Current and future ozone risks to global terrestrial biodiversity and ecosystem processes

Abstract Risks associated with exposure of individual plant species to ozone (O3) are well documented, but implications for terrestrial biodiversity and ecosystem processes have received insufficient attention. This is an important gap because feedbacks to the atmosphere may change as future O3 levels increase or decrease, depending on air quality and climate policies. Global simulation of O3 using the Community Earth System Model (CESM) revealed that in 2000, about 40% of the Global 200 terrestrial ecoregions (ER) were exposed to O3 above thresholds for ecological risks, with highest exposures in North America and Southern Europe, where there is field evidence of adverse effects of O3, and in central Asia. Experimental studies show that O3 can adversely affect the growth and flowering of plants and alter species composition and richness, although some communities can be resilient. Additional effects include changes in water flux regulation, pollination efficiency, and plant pathogen development. Recent research is unraveling a range of effects belowground, including changes in soil invertebrates, plant litter quantity and quality, decomposition, and nutrient cycling and carbon pools. Changes are likely slow and may take decades to become detectable. CESM simulations for 2050 show that O3 exposure under emission scenario RCP8.5 increases in all major biomes and that policies represented in scenario RCP4.5 do not lead to a general reduction in O3 risks; rather, 50% of ERs still show an increase in exposure. Although a conceptual model is lacking to extrapolate documented effects to ERs with limited or no local information, and there is uncertainty about interactions with nitrogen input and climate change, the analysis suggests that in many ERs, O3 risks will persist for biodiversity at different trophic levels, and for a range of ecosystem processes and feedbacks, which deserves more attention when assessing ecological implications of future atmospheric pollution and climate change.

2012). Changing habitat conditions and disturbance are among the main causes of changes in plant communities at a global scale (Tilman & Lehman, 2001). Air pollution is recognized as an important factor affecting habitat conditions globally, while tropospheric ozone (O 3 ) has been identified as the most widespread phytotoxic gaseous pollutant causing significant long-term abiotic stress over large areas (Ashmore, 2005). As a result of increasing emissions of precursor gases (carbon monoxide [CO], oxides of nitrogen [NO x ], volatile organic compounds [VOC], and methane [CH 4 ]), mean concentrations have been growing since the 1950s, at a rate of 5 ppb/decade on average in the northern hemisphere (NH) and by 2 ppb/decade in the southern hemisphere (SH; Cooper et al., 2014). According to the four representative concentration pathways (RCPs) used in the Intergovernmental Panel on Climate Change Fifth Assessment Report (AR5; IPCC, 2013), by the middle of this century, both increases and decreases in tropospheric O 3 concentrations are possible, depending on the regional balance between processes leading to either formation or destruction of O 3 , and the extent of adoption of air pollution abatement measures underlying the different RCPs (Fiore et al., 2012). Greenhouse gas emissions differ between RCPs, and the consequent effects on climate and land use also alter the concentrations and distribution of O 3 , which also acts as important greenhouse gas.
The global threats to agricultural yields and food security posed by O 3 under different scenarios have been quantified and discussed by several studies (Chuwah, van Noije, van Vuuren, Stehfest, & Hazeleger, 2015;Tai, Martin, & Heald, 2014). In contrast, implications for biodiversity at the global scale are much less certain and have had little recognition. This is an important gap, which deserves attention when assessing ecological implications of future developments of atmospheric pollution and climate. Here, we provide a global evaluation of the current (year 2000) and future (2050) O 3 exposure of the Global 200 (G200) terrestrial ecoregions (ER), which are priority regions for conservation (Olson & Dinerstein, 1998;wwf.panda.org). ERs have relatively uniform climate with a characteristic set of ecological communities. They are typified by high numbers of endemic species, high taxonomic uniqueness, global rarity, and/or unique ecological phenomena. They have been selected for their irreplaceability and distinctiveness and represent all the major global biomes. We focus on the G200 ERs, rather than on biodiversity hot spots, because our focus is on broader issues of ecosystem structure and function, rather than the threat to individual species.
We link this evaluation of O 3 exposure to a critical focused review of the observational and experimental evidence for impacts of elevated O 3 exposure on terrestrial biodiversity, and on downstream ecosystem processes and related feedbacks to the atmosphere. This review is based mostly on evidence in temperate regions, and we discuss the extrapolation to regions for which little knowledge of O 3 effects currently exists. Finally, we assess possible risks and benefits of different climate and air pollution policies for the ERs, and for the major biomes within which they are situated, in different regions of the world.
