Simulated CO2-induced ocean acidification for ocean in the East China: historical conditions since preindustrial time and future scenarios

Since preindustrial times, as atmospheric CO2 concentration increases, the ocean continuously absorbs anthropogenic CO2, reducing seawater pH and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$[{{\rm{C}}{\rm{O}}}_{3}^{2-}]$$\end{document}[CO32−], which is termed ocean acidification. We perform Earth system model simulations to assess CO2-induced acidification for ocean in the East China, one of the most vulnerable areas to ocean acidification. By year 2017, ocean surface pH in the East China drops from the preindustrial level of 8.20 to 8.06, corresponding to a 35% rise in [H+], and reduction rate of pH becomes faster in the last two decades. Changes in surface seawater acidity largely result from CO2-induced changes in surface dissolved inorganic carbon (DIC), alkalinity (ALK), salinity and temperature, among which DIC plays the most important role. By year 2300, simulated reduction in sea surface \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$[{{\rm{C}}{\rm{O}}}_{3}^{2-}]$$\end{document}[CO32−] is 13% under RCP2.6, contrasted to 72% under RCP8.5. Furthermore, simulated results show that CO2-induced warming acts to mitigate reductions in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$[{{\rm{C}}{\rm{O}}}_{3}^{2-}]$$\end{document}[CO32−], but the individual effect of oceanic CO2 uptake is much greater than the effect of CO2-induced warming on ocean acidification. Our study quantifies ocean acidification induced by anthropogenic CO2, and indicates the potentially important role of accelerated CO2 emissions in projections of future changes in biogeochemistry and ecosystem of ocean in the East China.

Atmospheric CO 2 concentration has reached 407.44 ± 0.10 ppm (parts per million) by year 2018, increased by 46% since preindustrial time 1 , which is mainly due to human activities of fossil fuel burning and land use changes. Observational-based estimates show that, during year 1750 and 2017, total anthropogenic CO 2 emission is 660 ± 95 PgC (1 PgC = 10 15 grams of carbon = 1 billion tons of carbon) 2 . About 42% of these emissions stayed in the atmosphere, meanwhile, about 25% and 33% of the emissions were absorbed by the ocean and terrestrial biosphere, respectively 2 .
The rise of atmospheric CO 2 concentration results in global warming through trapping long wave radiation, a process known as greenhouse effect [3][4][5][6] . Global warming could alter physical, chemical, and biological processes in the ocean [7][8][9][10] . Oceanic uptake of CO 2 could buffer part of the global warming; however, not only global warming, the penetration of anthropogenic CO 2 would also perturb ocean chemistry by making seawater more acidic, which is termed as ocean acidification 11 .
Generally, ocean acidification is caused primarily by oceanic CO 2 uptake from the atmosphere. In addition, especially in the coastal seas, there are some other factors that could also lead to the acidification of seawater. For example, increasing input of anthropogenic nitrogen to the ocean, and the resultant changes in organic matter production, oxidation, and deoxygenation, may have effects on ocean acidification 12 . However, many recent studies suggested that impacts of biological nitrogen assimilation and release on ocean acidification are negligible compared with impacts of CO 2 absorption 13,14 . Variations in riverine carbon fluxes to the ocean could either accelerate or offset seawater acidity in coastal areas, revealing large uncertainties 15 . Also, some processes in the ocean CaCO 3 cycle, including calcification and ballast effects, could also trigger feedbacks to ocean acidification [16][17][18] . This study will focus on the individual effect of oceanic CO 2 uptake on ocean acidification, which is generally considered a key process affecting seawater acidity and the ocean carbon cycle in the East China. In addition, we also assess the effect of CO 2 -induced warming on ocean acidification, and compare the strength of individual effects of oceanic CO 2 uptake and CO 2 -induced warming on ocean acidification in the East China.
