A humidity-based exposure index representing ozone damage effects on vegetation

Surface ozone (O3) is detrimental to plant health. Traditional exposure indexes, such as accumulated hourly O3 concentrations over a threshold of 40 ppb (AOT40), are easy to be derived and widely used to assess O3 damage effects on vegetation. However, the regulation of environmental stresses on O3 stomatal uptake is ignored. In comparison, the dose-based indexes are much more reasonable but require complex parameterization that hinders further applications. Here, we propose a new humidity-based index (O3RH) representing O3 damage effects on vegetation, which can be simply derived using ground-level O3 and relative humidity (RH). Compared with O3 damages to gross primary productivity (GPPd) derived from a process-based scheme over May to October in 2015–2018, the O3RH index shows spatial correlations of 0.59 in China, 0.62 in U.S., and 0.58 (P< 0.01) in Europe, much higher than the correlations of 0.16, −0.22, and 0.24 (P< 0.01) for AOT40. Meanwhile, the O3RH index shows temporal correlations of 0.73 in China, 0.82 in U.S, and 0.81 (P< 0.01) in Europe with GPPd, again higher than the correlations of 0.50, 0.67, and 0.79 (P< 0.01) for AOT40. Analyses of O3RH reveal relatively stable trend of O3 vegetation damages in eastern U.S. and western Europe, despite the long-term reductions in local O3 pollution levels. Our study suggests the substitution of traditional exposure-based indexes such as AOT40 with O3RH for more reasonable assessments of O3 ecological effects.


Introduction
Tropospheric ozone (O 3 ) is a secondary air pollutant generated by photochemical reactions of nitrogen oxide (NO x = NO + NO 2 ) and volatile organic compounds (Atkinson 2000, Kleinman 2005, Jacob and Winner 2009. Ambient surface O 3 concentrations ([O 3 ]) kept increasing by 0.5%-2% yr −1 at the middle latitudes of the Northern Hemisphere over 1970-2000(Vingarzan 2004. Since the 1990s, [O 3 ] decreased in rural areas in North America and Europe (on average 0.23 ppbv yr −1 ) but increased in urban areas worldwide (on average 0.31 ppbv yr −1 ) (Sicard 2020). O 3 exposure (including acute exposure with high [O 3 ] and chronic exposure) leads to foliar injury and reductions in plant productivity (Paakkonen et al 1998, Lombardozzi et al 2012, Yue et al 2017, De Marco et al 2020, which further influence the land carbon budget as well as the climate (Tian et al 2011, Arnold et al 2018, Gong et al 2020.
The intensity of O 3 vegetation damage depends not only on [O 3 ], but also on environmental stresses. For example, the drought conditions with low air relative humidity (RH) and low soil-water contents lead to closure of plants stomata, further reducing stomatal O 3 uptake and O 3 injury (Khan and Soja 2003, Hayes et al 2012, Gao et al 2017. The differences in carbon dioxide concentrations, and nitrogen loads may result in different vegetation responses to O 3 even with the same [O 3 ] (Topa et al 2004, Thomas et al 2006, Mishra et al 2013. Furthermore, differences in plant function types (PFTs) as well as phenological stages also lead to different stomatal O 3 uptakes (Clifton et al 2020b) and the consequent vegetation damages (Sitch et al 2007. As a result, environmental stresses such as air temperature and solar radiation would indirectly regulate O 3 vegetation damages by influencing PFT distribution and plants phenology. To assess the O 3 risks to ecosystem functions, various damaging indexes have been proposed and applied. In general, these indexes can be classified into exposure-based or dose-based groups. The accumulated O 3 over a threshold of 40 ppb (AOT40) is a typical exposure-based index adopted by the Long-Range Transboundary Air Pollution (CLRTAP) Convention to assess the ecological impacts of O 3 (Fuhrer et al 1997, Spranger et al 2004. AOT40 represents the O 3 exposure level using a simplified formula but is insufficient to quantify O 3 -induced vegetation damage since the influences of environmental stresses are not considered (Emberson et al 2000, Mills et al 2011a. Many studies found that the vegetation damage was more determined by 'O 3 uptake fluxes' entering stomata rather than O 3 exposure (e.g. Musselman et al 2006, Karlsson et al 2007, Mills et al 2011a, Bueker et al 2015, De Marco et al 2020, Clifton et al 2020b. As a result, the dose-based index such as phytotoxic ozone dose over a threshold flux of Y nmol m 2 PLA s −1 (POD Y , and PLA is the projected leaf area) is proposed to represent the stomatal flux of ozone. POD Y includes the influences of environmental stresses on stomata and thus describes O 3 damage effects more mechanistically and precisely (Mills et al 2011b).
The key step of estimating POD Y is the calculation of stomatal conductance (g s ), which is generally obtained by two kinds of model: the Jarvis model (Jarvis 1976, Emberson et al 2000, Buckley and Mott 2013 or photosynthesis-stomata (A net -g s ) model (e.g. Farquhar et al 1980, Ball et al 1987. The Jarvis model calculate g s by multiplying PFT-specific maximum g s (provided by published observational data) and a series of factors representing influences of environmental stresses (including temperature, vapor pressure deficit (VPD), soil water content and solar radiation) and phenology (Spranger et al 2004). It is effective to assess O 3 vegetation damages at singlesite level (e.g. Bueker et al 2012Bueker et al , 2015, but the complexity in calculating each factor and the difficulties in obtaining observed input data (such as photosynthetic photon flux density, soil water potential, quasilaminar resistance (r b ) and leaf surface resistance (r c )) limit the application of Jarvis model when upscaling to regional or global scales. The A net -g s model derives g s by coupling photosynthesis rates based on physiological relationships (Clifton et al 2020a). It has been widely applied in dynamic global vegetation models (DGVMs) or land-surface models (Yue andUnger 2015, Sadiq et al 2017), making the largescale evaluation possible but requiring proficient coding skills and high computing resources.
Because of the complexity in deriving POD Y , the exposure-based indexes are still widely used to assess O 3 ecological effects, especially in the atmospheric chemistry community (e.g. Sicard et al 2016, Lin et al 2018, Lu et al 2018, Mills et al 2018, Feng et al 2019, though POD Y is a better metric to assess O 3 ecological effects (e.g. Mills et al 2011a, Shang et al 2017. Karlsson et al (2004) attempted to modify AOT40 as a new index named AOT30 VPD by considering humidity impacts on g s . However, the AOT30 VPD based on subterranean clover was designed to describe the short-term visible ozone injury and thus unable to assess O 3 damages on ecosystem productivity (Spranger et al 2004). In this study, we propose a new index based on DGVMs simulations with A net -g s model to indicate the long-term O 3 damage effects to ecosystem productivity. The new index named O 3 RH has two main advantages: (a) calculations of O 3 RH are as easy as AOT40 and (b) the index can represent spatiotemporal pattern of O 3 damage as efficient as the dose-based method. In particular, we are not denying the advances of POD Y metric, instead we propose the simplified but comparably effective O 3 RH index to facilitate the current assessments of O 3 ecological risks for atmospheric chemistry community.

