Dynamics of Foliar Responses to O3 Stress as a Function of Phytotoxic O3 Dose in Hybrid Poplar

With background concentrations having reached phytotoxic levels during the last century, tropospheric ozone (O3) has become a key climate change agent, counteracting carbon sequestration by forest ecosystems. One of the main knowledge gaps for implementing the recent O3 flux-based critical levels (CLs) concerns the assessment of effective O3 dose leading to adverse effects in plants. In this study, we investigate the dynamics of physiological, structural, and morphological responses induced by two levels of O3 exposure (80 and 100 ppb) in the foliage of hybrid poplar, as a function of phytotoxic O3 dose (POD0) and foliar developmental stage. After a latency period driven by foliar ontological development, the gas exchanges and chlorophyll content decreased with higher POD0 monotonically. Hypersensitive response-like lesions appeared early during exposure and showed sigmoidal-like dynamics, varying according to leaf age. At current POD1_SPEC CL, notwithstanding the aforementioned reactions and initial visible injury to foliage, the treated poplars had still not shown any growth or biomass reduction. Hence, this study demonstrates the development of a complex syndrome of early reactions below the flux-based CL, with response dynamics closely determined by the foliar ontological stage and environmental conditions. General agreement with patterns observed in the field appears indicative of early O3 impacts on processes relevant, e.g., biodiversity ecosystem services before those of economic significance – i.e., wood production, as targeted by flux-based CL.


INTRODUCTION
The ground-level concentrations of ozone (O 3 ) have increased during the past century (Maas and Grennfelt, 2016), and are predicted to remain stable or increase during the 21st century (Revell et al., 2015;Fu and Tian, 2019). They have already reached levels negatively affecting crop plants and the natural vegetation (Wittig et al., 2009;Jolivet et al., 2016;Proietti et al., 2016;Li et al., 2017), and steady or increasing impacts are expected over the course of next decades (Karlsson et al., 2017).
Once entering the leaf through stomata, O 3 degradation causes the formation of reactive oxygen species (ROS), the accumulation of which triggers rapid oxidative bursts (Schraudner et al., 1998;Pasqualini et al., 2003;Moura et al., 2018). ROS can also act as elicitors of programed cell death (PCD) reminiscent of plant responses during defensive plant/pathogen interactions which are subsequently designated as hypersensitive responselike (HR-like; Vollenweider et al., 2002;Bhattacharjee, 2005;Günthardt-Goerg and Vollenweider, 2007;Moura et al., 2018). In parallel, an acceleration of cell senescence (ACS), with distinct apparent mechanisms, can be observed (Pell et al., 1999;Günthardt-Goerg and Vollenweider, 2007;Vollenweider et al., 2019). The characteristic symptoms thus include marked degenerative injuries in chloroplasts, in apparent relation to an increase in the constitutive ROS load resulting from the daily photosynthetic activity. As a consequence, these latter organelles are particularly sensitive to O 3 stress (Joo et al., 2005;Kangasjarvi et al., 2005). However, the sequence of plant reactions in response to O 3 stress remains unclear, especially given the driving -but still partially understood -effects of interacting environmental conditions and ontological development. In field vs. climate chamber conditions, for example, the high vs. low-intensity illumination can lead to contrasted symptom expression, with clear synergies between photooxidative and O 3 stress in the former case only (Günthardt-Goerg and Vollenweider, 2007;Paoletti et al., 2009;Moura et al., 2018;Vollenweider et al., 2019). Hence, the dynamics of responses to O 3 stress as a function of environmental conditions needs further research.
Although the effects of O 3 stress have been observed in both mature and developing foliage, their intensity is strongly related to the leaf ontology, the mature leaves being more sensitive than those still in expansion. However, the younger vs. older leaves can show higher rates of stomatal conductance and O 3 uptake (Reich, 1983;Strohm et al., 1999;Bagard et al., 2008;Zhang et al., 2010;Guerrero et al., 2013), suggesting an enhanced detoxification capacity (Bellini and De Tullio, 2019). Still, the mechanisms underlying the higher O 3 tolerance in developing foliage remain largely obscure (Strohm et al., 2002) and the differences in response dynamics as a function of leaf ontogenetic development require further investigations.
To assess and prevent O 3 injury on vegetation and forest trees, a concentration-based index, namely, the accumulated O 3 exposure threshold over 40 ppb (AOT40), was initially proposed (Fuhrer et al., 1997). Given the dependency of O 3 phytotoxicity on stomatal conductance, the biologically (Karlsson et al., 2007;Mills et al., 2011) and environmentally (Musselman et al., 2006;Dizengremel et al., 2013;Büker et al., 2015) more relevant flux-based approach has been increasingly implemented. Nowadays, the O 3 critical level (CL) is defined for given vegetation types or plant species and calculated as the Phytotoxic O 3 Dose over a Y threshold for a specific species or group of species (POD Y_SPEC (Mills et al., 2017). Based on empirical evidence from risk assessment studies -linking POD Y_SPEC values to tree biomass loss or foliar injury -the current CL typically targets 4% maximum, i.e., growth reductions by oxidative stress. However, such markers represent some late O 3 stress effects, at least partly resulting from earlier processes in foliage (i.e., reduced physiological activity/extensive cellular injury) which dynamics primarily depends on detoxification processes (Dghim et al., 2013;Dumont et al., 2014;Dusart et al., 2019a). With a view to the larger implementation and acceptance of flux-based approach, there is then an important knowledge gap regarding the dynamics and effective POD x of first effective O 3 stress effects, prior to the appearance of current risk assessment markers.
In this study, our main objective was to characterize the dynamics of early physiological and structural responses to O 3 stress in poplar trees as a function of flux-based O 3 dose and before, e.g., growth reduction and extensive foliar injury, the primary markers of O 3 stress for defining O 3 CL (Sanz and Catalayud, 2011;Mills et al., 2017). The tested hypotheses (H) included: (H1) the development of injury and growth response to O 3 stress, as well as physiological and structural changes, proceeds in sequential order, with each response showing specific dynamics; (H2a) O 3 elicits different injury responses within the foliage of trees (H2b) with ACS occurring before the development of HR-like lesions (Vollenweider et al., 2019); (H3a) at comparable O 3 dose and irrespective of the applied O 3 concentration, leaves show similar responses and (H3b) response dynamics; (H4) the dynamics of responses depends on the leaf developmental stage (Moura et al., 2018). Therefore, rooted cuttings of hybrid poplar (Populus tremula x alba) were exposed to three O 3 concentrations in fully controlled conditions for a month. The leaf physiology, development of ACS and HR-like lesions, and appearance of visible injuries were monitored over the course of 29 days. The interaction between foliar response dynamics and leaf ontological development was evaluated by assessing the responses to treatments at two distinct leaf positions.

