Seawater acidification can be induced both by absorption of atmospheric carbon dioxide (CO2) and by atmospheric deposition of sulfur and nitrogen oxides and ammonia. Their relative significance, interplay, and dependency on water column biogeochemistry are not well understood. Using a simple biogeochemical model we show that the initial conditions of coastal systems are not only relevant for CO2-induced acidification but also for additional acidification due to atmospheric acid deposition. Coastal areas undersaturated with respect to CO2 are most vulnerable to CO2-induced acidification but are relatively least affected by additional atmospheric deposition-induced acidification. In contrast, the pH of CO2-supersaturated systems is most sensitive to atmospheric deposition. The projected increment in atmospheric CO2 by 2100 will increase the sensitivity of coastal systems to atmospheric deposition-induced acidification by up to a factor 4, but the additional annual change in proton concentration is at most 28%.
Combustion of biomass and fossil fuels releases SOx and NOx to the atmosphere. Due to the short (i.e., ~1 week) residence time of atmospheric sulfur (S) and nitrogen (N) oxides, their deposition in the marine realm occurs mainly in coastal areas downwind of the principal terrestrial source regions [Rodhe et al., 2002] and in waters with intensive shipping [Tsyro and Berge, 1997]. In marine systems, atmospheric S and N oxides eventually end up as H2SO4 and HNO3, either through chemical alterations in the atmosphere and wet deposition of H2SO4 and HNO3, or through dry deposition of SOx and NOx and subsequent fast hydration in the seawater [Doney et al., 2007]. Absorption of S and N oxides by seawater thus causes a decrease in total alkalinity (TA) and hence a lowering in pH, whereby the effect of S oxide deposition on TA is twice as high as the effect of N oxide deposition [Doney et al., 2007; Hunter et al., 2011].
Ammonia (NH3) emissions originate mostly from animal husbandry, as well as fertilizer use, and its deposition to marine systems causes an increase in TA and hence a rise in pH. However, in the atmosphere NH3 can be transformed to acidic N through various processes [Dentener and Crutzen, 1994] and when any of the deposited NH3 is nitrified, the alkaline flux effectively changes to an acidity flux with the same stoichiometry as N oxide deposition [Doney et al., 2007; Hunter et al., 2011].
While acidification of terrestrial and freshwater systems due to atmospheric deposition has been investigated for decades [Schindler, 1988], there are very few studies explicitly dealing with the effects of atmospheric deposition on seawater pH. Seawater acidification induced by absorption of CO2 and by atmospheric deposition co-occur, but their relative importance and interactions (synergistic or antagonistic) have been poorly documented. Using time series of rainfall and seawater CO2 parameters in the North Atlantic Ocean near Bermuda, Bates and Peters  showed that wet deposition of acids contributes at most 2–5% to surface water acidification. By coupling a biogeochemical model to a global circulation model, Doney et al.  found a similar magnitude of the contribution of anthropogenic S and N deposition to surface water acidification, but they calculated changes in carbonate chemistry as high as 10–50% of the CO2-induced changes close to source regions and in shallow seas. Hunter et al.  focused on three coastal seas and reported that atmospheric inputs of strong acids compensate the CO2-induced acidification. This implies that ocean acidification trajectories based on CO2 absorption alone provide an incomplete, biased picture. Hassellöv et al.  focused on global acidification due to shipping activities and found that in certain coastal hot spots shipping-based acidification can be of the same order of magnitude as CO2-induced acidification.
Although the three modeling studies [Doney et al., 2007; Hunter et al., 2011; Hassellöv et al., 2013] used a similar approach, their estimated contribution of atmospheric deposition to ocean acidification varies from negligible to substantial. This is due to differences in the systems and processes considered and in the way the effects of atmospheric S and N deposition on seawater pH are quantified. Here we use a simple model to identify key factors and to explain the effect of different initial conditions on the additive pH change due to atmospheric acid deposition.
2 Methodological Considerations
2.1 The Fraction of Atmospheric NH3 Deposition That Will Be Nitrified
The fraction of deposited NH3 that is nitrified is highly relevant for the calculated changes in TA and pH [Doney et al., 2007; Hunter et al., 2011]. As this fraction is poorly known, likely variable, and time scale dependent, these studies considered the two extreme cases (i.e., zero and complete nitrification of the deposited NH3). Including biological feedbacks resulting from enhanced N availability, which may especially be of importance in N-limited coastal marine ecosystems, Doney et al.  found that globally approximately 98% of the deposited NH3 will be nitrified over a time span of 10 years.
For studies focusing on the seasonal to interannual scales, we propose to use the parameterization by Yool et al.  who, based on a compilation of a global data set, suggested a mean specific nitrification rate of 0.2 d−1. If we include their lower (0.02 d−1) and upper (2 d−1) limits and convert to percentages of ammonia being used for nitrification, this leads to values of 43, 16, and 81%, respectively, of which the lower estimate is close to the 15% suggested by Middelburg and Soetaert  for the North Sea.