The Yale Interactive terrestrial Biosphere (YIBs) model
The YIBs model includes nine PFTs and can dynamically simulate vegetation biophysical processes, including leaf photosynthesis (A tot ), respiration, transpiration, phenology, and carbon allocation at the global scale (Yue and Unger 2015). Stomatal conductance (g s ) is dependent on A tot following the Ball-Berry scheme (Farquhar et al 1980, Ball et al 1987: where R d is the respiration rate. RH and C s indicate the RH and CO 2 concentration at the leaf surface, respectively. m and b are PFT-dependent parameters regulating stomatal conductance (see details in Yue and Unger (2015) and Gong et al (2020)

O 3 damage scheme in YIBs
The O 3 damage ratio (F) to the original photosynthesis is calculated as a linear function of stomatal O 3 uptake fluxes (F O3 ) (Sitch et al 2007): where the PFT-specific parameters a and F O3,crit are derived from observations (Sitch et al 2007, Yue andUnger 2015). The parameter a has two sets of values representing varied sensitivities from low to high (see details in Sitch et al (2007) and Gong et al (2020)). F O3 is calculated by the following formula: where [O 3 ] is the ambient O 3 concentration and R a is the aerodynamic resistance. k O3 is set as 1.67 to represent the ratio of leaf resistance for O 3 to leaf resistance for water vapor. The stomatal conductance g s is derived from equation (1). Evaluations showed that this scheme was able to simulate reasonable GPP-O 3 and g s -O 3 relationships for various PFTs (Yue et al 2016, Yue and Unger 2018).

Definition of O 3 damaging indexes
Three widely used indexes, including maximum daily 8 h (MDA8) [O 3 ], AOT40, and POD 1 , are compared for O 3 vegetation damage:  (3)), the latter of which is related to RH (equation (1)), we propose a new RH-based O 3 damage index O 3 RH as follows: where the f(O 3 ) and f(RH) are expressed following the thresholds described in section 3.3: For each coefficient b (b 1 , b 2 , b 3 and b 4 ), the statistical significances (P values) are examined by t-test. Finally, the factor with the minimum P value is determined as the key factor at each grid.

