Integrating O3 influences on terrestrial processes: photosynthetic and stomatal response data available for regional and global modeling

Plants have a strong influence on climate by controlling the transfer of carbon dioxide and water between the biosphere and atmosphere during the processes of photosynthesis and transpiration. Chronic exposure to surface ozone (O3) differentially affects photosynthesis and transpiration because it damages stomatal conductance, the common link that controls both processes, in addition to the leaf biochemistry that only affects photosynthesis. Because of the integral role of O3 in altering plant interactions with the atmosphere, there is a strong motivation to incorporate the influence of O 3 into regional and global models. However, there are currently no analyses documenting both photosynthesis and stomatal conductance responses to O 3 exposure through time using a standardized O 3 parameter that can be easily incorporated into models. Therefore, models often rely on photosynthesis data derived from the responses of one or a few plant species that exhibit strong negative correlations with O 3 exposure to drive both rates of photosynthesis and transpiration, neglecting potential divergence between the two fluxes. Using data from the peer-reviewed literature, we have compiled photosynthetic and stomatal responses to chronic O 3 exposure for all plant types with data available in the peer-reviewed literature as a standardized function of cumulative uptake of O3 (CUO), which integrates O 3 flux into leaves through time. These data suggest that stomatal conductance decreases ∼ 11 % after chronic O3 exposure, while photosynthesis independently decreases ∼ 21 %. Despite the overall decrease in both variables, high variance masked any correlations between the decline in photosynthesis or stomatal conductance with increases in CUO. Though correlations with CUO are not easily generalized, existing correlations demonstrate that photosynthesis tends to be weakly but negatively correlated with CUO while stomatal conductance is more often positively correlated with CUO. Results suggest that large-scale models using data with strong negative correlations that only affect photosynthesis need to reconsider the generality of their response. Data from this analysis are now available to the scientific community and can be incorporated into global models to improve estimates of photosynthesis, global landcarbon sinks, hydrology, and indirect radiative forcing that are influenced by chronic O 3 exposure.


