Ozone risk for vegetation in the future climate of Europe based on stomatal ozone uptake calculations


Corresponding author.
e-mail: jenny.klingberg@dpes.gu.se


The negative impacts of surface ozone (O3) on vegetation are determined by external exposure, leaf gas exchange and plant antioxidant defence capacity, all dependent on climate and CO2 concentrations. In this study the influence of climate change on simulated stomatal O3 uptake of a generic crop and a generic deciduous tree at ten European sites was investigated, using the LRTAP Mapping Manual stomatal flux model. O3 concentrations are calculated by a chemistry transport model (MATCH) for three 30-yr time-windows (1961–1990, 2021–2050, 2071–2100), with constant precursor emissions and meteorology from a regional climate model (RCA3). Despite substantially increased modelled future O3 concentrations in central and southern Europe, the flux-based risk for O3 damage to vegetation is predicted to remain unchanged or decrease at most sites, mainly as a result of projected reductions in stomatal conductance under rising CO2 concentrations. Drier conditions in southern Europe are also important for this result. At northern latitudes, the current parameterisation of the stomatal conductance model suggest O3 uptake to be mainly limited by temperature. This study demonstrates the importance of accounting for the influences by climate and CO2 on stomatal O3 uptake, and of developing their representation in models, for risk assessment involving climate change.

1. Introduction

Surface ozone (O3) is an air pollutant of major concern, causing reduced crop yield and quality, impaired forest growth and negative effects on human health (The Royal Society, 2008). Due to its strong dependence on meteorological conditions, surface O3 is sensitive to climate change (Jacob and Winner, 2009). A number of studies have reported increasing background O3 concentrations over the mid-latitudes of the Northern Hemisphere during the last decades (Vingarzan, 2004; Derwent et al., 2007) and modelling studies indicate a significant rise in global mean O3 concentration in the future unless large emission reductions are implemented (Prather et al., 2003; Dentener et al., 2006; Stevenson et al., 2006). For Europe, regional air quality models, simulating the conditions during future climate, generally show increasing O3 concentrations despite constant anthropogenic precursor emissions (Langner et al., 2005; Meleux et al., 2007; Andersson and Engardt, 2010). The increase is mostly explained by increased temperature, decreased cloudiness (Meleux et al., 2007) and reduced dry deposition (Andersson and Engardt, 2010). Thus, climate change has the potential to counteract emission reductions aimed to limit surface O3 concentrations. This is important to consider in future air quality and emission control policies (Meleux et al., 2007).

Current global yield losses has been estimated to be in the range of 3–16% for four major food crops due to exposure to surface O3 concentrations, with additional reductions under possible higher O3 concentrations in the future, posing a serious threat to food security (Van Dingenen et al., 2009). However, the negative effects of O3 on vegetation are more closely related to the uptake of O3 through the stomata than to the concentration in the ambient air (Emberson et al., 2000b; Ashmore et al., 2004; Pleijel et al., 2004; Uddling et al., 2004; Karlsson et al., 2007a). Factors influencing the stomatal uptake of O3 therefore have to be considered in risk assessment. A flux-based index (PODY, phytotoxic O3 dose above a flux threshold Y, earlier called AFstY, accumulated stomatal flux of O3) has been developed, which takes into account the influence of temperature, solar radiation, water vapour pressure deficit (VPD), soil water potential (SWP), atmospheric O3 concentration and plant development stage (phenology) on stomatal O3 uptake (LRTAP Convention, 2004; Pleijel et al., 2007). For a given O3 concentration, the stomatal flux (O3 uptake) will be greater under humid conditions since dry air and soil moisture deficit induce stomatal closure to minimize plant water loss through transpiration. Evaluations have shown the PODY index to be superior to the concentration-based index AOT40 in explaining yield reductions for wheat and potato (Pleijel et al., 2004) as well as biomass reductions and visible leaf injury for O3 sensitive tree species (Uddling et al., 2004; Karlsson et al., 2007a). Unlike the concentration-based index AOT40, the flux-based approach allows modification of O3 sensitivity by climatic conditions to be incorporated into the risk assessment.

Models of stomatal conductance (gs) have been parametrized for a limited number of crop and tree species. Within the UNECE LRTAP Convention, a simplified multiplicative gs model has been developed and employed to quantify stomatal O3 flux and estimate the risk for O3 damage to a generic crop and generic tree species across Europe (LRTAP Convention, 2004; Simpson and Emberson, 2006) to be used in large-scale modelling. Maps of modelled flux-based O3 risk for a generic crop have been shown to better correspond with field-based evidence of adverse affects compared to AOT40 (Hayes et al., 2007).