Plant Material and Controlled O 3 Exposure
Young trees from a hybrid Populus tremula x alba clone (INRAE 717-1b4) were cultivated similarly to Cabane et al. (2004). Before the experimental exposure, micro propagated cuttings were grown for 2 weeks in 0.5 L pots containing compost (Gramoflor Universel) and perlite [1:1 (v/v)], and placed in containers covered with transparent acrylic hoods inside a growth chamber. The environmental conditions were set at 22°C/18°C day/night temperature, 350 μmol m −2 s −1 photosynthetic active radiation (PAR, 1 m below lamps) during a 14-h photoperiod (Philips Son-T Agro 400 W lamps), 75%/85% relative humidity (day/night). The young trees were then transplanted into 10 L pots filled with compost (Gramoflor Universel) and fertilized with 3 g l −1 of slow-release Nutricot T 100 granules (13:13:13:2 N:P:K:MgO, Fertil, Boulogne-Billancourt, France). They were further cultivated for 1 month in the same growth chamber and watered to field capacity every day. The trees retained (n = 48), with a view to the forthcoming O 3 exposure experiments, were 29.5 ± 0.2 cm high, with 13.1 ± 0.1 leaves. During experiments, all foliar assessments were repeated at the third and tenth leaf position from the tree base, thus selecting the youngest fully expanded leaf and that still in expansion at treetop by the start of exposure.
Before exposure, the selected trees were randomly distributed among six ventilated phytotron chambers (1 air change min −1 ; 120 cm × 117 cm and 204 cm high) within the O 3 -exposure facility of the PEPLor platform (Faculty of Sciences and Technologies, University of Lorraine). Within each chamber (N = 8 trees), the plant position was randomized by each assessment. The transferred poplars were then left to acclimate for 1 week, with environmental conditions similar to those in the growth chamber. The O 3 exposure experiment included three treatments [charcoal-filtered (CF) air; CF + 80 ppb O 3 ; CF + 100 ppb O 3 ] replicated in two chambers each and performed for 30 days (N = 16 trees per treatment). O 3 was generated from pure oxygen using an O 3 generator (Innovatec II, Rheinbach, Germany), and provided to the chambers during the daytime period in the form of a 13 h square wave, starting 1 h after the light was switched on. The O 3 concentrations within each phytotron chamber were monitored twice an hour using a computer-assisted automatic O 3 analyzer (O341M, Environment SA, Paris).

