Feedbacks of CaCO3 dissolution effect on ocean carbon sink and seawater acidification: a model study

The oceanic absorption of atmospheric CO2 acidifies seawater, which accelerates CaCO3 dissolution of calcifying organisms, a process termed dissolution effect. Promoted CaCO3 dissolution increases seawater ALK (alkalinity), enhancing ocean carbon sink and mitigating ocean acidification. We incorporate different parameterizations of the link between CaCO3 dissolution and ocean acidification into an Earth System Model, to quantify the feedback of the dissolution effect on the global carbon cycle. Under SRES A2 CO2 emission scenario and its extension with emissions of 5,000 PgC in ∼400 years, in the absence of the dissolution effect, accumulated ocean CO2 uptake between year 1800 and 3500 is 2,041 PgC. The consideration of the dissolution effect increases ocean carbon sink by 195–858 PgC (10%–42%), and mitigates the decrease in surface pH by 0.04–0.17 (a decrease of 10%–48% in [H+] (hydrogen ion concentration)), depending on the prescribed parameterization scheme. In the epipelagic zone, relative to the Arc-Atlantic Ocean, the Pacific-Indian Ocean experiences greater acidification, leading to greater dissolution effects and the resultant stronger feedbacks on ocean carbon sink and acidification in the Pacific-Indian Ocean. Noteworthy, the feedback of dissolution effect on ocean carbon sink can be comparable with or stronger than the feedback from CO2-induced radiative warming. Our study highlights the potentially critical role played by CaCO3 dissolution effect in the ocean carbon sink, global carbon cycle and climate system.


Introduction
Since the preindustrial time, CO 2 emissions to the atmosphere have been continuously increasing, mainly resulting from human activities of fossil fuel burning, industrial processes and land use changes. Data-based estimates revealed a total anthropogenic CO 2 emission of 690 ± 80 PgC (1 PgC = 1 petagrams of carbon = 10 15 gC) between year 1750 and 2020 [1]. About 42% of these emissions stayed in the atmosphere, increasing atmospheric CO 2 concentration, inducing global warming through trapping longwave radiation, which is termed greenhouse effect [2]. Meanwhile, 26% of the emissions were absorbed by the ocean, which is considered as a main sink of atmospheric CO 2 [1]. Ocean carbon sink is considered as one of the most important pathways to regulate atmospheric CO 2 concentration increase and global warming. Nevertheless, ocean carbon sink is also perturbing marine biogeochemistry by acidifying seawater, a process defined as ocean acidification [3][4][5].
A series of marine dynamical and biogeochemical processes could trigger strong feedbacks on ocean carbon sink, atmospheric CO 2 and climate system. For instance, CO 2 -induced warming would weaken ocean mixing and circulation, decelerating the transportation of anthropogenic carbon to the deep ocean, inhibiting ocean carbon sink, and triggering a positive feedback to the rising atmospheric CO 2 contents [6,7]. Also, CO 2 -induced warming would decrease CO 2 solubility, reducing chemical buffering capacity of seawater, weakening ocean's ability as a carbon sink [8,9]. In addition, growing seawater CO 2 partial pressure could increase the ocean C:N:P ratio, accelerating extracellular organic carbon production, promoting marine biological carbon pump and ocean carbon sink [10,11].
Ocean acidification would also influence the global carbon cycle and climate system. Over the last 40 years, global mean sea surface pH has diminished by 0.068-0.108 [12], corresponding to an increase of 17%-28% in [ where [ ] + Ca 2 denotes the calcium ion concentration, and * K sp represents the CaCO 3 stoichiometric solubility product constant [2].