Introduction
Surface ozone (O 3 ) is a greenhouse gas that has increased by 30-60 % globally since industrialization (Karnosky, 2005), resulting in large direct economic costs because O 3 is a strong oxidant that causes foliar damage and reductions in crop yield (Skärby et al., 1987;Mortensen, 1992;Ashmore, 2005;Morgan et al., 2006;Feng et al., 2008).Fossil fuel combustion and industrial processes have increased atmospheric concentrations of nitrogen oxides (NO x ) and volatile organic compounds (Denman, 2007) that produce O 3 through photochemical reactions.Ozone concentrations are predicted to continue increasing in polluted regions by 1-10 ppb in coming decades due to climate change (Jacob and Winner, 2009).At the same time, higher water vapor concentrations in future climates are expected to decrease global background O 3 concentrations (Jacob and Winner, 2009), Published by Copernicus Publications on behalf of the European Geosciences Union.
with expected summertime decreases in the United States of 2-15 ppb (Wu et al., 2008).
Overall photosynthetic reductions after chronic O 3 exposure are common in many types of plants, though stomatal responses are more variable.Meta-analyses of trees (Wittig et al., 2007), wheat (Feng et al., 2008), soybeans (Morgan et al., 2003) and studies comparing plants from multiple functional groups (Reich and Amundson, 1985;Volin et al., 1998) all suggest approximately a 20 % average decrease in photosynthesis after chronic exposure (typically defined as exposures longer than seven days) to O 3 .In contrast, average stomatal responses to O 3 exposure show decreases of 6-10 % in trees (Wittig et al., 2007) and ∼ 20 % in crops (Morgan et al., 2003;Feng et al., 2008), suggesting that stomatal conductance does not always respond similarly to photosynthesis and that responses are variable among plant types.Ozone exposure can also have inconsistent effects on leaf water loss.Several studies have reported that canopy evapotranspiration decreases with O 3 exposure (Bernacchi et al., 2011;VanLoocke et al., 2012), while others suggest that chronic O 3 exposure results in sluggish stomatal responses in many plants types (Mills et al., 2009;Paoletti and Grulke, 2010), (2) CO 2 travels through the intercellular spaces to the chloroplasts (mesophyllic processes); and (3) inside the chloroplasts, CO 2 is changed into sugars through biochemical carbon fixation.At the leaf-level, the process of transpiration (b) is only controlled by stomatal conductance (1).Stomatal conductance is a common mechanism that controls both photosynthesis and transpiration, so these processes are often closely coupled and will continue to be coupled if O 3 primarily effects stomatal conductance.However, experimental evidence shows that O 3 damage primarily effects biochemical carbon fixation (3) and mesophyllic processes (2), leading to a decoupling of photosynthesis and transpiration.
and can increase leaf-level and ecosystem-scale transpiration rates (Paoletti, 2005;McLaughlin et al., 2007).Despite differences in photosynthetic and stomatal responses to O 3 , many modeling studies assume a fixed relationship between photosynthesis and stomatal conductance (Ollinger et al., 1997;Felzer et al., 2004;Sitch et al., 2007) and typically modify photosynthesis values using data from four or fewer species per plant functional type.Simulating O 3 changes to photosynthesis in this manner typically overestimates the decrease in stomatal conductance (Lombardozzi et al., 2012a).If stomatal conductance decreases less than photosynthesis during O 3 exposure as suggested by experiments, then hydrologic changes regulated by the biosphere, including precipitation, latent heat flux and surface runoff, are incorrectly predicted in most current model formulations.Predictions of O 3 damage and impacts on carbon and water cycling can be improved by separately incorporating photosynthetic and stomatal responses to chronic O 3 exposure (Lombardozzi et al., 2012b).However, few data are available documenting responses of stomatal conductance to O 3 exposure through time, and large-scale models use specific photosynthetic responses of plants sensitive to O 3 exposure rather than a photosynthetic response generalized over multiple plant species.
Plant responses to chronic O 3 exposure through time depend on the concentration of O 3 , length of exposure time, and the vulnerability of the plant to O 3 .Plant vulnerability is a function of both stomatal conductance, which can limit the amount of O 3 entering the leaf, and antioxidant defenses within the leaf, which prevent oxidative damage by scavenging O 3 (Dizengremel et al., 2008).One metric of standardizing plant responses to chronic O 3 exposure is to calculate cumulative uptake of O 3 (CUO), which integrates O 3 flux into the leaf through time.Because CUO accounts for changes in stomatal conductance and therefore describes the amount of O 3 that enters the leaf, it estimates the ability of O 3 to oxidize biochemical components of the leaf that regulate photosynthesis and stomatal conductance.A flux threshold is sometimes used in CUO calculations to account for the ability of the plant to detoxify using antioxidants found within plant leaves.However, CUO does not directly account for internal defenses that can vary widely by plant and assumes that plants exposed to high concentrations for short periods of time will have similar CUO to plants exposed to lower concentrations for longer periods of time.Regardless, several studies have demonstrated strong negative correlations between CUO and photosynthesis (Reich, 1987;Wittig et al., 2007), biomass (Karlsson et al., 2004;Uddling et al., 2004) or relative crop yield (Pleijel et al., 2002;Pleijel et al., 2004), making CUO a better metric for predicting plant responses to O 3 than concentration or cumulative O 3 exposure (Reich, 1987;Wieser, 1997;Musselmann and Massmann, 1999;Matyssek et al., 2007).To our knowledge, there is no comprehensive analysis that documents responses of stomatal conductance to CUO in multiple plant types.Additionally, most available CUO data are based on responses of temperate deciduous trees and crops, though other plant functional types may respond differently.
