Ocean acidification evolves on the background of a natural ocean pH gradient that is the result of the interplay between ocean mixing, biological production and remineralization, calcium carbonate cycling, and temperature and pressure changes across the water column. While previous studies have analyzed these processes and their impacts on ocean carbonate chemistry, none have attempted to quantify their impacts on interior ocean pH globally. Here we evaluate how anthropogenic changes and natural processes collectively act on ocean pH, and how these processes set the vulnerability of regions to future changes in ocean acidification. We use the mapped data product from the Global Ocean Data Analysis Project version 2, a novel method to estimate preformed total alkalinity based on a combination of a total matrix intercomparison and locally interpolated regressions, and a comprehensive uncertainty analysis. We find that the largest contribution to the interior ocean pH gradient comes from organic matter remineralization, with CaCO3 cycling being the second most important process. The estimates of the impact of anthropogenic CO2 changes on pH reaffirm the large and well-understood anthropogenic impact on pH in the surface ocean, and put it in the context of the natural pH gradient in the interior ocean. We also show that in the depth layer 500–1,500 m natural processes enhance ocean acidification by on average 28 ± 15%, but with large regional gradients.
- Globally the most important natural process changing interior ocean pH is organic matter remineralization
- CaCO3 cycling has the largest impact on interior ocean pH in the northern Pacific and Indian Oceans
- Remineralization enhance anthropogenic pH change and while today this enhancement is small it may grow more important as ocean acidification continues
Ocean pH is decreasing as a direct consequence of ocean uptake of carbon dioxide (CO2) emitted during the burning of fossil fuels, cement production, and land use changes. This is commonly referred to as ocean acidification (Caldeira & Wickett, 2003; Feely et al., 2004, 2009). The decrease is well documented globally (e.g., Bates et al., 2014; Lauvset et al., 2015) and there is active research focused on understanding its consequences (e.g., Gattuso et al., 2015; Kroeker et al., 2013). The impacts of ocean acidification on marine life have proven to be complex, and there is now ample evidence that other factors such as food availability, predator-prey dynamics, and temperature changes affect an organism's response to ocean acidification (e.g., Kroeker et al., 2013). In a recent metaanalysis, Hurd and Cornwall (2015) highlighted the importance of considering natural gradients in carbonate chemistry for experimental design, in addition to ocean acidification-related chemistry changes.
Understanding the natural processes that determine ocean pH is also important when communicating ocean acidification research. Ocean acidification happens on top of several natural processes, and the combination of the two make some regions particularly vulnerable. One such region is the Southern Ocean, where upwelling of waters rich in remineralized carbon create a local minimum in carbonate ion concentrations in the thermocline. As shown by Negrete-García et al. (2019) this natural gradient means that a very small anthropogenic perturbation is sufficient to shift the aragonite saturation horizon by several hundred meters from one year to the next. A similar effect has been shown by Hauri et al. (2013), Feely et al. (2018), and Franco et al. (2018) for Eastern Boundary Upwelling Systems, which are also regions where natural processes act to enhance the impact of ocean acidification. Such research highlights the fact that ocean acidification decreases natural pH distributions that result from ocean mixing, biological production and remineralization, mineral dissolution, temperature changes, and gas exchange.
Several studies have addressed how natural processes determine the ocean mean distribution of pH (e.g., Feely et al., 2004, 2009), but they are either descriptive or regionally focused. There is a lack of studies quantifying how natural processes control pH globally. Here we assess the subsurface patterns in global climatological ocean pH, and determine the degree to which the pH distribution and its spatial variability are governed by anthropogenic changes, the advection of surface waters into the interior (the so-called preformed component), organic matter remineralization, and carbonate mineral (mostly CaCO3, but also, e.g., MgCO3) dissolution. The goal is to properly evaluate how anthropogenic changes and natural gradients collectively act on ocean pH and how this sets the regional vulnerability to future changes in ocean acidification.
2 Data and Methods
Unless otherwise noted we use the Global Ocean Data Analysis Project version 2 (GLODAPv2) mapped climatologies (Lauvset et al., 2016) throughout this study, and all additional calculations are based on this product. The baseline value from which the various pH perturbations are estimated is pH2002, which is calculated from the observed total alkalinity (TA) and dissolved inorganic carbon (DIC) in GLODAPv2 (Olsen et al., 2016), where DIC is first adjusted to year 2002 using the atmospheric perturbation method as detailed in Appendix B of Lauvset et al. (2016), and the assumption that TA does not change over time. Since few ship-based observations of ocean carbonate chemistry are made in winter there is a recognized seasonal bias in the GLODAPv2 data product (Olsen et al., 2019), and hence in all calculations made using these data, which makes the data product challenging to use for the surface ocean. No attempt has been made to correct for this bias in this present work, although we note that this bias is primarily an issue for surface measurements and this analysis focuses on the ocean interior. We nevertheless note that previous studies (e.g.,Fassbender et al., 2018 ; Landschützer et al., 2018 ; Lenton et al., 2012) have indicated that there is a nonnegligible difference between summertime and wintertime trends in surface ocean pCO2, and it is possible that this is also the case for pH. Therefore, the anthropogenic CO2 impacts on pH (and Ω) estimated here, which align with previous estimates (e.g., Caldeira & Wickett, 2003), should be considered representative of the summertime. Relatedly, warming of the ocean also affects the amplitude of the seasonal cycling and ocean stratification, which has been shown to affect ocean pCO2 (Fassbender et al., 2018; Landschützer et al., 2018). The same effect is likely for ocean acidification, but the GLODAPv2 data product, due to the seasonal bias, cannot be used to assess this effect. This effect has therefore not been accounted for or discussed in detail in this study.
All GLODAPv2 data are available for download through www.glodap.info. In this study, pH and pH changes are always on the total hydrogen scale and, unless otherwise stated, always calculated for in situ temperature and pressure.
2.2 Preformed Values
To estimate preformed total alkalinity (TA0), we used a novel combination of the total matrix intercomparison (Gebbie & Huybers, 2011) and the Locally Interpolated Alkalinity Regression (LIARv2) methods (Carter et al., 2018). The approach used here is an evolution of the TA0 estimate employed by Feely et al. (2002), which relied on basin-specific surface ocean regressions of TA with properties that are conserved in the interior ocean. It addresses two problems with the earlier approach: first, ocean interior ventilation tends to occur during the wintertime when seawater properties are substantially different (e.g., colder) than in the summer. Since the data used to train the regressions have mostly been collected during summer months, this can substantially bias the regressions toward that season, that is, conditions during winter, the time of water mass formation, is not well represented (Carter et al., 2014). Secondly, available TA regressions are valid for specific regions of the surface ocean, and since interior ocean water often originates from distant regions or other ocean basins entirely, selecting which regression to use is problematic. This new approach addresses these challenges by estimating seawater properties just below the shallow summer mixed layers and then propagating these estimates into the global ocean using the total matrix intercomparison approach (Gebbie & Huybers, 2011).