2.2 Open Versus Closed System Calculations
So far, modeling studies have adopted different strategies to calculate the effect of atmospheric deposition on seawater pCO2 (pCO2,sw) and thus acidification:
Hunter et al. , in four of their scenarios (“CO2 + SOx,” “CO2 + SOx + NOx,” “CO2 + SOx + NOx + NH3,” and “after nitrification”), and Hassellöv et al.  did not allow reequilibration of pCO2,sw with the atmosphere (closed system).
In their “after buffering” scenarios, Hunter et al.  restored pCO2,sw values to their original values, i.e., full reequilibration of pCO2,sw with the atmosphere (open system).
Doney et al.  dynamically calculated pCO2,sw values at each time step taking CO2 equilibration and atmospheric deposition into account.
The first approach, which neglects air-sea exchange of CO2, is limited to the additive effect of atmospheric CO2 uptake and acid deposition and results in too much acidification. The second approach restores pCO2,sw values and in this way numerically generates CO2 effluxes that modulate the decrease in pH. The third approach, in contrast, allows for simultaneous calculation of all processes, thus producing the most accurate pCO2,sw values. However, the choice for a specific approach highly depends on the time scale of the study. Doney et al.  looked at annual changes in a time span of 10 years using small time steps, highlighting the need for an iteratively determined pCO2,sw. Hunter et al.  considered 1 year, an interval in which pCO2,sw may be assumed to be in equilibrium with atmospheric pCO2 (pCO2,atm). However, Hassellöv et al.  used monthly intervals over a 1 year period where pCO2,sw was updated using measured rather than calculated data. This approach led to an overestimation of the calculated total surface water acidification in CO2-supersaturated systems.
2.3 Supersaturated Versus Undersaturated Systems
In natural systems, pCO2,sw can be out of equilibrium with pCO2,atm [Takahashi et al., 2009], either seasonally due to biological activity and/or temperature effects, or more permanently as a result of their hydrodynamic setting. The time scale on which pCO2,sw equilibrates with the atmosphere (approximately weeks) is long compared to the residence time of atmospheric S and N oxides (approximately days), indicating the potential for atmospheric deposition to influence the equilibration process. The decrease in TA caused by atmospheric S and N deposition and associated increase in pCO2,sw will have different effects on CO2-supersaturated and undersaturated systems. The direction and rate of CO2 air-sea exchange are thus key to accurate projection of seawater acidification. We illustrate this with a simple case study.
A comparison is made of three systems which have the same TA, salinity, and temperature but differ in their initial pCO2,sw. System one is initially in equilibrium with pCO2,atm. In the second, undersaturated system 11.5 µmol kg−1 yr−1 of dissolved inorganic carbon (DIC) is removed by biological processes, resulting in an initial pCO2,sw of 250 ppmv. The third, supersaturated system has an initial pCO2,sw of 600 ppmv, corresponding to a net DIC production of 17.4 µmol kg−1 yr−1. Using a time step of 1 year, we calculated the net effect on pCO2,sw and pH of (1) a constant acid deposition flux where either zero or complete nitrification of the deposited NH3 results in a TA decrease of 1.34 and 3.94 µmol kg−1 yr−1, respectively, and (2) increasing pCO2,atm to 936 ppmv in 2100 according to the output of the highest Representative Concentration Pathway, RCP8.5 [Meinshausen et al., 2011]. Air-sea exchange of CO2 was calculated using a simple kinetic rate law
with a rate constant k of 2 yr−1, so that there was incomplete equilibration with the atmosphere.
All three systems show an increase in pCO2,sw and a decrease in pH (Figure 1). Atmospheric deposition causes further decreases in pH, because the additional lowering in TA reduces the capacity of the system to buffer changes. Consistent with Doney et al.  and Hunter et al. , nitrification of atmospheric NH3 leads to lower pH values.
The sensitivity toward ocean acidification and the interaction with atmospheric deposition depend on the biogeochemical context, i.e., the biological (or physical) processes adding or removing DIC, moderated by CO2 air-sea exchange. The undersaturated system is most sensitive to CO2-induced acidification (∆pH is −0.431), as the enhanced gradient between pCO2,atm and pCO2,sw increases the CO2 influx. Atmospheric deposition, however, lowers this CO2 gradient, resulting in a reduced influx. As a result, this system is relatively least vulnerable to atmospheric deposition (additional decrease in proton concentration ([H+]) of 11.5% and 38.0% excluding and including nitrification, respectively). The reverse holds for the supersaturated system, where the CO2-induced ∆pH is −0.255. Here a higher pCO2,sw resulting from atmospheric deposition increases the gradient between pCO2,sw and pCO2,atm and thus leads to a higher CO2 efflux, i.e., it is more perturbed. This results in a relatively higher additive [H+] decrease of 14.4% and 47.8% excluding and including nitrification, respectively.
