991;Wittig et al., 2007). A partial stomatal closure prevented excessive water loss through stomatal openings (Pinheiro and Chaves, 2011;Mc Dowell et al., 2008;Allen et al., 2008) 265 during drought stress, and enhanced the closure with ozone allowing DS×OS plants to better survive the increased drought. Kobayashi et al. (1993) considers the interactive effects of O 3 and drought stress using a growth model of soybean, finding that ozone fumigation reduces or postpones drought stress, similar to the findings of this experiment. Figure 1 (b) shows a decrease of net photosynthesis (A) with the increase of the stress for both set, especially between R1 and R2, whereas the values in R3 and R4 are close to zero. In R1, A presented the same differences exposed for g S between 270 the sets. Our results are different from the finding of Tjoelker et al. (1995) and Paoletti (2005), where stomatal conductance and photosynthesis are shown to decouple at moderate ozone exposure due to direct damage to biochemical carboxylation, caused by chronic ozone exposure.
The ratio of sum of carbon lost in form of BVOC (C BVOCs ) and the uptake of carbon from net photosynthesis (C A ) is shown in Fig. 2. Initially, at low drought stress (R1), 3-7 % of the assimilated carbon was lost as emitted BVOC, which matches 275 findings in other studies Baldocchi et al., 1995;Monson and Fall, 1989;Fang et al., 1996), showing that ~2 % of carbon assimilated is lost as IS (C IS /C A ) under unstressed conditions and at 30°C. As CO 2 assimilation rate decreased quickly, and BVOC emission (especially isoprene emission) stayed elevated the ratio of lost vs. fixed carbon increased to 20 % for DS and 16 % for DS×OS in R2. Pegoraro et al. (2004) reported a carbon loss in the order of 50 % for SWP of -2 MPa, in a drought experiment with Quercus virginiana. In R3, the increasing stress corresponded to ratios of 0.7 280 and 1.03 for DS and DS×OS respectively. Alternative carbon sources for isoprene biosynthesis under drought stress are thus https://doi.org/10.5194/bg-2020-260 Preprint. Discussion started: 6 August 2020 c Author(s) 2020. CC BY 4.0 License.
proposed for DS×OS. For example, extra-chloroplastic origin or chloroplastic starch (Karl et al., 2002;Kreuzwieser et al., 2002;Funk et al., 2004;Affek and Yakir, 2003;Schnitzler et al., 2004;Rosenstiel et al., 2003) can sustain carbon sources for isoprene production. The increasing ratio was driven by the decline of fixed carbon with increasing SWP; for IS, the dominant BVOC (averagely 96 % of the total emissions), mean standardized IS emissions of DS×OS treated plants were 285 consistently higher in all SWP ranges compared to DS alone (Figure 3), thus showing the difference between DS and DS×OS in carbon loss ratio in the highest SWP ratio range.
At very high drought stress this ratio decreased again to 0.4 in DS and 0.8 in DS×OS.

BVOCs emissions
To give a general overview on BVOC emissions for both sets Fig. 4 (a) and (b) show the total mass spectra ranging from 40-290 220 amu for the first and last day of measurement for DS and DS×OS respectively. Figure 4 (c) shows relative changes of the mass spectra between the first and last day of measurements. The mass range 80-110 amu, hosting many mass to charge ratios associated with GLVs, showed the strongest difference between the two sets. Plants exposed to ozone and drought stress (DS×OS) exhibited smaller increases in this mass range compared to drought stressed (DS) plants. Changes in emissions or lack thereof for IS, MT, SQT and stress related BVOCs are investigated in further detail below and are 295 summarized in Table 2.