To visualize our new preformed alkalinity estimates (TA0) we compare these values to the mapped TA in the GLODAPv2 data product (Lauvset et al., 2016; Figures S12 and S13). On the surface the difference is, as expected, very small (1 ± 20 μmol kg−1, where the uncertainty is 1 standard deviation of the mean; Figure S11a) while it increases with depth to an average 103 ± 6 μmol kg−1 at 3,000 m in the North Pacific Ocean (Figures S12c and S14; uncertainty is 1 standard deviation of the mean) and water mass age (Figure S1), as a consequence of carbonate mineral dissolution.
2.3 Anthropogenic Carbon and Preindustrial Values
Anthropogenic carbon (Canth) is calculated using the transit time distribution (TTD) method (Hall et al., 2002; Waugh et al., 2006) on all CFC-12 observations in GLODAPv2, following the method used by Lauvset et al. (2016), under the assumption that the ratio between the mean age and width of the TTD is unity (Γ/Δ=1). We used the TA0 estimated as described in section 2.2 to calculate Canth but note that the TTD based Canth estimates are rather insensitive to how TA0 is estimated (not shown). Canth was normalized to year 2002 following the method outlined in Lauvset et al. (2016). Preindustrial pH (pHPI), and aragonite (ΩArPI) and calcite saturation state (ΩCaPI) are calculated in CO2SYS (Lewis & Wallace, 1998) version 2.0 for MatLab® (Orr et al., 2018; van Heuven et al., 2009) from DICPI and TA using the carbonate dissociation constants of Lueker et al. (2000), the bisulfate ion dissociation constant of Dickson (1990), the total borate-salinity relationship of Uppström (1974), and the mapped phosphate and silicate fields from GLODAPv2.
The TTD method used to estimate Canth for this study is known to overestimate Canth in the Southern Ocean due to the assumption of constant equilibrium (He et al., 2018; Khatiwala et al., 2013; Waugh et al., 2006). Waugh et al. (2006) suggests a 20% global overestimate in Canth calculated using the TTD method, while Vazquez-Rodriguez et al. (2009) estimated that the TTD method overestimated Canth in the Southern Ocean by 2–3 μmol kg−1 relative to the Sabine et al. (2004) inventory. As a result, the numbers presented here are likely slightly overestimated.
2.4 Decomposition of pH
In the ocean interior pH changes are (i) thermodynamically driven, primarily through changes in temperature (ΔpHT) and pressure (ΔpHpress), and biogeochemically driven through (ii) net remineralization of organic carbon to inorganic carbon (ΔpHorg) and (iii) net dissolution of CaCO3 minerals (ΔpHCaCO3). The two latter processes do not directly change pH but rather add DIC or release TA (for carbonate dissolution) or titrate it away (for organic matter remineralization). These chemical changes then cause pH to change. In this paper, we consider the measured pH (pH2002) to be the baseline and explicitly calculate how much the interior ocean pH is changed by each process, as well as the accumulation of anthropogenic carbon (ΔpHanth). Our analysis is based on well-established methods that have been used to decompose DIC and TA in several previous studies (e.g.,Cameron et al., 2005 ; Couldrey et al., 2019 ; Eggleston & Galbraith, 2018 ; Gruber et al., 1996 ; Sarmiento & Gruber, 2006; Schmittner et al., 2013). In addition, a more simplified way to decompose interior ocean pH is detailed in Millero (2013), following Park (1969). All components were calculated in CO2SYS (Lewis & Wallace, 1998), using version 2.0 of CO2SYS for Matlab® (Orr et al., 2018; van Heuven et al., 2009), which includes routines for error propagation (Orr et al., 2018) used to estimate uncertainties for each component. The choice of dissociation constants is as described in section 2.3. Our results are visualized as cross sections for the four transects shown in Figure 1: (i) a line through the Nordic Seas along the zero meridian south to the Iceland-Shetland ridge; (ii) the Global Ocean Ship-based Hydrographic Program (GO-SHIP) lines A16 southward in the Atlantic Ocean, SR04 and S04P westward in the Southern Ocean, and P16 northward in the Pacific Ocean; (iii) the GO-SHIP line I09 southward in the Indian Ocean; and (iv) a line that crosses the Arctic Ocean from the Bering Strait to North Pole along 175°W and from the North Pole to the Fram Strait along 5°E. In addition, results are shown as global maps on three different depth horizons (20, 1,000, and 3,000 m).
Note that ΔpHorg is calculated using DIC and TA adjusted for the amounts gained through CaCO3 cycling (equation 3) following Sarmiento and Gruber (2006) under the assumption that CaCO3 dissolution happens after organic matter remineralization has changed both DIC and TA. This minimizes biases brought about by the nonlinear dependence of pH to DIC and/or TA perturbations.
As mentioned, all calculations are conducted at in situ temperature, under the assumption that this is conserved as water circulates through the interior ocean. Effects of hydrothermal heating and adiabatic expansion and compression of seawater are small and neglected. For example, the impact on pH of the heating water masses experience as they are transported from surface to higher pressures (adiabatic heating), is very small. Typically, a change of pressure from surface to 4,000 m generates a heating of about 0.4 °C, which would lead to a pH drop of ~0.006 units (as the pH change per 1 °C is ~0.015 under constant DIC and TA). The impacts of ocean mixing are captured in our preformed pH distributions (at in situ temperature). However, we caution that temperature plays a major role in driving the processes that lead to surface ocean variability in carbonate chemistry (Jiang et al., 2015) while only having a small net impact on the pH of surface waters that are well-equilibrated with the atmosphere (Jiang et al., 2019).
