Tropospheric ozone is known to have a damaging effect on carbon uptake in the terrestrial biosphere. We show that limitations of available nitrogen for sufficient plant growth reduce the negative impact of tropospheric ozone on carbon uptake in plants, leading to a smaller indirect change in radiative forcing than previously calculated. Transient climate simulations between 1900 and 2004 where plant growth is affected by tropospheric ozone have been performed by the National Center for Atmospheric Research Community Land Model with and without a coupling to the nitrogen cycle. When the land model includes nitrogen limitation on plant growth, the negative effect from tropospheric ozone on carbon uptake in plants is reduced by up to a factor of four compared to model simulation without nitrogen limitation. Only 2–5% of the radiative forcing from CO2 between 1900 and 2004 can be attributed to the indirect effect of tropospheric ozone which is a factor of six lower than results from previous studies.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Forests have an important role in the global carbon cycle. Through photosynthesis, plants and trees use sunlight to convert CO2, water, and nutrients into sugars and carbohydrates, which accumulate carbon in leaves, stems, and roots. CO2 is also released by plants through the reverse process respiration. Overall, the terrestrial biosphere acts as a global net carbon sink [Ballantyne et al., 2012; Pan et al., 2011]. However, human activities have influenced the carbon cycle by increasing levels of greenhouse gases and toxic air pollutants in the atmosphere [Denman et al., 2007; Fowler et al., 2009]. Tropospheric ozone (O3) is one climate compound that has increased since preindustrial times and reached levels that are damaging to vegetation. It is currently considered to be the greenhouse gas and air pollutant which affects terrestrial ecosystems the most [Ashmore, 2005; Karnosky et al., 2007]. O3 is not directly emitted into the atmosphere but is a product of photochemical reactions with precursors such as NOx, CH4, CO, and volatile organic compounds. O3 is ranked third among the greenhouse gases causing a radiative forcing of 0.35 Wm−2 since preindustrial time [Forster et al., 2007]. In addition, O3 is an air pollutant of major concern, having negative effects on human health, causing reduced crop yield, and weakening forest growth [Karnosky et al., 2007]. It enters the leaves of plants primarily through the stomata, and it then reduces productivity through cellular damage and biochemical aspects of photosynthesis. This damage results in decreased rates of photosynthesis and stomatal conductance [Wittig et al., 2007]. When O3 supresses carbon uptake through reduced photosynthesis, more CO2 is accumulating in the atmosphere. O3 has therefore an indirect radiative forcing on CO2 by its effect on the terrestrial biosphere. The indirect effect of O3 on the global carbon cycle operates on a much longer timescale than the tropospheric ozone which itself has a short lifetime and can thus have a significant impact on global climate [Unger and Pan, 2012]. This indirect effect has been calculated to be comparable to the direct forcing of O3 which is estimated to be in the range between 0.2 Wm−2 and 0.4 Wm−2 at present-day relative to 1900 [Sitch et al., 2007]. However, a constraint on carbon sequestration in the terrestrial biosphere is sufficient nitrogen supply for growth. Nitrogen control of photosynthesis has been revealed as a fundamental process to be included in climate-carbon cycle model simulations [Thornton et al., 2007].
 Although several previous studies have examined the individual effects of changes in atmospheric CO2 [Friedlingstein et al., 2006] and O3 [B Felzer et al., 2004; Sitch et al., 2007] on the land carbon sink, information on their combined effect and the importance of available nitrogen in the soil is not well understood and not accounted for in current generation of climate-carbon cycle models [Friedlingstein et al., 2006]. Global studies of the combined carbon-nitrogen (CN) coupling and O3 effect on the land carbon sink do not yet exist. Such studies are very useful when we want to use radiative forcing as a measure in order to compare the results with other forcing mechanisms. However, model studies of changes in evapotranspiration, runoff [Felzer et al., 2009], and carbon sequestration [Felzer et al., 2004] in United States, which take the N-cycles into account, exist and emphasize the importance of linking these effects together in climate modeling.