Dynamics of Leaf Physiology Responses and Estimation of Phytotoxic O 3 Dose
The effect of treatments on the dynamics of leaf gas exchanges was assessed by measuring the net CO 2 assimilation rates (A net ) and stomatal conductance to water vapor (g w ) every 2 days, 3 h after starting the O 3 exposure. Selecting six trees per treatment, the measurements were performed using a Li-6400XT portable photosynthesis system (LiCor, Inc., Lincoln, NE, United States), with cuvette temperature set at 22°C, light intensity (PAR) at 300 and 320 μmol m −2 s −1 for measurements at the third and tenth leaf position, respectively, airflow at 300 μmol s −1 , CO 2 concentration at 400 ppm, and leaf vapor pressure deficit (VPD leaf ) < 1 kPa. The values were recorded once g w and A net remained stable for 30 s.
The g w estimates (mol H 2 O m −2 s −1 ) were used to calculate the instantaneous O 3 uptake into the leaf under environmental stable conditions ( F O 3 ), according to Bagard et al. (2015): [O 3 ] atm as the O 3 concentration (ppb) in the phytotron chamber, and g O 3 (O 3 m −2 s −1 ) as the stomatal conductance to O 3 , according to (Lamaud et al., 2009): as the O 3 and water molecular diffusivity (cm −2 s −1 ) respectively (Massman, 1998). The hourly O 3 uptake (mmol O 3 m −2 h −1 ) was calculated by integrating F O 3 over an hour and the POD 0 (mmol O 3 m −2 ), by cumulating the hourly O 3 uptake since the beginning of experiment. Missing g w measurements were estimated based on values from flanking days (Bagard et al., 2008).
The effect of treatments on the dynamics of leaf chlorophyll content was assessed by measuring estimates of surface-based concentrations of chlorophylls (total chlorophyll index) every day, 1 h after switching the light on and before the start of O 3 treatment. Selecting six trees per treatment, the estimates were obtained averaging 10 measurements per leaf performed with a leaf clamp sensor device (Dualex Force-A, Orsay, France).

Dynamics of Microscopic and Visible Leaf Injury
The development of HR-like lesions within the mesophyll and that of O 3 symptoms throughout foliage was monitored using completing microscopic assessments and visible injury observations. For microscopic assessments, two discs (diameter = 6 mm) per leaf position in two trees per chamber were sampled every 2 days, until HR-like lesions were detected in the 100 ppb O 3 treatment at both leaf positions; the sampling interval was then extended (3-6 days). The harvested discs were processed immediately after sampling.
Necrotic cells within mesophyll as a consequence of HR-like lesions were evidenced using the Trypan blue assay (Pasqualini et al., 2003;Joo et al., 2005;Faoro and Iriti, 2009). Briefly, the leaf discs were stained for 3 min in a hot lactophenol Trypan blue mixture (60 ml staining solution: 10 g phenol, 10 mg Trypan blue, 30 ml ethanol, 10 ml glycerol, 10 ml lactic acid, and 10 ml distilled water) and the necrotic cells contrasted for 20 min against a clear background using 2.5 g ml −1 hot chloral hydrate, before mounting in 60% glycerol (Pasqualini et al., 2003). The preparations were then transferred to WSL where all microscopy assessments were performed. The disk's central part, free of staining artifacts, was observed using the 5× objective of a Leica microscope (Leitz DM/RB). Given the disk thickness (>200 μm) and to create high contrast pictures, the preparations were imaged after inserting the 10x condenser and removing most filters and diaphragms, using the INFINITY 2-1R camera and Lumenera Infinity Analyze (release 6.4) software (Lumenera Corp., Ottawa, ON, Canada). The center of each disk preparation was photographed, creating composite images made of nine tiles each. The percentage area, particle size, and shape properties of HR-like lesions inside of stitched images were quantified using computer-assisted color image analysis (software WinCELL ™ 2004, Regent Instruments Inc., Québec, QC, Canada). Briefly, the software attributed the whole lesion or part of it to one of two color classes (non-oxidized: violet hue; oxidized: dark blue hue) made of 10 shades each, defined based on a representative batch of images and contrasting with the background color class (grayish hue, based on 10 white to gray shades). The quantified parameters characterized the size and shape properties of total and individual lesion particles.
The HR-like lesions and oxidation diagnosis were verified based on hand, and semi-thin sections from samples collected in all treatments at the two leaf positions during the whole study and subsequently processed and observed as described previously (Moura et al., 2018). Briefly, supplementary leaf discs were infiltrated upon sampling with EM-grade 2.5% glutaraldehyde buffered at pH 7.0 with 0.067 M Soerensen phosphate buffer, renewed after vacuum infiltration. Sections (60 μm) obtained using a custom-made hand microtome and kept unstained were used for visualizing chlorophylls and the oxidation of HR-like lesions. Technovit-embedded 1.5 μm sections, obtained using a Supercut Reichert 2050 microtome and stained with Toluidine blue (Vollenweider et al., 2016), were used to identify HR-like markers within necrotic mesophyll cells, after observation with phase contrast illumination in bright field microscopy using the 5-100× objectives of the Leica microscope and imaged using the Infinity camera, as mentioned above. The development of O 3 injury in response to treatments was surveyed in the morning, before the start of O 3 exposure, daily. Upon appearance, the development of visible injury was monitored with pictures of symptomatic leaves. The percentage area of necrosis per leaf within the latter material was quantified using color image analysis, using the Color Segmentation plugin in Fiji freeware (ver. 2.0.0; Schindelin et al., 2012).

