Increased food production for the ever-growing world population has been enabled by the widespread agricultural expansion and intensification with heavy fertilizer applications, which have correspondingly led to an enhancement in ammonia (NH3) emissions from the land by a factor of 2 to 5 since preindustrial times (Behera et al., 2013; Gu et al., 2015; Zhu et al., 2015). For instance, Asia (excluding Siberia), home to more than 60 % of the world population (FAOSTAT, 2021), has experienced a rapid expansion of agricultural activities (Liu and Tian, 2010; Tian et al., 2014), accounting for ∼50 % of the global total consumption of synthetic fertilizer and 30 %–40 % of global manure production (FAOSTAT, 2016). Agriculture-related activities are known to be the most significant sources of atmospheric NH3, of which the vast majority (∼60 %) originates from the excessive use of nitrogenous fertilizer and concentrated operations of livestock feeding on a global scale (Huang et al., 2012; Paulot et al., 2014; Zhang et al., 2018). For Asia, the percentage is even higher (80 %–90 %; Streets et al., 2003; Reis et al., 2009; Gu et al., 2012; Kang et al., 2016; Zhang, 2017; Zhang et al., 2018). Crops typically take up only about 40 %–60 % of the nitrogen fertilizer applied to croplands (Tilman et al., 2002; Zhang et al., 2015; Liu et al., 2016; Mueller et al., 2017), and only 25 %–35 % of the nitrogen fed to dairy cows is converted into milk (Bittman and Mikkelsen, 2009), while most of the remainder is chemically transformed into a variety of simple and complex forms and leaked to the environment. The release of gaseous NH3 into the atmosphere is one of the major nitrogen leakages from agricultural soils. Under a business-as-usual scenario, where future nitrogen use efficiency (NUE; i.e., the fraction of nitrogen input finally harvested as output) in agricultural systems is not expected to be substantially improved, increasing food production will undoubtedly continue to intensify agricultural NH3 emissions into the overlying air (Erisman et al., 2008; Lamarque et al., 2011; Zhang, 2017).

Reactive nitrogen, from the emissions of nitrogen oxides (NOx; NO + NO2) and NH3, is deposited over the land and ocean through a variety of processes collectively known as wet and dry deposition. As the combustion-driven NOx emission is projected to slow down due to regulatory efforts (van Vuuren et al., 2011), while agricultural NH3 emission will continue to increase (Lamarque et al., 2011), future nitrogen deposition is expected to increase overall in the global budget (Galloway et al., 2003; Paulot et al., 2013; Lamarque et al., 2013; Kanakidou et al., 2016) and shift from a nitrate-dominated to ammonium-dominated condition (Ellis et al., 2013; Paulot et al., 2013; Li et al., 2016). Atmospheric nitrogen deposition onto the land surface is an important source of soil mineral nitrogen and, thus, enhances plant growth; this is known as the nitrogen fertilization effect (Reay et al., 2008; Templer et al., 2012). The fertilization effect depends on the soil nitrogen limitation defined as the nitrogen constraint on the productivity of many terrestrial ecosystems (Vitousek et al., 2002; Gruber and Galloway, 2008; LeBauer et and Treseder, 2008; Heimann and Reichstein, 2008; Reay et al., 2008; Zaehle et al., 2010). Nitrogen limitation is often found in natural soils where severe nitrogen competition among plants and microbes exists, and the unmet plant nitrogen demand can be translated to a reduction in the potential gross primary production (GPP) of the terrestrial ecosystems, representing a direct downregulation of photosynthetic carbon gain.

Nitrogen deposition affects the terrestrial carbon and nitrogen cycle, but much less is known about how nitrogen deposition affects atmospheric chemistry via terrestrial changes and feedbacks. As nitrogen limitation is relaxed, enhanced carbon assimilation can be translated to changes in the carbon mass allocated to different plant parts, ultimately manifested as an enhancement in vegetation structural variables such as leaf area index (LAI) and canopy height. Meanwhile, nitrogen deposition can also alter soil inorganic nitrogen composition and a variety of abiotic and biotic processes, including uptake by plants, nitrification, denitrification, immobilization by microbes, and fixation in clay minerals. Soil NOx is produced as a byproduct of nitrification and denitrification, which are two microbial processes that first convert NH3 aerobically to nitrate (NO3-) and then NO3- to nitrous oxide (N2O) or nitrogen gas (N2) under anoxic conditions. As LAI, canopy height and soil NOx are known to affect surface air quality, nitrogen deposition can potentially affect atmospheric chemistry through affecting vegetation structure and ecophysiology, as well as soil biogeochemistry.