2.5 Uncertainty Analysis
The uncertainties in each input variable and parameter are listed in Table 1, and the error propagation equations for equations 3–13 are given in equations S2–S15 in the supporting information. Following recommendations by Orr et al. (2018) these uncertainties are defined as standard uncertainties (1σ), and we assume that the uncertainties in DIC and TA are uncorrelated. Since we include the mapping error the assumption of uncorrelated uncertainties is not strictly true since where mapping error for DIC is large it is also large for TA. This leads to an overestimation of the overall uncertainty (Orr et al., 2018). Throughout the text and in all figures the final combined uncertainties (Figures S4–S8) are presented as 2σ. Some important caveats regarding our uncertainty analysis are detailed in the supporting information.
|Input variable||Uncertainty (1σ)|
|σ (TA)0||5.5 μmol kg−1a|
|σ (DIC)||Combination of measurementb uncertainty (4 μmol kg−1) and spatially variable mappingc error summed in quadrature|
|σ (TA)||Combination of measurementb uncertainty (6 μmol kg−1) and spatially variable mappingc error summed in quadrature|
|σ (salinity)||Combination of measurementb uncertainty (0.005) and spatially variable mappingc error summed in quadrature|
|σ (temperature)||The mapping errorc|
|σ (phosphate)||Combination of measurementb uncertainty (2%) and spatially variable mappingc error summed in quadrature|
|σ (silicate)||Combination of measurementb uncertainty (2%) and spatially variable mappingc error summed in quadrature|
|σ (Canth)||Combination of estimation uncertainty (10%d) and the spatially variable mappingc error summed in quadrature|
|σ (AOU)||Combination of oxygen measurementb uncertainty (1%), spatially variable mappinge error, and uncertainty in oxygen saturation calculationf|
|σ (1.36 coefficient)||10%g|
|σ (pKsp-c) h,l||0.02|
|σ (total boron)h||1%|
- a RMSE of TA from Table 1 in Carter et al. (2018).
- b Measurement uncertainties are taken from the GLODAPv2 (Olsen et al., 2016). We assume zero measurement uncertainty in temperature.
- c Mapping uncertainties are taken from the GLODAPv2 mapped data product (Lauvset et al., 2016).
- d Estimated based on the results of He et al. (2018).
- e Assumed based on range of published, commonly used stoichiometric ratios (e.g., Redfield et al., 1963; Takahashi et al., 1985; Li & Peng, 2002; Körtzinger et al., 2001).
- f 0.3% following Garcia and Gordon (1992).
- g Estimated based on published values of this coefficient (Kanamori & Ikegami, 1982; Wolf-Gladrow et al., 2007).
- h Default uncertainty as given in the errors.m subroutine of the CO2SYS program (Orr et al., 2018).
- i Dissociation constant for boric acid.
- j Water dissociation constant.
- k Solubility product of aragonite.
- l Solubility product of calcite.
3.1 pH Gradients Due to Natural Processes
Interior ocean pH2002 gradients are substantially larger than surface pH2002 gradients, with an in situ pH2002 range between 7.52 ± 0.05 and 8.33 ± 0.10 in the interior ocean (Figures 2a and S4). Mostly, this range is due to biological processes that affect the interior ocean pH as outlined in section 2.4, and pressure effects. Pressure has a significant impact on pH in the deep ocean interior, primarily due to the pressure effects on the dissociation constants for carbonic acid (K1) and bicarbonate ion (K2), and causes a decrease in pH with depth (Figure 3a). Along our section (Figure 1), which is never deeper than 5,500 m, ΔpHpress has a maximum impact of −0.27 ± 0.11, that is, the chemical speciation changes that result from increased pressure at depth decrease the pH in the deepest parts of our section by 0.27.
Organic matter remineralization leads to a decrease in interior ocean pH (Figure 3b) because remineralization increases the concentration of inorganic carbon. Of the four components that create the interior ocean pH gradients (equation 2) ΔpHorg has the largest contribution overall, and has its largest impacts in the northern Pacific and the northern Indian Oceans where subthermocline water masses are typically older than 1,000 years (Figure S1): ΔpHorg is up to −0.79 ± 0.28 and −0.61 ± 0.23 in these regions, respectively (Figure 3b). As we use AOU to estimate organic matter remineralization there is naturally also a close correlation between maximum ΔpHorg and oxygen minimum zones, which are often not caused by high rates of organic matter remineralization, but rather lack of mixing and ventilation due to slow or stagnant circulation (Keeling et al., 2010). Along our transect, the spatial pattern ΔpHorg is also very similar to that in both water mass age (Figure S1) and DICorg (Figure S9) since the amount of remineralized carbon in seawater increases as the product of the age of the water mass and the average net rate of remineralization. That the age of the seawater is an important factor (r = 0.65) can clearly be seen in the steady decrease in the magnitude of ΔpHorg moving from south to north across the Pacific (Figure 3b). Both from the AOU (not shown) and the water mass age, the Southern Ocean (south of 45°S) appears quite recently ventilated, as expected due to active water mass formation in this region. Antarctic Intermediate Water and Subantarctic Mode Water are formed in the Antarctic Circumpolar Current, and spread northward to ventilate the Southern Hemisphere at ~1,000-m depth. These water masses show up with comparatively small ΔpHorg signals in all ocean basins (Figure 4b). There is nevertheless a significant subsurface ΔpHorg component (up to −0.37 ± 0.15) due to upwelling of quite old Pacific Deep Water and Upper Circumpolar Deep Water (Talley, 2013) near the polar front of the Antarctic Circumpolar Current (Figure 3b).
The importance of the rate of remineralization, approximated by dividing ΔpHorg by water mass age (Figures S10 and S11), can also be seen, for example, in the equatorial Atlantic where a ΔpHorg maximum is found in northward penetrating Antarctic Intermediate Water, which is younger than the North Atlantic Deep Water below it (Figures 4b and 4c), but shallower and subjected to higher rates of remineralization (Figures S11b and S11c). Local ΔpHorg maxima found in the upper 1,000 m of the ocean (Figures 3b and 4a–4c) are all associated with the thermocline or upwelling of older water and highlight that the largest vertical gradients in ΔpHorg are in the upper ocean, which is consistent with the Martin curve (Martin et al., 1987). Upwelling deep waters show up as patches of strong remineralization signals at 100 m on both sides of the equator in the Pacific, and along the U.S. western seaboard (Figure 4a). The deep ocean, on the other hand, generally has a very small, and spatially uniform, rates of remineralization (Figure S11c), but also here there are regional differences with well-ventilated areas generally having larger rates than old water masses. Related to this there is a clear inverse correlation (r = −0.69) between the rate of remineralization and Canth (not shown).