The objective of this work was to determine generalized responses of photosynthesis and stomatal conductance to chronic O 3 exposure in multiple plant types, using CUO as an index for damage to plant physiological processes.This study improves upon work presented in previous syntheses, which document specific plant type responses to O 3 concentrations (Morgan et al., 2003;Wittig et al., 2007;Feng et al., 2008), because it determines the physiological responses to chronic O 3 exposure as a function of both O 3 concentration and time in multiple plant functional types.We collected all available data from the peer-reviewed literature to determine whether photosynthesis and stomatal conductance responded differently to chronic O 3 exposure, and whether responses varied by plant functional types.Modeling CUO using continuous environmental data is the ideal method to estimate O 3 uptake and damage (e.g.Emberson et al., 2000;Karlsson et al., 2004;Mills et al., 2011;Pleijel et al., 2004).However, simultaneous high-resolution environmental data and O 3 concentrations are rarely presented in conjunction with response variables within manuscripts.In our assessment of the literature, only eleven publications that documented photosynthetic and stomatal responses to O 3 also calculated high resolution CUO data, so an alternative method for calculating CUO was needed to determine generalized responses across many plant species and functional types.CUO was calculated using stomatal conductance, O 3 concentration, and exposure duration data available within the literature (see Appendix A).Through incorporating all available information in the literature, these data will provide a comprehensive data set for use in models that will greatly expand the number of species that are represented in modeled photosynthetic and stomatal responses to O 3 exposure.

Data collection
A database documenting the effects of O 3 on photosynthesis and stomatal conductance was compiled by surveying peerreviewed literature using key word searches in the Web of Science (ISI).A total of 156 papers published between 1970 and June 2011 that chronically (> 7 days, consistent with methods used by Morgan et al., 2003, Wittig et al., 2007) exposed plants to O 3 using growth chambers, open-top chambers, branch chambers, or other fumigation methods contained data sufficient to calculate CUO (see Eqs. 1 and 2) and reported changes in photosynthesis and stomatal conductance.Individual measurements within papers were considered independent if they were not previously reported and were measured for different species or genotypes or on different days, similar to methods used by Wittig et al. (2007).Data from papers were eliminated if O 3 concentration or exposure duration used to calculate CUO were unclear; if both control and treatment responses were not reported; if photosynthesis was not reported in conjunction with stomatal conductance; if units of stomatal conductance or photosynthesis were not reported per leaf area; if the type of gas that stomatal conductance was measured for was unclear; if other environmental interactions were included so that the effect of just O 3 was unclear; or if studies were shorter than seven days.In total, data were collected from 108 papers representing 79 species providing sample sizes of 670 for photosynthesis and 772 for stomatal conductance for analyses.A list of publications used in this analysis is provided in Appendix A and data are available upon request.
Values of photosynthesis, stomatal conductance, V cmax , which reflects the amount and activity of enzymes used in biochemical carbon fixation and is often damaged with O 3 exposure (Calatayud et al., 2010;Cardoso-Vilhena et al., 2004;Farage and Long, 1999;Feng et al., 2008;Fiscus et al., 2005;Noormets et al., 2001;Ojanperä et al., 1998;Pellegrini et al., 2010;Zheng et al., 2002), and CUO (if available) from control and elevated O 3 treatments were collected from tables, figures, and text within papers and compiled into a database.If data were presented graphically, data were extracted using PlotDigitizer (PlotDigitizer 2.5.1,Sourceforge, Japan).Additional information about factors that might influence the response of photosynthesis and  1).
Confidence levels for CUO calculations were assigned based on the quality of data presented in the publication.Data were assigned high confidence if CUO was presented in the publication or medium confidence if the publication contained multiple stomatal conductance measurements throughout the course of the experiment, allowing calculations to account for changes in stomatal conductance through time and resulting in more accurate calculations of CUO.Data were assigned low confidence if only an end-point stomatal conductance measurement was reported or if any assumptions were made about duration of O 3 exposure.
Vulnerability was estimated using the ratio of photosynthesis to stomatal conductance in control plants, assuming that higher stomatal conductance allowed higher flux of O 3 into the leaf (Reich and Amundson, 1985) and that leaf internal defense was a function of photosynthesis (Massman, 2004).Volin et al. (1998) found a strong positive correlation between reduction in biomass due to O 3 and the ratio of photosynthesis to stomatal conductance.Therefore, a plant with low photosynthesis relative to stomatal conductance (i.e., photosynthesis:conductance < 0.02 µmol CO 2 mol −1 H 2 O) was considered vulnerable because it was susceptible to higher CUO relative to its ability to defend itself against O 3 damage internally.Ratios of photosynthesis to stomatal conductance (photosynthesis : conductance) were grouped into quartiles and data with high values assigned to low vulnerability (high photosynthesis per unit conductance, > 0.06 µmol CO 2 mol −1 H 2 O) and low values assigned to high vulnerability (high conductance per unit photosynthesis, < 0.02 µmol CO 2 mol −1 H 2 O).