Morphological Assessments
After 30 days of exposure, all trees were harvested and biometric assessments were conducted. Tree height was recorded and stem diameter 1.5 cm above root collar was measured, using a hand caliper. The number of leaves per tree, shed or still attached, was recorded before harvest. Leaf and stem material was oven-dried to constant weight, before determining the dry mass of each fraction.

Statistical Analysis
The dynamics of physiological and structural responses to treatments in foliage and the differences in whole-tree morphology and biomass between groups by the end of the experiment were analyzed using linear mixed-effects models (lmem). The fixed-effect factors included the O 3 treatment, leaf position, time or POD 0 and interactions, whereas the tree nested in the chamber (leaf data; with the leaf position as the statistical unit) or the chamber (morphology/biomass data; with the tree as the statistical unit) were introduced in models as random terms. Homoscedasticity and normality of residuals were verified graphically, and the dependent variables were log-or square-transformed to meet the model assumptions, as needed. The differences between treatments at given assessment dates were tested using post-hoc tests (Tukey's honest significant difference). All statistical analyses were performed using R statistical software, version 3.5.0 (R Development Core Team, 2017), with the packages lme4 (Bates et al., 2015) for linear mixed-effects models, and emmeans (Lenth, 2016) for posthoc testing.

Morphological Responses
After 30 days of exposure, no change in tree height, stem diameter, or foliar dry mass in response to O 3 exposure was observed (Table 1). However, the stem biomass (p = 0.014), amount of leaves (p = 0.003), and leaf shedding (p = 1.2 × 10 −6 ) were increased, with significant differences between the 80 and 100 ppb O 3 treatments in the case of leaf shedding.

Dynamics of Stomatal Responses and Changes in the Phytotoxic O 3 Dose
At both leaf positions, the 100 and 80 ppb O 3 treatments significantly reduced g w ( Figure 1A; O 3 treatment: p < 0.001) and accelerated its leaf ontology-driven decrease (O 3 treatment * Time: p < 0.001). This reduction was delayed at the tenth vs. third leaf position (O 3 treatment*leaf position, O 3 treatment * time * leaf position: p < 0.001), with a 50% decrease in g w reached in 15 vs. 6 days, respectively, in the 100 ppb treatment. As indicated by increasing g w in maturing leaves (tenth leaf position) during the first 10 days of exposure irrespective of O 3 exposure (leaf position: p < 0.001), the O 3 treatment affected g w only once the ontological development had been achieved (latency phase, Figure 1A). By the end of the experiment and only at the tenth leaf position, the differences in g w between the 100 and 80 ppb treatments were significant.
By the end of the experiment, trees in the 100 vs. 80 ppb O 3 treatment showed a larger POD 0 , as a consequence of their higher O 3 concentrations and mostly similar g w ( Figure 1B; O 3 treatment, O 3 treatment*time: p < 0.001). After 30 days of exposure, the POD 0 at the third and tenth leaf positions was thus 1.4 and 1.2 times higher in the 100 vs., 80 ppb O 3 treatments. The POD 0 was also higher in leaves at the tenth vs. third leaf position (leaf position: p < 0.001), as a consequence of the delayed leaf ontogeny and higher g w (O 3 treatment*leaf position: p < 0.001). After 10 days of exposure, the POD 0 levels in younger foliage thus exceeded those in older material by approximately 25% and outpaced them by 40% by the end of exposure (O 3 treatment * time * leaf position: p < 0.001).