In contrast to organic matter remineralization, CaCO3 dissolution leads to an increase in pH (Figure 3c) because dissolution of CaCO3 releases carbonate ions, thereby elevating TA more than DIC and increasing the buffer capacity (Feely et al., 2002, 2004, 2012; Zeebe & Wolf-Gladrow, 2001). Note that the spatial pattern in ΔpHCaCO3 closely resembles that of ΔDICCaCO3 (see Sarmiento & Gruber, 2006, Figure 9.3.4), although our maximum is somewhat shallower than that shown by Sarmiento and Gruber (2006). The impacts of carbonate mineral cycling are typically much smaller than the impacts of organic matter cycling both because more carbon is cycled as organic matter than as carbonate minerals and because the effects on pH of increased TA (2 per CaCO3 dissolved) are partially offset by the effects of increased DIC (1 per CaCO3 dissolved). Interior ocean net carbonate dissolution lead to an increase in pH of at most 0.24 ± 0.12 in the North Pacific, and 0.22 ± 0.10 in the northern Indian Ocean. In both these regions, the maximum impact is centered around 2,500 m. In the North Pacific, the calcite saturation horizon is very shallow (200–700 m), although still approximately 100–300 m deeper than the aragonite saturation horizon (Figure 3c), which is consistent with previous studies (Feely et al., 2002, 2012). Though well-known and much described, the shallow and similar depths for aragonite and calcite saturation horizons are unusual since everywhere else the calcite saturation horizon is deeper than 3,000 m, and ~2,000 m deeper than the aragonite saturation horizon. This pattern is ultimately governed by the remineralization of organic matter, the DIC accumulation in the oldest water masses leads to greater decreases in pH—as discussed above—and corrosive conditions for calcite and aragonite. The global pattern of ΔpHCaCO3 therefore resembles the pattern of ΔpHorg, but with an opposite sign and reduced magnitude for reasons mentioned above. Interestingly, the area of high ΔpHCaCO3 found in the northern Indian Ocean does not have a corresponding shallow calcite saturation horizon, as found in the North Pacific, but the aragonite saturation horizon does shoal toward the surface.
Distributions of pHPI,0P = 0, the preindustrial pH at time of water mass formation, which are subsequently only changed by mixing in the interior ocean, are displayed in Figures 3d and 4g–4i. pHPI,0P = 0 is highly homogeneous in the ocean interior with an average of 8.20 ± 0.14 at both 1,000 and 3,000 m (Figures 4h and 4i). Given the sensitivity of pH to temperature, and the relatively large temperature gradients near the surface (i.e., at 100 m), the surface gradients in pHPI,0P = 0 are large (Figure 4g). However, below 1,000 m pHPI,0P = 0 is 8.20 ± 0.13 throughout the transects chosen here (Figure 3d). This indicates that the major natural processes, which add or remove DIC and TA in the ocean interior and thus change pH, have been accounted for and quantified fairly accurately. As an additional control for the validity of our decomposition we use the estimated TA0 (equation 1) and DICPI,0 (equation 11) to calculate preformed preindustrial partial pressure of CO2 (pCO2PI,0; Figure S14). Given that DICPI,0 accumulates all sources of error, it lends confidence in our preformed properties, and thus our decomposition, that the difference between our pCO2PI,0 estimate and the approximate preindustrial atmospheric CO2 (xCO2PI,atm ~280 ppm) is consistent with theoretical expectations in having higher undersaturation signals in the high latitudes compared to the lower latitudes. Note that nowhere is our pCO2PI,0 significantly different from xCO2PI,atm.
3.2 Anthropogenic Changes
The anthropogenic change in ocean pH is due to the ocean's increasing absorption of CO2 from the atmosphere. This uptake is driven by the increasing partial pressure of CO2 in the atmosphere and has led to more carbon being stored in the ocean. The estimates of the impact of anthropogenic CO2 changes on pH (Figure 5b) reaffirm the large and well-understood anthropogenic impact on pH in the surface ocean (Jiang et al., 2019), and put it in the context of natural pH gradients in the interior ocean. In the top 200 m ΔpHanth is >80% of ΔpHorg, and in the convergence zones ΔpHanth is at least 40% of ΔpHorg down to 800 m. Along the section shown here, the 0.05 ΔpHanth contour only extend deeper than 1,000 m in the North Atlantic, Arctic, and parts of the Southern Ocean (Figures 2a and 5b)—regions where the water column inventory of Canth (Figure 5a) is large. Thus, 1,000 m is generally the depth below which natural gradients dominate and anthropogenic changes are small.
This paper focuses mainly on interior ocean carbonate chemistry gradients, but it is important to consider surface ocean changes for context and because anthropogenic changes in pH originate at the surface. Surface ocean pH is clearly lower in year 2002 than it was in preindustrial times (Figures 6a and 6b). The decrease in surface waters is a globally quite uniform 0.10 ± 0.10 (Figure 6c and see Text S3 in the supporting information for notes on the uncertainty), although the uncertainty is spatially more variable than the change itself (Figure S7c). Since the pH scale is logarithmic this change represents a 25% increase in surface ocean acidity (total hydrogen ion concentration) relative to preindustrial times (approximately year 1800). The Intergovernmental Panel on Climate Change estimates 26% increased total hydrogen ion concentration since 1850 (IPCC, 2013), which, given the uncertainties, is indistinguishable. Corresponding to the change in DIC and pH the surface ocean saturation states of both aragonite (ΩAr; Figures 2b and 6d–6f) and calcite (ΩCa; Figures 2c and 6g–6i) have also decreased since preindustrial times, by on average −0.53 ± 0.71 and −0.81 ± 1.10, respectively (where the uncertainties are the average of the 2σ shown in Figures S8f and S8j).
4.1 Anthropogenic Change and Its Drivers
Anthropogenic ocean pH changes are a direct function of the accumulation of Canth in the ocean so any quantification of ΔpHanth depends to some degree on how Canth is estimated. In this study we use the TTD method (see section 2.3) and the estimated water column inventory of Canth (Figure 5a) of 167 ± 29 PgC is (just) within the uncertainties of the estimate by Khatiwala et al. (2013) for 2010 (155 ± 31 PgC) and the more recent of estimate of 160 ± 20 PgC for 2010 by Gruber et al. (2019) based on the eMLR(C*) approach (Clement & Gruber, 2018). Our estimate is for an earlier year (2002) and yet is the highest of the three, but regardless, the comparability with previous estimates, along with the spatial pattern being consistent with previous work (e.g.,Khatiwala et al., 2013 ; Sabine et al., 2004), lends confidence to the estimates of the impact of anthropogenic CO2 on pH presented here.