2.4 Model Description
Based on the discussion above, we extended the model of Hunter et al. . Rather than assuming two extreme scenarios for nitrification and outgassing of CO2 to the atmosphere (zero versus complete), we implemented kinetic descriptions for both processes: the Yool et al.  parameterization for nitrification (section 2.1) and a simple air-sea exchange expression
with a transfer velocity kd of 2.7 m d−1, which corresponds to a wind speed of 7.6 m s−1 [Liss and Merlivat, 1986], within the range of median wind speeds for the European coastal zone [Gazeau et al., 2004]. The model was run until the end of the 21st century assuming that atmospheric deposition fluxes remain constant within this time period and the increase in pCO2,atm follows the RCP8.5 scenario. Horizontal advection and exchanges with the deep water were not taken into account. Finally, in addition to the southern North Sea, Baltic Sea, and South China Sea case studies presented by Hunter et al. , we included the northwestern (NW) Mediterranean Sea as a fourth case study (for parameterization see Table S1 in the supporting information).
We calculated the evolution of pH using the explicit pH modeling approach of Hofmann et al. [2010b] (see supporting information for extensive explanation). Briefly, the main advantage of this approach is that changes in pH can be directly attributed to the different processes affecting pH:
Here Rp is the rate of a process p (mol kg−1 yr−1) and sensitivity (Sp) is defined as the ratio of a stoichiometric coefficient for the proton in the reaction and a buffer factor (β):
The buffer factor is defined as
This way, all processes affecting pH can be included simultaneously while their individual contribution can still be extracted. All calculations were performed on the total pH scale using the R package AquaEnv [Hofmann et al., 2010a] with a time step of 1 year, using the equilibrium constants of Mehrbach et al.  as refitted by Dickson and Millero  for the carbonate system and of Dickson  for sulfate.
With 43% of the atmospheric NH3 being nitrified ∆pH is smallest (−0.336) in the NW Mediterranean Sea and largest (−0.386) in the southern North Sea (Figure 2a). The latter coastal system is also most sensitive to changes in the fraction of atmospheric NH3 deposition that is nitrified. With 16% nitrification the total pH decrease for the North Sea is 0.0128 smaller, while pH decreases by an additional 0.0187 with a fraction of 81%. For the South China Sea these numbers are slightly lower, while for the other two seas varying the degree of nitrification leads to a negligible change in the total acidification (data not shown). Comparing our results with model runs with CO2-induced acidification only (Figure 2b), we see that the additive ∆pH due to atmospheric acid deposition is also highest (0.0430 until 2100) in the southern North Sea. The contribution of acid deposition is smallest in the NW Mediterranean Sea (0.0006 additional pH decrease until 2100), probably due to its deep mixed layer and high buffering capacity (Figure 2f).
The contribution of the processes involved in proton cycling to the net change in [H+] is shown for the southern North Sea in Figure 2c. The net change (gray line) increases with time until 2071, after which it slows down. The magnitude of [H+] change induced by each process increases with time. Atmospheric acid deposition rates are constant, so the increase in d[H+]/dt with time can be attributed directly to increased sensitivity (Sp). Air-sea exchange of CO2 (Rexch), however, does not remain constant (Figure 2e) since both atmospheric and seawater CO2 change with time. This explains why the relative contribution of CO2 air-sea exchange to acid deposition first decreases and then increases with time (Figure 2c), following the trend in Rexch (Figure 2e). However, the absolute value of [H+] change is primarily controlled by the increase of Sexch (Figure 2d) rather than changes in Rexch.
Using CO2 air-sea exchange as an example, we show that changes in sensitivity predominantly result from changes in the buffering capacity. The stoichiometric coefficient for the proton decreases slightly until 2100, ranging from 3.9% in the Baltic Sea to 5.8% in the South China Sea (Figure 2d). Changes in the stoichiometric coefficient for the proton of nitrification are even smaller because the equilibrium constant of the NH4+/NH3 acid-base pair is far from the range of pH considered here. The buffer factor β decreases by 57% in the Baltic Sea and roughly fourfold in the southern North Sea (Figure 2f). The fourfold decrease in buffering (β) by the end of the 21st century combined with minor changes in stoichiometric coefficients for the proton means that the southern North Sea becomes about 4 times more sensitive to any process involving proton transfer. Differences in buffering capacity also explain why the Baltic Sea is the most sensitive and the South China Sea and NW Mediterranean Sea least sensitive to processes consuming or producing protons, as β is inversely related to Sp (equation (5)).