In assessing future O3 risk for vegetation it is important to consider the reduction of gs in elevated CO2 concentrations (Ainsworth and Rogers, 2007). Plants do not maximise the CO2 uptake, but rather optimise the water use efficiency to loose as little water as possible per CO2 taken up (Jones, 1992). In higher CO2 concentrations, the optimum water use efficiency tends to be achieved with smaller stomatal opening. In a case study for winter wheat, Harmens et al. (2007) assumed a 35% reduction in gs due to elevated CO2 concentrations and modified current (1997) meteorology and O3 concentrations input data to POD6 calculations for five grid squares in the EMEP model. The results showed that with a 3°C increase in temperature and constant absolute humidity, the absorbed O3 dose decreased, despite an assumed 5 ppb increase in O3 concentrations. The result is however based on simplified assumptions of the future climate. Sitch et al. (2007) estimated the impact of projected O3 concentrations on the land-carbon sink, by including the effect of O3 deposition on photosynthesis and the interactions between O3 and CO2 through stomatal closure in their global land-carbon cycle model. Their results suggest a significant suppression of the global land-carbon sink due to negative effects of O3 on plant productivity. CO2-induced stomatal closure was found to offset the O3 suppression of gross primary production by more than one third. Hence, the combined effects of climatic conditions and elevated CO2 concentrations on future O3 risk for vegetation are important for future crop and forest production and should be accounted for in coupled biosphere-climate models.

In this study, O3 concentrations from a regional chemistry transport model (MATCH), driven by meteorological data from a regional climate model (RCA3), was used to estimate reference (1961–1990), near future (2021–2050) and far future (2071–2100) flux-based O3 risk (PODY) in Europe using the Mapping Manual stomatal flux model (LRTAP Convention, 2004). Anthropogenic precursor emissions and boundary concentrations were held constant in the chemistry transport model, in order to limit the investigation to the influence of climate change.

The aims of the study were:

  • • To assess the influence of climate change and elevated CO2 concentrations on the flux-based risk for O3 damage to a generic crop and a generic deciduous tree in Europe. The hypotheses were that even without changes in O3 precursor emissions, climate change could significantly modify the flux-based O3 risk and that the plant stomatal response to elevated CO2 concentrations would have the potential to significantly reduce the O3 risk.
  • • To describe the spatial differences in the modelled reference and future flux-based risk at ten sites along a transect from northeast to southwest in Europe. The hypothesis was that dry air and high soil moisture deficit would substantially limit stomatal flux of O3 in the south, while low temperatures would be limiting in the north.
  • • To compare the flux-based risk assessment with that based on AOT40. The hypothesis was that AOT40 would predict a relatively smaller risk in the northern part of Europe and larger risk in the southern part of Europe in both reference and future climate, compared to the flux-based risk assessment.

2. Methods

2.1. Stomatal O3 flux

The stomatal O3 flux was calculated using a multiplicative algorithm (an extension of the concepts presented in Jarvis (1976), Emberson et al. (2000a) and Emberson et al. (2000b)), which includes functions accounting for the limiting effects of various abiotic factors on stomatal conductance, thereby regulating the O3 flux into the plant leaf. The multiplicative algorithm is (LRTAP Convention, 2004)


where gs is the stomatal conductance [mmol O3 m−2 sunlit projected leaf area (PLA) s−1] and gmax is the species-specific maximum gs. The functions fphen, inline image, flight, ftemp, fVPD and fSWP are expressed in relative terms (take values between 0 and 1) as a proportion of gmax. These parameters allow for the influence of phenology and O3, and the environmental variables (irradiance, temperature, water VPD and SWP) on gs to be estimated. The part of eq. (1) related to fphen and inline image is a most limiting factor approach; i.e. gs is limited by either senescence due to normal aging or premature senescence induced by O3.

The stomatal flux (Fst) of O3 to a plant leaf is calculated using a resistance analogue


The O3 concentration at the top of the canopy [C(z)] is assumed to be a reasonable estimate of the concentration at the surface of the laminar leaf boundary layer near the sunlit upper canopy leaves. The 1/(rb+rc) term is the deposition rate to the leaf determined by the quasi-laminar resistance (rb) and the leaf surface resistance (rc) gs/(gs+gext) represents the fraction of O3 taken up through the stomata, where 1/gext is the external leaf resistance.

The POD accumulated per unit projected sunlit leaf area above a threshold of Y nmol m−2 s−1 was calculated as


The PODY is accumulated over a period of time corresponding to the part of the growing season when the plant is considered to be sensitive to O3.

In this study a simplified flux-based method, recommended for large-scale modelling (LRTAP Convention, 2004; Simpson and Emberson, 2006) was used to indicate the risk for O3 damage to a generic crop (POD3crop) and a generic deciduous tree (POD1.6tree). Important simplifications were that O3-induced premature senescence was assumed to have no effect on gs (i.e. inline image) and that a lower threshold were used for the generic crop than recommended by the LRTAP Convention (2004) for specific crops (Y= 3 nmol m−2 PLA s−1 for a generic crop compared to Y= 6 nmol m−2 PLA s−1 for wheat and potato), making the method numerically more robust (Tuovinen et al., 2007). The threshold for trees is Y= 1.6 nmol m−2 PLA s−1. For a generic crop a 3-month time-window for stomatal O3 flux accumulation based on latitude (LRTAP Convention, 2004), was used to bypass the uncertainty in the timing of the relevant time-interval, which, especially for crops, is rather short. The start and length of the generic deciduous tree stomatal O3 accumulation period was based on latitude and altitude (LRTAP Convention, 2004). Due to uncertainties caused by difficulties in modelling a plant-relevant SWP, the potentially large variations in soil moisture within a model grid and possible irrigation practices, soil moisture was assumed not to limit gs (i.e. fSWP= 1). In one PODY calculation for the generic deciduous tree, however, a SWP function was included to estimate the potential influence of this factor on gs and the stomatal O3 flux. It was assumed that water was freely available from 0 to –0.05 MPa SWP with a linear decrease in water availability (and gs) below –0.05 MPa down to a minimum at –1.5 MPa SWP (Hall et al., 1977). The stomatal flux algorithm and the parameterisation of the simplified flux-based method for a generic crop and a generic deciduous tree are further described in the LRTAP Convention (2004).