The difference between ΔpHanth and is presented in Figure 9. Clearly, the inclusion of DIC from the remineralization of organic matter (DICorg) in the estimate gives a greater pH change for any given increase in Canth. This result reflects the changing ∂pH/∂DIC as water circulates in the deep ocean and accumulate DICorg: Canth is added to the ocean at the surface, water then sinks out of the surface, DICorg is added as the waters circulate in the deep ocean, ∂pH/∂DIC increases because of the increased DIC, and the increased sensitivity leads to a strengthening of the the initial pH decline brought about by the uptake of Canth at the surface. Note that comparing equations 6 and 14 does not directly tell us how much more sensitive pH is to addition of DICorg, which would have to be quantified by modifying equation 4, only how much the initial acidification is increased when remineralization subsequently happens. Increased TA from carbonate mineral dissolution, the other natural process discussed in this paper, would be expected to decrease ∂pH/∂DIC and thereby reduce the pH change due to addition of Canth, but these chemical changes are more than compensated by the release of CO2 through organic matter remineralization, which is clearly seen when comparing ∂pH/∂DIC at 3,000 m in an ocean without biology (labelled 3,000 m; TA0, DIC0 in Figure 7) and ∂pH/∂DIC at 3,000 m in an ocean with biology (labelled 3,000 m in Figure 7). Hereafter we refer to the strengthened ΔpHanth due to the addition of DICorg as an enhancement.
We find that in the depth layer 500–1,500 m the net enhancement is on average 28 ± 15% along our section, but there are large regional and vertical gradients (Figure 9). In the deep ocean, except in the Southern Ocean, the differences between and ΔpHanth are generally very small (<0.005; Figure 9a) and have therefore been removed from further analysis (Figure 9b). In the Southern Ocean there is a significant amount of Canth at the bottom, and here the ocean acidification enhancement decreases from 71 ± 4% in the 800–1,500-m range to 25 ± 4% deeper than 4,000 m. In the upper ocean (<1,500 m), however, the largest enhancements are found in low-oxygen regions like the equatorial Atlantic where the difference is 48 ± 13%. In the Indian Ocean the strongest enhancement (70 ± 10%) is found between 1,200 and 1,750 m just north of the Broken Plateau, where high AOU corresponds with the shoaling of the aragonite saturation horizon. In the Pacific Ocean the maximum along our section is 35 ± 14% at 300–1,750 m. While these relative changes appear large, the largest relative enhancements are found in areas of low oxygen that tend to have low concentrations of Canth. However, as Canth continues to be added to the ocean and propagate into poorly ventilated portions of the ocean this enhancement will grow more pronounced. In this regard the regional patterns are important to quantify as they render some ecosystems more vulnerable than others.
4.2 Ecosystem Vulnerabilities
Research is ongoing as to at what point the decrease in pH, ΩAr, and ΩCa will exceed ecosystems' natural thresholds for sustainability. In that respect, it is interesting that a recent study (Sulpis et al., 2018) found that there is already an observable anthropogenically induced CaCO3 dissolution at the ocean bottom in the North Atlantic, where deep-sea corals persist in comparatively high-pH deep waters with little naturally accumulated carbon from organic matter remineralization. By contrast, in the North Pacific Ocean pH is naturally low because of the strong effect of organic matter remineralization which is only partially compensated by CaCO3 dissolution. The northern North Pacific also has a very shallow aragonite saturation horizon (<200 m in some places). Thus, despite the anthropogenic impact overall being smaller in the North Pacific compared to the North Atlantic, it is possible that Pacific ecosystems are more vulnerable to ocean acidification because the changes compound on naturally low pH; that is, they are already close to CaCO3 undersaturation. However, assessing the impact of ocean acidification on marine organisms is difficult since the impacts are often species specific and regionally varying (e.g.,Fabry et al., 2008 ; Gattuso et al., 2015 ; Kroeker et al., 2013), but what is clear is that heavily calcified organisms are particularly sensitive, while organisms that are able to move themselves have higher tolerances (Kroeker et al., 2013). It has been shown that it is the CaCO3 saturation state rather than [CO32−] or pH which controls calcification (e.g.,Bednaršek et al., 2017; Fabry et al., 2008; Osborne et al., 2019; Waldbusser et al., 2015), and recent research from the North Pacific indicate that some calcifying species are quite vulnerable to anthropogenic change (Bednaršek et al., 2014, 2016, 2017, 2018; Feely et al., 2016; Osborne et al., 2019). Other studies (e.g., Fabry et al., 2008) have found that organisms adapted to strong natural gradients in carbonate chemistry have better mechanisms for buffering such variations, but it is still unclear whether this means they are also more resilient to anthropogenic changes that shift the natural gradient toward lower pH and shallower saturation horizons. Furthermore, naturally low-pH, high-DIC waters have a higher ∂pH/∂DIC and are thus more sensitive to further Canth accumulation (Figures 7-9). In the Southern Ocean, a region sensitive to anthropogenic changes due to the natural shallow minimum in (CO32−), a recent model study (Negrete-García et al., 2019) shows that there is a high likelihood of a very rapid emergence of shallow (<100 m) saturation horizon for aragonite, largely due to the accumulation of CO2 in the thermocline, which decreases the natural minimum in [CO32−] enough to cause undersaturation.
While this paper focuses on pH in the interior ocean the anthropogenic change is largest at the surface; thus, any ecosystem impact is likely to first be seen there. As an example application of the quantified anthropogenic changes in pH and Ω presented in this study we have estimated the potential loss of suitable habitat for shallow calcifying corals. In today's ocean coral reef ecosystems do not thrive where the water has ΩAr < 3 (Guinotte et al., 2003), and Figure 10 shows the extent of the ocean areas with surface ΩAr ≥ 3 in 2002 (Figure 10a) and in preindustrial times (Figure 10b). The potential habitat loss, defined as the preindustrial area minus the area in 2002, is 35,300,000 km2. The calculation uncertainty for ΩAr is substantial, largely due to the large uncertainty in the solubility product (Orr et al., 2018). Thus, the smallest possible extent in preindustrial times approximately equals the largest possible extent in 2002 and the uncertainty in the loss of area (ranging from 35,000,000 to 83,000,000 km2) is very large. Regardless of the change in habitat with ΩAr ≥ 3 an interesting feature in Figure 10 is that there are large regions today which appear to be very vulnerable to further anthropogenic changes. If the minimum 2002 area is used then the South China Sea and the East Indian Archipelago as well as most of the west coast of the Americas all have ΩAr < 3. If the maximum area is used, these regions mostly have ΩAr > 3, but the fact that just the uncertainty in the global mapped ΩAr yield such large differences indicate that these regions are just within the “healthy” zone and thus particularly vulnerable to additional anthropogenic changes. The above estimate of area loss, calculated using the best available observations of ocean carbonate chemistry, highlights the need for reducing the uncertainty in both observations and calculations. Until then, fully quantifying the environmental stress all ocean ecosystems are subject to, both today and in the future, remain difficult.