Marine calcifying organisms that utilize CaCO 3 to form skeletons or shells may be incapable of acclimatizing themselves to the decreases in Ω. An experimental study conducted by Keir [13] showed that, CaCO 3 dissolution rate would increase due to the drop of Ω, and the nonlinear dependence of CaCO 3 dissolution on Ω differs between aragonite and calcite. Employing Atomic Force Microscopy, Dong et al [14] also demonstrated higher calcite dissolution rates in seawater of low Ω. In-situ observational data in the North Pacific Ocean provided by Naviaux et al [15] confirmed the correlation between CaCO 3 dissolution and Ω, proposing a range of 0.11 ± 0.1 to 4.7 ± 0.7 for the reaction order (n) of the empirical dissolution rate = k(1−Ω) n equation. Some other observational and experimental studies also demonstrated the correlation between CaCO 3 dissolution rate and CaCO 3 saturation state [16][17][18][19]. Nevertheless, owing to different species of calcifying organisms concerned or different manipulation methods used in different studies, the sensitivity of the response of CaCO 3 dissolution to acidification varies dramatically.
The process of dissolution increases [ ] -CO 3 2 and [ ] -HCO 3 (bicarbonate ion concentration), increasing ocean ALK, amplifying the ocean's absorption of atmospheric CO 2 . Therefore, seawater acidification-accelerated CaCO 3 dissolution would promote the ocean carbon sink, triggering a negative feedback on the increased atmospheric CO 2 [2]. We term the dependence of CaCO 3 dissolution on ocean acidification as 'dissolution effect' here. Recent years, several modelling studies was conducted to quantify the feedbacks of dissolution effect on the carbon cycle and climate system. However, the parameterization schemes of dissolution effect stemmed from different observational or experimental studies, leading to varied estimates of the feedback of dissolution effect on the ocean carbon cycle [20][21][22].
Previous studies reveal that the timescales of atmospheric CO 2 adjustment processes in the ocean carbon cycle range from decades to millennia [23,24]. For instance, invasion of CO 2 to the ocean are on timescales of decades to millennia [25][26][27]. The ocean CaCO 3 cycle triggers feedbacks to the carbon cycle and climate system on timescales of centuries to millennia [22,[28][29][30]. Investigating the mechanisms of the processes and feedbacks in climate and carbon cycle on the appropriate timescales are vital for dependable projections of future changes in ocean carbon sink, atmospheric CO 2, and climate system [23,31]. Therefore, to develop a better understanding of the role played by the dissolution effect in the global carbon cycle and climate system, model simulations should be conducted on millennial timescales.
Building on previous studies, we further assess the feedbacks of dissolution effect on the carbon cycle and climate system on millennial timescales. We integrate the CaCO 3 dissolution-Ω correlation into an intermediate complexity Earth System Model to assess the strength of the dissolution effect in regulating atmospheric CO 2 increases. Usually, previous modelling studies concerning the dissolution effect assume a single CaCO 3 dissolution-Ω parameterization scheme. Due to different model structures in different studies, we could not directly compare their simulated results. In this study, different parameterization sensitivities with different reaction orders of CaCO 3 dissolution-Ω dependence are employed in a unified model structure, which allow us to investigate the importance of parameter uncertainty on the dissolution effect. Additionally, we compare the magnitude of feedbacks of dissolution effect with feedbacks from CO 2 -induced climate change. Furthermore, we quantify the feedback of different parameterizations of dissolution effect on ocean acidification in different ocean basins, which was barely analysed in previous studies. In this study, we intend to obtain a better understanding of the role of dissolution effect in the ocean carbon sink, global carbon cycle and climate system.  [32].
The model comprises the MOM 2 (Modular Ocean Model 2), an ocean general circulation model with a spatial resolution of 3.6°in longitude, 1.8°in latitude, and 19 levels in the vertical direction. MOM 2 is coupled with a dynamic-thermodynamic sea ice component [33]. UVic also includes an energy-moisture balanced atmosphere model [33]. The land carbon cycle module of the UVic is based on the TRIFFID (Top-down Representation of Interactive Foliage and Flora Including Dynamics), a dynamic vegetation model [34,35], and the MOSES (Met Office surface exchange scheme), a land surface model [35]. The ocean carbon cycle module of UVic includes an OCMIP-based (Ocean Carbon-cycle Model Intercomparison Project) inorganic carbon cycle component [36], and an ocean ecosystem/biogeochemical NPZD (nutrient, phytoplankton, zooplankton, detritus) component [32].