2.2. The CO2 function

Several studies have shown reduced gs (on average 0.66 of that in 370 ppm CO2) for wheat grown in 550–570 ppm CO2 concentrations (Wall et al., 2000, 2006). For wheat grown in 700 ppm CO2, the average gs ratio at elevated to ambient (∼350 ppm during daytime) CO2 was 0.56 (Bunce, 2000). In a meta-analysis, Ainsworth and Rogers (2007) found the reduction in stomatal conductance of C3 crops to be 25% in 567 ppm CO2 compared to ambient (366 ppm) and Pleijel et al. (2002) found up to a 20% reduction in gs for potato grown in 680 ppm CO2 compared to 360 ppm. Ainsworth and Rogers (2007) also found a 19% reduction of gs in trees grown in elevated CO2 concentrations (567 ppm on average) compared to ambient CO2 concentrations (366 ppm on average). Medlyn et al. (2001) reported a 21% decrease in gs for trees in response to growth in 700 ppm CO2 compared to 350 ppm CO2. Based on these studies, CO2 response functions (inline image) for a generic crop and a generic deciduous tree were parametrized and added to eq. (1). The influence of increasing CO2 on gs was, in this study, assumed to linearly decrease between 360 and 560 ppm CO2 concentration from 1 to 0.66 for a generic crop and to 0.8 for a generic deciduous tree, with no further reductions in gs above 560 ppm CO2 (Fig. 1). However, there is a large uncertainty with respect to the effect of elevated CO2 on gs in closed forest stands (Uddling et al., 2009 and references therein). Because of this uncertainty, the possible effect of CO2 on PODY was explored as a sensitivity study, by performing the calculations with and without inclusion of inline image. This was done also for the generic crop in order to separate effects of climatic changes from effects of rising CO2 concentrations on stomatal O3 flux.

Figure 1.

The influence of atmospheric CO2 concentrations on stomatal conductance for a generic crop (GEN-C) and for a generic deciduous tree (GEN-DT). Relative stomatal conductance (inline image) in relation to (a) CO2 concentration and (b) year using the A2 emission scenario. The data points in (a) are the observations on which the CO2 functions were based, see Section 2.2 for references.

2.3. MATCH and RCA3 model set-up and PODY input data

MATCH is a Eulerian, off-line, regional chemistry transport model. A detailed description of the model is available in Robertson et al. (1999). A number of previous studies have demonstrated the ability of MATCH to realistically simulate O3 concentrations over Europe (e.g. Tilmes et al., 2002; Andersson et al., 2007; van Loon et al., 2007). In order to investigate the effect of future climate change on surface O3, MATCH was driven by meteorology from the Rossby Centre's regional climate model (RCA3), described in Kjellström et al. (2005). In this study RCA3 was forced with climate data from the ECHAM4/OPYC3 global model on its boundaries simulating the IPCC SRES A2 and B2 emission scenarios (Nakicenovic et al., 2000). A2 is one of the more pessimistic greenhouse gas emission scenarios. The two emission scenarios yield similar climate change evolution for the first half of the 21st century, but in the second half of the 21st century there is an increasingly larger signal in the A2 scenario. The canopy-scale O3 dry deposition to vegetation in MATCH is a function of soil moisture, air humidity, temperature and irradiance. Also, emissions of biogenic isoprene are calculated online in MATCH. Natural emissions of other VOCs, sulphur or nitrogen containing compounds are not included in the present set-up. Hence, modelled future O3 concentrations are affected by changes in photochemistry, transport patterns, emissions and uptake by vegetation. Anthropogenic emissions and tracer boundary concentrations were set constant in MATCH, representing the year 2000, in order to isolate the influence of climate change and not changes in emission patterns. In MATCH three 30-yr periods were simulated: reference (1961–1990), near future (2021–2050) and far future (2071–2100). The set-up of the model system is further described in Andersson and Engardt (2010). To test if average PODY for the far future (A2 emission scenario) was significantly different compared the reference period, a two-sided Student's t-test was applied, assuming unequal variances, with 58 degrees of freedom (2 ×N−2; N= 30 yr).