There are some notable caveats regarding possible sources of bias in our study. First, it is clear from much previous work that pHPI,0P = 0 is sensitive to how DICorg and TAorg are calculated. AOU, which is used in our study, is the most commonly used method because it does not require knowledge about preformed values. However, previous studies (e.g., Carter et al., 2014; Ito et al., 2004; Talley et al., 2003; Russell & Dickson, 2003) have raised questions about the appropriateness of AOU as a proxy for organic matter remineralization given the assumption that a water mass is fully saturated with atmospheric oxygen at the time of subduction. This has for a long time been known to be incorrect (e.g., Redfield et al., 1963) since the degree to which saturation is reached depends on many factors including, but not limited to, wind speed, cooling, ice cover, and biological processes on the surface (e.g., Duteil et al., 2012). In addition, the use of AOU as a proxy for organic matter remineralization does not account for the fraction that occurs through denitrification. Deutsch et al. (2001) estimated the impacts of water column denitrification in the North Pacific, finding a maximum nitrate loss in oxygen-deficient zones of ~10 μmol kg−1, which implies that our AOU-based DICorg is off by a comparable number. Denitrification can therefore be assumed to result in a ΔpHorg estimate error of ~0.03 within oxygen-deficient zones and considerably less elsewhere. Second, constant stoichiometric ratios have been used throughout this study, while there is ample evidence that these ratios vary both spatially and temporally (e.g., Arrigo et al., 1999; Frigstad et al., 2011; Geider & La Roche, 2002; Li & Peng, 2002). However, as discussed in Deutsch and Weber (2012), synthesis efforts show that there appears to be a global mean stoichiometry similar to the classical Redfield ratio. Using spatially varying stoichiometry would likely change the results presented here, so we conservatively assume a 30% uncertainty applied to our globally uniform rC:O value (Table 1) to account for the possible latitudinal range of variability in stoichiometric ratios.
5 Summary and Conclusions
We decompose pH into components describing the anthropogenic impact (ΔpHanth), organic matter remineralization (ΔpHorg), CaCO3 cycling (ΔpHCaCO3), thermodynamics (ΔpHT and ΔpHpress), and ocean circulation (pHPI,0P = 0). The ΔpHorg component is the most important (maximum reduction of 0.79 ± 0.26) in explaining the interior ocean pH gradients, while ΔpHCaCO3 has a significant effect only at quite large depths (centered around 2,500 m) in the North Pacific and the northern Indian Oceans. The Atlantic Ocean sector of the Southern Ocean, which has a naturally shallow aragonite saturation horizon, is one area where the gradient between the surface ocean pH (~8.2) and the naturally low subsurface pH (~7.9) is quickly vanishing due to invasion of anthropogenic carbon. Here pH is naturally low due to strong organic matter remineralization, and anthropogenic changes are penetrating very deep (ΔpHanth ~0.07 at 1,000 m), and this is an area recently identified as being particularly sensitive to small changes in ocean carbonate chemistry (Negrete-García et al., 2019). When removing the organic matter remineralization and CaCO3 cycling from the estimate ΔpHanth becomes smaller, indicating an enhancement of anthropogenic pH changes by natural processes, which is likely to continue and grow stronger as ocean acidification continues. With the exception of the Southern Ocean, we currently see this enhancement only in the upper ocean (<1,500 m) and most pronounced in low-oxygen areas. Anthropogenic emissions of CO2 have led to both decreased pH and decreases in the saturation states of CaCO3 minerals. However, given the uncertainties in calculating the ocean carbonate chemistry it is difficult to conclude definitively how large this impact is on ocean ecosystems. Thus, fully quantifying the environmental stress ocean ecosystems are subject to both today and in the future through the reduction of uncertainty in the observations and calculations remains a priority.
The authors want to thank Steven van Heuven (Royal Netherlands Institute for Sea Research), Robert Key (Princeton University), and Jake Gebbie (Woods Hole Oceanographic Institution) for the valuable and useful discussions, as well as Nicolas Gruber and one anonymous reviewer for the thoughtful and constructive reviews which helped improve the paper. B.R.C., L.J., and R.A.F. acknowledge funding support from the National Oceanic and Atmospheric Administration (NOAA)'s Ocean Observations and Monitoring Division (OOMD) and Ocean Acidification Programs (OAP). B.R.C. is funded by the OOMD Carbon Data Management and Synthesis Grant (N8R3CEA-PDM), fund reference 100007298. F.F.P. was supported by Ministerio de Ciencia, Investigaciones y Universidades through the ARIOS (CTM2016-76146-C3-1-R) project co-funded by the Fondo Europeo de 415 Desarrollo Regional 2014-2020 (FEDER). S.K.L., A.O., and F.F.P. were supported through EU Horizon 2020 through the AtlantOS project (grant agreement 633211). L.J. was supported by NOAA's Ocean Acidification Program under the Ocean Acidification Data Stewardship Project. This is contribution 4936 from the Pacific Marine Environmental Laboratory of NOAA and 2019-1042 from JISAO. All data used in this study are available for download through https://www.glodap.info/index.php/mapped-data-product/ (last accessed 16 December 2019).
|gbc209419-sup-0001-Text_SI.pdfPDF document, 3.6 MB||
Supporting Information S1
Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
- 1994). Redfield ratios of remineralization determined by nutrient data-analysis. Global Biogeochemical Cycles, 8(1), 65–80.
- 1999). Phytoplankton community structure and the drawdown of nutrients and CO2 in the Southern Ocean. Science, 283(5400), 365–367. https://doi.org/10.1126/science.283.5400.365
- 2014). A time-series view of changing ocean chemistry due to ocean uptake of anthropogenic CO2 and ocean acidification. Oceanography, 27(1), 126–141.
- 2018). El Niño-related thermal stress coupled with upwelling-related ocean acidification negatively impacts cellular to population-level responses in pteropods along the California Current System with implications for increased bioenergetic costs. Frontiers in Marine Science. https://doi.org/10.3389/fmars.2018.00486
- 2014). Limacina helicina shell dissolution as an indicator of declining habitat suitability due to ocean acidification in the California Current Ecosystem. Proceedings of the Royal Society B: Biological Sciences, 281(1785), 20140123. https://doi.org/10.1098/rspb.2014.0123
- 2016). Pteropods on the edge: Cumulative effects of ocean acidification, warming, and deoxygenation. Progress in Oceanography, 145, 1–24. https://doi.org/10.1016/j.pocean.2016.04.002
- 2017). New ocean, new needs: Application of pteropod shell dissolution as a biological indicator for marine resource management. Ecological Indicators, 76, 240–244. https://doi.org/10.1016/j.ecolind.2017.01.025
- 2003). Anthropogenic carbon and ocean pH. Nature, 425(6956), 365–365. https://doi.org/10.1038/425365a
- 2005). A factorial analysis of the marine carbon cycle and ocean circulation controls on atmospheric CO2. Global Biogeochemical Cycles, 19, GB4027. https://doi.org/10.1029/2005GB002489
- 2018). Updated methods for global locally interpolated estimation of alkalinity, pH, and nitrate. Limnology and Oceanography: Methods, 16(2), 119–131.