CUO calculations
If available, CUO was collected from tables, figures, or text in publications.Otherwise, CUO was estimated using data contained within the papers.From each publication, stomatal conductance and cumulative exposure to O 3 (CEO 3 ) was recorded.Cumulative O 3 exposure was either presented in the paper (i.e., AOT00, SUM00, etc.) or calculated as where H was the number of daytime hours the plant was exposed to elevated O 3 ; D was the total number of days; and [O 3 ] is the external O 3 concentration in ppb that plants were exposed to during daytime hours.AOT00 and SUM00 values are calculations of O 3 concentrations above 0 ppb over total exposure time and are therefore equivalent to calculations of CEO 3 ; AOT and SUM values with thresholds higher than 0 ppb (e.g., AOT40 or SUM06) were not used in this analysis.Cumulative uptake of O 3 was calculated using CEO 3 (similar to Reich, 1987;Nunn et al., 2006;Wittig et al., 2007;Lombardozzi et al., 2012a): where g s is the mean leaf-level stomatal conductance in units of mol H 2 O m −2 s −1 ; k O 3 = 1.67 and is the ratio of leaf resistance for O 3 to leaf resistance for water vapor (Sitch et al., 2007;Wittig et al., 2007); 3600 is the number of seconds per hour; and 10 −6 is the conversion from nmol to mmol.Because stomatal conductance likely changes over the duration of the experiments, CUO calculations were adjusted to account for these changes when papers reported multiple measurements of stomatal conductance.For example, when stomatal conductance values were presented at day 10 and day 20 of an experiment, uptake was calculated between days 0 and 10 using the mean stomatal conductance value from day 10, and uptake between days 10 and 20 was calculated using the stomatal conductance value from day 20.The uptake from days 0 through 10 was added to the uptake from days 10 through 20 to get a CUO value at day 20.This process was repeated at every time point for each paper reporting multiple stomatal conductance values, allowing for the most accurate CUO estimates possible with the available data by accounting for changes in stomatal conductance throughout each experiment.Calculations of CUO using this method were strongly correlated to CUO presented within manuscripts (r 2 = 0.92, Fig. B1).

Statistical analyses
A linear regression framework was used to analyze data for relationships between change in photosynthesis or stomatal conductance relative to plants exposed to little or no O 3 (% of control) with CUO (an indicator of O 3 plant physiological damage through time) using the generalized linear model (glm) function in R © version 2.11.1.Both response variables were normally distributed, though graphs depict logtransformed CUO for visual purposes.Linear and log-linear relationships between the variables and CUO were tested, though linear relationships were almost always stronger and used in all analyses.Linear models were constructed to test relationships using individual and combinations of factors (listed above) and their interactions, with models having p values ≤ 0.05 considered significant.The best combination of predictors that explained the relationships between photosynthesis or stomatal conductance and CUO was selected using r 2 values.In addition to linear analyses, overall means were compared using Student's t tests.3 Results

All data
Across all data, average values for photosynthesis and stomatal conductance significantly decreased after chronic O 3 exposure (p < 0.001 for each, data not shown); photosynthesis decreased by 21 %, which was significantly greater than the 11 % decrease in stomatal conductance (p < 0.001).Additionally, there was a weak but significant positive relationship between photosynthesis and CUO (Fig. 2a; r 2 = 0.009, p = 0.01) that was strongly influenced by a single study measuring guava responses to O 3 .There was no correlation between stomatal conductance and CUO (Fig. 2b; p = 0.11).Though photosynthesis is positively correlated with CUO, photosynthesis values were negative in response to chronic O 3 exposure until a CUO of ∼ 538 mmol m −2 ; at CUO values above ∼ 538 mmol m −2 , which were only calculated in two studies, photosynthesis was stimulated by O 3 exposure.Removing the single study measuring guava responses to O 3 changed the results, leading to no significant correlation between photosynthesis and CUO (p = 0.38; data not shown).
Since V cmax data were only available for a small subset of studies, data were not separated into sub-categories.Overall, V cmax responded similarly to photosynthesis responses documented in the same study (Fig. 3; p = 0.41).Within these studies, neither V cmax nor photosynthesis was significantly correlated with CUO (V cmax : p = 0.31; photosynthesis: p = 0.41), though mean responses of both significantly decreased from control values (data not shown, p < 0.001).