Our simulations used the Community Earth System Model (CESM; Appendix S1), including changes in anthropogenic emissions of precursor gases and climate, but not land use (Val Martin et al., 2015).
The CESM model reproduces global surface O 3 levels well, although values at any location may differ by up to 15% from measured values (Tilmes et al., 2016). We considered results for two contrasting scenarios (Table S1): RCP4.5, which aims to stabilize global radiative forcing at 4.5 W/m 2 by the end of the century, and RCP8.5, in which greenhouse gas emissions continue to increase over this century, and there is no climate stabilization. The global precursor emissions of CO, NO x , and VOCs in 2050 are similar in these two RCPs, but 2050 concentrations of CH 4 are much higher under RCP8.5; this is relevant because CH 4 contributes to background tropospheric O 3 levels both as an O 3 precursor and by its effect on global warming (West & Fiore, 2005).   (CLRTAP) uses AOT40 as an exposure index for estimating ecological risks (Appendix S1) and defines an AOT40 of 3 ppm hr accumulated over 3 months as a critical level of O 3 "above which adverse effects may occur on the growth of the most sensitive species of (semi-) natural communities dominated by annuals" (CLRTAP, 2015). We estimate that at a three-month mean M12 value of 35-40 ppb, there is a high likelihood that the AOT40 critical level will be exceeded ( Figure S1).

| O 3 EXPOSURE OF G200 ECOREGIONS
The three-monthly M12 values in individual G200 ERs ranged from 15.8 to 65.2 ppb (Table S2). The 10 ERs with the highest exposure all have maximum three-monthly M12 values ≥60 ppb (Table 1).
These ERs fall within five major biomes; four are in the temperate broadleaf and coniferous forests biome, two in the Mediterranean forests, woodlands, and shrubs biome, and three are in the montane grasslands biome; only the Terai-Duar savannas and grasslands, at the base of the Himalayas, are in a subtropical biome. Three are in North America, five in Asia, and two stretch across Eurasia. Some of these ERs are also associated with global biodiversity hot spots, as identified by Myers, Mittermeier, Mittermeier, de Fonesca, and Kent (2000), for example, in California, the Mediterranean Basin, the Caucasus, and the Himalayas. As CESM model predictions may differ from measured values (see above), there is some uncertainty in the precise rankings of ERs in Table 1, and in Table S2, which provides M12 values for all ERs.  1990(Mills, Hayes, et al., 2011Mills, Pleijel, et al., 2011), and in forests, a systematic assessment of observational data for 2009 revealed symptoms in 17 different species across 13 European countries (Schaub & Calatayud, 2013

| CHANGES IN PLANT COMMUNITIES
Some species are better protected from O 3 stress than others due to differences in leaf diffusive properties, cellular detoxification capacity, compensatory biomass production and allocation, or isoprene emission. However, the genetic basis for the differential sensitivity remains elusive, although some recent studies with O 3 -sensitive and O 3 -resistant Arabadopsis thaliana (Xu et al., 2015) and crops such as rice (Frei, 2015) (Hayes, Jones, Mills, & Ashmore, 2007). However, another analysis (Jones, Hayes, Mills, Sparks, & Fuhrer, 2007) of the same database suggested that light-loving species tend to be more sensitive than those that normally occur in the shade, plants of dry sites tend to be more sensitive than those found in more moist soils, and plants tolerant of moderately saline conditions are more sensitive than those of nonsaline habitats. The extent to which these findings can be generalized to species in other ERs is uncertain, as the sensitivity of the species in many of the ERs with high O 3 exposure is unknown. But the fact that species from the Fabacea (or Leguminosae) family have consistently been found to be relatively more sensitive than those of other families, and because Fabacea, including many trees, shrubs, and herbaceous plant species, are an ubiquitous component of both temperate and tropical ERs, O 3 -sensitive species are likely to be present in ERs that so far have not been monitored.