Isoprene emissions
Q. robur is generally classified as a high IS emitting (Benjamin and Winer, 1998;Lehning et al., 2002) and medium to low MT and SQT emitting species (Owen et al., 1997;Karl et al., 2009;Steinbrecher et al., 2009). Figure 3 shows standardized isoprene emissions (IS S ) as a function of drought stress for all investigated trees. In the range of SWP R1 the plants were in a 300 low-to no-water-stress condition (Brüggemann and Schnitzler, 2002). Whereas g S and A ( Fig. 1 (a),(b)) decreased rapidly with increasing drought stress and bottom out at -3 MPa, isoprene emissions decreased much slower reaching close to zero emissions at SWP -6 MPa. IS S in R1 was 12.8 nmol m -2 s -1 and 18.0 nmol m -2 s -1 for DS and DS×OS respectively. In R4 the mean IS S was 1.7 nmol m -2 s -1 for DS and 3.9 nmol m -2 s -1 for DS×OS 305 Given that IS S emissions remain higher in DS×OS for R1 and R2, compared to DS suggests that overall isoprene production within the leaves must have remained high in response to ozone. High IS fluxes due to ozone treatment are also reported in other studies (Fares et al., 2006;Velikova et al., 2005;Kanagendran et al., 2018).
An increase in IS with moderate stress was observed by Pegoraro et al. (2004) and Beckett et al. (2012), who related this finding to an increase in leaf temperatures as a consequence of stomatal closure. In contrast, leaf temperatures did not change 310 significantly in our study, suggesting IS emissions of DS×OS in R2 being a result of a temperature-independent isoprene production.
The decrease of A with decreasing SWP, particularly at mild drought stress (Fig 1b), is much more pronounced than the decrease of IS S emission rates (Fig 3). Similar results are found for leaf level measurements of Q. robur (Brüggemann and 315 Schnitzler, 2002), Populus alba (Brilli et al., 2007) and Quercus virginia (Pegoraro et al., 2004) as well as on the ecosystem scale in the Ozark region in the central U.S. (Seco et al., 2015).
Even though the rate of photosynthetic carbon assimilation declined much faster under drought than IS, a substantial decline of IS was also seen as drought progressed. Drought stress has been found to be one of the stronger influencing factors 320 affecting photosynthesis but had often only limited influence on IS emission rates Sharkey and Loreto, 1993;Fang et al., 1996). https://doi.org/10.5194/bg-2020-260 Preprint. Discussion started: 6 August 2020 c Author(s) 2020. CC BY 4.0 License.
In young hybrid poplars (Populus deltoides cv. 55/56 x P. deltoides cv. Imperial), the combined application of elevated ozone and drought decreases isoprene emission, whereas drought alone increases the emission, and ozone alone decreases it (Yuan et al., 2016). 325 Studies report that volatile isoprenoids strengthen cellular membranes, thus maintaining the integrity of the thylakoidembedded photosynthetic apparatus and have a generic antioxidant action by deactivating ROS around and inside leaves and thus indirectly reduce the oxidation of membrane structures and macromolecules (Singsaas et al., 1997;Loreto and Velikova, 2001;Affek and Yakir, 2002;Loreto and Schnitzler, 2010;Velikova et al., 2012). Figure 5 shows MT S (a) and SQT S (b) as a function of SWP. Mean MT S for DS and DS×OS were 1.0 × 10 -2 nmol m -2 s -1 and 3.6 × 10 -2 nmol m -2 s -1 respectively at R1. With the increase of drought stress (R3) DS×OS decreased to 1.5 × 10 -2 nmol m -2 s -1 while DS emissions remained stable (1.0 × 10 -2 nmol m -2 s -1 ). For higher drought stress (R4) both sets showed an increase in MT emissions reaching 3.3 × 10 -2 nmol m -2 s -1 for DS and 4.7 × 10 -2 nmol m -2 s -1 for DS×OS. Loreto et al. (2004), demonstrated that ozone can stimulate the emission of monoterpenes in Q. ilex, but that ozone has no 335 effect on photosynthesis nor on any other physiological parameter, when Mediterranean oak plants are exposed to mild and repeated, as well as acute ozone stress.

Terpenoid emissions 330
In this experiment MT emissions from Q. robur, increased in DS and DS×OS trees. In the case of DS, there was a positive effect of drought, with an increase in MT emissions, although there was a drastic decrease of IS emissions when the water deficit was severe. These observations contrast those by Lluisá and Peñuelas (1998) for Q. coccifera reporting a decrease of 340 MT emissions. This could be due to the fact that in the case of Q. coccifera no specific terpene storage structures are present in leaves, while they are present in Q. robur (Karl et al., 2009).
In both sets SQT S emissions remained close to zero down to a SWP of -3 MPa. SQT S emissions increase with increasing drought stress reaching a mean value of 1.4 × 10 -2 nmol m -2 s -1 for DS and 3.5 × 10 -2 nmol m -2 s -1 for DS×OS in R4. The 345 increase of SQT S in the set with ozone began one day later than in the set without ozone fumigation. (Toome et al., 2010;Maes and Debergh, 2003;Ibrahim et al., 2006). Ormeño et al. (2007) observe a reduction of sesquiterpenes with drought stress for a variety of plant species including Q. coccifera. For Q. robur we see an increase of SQT emissions under conditions of severe drought. 350