UVic model has been included in numerous international model intercomparison projects, such as the C 4 MIP (Coupled Carbon Cycle Climate Model Intercomparison Project) [37], coordinated THC (thermohaline circulation) experiments [38,39], and the Paleoclimate Modelling Intercomparison Project [40]. UVic also participated in the EMICs (Earth System Models of Intermediate Complexity) model intercomparison project in the framework of the IPCC (Intergovernmental Panel on Climate Change) AR5 (Fifth Assessment Report) [41], and simulated results of the UVic are also analysed in IPCC AR6 (the Sixth Assessment Report) [42].
The key biological components in the UVic are as follows: Here, ( ) S PO , is the quadratic mortality of nondiazotrophic phytoplankton. μ P P D is the mortality of diazotrophs, γ 1 [G(P O )+G(P O )]Z represents the portion of phytoplankton (including diazotrophs and nondiazotrophs) assimilated by zooplankton, m Z Z 2 is the mortality of zooplankton, and ¶ ¶ w D z D denotes the export minus import of detritus at the ocean depth of z. The model calculates CaCO 3 dissolution as follows: denotes CaCO 3 dissolution at the depth of z, z b is the depth of seafloor, Pr CaCO 3 denotes CaCO 3 production, D CaCO 3 represents the e-folding depth of the vertically integrated production of CaCO 3 (D CaCO 3 = 6,500m), ( represents the portion of nondiazotrophic phytoplankton grazed but not assimilated by zooplankton, denotes the production ratio of CaCO 3 /POC, while / R C N is the Redfield ratio of C/N (carbon/nitrogen) [32].

Parameterization of CaCO 3 dissolution in water column
In the original UVic model, there is no direct relation between ( ) Di z CaCO 3 and ocean acidification. In this study, on the basis of previous studies about the dissolution effect, a parameterization scheme that links CaCO 3 dissolution to calcite saturation state (Ω C ) is integrated to the model: Here, denotes the export flux of CaCO 3 , the exponent n denotes the reaction order, ( ) CaCO z 3 is the concentration of CaCO 3 in water column, v CaCO 3 represents CaCO 3 settling velocity, and k CaCO 3 denotes dissolution rate constant. We term 3 as a whole. In equation (10), Di CaCO 3 is Ω C -related only when Ω C < 1; otherwise, when Ω C 1, Di CaCO 3 is set to be identical to the original model.
For the apparent order 'n' in the CaCO 3 dissolution-Ω C correlation (equation (10)), previous observational and experimental studies provided values of 1.0-4.7 [2,15,16,18,43]. In this study, the dissolution reaction order n are set at 1.0 (set L, denoting the set with a low reaction order n) and 4.7 (set H, denoting the set with a high reaction order n), covering the range of n values provided by observational and experimental studies (table 1). For each n, three different values of K are selected, approximately covering the range of dissolution rate constant provided by previous observational and experimental studies, to represent varying sensitivities of CaCO 3 dissolution rate to ocean acidification (table 1) [15,[43][44][45]. For the six model versions, the dependence of ( ) Moreover, a reference model version which uses the original parameterization scheme in the model is also employed (REF in table 1). Therefore, by comparing simulated results of set L and set H with REF, we can quantify the feedbacks of dissolution effect on the oceanic CO 2 absorption and ocean carbon cycle.