Dynamics of Leaf Physiology Responses
Irrespective of the leaf position, A net showed responses to treatments and response dynamics similar to g w (Figure 2A vs. Figure 1A). After 30 days of exposure, A net in the 100 ppb O 3 vs. CF treatment was decreased by 70 and 35% at the third and tenth leaf position, respectively. Once the leaf ontogenetic differentiation achieved, A net decreased with POD 0 in a constant and monotonic manner ( Figure 2B; POD: p < 0.001), regardless of the treatment or leaf position. The significant effects of O 3 treatment and O 3 treatment * POD factors (p < 0.001) could then be related to the less affected A net values in the 80 vs. 100 ppb treatment at the highest POD 0 reached by the end of the exposure. The observed reduction in A net as a function of POD 0 was stronger at the third vs. tenth leaf position ( Figure 2B; leaf position p < 0.001) but the dynamics at both leaf positions was similar (leaf position * POD: ns). Hence, at POD 0 of 5 mmol O 3 m −2 , A net at the tenth leaf position did not show any reduction relative to CF treatment yet, vs. 50% A net loss in older leaves, irrespective of the O 3 treatment. The response differences between the two leaf positions were further observed at higher POD 0 . With POD 0 above 9 mmol O 3 m −2 , as recorded in younger leaves only, and also as a consequence of the aforementioned latency effect (Figure 1B), A net never dropped to levels observed at the third leaf position for lower POD 0 . Consequently, the photosynthetic activity in younger vs. older foliage appeared less sensitive to the absorbed O 3 dose.
An O 3 impact on the chlorophyll content index of leaves was detected after 24 days of the experiment. The impact was restricted to the third leaf position (Figure 3A; O 3 treatment: ns; O 3 treatment * time, O 3 treatment*leaf position: p < 0.001), showing a decrease of 30% for the total chlorophyll index in the 100 ppb O 3 vs. CF treatment. These findings primarily related to latency effects due to leaf ontological maturation, which were observed at the third as well as the tenth leaf position in the case of this parameter, lasting 7 and 17 days, respectively. Accordingly, a larger latency peak was observed in younger than older foliage. The total chlorophyll index decreased with higher POD 0 ( Figure 3B; POD: p < 0.05), irrespective of the O 3 treatment (O 3 treatment: ns, O 3 treatment * POD: ns). At low POD 0 , the decline was rather monotonic, but accelerated with values exceeding 8 mmol m −2 at the third leaf position, thus contrasting with the nearly linear drop observed in younger leaves. Confirming a higher O 3 tolerance leaves at the tenth leaf position showed smaller (leaf position: p < 0.05) and slower (leaf position * POD 0 p < 0.001) drops with higher POD 0 .

Dynamics of HR-Like Lesion Spread and Development
The microscopic necrosis observed in mesophyll using the Trypan blue assay was diagnosed as being caused by HR-like processes (Figures 4A-D), based on several typical O 3 -stress markers (Paoletti et al., 2009;Vollenweider et al., 2019). They included (1)  the lower palisade parenchyma (Figures 4I-L). Non-oxidized lesions ( Figure 4G) showed up first (Figures 5A,C, 6C), with, for instance, still green chloroplasts visible within collapsed dead cells (Figures 4I,J). The dark hues of oxidized lesions ( Figure 4L) were enhanced by staining with Trypan blue (Figures 4E,G). Oxidized cells showed sharp wall angles indicative Values represent means ± SE, n = 6. Different letters indicate significant differences between treatments at a given assessment date (post-hoc Tukey's Honest significant difference, p ≤ 0.05).
of breaks and disrupted cell content (Figures 4K,L). All these typical HR-like traits showed little variation, regardless of the O 3 treatment or leaf position. The first HR-like lesions at the third and tenth leaf position were observed after 6 and 13 days of exposure, respectively. This was much earlier than reductions in the chlorophyll content index (Figure 5A). Despite large response variability among trees, the effect of O 3 treatment was significant (Figure 5A; O 3 treatment: p < 0.05; O 3 treatment*time: p < 0.001). A larger leaf percentage area showing HR-like lesions was observed for the 100 vs. 80 ppb O 3 treatment, with differences between the two treatments at the third leaf position becoming significant after 20 days of exposure (O 3 *time; p < 0.001). After 27 days of treatment, the percentage area of lesions in the 100 vs. 80 ppb O 3 treatment was two and five times higher at the third and tenth leaf positions, respectively. However, each leaf position showed specific response dynamics (O 3 treatment * leaf position: p < 0.01, O 3 treatment * time * leaf position: p < 0.05), rather sigmoidal-like vs. linear -once lesions appeared -in older vs. younger foliage ( Figure 5A). Moreover, the HR-like lesions in response to the two O 3 concentrations showed up simultaneously at the third leaf position whereas a 7-day delay was observed at the tenth leaf position (Figure 5A).
When expressed as a function of POD 0 , the differences between the two O 3 treatments in the leaf percentage area showing HR-like lesions were leveled out, especially at the third leaf position (Figure 5B; O 3 treatment: ns). However, the dependency on POD 0 was lessened at the tenth vs. third leaf position (leaf position * POD: p < 0.001), and distinctly higher lesion percentage areas in response to similar POD 0 were observed in the 100 vs. 80 ppb O 3 treatment in younger leaves (O 3 treatment * POD * leaf position: p < 0.05). Hence, not only the O 3 dose but also the O 3 absorption rate then determined the lesion severity. The oxidized HR-like lesions, expressed as a function of time or POD 0 , showed responses and response dynamics similar to HR-like lesions taken globally (Figures 5C,D  vs. Figures 5A,B). The main differences included a smaller percentage of the injured area and a weaker symptom dynamics. At the tenth leaf position, oxidized HR-like lesions in the 80 ppb O 3 treatment were observed only occasionally. Analyzing single HR-like injuries, the shape (data not shown) and size of lesions remained stable over time or with increasing POD 0 . Furthermore, they did not respond to the O 3 treatment, leaf position, or interaction factors (Figures 6A,B; all factors: ns). The only change observed was increasing oxidation with longer exposures and at higher POD 0 (Figures 6C,D; Time, POD: p < 0.05). Hence, the observed increases in the percentage area of HR-like lesions with time, or higher POD 0 and in response to the O 3 treatment (Figures 5A,B) resulted as a consequence of the multiplication of single HR-like reactions and higher lesion density, rather than from increased growth of already developed injuries. However, both the higher lesion density and the growing oxidation of individual HR-like lesions could contribute to the observed increase in the leaf percentage area showing oxidation (Figures 5C,D).