Seawater pH is known as a measurement to quantify the degree of ocean acidification. Since the industrial revolution, sea surface pH (pH = −log 10 [H + ]) has dropped by about 0.1 units by year 2013, corresponding to an increase of 26% in hydrogen ion concentration ([H + ]) 19 . The current pH reduction rate is likely to be the highest during the past hundreds of thousands of years 20 . The elevated hydrogen ion concentration tended to reduce carbonate ion concentration ( − [CO ] 3 2 ) via: The reduction in carbonate ion concentration would lower seawater calcium carbonate (CaCO 3 ) saturation state (Ω) for aragonite or calcite (two different polymorphs of CaCO 3 ), which is defined as: where K sp ⁎ is the stoichiometric solubility product constant with respect to aragonite or calcite 3,21 . The main concern of ocean acidification originates from the potentially adverse impacts of CaCO 3 saturation state reductions on marine calcifying organisms, which use CaCO 3 to form their skeletons or shells. For instance, reduced Ω could decrease calcification rate of calcifying organisms [22][23][24][25][26][27][28] , and increase CaCO 3 dissolution rate 29-34 , making their skeletons or shells vulnerable. In addition, by conducting meta-analyses of case studies, Jin et al. (2015) suggested, during the past decade, the echinoderm/microbenthic productivity has been altered in the seas around China, which has implications for the effects of ocean acidification on marine ecosystem 35 . Concluded from laboratory experiments, Liu and He (2012) also provided evidences of the potentially important impacts of ocean acidification on metabolic processes of some calcifying organisms in the China seas 36 . Calcifying organisms in the China seas could be of major ecological and economic importance. China is known as one of the most important countries of marine aquaculture industry. At year 2016, aquaculture production (excluding aquatic plants) from China is 4.97 × 10 7 t, accounting for about 61% of the global total aquaculture production, and large quantities of the marine aquaculture production are calcifying organisms 37 . Therefore, acidification in the China seas could lead to reductions in the aquaculture production and the consequent economic losses, meanwhile, posing threats to marine ecosystems.
For lack of historical observational data and the requirement for future projections in ocean chemistry fields, numerical models were used to investigate the changes in ocean acidification and biogeochemical processes in previous studies. For example, modeling studies show consistently that the drop in sea surface mean pH since preindustrial times is about 0.1 11,38,39 , which is compared well with observational-based estimates 19 . By forcing the Lawrence Livermore National Laboratory ocean general-circulation model under IPCC SRES scenario, Caldeira and Wickett (2005) indicated that ocean surface pH could be reduced by 0.3-0.5 units by the end of this century 40 . By forcing 13 ocean-carbon cycle models under IPCC IS92a scenario, Orr et al. (2005) projected that surface seawater would be undersaturated with respect to aragonite in Mid-21st century 38 .
In addition to the global scale, projections of future ocean acidification for specific spots of ocean in the East China are also provided in some other studies [41][42][43][44] . For example, Chou et al. reported that under the IPCC IS92a emission scenario, under the total effects of eutrophication and elevated atmospheric CO 2 , the bottom water of the Yangtze River plume area would be undersaturated with respect to aragonite (Ω A ≈0.8) by the end of this century, threatening the benthic ecosystem 43 . Zhai, using a predicted scenario that atmospheric CO 2 increases by 100 ppm for the 2050 s since present, proposed that half of the Yellow Sea benthos would be covered by acidified seawater having a critical Ω A of less than 1.5 41 45 , the ocean carbon cycle [46][47][48] , and historical oceanic uptake of carbon and its isotopes 49 . Refer to Methods section for detailed descriptions of the UVic model.
In this study, we extend previous studies by using UVic model to quantify ocean acidification induced by anthropogenic CO 2 for ocean in the East China (115-130°E, 20-40°N). Usually, previous studies would only focus on analysing the changes in seawater acidity, e.g., seawater pH; this study further quantifies the impacts of changes in ocean chemistry (i.e., DIC, ALK, salinity and temperature) on ocean acidification, as well as analyses the spatial heterogeneity of ocean acidification. In addition, we investigate the effects of CO 2 -induced warming on different ocean chemistry fields, and compare the strength of individual impacts of oceanic CO 2 uptake and CO 2 -induced warming on ocean acidification, which was ignored by previous relevant studies. Furthermore, in this study, we analyse the nonlinearity relationship between atmospheric CO 2 scenario used and ocean acidification, which enable us to have a better estimate of the extent of ocean acidification in the East China under different CO 2 emission policies. We aim to further our understanding of the role played by accelerated anthropogenic CO 2 emissions in the carbon cycle of ocean in the East China, which is also important for reliable projections of future changes in marine biogeochemistry and ecosystem in west Pacific.

