Elevated Ozone Concentration Reduces Photosynthetic Carbon Gain but Does Not Alter Leaf Structural Traits, Nutrient Composition or Biomass in Switchgrass

Elevated tropospheric ozone concentration (O3) increases oxidative stress in vegetation and threatens the stability of crop production. Current O3 pollution in the United States is estimated to decrease the yields of maize (Zea mays) up to 10%, however, many bioenergy feedstocks including switchgrass (Panicum virgatum) have not been studied for response to O3 stress. Using Free Air Concentration Enrichment (FACE) technology, we investigated the impacts of elevated O3 (~100 nmol mol−1) on leaf photosynthetic traits and capacity, chlorophyll fluorescence, the Ball–Woodrow–Berry (BWB) relationship, respiration, leaf structure, biomass and nutrient composition of switchgrass. Elevated O3 concentration reduced net CO2 assimilation rate (A), stomatal conductance (gs), and maximum CO2 saturated photosynthetic capacity (Vmax), but did not affect other functional and structural traits in switchgrass or the macro- (except potassium) and micronutrient content of leaves. These results suggest that switchgrass exhibits a greater O3 tolerance than maize, and provide important fundamental data for evaluating the yield stability of a bioenergy feedstock crop and for exploring O3 sensitivity among bioenergy feedstocks.


Introduction
Obtaining renewable energy from biomass feedstocks is projected to reduce reliance on traditional fossil fuels and emissions of greenhouse gases while benefitting economic growth and energy security [1][2][3]. Currently, the production of corn-based ethanol is the most common biofuel feedstock in the USA, but ethanol can also be derived from woody feedstocks or other dedicated bioenergy crops [1,[3][4][5]. Switchgrass, a native perennial warm-season C 4 grass of North America [6], has been recognized as an emerging and promising bioenergy feedstock [4,7,8]. With broad adaptability, switchgrass can produce high biomass yields under limited water and nutrient supply on marginal croplands [9][10][11]. Switchgrass also has the potential to produce greater biomass yields (13 Mg ha −1 ) than maize grain (11 Mg ha −1 ) given similar inputs [4]. Several studies have examined the impact phyologenetic relationship with maize [59], we hypothesized that elevated O 3 would lead to: (a) reductions in photosynthetic traits and capacity; (b) alterations in leaf structure; and (c) changes in biomass and nutrient composition.

Leaf Photosynthetic and Chlorophyll Fluorescence Responses to Elevated O 3
On 25 July (DOY 206) and 13 August (DOY 225), 2018, elevated O 3 concentration significantly reduced in situ net CO 2 assimilation rates (A) and stomatal conductance to water vapor (g s ), but there was no significant effect of elevated O 3 on intercellular CO 2 concentration (C i ) or instantaneous water use efficiency (iWUE) (Figure 1). Chlorophyll fluorescence parameters were not as consistently altered by elevated O 3 . A significant reduction in PSII maximum efficiency (F v '/F m ') was observed on DOY 206, but not DOY 225 (Figure 2a), while significant reductions in quantum yield of PSII (Φ PSII ) and electron transport rate (ETR) were only observed on DOY 225 (Figure 2b reductions in photosynthetic traits and capacity; (b) alterations in leaf structure; and (c) changes in biomass and nutrient composition.

Leaf Photosynthetic and Chlorophyll Fluorescence Responses to Elevated O3
On 25 July (DOY 206) and 13 August (DOY 225), 2018, elevated O3 concentration significantly reduced in situ net CO2 assimilation rates (A) and stomatal conductance to water vapor (gs), but there was no significant effect of elevated O3 on intercellular CO2 concentration (Ci) or instantaneous water use efficiency (iWUE) (Figure 1). Chlorophyll fluorescence parameters were not as consistently altered by elevated O3. A significant reduction in PSII maximum efficiency (Fv'/Fm') was observed on DOY 206, but not DOY 225 (Figure 2a), while significant reductions in quantum yield of PSII (ΦPSII) and electron transport rate (ETR) were only observed on DOY 225 (Figure 2b