Data subsets
While several sources of variation were considered in these analyses, two were considered important to remove from the analyses.One was the confidence in the CUO calculations.CUO data presented within a paper, or "high" confidence data, were only available for 11 studies, and there was no correlation between CUO and photosynthesis or stomatal conductance for high-confidence data (Fig. 4).Calculations of CUO in data categorized as "low" confidence relied on a single stomatal conductance value measured at the end of an experiment.Since stomatal conductance decreases in response to O 3 , using the stomatal conductance value from the end of the experiment results in calculations of CUO that are likely smaller than the CUO the plants actually experienced.Therefore, analyses below remove all low-confidence data and instead subset the data to use only medium-and high-confidence data.
The second important source of variation considered in analyses was whether control plants were exposed to ambient air, which had background levels of O 3 specific to a particular location, or charcoal-filtered (CF) air, which removed nearly all O 3 from the environment.It is essential to separate analyses based on whether control plants were exposed to CF or ambient air because responses of photosynthesis and stomatal conductance in this analysis are made relative to a control treatment.Since ambient air varies by location, data presented within the rest of this manuscript focuses on responses relative to CF air so that comparisons are made to plants exposed to similar O 3 concentrations.Comparisons of elevated O 3 to ambient air are presented in Tables B1 and B2.It should be noted that all free-air enrichment exposure systems compare responses to ambient control air, so data collected using this type of exposure system are removed from further analyses.When data were subset to remove low-confidence data and studies using ambient air, average photosynthesis decreased by ∼ 18 % and average stomatal conductance decreased by 16 % (Tables 2 and 3).There was a weak negative correlation between photosynthesis and CUO (Table 2; r 2 = 0.02, p = 0.01), but no correlation between stomatal conductance and CUO (Table 3; p = 0.63).Photosynthesis with large numbers of data points within categories of plant type (crop, n = 134), plant age (< 1, n = 234), O 3 exposure system (open-top chamber, n = 146), and rooting environment (pot, n = 271) were all weakly negatively correlated with CUO, with regression equations for each very similar to the overall regression (Table 2).There was a strong negative relationship between photosynthesis and CUO in plants grown in the ground (r 2 = 0.17, p < 0.001), though no correlations between CUO and photosynthesis within any other subcategory.
There were strong positive correlations between stomatal conductance and CUO in plants exposed to O 3 in growth chambers (r 2 = 0.12, p < 0.001) and plants with low vulnerability to O 3 (r 2 = 0.23, p < 0.001).There was also a strong positive correlation between stomatal conductance and CUO in temperate evergreen trees (r 2 = 0.32, p < 0.001), though this trend was largely driven by four data points above CUO values of 50 mmol m −2 .There was a strong negative correlation between stomatal conductance and CUO in plants older than 5 yr (r 2 = 0.34, p = 0.02), though data were limited in this category (n = 15) (Table 3).O 3 concentration also did not influence the correlations between CUO and photosynthesis or stomatal conductance (Fig. 5).
When plant functional types were considered, mean photosynthesis was lower in crops (Table 2; p = 0.05) but did not differ among other plant functional types.Crops are the only plant functional type where photosynthesis was significantly, though weakly, correlated with CUO (Table 2, Fig. 6c; r 2 = 0.08, p < 0.001).Mean stomatal conductance was significantly higher in temperate deciduous trees (Table 3; p = 0.02) and was significantly lower in crops (Table 3; p = 0.007), as compared to the overall mean.Stomatal con-ductance was positively correlated with CUO in temperate evergreen trees (Table 3, Fig. 6e; r 2 = 0.32, p < 0.001), but correlations were not significant for other plant types.Data pooled by O 3 concentrations demonstrated similar patterns (Fig. 7), with no strong correlations between O 3 concentration and photosynthesis or stomatal conductance in any plant functional type.Data presented in Figs. 6 and 7   Very few studies (9 of 108) contained data for other plant functional types (Table 1).Since correlations were either weak or not evident in plant functional types within the data set that removed both low-confidence data and studies using ambient air, smaller, more specific subsets were created.Due to the small number of data that fall within each category, the subsets focused on categories with the largest number of data: temperate deciduous trees and crops grown in pots that were exposed to 100-125 ppb O 3 .Within this subset, photosynthesis and stomatal conductance in temperate deciduous trees that were 1 to 5 yr old were not correlated to CUO (pho-tosynthesis: Fig. 8a, r 2 = 0.02, p = 0.40; stomatal conductance: Fig. 8c, r 2 = 0.003, p = 0.71).Similarly, photosynthesis in crops less than one year old (grown in pots, exposed to 100-125 ppb O 3 ) was not correlated with CUO (Fig. 8b,  r 2 = 0.004, p = 0.53).However, stomatal conductance for the same subset of crops was positively correlated with CUO (Fig. 8d, r 2 = 0.29, p = 0.004).Though our generalized responses largely do not demonstrate negative correlations between photosynthesis and CUO, several individual studies report negative correlations between photosynthesis and CUO.To determine how easily the negative relationship was lost, data from a single study was considered, and additional data added sequentially.In a single study measuring the response of wheat to chronic O 3 exposure that reported CUO values (plants > 1 yr old were grown in pots, compared to CF air, and exposed to 100 ppb O 3 in a growth chamber), photosynthesis and stomatal conductance were both negatively correlated with CUO (Fig. 9, open circles; photosynthesis: r 2 = 0.59, p < 0.001; stomatal conductance: r 2 = 0.20, p = 0.05).When other studies measuring responses of wheat to O 3 under similar conditions (plants > 1 yr old were grown in pots, compared to CF air, and exposed to 100 ppb O 3 in a growth chamber) are included, there is no correlation between photosynthesis and CUO (Fig. 9a; p = 0.59) and the correlation between stomatal conductance and CUO becomes positive (Fig. 9b; r 2 =0.20, p = 0.002).