- 2014). Mixing and remineralization in waters detrained from the surface into Subantarctic Mode Water and Antarctic Intermediate Water in the southeastern Pacific. Journal of Geophysical Research, Oceans, 119(6), 4001–4028. https://doi.org/10.1002/2013JC009355
- 2018). The eMLR(C*) Method to determine decadal changes in the global ocean storage of anthropogenic CO2. Global Biogeochemical Cycles, 32(4), 654–679. https://doi.org/10.1002/2017GB005819
- 2019). Drivers of 21st century carbon cycle variability in the North Atlantic Ocean. Biogeosciences Discussions, 2019, 1–33.
- 2001). Denitrification and N2 fixation in the Pacific Ocean. Global Biogeochemical Cycles, 15, 483–506.
- 2012). Nutrient ratios as a tracer and driver of ocean biogeochemistry. Annual Review of Marine Science, 4, 113–141. https://doi.org/10.1146/annurev-marine-120709-142821
- 1990). Standard potential of the reaction: AGCL(S) + 1/2H − 2(G) = AG(S) + HCL (AQ) and the standard acidity constant of the ion HSO4− in synthetic sea-water from 273.15-K to 318.15-K. Journal of Chemical Thermodynamics, 22(2), 113–127.
- 2012). Preformed and regenerated phosphate in ocean general circulation models: Can right total concentrations be wrong? Biogeosciences, 9(5), 1797–1807. https://doi.org/10.5194/bg-9-1797-2012
- 2018). The devil's in the disequilibrium: Multi-component analysis of dissolved carbon and oxygen changes under a broad range of forcings in a general circulation model. Biogeosciences, 15(12), 3761–3777.
- 2008). Impacts of ocean acidification on marine fauna and ecosystem processes. ICES Journal of Marine Science, 65(3), 414–432.
- 2018). Seasonal asymmetry in the evolution of surface ocean pCO2 and pH thermodynamic drivers and the influence on sea-air CO2 flux. Global Biogeochemical Cycles, 32(10), 1476–1497. https://doi.org/10.1029/2017GB005855
- 2016). Chemical and biological impacts of ocean acidification along the west coast of North America. Estuarine, Coastal and Shelf Science, 183(A), 260–270. https://doi.org/10.1016/j.ecss.2016.08.043
- 2009). Ocean acidification: Present conditions and future changes in a high-CO2 world. Oceanography, 22(4), 36–47.
- 2018). The combined effects of acidification and hypoxia on pH and aragonite saturation in the coastal waters of the Californian Current Ecosystem and the northern Gulf of Mexico. Continental Shelf Research, 152, 50–60. https://doi.org/10.1016/j.csr.2017.11.002
- 2012). Decadal changes in the aragonite and calcite saturation state of the Pacific Ocean. Global Biogeochemical Cycles, 26, GB3001. https://doi.org/10.1029/2011GB004157
- 2004). Impact of anthropogenic CO2 on the CaCO3 system in the oceans. Science, 305(5682), 362–366. https://doi.org/10.1126/science.1097329
- 2002). In situ calcium carbonate dissolution in the Pacific Ocean. Global Biogeochemical Cycles, 16(4), 1144. https://doi.org/10.1029/2002GB001866
- 2018). Contrasting impact of future CO2 emission scenarios on the extent of CaCO3 mineral undersaturation in the Humboldt Current System. Journal of Geophysical Research: Oceans, 123(3), 2018–2036. https://doi.org/10.1002/2018JC013857
- 2011). Seasonal variation in marine C:N:P stoichiometry: Can the composition of seston explain stable Redfield ratios? Biogeosciences, 8(10), 2917–2933.
- 1992). Oxygen solubility in seawater: Better fitting equations. Limnology and Oceanography, 37(6), 1307–1312.
- 2015). Contrasting futures for ocean and society from different anthropogenic CO2 emission scenarios. Science, 349(6243), aac4722. https://doi.org/10.1126/science.aac4722
- 2011). How is the ocean filled? Geophysical Research Letters, 38(6), L06604. https://doi.org/10.1029/2011GL046769
- 2002). Redfield revisited: Variability of C:N:P in marine microalgae and its biochemical basis. European Journal of Phycology, 37(1), 1–17.
- 2019). The oceanic sink for anthropogenic CO2 from 1994 to 2007. Science, 363(6432), 1193–1199. https://doi.org/10.1126/science.aau5153
- 1996). An improved method for detecting anthropogenic CO2 in the oceans. Global Biogeochemical Cycles, 10(4), 809–837.
- 2003). Future coral reef habitat marginality: Temporal and spatial effects of climate change in the Pacific basin. Coral Reefs, 22(4), 551–558.
- 2002). Inferring the concentration of anthropogenic carbon in the ocean from tracers. Global Biogeochemical Cycles, 16(4), 1131. https://doi.org/10.1029/2001GB001835
- 2013). Spatiotemporal variability and long-term trends of ocean acidification in the California Current System. Biogeosciences, 10(1), 193–216.
- 2018). A model-based evaluation of the inverse Gaussian transit-time distribution method for inferring anthropogenic carbon storage in the ocean. Journal of Geophysical Research, Oceans, 123, 1777–1800. https://doi.org/10.1002/2017JC013504
- 2015). Experimental design in ocean acidification research: Problems and solutions. ICES Journal of Marine Science, 73(3), 572–581.
IPCC (2013). Summary for policymakers. In T. F. Stocker, D. Qin, G.-K. Plattner, M. Tignor, S. K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex, & P. M. Midgley (Eds.), Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge, United Kingdom and New York, NY, USA: Cambridge University Press.
- 2004). Is AOU a good measure of respiration in the oceans? Geophysical Research Letters, 31, L17305. https://doi.org/10.1029/2004GL020900
- 2019). Surface ocean pH and buffer capacity: Past, present and future. Scientific Reports, 9(1), 18624. https://doi.org/10.1038/s41598-019-55039-4
- 2015). Climatological distribution of aragonite saturation state in the global oceans. Global Biogeochemical Cycles, 29(10), 1656–1673. https://doi.org/10.1002/2015GB005198
- 1982). Calcium-alkalinity relationship in the North Pacific. Journal of the Oceanographical Society of Japan, 38(2), 57–62.