The existence of wide differences in sensitivity between species implies that O 3 stress can cause long-term shifts in species evenness or richness in diverse plant communities. There is some observational evidence of such effects within the North America ERs listed in Table 1; for instance, changes in species richness in coastal shrub vegetation (Artemisia californica Less.) within the "California chaparral and woodlands" ER were attributed to O 3 (Westman, 1979), and significant changes in stand composition have been reported along O 3 gradients in the San Bernardino Mountains (Miller, 1973), within the "Sierra Nevada coniferous forest" ER (Arbaugh & Bytnerowicz, 2003), although effects of O 3 in these areas may be difficult to separate from other influencing factors such as high nitrogen (N) deposition (Fenn, Poth, Bytnerowicz, Sickman, & Takemoto, 2003). Payne et al. (2011) identified O 3 as a key driver of compositional changes in species in British acid grassland, in addition to N deposition, although it was not associated with a reduction in species richness or diversity indices. In  (Evans & Ashmore, 1992). Exposure to elevated O 3 of an upland mesotrophic grassland in the UK that was managed to increase species diversity significantly decreased the biomass of Ranunculus species; this was attributed to reduced performance of the hemi-parasitic species Rhinanthus minor (yellow rattle), a species that reduces the productivity of grasses and opens up the grassland canopy, suggesting that O 3 stress may be a significant barrier to achieving increased species diversity in managed grasslands because of its effects on this keystone species (Wedlich et al., 2012). In an early succession pine forest community, O 3 -sensitive blackberry (Rubus cuneifolius) reached the highest cover under high O 3 exposure (Barbo, Chappelka, Somers, Miller-Goodman, & Stolte, 1998), either because growth of blackberry was less affected by O 3 than its leaf injury indicated, or it was more effective in out-competing other, less O 3 -sensitive species for resources.
In general, effects of O 3 on the competitive balance between species are not uniform and may depend on the species mixture (Nussbaum, Bungener, Geissmann, & Fuhrer, 2000).
In herbaceous species, short-term sensitivity of growth to O 3 is positively related to inherent relative growth rate (Bungener, Nussbaum, Grub, & Fuhrer, 1999;Danielsson, Gelang, & Pleijel, 1999;Davison & Barnes, 1998) suggesting that faster growing species tend to be more O 3 -sensitive than slower growing species. Thus, in ERs where relative growth rates are generally low, O 3 stress would be less damaging than in ERs dominated by faster growing species. In fact, after several years, changes in the functional group composition of subalpine grassland at high O 3 (Volk, Bungener, Contat, Montani, & Fuhrer, 2006) could not be separated statistically from nutrient gradient effects (Stampfli & Fuhrer, 2010). Similarly, a montane Geo-Montani-Nardetum proved resilient to long-term O 3 exposure, regardless of extra N input (Bassin, Volk, & Fuhrer, 2013); this was not caused by low canopy O 3 uptake (Volk, Wolff, Bassin, Ammann, & Fuhrer, 2014). However, in the absence of changes in species, micro-evolutionary adaptation to O 3 stress might be involved in these permanent old grasslands (Kölliker, Bassin, Schneider, Widmer, & Fuhrer, 2008) and also in some forests (Moran & Kubiske, 2013). For instance, because of a competitive disadvantage, the most sensitive aspen genotype was eliminated in a seven-year exposure to elevated O 3 from the seedling stage through to maturity, although total growth of the stand was not affected (Kubiske, Quinn, Marquardt, & Karnosky, 2007).
Shifts in community composition could also result from specific changes in reproductive success caused by decreased biomass allocation (Bender, Bergmann, & Weigel, 2006;Wang et al., 2015) impairing reproductive growth and development (Leisner & Ainsworth, 2012) and seed production (Bender et al., 2006;Harward & Treshow, 1975), or from direct effects of O 3 on reproductive structures (Black, Black, Roberts, & Stewart, 2000). In temperate grasslands, experimental O 3 treatment reduced seed number, fruit number, and weight, but increased flower number and flower weight in a number of species, for example, in paper birch (Betula papyrifera) (Leisner & Ainsworth, 2012), and decreased seed weight and germination rate (Darbah et al., 2008) with implications for the establishment and survival of the progeny.
Where plant composition greatly depends on the belowground seed pool, declining reproductive success can be caused by elevated O 3 exposure, such as in the Dehesa annual grasslands, which cover several million hectares in the Iberian Peninsula within the highly O 3 -exposed "Mediterranean forests, woodlands, and shrubs" ER (  (Hayes, Wagg, Mills, Wilkinson, & Davies, 2012;Hayes, Williamson, & Mills, 2012), and such subtle shifts play an important role when flowering is closely synchronized with pollinating species (Black et al., 2000).