Changes in the BWB Relationship due to Elevated O3
To further estimate the effect of elevated O3 on switchgrass carbon and water fluxes, the Ball-Woodrow-Berry (BWB) model was applied to gas exchange data collected in the field. As predicted, AH s C s was strongly correlated with gs in both ambient (p < 0.0001) and elevated (p < 0.0001) O3 ( Figure   4). However, there was no significant difference in the slope or intercept of the relationship between gs and

Changes in the BWB Relationship due to Elevated O3
To further estimate the effect of elevated O3 on switchgrass carbon and water fluxes, the Ball-Woodrow-Berry (BWB) model was applied to gas exchange data collected in the field. As predicted,

Changes in the BWB Relationship due to Elevated O 3
To further estimate the effect of elevated O 3 on switchgrass carbon and water fluxes, the Ball-Woodrow-Berry (BWB) model was applied to gas exchange data collected in the field. As predicted, AH s C s was strongly correlated with g s in both ambient (p < 0.0001) and elevated (p < 0.0001) O 3 ( Figure 4). However, there was no significant difference in the slope or intercept of the relationship between g s and AH s C s in ambient and elevated O 3 ( Figure 4).

Leaf Respiration and Dark Adapted Chlorophyll Fluorescence Responses to Elevated O3
Leaf dark respiration did not differ significantly between ambient and elevated O3 (Figure 5a). Although elevated O3 treated leaves had significantly greater dark adapted chlorophyll fluorescence (Fv/Fm) than ambient leaves (Figure 5b), the Fv/Fm values were very similar and both were higher than 0.7 (0.70 ± 0.0084 vs. 0.72 ± 0.0039) at ambient and elevated O3, indicating that leaves under both treatments were not experiencing photodamage.

Leaf Morphology and Anatomy were not Altered by Elevated O3
Leaf thickness, conduit size, inner bundle sheath size, vein size and sclerenchyma size tended to be greater in ambient compared to elevated O3, however the trends were not statistically significant (Table 1). There were no significant effects of elevated O3 on other traits of leaf anatomy (Table 1). In addition, elevated O3 did not alter stomatal and minor vein characteristics in switchgrass (Table 1). In both ambient and elevated O3 treatment, leaf minor vein length per leaf area was not correlated

Leaf Respiration and Dark Adapted Chlorophyll Fluorescence Responses to Elevated O 3
Leaf dark respiration did not differ significantly between ambient and elevated O 3 (Figure 5a). Although elevated O 3 treated leaves had significantly greater dark adapted chlorophyll fluorescence (F v /F m ) than ambient leaves (Figure 5b), the F v /F m values were very similar and both were higher than 0.7 (0.70 ± 0.0084 vs. 0.72 ± 0.0039) at ambient and elevated O 3 , indicating that leaves under both treatments were not experiencing photodamage.

Leaf Respiration and Dark Adapted Chlorophyll Fluorescence Responses to Elevated O3
Leaf dark respiration did not differ significantly between ambient and elevated O3 ( Figure 5a). Although elevated O3 treated leaves had significantly greater dark adapted chlorophyll fluorescence (Fv/Fm) than ambient leaves (Figure 5b), the Fv/Fm values were very similar and both were higher than 0.7 (0.70 ± 0.0084 vs. 0.72 ± 0.0039) at ambient and elevated O3, indicating that leaves under both treatments were not experiencing photodamage.

Leaf Morphology and Anatomy were not Altered by Elevated O3
Leaf thickness, conduit size, inner bundle sheath size, vein size and sclerenchyma size tended to be greater in ambient compared to elevated O3, however the trends were not statistically significant (Table 1). There were no significant effects of elevated O3 on other traits of leaf anatomy (Table 1). In addition, elevated O3 did not alter stomatal and minor vein characteristics in switchgrass (Table 1). In both ambient and elevated O3 treatment, leaf minor vein length per leaf area was not correlated

Leaf Morphology and Anatomy Were Not Altered by Elevated O 3
Leaf thickness, conduit size, inner bundle sheath size, vein size and sclerenchyma size tended to be greater in ambient compared to elevated O 3 , however the trends were not statistically significant (Table 1). There were no significant effects of elevated O 3 on other traits of leaf anatomy (Table 1). In addition, elevated O 3 did not alter stomatal and minor vein characteristics in switchgrass (Table 1). In both ambient and elevated O 3 treatment, leaf minor vein length per leaf area was not correlated