- 2010). Ocean deoxygenation in a warming world. Annual Review of Marine Science, 2(1), 199–229.
- 2013). Global ocean storage of anthropogenic carbon. Biogeosciences, 10(4), 2169–2191. https://doi.org/10.5194/bg-10-2169-2013
- 2013). Impacts of ocean acidification on marine organisms: Quantifying sensitivities and interaction with warming. Global Change Biology, 19(6), 1884–1896. https://doi.org/10.1111/gcb.12179
- 2001). Redfield ratios revisited: Removing the biasing effect of anthropogenic CO2. Limnology and Oceanography, 46(4), 964–970.
- 2018). Strengthening seasonal marine CO2 variations due to increasing atmospheric CO2. Nature Climate Change, 8(2), 146–150.
- 2015). Trends and drivers in global surface ocean pH over the past 3 decades. Biogeosciences, 12(5), 1285–1298.
- 2016). A new global interior ocean mapped climatology: The 1° × 1° GLODAP version 2. Earth System Science Data, 8(2), 325–340. https://doi.org/10.5194/essd-8-325-2016
- 2012). The observed evolution of oceanic pCO2 and its drivers over the last two decades. Global Biogeochemical Cycles, 26, GB2021. https://doi.org/10.1029/2011GB004095
- 1998). Program developed for CO2 system calculations, in ORNL/CDIAC-105, edited, Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of Energy, Oak Ridge, Tennessee.
- 2002). Latitudinal change of remineralization ratios in the oceans and its implication for nutrient cycles. Global Biogeochemical Cycles, 16(4), 1130. https://doi.org/10.1029/2001GB001828
- 2000). Ocean pCO(2) calculated from dissolved inorganic carbon, alkalinity, and equations for K-1 and K-2: Validation based on laboratory measurements of CO2 in gas and seawater at equilibrium. Marine Chemistry, 70(1-3), 105–119.
- 1987). VERTEX: Carbon cycling in the northeast Pacific. Deep Sea Research Part A. Oceanographic Research Papers, 34(2), 267–285. https://doi.org/10.1016/0198-0149(87)90086-0
- 2013). Chemical Oceanography ( Fourth ed.). Boca Raton, FL: CRC Press Taylor & Francis.
- 2019). Sudden emergence of a shallow aragonite saturation horizon in the Southern Ocean. Nature Climate Change, 9(4), 313–317. https://doi.org/10.1038/s41558-019-0418-8
- 2016). The Global Ocean Data Analysis Project version 2 (GLODAPv2): An internally consistent data product for the world ocean. Earth System Science Data, 8(2), 297–323. https://doi.org/10.5194/essd-8-297-2016
- 2019). GLODAPv2.2019: An update of GLODAPv2. Earth System Science Data, 11(3), 1437–1461. https://doi.org/10.5194/essd-11-1437-2019
- 2018). Routine uncertainty propagation for the marine carbon dioxide system. Marine Chemistry, 207, 84–107. https://doi.org/10.1016/j.marchem.2018.10.006
- 2019). Decadal variability in twentieth-century ocean acidification in the California Current Ecosystem. Nature Geoscience, 13(1), 43–49. https://doi.org/10.1038/s41561-019-0499-z
- 2011). CO2 air–sea disequilibrium and preformed alkalinity in the Pacific and Indian Oceans calculated from subsurface layer data. Journal of Marine Systems, 84(3), 67–77.
- 1969). Oceanic CO2 system: An evaluation of ten methods of investigation. Limnology and Oceanography, 14(2), 179–186.
- 1963). The influence of organisms on the composition of sea water. In M. N. Hill (Ed.), The Sea (pp. 26–77). Hoboken, NJ: Wiley-Interscience.
- 2003). Variability in oxygen and nutrients in South Pacific Antarctic Intermediate Water. Global Biogeochemical Cycles, 17(2), 1033. https://doi.org/10.1029/2000GB001317
- 2004). The oceanic sink for anthropogenic CO2. Science, 305(5682), 367–371. https://doi.org/10.1126/science.1097403
- 2006). Ocean biogeochemical dynamics, XII (p. 503). Princeton, NJ: Princeton University Press.
- 2013). Biology and air-sea gas exchange controls on the distribution of carbon isotope ratios (delta C-13) in the ocean. Biogeosciences, 10(9), 5793–5816.
- 2018). Current CaCO3 dissolution at the seafloor caused by anthropogenic CO2. Proceedings of the National Academy of Sciences of the United States of America, 115(46), 11700–11705. https://doi.org/10.1073/pnas.1804250115
- 1985). Redfield Ratio Based on Chemical Data from Isopycnal Surfaces, Journal of Geophysical Research, 90(C4), 6907–6924.
- 2013). Closure of the global overturning circulation through the Indian, Pacific, and Southern Oceans: Schematics and transports. Oceanography, 26(1), 80–97. https://doi.org/10.5670/oceanog.2013.07
- 2003). Deep convection and brine rejection in the Japan Sea. Geophysical Research Letters, 30(4), 1159. https://doi.org/10.1029/2002GL016451
- 1974). The boron/chlorinity ratio of deep-sea water from the Pacific Ocean. Deep Sea Research and Oceanographic Abstracts, 21(2), 161–162.
- 2009). MATLAB Program developed for CO2 system calculations, ORNL/CDIAC-105b, Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, US Department of En ergy, Oak Ridge, Tennessee.
- 2009). Anthropogenic carbon distributions in the Atlantic Ocean: Data-based estimates from the Arctic to the Antarctic. Biogeosciences, 6(3), 439–451.
- 2015). Calcium carbonate saturation state: On myths and this or that stories. ICES Journal of Marine Science, 73(3), 563–568.
- 2006). Anthropogenic CO2 in the oceans estimated using transit time distributions. Tellus Series B: Chemical and Physical Meteorology, 58(5), 376–389.
- 2007). Total alkalinity: The explicit conservative expression and its application to biogeochemical processes. Marine Chemistry, 106(1-2), 287–300.
- 2001). CO2 in seawater, equilibrium, kinetics, isotopes, XIII (p. 346). Amsterdam, PAYS-BAS: Elsevier.
References From the Supporting Information
- 2004). A global ocean carbon climatology: Results from Global Data Analysis Project (GLODAP). Global Biogeochemical Cycles, 18, GB4031. https://doi.org/10.1029/2004GB002247