Results
To quantify the effect of increasing atmospheric CO 2 concentration on ocean acidification in the East China, a series of 500-year Earth system model simulations are designed. From year 1800 to 2017, all model simulations are forced by observational-based atmospheric CO 2 concentration. After 2017, simulations are forced by Representation Concentration Pathway scenarios (RCPs, including RCP2.6, RCP4.5, RCP6.0, and RCP8.5) and their extensions up to year 2300 (Fig. 1a). To quantify the influences of CO 2 -induced warming on ocean acidification in the East China, we conducted an additional set of simulations in which CO 2 -induced warming is not allowed to affect the ocean carbon cycle. Refer to Methods section for detailed descriptions of the model and simulation experiments.
In the following, we first present results from the simulations including CO 2 -induced warming effects on the ocean carbon cycle. Then, in the "Impacts of CO 2 -induced warming on ocean acidification in the East China"  www.nature.com/scientificreports www.nature.com/scientificreports/ section, by comparing the results from simulations with and without CO 2 -induced warming, we analyse the effects of CO 2 -induced warming on ocean acidification in the East China.
Model evaluation. UVic-simulated global oceanic CO 2 uptake during the historical period since preindustrial time is consistent with observational-based estimates reported by IPCC AR5 8 (Table 1). For example, simulated cumulative oceanic CO 2 uptake during preindustrial time-year 2011 is 147 PgC, within the observational range of 155 ± 30 PgC reported by IPCC AR5 8 (Table 1). Model-simulated averaged oceanic CO 2 uptake during 2002-2011 is 2.4 PgC yr −1 , which compares well with the observed value of 2.4 ± 0.7 PgC yr −1 (Table 1). Model-simulated atmospheric CO 2 concentration is also consistent with observations. For example, simulated annual mean atmospheric CO 2 concentration at year 2017 is 405.0 ppm (Fig. 1a), compared well with observational-based estimate of 405.0 ± 0.1 ppm 1 .
Model-simulated carbon-related tracers are also compared with observed estimates from the Global Ocean Data Analysis Project (GLODAP) 50 . As shown in Supplementary Fig. S1, simulated vertical profiles of dissolved inorganic carbon (DIC) and alkalinity (ALK) for ocean in the East China agree well with observational-based estimates. In addition, the UVic model can capture the observed large-scale distributions of key tracers for the global ocean, as well as different ocean basins (refer to Supplementary Material of Cao et al. 48 ).

Ocean acidification in the East China: historical conditions since preindustrial time.
Under the CO 2 concentration scenario depicted in Fig. 1a and Supplementary Fig. S2a, by year 2017, the atmospheric CO 2 concentration increases to 404 ppm (see Supplementary Table S1). With the increasement in atmospheric CO 2 concentration, the ocean continuously absorbs anthropogenic CO 2 from the atmosphere, leading to acidification in the global ocean (see Supplementary Fig. S3). As shown in Fig. S3, generally, the mid-latitude ocean surface shows greater decreases in pH,  Table S1). Over the same time period, sea surface mean [CO ] 3 2− in the low latitudes reduced by 16%, contrasted to reductions by 18% for ocean in the East China ( Fig. 1e, Supplementary Fig. S3b, and Table S1). This study will focus on analysing ocean acidification conditions in the East China (unless otherwise stated), in which ecosystems could be especially vulnerable to ocean acidification, triggering important effects on global fishery and marine aquaculture industries. . For example, surface pH in the Yellow Sea shows a greater decrease (1800-2017) relative to the rest of ocean in the East China, which is largely due to the www.nature.com/scientificreports www.nature.com/scientificreports/ rapid elevation in surface temperature, while surface DIC and ALK over the Yellow Sea does not show relatively significant trends (Fig. 2b, Supplementary Fig. S4).
In previous projects and studies, observational-based estimates of ocean acidification for specific areas in the East China have also been reported. For instance, field surveys conducted by Zhai    Ocean experiences the greatest reductions in sea surface pH, mainly due to the effects of rising temperature (Fig. 4b).
Simulated results show that responses of ocean acidification could be sensitive to the atmospheric CO 2 scenarios used. For example, in the simulation under RCP2.6 scenario, by year 2300, the cumulative CO 2 uptake for ocean in the East China is 19.7 PgC, leading to reductions in sea surface pH and [CO ] 3 2− by 1% and 13% (Fig. 1,  Supplementary Table S1). In contrast, in the simulation under RCP8.5 scenario, by year 2300, the cumulative oceanic CO 2 uptake is 23.4 PgC, resulting in reductions in surface pH and − [CO ] 3 2 by 9% and 72%, respectively (Fig. 1, Supplementary Table S1).