Increasing pCO2,atm is by far the most important process contributing to the total change in pH (Figure 2b), but this does not imply that CO2 air-sea exchange dominates proton cycling (Figure 2c). The cumulative effect of atmospheric deposition is a production of protons (Figure 2c), inducing an increase of pCO2,sw. However, at the same time the increment in pCO2,atm lowers the air-sea gradient of CO2. The balance between both processes determines whether the net CO2 flux is in or out. In the southern North Sea, the increase of pCO2,sw by acid deposition is stronger than the increase of pCO2,atm, resulting in an efflux and implying proton consumption due to CO2 air-sea exchange (negative Rexch, Figure 2e). Thus, the acidifying effect of a growing pCO2,atm is here masked in the proton balance. In the other three seas, the increment of pCO2,atm dominates, resulting in net proton production (positive Rexch, Figure 2e).
4.1 Comparison With Previous Studies
The global annual mean (0.00037) and median (0.00018) decrease in pH due to total atmospheric acid inputs as calculated by Hassellöv et al.  are an order of magnitude smaller than the measured annual open-ocean acidification [Santana-Casiano et al., 2007; Midorikawa et al., 2012]. With the RCP8.5 scenario for pCO2,atm, pH at our four coastal sites is expected to decrease at a magnitude roughly similar to global estimates [Bopp et al., 2013]. The inclusion of atmospheric acid deposition leads to an additional change in [H+] of at most 28%, which appears to be higher than the average global contribution. However, annual acidification at two distinct coastal sites, i.e., the southern North Sea [Provoost et al., 2010] and Tatoosh Island, Washington state, USA [Wootton and Pfister, 2012], currently occurs at a rate 1 order of magnitude higher than in the open ocean. This is an indication that some important processes that affect pH in coastal areas are not included in our model. Possible mechanisms are site specific and may include alkalinity inputs from shelf sediments [Thomas et al., 2009], increased upwelling [Feely et al., 2008], increased inputs of riverine dissolved organic carbon [Wootton and Pfister, 2012], and changes in the production-respiration balance [Borges and Gypens, 2010; Provoost et al., 2010].
4.2 Regional and Saturation State-Related Differences
Regionally distinct initial conditions are known to influence the response of a system to increasing pCO2,atm. For example, the Arctic Ocean is widely recognized to be one of the systems most vulnerable to increasing pCO2,atm, because of its naturally low pH, buffering capacity, and temperature [Orr et al., 2005], while low-latitudinal regions are generally less susceptible to CO2 invasion [Egleston et al., 2010]. This is the first study showing that initial conditions also play a role in the effect of atmospheric acid deposition on pH.
Systems supersaturated with respect to pCO2,atm, such as heterotrophic ecosystems where respiration exceeds gross production, are most sensitive to additive acidification by acid deposition, whereas CO2- undersaturated systems, in particular autotrophic ecosystems, are least sensitive. The metabolic balance and CO2 saturation conditions of coastal ecosystems are spatially and temporally heterogeneous and their role as a source or sink of atmospheric CO2 is debated [Cai et al., 2006; Chen and Borges, 2009]. The southern North Sea and northern part of the South China Sea are reported to be sources of CO2 to the atmosphere [Thomas et al., 2004; Zhai et al., 2005]. Thus, atmospheric acid deposition strengthens the outgassing fluxes from these seas. In contrast, both the northwestern Mediterranean Sea [Durrieu de Madron et al., 2003] and the Baltic Sea [Thomas et al., 2010] are generally sinks of atmospheric CO2, and atmospheric acid deposition leads to a weakened CO2 uptake in these seas.
The spatial and temporal variability in coastal air-sea CO2 gradients, which ranges, e.g., from approximately −100 to 100 ppmv in the southern North Sea [Thomas et al., 2004], introduces uncertainty to our calculated annual CO2 air-sea gas exchange rates. Additionally, selecting one from the several parameterizations that exist for kd adds a factor of 2 uncertainty [Garbe et al., 2014].
4.3 Future Projections
Globally, atmospheric S and N emissions are expected to decline within the next century due to air-quality regulations and reduced fertilizer and fossil fuel use. However, emissions tend to become relatively more concentrated in economic growth regions such as China, India, and Brazil [Van Vuuren et al., 2011]. Generally speaking, low-latitudinal shelf seas and near-shore coastal areas are CO2-supersaturated, while CO2-undersaturated shelf seas are mostly found at middle to high latitudes [Chen and Borges, 2009]. Since most economic growth regions are present at low latitudes, their CO2-supersaturated coastal systems are especially vulnerable to future acid deposition.
This research is supported by a Sea and Coastal Research fund (83910502) of the Netherlands Organisation for Scientific Research (NWO). The idea for this paper came from the deliberations of GESAMP Working Group 38, the Atmospheric Input of Chemicals to the Ocean. We thank both reviewers for their constructive comments that have significantly improved the paper.
The Editor thanks two anonymous reviewers for their assistance in evaluating this paper.