For risk assessment purposes it is the stomatal flux of O3 to the sunlit leaf level of specified vegetation types and not the canopy-scale flux that is important. The PODY was, in this study, calculated off-line for a generic crop and a generic deciduous tree at 10 sites based on modelled data from the corresponding grid-cell in MATCH (O3) and RCA3 (meteorology). The ten monitoring sites within the European Monitoring and Evaluation Programme (EMEP) within CLRTAP were selected in a transect from northeast to southwest to represent different climatic conditions in Europe. Site characteristics and estimated growing seasons can be found in Table 1 and site locations in Fig. 2 which also shows the modelled April to September average daily maximum O3 in reference climate (1961–1990). PODY was calculated for both a generic crop and a generic deciduous tree at all sites although, for example, wheat is presently not a suitable crop at the northernmost site. However, the distribution of crops and trees are likely to change in response to climate changes.

Table 1.  Site coordinates and characteristics according to the EMEP website (http://www.emep.int/) and soil texture class given by RCA3
SiteCountryLatitude LongitudeAltitude (m a.s.l.)GEN-C growing season (day number)GEN-DT growing season (day number)Soil texture class (RCA3)
  1. Note: The growing season for a generic crop (GEN-C) is based on latitude and the growing season for a generic deciduous tree (GEN-DT) is based on latitude and altitude (LRTAP Convention, 2004).

FI22N Finland66°19’N
310165–254133–261Sandy loam
FI37S Finland62°33’N
162154–243125–271Sandy loam
SE32S Sweden57°49’N
SE11S Sweden56°01’N
DE02N Germany52°48’N
DE03S Germany47°55’N
1205 118–207115–288Loam
FR10C France47°16’N
620116–205107–296Silt loam
FR13S France43°37’N
236106–195 97–308Silt loam
ES09N Spain41°16’N
1360 100–189106–301Loam
ES07S Spain37°14’N
1230  90–179 98–310Silt loam
Figure 2.

Location of the ten EMEP sites considered in the PODY calculations. Also shown is the modelled average summer-time (April–September) daily maximum O3 in reference climate (1961–1990).

O3 and meteorological input data for the PODY calculations should be valid for the height of the canopy which is assumed to be 1 m above ground level for a generic crop and 20 m for a generic deciduous tree. In MATCH, the O3 concentrations representing the lowest model layer were downscaled to 3 m at every time-step based on local stability and surface deposition velocity using similarity theory. To avoid systematic errors, a further correction was applied to adjust the 3 m O3 concentrations from MATCH to canopy height. Since values of u* and canopy-scale deposition to vegetation in MATCH were not available (not saved as output data), the simplified method in the Mapping Manual (LRTAP Convention, 2004) was used, assuming that the O3 concentration at 1 m was 93% of the 3 m concentration and that the O3 concentration at 20 m was 105% of the 3 m concentration). Modelled wind speed (3 h resolution, from RCA3) at 10 m was adjusted to canopy height using the logarithmic wind law, assuming neutral stability. Canopy height wind speed was used to estimate the leaf boundary layer resistance required in the flux calculation. Modelled temperature and relative humidity (from RCA3) corresponding to 2 m had 3 h temporal resolution. Linear interpolation was applied to receive the hourly temporal resolution needed for the PODY calculations. Incoming short-wave radiation data with 30 min resolution was recalculated to photosynthetic photon flux density (PPFD) by multiplying with a factor of 2 (Monteith and Unsworth, 2008). Soil water content summed over the top two soil layers in RCA3 (2.3 m) with six hour resolution was recalculated to SWP using the Groenevelt-Grant soil water retention model (Groenevelt and Grant, 2004) anchored in the point of saturation (Grant et al., 2010)


where θ is the relative water content (m3 m−3), θs is the relative water content at the point of saturation (m3 m−3) and h is the matric suction (m). k0, k1 and n are freely adjustable fitting parameters. Values of the parameters used to fit 20 Dutch soils to the Groenevelt-Grant model (Grant et al., 2010) was used for the soil types most similar to those in RCA3 (see Table 1).

2.4. Comparison of simulations and observations

The MATCH-RCA3 performance have earlier been evaluated and shown good agreement with O3 concentrations measured at EMEP stations in Europe (Engardt et al., 2009; Andersson and Engardt, 2010). The bias for summer and yearly average and daily maximum concentrations was within ±10% and correlations were 0.66–0.90. In this study, average and daily maximum O3 concentrations during the generic crop O3 accumulation period were used to estimate the performance of the MATCH-RCA3 modelling system for the ten selected EMEP monitoring sites. Modelled 3 m O3 concentrations from the last 10 yr of the reference period (i.e.1981–1990) were compared to O3 measurements during 1995–2006. The measurements were centred around the year of precursor emissions used in MATCH (year 2000). Only years with less than 10% missing data during the generic crop-growing season were included. The (spatial) correlations were high, 0.60 for average O3 and 0.85 for daily maximum, and the average bias small (+1.5% bias of average O3 and –3.7% bias of the daily maximum).