Simulation experiments
First, by using the constant preindustrial atmospheric CO 2 content (280 ppm), we integrate the seven versions of model listed in table 1 for 10,000 model years, which is called the spinup simulations. In the last 100 years of the seven spinup simulations, mean air-sea CO 2 fluxes are −0.0008-0.0058 PgC yr −1 , confirming the quasiequilibrium states for all versions of model, which could be taken as the initial fields for the preindustrial year 1800. Subsequently, two series of transient simulations are conducted for 1,700 model years (during years 1800-3500). In the former series of experiments, the increasing atmospheric CO 2 influences both radiative forcing and the ocean carbon cycle. In the latter series of simulations, the increasing atmospheric CO 2 would not affect atmospheric radiative forcing, that is, the ocean carbon cycle would not be influenced by CO 2 -induced radiative warming. Both series of experiments consist of seven simulations, corresponding to the seven versions of model depicted in section 'Parameterization of CaCO 3 dissolution in water column' (table 1). During 1800-2100, all these 14 simulations are under the scenario of IPCC SRES (Special Report on Emissions Scenarios) A2, a 'business-as-usual' CO 2 emission pathway. After 2100, we presume a linear decrease in CO 2 emissions, reaching zero at year 2319, and the accumulated CO 2 emission after year 2000 is 5,000 PgC ( figure  S2). In the spinup stage, terrestrial weathering input of ALK is prescribed to be equal to ALK output through CaCO 3 burial in marine sediments. In the following transient simulations, we keep the weathering input fixed at its preindustrial quasi-equilibrium value, and enable CaCO 3 burial flux to change freely [26]. In consequence, in transient simulations, the variations in global ocean ALK would be dominated by the balance between burial flux of CaCO 3 and weathering flux of ALK.

Calculation of ocean key carbonate chemistry fields
We calculate key ocean carbonate fields, including ocean pH, [ ] -CO 3 2 and Ω, following the protocol of the OCMIP (http://ocmip5.ipsl.jussieu.fr/OCMIP/). UVic-simulated seawater temperature, salinity, dissolved inorganic carbon (DIC) and ALK concentrations, as well as observation-based fields of seawater silicate and phosphate concentrations in the GLODAP (Global Ocean Data Analysis Project) [46] are used. Based on the above four equations (equations (13)- (16)) and six variables (DIC and ALK concentrations, HCO , 3 and [H + ]) in marine carbonate chemistry system, specifying two variables, we could compute the remaining four fields [47,48]. Thus, changes in temperature and salinity (related to * K 1 and * K 2 ), or changes in DIC and ALK concentrations, could lead to the changes in pH (pH = −log 10 [H + ]) and [ ] -CO , 3 2 that is, changes in ocean acidity.

Results
To estimate the feedbacks of dissolution effect, we incorporate different parameterization schemes linking dissolution rate with saturation state of CaCO 3 into the original model (refer to Methods section). After modifying the original UVic model, we conduct two series of transient model simulations, representing simulations with or without effects of CO 2 -induced warming on the ocean carbon cycle, respectively. Both series of experiments consist of seven simulations, which corresponds to the seven model configurations introduced in the Methods section. In the following, we only show results of simulations with the impacts of CO 2 -induced radiative warming, unless explicitly stated otherwise.

Model-observation comparison in different ocean basins
Model-simulated oceanic CO 2 absorption in each simulation is compared with model-based and observationbased estimates presented in the IPCC AR5 and AR6 (  [49]. We also compare the model-simulated results with observation-based estimates from the GLODAP [46]. Figure S3 presents the model-simulated and data-based estimates of vertical profiles of DIC and ALK concentrations in the Pacific-Indian Ocean, the Arc-Atlantic Ocean, and the global ocean. As shown in figure S3,   S1). Atmospheric CO 2 content increases to a peak of 2,124 ppm around year 2300 ( figure 1(c)). After CO 2 emission ceases, atmospheric CO 2 content drops to 1,771 ppm by year 3500, when sea surface temperature rises by 8.9 ℃ relative to year 1800 (figures 1(c), (d)). As an important carbon sink, the ocean continuously consumes atmospheric CO 2 , which also causes ocean acidification. For example, global mean sea surface pH decreases from 8.16 to 7.39 during years 1800-2300, and after the cessation of CO 2 emissions, recovering to 7.44 by year 3500 (figure 2(e)).