| CHANGES IN SOIL MICROBIOTA AND NUTRIENT CYCLING
The belowground ecosystem compartment is insulated from direct O 3 exposure, but there is an accumulating body of evidence that effects aboveground translate into changes in soil microbial communities, and further propagate through the microbial food web to alter carbon (C) and N cycling (Lindroth, 2010). The main pathways considered here, and their implications for ecosystem processes, and feedbacks to the atmosphere, are depicted in Figure 2.
A general, but highly variable, trend is that under high O 3 , relatively less biomass is allocated to roots compared to shoots, with a mean reduction by 5.6% across all species covered in a meta-analysis (Grantz, Gunn, & Vu, 2006). This reduces the amount of root detrital inputs and consequently may significantly affect long-term soil C formation rates (Loya, Pregitzer, Karberg, King, & Giardina, 2003). In addition to litter input, litter decomposition is a key process in nutrient cycling, which in complex ways depends on the diversity of litter, the decomposer community (Gessner et al., 2010), and environmental and soil conditions.
It has been suggested that plant species richness is not related to the diversity of litter composition (Meier & Bowman, 2008), and thus, O 3 effects at the species diversity level may be of limited relevance for litter decomposition in the soil. But, evidence exists that, in the absence of changes in plant diversity, O 3 slows decomposition, although a general pattern is lacking and a range of different mechanisms could be involved (Couture & Lindroth, 2013).
Slower decomposition could be related to changing soil microbial functional diversity caused by altered litter quality (Aneja et al., 2007).
Mycorrhizae are ubiquitous in all terrestrial ecosystems and play an essential role in soil-plant nutrient exchange and via the turnover of external mycelium for the transfer of root-derived C to SOM (Godbold et al., 2006). Lower O 3 stress would thus not only benefit ectomycorrhizal diversity and richness (Katanić, Paoletti, Orlović, Grebenc, & Kraigher, 2014), but also soil nutrient and C cycling, particularly in very dry, wet, or cold habitats where plant productivity is limited by environmental conditions, such as those at high latitudes or in montane regions.

| IMPLICATIONS FOR TERRESTRIAL FEEDBACKS TO THE ATMOSPHERE
Lindroth (2010)  it can be hypothesized that in spite of lower soil C inputs associated with reduced net primary production, soil C stocks could increase due to lower degradability of the litter and reduced microbial activity.
However, data from sufficiently long O 3 exposure studies are extremely rare, and findings are variable. In experimental forests, O 3 reduced the C content in woody tissues and in the near-surface mineral soil (Talhelm et al., 2014), and in more stable SOM pools (Hofmockel, Zak, Moran, & Jastrow, 2011), but data from a high-elevation grassland experiment indicated that soil C remains unchanged, possibly because a low C input was compensated by reduced turnover (Volk et al., 2011), as discussed above. Similarly, in a modeling study, the replacement of sensitive by more tolerant plant species or genotypes (as also discussed above) in a temperate deciduous forest led to unchanged biomass F I G U R E 2 Diagram summarizing main downstream processes affected by O 3 uptake in plant communities, starting either with or without changes in species composition (box), and ultimately feeding back to atmospheric composition. 1, Reduced litter input and root exudation, lower degradability; 2, altered microbiota and slower decomposition; 3, increased immobilization of C and N; 4, reduced nutrient availability; 5, altered methanogenic activity in wetlands; 6, reduced soil respiration and N availability for denitrification; 7, loss of water flux control under drought; 8, emission of biogenic volatile organic compounds C stocks in the long term (>100 years) (Wang, Shugart, Shuman, & Lerdau, 2016  Impacts of O 3 on N 2 O emissions are even less certain, but N immobilization due to decreased decomposition not only limits the availability of N for plants, as reviewed above, but also for the denitrifier community, which could reduce the potential for nitrification and denitrification Kou, Cheng, Zhu, & Xie, 2015   show a very small decrease due to changes in emissions.   Table 3a were identified as having high future N deposition. Based on implementation of current legislation to 2030, highest future N deposition is projected for Chota-Nagpur dry forests and Terai-Duar savannas and grassland.

| INTERACTIONS WITH OTHER ABIOTIC STRESSES IN A FUTURE CLIMATE
Bleeker, Hicks, Dentener, Galloway, and Erisman (2011) predicted that by 2030, 62 biodiversity hot spots and G200 ERs are projected to receive >30 kg N ha −1 year −1 , with forest and grassland ecosystems in Asia most exposed.