Hourly temperature and VPD from RCA3 (1981–1990) were compared to the surface analysis (1998–2005) from a high resolution (11 km) analysis system (MESAN, Häggmark et al., 2000). MESAN-data were only available for the five northernmost sites (FI22, FI37, SE32, SE11 and DE02). Comparison of data from the generic crop O3 accumulation period showed a good agreement of average temperature from RCA3 (–2.5% bias in °C) but consistently smaller range and standard deviation. The 90-percentile temperature was 3.2°C lower in RCA3 compared to the surface analysis on average. Since high VPD occurs at high temperatures, the underestimation in temperature results in a large VPD underestimation (–43% bias on average). An explanation could be smaller diurnal variation in the regional scale RCA3 compared to the surface analysis with higher resolution and a better capture of local scale processes. The large underestimation of VPD may cause overestimation of gs and stomatal O3 flux. The sensitivity of the calculation of POD3crop due to underestimated VPD input data was analysed by the calculation of the percentage change in the POD3crop index caused by a 50% increase in the modelled VPD from RCA3. Modelled temperature, VPD and O3 used for the PODY calculations are shown in Fig. 3.

Figure 3.

Box- and whisker plots of the RCA3 and MATCH data for the accumulation period of the generic crop used as input in the stomatal conductance model during (a) 1961–1990 and (b) 2171–2100 using the SRES A2 emission scenario.

3. Results

3.1. Concentration-based versus flux-based risk for O3 damage to a generic crop

In general, model calculated O3 concentrations increase towards southern Europe (Fig. 2). The concentration-based AOT40 index, calculated for daylight hours during the 3 months corresponding to the generic crop growing season (Fig. 4), indicates a higher O3 risk in southern compared to northern Europe. Already in the reference climate the current AOT40 critical level of 3 ppm h (LRTAP Convention, 2004) is exceeded at many sites. Note, however, that the 3 month growing season depends on latitude in this study, which can differ from the May–July AOT40, used e.g. in EU directives for vegetation risk assessments. At 6 of the 10 sites investigated, there is a large increase in future AOT40, especially in the 2071–2100 A2 emission scenario, where five sites (DE03, FR10, FR13, ES09 and ES07) will experience more than a doubling of AOT40. The four Nordic sites (FI22, FI37, SE32 and SE11) exhibit small changes in AOT40. Since emissions were held constant during the simulations, the increase in O3 concentration is entirely due to changes in the climate. The increase in southern Europe O3 concentrations has been explained by reduced dry deposition caused by a changed climate affecting the uptake to vegetation, while increasing isoprene emissions played a minor role (Andersson and Engardt, 2010).

Figure 4.

Average 3-month daylight AOT40 (corresponding to the generic crop growing season) at the different sites, based on O3 concentrations at 1 m. Error bars show standard deviation (N= 30 yr). Current critical level is 3 ppm h, which is indicated by the horizontal line.

The geographical pattern of the generic crop POD3crop is considerably different compared to that of AOT40 (Fig. 5). Both in reference and future climate, the stomatal O3 flux is largest in central Europe. The reduction in gs is 34% in the 2071–2100 A2 period due to CO2 concentrations above 550 ppm (Fig. 1). As a result, the sites with the largest increase in AOT40 by 2071–2100 in the A2 emission scenario show non-significant changes in POD3crop (DE03 and FR10) or a large decrease in POD3crop (statistically significant with p < 0.0001 for FR13, ES09 and ES07). In the same period, the two Swedish sites show a small but significant increase in POD3crop (p= 0.0119 for SE32 and p= 0.0268 for SE11). The same geographical pattern and direction of changes are found for POD0crop (data not shown), indicating that the result is not dependent on the choice of threshold.

Figure 5.

Average phytotoxic O3 dose for a generic crop (POD3crop) during 1961–1990, 2021–2050 and 2071–2100 following the SRES A2 emission scenario. The striped part of the bars show average POD3crop when the gs response function for CO2 (inline image) is included in the calculation. Error bars show standard deviation (N= 30 yr). Circles are average O3 concentration (scale on the right) for the same time periods as POD3crop is accumulated.

At the northern sites the O3 concentration is low and the temperature function significantly limits the gs (average ftemp substantially below 1, see Table 2). With no inclusion of inline image, POD3crop increase significantly at the Nordic sites in the 2071–2100 A2 period compared to the reference period (p < 0.0001 for FI22, FI37, SE32 and SE11). The increase can be explained by higher temperatures and, for SE11, to some extent higher O3 concentrations. For example, at the northernmost site (FI22) the average ftemp in reference climate is only 0.24 but increase to 0.49 in 2071–2100 A2, while the average VPD influence on stomatal conductance does not change (average fVPD does not fall below 1.00 in reference or future climate, see Table 2). Without the influence of inline image, POD3crop more than double at FI22 while the average O3 concentration decrease by approximately 1 ppb.

Table 2.  Average daylight ftemp and fVPD for a generic crop during the 30-yr reference period and year 2071–2100 following the SRES A2 emission scenario
Site1961–19902071–2100 A2

In central Europe, the O3 concentrations are high and climatic conditions favour stomatal O3 uptake. The large increase in O3 concentrations by 2071–2100 A2, together with considerably higher ftemp, but only slightly lower fVPD (Table 2) result in a significantly increased stomatal O3 flux compared to reference climate (p < 0.0001 for DE02, DE03 and FR10), when inline image is not included in the calculation.