Stress on plants can induce SQT emissions
The release of SQT from leaves can be triggered when plants face stress due to oxidative processes in leaves, indicating that damaging effects inside the plants start to occur (Beauchamp et al., 2005;Bourtsoukidis et al., 2012). Unlike MT, SQT don't provide an additional barrier to plant damage during severe water stress (Palmer-Young et al., 2015). This is due to their different physico-chemical characteristics and the different pathways that produce them (Niinemets et al., 2004;Umlauf et al., 2004). In the case of SQT emissions, the parallel occurrence of two stresses (ozone and increased drought) generally led 355 to an increase in emissions. In fact, the higher SQT emissions in DS×OS compared to DS may have been due to ozone, similar to those reported in Beauchamp et al. (2005).

GLV and SHIKIMATE emissions
The ΣGLV increased for both sets in R4 (Fig. 6 (a)). Within ΣGLV m/z 99.080, attributable to hexenal isomers, showed the strongest increase in DS (mean value of m/z 99.080 in R4 was 68 % of the Σ GLV emission). Within the cascade of GLV 360 production, (E)-2-hexenal and (Z)-3-hexenal are typically the ones appearing first (Fall et al., 1999). DS×OS, on the other hand, showed an increase in Shikimate compounds (Fig. 6 (b)) at SWP < -3 MPa, DS showed similar but less pronounced trend. The ΣShikimate was dominated by Methyl Salicylate (MeSa) across the entire SWP range for DS https://doi.org/10.5194/bg-2020-260 Preprint. Discussion started: 6 August 2020 c Author(s) 2020. CC BY 4.0 License. and in R1-R3 for DS×OS. R4 of DS×OS was dominated by m/z 95.050 (matching the exact mass of protonated phenol, MeSa is considered as a volatile stress signaling molecule from plants (Karl et al., 2008). High emissions of MeSa are also found in the case of the tobacco plant (Nicotiana tabacum L. cultivars) in both O 3 sensitive and O 3 tolerant, exposed to ozone at high concentrations (Heiden et al., 1999;Beauchamp et al., 2005).
Observing the increase in GLV emissions in DS and Shikimate emissions in DS×OS was important to understand how ozone 370 affected the Q. robur trees exposed to drought stress. The impact of exposure to high ozone concentrations on ROS production was not significant and not associated with membrane lesions in Pellegrini et al. (2019). In this experiment, GLV emissions in R4 were low in ozone treated plants (DS×OS), while plants that were exposed to drought only (DS) exhibited higher emissions. The observations of this experiment can be interpreted such that plants did not suffer from detrimental effects due to acute ozone exposure yet (e.g. Beauchamp et al., 2005), but that mild ozone exposure can potentially delay 375 effects of drought stress and help maintain membrane structure and integrity.
The activation of an efficient free radical scavenging system can minimize the adverse effects of a general peroxidation (Miller et al., 1999). This was not the case in DS, where exposure to severe water stress alone led to an increase of GLV emissions suggesting the onset of physical membrane damage, as the enhancement of the lipoxygenase activity, in accordance with other studies (Ebel et al., 1995;Wenda-Piesik, 2011). In addition to the lipoxygenase and hydroperoxide 380 lyase systems producing GLVs, the phenylpropanoid pathway signals plant responses to stimuli induced by abiotic factors (Dixon and Paiva, 1995;Baier et al., 2005;Heath, 2008;Vogt, 2010), but drought stress alone does not induce the phenylpropanoid pathway in Q. robur (Pellegrini et al., 2019).
On the other hand, DS×OS, showed a small increase of GLV only at the highest stress level. We take this to indicate that ozone has the potential to inhibit drought stress damage and therefore the emissions of GLV, by stimulating the 385 phenylpropanoid pathway to form an antioxidant protection for chloroplasts (Pellegrini et al., 2019). Cabané et al. (2004) report that, in poplar leaves, ozone exposure not only stimulate the enzymes of the phenylpropanoid pathway, but also the activity of the enzyme SHDH of the shikimate pathway that yield TPhe in fully developed leaves.
To better understand the emissions of GLVs and Shikimate volatiles, we looked at antioxidant capacity, total phenol content and peroxidase activity summarized in Table 3. No significant differences were found for antioxidant capacity between the 390 sets DS, DS×OS and their corresponding references C, and OS. However, it appeared that the OS had the highest oxidizing capacity. TPhen in the fully developed leaves was significantly higher in the two groups experiencing drought stress (DS, DS×OS) than in those with no drought stress (C, OS). Pellegrini et al. (2019) found, a significant difference in TPhen content in well-watered plants with the increase of ozone, and a decrease at moderate drought and no significant influence of ozone on TPhen during severe drought in Q. robur. The results of our study showed no significant decrease in TPhen due to ozone 395 fumigation both in well-watered and severe drought condition (OS, DS×OS). Peroxidase activity analysis did not show significant differences between the four sets. This is in accordance with the finding of Schwanz and Polle (2001) who found that unspecific peroxidase activities are not affected by drought stress in Q. robur.