Discussion
Chronic O 3 exposure causes the plant physiological processes of photosynthesis and stomatal conductance to change, though how these processes change through time is not well known for many plant functional types.The results of this analysis suggest that chronic O 3 exposure depressed leaf-level photosynthesis and stomatal conductance in all plant types, ages, rooting environments and estimated sensitivities, though the variability within these data was too large to determine relationships with CUO.Overall, the 21 % average decrease in photosynthesis was larger than the 11 % average decrease in stomatal conductance, demonstrating that these variables respond differently to chronic O 3 exposure.A meta-analysis of tree responses to O 3 compared to charcoal-filtered air found a 19 % decrease in photosynthesis and a 10 % decrease in stomatal conductance (Wittig et al., 2007), similar to the means reported here.The difference between average photosynthesis (−18 %) and average stomatal conductance (−16 %) was less dramatic when low-confidence data and comparisons to ambient air were removed from the data set, but accounting for these factors did not strongly improve the correlations of either variable with CUO.This work additionally found that decreases in photosynthesis were typically larger than the decreases in stomatal conductance in most plant type, age, rooting environment, exposure system, O 3 concentration, and estimated vulnerability categories, though the magnitude of decrease in each variable differed based on the category.
To our knowledge, this is the first study that widely describes the relationship of stomatal conductance to chronic O 3 exposure through time, despite the known importance of stomatal conductance responses to O 3 on ecosystem-scale water dynamics like soil water content and stream flow (McLaughlin et al., 2007).This study found that stomatal conductance decreased less than photosynthesis after chronic O 3 exposure, but, unexpectedly, there was no strong correlation between stomatal conductance or photosynthesis and CUO in the overall data or in data subsets (Fig. 2, Tables 2  and 3).The only physiological responses that can be easily generalized from these data are that photosynthesis and stomatal conductance decrease after chronic O 3 exposure, suggesting that models incorporating negative correlations between photosynthesis and O 3 need to reconsider the generality of their responses.While calculations of CUO were not independent of stomatal conductance, there was also no consistent decrease in stomatal conductance based on O 3 concentration (Fig. 5), similar to results found by Wittig et al. (2007).Ozone concentration also did not affect the decreases in photosynthesis in a systematic manner (Fig. 5) in our study or in results found by Wittig et al. (2007).
Initial O 3 damage usually decreases enzyme activity and RuBP regeneration (Fiscus et al., 2005) and is apparent in V cmax and J max reductions after chronic O 3 exposure.Accounting for decreases in V cmax and/or J max is one CUO is shown on a log scale, but linear analyses (r 2 and line equations) were performed on non-log transformed data.r 2 values are only included for significant correlations, and n values are the number of data points included in the analyses.
mechanistic approach that physiological models can use to estimate O 3 damage to photosynthesis (Martin et al., 2000;Lombardozzi et al., 2012b), though little data are available to parameterize models in this way.Here, we chose to focus on decreases in V cmax due to the higher availability of data.In the available data, average V cmax followed similar trends as in photosynthesis and decreased in response to chronic O 3 , similar to V cmax measured in other studies (Cardosa-Vilhena et al., 2004;Calatyud et al., 2010;Kellomäki et al., 1997;Löw et al., 2007), though was not significantly correlated with CUO (Fig. 3).Cardosa-Vilhena et al. ( 2004) likewise found that decreases in V cmax were not linearly correlated with CUO in wheat leaves exposed to O 3 .Responses of V cmax were not significantly different from photosynthetic responses, suggesting that either can be used to estimate photosynthetic decreases.However, photosynthesis responses can be used in models with more confidence given the lack of data estimating responses of V cmax to chronic O 3 exposure.
Although we observed a similar overall decrease in photosynthesis, the lack of correlation between photosynthesis and CUO was different from the strong negative correlations found by Reich (1987) and Wittig et al. (2007) and was contrary to parameterizations used in many models incorporat-ing the effects of O 3 on leaf-level physiology (e.g.Ollinger et al., 1997;Ollinger et al., 2002;Felzer et al., 2004;Felzer et al., 2005;Sitch et al., 2007;Ren et al., 2011).Demography of the data might play a role in the difference as the present study had a larger sample size (n = 670) and incorporated multiple types of plants, though temperate trees (evergreen and deciduous) comprised 58 % of these data and correlations were weak or non-existent when separated by plant type.While the methods used to calculate CUO were similar in all studies, the present study imposed strict standards on collected data to maintain confidence in CUO calculations and therefore rejected some data that were used by Wittig et al. (2007).It also seems likely that the differences between these studies is in part because the method used in the present study for calculating CUO accounted for changes in stomatal conductance through time whenever possible and resulted in a strong correlation with CUO presented in publications (Fig. B1, r 2 = 0.92), which was not factored into CUO calculations in other studies.
Most studies reporting responses of photosynthesis and stomatal conductance to chronic O 3 exposure do not provide CUO or fine-scale measurements of stomatal conductance to allow for accurate estimations of CUO, making it difficult to document responses of photosynthesis and stomatal conductance over a broad range of CUO.While statistical and modeling techniques have improved the accuracy of CUO by simulating hourly stomatal conductance values that can be incorporated into CUO calculations, (e.g.Grulke et al., 2002;Pleijel et al., 2002;Karlsson et al., 2004;Pleijel et al., 2004;Uddling et al., 2004), those techniques could not be employed in the present study due to the limitation of available environmental data within the studies.Though the accuracy of CUO calculations can be improved by accounting for hourly stomatal conductance and O 3 concentration, our methods of calculating CUO were strongly correlated with studies that presented CUO within their publication (r 2 = 092; Fig. B1), suggesting that any error in CUO calculations was systematic and did not influence correlations.Even combining the few studies that provided CUO values (high-confidence data) did not result in correlations between CUO and stomatal conductance or photosynthesis (Fig. 4).To improve the confidence of our findings, however, further analyses focus on a subset of data that contain only high-and medium-confidence data so that all CUO calculations account for the changes in stomatal conductance through time.
Plant functional type did not have a strong effect on the relationship of photosynthesis or stomatal conductance with either CUO (Fig. 6) or the O 3 concentration that plants were exposed to (Fig. 7).Wittig et al. (2007) found photosynthesis decreases that were similar in both angiosperms and gymnosperms in experiments that artificially increased O 3 concentration compared to charcoal-filtered control air, similar to the mean decreases found in this study (p = 0.66).Further, Reich and Amundson (1985) found that crop photosynthesis was more negatively affected than temperate evergreen trees after chronic O 3 exposure, also similar to our results (p = 0.05).Data used in this analysis additionally suggest that the decrease in mean crop stomatal conductance was 16 % larger than stomatal conductance in temperate deciduous trees.This result is consistent with responses of stomatal conductance determined in meta-analyses of multiple tree types and wheat; tree stomatal conductance decreased by 10 % (Wittig et al., 2007) and wheat stomatal conductance decreased by 22 % (Feng et al., 2008).