Emergence of Visible Symptoms
The first visible symptoms in leaves exposed to the O 3 treatments appeared by the end of the experiment, that is, 23 days after the start of exposure. The first HR-like lesions had been detected more than 2 weeks earlier, whereas the observed drops in foliar chlorophyll content index were rather synchronous ( Figure 7A vs. Figures 3A, 5A). These visible symptoms consisted of intercostal necrotic dark spots spread in the leaf Frontiers in Plant Science | www.frontiersin.org blade, as previously observed in poplar (Figure 7A; Cabane et al., 2004;Giacomo et al., 2010;Dghim et al., 2013). By the end of exposure, only low levels of injury could develop (O 3 treatment: ns; O 3 treatment*Time p < 0.001), with significantly higher percentages in the 100 vs. 80 ppb O 3 treatment at the third leaf position only. In younger leaves, the visible injury appeared 4 days later than at the third leaf position, and differences between treatments throughout the experiment remained non-significant (leaf position, O 3 treatment * time * leaf position: p < 0.05; O 3 treatment*leaf position: p < 0.01).
When expressed as a function of POD 0 , the visible injuries appeared at a lower O 3 threshold at the third than tenth leaf position (9 and 12 mmol O 3 m −2 in the case of 100 ppb treatment; Figure 7B). With only the beginning of injury development assessed in a 30-day experiment, the O 3 treatment effects could not reach any significance, and only preliminary information on the dynamics of visible symptom development was thus obtained. The simultaneous detection of early injury in the two O 3 treatments ( Figure 7A) thereby resulted in higher injury values for similar POD 0 in the 80 vs. 100 ppb O 3 treatment ( Figure 7B; O 3 treatment * POD: p < 0.05). Similarly, differences between younger and older leaves were detected as a trend only, with still very low injury values recorded at the tenth leaf position and for the higher POD 0 values only (leaf position: ns; leaf position * POD: p < 0.001).