Conclusions
The changes in BVOC emissions of Q. robur subject to continuously increasing drought were investigated and differences in 400 the drought progression were observed in plants with and without ozone fumigation. Stomatal conductance and net photosynthesis showed a fast reaction to increasing drought closing stomata and reducing CO 2 uptake strongly. IS S emissions, on the other hand, stayed high down to a SWP of -3 MPa and then decreased gradually. We believe that leaves must have maintained a high production of IS to sustain similar emissions at a SWP of -2 MPa. MT S and SQT S emissions https://doi.org/10.5194/bg-2020-260 Preprint. Discussion started: 6 August 2020 c Author(s) 2020. CC BY 4.0 License. increased under high drought stress. Plants that were subject to one hour of ozone fumigation (~100 ppbv) every day in 405 addition to reduced watering showed lower stomatal conductance at mild drought stress compared to those with no ozone fumigation, and consecutively the effect of drought was slowed down. The Shikimate pathway, producing antioxidants, was stimulated earlier in the set with ozone. The combination of (i) sustained isoprene emissions, (ii) increase of antioxidants due to the higher stimulation of the two pathways (phenylpropanoid and shikimate) and (iii) early closure of the stomata resulted in a longer endurance of drought stress in the set exposed to ozone. Therefore, it was concluded that fumigation with ozone 410 (~100 ppbv) decelerated the effect of drought in Q. robur. Overall Q. robur leaves appeared very resistant to drought stress.
Consequently GLVs indicating cell damage were only emitted at SWP < -5 MPa.
As seasonal drought events and elevated ozone concentrations often occur in parallel in mid latitudes (Löw et al., 2006;Panek et al., 2002) it is important to study their combined stress effects. In this study we observe that a combination of stresses can lead to opposing feedbacks that alter BVOC emissions. These effects are compound specific and reflect 415 biochemical changes in the plant.

Author contribution
AP, LK, TGK, GW, HH drafted the manuscript, which was commented by all co-authors. Laboratory work was performed by AP, LK, ACF, MG, TGK, HS and JG. AP, LK, ACF and HH analyzed and interpreted the data.

Competing interests 420
The authors declare that they have no conflict of interest.

Acknowledgments
This work was supported by the Vienna Science and Technology Fund (WWTF, project number: ESR17-027). In addition AP was supported by a doctoral grant fellowship of the LFU. We are grateful to Polona Jamnik for kindly providing

845
https://doi.org/10.5194/bg-2020-260 Preprint. Discussion started: 6 August 2020 c Author(s) 2020. CC BY 4.0 License. Figure A1: Scheme of a custom-made plant enclosure and set up of the experiment. In brief, the chambers consisted of a PTFEcovered bottom plate with an opening mechanisms to insert and seal the plant stem using PTFE-plugs; furthermore, the bottom plate featured three in-and outlets for gas sampling and ozone exposure; the inlet was raised above the bottom plate to allow for 850 air mixing. The upper part of the chamber consisted of a transparent, 12-liter PET-bag, holding most of the tree crown. The bags were tightly sealed towards the bottom plate.