Key factors determine O 3 vegetation damage
Following the multi-linear regression method, figure 1 shows the key factors that dominate O 3induced GPP damages in China, the U.S. and Europe over May to October in 2015-2018. In almost all

A review of O 3 vegetation damage with water stress
We explore the impacts of drought on O 3 vegetation damage from literature (    (3)), the latter of which is positively correlated with RH (equation (1)). For the wet conditions with RH >80%, GPP damages are mainly influenced by [O 3 ], which decreases with increasing RH (figure 3).  different plant species, leading to lower spatial consistency between O 3 RH and GPP damages than that between POD 1 and GPP damages (figures 2(j)-(l)). Figure  . It reveals that the traditional exposure-based index AOT40, which is originally proposed to evaluate the O 3 ecological effects over Europe (Fuhrer et al 1997), should be used with cautions over regions outside Europe due to the missing of regulations by water stress. As a comparison, the new O 3 RH index shows regionally consistent high temporal correlation coefficients (0.73-0.82) with local GPP damages, highlighting the importance of water stress in regulating O 3 -induced vegetation damages.

Application of the O 3 RH index
With the O 3 RH index, we expect to estimate O 3 -induced GPP damages as reasonable as DGVM simulations. Figure S3 compares the O 3 RH with GPP percentage damages (GPP d (%)) over May to October simulated by YIBs model over China, the U.S and Europe. Over these regions, high R 2 from 0.60 to 0.75 are predicted between O 3 RH and GPP d (table S2). Such correlations are much higher than the R 2 of 0.24-0.47 between AOT40 and GPP d , indicating the improvement of O 3 RH in describing the O 3 ecological effects. We further calculate the linear regression between O 3 RH and GPP d (%) as follows: The average slope of 0.65 is accompanied by a range of uncertainties from 0.51 to 0.77 for different years at different regions (table S2) 4(b)). However, trends of O 3 RH are moderate and insignificant for both eastern U.S. and western Europe (figure 4(c)), in part attributed to the positive trends in RH (not shown). As a result, estimates based on O 3 RH suggest that declining [O 3 ] failed to bring expected lower O 3 vegetation damages in these two regions. Such conclusion is consistent with regional studies using observations (Ronan et al 2020) and DGVMs (Yue et al 2016).  (1980( -2018( , blue lines) and western Europe (1990. The locations of observational sites are shown in figure S1. Colored shades indicate the ranges between the first and third quartiles of the site samples in each region. Trends significant at the P < 0.01 level are indicated by the asterisks.

Discussions
In this study, we proposed a new humidity-based index O 3 RH to simplify the calculation of O 3 vegetation damage but with comparable accuracy to dosebased indexes or simulations by DGVMs. However, some studies also reported the sluggish effects that stomata lost functions under O 3 exposure (Paoletti and Grulke 2010, Hoshika et al 2014, Lombardozzi et al 2015, which is difficult to be represented in Sitch et al (2007) scheme. The extraordinary high correlations between GPP damages and POD 1 (figures 2(j)-(l)) are also attributed to the F O3 dependent scheme even though most filed experiments show correlations coefficients between vegetation damages and POD Y generally lower than 0.8 (Bueker et al 2015, Convention et al 2017.
Despite these uncertainties, we demonstrate that the O 3 RH index is a simplified but effective way to assess regional O 3 vegetation damages. We suggest the substitution of traditional AOT40 with O 3 RH to account for the regulation by water stress. Analyses using O 3 RH show that O 3 vegetation damages continue increasing in China and remain stable in eastern U.S. and western Europe during the past several years and decades. Such trends pose a long-lasting threat by surface O 3 to global ecosystems.

Conclusion
To overcome the poor spatiotemporal representations of traditional O 3 exposure indexes and disadvantages of dose-based indexes in complex and skilled calculations, a new humidity-based index O 3 RH was proposed in this study to better assess the O 3 ecological effects. We firstly selected RH and [O 3 ] as the key factors determining the magnitudes of O 3 -indced vegetation damages by multi-linear regressions, and then explored the relationships among [O 3 ], RH and GPP damages with the help of YIBs model. The simulation as well as field experiments from literature both supported that moist conditions enhance O 3 vegetation damages. Based on these analyses, O 3 RH was proposed and evaluated, which showed better spatiotemporal variation of O 3 -induced GPP reductions than the AOT40 index. Applications of O 3 RH index show that the decline of [O 3 ] over the past several decades cannot relieve O 3 vegetation damages in eastern U.S and western Europe. Meanwhile, the fast increases of surface [O 3 ] boost damages to vegetation in China. Our results showed that O 3 RH was able to be calculated as easy as exposure-based indexes (not dependent on any expensive observations or numerical models) and had similar spatiotemporal representation of O 3 damage as dose-based method, which greatly facilitated the assessment of O 3 vegetation damages, especially for policy makers and researchers without ecological backgrounds.

Data availability statement
The data that support the findings of this study are available upon reasonable request from the authors.