Ozone exposure is expected to interact with N addition and/or warming, as reviewed by Mills et al. (2016). Effects of climate change on stomatal O 3 flux and canopy uptake of O 3 can be either direct-for example, temperature, CO 2 , and humidity effects on stomatal conductance-or indirect via an influence on soil water potential and plant development (Harmens, Mills, Emberson, & Ashmore, 2007;Mills et al., 2016). In addition, O 3 itself can, for example, modify the responses of plants to naturally occurring environmental stresses such as drought (Hayes, Wagg, et al., 2012;Hayes, Williamson, et al., 2012;Mills, Hayes, Wilkinson, & Davies, 2009;, 2010 via effects on the hormonal control of stomatal functioning (Dumont et al., 2013) and plant development (canopy and roots), which can feedback to global warming (Sitch et al., 2007). Under O 3 exposure, many species have smaller roots (Grantz et al., 2006), thereby enhancing drought sensitivity. Depending on species, O 3 might induce stomatal closure, increased stomatal opening or sluggishness (Hoshika, Omasa, & Paoletti, 2013;Hoshika, Katata, et al., 2015), or have no effect . Differences in the specific response to O 3 of stomatal control may thus affect species composition indirectly through variable soil moisture changes (Jäggi & Fuhrer, 2007). With progressive global climate change, drought episodes are projected to become more frequent in many world regions, and subtle interactions of O 3 with water flux regulation may thereby influence community dynamics and species dominance. Sun et al. (2012) suggested that loss of stomatal sensitivity in a Southern Appalachian forest in the USA will not only increase drought severity in the region, thus affecting ecosystem hydrology and productivity, but it will also have negative implications for flow-dependent aquatic biota. When occurring over sufficiently large areas, high O 3 effects on stomata could shift catchment water balances through altered canopy water fluxes (Lombardozzi, Levis, Bonan, Hess, & Sparks, 2015;McLaughlin, Wullschleger, Sun, & Nosal, 2007;Sun et al., 2012), with possible implications for the surface energy balance (Super, Vilà-Guerau De Arellano, & Krol, 2015).
Some of the Asian regions such as the Tibetan plateau have also been identified as a hot spot of climate change impacts, both in terms of recent observed change (Shen et al., 2015;Turco, Palazzi, von Hardenberg, & Provenzale, 2015) and model projections (Diffenbaugh & Giorgi, 2012). Combining projections for both mean changes in temperature and precipitation with changes in the interannual variability of these parameters, simulations by Li et al. (2013) revealed that by the end of the 21st century, 96% of G200 ERs will face moderate to pronounced climatic changes relative to the change in the past five decades, with ERs at high northern latitudes being most exposed to change, followed by those in the Mediterranean Basin, Amazon Basin, East Africa, and South Asia. Hence, some of the priority ERs, which are highlighted in our analysis as being of greatest threat from increased O 3 exposure, are also at high risk from N deposition and climate change, emphasizing the need to assess the effects of O 3 together with other key components of environmental change.
Increasing CO 2 in controlled environments or open-top chambers often ameliorates effects of O 3 on leaf physiology, growth, and C allocation; however, evidence from field-based experiments does not support that they have fully compensatory effects when co-occurring . Combined responses to elevated temperature and O 3 have rarely been studied even though some critical growth stages such as seed initiation are sensitive to both. Kasurinen et al. (2012) showed that O 3 modifies the response of temperate silver birch to warming, but the magnitude of response varies among genotypes.
Although the review by Mills et al. (2016) (Bassin et al., 2013;Volk et al., 2014). Under climatically challenging conditions, added N to low background N deposition caused large changes in the community composition, with sedges becoming particularly dominant, while added O 3 had no effect on functional group composition and few effects on productivity (see above

| CONCLUSION: IMPLICATIONS OF DIFFERENT CLIMATE AND AIR POLLUTION POLICIES
In spite of the limited direct evidence for O 3 effects on terrestrial biodiversity, and of sufficient experimental and observational data from the full global range of ERs with high conservation value, the information presented in this study leads us to conclude that O 3 levels are sufficiently high today, or will become so in the future, to exert a large-scale influence on community composition at different trophic levels, and to alter nutrient and C cycling with possible feedbacks to the climate.