 This study focuses on the effect of the carbon-nitrogen cycle on the reduced land carbon sink since 1900 caused by O3. We estimate the accumulated atmospheric CO2 and associated radiative forcing due to O3 in a land model with and without nitrogen limitation on plant growth. While most previous land model studies have investigated how the nitrogen cycle influences the responses in the terrestrial carbon cycle [Thornton et al., 2007; Zaehle et al., 2010], here we take the next step to evaluate how the nitrogen cycle has influenced the twentieth century global terrestrial carbon cycle responses to both CO2 and O3.
 The National Center for Atmospheric Research Community Land Model version 4 (NCAR CLM4) is the land component of the Community Earth System Model and includes a carbon-nitrogen biogeochemistry model [Lawrence et al., 2011; Oleson et al., 2010; Thornton and Rosenbloom, 2005]. When the carbon-nitrogen biogeochemistry is active (CLM4CN), a potential gross primary production (GPP) is first calculated without any constraints to available nitrogen. The decomposition of soil organic matter in the carbon and nitrogen cycle does not only release carbon but also soil mineral nitrogen that is used to supply two different processes: plant growth and further decomposition by microorganisms (immobilization). After supplying immobilization with nitrogen, remaining nitrogen is used to plant growth. If the plant specific demand of nitrogen is not met due to the remaining nitrogen, the potential GPP is reduced accordingly to an actual GPP. The actual GPP is thus down regulated by the dynamics of available soil mineral nitrogen. When the carbon and nitrogen cycle interactions are included in CLM, the effect of elevated CO2 on land carbon sink is reduced by 74% [Thornton et al., 2007]. For more detailed description of the CN coupling in CLM4CN, see Thornton et al.  and Thornton et al. . The carbon-only version of the model, CLM4C, includes no down regulation of GPP and nitrogen limitation. We use the option of turning on and off the nitrogen cycle in order to study the importance of carbon-nitrogen coupling for the effect of tropospheric ozone on carbon uptake.
 Previous studies on the O3 effect on the land carbon sink using global climate models assume that increasing surface ozone concentration reduces the photosynthetic rate [Collins et al., 2010; Sitch et al., 2007]. We use the method introduced in Sitch et al.  where the net photosynthesis is modified by a fractional f-factor that is controlled by the following model parameters: surface ozone concentration (nmol/m3), model-dependent parameters such as leaf conductance of H2O (sm−1), boundary layer resistance between leaf surface and reference level (sm−1), and two plant functional type specific sensitivities to O3 uptake (low and high). The low and high sensitivities for each plant type are introduced as lower and upper limits in the range of how O3 impacts the photosynthesis for specific plant types. The solution of the f-factor equation is calculated numerically within the model for each time step. Figure 1 shows the f-factor for different vegetation types included in this study and how it decreases for increasing levels of surface ozone. The f-factor for broad-leaf trees, needle-leaf trees, and shrubland with low plant sensitivity is close to having a linear relationship for increasing O3 concentrations as suggested by Wittig et al. . However, C3 and C4 grass with low sensitivity and for all plant types with high sensitivity, the f-factor is nonlinear and declines with a stronger gradient toward increasing levels of O3 up to approximately 1000 nmol/m3 and then flattens out. As Wittig et al.  also pointed out, O3 has different impact on photosynthesis for different plant types and concentration levels which the modified photosynthesis approach takes into account. The temporal resolution of the surface ozone fields in the simulations including the modified photosynthesis approach in CLM4 is monthly averages from 1900 to 2004 calculated by the Oslo-CTM2 chemistry transport model [Skeie et al., 2011]. In a sensitivity test, we also use monthly averaged surface O3 concentrations from the Model for Ozone and Related chemical Tracers (MOZART) chemistry transport model [Horowitz et al., 2003].