No Changes in Biomass and Nutrient Composition between Ambient and Elevated O3
There was no significant effect of elevated O3 on leaf area, biomass or tiller number in switchgrass after growing in chronic elevated O3 for two months (Table 2). Additionally, there was no significant effect of elevated O3 on leaf mass per area (LMA) (64.8 ± 1.12 vs. 69.4 ± 2.60) ( Table 2). Leaf and stem N content, and leaf and stem C:N were also unchanged by elevated O3 (Table 2). Elevated O3 led to a significant decrease in potassium (K) and in leaf stable carbon isotope composition (δ 13 C) (Table 3). However, no changes in the content of micronutrients or leaf stable nitrogen isotope composition (δ 15 N) were observed (Table 3).

No Changes in Biomass and Nutrient Composition between Ambient and Elevated O 3
There was no significant effect of elevated O 3 on leaf area, biomass or tiller number in switchgrass after growing in chronic elevated O 3 for two months (Table 2). Additionally, there was no significant effect of elevated O 3 on leaf mass per area (LMA) (64.8 ± 1.12 vs. 69.4 ± 2.60) ( Table 2). Leaf and stem N content, and leaf and stem C:N were also unchanged by elevated O 3 ( Table 2). Elevated O 3 led to a significant decrease in potassium (K) and in leaf stable carbon isotope composition (δ 13 C) (Table 3). However, no changes in the content of micronutrients or leaf stable nitrogen isotope composition (δ 15 N) were observed (Table 3). Table 2. Leaf and stem biomass, N content and C:N of switchgrass exposed to ambient and elevated O 3 in 2018. Data are presented as means ± SE (n = 3). Significant differences between ambient and elevated O 3 are indicated by different letters.

Impact of Elevated O 3 on Photosynthesis and Stomatal Conductance
It is well known that elevated O 3 negatively influences the growth, development, production and yield of C 3 plants. In contrast, there is a much more limited body of information about the impacts of elevated O 3 on photosynthesis and performance of C 4 species. Here, we studied the effects of elevated O 3 on leaf photosynthetic and structural traits using a promising C 4 bioenergy crop, switchgrass, which was grown under season-long elevated O 3 in the field with FACE technology. We found that elevated O 3 significantly reduced midday A and g s (Figure 2), consistent with past observations in maize [50][51][52][53][54] and sugarcane [56,57]. Additionally, maximum photosynthetic capacity (V max ) was lower in elevated O 3 (Figure 3), also consistent with previous observations in maize [53]. However, intercellular CO 2 concentration (C i ) and instantaneous water use efficiency (iWUE) did not statistically differ between ambient and elevated O 3 (Figure 1), and the slope between g s and AH s C s was also different ( Figure 4). In C 3 species, it is commonly observed that elevated O 3 impairs photosynthetic capacity, with reduced g s being a consequence rather than a driver of lower A [29,60]. Additionally, stomata can be damaged by O 3 exposure leading to sluggish response to other environmental parameters [61]. Thus, in elevated O 3 , greater g s may be required to support a given A, which decreases water use efficiency [61]. In switchgrass, this was not observed, and both A and g s were proportionally affected, leading to no change in iWUE or the slope of the BWB model. To our knowledge, this is the first study to test how the slope of the BWB is affected by elevated O 3 in C 4 species, but, in rice, elevated O 3 -induced changes in the BWB relationship were observed in O 3 sensitive cultivars, but not more tolerant cultivars [43]. In sugarcane, the degree to which A and g s were affected by elevated O 3 varied with genotype [57], thus it is possible that slope of the BWB relationship was also impacted, but this was not explicitly tested. In this study on switchgrass, only one genotype was investigated, but it is also possible that there is intraspecific genetic variation in O 3 response within switchgrass.
Long-term exposure to elevated O 3 stress often significantly reduces either light-and/or dark-adapted chlorophyll fluorescence parameters [39][40][41][42]. In switchgrass, the effects of elevated O 3 on fluorescence were inconsistent. Reductions in F v '/F m ' were only found on DOY 206, while quantum yield of PSII (ΦPSII) and electron transport rate (ETR) were reduced later in the growing season. No changes in photochemical quenching (qP) were observed in O 3 -exposed leaves (Figure 2), indicating PSII photochemistry did not change in the O 3 -treated leaves of switchgrass. Although maximum dark-adapted quantum yield of photosystem II (F v /F m ) was significantly increased in O 3 -exposed leaves, both values of ambient and elevated O 3 were higher than 0.7 (Figure 5b), which further confirmed that PSII reaction center was not damaged by elevated O 3 . Overall, PSII photochemistry in switchgrass was not strongly impacted by O 3 stress, even though there were reductions on photosynthetic capacity and stomatal conductance.