Manipulating the atmosphere around adult plants is difficult, making it hard to assess physiological responses of mature plants to chronic O 3 exposure.Given these challenges, literature data are most scarce for mature trees (in plants older than 5 yr, stomatal conductance n = 15, Table 3; photosynthesis n = 7, Table 2), which make comparisons across plant ages challenging and modeling responses of different age classes uncertain.Within the published literature, a few studies found that O 3 decreased photosynthesis in mature trees more than juvenile trees (Edwards et al., 1994;Rebbeck et al., 1993), while Grulke and Miller (1994) conversely found that sensitivity to O 3 decreases as the age of trees increases.Results in the present study, which are generalized over several plant species, demonstrate no overall photosynthetic differences between mature and juvenile trees in response to O 3 (Table 2, p = 0.93), similar to measured responses in Ponderosa pine trees (Momen et al., 1997).
Though photosynthetic correlations with CUO were not significant in mature plants (i.e., older than 5 yr; p = 0.8), similar to responses of mature giant sequoia (Grulke et al., 1996), stomatal conductance in mature plants decreased with CUO (r 2 = 0.34, p = 0.02).This strong negative relationship contrasts with the variable mean stomatal responses of mature trees documented by Karnosky et al. (2007) and suggests that CUO can help clarify stomatal responses in mature trees.
Though several individual studies find negative correlations between photosynthesis and CUO, driving the assumption in models that photosynthesis decreases linearly with CUO, the relationship is lost when generalizing across plant species (Fig. 8) and even within the same plant species (Fig. 9) when plant age, O 3 concentration, rooting environment and exposure system were similar.These results demonstrate that correlations with CUO can be difficult to document, even within a seemingly narrow range of variability.For example, photosynthesis in temperate deciduous trees from 1 to 5 yr old, and in crops that were less than one year old was not correlated with CUO (Fig. 8a and  b; p = 0.4, 0.53).However, stomatal conductance was positively correlated with CUO in crops less than one year old (Fig. 8d; r 2 = 0.29, p = 0.004), suggesting that the mechanisms regulating stomatal closure in crops become damaged with chronic O 3 exposure, causing stomatal conductance to increase (Freer-Smith and Dobson, 1989;Hassan et al., 1994; Maiermaercker and Koch, 1991;Manes et al., 1998Manes et al., , 2001;;McLaughlin et al., 2007;Mills et al., 2009).The trends are similar within data sets that are even more specific.Though all photosynthesis data used in Fig. 9a decreased in response to O 3 , data from each study was not necessarily negatively correlated with CUO.Two of the studies were conducted by the same author and reported CUO values in the manuscript (circles and triangles in Fig. 9), though photosynthesis was significantly correlated with CUO in only one of these studies (circles, Fig. 9a).While generalized negative correlations between photosynthesis and CUO might have been evident if only studies with negative correlations between photosynthesis and CUO were used, handpicking data in this fashion can lead to biased responses that, when used in models, do not accurately represent generalized responses of plants to chronic O 3 exposure.
Correlations of CUO with photosynthesis or stomatal conductance were not evident within refined data sets that accounted for plant age, type, rooting environment, O 3 concentration, and exposure system (Figs.8 and 9), although Karlsson et al. (2007) found negative correlations between biomass and CUO above a threshold of 1.6 mmol O 3 m −2 in three species of temperate deciduous trees.While direct comparisons for decreases in biomass and photosynthesis cannot be made, the strong negative correlations found by Karlsson et al. (2007) suggest that including an O 3 threshold, which was not possible in this study due to limitation of available data, might improve correlations with CUO.However, Wittig et al. (2007) found negative correlations between photosynthesis and CUO without accounting for an O 3 threshold.Additionally, changes in biomass reflect the whole-plant response to O 3 exposure, whereas measurements of photosynthesis, the focus of this study, only capture leaf-level responses.Ozone-induced plant-level phenological changes, such as the observed linear decreases in LAI with chronic O 3 exposure (Betzelberger et al., 2012), possibly contribute to the difference in plant-level compared to leaf-level response to O 3 .
In addition to its role as a greenhouse gas, O 3 is likely to cause indirect changes in climate (Sitch et al., 2007) through altering carbon and water exchange between the biosphere and atmosphere, yet our ability to predict these responses on ecosystem or global scales is inadequate for several reasons.First, our understanding of plant responses is limited to temperate and crop ecosystems, which cover only 25 % of earth's land surface (Grace, 2004), leading to large uncertainty in global predictions.In the current literature, only three studies report data for tropical tree species or herbaceous plants, and two or fewer studies report data for grasses, and deciduous and evergreen shrubs (Table 1).The lack of available data for these ecosystems forces regional and global models (e.g.Felzer et al., 2004;Sitch et al., 2007) to base responses of tropical and grassland ecosystems on temperate tree and crop data (e.g., Reich, 1987;Karlsson et al., 2004;Pleijel et al., 2004).Further, most models incorporating plant responses to O 3 assume that photosynthesis declines linearly with CUO (e.g.Ollinger et al., 1997;Ollinger et al., 2002;Felzer et al., 2004;Felzer et al., 2005;Sitch et al., 2007;Ren et al., 2011), although the results of this study suggest this assumption is not accurate when generalized across several plant species.Finally, the common assumption that stomatal conductance declines linearly with photosynthesis should also be reconsidered since the responses of 750+ independent measurements collected in this study largely suggests that photosynthesis and stomatal conductance do not change at the same rate or with the same magnitude during chronic O 3 exposure.Simulating robust responses of both photosynthesis and stomatal conductance to chronic O 3 exposure using data collected by this study, rather than responses of a single or few studies, will improve the accuracy of carbon, water, and climate predictions.Nevertheless, global responses www.biogeosciences.net/10/6815/2013/Biogeosciences, 10, 6815-6831, 2013 to O 3 , particularly for grassland and tropical forest ecosystems, will remain uncertain until more data are collected.
In conclusion, both photosynthesis and stomatal conductance decrease in response to chronic O 3 exposure, but there is little ability to predict a generalized response of photosynthesis or stomatal conductance through time.Relationships might be improved if future studies report physiological data in conjunction with CUO that is calculated based on fine-resolution O 3 concentrations and stomatal conductance measurements.Models can continue to use data that contain negative correlations between photosynthesis and CUO, though these responses are not necessarily representative of all plant responses, even within a single species.Data presented here suggest that studies using regional and global models, which generalize many plant physiological parameters, should also consider using generalized physiological responses to O 3 based on the modeled plant functional types.Ultimately, scientists can select data that best represents the needs of the model from data subsets included in this analysis, whether it is a generalized response, responses for particular plant types, or plants of a particular age, and allow the models to incorporate responses of photosynthesis and stomatal conductance independently (see methods in Lombardozzi et al., 2012b).
Additionally, data suggest that linear changes with CUO should be made when there are strong correlations, but do not support applying a linear relationship when there is no correlation between a variable and CUO.Including linear decreases in photosynthesis where data do not support this trend can possibly overestimate the decreases in the global land-carbon sink in the long term, and potentially underestimate the decreases in the short term.Instead, when correlations with CUO do not exist, models should reduce each variable by a specific proportion at every model time step based on the appropriate plant functional type data, regardless of CUO.Though these analyses were not able to incorporate any O 3 concentration or CUO threshold, the use of a threshold can be developed within a model to tune the predicted responses to match observations.Adjusting modeled values of photosynthesis and stomatal conductance independently using generalized data such as these allows for models to better capture the physiological responses of plants to chronic O 3 exposure and can potentially improve the accuracy of predicted responses in global carbon and hydrologic cycles.