Nitrogen-mediated changes in vegetation and soil can affect surface ozone air quality via various pathways (Fig. 1). Among them, biogeochemical effects are processes mediated via direct exchange (i.e., emissions or deposition) of relevant chemical species between the terrestrial biosphere (vegetation and soil microbes) and the atmosphere, while biogeophysical or meteorological effects are mediated through a modification of the overlying meteorological environment (i.e., temperature, humidity, turbulence structure, etc.), as defined in Sadiq et al. (2017), Zhou et al. (2018), and Wang et al. (2020). One possible biogeochemical pathway is that LAI enhancement could elevate surface ozone by increasing biogenic volatile organic compound (VOC) emissions in high-NOx environments but could also reduce ozone by increasing the dry depositional uptake via leaf stomata (Zhao et al., 2017; Fu and Tai, 2015). Another possible biogeochemical effect is via the increase in canopy height, which further enhances surface roughness length, turbulent mixing, and, thus, higher aerodynamic conductance for land–atmosphere exchange including, ozone dry deposition (Bonan, 2016; Oleson et al., 2013). Another possible biogeochemical effect is that increased inorganic nitrogen availability facilitates soil NOx emission through nitrification and denitrification processes, which further causes rapid NO and NO2 cycling for ozone formation. Biogeophysical effects or meteorological effects are through vegetation-induced changes in the surface energy balance (e.g., absorbed solar radiation and sensible and latent heat fluxes) and subsequent changes in surface temperature, precipitation, humidity, circulation patterns, and moisture convergence (Wang et al., 2020). Higher temperature enhances ozone mainly through increased biogenic emissions and higher abundance of NOx, while lower humidity reduces the chemical loss rate of ozone (Jacob and Winner, 2009). Surface ozone changes via each individual process are heterogeneous over the globe, and the overall ozone response through various biogeochemical and biogeophysical pathways is highly complex (Zhao et al., 2017).

Here we present a study that investigates how agriculture-induced increases in NH3 emissions and subsequent nitrogen deposition could affect surface ozone air quality via terrestrial ecosystem changes in terms of LAI, canopy height and soil NOx emissions. We used an asynchronously coupled modeling framework based on the atmosphere (Community Atmosphere Model with Chemistry – CAM-Chem) and land (Community Land Model – CLM) components of the Community Earth System Model (CESM) to quantify the corresponding responses of surface ozone air quality to terrestrial changes. We first examined the responses of vegetation and soil variables to the present-day vs. future scenarios of nitrogen deposition and then used those terrestrial changes to drive factorial simulations for surface ozone. To evaluate the relative importance of LAI, canopy height, and soil NOx emissions, we evaluated ozone responses to the three individual effects and the overall combined effects using prescribed meteorology (i.e., large-scale meteorological responses to terrestrial changes are not allowed). Furthermore, we evaluated the effects of changing meteorology to surface ozone by conducting simulations using dynamic meteorology (i.e., where the overlying boundary layer meteorology and large-scale circulation also respond to terrestrial changes). Model configuration with dynamic meteorology represents the overall effects from regional terrestrial changes and associated meteorological changes (an integration over both biogeochemical and biogeophysical effects to surface ozone), whereas the setting with prescribed meteorology provides limited above-surface-layer meteorological changes directly caused by terrestrial changes and represents the biogeochemical effects only. Our study emphasizes the complexity of biosphere–atmosphere interactions and their indirect modulating effects on air quality and atmospheric chemistry, which are important for evaluating the impacts from future food production trends on air quality and health beyond the direct effects of agricultural emissions alone.

We first show the year 2000 emissions of reactive nitrogen as NOx (48 Tg N yr-1; Fig. 2a) and NH3 (53 Tg N yr-1; Fig. 2b), with a global budget of 101 Tg N yr-1, which is in good agreement with Ciais et al. (2013). NOx is densely emitted from industrial and populated regions, while hotspots for NH3 emission are India and eastern China, with intensive agricultural activities and inefficient fertilizer use. The global year 2050 NH3 emission is projected to reach 67, 57, 65, and 71 Tg N yr-1 in the Representative Concentration Pathways (RCPs) of RCP2.6, RCP4.5, RCP6.0, and RCP8.5, respectively, mainly due to rising agricultural production (RCP database version 2.0.5). Yet, RCP projections did not include a sufficient representation of the spatial patterns of agricultural NH3 emissions worldwide – especially in Asia, which has the world's most productive croplands (RCP database version 2.0.5). To capture the year 2000 to year 2050 agricultural intensification, we, therefore, estimated future NH3 emissions based on the FAO year 2000 to year 2050 crop production changes.