Dynamics of Physiological and Structural Responses to O 3 Stress
The physiological and structural responses detected during 30 days of O 3 exposure developed mostly before and in some cases at the same time as the initial visible injury and first leaf shedding, whereas no O 3 effect on the gross morphology of trees was observed. The two O 3 treatments accelerated the ontological decline of leaf gas exchange at the two-leaf positions, once leaf physiology had reached maturity. O 3 -induced reductions in the stomatal conductance and net CO 2 assimilation are well-documented in the case of various species, including poplars (Pell et al., 1992;Bagard et al., 2015;Dusart et al., 2019a) but more rarely with a leaf ontogenetic perspective. The decrease in net CO 2 assimilation could result from a smaller stomatal aperture, limiting CO 2 availability. Indeed, the O 3 effects on stomata are well-established (Kangasjarvi et al., 2005), including slower movements of cell guard cells upon exposure referred to as stomatal sluggishness, as observed in different species but with higher measurement frequency than in our case (Paoletti and Grulke, 2010;Dumont et al., 2013;Dusart et al., 2019b). Lower mesophyll conductance could also contribute to the observed acceleration of A net reduction, as compared with the ontogenetic decrease observed in CF trees (Xu et al., 2019). Although not significant in older foliage, the drop in A net tended to be more expressed in the 100 vs. 80 ppb treatment after 20 days, suggesting an additional reduction in carboxylation efficiency and rate of electron transport (Bagard et al., 2008;Shang et al., 2017). A possible cause could be the starting degradation of the photosynthetic machinery, as suggested by the concomitant reduction in the chlorophyll content index. The reduction in leaf chlorophyll content is another well-documented leaf physiological response to elevated O 3 (Reich, 1983;Bagard et al., 2015;Dusart et al., 2019b), indicative of ACS, together with other markers of chloroplast degeneration (Mikkelsen and HeideJorgensen, 1996;Günthardt-Goerg and Vollenweider, 2007;Moura et al., 2018). Also typical of ACS and degenerative processes was the progressive and mostly monotonic reductions observed in the case of all physiology parameters (g w , A net , and chlorophyll index). This contrasted with abrupt responses upon exceedance of a threshold, as observed in the case of HR-like reactions. In the field, the ontogeny-driven ACS latency observed in both younger and older foliage (total chlorophyll index) may also contribute to delaying the onset of degenerative events in response to O 3stress, as suggested by the development of ACS traits and visible injury rarely occurring before summer, once the foliage has fully matured (Vollenweider et al., 2019).
The new cell necrosis assay, using computer-assisted color image analysis, allowed us to monitor the emergence and development of validated HR-like reactions for the first time. It provided unprecedented capacity for quantitative assessments of cell death reactions in experimental conditions, overcoming limitations and uncertainties regarding the observation of visible injury only. Typical HR-like markers were detected in the lesions (Paoletti et al., 2009;Vollenweider et al., 2013Vollenweider et al., , 2019Feng et al., 2016). However, there were marked differences in classical traits as well, including the mid-rather than the upper-mesophyll location of HR-like lesions or a missing intra-and intercellular gradient of injury. Such features indicated low levels of photo-oxidative stress (Foyer et al., 1994;Günthardt-Goerg and Vollenweider, 2007;Guerrero et al., 2013). Given maximum PAR above 2000 vs. 350 μmol m−2 s −1 with high (Ritchie, 2010;Poorter et al., 2019) vs. low light conditions, HR-like reaction peculiaritiestogether with the late onset of ACS -can be attributed to specifics in the environmental conditions, especially regarding PAR supply. This finding thus provides further confirmation of the close dependency relating the O 3 symptom expression in foliage and precise experimental and exposure conditions of tested material (Paoletti et al., 2009;Moura et al., 2018;Vollenweider et al., 2019).
The rather sigmoidal-like injury dynamics observed in older foliage was in good agreement with already existing molecular and trait evidence on HR-like processes. The 6 days/3 mmol O 3 m −2 s −1 delay between the start of exposure and occurrence of first lesions was thus indicative of the O 3 dose-dependent onset of genetically controlled PCD (Rao et al., 2000;Overmyer et al., 2005). The steep increase in injury percentage area reflected the rapid cell death completion once PCD started (Overmyer et al., 2005;Günthardt-Goerg and Vollenweider, 2007). Finally, the plateau reached was indicative of lesion containment, blocking its further spread (Overmyer et al., 2003;Kangasjarvi et al., 2005;Marchica et al., 2019). In younger foliage, the experiment was terminated before a plateau could be reached, with values of lesion percentage area which would probably have been sizably higher than at the third leaf position. Further suggesting the genetic control of PCD, in older foliage, the first HR-like reactions occurred 17 days/at 5.3 mmol O 3 m −2 s −1 before any evidence of biochemical limitation and chloroplast injury, as indicated by low levels of chlorophyll content indexes. The early HR-like responses, and their antecedence concerning ACS and first visible injury, contrasted with field evidence (Vollenweider et al., 2019), further outlining how important the environmental conditions can be regarding response order and dynamics.
Image analysis in WinCELL based on three color classes allowed us to quantify the structurally contrasted non-oxidized and oxidized lesions, based on constitutive and stained-color characteristics. REDOX changes during oxidative stress and cell death form an important cell physiology process (Foyer and Noctor, 2005) that is well-documented in the case of O 3 stress (Ranieri et al., 2000;Baier et al., 2005;Chen and Gallie, 2005;Bellini and De Tullio, 2019), detectable with different structural and ultrastructural markers (Moura et al., 2018;Vollenweider et al., 2019) and underlying changes in visible symptom expression. Expressed as leaf area percentages, the oxidized and total HR-like lesions showed similar dynamics and responses to increasing POD 0 (same results for non-oxidized lesions, data not shown). The main differences in oxidized vs. non-oxidized lesions included their (1) lower percentage area, (2) delay in development, and (3) higher severity (i.e., cell wall breaks, cell content disruption). Non-oxidized and oxidized lesions may thus correspond to two types or two stages of HR-like reactions. However, the first hypothesis appears unlikely, given lacking molecular evidence for alleged PCD severity variation. In favor of the second, oxidized vs. non-oxidized lesions appeared later, and the oxidation degree of a lesion increased with time. However, it implies the further evolution of HR-like lesions after cell collapse and death, which needs further structural and ultrastructural confirmation.
The visible injury was detected only once 2-5% of the leaf percentage area showed oxidized lesions, thus with a detection delay and a POD 0 gap compared to the onset of HR-like reactions amounting to 18 days and 4.9 mmol O 3 m −2 s −1 . Similarly, risk assessment studies using visible injury markers rather target the late and final structural evolution of responses to O 3 stress in foliage, with possible interspecific variation, instead of the injury appearance in foliage.