The dissolution effect in simulations of different parameterization schemes
Owing to ocean acidification, calcite saturation state (Ω C ) would decrease (equation (1), figures 3(b), (d)). In REF simulation, CaCO 3 dissolution rate is not related to Ω C , therefore, the warming of sea surface might boost CaCO 3 production and the resultant CaCO 3 export flux (figures 4(a), (d), (g)), by increasing phytoplankton growth and biomass, which CaCO 3 production is related to (equation (9)).  Ocean CaCO 3 cycle changes would lead to changes in seawater ALK. In REF simulation, the enhanced CaCO 3 production due to sea surface warming leads to a reduction in ocean ALK (figures 2(c), (d)). In the simulation considering dissolution effect, enhanced CaCO 3 dissolution due to decreased Ω C could increase the seawater ALK. For instance, in H1 simulation, surface/ocean mean ALK would rise by 35/68 μmol kg −1 during years 1800-3500. While for L3 simulation with stronger dissolution effect, in the same time period, surface/ ocean mean ALK rises by 181/142 μmol kg −1 (figures 2(c), (d), table S1).   dissolution-Ω C in different ocean depths, dissolution effect shows different magnitude among simulations. For instance, in simulations L3 and H3, the greatest drop of Ω C with Ω C < 1 locates in ocean depth of ∼500-2,000m, where Ω C ranges in ∼0.8-1.0 (figures 3, S1(b), (d)). Here we only consider the depth with Ω C <1, since in our simulations, dissolution effect only takes effect when Ω C <1 (equation (10) in Methods section).

Feedbacks of dissolution effect on the ocean carbon cycle
The dissolution effect increases seawater ALK, further amplifying ocean carbon sink, acting as a negative feedback on the rising atmospheric CO 2 concentration [2,20,22]. For instance, relative to REF simulation, the incorporation of dissolution effect enhances accumulated ocean CO 2 uptake by 195-858 PgC by year 3500, accrossing the six forms of CaCO 3 dissolution parameterization ( figure 1(b), table S1). Therefore, modelled atmospheric CO 2 concentration ranges in 1,333-1,670 ppm with the incorporation of dissolution effect, relative to 1,770 ppm in the REF simulation by year 3500 (figure 1(c)). As a results, by year 3500, the dissolution effect mitigates surface warming by 0.22-1.06 ℃ compared to the REF simulation, acrossing different forms of CaCO 3 dissolution parameterization applied ( figure 1(d)).
The dissolution effect also provides feedbacks to ocean acidification. On one hand, the dissolution effect increases ocean CO 2 uptake, increasing ocean DIC, tending to enhance ocean acidification (figures 1(a), (b),  2(a), (b)). On the other hand, the dissolution effect also increases ocean ALK, which tends to mitigate ocean acidification (figures 2(c), (d)). For the global ocean, impacts of dissolution effect-induced increase in ALK dominate impacts of dissolution effect-induced increase in DIC, leading to a combined impact of mitigating ocean acidification (figures 5(a)-(c)). Therefore in the global scale, the dissolution effect enhances ocean carbon sink, and simultaneously, mitigates ocean acidification. For instance, surface mean pH in H1 is 0. 09 3 2 in L3 (figures 2(e), (g), table S1). Relative to the Arc-Atlantic Ocean, due to the greater acidification in the epipelagic zone of the Pacific-Indian Ocean, the dissolution effect has greater effects on ocean ALK in the Pacific-Indian Ocean (figures 3(f), (j), 5(e), (f), (h), (i)). Therefore, in the Pacific-Indian Ocean, the dissolution effect triggers stronger feedbacks on ocean carbon sink and ocean acidification ( figure S4). Figure S4 presents the spatial distributions of feedbacks of dissolution effect on ocean CO 2 uptake and ocean acidification, which shows especially great magnitude in the Equatorial East-Pacific and the Northeast Indian Ocean. For instance, at year 2600, in L3 simulation relative to the REF simulation, the inclusion of dissolution effect increases the ocean CO 2 uptake by 2.42gC m −2 yr −1 in the Pacific-Indian Ocean, which amounts to 1.81gC m −2 yr −1 in the Arc-Atlantic Ocean. In the depth of Arc-Atlantic Ocean, ocean acidification is exacerbated in H1 and L3 relative to the REF simulation, mainly due to increased DIC, which dominates the impacts of increased ALK on seawater pH (figures 5(d)-(f)).