The FAO projects global year 2050 crop production to be higher than the year 2000 level due to changes in yield, crop intensity (i.e., multiple cropping and shortening of fallow periods), and arable land (Alexandratos and Bruinsma, 2012). The major increases occur in South America and central Africa due to yield increases and arable land expansion. Production growth factor g in Fig. 2c can go up to 2–3 for South America and 3–5 for central Africa, while it is 1.5–2 for some of the world's most productive croplands at northern midlatitudes, suggesting that the Southern Hemisphere will be playing an increasingly important role in producing food for the future global population. By scaling up year 2000 NH3 emissions by the growth factor in FAO crop production, we estimated that the year 2050 NH3 budget will be 71 Tg N yr-1, a 34 % increase (18 Tg N yr-1) compared to the year 2000 emission, with major increases over East China, India, the midwestern United States, Brazil, Argentina, and East Africa (Fig. 2d). This estimate is comparable to the RCP8.5 estimate of 71 Tg N yr-1, as both studies assumed a business-as-usual scenario, where future NUE in agroecosystems is not expected to be improved much. We fed both year 2000 and year 2050 NH3 emissions into the CESM model to simulate the corresponding nitrogen deposition. The global budget of both reduced (NHx) and oxidized (NOy) nitrogen deposition is 101 Tg N yr-1 in year 2000 (Fig. 3a), which almost balances out the emission totals of both NH3 and NOx. Nitrogen deposition in year 2050 is 121 Tg N yr-1, a 20 % (20 Tg N yr-1) increase from the year 2000 total (Fig. 3b). Increases in year 2000 to year 2050 nitrogen deposition mostly result from increased NHx deposition, since we fixed the NOx emission at the year 2000 level to isolate the deposition changes due to agricultural intensification alone. This increased nitrogen deposition serves as an important input of mineral nitrogen from the atmosphere to the biosphere.

We present, in this section, the fertilization effect of year 2050 nitrogen deposition and associated enhancements in vegetation structure (i.e., LAI and canopy height) and soil NOx emissions compared with those of the year 2000 nitrogen deposition. Nitrogen uptake from the soil is an important determinant of plant growth, as nitrogen is a major component of chlorophyll (i.e., pigments absorbing light energy for photosynthesis) and RuBisCo (i.e., the enzyme necessary for carbon fixation). Meanwhile, mineral nitrogen availability is also vital for nitrification and denitrification microbial processes where NOx is produced as a byproduct. In CLM, the plant nitrogen demand for new growth is calculated by the carbon available for allocation to new growth allocation, given the C:N stoichiometry of a given plant type and plant part. From the soil side, soil mineral nitrogen supply is calculated by adding various nitrogen sources (e.g., atmospheric nitrogen deposition, fertilizer, and biological nitrogen fixation) and subtracting nitrogen sinks (e.g., leaching and assimilation by heterotrophs). When the plant nitrogen demand is greater than the soil nitrogen supply, the plants are not able to take up enough nitrogen to support the carbon allocation for new growth, which would then be reduced (downregulated) by a percentage in the model, which we refer to as the soil nitrogen limitation on plant growth here. When the soil is nitrogen limited, the plants are not able to take up enough nitrogen for maximum photosynthesis, and unmet plant nitrogen demand is translated back to a carbon supply surplus, which is eliminated through the reduction of the GPP in the CLM model. Figure 3c shows the year 2000 GPP percentage reductions due to nitrogen limitation. Most of the nitrogen-limited soils are found over the boreal forests because of slow soil decomposition and turnover, with litter of a high C:N content and a cold climate. Savannas and grasslands in the tropics are also mildly nitrogen limited because of low foliar nitrogen concentrations and plant density. Figure 3d shows the differences in GPP reductions, i.e., year 2050 GPP reductions minus year 2000 GPP reductions. We found smaller GPP reductions induced by nitrogen limitation in the year 2050 than in the year 2000, reflecting higher plant productivity and growth over year 2000 to year 2050. However, this nitrogen fertilization effect is found only over nitrogen-limited regions but not over nitrogen-abundant regions, such as India and northern China, where the critical nitrogen loads are almost always exceeded (Zhao et al., 2017) despite substantial increases in nitrogen deposition over year 2000 to year 2050.