Effect of Elevated O 3 on Leaf Structure
Leaf structural traits such as leaf mass per area (LMA) are predicted to contribute to O 3 sensitivity among species [62,63], but the effects of elevated O 3 on leaf anatomical traits have not been well studied, especially in C 4 species. There was no significant effect of O 3 in switchgrass foliar anatomy (Table 1), which may result from the unique leaf structural features of C 4 species including large bundle sheath volumes that enable greater Rubisco content than needed for photosynthetic saturation [64]. Feng et al. (2018) showed that tree species with greater LMA tended to have more O 3 tolerance [63]. Switchgrass has greater LMA than maize [53], and showed greater tolerance to O 3 , although only a single genotype of switchgrass was investigated. A more thorough characterization of the relationship between LMA and O 3 tolerance in grasses would be needed to test if the relationship found in trees translates to other functional groups.
Leaf minor vein density is an important determinant of leaf water and nutrient transport efficiency, which together are essential for hydraulic conductance and stomatal function. Previous work in other species reported that elevated O 3 decreases whole plant hydraulic conductance [65], but studies have not examined how elevated O 3 impacts the anatomical determinants of hydraulic conductance such as leaf minor vein density. Under temperature stress, leaf minor vein density and stomatal density increased in parallel supporting greater leaf hydraulic conductance [66]. Other studies have also shown that the correlation between leaf minor vein density and stomatal density varies with environmental factors including temperature, atmospheric humidity and altitude [66][67][68][69]. In this study on switchgrass, there was no correlation between leaf minor vein density and stomatal density (Figure 6a), but there was an unexpected negative correlation between leaf minor vein density and guard cell length as well as between the leaf minor vein density and stomatal pore area index (Figure 6b,c). Across a diverse range of species, leaf minor vein density is positively correlated with stomatal density [70], however the opposite pattern of what was observed here. Study of additional genotypes and conditions would be needed to more broadly understand this relationship in switchgrass.

Effect of Elevated O 3 on Biomass and Nutrient Composition
Many previous studies have shown that elevated O 3 negatively affects both biomass and yield production across plant species [27,29]. A review of woody species estimated that elevated O 3 reduces biomass by 7% across diverse tree species [28]. Similarly, a review of the effects of elevated O 3 on reproductive processes suggested that yield and seed weight are reduced to a similar extent in both C 3 and C 4 species [71]. However, few C 4 species have been studied in detail, and most of the prior work focused on maize. In tobacco, growth at high N treatment protected from O 3 damage [72] suggesting that the negative impacts of O 3 on biomass may be improved by soil nutrient conditions. Similar results were observed in switchgrass which was grown under high fertility at the FACE site in this study ( Table 2). No differences in leaf and stem N content or above-ground biomass were observed in ambient and elevated O 3 . Given the decrease in photosynthesis, it is somewhat surprising that no differences in above-ground biomass were observed. However, in wheat, root biomass is reduced more than shoot biomass at elevated O 3 [27], and it is possible that there was a change in allocation in switchgrass at elevated O 3 as well.
Other nutrients including magnesium (Mg), phosphorus (P), sulfur (S), potassium (K), zinc (Zn), calcium (Ca) and iron (Fe) are important components of the photosynthetic apparatus and reactions [73][74][75] and also impact the efficiency of biomass combustion systems [76]. Here, a significant decrease in elevated O 3 was only observed for K (Table 3), which may be associated with reductions in net CO 2 assimilation (A) and stomatal conductance (g s ) (Figure 1) [75]. Indeed, changes in nutrient composition highly depend on the soil properties and on the O 3 impact on plant metabolism [77][78][79]. There was a significant, but small, reduction in leaf stable carbon isotope composition (δ 13 C) but no change in nitrogen isotope composition (δ 15 N) at elevated O 3 . Generally, δ 13 C is positively correlated with water use efficiency [80][81][82], while δ 15 N serves as an indicator of plant N acquisition, fixation and cycling [83][84][85]. Both δ 13 C and δ 15 N are strongly controlled by environmental conditions. Although elevated O 3 did not alter iWUE on DOY 206 and DOY 225 (Figure 1 and Table 1), decreased δ 13 C suggests that there was an accumulated effect of elevated O 3 over the life-time of the leaf, albeit small. As discussed above, plants grown in sufficient N were not compromised by elevated O 3 , which could partially explain the limited effects of elevated O 3 on δ 15 N [72].