Notably, for the six model versions with different parameterizations of dissolution effect (L1-L3 and H1-H3, table 1, figures S1(a), (b)), resulting from different magnitude of dissolution effect on ocean ALK, the magnitude of feedbacks of dissolution effect on the ocean carbon cycle vary among simulations (figures 1, 2). In addition, the relationship between parameterization scheme used and feedbacks of dissolution effect is nonlinear. For the parameterization scheme of dissolution effect in model versions L1-L3, as the reaction order n remains the same (n = 1), However, the feedbacks of dissolution effect in simulations L1-L3 are nonlinear. For instance, at year 3500, ( Therefore, the nonlinearity between parameterization scheme used and feedbacks of dissolution effect could be significant.

Comparison between feedbacks of dissolution effect and feedbacks of CO 2 -induced radiative warming
The former sections analyse the modelled results with the incorporation of the warming effect. To quantify the radiative effect of increasing atmospheric CO 2 on the ocean carbon cycle, the second series of simulations without the effects of CO 2 -induced warming are also conducted. Simulated results reveal that, in the REF case, cumulative ocean CO 2 uptake in the simulation without the warming effect is 658 PgC larger than that in the simulation with the warming effect by year 3500 (figure 6(a), S5 and table S1). In comparison, the cumulative ocean CO 2 uptake in simulations with the dissolution effect is 195-858 PgC larger than that in REF simulation (figure 6(a), S5 and table S1). This comparison indicates that, on a timescale of longer than 1,500 years, in terms of ocean carbon sink, the feedback of dissolution effect can be comparable to or even stronger than feedbacks from the warming effect, showing the potentially importance of the dissolution effect.
Noteworthy, the feedback of dissolution effect on ocean CO 2 uptake operates ∼200 years later than feedbacks from the warming effect ( figure S5). Therefore, within a timescale of ∼400 years, in terms of oceanic CO 2 uptake, feedbacks from the warming effect would dominate feedbacks from dissolution effect; While on longer timescales, the feedbacks of dissolution effect become more important and eventually comparable to feedbacks from warming effect on millennial timescales (figure S5).

Discussion and conclusion
In this study, model simulations are performed to quantify the potential feedbacks of dissolution effect on the climate change and ocean carbon cycle. Based on the reference model version (REF), we incorporate six parameterizations of the dissolution effect into the original UVic model. Under SRES A2 CO 2 scenario and its extension with total emissions of 5,462 PgC, with the growth of atmospheric CO 2 concentration and the resultant seawater acidification, the inclusion of dissolution effect tends to elevate seawater ALK and intensify ocean carbon sink. Consequently, the dissolution effect provides negative feedbacks on atmospheric CO 2 increase and mitigates global warming. Our results show that, compared to the reference simulation (REF in  table 1), the simulations which incorporated dissolution effect diminish simulated atmospheric CO 2 concentration by 100-437 ppm by year 3500 (table S1). The strength of feedbacks of the dissolution effect relies on the exact form of dissolution-Ω C parameterization applied, highlighting the importance of parameter uncertainty of the dissolution effect in affecting the global carbon cycle. While the dissolution effect acts to intensify the ocean carbon sink, it also eases seawater acidification primarily caused by the rising ALK. For instance, by year 3500, in simulation L3, the dissolution effect increases surface mean pH by 0.17, corresponding to a decrease of 48% in [ ] + H relative to simulation REF. Noteworthy, in the epipelagic zone, relative to the Arc-Atlantic Ocean, the Pacific-Indian Ocean experiences greater acidification, leading to greater dissolution effects and the resultant stronger feedbacks on ocean carbon sink and acidification in the Pacific-Indian Ocean. For instance, at year 2600, in L3 simulation relative to the REF simulation, the inclusion of dissolution effect increases the ocean CO 2 uptake by 2.42gC m −2 yr −1 in the Pacific-Indian Ocean, which is 34% larger than the dissolution effect in the Arc-Atlantic Ocean. Additionally, our results show that the feedback of dissolution effect on ocean carbon sink is comparable to, and even stronger than the feedbacks from CO 2 -induced warming in some cases.