As shown in Table 2, fVPD substantially limits the gs at the three southernmost sites (FR13, ES09 and ES07) already in reference climate. For example, average fVPD at the southernmost site (ES07) is 0.75 compared to 1.00 at the site with largest POD3crop (DE02). Even without the inclusion of inline image, the stomatal O3 flux does not increase in the future climate, despite a large increase in O3 concentration (5 ppb on average) by 2071–2100 A2, due to increased VPD.

The pattern is similar for the B2 emission scenario (see Appendix S1 in the Supporting Information). The increase in O3 concentration at the five southernmost sites is smaller than in the A2 emission scenario (2.7 ppb difference between the 2071–2100 B2 and the reference period, compared to 5.3 ppb in A2). The fVPD does not limit gs at the southernmost sites to the same extent, while the increase in gs due to increasing temperature at the northernmost sites is not as large.

The effect of increasing the calculated VPD from RCA3 with 50% on average daylight fVPD and POD3crop is shown in Table 3. The gs response to elevated CO2 was not included in this calculation. The average fVPD is mainly influenced at the southern sites in the reference period and the central and southern sites in the 2071–2100 A2 period. The decrease in fVPD result in a considerable reduction of POD3crop compared to the base calculations at the southern sites. However, the temporal trend and geographical pattern is not affected.

Table 3.  The effect of increasing the calculated VPD from RCA3 with 50% on the average daylight fVPD and average POD3crop (inline image not included) for a generic crop during the 30-yr reference period and year 2071–2100 following the SRES A2 emission scenario
Site1961–1990 (VPD×1.5)2071–2100 A2 (VPD×1.5)
  1. Note:% change refers to the change in POD3crop compared to the calculation without the increase in VPD (shown in Fig. 5).

FI220.99 2.2−141.00 5.1 −3
FI370.99 6.2 −70.9911.7 −2
SE320.99 7.9 −50.9617.9 −8
SE110.9914.0 −20.9827.3 −5
DE020.9819.7 −70.8728.3−18
FR130.70 9.5−420.37 8.4−47
ES090.73 7.2−500.38 7.3−54
ES070.55 5.6−610.26 5.0−62

3.2. Concentration-based versus flux-based risk for O3 damage to a generic deciduous tree

The spatial pattern of AOT40 calculated over the generic deciduous tree growing season (data not shown) is very similar to that of the generic crop, with large increases in AOT40 at the six southernmost sites, especially for the 2071–2100 A2 period. The spatial pattern of the generic deciduous tree POD1.6tree following the A2 scenario (Fig. 6) is also similar to that for a generic crop (POD3crop), with largest stomatal O3 flux in central Europe. At all sites POD1.6tree was significantly reduced in the 2071–2100 A2 period compared to the reference period (p < 0.0001) due to the assumed reduction of gs by 20% with CO2 concentrations above 550 ppm (Fig.1).

Figure 6.

Average phytotoxic O3 dose for a generic deciduous tree (POD1.6tree) during 1961–1990, 2021–2050 and 2071–2100 following the SRES A2 emission scenario. The striped part of the bars show average POD1.6tree when the gs response function for CO2 (inline image) is included in the calculation. Error bars show standard deviation (N= 30 yr). Circles are average O3 concentration (scale on the right) for the same time periods as POD1.6tree is accumulated.

Low and decreasing O3 concentrations as well as low average ftemp (see Table 4) result in small POD1.6tree at the two northernmost sites (FI22 and FI37), and no significant change by 2071–2100 A2 when the influence of inline image is not included in the calculation. SE32, SE11 and DE02 experience increased stomatal O3 flux in the 2071–2100 A2 period (p < 0.0001) compared to the reference period, due to climate changes, explained by increases in O3 concentration and average ftemp.

Table 4.  Average daylight ftemp, fVPD and fSWP for a generic deciduous tree during the 30-yr reference period and year 2071–2100 following the SRES A2 emission scenario
Site1961–19902071–2100 A2

The gs of the generic deciduous tree has a lower temperature optimum compared to the generic crop (Topt= 21°C instead of 26°C) and at the southernmost sites (FR13, ES09 and ES07) high temperatures limit gs in 2071–2100 A2. In addition to the dry air conditions (low average fVPD, see Table 4) this result in a significantly reduced stomatal O3 flux (p < 0.0001) despite a large increase in average O3 concentration even without the inclusion of inline image in the calculations. The result for the B2 emission scenario can be seen in Appendix S2 in the supporting information

To estimate the potential influence of soil moisture, a SWP function was included in one POD1.6tree calculations (Fig. 7). The gs response to elevated CO2 was not included. The results indicate that fSWP mainly limit O3 uptake at the southern sites (Table 4). The SWP function is especially limiting at FR13 and ES07.

Figure 7.

Average phytotoxic O3 dose for a generic deciduous tree (POD1.6tree) during 1961–1990, 2021–2050 and 2071–2100 following the SRES A2 emission scenario. Included in the POD1.6tree calculation is the gs response function for SWP (fSWP), but not for CO2 (inline image). Error bars show standard deviation (N= 30 yr).