Implications for Bioenergy Feedstock Development
Although a successful bioenergy industry will require high productivity and yield stability of bioenergy feedstocks, how the bioenergy crops acclimate to a rapidly changing, more polluted environment should be considered seriously. Our results provide evidence that switchgrass exhibits O 3 tolerance, and suggest that C 4 bioenergy crops including maize and switchgrass differ in O 3 tolerance. However, the year of our experiment was extremely wet, and previous work in maize also showed that O 3 sensitivity was greater in dry years [31], thus additional side-by-side experiments with more genotypes and species are needed for a definitive comparison. In natural environments ambient O 3 concentrations strongly vary over the land surface throughout the day and over the season, resulting in geographic variation in O 3 pollution. Therefore, understanding variation in C 4 bioenergy feedstock responses to elevated O 3 could be used to better place specific feedstocks on a dynamic landscape.

Field Site, Plant Material and Growth Condition
The study was conducted at the Free Air Concentration Enrichment (FACE) facility in Champaign, IL, USA (www.igb.illinois.edu/soyface/, 40 • 02'N, 88 • 14'W) in 2018. Six plots in octagonal shape of 20 m diameter were designed for this study: three at ambient O 3 concentration (30-50 nmol mol −1 ) and three fumigated to elevated O 3 concentration (~100 nmol mol −1 ). The weather conditions including daily maximum and minimum air temperature, averaged light intensity (9:00-18:00), precipitation, averaged daily relative humidity and O 3 concentration (10:00-18:00) during growing season of 2018 were monitored by an on-site weather station at the FACE facility and shown in Figure 7.
Seedlings octagonal shape of 20 m diameter were designed for this study: three at ambient O3 concentration (30-50 nmol mol −1 ) and three fumigated to elevated O3 concentration (~100 nmol mol −1 ). The weather conditions including daily maximum and minimum air temperature, averaged light intensity (9:00-18:00), precipitation, averaged daily relative humidity and O3 concentration (10:00-18:00) during growing season of 2018 were monitored by an on-site weather station at the FACE facility and shown in Figure 7.  [86] and Yendrek et al. [53,54]. Elevated O3 fumigation was carried out using the O3-enriched air that was delivered to and released within the experimental plots with FACE technology. O3 was generated by an O3 generator (CFS-3 2G; Ozonia) using pure oxygen and monitored by a chemiluminescence O3 sensor (Model 49i,