Fig. 2 .Fig. 3 .
Fig. 2. The correlation of photosynthesis (open symbols; a) and conductance (closed symbols; b) to CUO across all plant types, ages, and vulnerabilities for all data.Response values are reported as % of control (treatment/control).CUO is shown on a log scale, but linear analyses (r 2 and line equations) were performed on nonlog transformed data.r 2 values are only included for significant correlations, and n values are the number of data points included in the analyses.

Fig. 4 .
Fig. 4. The correlation between cumulative O 3 uptake (CUO) and photosynthesis (open symbols; a) or stomatal conductance (closed symbols; b) in all data that is compared to control plants exposed to charcoal-filtered air where CUO values were reported in the publication (e.g., high-confidence data).

Fig. 6 .
Fig. 6.The correlation of photosynthesis (open symbols; a-c) and conductance (closed symbols; d-f) to CUO for temperate deciduous trees (a, d), temperate evergreen trees (b, e) and crops(c, f) for all data that is compared to control plants exposed to charcoal-filtered air with medium or high confidence in cumulative O 3 uptake (CUO) calculations.Response values are reported as % of control (treatment/control).CUO is shown on a log scale, but linear analyses (r 2 and line equations) were performed on non-log transformed data.r 2 values are only included for significant correlations, and n values are the number of data points included in the analyses.

Fig. 7 .
Fig. 7.The correlation of photosynthesis (open symbols; a-c) and conductance (closed symbols; d-f) to 25 ppb ozone concentration bins for temperate deciduous trees (a, d), temperate evergreen trees (b, e) and crops (c, f) for all data that is compared to control plants exposed to charcoal-filtered air with medium or high confidence in cumulative O 3 uptake (CUO) calculations.Response values are reported as % of control (treatment/control). r 2 values are only included for significant correlations, and n values are the number of data points included in the analyses.

Fig. 8 .
Fig. 8.The correlation of photosynthesis (open symbols; a-b) and conductance (closed symbols; c-d) to CUO for all temperate deciduous trees between one and five years old (a, c) and crops less than one year old (b, d) to chronic O 3 exposure for all plants grown in pots, exposed to 100-125 ppb O 3 in growth chambers, and are compared to control plants exposed to charcoal-filtered air with medium or high confidence in cumulative O 3 uptake (CUO) calculations.Response values are reported as % of control (treatment/control).CUO is shown on a log scale, but linear analyses (r 2 and line equations) were performed on non-log transformed data.r 2 values are only included for significant correlations, and n values are the number of data points included in the analyses.

Fig. 9 .
Fig. 9.The correlation of photosynthesis (a) and stomatal conductance (b) to CUO from all studies that measured responses of wheat plants grown in pots exposed to 100 ppb O 3 in growth chambers and compared to control plants exposed to charcoal-filtered air.CUO is shown on a log scale, but linear analyses (r 2 and line equations) were performed on non-log transformed data.Different symbols represent different studies.Initial data considered (open circles and dashed line) were from a single study and demonstrated negative correlations between CUO and both photosynthesis (r 2 = 0.59, p < 0.001) and conductance (r 2 = 0.19, p = 0.05).Adding data from other studies measuring the responses of wheat plants to O 3 in similar environments (closed symbols) resulted in no overall correlation with CUO, though overall correlations exist in some individual studies (photosynthesis: triangles and dotted line, r 2 = 0.15, p = 0.15; squares and dot-dashed line, r 2 = 0.61, p = 0.06; conductance: triangles and dotted line, r 2 = 0.28, p = 0.03; squares and dot-dashed line, r 2 = 0.004, p = 0.91).The text in the figure and the solid regression line summarizes the information for all data points included in the figure.r 2 values are only included for overall significant correlations, and n values are the number of data points included in the analyses.

Table 1 .
Categories and levels describing the data collected from experiments studying O 3 effects on photosynthesis and stomatal conductance.Each row shows a category and the columns show various levels of that category.All tree categories are temperate unless otherwise noted.Numbers in parentheses are the number of studies and the number of data points within the associated categorical level: (# of studies, n).

Biogeosciences, 10, 6815-6831, 2013 www.biogeosciences.net/10/6815/2013/ D. Lombardozzi et al.: O 3 influences on terrestrial processes 6821Table 2 .
. Correlations between CUO and stomatal conductance were either weak or not significant in other subcategories or in the rooting environment category The number of data points (n), mean, regression, and statistics for photosynthesis when sorted by categories of plant type, plant age, O 3 exposure system, rooting environment, and vulnerability for all data that is compared to control plants exposed to charcoal-filtered air with medium or high confidence in cumulative O 3 uptake (CUO) calculations.All means are reported as the percent of control treatment.The p value for the overall data set (All data) designates whether the overall mean is significantly different from 100 (i.e., 0 % change).The p value for the mean within each category designates whether the mean of the category is statistically different from the mean of the overall data set (All data).All regressions are based on the correlation of photosynthesis to CUO for data within the category and are not included if the relationship is not significant (NS).The r 2 is calculated for each regression, and p values designate whether the relationship between photosynthesis and CUO for data within the category is statistically significant.P values are considered significant at p = 0.05, and significant values are indicated with * .
focused only on plant functional types with abundant data and more than three publications reporting values (see Table1), including temperate deciduous trees, crops and temperate evergreen trees (39, 35, and 13 % of the total data subset, respectively).

Table 3 .
The number of data points (n), mean, regression, and statistics for stomatal conductance when sorted by categories of plant type, plant age, O 3 exposure system, rooting environment, and vulnerability for all data that is compared to control plants exposed to charcoalfiltered air with medium or high confidence in cumulative O 3 uptake (CUO) calculations.All means are reported as the percent of control treatment.The p value for the overall data set (All data) designates whether the overall mean is significantly different from 100 (i.e., 0 % change).The p value for the mean within each category designates whether the mean of the category is statistically different from the mean of the overall data set (All data).All regressions are based on the correlation of stomatal conductance to CUO for data within the category and are not included if the relationship is not significant (NS).The r 2 is calculated for each regression, and p values designate whether the relationship between stomatal conductance and CUO for data within the category is statistically significant.P values are considered significant at p = 0.05, and significant values are indicated with * .