Leaf Position Dependency of Responses to O 3 Stress
Reductions in leaf gas exchanges or the development of HR-like lesions and visible symptoms at the tenth vs. third leaf position occurred later and for larger POD 0 . Given the higher stomatal conductance and POD 0 in younger leaves, their greater physiological activity and lower levels of injury suggest higher O 3 tolerance, while a contribution by enhanced stomatal closure can be excluded. This finding is confirmed by similar reports on enhanced O 3 tolerance in maturing leaves (Reich, 1983;Paakkonen et al., 1996;Strohm et al., 2002;Bagard et al., 2008;Zhang et al., 2010;Guerrero et al., 2013). This may be related to sink functional properties and larger resource availability for defense and repair (Coleman, 1986). Resource availability in young leaves could be increased by the supply of nutrients (such as nitrogen, potassium, and phosphorus) coming from senescent leaves (Maillard et al., 2015;Have et al., 2017). Hence, the concentration of phenolics with antioxidant properties and other antioxidative capacities decline during the sink-to-source transition in maturing foliage, thus increasing leaf susceptibility to oxidative stress (Coleman, 1986;Strohm et al., 2002;Blokhina and Fagerstedt, 2006;Bellini and De Tullio, 2019). However, and in contrast to older foliage, the development of HR-like lesions as a function of POD 0 at the tenth leaf position depended on the O 3 treatment, with a higher O 3 tolerance in the 80 ppb O 3 treatment. Given the high O 3 dose in maturing leaves, this finding highlights the importance of the O 3 absorption rate given the saturation of the antioxidative system. The higher O 3 tolerance in younger vs. older foliage was further confirmed by their still comparable leaf percentage areas showing HR-like lesions in the 100 ppb O 3 treatment despite 1.8 times higher POD 0 at the tenth leaf position.

Reaction Gradient in Foliage Concerning Critical O 3 Levels
In our experiment, the current CL (POD Y_SPEC for beech and birch = 5.2 mmol O 3 m −2 ; Mills et al., 2017) was equivalent to a POD 0 of 5.7 mmol O 3 m −2 . By the end of exposure, this CL had thus been exceeded by 1.54 and 2.35 times at the third and tenth leaf positions, respectively. If any impairment of tree morphology and biomass was still lacking, reductions in leaf gas exchange, development of structural injury, and the emergence of visible symptoms at the third leaf position had already been observed for O 3 dose, amounting to 0.82, 0.69, and 1.46 times the current CL, respectively. At the tenth leaf position, these responses were detected for POD Y_SPEC 1.83, 1.18, and 2.32 times above CL. These findings highlight the high dependency of sensitivity evaluations on the selected parameters and scale of observation. They also outline the within-tree gradient of sensitivity to O 3 stress, given the large size of such organisms, which, as a result, complicates O 3 risk assessment. They finally indicate that below CL, significant effects in the foliage of trees, such as in the impairment of leaf physiology and development of microscopic necrosis in extended parts of mesophyll, can be expected. These responses may already contribute to reduced carbon uptake and storage in foliage and other tree organs before reaching CL thresholds.

CONCLUSION
In this study, we characterized the dynamics of physiological, structural, and morphological responses to two levels of O 3 exposure and as a function of time, POD 0 and leaf position, in fully controlled conditions. We observed contrasting dynamics, monotonic or sigmoidal-like, as a function of plant responses but irrespective of leaf position, before any visible symptoms and effects on the gross morphology of trees. The first microscopic necrosis developed weeks before the appearance of visible symptoms and at half the O 3 dose. Concerning experimental hypotheses (H), the sequential development and distinct dynamics of physiological, structural, and morphological responses to O 3 stress was confirmed (confirmation of H1); both HR-like and ACS responses were elicited, the former occurring first (confirmation of H2a, rejection of H2b). When expressed as a function of POD 0 , leaf responses did not depend on the O 3 treatment (confirmation of H3a), except for the development of structural injury that depended on the O 3 absorption rates in younger foliage (partial rejection of H3b). Finally, response dynamics were strongly related to leaf age as a function of time or POD 0 , showing delay in younger foliage (confirmation of H4). This study thus sheds light on the syndrome of early reactions to O 3 stress and disentangles the specific dynamics of distinct but co-occurring plant responses, before CL exceedance. The resulting variety of symptoms, as observed by the end of the experiment, provides an exemplary experimental demonstration for integrative injury display, as found in the field late in summer. Given ACS and HR-like timing inversion, compared to field conditions, the ontogenetical and environmental drivers also appear to have a prevailing effect over the sensitivity of affected markers, regarding the timing and dynamics of each cellular response. Whatever the exact sequence order of early reactions to O 3 stress below CL -in the field or controlled conditions, they will modify key foliage properties. Such impacts can be relevant for some, e.g., biodiversity ecosystem services before those of economic significance -i.e., wood production, as targeted by flux-based CL.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.