4. Discussion

This study investigated how the changing climatic conditions and elevated atmospheric CO2 concentrations may modify leaf uptake of O3 and, hence, the risk for negative impacts on crops and forest trees. In agreement with earlier studies (Simpson and Emberson, 2006; Emberson et al., 2007; Simpson et al., 2007) the spatial flux-based risk pattern for O3 damage differed substantially from the AOT40-based risk, in the reference as well as future climate and irrespective of emission scenario (A2 or B2). It is important to take the influence of climate change into account in future risk assessment of surface O3, since factors such as warming, changes in amount and distribution of precipitation, shifts in growing season and elevated CO2 concentrations affect the stomatal uptake of O3 into the leaves (Harmens et al., 2007). Flux-based models are generally considered to be more physiologically relevant compared to exposure-based indices such as AOT40. The AOT40 index only reflects changes in the ambient O3 concentration, while the flux-based PODY index also allows climatic conditions and the CO2 concentration to modify stomatal uptake rates of O3, in line with important physiological mechanisms.

The results in this study clearly showed that the expected reduction in gs with rising atmospheric CO2 concentrations is of large importance for the decrease in projected future flux-based O3 risk. Harmens et al. (2007) also showed that the reduced gs under higher CO2 concentrations has the potential to significantly reduce the risk for O3 damage. For crops, there seem to be consensus among studies that plants growing under elevated CO2 concentrations have reduced gs (Bunce, 2000; Wall et al., 2000; Ainsworth and Rogers, 2007), but the magnitude of this effect under realistic agronomic conditions is less certain. For forests, gs may not at all decrease with rising atmospheric CO2 concentrations in many tree species. In four out of five free-air CO2 enrichments experiments in closed forests, gs was not significantly reduced under elevated CO2 concentrations (Ellsworth, 1999; Bernacchi et al., 2003 (pre-coppice canopy closure); Keel et al., 2007; Maier et al., 2008; Uddling et al., 2009; but see Gunderson et al., 2002 and Domec et al., 2009). In the long term, cumulative effects of elevated CO2 on plant growth and stand structure may be more important than the primary stomatal closure response to increased CO2 in determining gs in closed forests (Uddling et al., 2008; Uddling et al., 2009).

In a flux-based risk assessment for pine, beech and oak during one year at three sites in Europe, Emberson et al. (2007) found phenology (fphen) and soil water (fSWP) to be the key drivers limiting the seasonal profile of O3 flux. Tuovinen et al. (2009) emphasised the urgent need to develop soil moisture status modelling and its effect on gs for forests (crops may be irrigated). In the base case of our results, soil moisture is assumed to never limit gs, mainly because of difficulties in estimating the plant-relevant soil water potential valid for the integrated rooting zone of vegetation inside an entire model grid. The simulation including fSWP for the generic deciduous tree indicated, however, that soil water deficits have the potential to limit stomatal uptake of O3 in southern Europe and that this limitation is likely to increase in the future (Fig. 7). The conversion (based on available data from RCA3) of mean soil water content to mean SWP across a 2.3 m soil profile may represent an oversimplification as it does not capture the vertical gradients in soil water status and its integrated, non-linear, effect on plant water availability. Therefore, the assessment of the impact of geographical differences and temporal changes in soil moisture on stomatal O3 flux presented here indicates directions rather than magnitudes of the likely effects.

Several studies have reported an earlier onset of spring and a prolongation of the growing season in mid and high latitudes associated with global warming (Linderholm, 2006; Menzel et al., 2006). Changes in the timing of the growing season were omitted in this study, which assumed a fixed growth interval based on latitude. For specific crops, such as wheat, cumulative exposure may decline in a warmer climate due to accelerated plant development (Fuhrer, 2009). An earlier start of the growing season for wheat could result in a reduced overlap between the O3 exposure accumulation period and the peak summer O3 levels (Harmens et al., 2007; Fuhrer, 2009). Conversely, an earlier growing season may lead to more frequent co-occurrence of sensitive stages and spring-time peaks in O3 in the northern part of the Nordic countries (Karlsson et al., 2007b; Klingberg et al., 2009). For the generic crop, the 3 month growing season, much longer than the approximately 55-d period of high O3 sensitivity for wheat (LRTAP Convention, 2004), allows the POD3crop index to be less sensitive to changes in the timing of the growing season and can be interpreted as maximum potential risk. For trees, the growth period, and thus the period of O3 uptake, is likely to be prolonged in a future warmer climate (Taylor et al., 2008), implying an underestimation of the future O3 risk in this assessment.

There are large uncertainties in both climate projections and parameters in models for O3 risk assessment, as pointed out by Fuhrer (2009). Leaf properties determining the rate of O3 uptake and the biochemical defence capacity can be modified by changes in temperature, air humidity, soil moisture and increasing CO2 concentration. O3 uptake calculations may only be accurate when applied under conditions representative of those under which the parameterisation was performed, increasing the uncertainty when extrapolating to future climatic conditions (Fuhrer, 2009). Ashmore (2005) requested the development of new models, linking stomatal flux and detoxification processes to carbon assimilation and allocation, to include the capacity of species to adapt to changes in nutrient and water availability. Also, Musselman et al. (2006) emphasized the need to consider detoxification mechanisms in flux-based models, but noted the large uncertainty in quantifying the various defence mechanisms in plants.