Historical observational studies provided references for our simulated results. For instance, our modelled export flux of CaCO 3 at ocean depth of 130m in the mid-1990s is ∼0.71 PgC yr −1 , which is certainly within the range of 0.5-1.8 PgC yr −1 according to observations combined with modelled results provided by Jin et al [50]. Our simulated CaCO 3 production is ∼0.74 PgC yr −1 in year 2000, which agrees with the observational range of 0.6-1.6 PgC yr −1 obtained from sediment trap and satellite data [51]. UVic-simulated CaCO 3 dissolution rate in year 2000 is ∼0.55 PgC yr −1 , which lies in the range of 0.3-0.7 reported by previous observational studies [52][53][54]. Dong et al [19] presented the CaCO 3 sinking flux data captured by sediment traps at 100m and 200m in the North Pacific in year 2017, which were in a range of ∼0.13-2.23mmolm −2 day −1 . Our simulated results show that, in year 2015, at depth of 177.5m in the North Pacific Ocean, the CaCO 3 sinking flux is 0.20-0.21mmolm −2 day −1 , which are within the observed range presented by Dong et al [19]. Consequently, our simulated results can capture the observed large scale ocean CaCO 3 cycle fields.
Our results present a potentially important feedback of dissolution effect on atmospheric CO 2 . Different model-based estimates of feedbacks from the dissolution effect are shown in several previous studies. For instance, by using low-sensitivity parameterizations of CaCO 3 dissolution rate to seawater [ ] -CO , 3 2 under IPCC SRES A1B scenario with total CO 2 emissions of 2,450 PgC, Ilyina and Zeebe [20] suggested that after 3,000 years since the preindustrial time, dissolution effect would decrease atmospheric CO 2 content by <10 ppm. By adopting nonlinear CaCO 3 dissolution rate-Ω parameterization schemes, Andersson et al [21] predicted that by the end of the twenty-first century, seawater acidification-induced CaCO 3 dissolution could buffer the carbon chemistry of pore waters, but overlying shallow-water acidification would not be noticeably buffered. By linking calcite dissolution rate with [ ] -CO , 3 2 under an Anthropocene-like scenario, Boudreau et al projected that after about 10,000 years since the preindustrial time, ocean acidification-induced CaCO 3 dissolution would buffer acidification basically back to preindustrial state, triggering notable effects on carbon cycle and climate system [22]. For comparison, in this study, under SRES A2 CO 2 emission scenario and its extension with an emission of 5,000 PgC in ∼400 years, by year 3500, the dissolution effect decline atmospheric CO 2 concentration by a noticeable 100-437 ppm, and buffers the decrease in sea surface pH by 0.04-0.17, accrossing different CaCO 3 dissolution parameterizations. Relative to previous modelling studies, our simulated results show the same sign, but greater magnitude in feedbacks of dissolution effect on the carbon cycle, which depends on the different representations of the dissolution effect. Therefore, to obtain more dependable predictions of the feedbacks of dissolution effect, more systematized observational and experimental studies about the responses of CaCO 3 dissolution rate to seawater acidification are required.