 The experimental model setup consists of four transient offline simulations (Table 1) between 1900 and 2004 forced with historical greenhouse gases and aerosol concentrations, changes in nitrogen deposition, and land use change. The simulations start in 1850 but in order to compare our results to Sitch et al. , we use 1900 as a reference year for changes up to present day. NCAR-NCEP reanalysis data for 1948–2004 [Qian et al., 2006] is used to represent the whole simulated period between 1900 and 2004 which means that climate change between 1900 and 1947 is not considered. We study the changes in total ecosystem carbon (ΔCTOT) caused by O3 with or without a carbon-nitrogen coupling in the two following cases: CN = CLM4CN_O3 − CLM4CN and C-only = CLM4C_O3 − CLM4C.
 To study the suppression of carbon sequestration on the atmospheric CO2 concentration, we also run the model in a land/atmosphere configuration using the NCAR Community Atmosphere Model (CAM4) [Neale et al., 2011] where the atmospheric CO2 concentration is able to respond to changes in the biosphere. Two online simulations (see Table 1) are performed with fixed sea surface temperatures, and we study the change in atmospheric CO2 concentration in response to CO2 between these two simulations (CN_CO2 = CLM4CN_O3_CO2 − CLM4CN_CO2). The prognostic atmospheric CO2 concentration in CAM4 is calculated based on three components: (1) the simulated net ecosystem exchange of carbon between land and atmosphere received from CLM4, (2) historical data on fossil fuel emissions (gridded annual data taken from http://cdiac.ornl.gov/), and (3) historical data on carbon uptake in the ocean [Takahashi et al., 2009]. The three model simulations which include the effect of O3 on the land carbon sink are run twice with either the low or high plant specific sensitivity.
 Radiative forcing calculations are based on higher levels of CO2 in the atmosphere caused by suppression of the land carbon sink. In offline simulations, the reduction in total ecosystem carbon (in PgC/yr) is converted into an increase of CO2 in ppm [Denman et al., 2007]. We assume that 50% of the additional CO2 in the atmosphere from reduced carbon uptake is taken up by the oceans [Sabine et al., 2004], i.e., only half of the CO2 remains in the atmosphere and contributes to a radiative forcing. In online simulations, the model calculates the change in atmospheric CO2 concentration interactively. The change in CO2 concentration in parts per million (ppm) contributes to a radiative forcing that we calculate according to the logarithmic approach taken from Myhre et al. .
3 Results and Discussion
 Figure 2 shows the reduction in ΔCTOT caused by the indirect effect of O3 for two different surface ozone data sets in C-only simulations. The results from the two data sets are shown in comparison with the results in Sitch et al. . Corresponding numbers can be seen in Table S1. The two ozone data sets we have used appear to have a smaller effect for both low and high plant sensitivities compared to the results between 1900 and 2004 in Sitch et al. , especially for the low sensitivity. This is caused by lower O3 concentrations at present day in Oslo-CTM2 and MOZART (not shown here) which causes a higher increase in surface O3 between 1900 and 2004 in Sitch et al.  than in the two data sets shown here. Using different ozone data sets did not significantly impact total ecosystem carbon for either low or high plant sensitivity to ozone, so the following discussion focuses on ozone concentrations from Oslo-CTM2 only.
 ΔCTOT is continuously reduced between 1900 and 2004 due to elevated O3, see Figure 3. In 2004, C-only cases show a reduction in ΔCTOT of 31–83 PgC/yr (low-high), and when the nitrogen limitation is included, the ΔCTOT is reduced to 8–26 PgC/yr (numbers are summarized in Table 2). The CN coupling has a large effect on the change of total ecosystem carbon caused in response to O3. ΔCTOT is 3.9–3.2 times lower in CN-simulations compared to the C-only. The corresponding ranges of the radiative forcing estimates for C-only and CN coupling is 0.13–0.33 and 0.03–0.11 Wm−2, respectively. The radiative forcing estimates in C-only simulations are comparable to results in Sitch et al.  where they calculated a present-day indirect RF from O3 to be in the same range as the direct RF of tropospheric ozone, i.e., 0.34–0.38 Wm−2 [Forster et al., 2007] (see Table S1). We find a weaker forcing than Sitch et al.  in the C-only simulations, especially for the low plant sensitivity, caused by the use of a different land model and ozone concentrations from Oslo-CTM2, as explained above and shown in Figure 2. Similar to ΔCTOT, the estimated radiative forcing from the CN coupling is 4.3–3.0 times lower than C-only cases. Results from the CN-cases are more realistic estimates because the inclusion of the N-cycle in CLM4 provides an improvement of the representation of the carbon fluxes [Thornton et al., 2007]. Zak et al.  also pointed out that Sitch et al.  may overestimate the effect of O3 compared to chamber experiments performed with elevated CO2 and O3, which we hereby confirm and attribute to the effect of nitrogen limitation.