In this study, the current LRTAP Convention (2004) parameterisation of the gs model indicated that ftemp was considerably more important than fVPD in limiting O3 uptake for the generic crop (Table 2) as well as for the generic tree at the northern sites (Table 4). This large importance of ftemp compared to fVPD in limiting gs could be questioned, considering the fact that minimizing water loss is a main function of stomata. The bell shaped optimum curve of the temperature function for crops seems uncertain as it predicts complete stomatal closure at temperatures below 12°C and above 40°C, which is hardly realistic for the Nordic and Mediterranean regions, respectively. Most physiological responses are known to acclimate to the prevailing temperature (e.g. Larcher, 2003; Körner, 2006) and both spatial (cold north vs. warm south) and temporal (during climate change) acclimation of the stomatal temperature response are thus likely. A certain degree of stomatal closure in response to low temperatures can be expected in colder conditions, but further evaluation of the stomatal temperature response function is required to improve predictions of the effect of climate on stomatal O3 flux. In addition, modelled stomatal O3 flux strongly depends on the quality of the gmax value (Fuhrer, 2009), which differs significantly between species and is likely to depend on climatic region.

The PODY calculations are strongly dependent on the quality of the input data. The coarse spatial resolution of regional climate models (50 × 50 km2) limits the details of important processes, such as diurnal range and extremes. Local scale characteristics, e.g. promoting nighttime air temperature inversions, may cause large diurnal O3 range deviating from the average conditions of the model grid (Klingberg et al., 2008). The large underestimation of VPD in RCA3 (–43% bias on average for the five northernmost sites) is of special concern for the stomatal O3 uptake calculations and may cause an overestimation of the flux-based O3 risk (Table 3). The importance of representative diurnal dynamics in temperature and VPD is emphasized when regional scale modelled data are used for stomatal O3 uptake calculations. Klingberg et al. (2008) estimated the sensitivity of POD0 and POD6 calculations for potato to small changes in input data in more detail for a site in southwest Sweden. A 10% change in O3 and temperature resulted in approximately 20% and 34–40% change in POD6, respectively. Without the flux threshold (POD0), the change was smaller. The sensitivity to a 10% change in VPD was much smaller (3–5% change in POD), but it was concluded that this factor presumably plays a much greater role in drier conditions. Other important simplifications are that the land-cover does not change with changing climate in MATCH and that meteorology is only based on one set-up of the climate model.

The uncertainties discussed above result in considerably uncertainty in the exact value of the stomatal O3 fluxes reported here. The most important uncertainties are considered to be the projections of future atmospheric changes and the parameterisation of the stomatal O3 flux model, neglecting both spatial and temporal acclimation and adaptation. More research is required to reduce the uncertainties before projected POD3crop and POD1.6tree can be used to give reasonable estimates of European yield or growth loss due to O3 damage. This study does therefore not intend to predict European yield or growth losses due to O3 damage in the future. The flux-based approach is, however, an improvement compared to AOT40. Notwithstanding the mentioned uncertainties we believe that there is high confidence in the general pattern and directions of change obtained in this study of how future climatic conditions will influence the O3 risk for vegetation in Europe.

5. Conclusions

  • • The increase in modelled future surface O3 concentrations is large in southern Europe, despite constant precursor emissions in MATCH. However, the future stomatal O3 uptake remain unchanged or decrease at most sites for a generic crop and decrease at all sites for a generic deciduous tree, mainly as a result of the modelled reduction in stomatal conductance with rising atmospheric CO2. Drier air and high soil moisture deficit as well as high temperatures are also important in southern Europe. However, if the CO2gs response function is excluded from the analysis, which may be realistic for forest trees (Uddling et al., 2009 and references therein), the future stomatal O3 flux is reduced only in southern Europe for the generic deciduous tree.
  • • The largest flux-based O3 risk for a generic crop and a generic deciduous tree is found at the central European sites, in both the reference and projected future climate. This pattern differs substantially from the concentration-based AOT40 index, indicating largest O3 risk in southern Europe.
  • • According to the current gs model parameterisation, the temperature function limits the stomatal O3 flux in northern Europe considerably, while the VPD and SWP functions are important mainly in southern Europe.
  • • This study demonstrates the importance to account for the climatic and atmospheric CO2 influences on stomatal O3 uptake for improved risk assessment in a changing atmosphere and climate.

6. Acknowledgments

Thanks are due to Patrick Samuelsson and Barry Broman at the Rossby Centre for providing climate data in a suitable format. Thanks are also due to the research programme Climate Change and Environmental Objectives (CLEO), funded by the Swedish Environmental Protection Agency, for supporting the work by Per Erik Karlsson, Magnuz Engardt and Håkan Pleijel.