The potentially important feedbacks of CaCO 3 dissolution-Ω C dependence on the global carbon cycle are assessed in this study. Some other processes in the ocean CaCO 3 cycle which are not discussed could also trigger noticeable feedbacks on the global carbon cycle. For instance, the reduction in calcification rate due to ocean acidification, which is called 'calcification effect', would also alter the ocean carbon cycle [55]. Ridgwell et al [51] reported, due to the calcification effect, atmospheric CO 2 burden would be reduced by 4%-13% by year 3000, under a CO 2 scenario which has total emissions of 4,000 PgC. Hofmann and Schellnhuber [28], using a scenario with total emissions of 4,075 PgC, predicted a 9% decrease in atmospheric CO 2 by year 3000 resulting from the calcification effect. Zhang and Cao [55] showed that by year 3500, under a scenario based on SRES A2, the calcification effect acts to decrease atmospheric CO 2 concentration by 1%-11%, depending on the prescribed parameter scheme. Other relevant modelling studies show weaker feedbacks of the calcification effect on atmospheric CO 2 by ∼0.1%-1.2% [30,56,57]. In addition, the dependence of F POC on F PIC (particulate organic carbon and particulate inorganic carbon export fluxes to the deep ocean), which is termed 'ballast effect', is not considered in the model. Inhibited CaCO 3 production could reduce PIC export flux, which hence reduces POC export flux, weakening the ocean carbon pump and inhibiting the oceanic absorption of atmospheric CO 2 [28,29]. Previous relevant modelling studies show ballast effect-induced feedbacks of ∼0.4%-5.2% on atmospheric CO 2 [28,29,56,58]. For comparison, our simulated results show a dissolution effect-induced feedback of 6%-25% on atmospheric CO 2 content by year 3500, which is a relatively larger uncertainty. Therefore, with respect to the feedbacks of ocean CaCO 3 cycle on atmospheric CO 2 , the model parameter uncertainties of dissolution effect could have important impacts on projections of the carbon cycle and climate system. This study conducts model simulations to examine the impacts of CaCO 3 dissolution effect on inter-annual changes in the carbon cycle and climate system. In addition, largely as a results of seasonal changes in ocean temperature, the ocean CO 2 uptake shows a high seasonal variability. The resultant seasonal changes in DIC and ALK concentrations, as well as temperature changes, lead to seasonal changes in ocean pH and other key carbonate chemical fields [59][60][61]. Compared to inter-annual responses of ocean chemical fields to atmospheric CO 2 and climate change, little is known about the seasonal changes in ocean chemical fields, which would need further experimental, observational, and modelling studies [59].
In our simulated results, whereas the CaCO 3 dissolution effect intensify the ocean carbon sink, it also buffers seawater acidification primarily caused by the rising ALK. In addition to CaCO 3 , there are other minerals that can increase seawater ALK when they dissolve, such as olivine and quicklime [62,63]. By distributing these minerals on land (enhanced weathering or EW), or in the ocean (ocean ALK enhancement or OAE), global warming, as well as ocean acidification, could be mitigated, which is an important mCDR (marine carbon dioxide removal) geoengineering strategy [64][65][66]. For instance, under a large olivine deployment of small grain size (10 μm), atmospheric CO 2 could be reduced by >800 PgC by year 2100 [67]. Andrew et al [68] also proposed that, artificial ocean alkalization could offset warming and mitigate ocean acidification at the global scale. Our results emphasize the potentially important feedbacks of the dissolutions of CaCO 3 and other alkaline minerals on the carbon cycle and climate system, providing new insights into mCDR geoengineering strategies, especially OAE.
Our study highlights the potentially crucial role played by CaCO 3 dissolution effect in the ocean carbon sink, global carbon cycle and climate system over millennial timescales. Meanwhile, we provide useful reference to the parameterization of dissolution effect for modelling studies. More relevant observational, experimental, and modelling studies are required to obtain deeper insights of the dissolution effect, which is vital for more dependable predictions of ocean carbon sink, atmospheric CO 2, and climate change.