Due to nitrogen fertilization, GPP, LAI, canopy height, and soil NOx emissions over nitrogen-limited regions are generally higher with the year 2050 nitrogen deposition (Fig. 4). Specifically, we found that the year 2050 nitrogen deposition to the land enhances global GPP by 2.1 Pg C yr-1 (Fig. 4e), and the enhanced carbon assimilation can be translated into changes in the carbon mass allocated to different plant parts such as leaves, stems, and roots. The two vegetation structural proxies in the CLM model, LAI, and canopy height, which characterize the carbon allocation to plant tissues, leaf, and stem, respectively. LAI was simulated to be higher by up to 0.3–0.4 m2 m-2 over tropical grasslands and croplands in Brazil, savannas in sub-Saharan Africa, and 0.1–0.2 m2 m-2 across boreal and temperate forests at midlatitudes (Fig. 4f). Canopy heights from broadleaf deciduous trees and needleleaf evergreen trees were simulated to be higher by up to 0.1–0.3 m over the eastern USA, southern Europe, southern Russia, and southeastern China. Increases of 0.3–0.4 m were found over broadleaf deciduous trees in South America, and ∼0.1 m increases were found for grasses and crops over sub-Saharan Africa (Fig. 4g). Meanwhile, the global soil NOx emission budget rises from 7.9 to 8.7 Tg N yr-1 (Fig. 4h) due to faster and greater nitrification and denitrification processes under the year 2050 atmospheric nitrogen deposition.

With the rising food need for the future world population, more intense agricultural activities are expected to cause substantial perturbations to the global nitrogen cycle, aggravating surface air pollution and imposing stress on terrestrial ecosystems. Much less studied, however, is how the ecosystem changes induced by agricultural nitrogen deposition may modify biosphere–atmosphere exchange and further exert secondary effects on global air quality. In this paper, we present a study to quantify the response of surface ozone air quality to vegetation structural (LAI and canopy height) and soil NOx emission changes under year 2000 vs. year 2050 agricultural ammonia emissions over centurial timescales by using an asynchronously coupled framework.

Agricultural ammonia emission in the coming decades is destined to increase. We estimated the year 2050 NH3 emissions to be 71 Tg N yr-1, a 34 % increase compared to the year 2000 emissions. Our estimate is comparable to 71 Tg N yr-1 made by RCP8.5, as both studies assumed a business-as-usual scenario where future NUE in agroecosystems is not expected to be improved by much. However, it should be acknowledged that increases in food production may also be obtained with a less-than-proportionate increase in fertilizer use as countries are developing a greater awareness of agriculture-related environmental impacts and adopting more efficient nutrient use practices in the coming decades. Gu et al. (2015) reported that reasonable changes in diet, NUE, and N recycling could reduce year 2050 N losses and anthropogenic reactive nitrogen creation to 52 % and 64 % of 2010 levels, respectively, in China. Fung et al. (2019) showed that the maize–soybean intercropping improves NUE by easing fertilizer application and NH3 volatilization in agricultural soils in China. Therefore, we acknowledge that the future paths of agricultural NH3 emissions and nitrogen deposition may differ from what we projected as a worst-case scenario in this study, but we do not expect the nature of the mechanisms and conclusions in this study to be altered significantly.