Table 2. Estimated Change in Total Ecosystem Carbon (ΔCTOT) and Radiative Forcing in C-Only and CN Offline Simulations
Radiative forcing (W/m2)
 The CLM4CN simulations with an online coupling to CAM4 provide estimates of the additional CO2 accumulation in the atmosphere caused by the reduced carbon sink in the biosphere. The annual global mean change in atmospheric CO2 concentration due to the indirect effect of surface ozone between 1900 and 2004 can be seen in Figure 4a, and the corresponding radiative forcings are shown in Figure 4b. Measurements of atmospheric CO2 show that annual global mean concentrations have increased with 80 ppm [Forster et al., 2007] (RF = 1.31 W/m2 [Skeie et al., 2011]) during this period. Our simulations estimate that, in 2004, 3–8 ppm of the additional CO2 is caused by the effect of tropospheric O3 (Table 3). Corresponding radiative forcing estimates are 0.03–0.07 W/m2. The terrestrial carbon uptake in online simulations is higher than in the offline simulations as expected due to the different treatment of carbon uptake in the ocean (see section 2).
Table 3. Estimated Change in Atmospheric CO2 and Radiative Forcing in Online Simulations
ΔCO2 in atmosphere (ppm)
Radiative forcing (W/m2)
 The simulations in this study include two counteracting effects caused by nitrogen: (1) increased growth by nitrogen deposition and (2) down regulated photosynthesis that is limited by available nitrogen in the soil. A sensitivity simulation (not shown here) without historical change in nitrogen deposition shows that the first effect of nitrogen causes only minor changes to plant growth (a 4% weaker radiative forcing) compared to simulations including historical change in nitrogen deposition. However, the second effect of nitrogen is particularly important to include in studies on O3 and plant growth because it reduces the radiative forcing by three to four times (see Table 2).
 The radiative forcing of the indirect effect of tropospheric ozone on plant growth is lower in land model simulations including limitations on available nitrogen. Estimates of radiative forcing calculated in NCAR CLM4CN offline simulations range from 0.03 to 0.11 Wm−2 depending on two different plant sensitivities to ozone (low-high). These results suggest that 2–8% of the radiative forcing from CO2 between 1900 and 2004 can be attributed to the indirect effect of O3. In additional online transient climate simulations, where the atmosphere is able to respond to changes in CLM4CN by influencing the atmospheric CO2 concentration, only 2–5% (0.03–0.07 W/m2) of the radiative forcing from CO2 between 1900 and 2004 can be attributed to the indirect effect of tropospheric ozone which is a factor of six lower than results from previous studies. These results highlight that nitrogen limitation to plant growth significantly decreases the impact of ozone on the land carbon sink and mitigates the indirect impact of ozone on radiative forcing. We therefore recommend that carbon-nitrogen coupling is included in future work that investigates ozone-climate interactions.
 This work has been funded by the Norwegian Research Council through the project CarboSeason. Computational facilities have been provided by the Norwegian Metacenter for Computational Science (NOTUR). We thank two anonymous reviewers for giving valuable comments to the manuscript.
 The Editor thanks two anonymous reviewers for their assistance in evaluating this paper.