Leaf Midday Gas Exchange, Chlorophyll Fluorescence and A/C i Curve
In situ midday gas exchange and chlorophyll fluorescence measurements were made on fully expanded leaves between 11:00 and 14:00 on sunny days of 25 July (DOY 206) and 13 August (DOY 225) in 2018. The net CO 2 assimilation rates (A), stomatal conductance to water vapor (g s ), intercellular CO 2 concentration (C i ) and chlorophyll fluorescence (F v '/F m ', Φ PSII , ETR, and qP) under illumination were measured with a portable photosynthesis system (LI 6400, LICOR Biosciences, Lincoln, NE, USA) following previously published protocols [53,87,88]. Briefly, the environmental conditions within the leaf cuvette were set to match ambient conditions: leaf cuvette temperature was 29 • C, CO 2 concentration was 400 µmol mol −1 , light intensity at the leaf surface was 1950 µmol m −2 s −1 and relative humidity was 60% for the DOY 206; leaf cuvette temperature was 31 • C, CO 2 concentration was 420 µmol mol −1 , light intensity at the leaf surface was 1750 µmol m −2 s −1 and relative humidity was 60% for DOY 225, 2018. The measurement was performed when photosynthesis had stabilized, typically 3-5 min after leaf enclosure. Considering the heterogeneity of physiology and structure within a given leaf [89,90], we measured photosynthesis and other functional traits (see below) in the middle part of all leaves. In all cases, 4-5 leaves of different individuals within each plot were measured, and were averaged for analyses. The instantaneous water use efficiency (iWUE) was calculated as A/g s .
Three sun-exposed leaves of different individuals within each plot were selected to measure the response of A to C i using a LI-6400. Predawn on DOY 206, leaves were excised and recut immediately under water to prevent leaf water potential decrease, chloroplast inorganic phosphate concentration or maximum photosystem II efficiency decrease [91]. With the cut end immersed, leaves were quickly transported to the laboratory where they were exposed to ambient CO 2 concentration and saturating light levels to achieve a steady-state. The middle part of the leaf was then enclosed in cuvette and measurements were initiated at a CO 2 concentration of 400 µmol mol −1 , air temperature of 25 • C, light intensity of 1800 µmol m −2 s −1 and relative humidity of 60%. CO 2 concentration within the cuvette was then changed sequentially as follows: 400, 300, 200, 100, 50, 400, 500, 600, 800, 1000, and 1200 µmol mol −1 . The maximum carboxylation capacity of phosphoenolpyruvate (V pmax ) and CO 2 saturated photosynthetic capacity (V max ) were calculated according to Farquhar et al. (1980), von Caemmerer (2000) and Markelz et al. (2011) [92][93][94].

Ball-Woodrow-Berry Relationship
The Ball-Woodrow-Berry (BWB) relationship was calculated as: where g s is stomatal conductance to water vapor (mol (H 2 O) m −2 s −1 ); A is net CO 2 assimilation rate (µmol (CO 2 ) m −2 s −1 ); H s and C s are relative humidity (Pa (air) Pa (Saturated) −1 ) and CO 2 concentration (Pa (CO 2 ) Pa (air) −1 ) at the leaf surface, respectively; and a and b are the slope (mol (H 2 O) mol (CO 2 ) −1 ) and intercept (mol (H 2 O) m −2 s −1 ) of the BWB relationship, respectively [43,48]. H s was calculated as: where E s is the partial pressure (Pa) of vapor at the leaf surface and E sat is the partial pressure (Pa) of vapor at saturation. E s was determined as: where T r is transpiration (mol (H 2 O) m −2 s −1 ) and P is air pressure (Pa). E i is the partial pressure of vapor (Pa) at substomatal cavity and is assumed to be saturated: where λ and R are the latent heat of vaporization and (set at 2,500,000 J kg −1 ) and the gas constant of vapor and (set at 461 J kg −1 K −1 ) [40], respectively, and T l is leaf temperature (K). C s was determined by the following equation: where 1.6 is the ratio of conductance for H 2 O to that for CO 2 and has dimensions of ((mol(H 2 O) m −2 s −1 )/ (mol(CO 2 ) m −2 s −1 )) and C i is intercellular CO 2 concentration (Pa (CO 2 ) Pa (air) −1 ). The values of A, g s , T r , C i , P and T l were observed from a portable photosynthesis analyzer Licor-6400. The slope a and intercept b of the BWB relationship were estimated using linear regression with observed g s and calculated AH s C s .