Atmospheric nitrogen deposition increases carbon uptake by terrestrial biosphere in nitrogen-limited areas and also stimulates release of NOx, nitrous oxide (N2O), and NH3 from soils (Reis et al., 2009; Zaehle et al., 2011). We found that nitrogen deposition increases by 20 % from year 2000 to year 2050, due to rising agricultural NH3 emissions, and this enhances global GPP by 2.1 Pg C yr-1. LAI was simulated to be higher by up to 0.3–0.4 m2 m-2 in the tropical grasslands and croplands and 0.1–0.2 m2 m-2 in the midlatitude boreal and temperate forests. Canopy height increases were found in boreal and temperate forests (by 0.1–0.4 m), as well as in tropical grasslands and croplands (by ∼0.1 m). The soil NOx emission budget rises to 8.7 Tg N yr-1 with the year 2050 nitrogen deposition because of intensive nitrification and denitrification processes. Due to decreasing trends of anthropogenic NOx emissions throughout this century (Myhre et al., 2013), soil NOx is expected to play an increasingly important role in the global NOx budget. Therefore, the inclusion of effects of soil NOx emissions to surface ozone is essential. These estimates are based on carbon and nitrogen interactions in CLM4.5 biogeochemistry (CLM4.5-BGC), which are widely used in estimating the long-term trajectory of terrestrial variations (Lombardozzi et al., 2012; Val Martin et al., 2014; Sadiq et al., 2017; Zhou et al., 2018). However, the internal soil nitrogen cycle, its coupling with the atmosphere, and reactive nitrogen gas emissions, other than N2O, are not fully represented in default CLM4.5-BGC. The soil NOx emission module that we added, which allows soil NOx to respond to nitrogen deposition from the atmosphere, partly improved the representation (Fung et al., 2021), but the NH3 emission we used was still based on inventories and scaling with future crop production and, thus, did not respond to nitrogen deposition. We expect, however, that the secondary effect of nitrogen deposition on NH3 should be much smaller than any perturbations due to agricultural changes (Fung et al., 2021). Moreover, fully coupled bidirectional nitrogen fluxes were not enabled in our model setting. Future work is needed to examine the overall downstream biogeochemical and biogeophysical effects in an Earth system model with a closed nitrogen cycle, where soil NOx and NH3 emissions to the atmosphere and nitrogen deposition from the atmosphere are fully coupled dynamically.

With only the biogeochemical effects of nitrogen-induced terrestrial changes (with prescribed meteorology where meteorological changes are not included), surface ozone is elevated by 1–3 ppbv in certain low NOx equatorial regions due to increased soil NOx emissions, while LAI and canopy height only modulate surface ozone by ±0.5 and 0.2 ppbv, respectively. With both the biogeochemical and biogeophysical effects under dynamic meteorology, changes in summertime surface ozone are within ±2–3 ppbv. Ozone responses due to vegetation changes are much higher with dynamic meteorology than prescribed meteorology, as vegetation changes shift surface energy balance, circulation patterns, and moisture flow and, thus, shape ozone concentrations. Local meteorological variations induced by vegetation structural changes are generally more important than the vegetation changes per se in terms of modulating surface ozone concentration and appear to be as important as the biogeochemical soil NOx effect. Furthermore, biogeophysical pathways related to canopy height changes have not been accounted for by most previous studies of ozone–vegetation interactions, which usually only considered LAI and other ecophysiological changes (Wang et al., 2020; Wong et al., 2018; Zhao et al., 2017; Fu and Tai, 2015). Global vegetation growth is altered by land use and land cover change, warming, CO2 fertilization, nitrogen deposition, ozone damage, etc., but the associated canopy height changes have usually been ignored, rendering an incomplete representation of terrestrial effects on surface air quality predictions. Here, we found that the effects of canopy height changes on surface ozone through the biogeophysical pathways are noticeable and can be as much as the effects associated with LAI changes alone.

One limitation of this study is that we did not consider ozone damage on stomatal conductance and photosynthesis, as in the study by Sadiq et al. (2017). If ozone damage on stomatal conductance is considered, higher ozone concentrations could have positive feedbacks on ozone itself via reduced dry deposition and enhanced isoprene emission. Meanwhile, ozone damage on plant productivity may also diminish the fertilization effect of nitrogen and foliar nitrogen content, which is itself vital for photosynthetic capacity (Franz and Zaehle, 2021). Therefore, if ozone damage is considered, lower LAI and canopy height are expected, thus compensating for some of the enhanced LAI and canopy height induced by higher nitrogen deposition found in this study. These changes in LAI and canopy height could further affect ozone via various biogeochemical and biogeophysical pathways, but such a secondary feedback effect is expected to be relatively minor (Zhou et al., 2018). More work is warranted to investigate the individual and combined effects of nitrogen deposition and ozone damage on plant growth and terrestrial carbon uptake, especially in light of the possible nonlinear interactions between ozone and nitrogen in plants (e.g., Shang et al., 2021).

Overall, our study demonstrates a novel link between agricultural activities and ozone air quality via the modulation of vegetation and soil biogeochemistry by nitrogen deposition and highlights the particular importance of considering meteorological changes following vegetation structural changes, including those in canopy height and soil NOx changes, in studying the effects of ozone–nitrogen–vegetation interactions in the future.