Dark Respiration and Dark-Adapted Chlorophyll Fluorescence
Leaf respiration rates and chlorophyll fluorescence under dark were also measured using the LI-6400. Immediately after each A/Ci curve was completed, the leaf was removed from the cuvette and kept in the cabinet under dark for at least 50 min. Environmental controls inside the cuvette were maintained to match the ambient conditions: leaf cuvette temperature was 27 • C, CO 2 concentration was 400 µmol mo −1 , relative humidity was 60% but light intensity at the leaf surface was 0 µmol m −2 s −1 . Leaf dark respiration was measured after readings stabilized, typically 3-10 min after leaf enclosure. To examine the effects of elevated O 3 on photosystem II (PS II) activity, dark-adapted chlorophyll fluorescence was measured. Following the respiration rates measurements, the leaf was further illuminated with a saturating irradiance (>7000 µmol m −2 s −1 ) to measure the minimum fluorescence yield (F 0 ) and the maximum dark-adapted fluorescence yield (F m ). The spatially averaged maximum dark-adapted quantum yield of photosystem II (PSII), F v /F m was calculated as the ratio of (F m − F 0 ) to F m .
To measure stomatal density and guard cell length, clear nail polish impressions were collected from abaxial surface of the lamina using other leaf discs from the same leaf sample used for leaf anatomical measurement and viewed and imaged under microscope. The stomatal pore area index (SPI) was calculated as (stomatal density) × (guard cell length) 2 [97].
Using a~2 cm 2 leaf disc from the same leaf sample used for leaf anatomical and stomatal measurement, minor vein density (i.e., minor vein length per leaf area) was determined. After the epidermis was removed with a sharp razor blade, the remaining leaf samples were put in bleach (Clorox Professional Products Company, Oakland, CA, USA) to clear mesophyll cells. The samples were then stained with toluidine blue (Electron Microscopy Sciences, Hatfield, PA, USA) and imaged under microscope. The length minor vein per leaf area was measured with Image J manually.

Biomass, C and N Content and Nutrient Composition Quantification
All plants were harvest on 23 August (DOY 235) in 2018. Three individuals of each plot were selected for the biomass, C and N content measurement. Leaf area was measured by an area meter (LI-2000, LICOR Biosciences, Lincoln, NE, USA) and the number of tillers of each plant was counted. Leaf and stem dry mass were determined after oven-drying for 1 week at 50 • C. Leaf dry mass per area (LMA) was calculated as dry mass/ area. Dried leaf and stem samples were then ground and weighted, and C and N content (%) was determined by a Costech 4010 elemental analyzer (Costech Analytical Technologies, Inc., Valencia, CA, USA).
Macro-and micronutrients were quantified as previously described [98] by inductively high-resolution coupled plasma mass spectrometry (Element 2 TM , Thermo Scientific). Briefly, samples were submitted to a microwave acid sample digestion (Multiwave ECO, Anton Paar, les Ulis, France) (1 mL of concentrated HNO 3 , 250 µL of H 2 O 2 and 900 µL of Milli-Q water for 40 mg DW). All samples were previously spiked with two internal standard solutions of gallium and rhodium for final concentrations of 10 and 2 µg L −1 . After acid digestion, samples were diluted to 50 mL with Milli-Q water to obtain solutions containing 2.0% (v/v) of nitric acid, then filtered at 0.45 µm using a teflon filtration system. Quantification of each element was performed using external standard calibration curves. The quality of mineralization and analysis were checked using a certified reference material of Citrus leaves (CRM NCS ZC73018, Sylab, Metz, France). Isotopic analysis of C and N was performed with a continuous flow isotope mass spectrometer (Isoprime, GV Instruments, Manchester, UK) linked to a C/N/S analyser (EA3000, EuroVector, Milan, Italy).

Statistical Analysis
The differences in physiological and structural traits between ambient and elevated O 3 were tested with one-way ANOVA followed by the Tukey's post hoc test using SPSS 16.0 (SPSS, Chicago, Illinois, USA). The differences in slope and intercept of BMB relationship (observed stomatal conductance vs. AH s C s ) between ambient and elevated O 3 were tested with standardized major axis tests using SMATR v2.0 [99]. All statistical tests were considered significant at p < 0.05.