In addition, the relationship between atmospheric CO 2 scenario used and ocean acidification in the East China is nonlinear (Fig. 6). [CO ] 3 2 RCP4 5 RCP2 6 /ΔCO 2 RCP4.5-RCP2.6 = −0.26 (Fig. 6, Table S1), showing greater nonlinearity than surface pH. The nonlinearity between www.nature.com/scientificreports www.nature.com/scientificreports/ atmospheric CO 2 scenario used and ocean acidification is noteworthy, which hints that if we aim to mitigate ocean acidification in the East China under a scenario of high atmospheric CO 2 content, a deeper reduction of anthropogenic CO 2 emission may be needed.
With the decrease of seawater [CO ] 3 2− in the East China, the aragonite saturation state (Ω A ) would also diminish (Fig. 1g,h). We take the simulations under RCP8.5 and RCP4.5 scenarios for example, to investigate the possible effects of ocean acidification on seawater chemistry in the East China under intensive and medium atmospheric CO 2 scenarios. Under RCP8.5, due to the oceanic CO 2 uptake and the resultant reduction in − [CO ] 3 2 , ocean Ω A drops from the preindustrial value of 1.6 to 0.7 by year 2300, and seawater Ω A would become less than 1 at nearly all ocean depths (Fig. 1h, Supplementary Fig. S7a and Table S1). The decreases in seawater Ω A would pose threats to calcifying organisms over ocean in the East China. For example, seawater that surrounding coral reefs becomes more and more acidic, from surface to depth (see Supplementary Figs. S7a, and S8a-c). Aragonite is the main constituents of calcareous endoskeleton of corals, therefore, the corals surrounded by undersaturated seawater with respect to aragonite (Ω A < 1) would encounter adverse impacts. There are also intriguing evidences that even in supersaturated seawater, CaCO 3 also dissolves 3,51 . Seawater chemistry fields at different depths have different responses to oceanic CO 2 uptake. For ocean in the East China, by year 2100, the reduction in Ω A is 2.0 at depth of 17.5 m, which becomes 0.3 at depth of 642.5 m (see Supplementary Fig. S7a). Changes in ocean chemistry at depths lag behind changes in the surface ocean because of the long time scale associated with the penetration of CO 2 into the deep ocean.
In comparison, in the simulation under RCP4.5 scenario, ocean Ω A drops from the preindustrial value of 1.6 to 1.2 by year 2300 (Fig. 1h, Supplementary Table S1). Compared with simulated results under RCP8.5, simulation under RCP4.5 presents higher ocean Ω A at years 2100 and 2300 (see Supplementary Figs. S7 and S8). In addition, from year 2100 to 2300, in simulation under RCP8.5, Ω A at different depths continues to decrease, while in simulation under RCP4.5, the reductions in Ω A at different depths, especially the decreases in surface Ω A are slight (Fig. 1, Supplementary Figs. S7 and S8). Therefore, responses of ocean acidification in the East China would be sensitive to the changes in atmospheric CO 2 , demonstrating the important impacts of atmospheric CO 2 changes on marine chemistry.

Impacts of CO 2 -induced warming on ocean acidification in the East China.
The impact of CO 2 -induced warming on ocean acidification in the East China could be inspected by comparing modeled ocean chemistry fields from simulations with and without CO 2 -induced warming (see Supplementary Table S2 and Fig. S9). Generally, CO 2 -induced warming would decrease the amount of ocean uptake of atmospheric CO 2 , mitigating ocean acidification (see Supplementary Table S2). Previous studies concluded that, this warming-induced reduction in oceanic CO 2 uptake is mainly as a result of warming-induced decreases in CO 2 solubility and ocean www.nature.com/scientificreports www.nature.com/scientificreports/ ventilation (ocean mixing and circulation) 9,52,53 . Effects of warming-induced changes in ocean biology processes (including phytoplankton growth and mortality rates, and detritus remineralization) offset with each other, therefore the total warming-induced biological impact on oceanic CO 2 uptake and ocean acidification is small (refer to Cao and Zhang, 2017) 9 .
For regional ocean in the East China, the responses of seawater pH and [CO ] 3 2− to CO 2 -induced warming are different (Fig. 7). The decoupled effects of CO 2 -induced warming on pH and [CO ] 3 2− are mainly due to their different thermodynamic dependence on temperature 39 (Fig. 4). For pH, the direct effect of rising temperature tends to reduce seawater pH (Fig. 4b), while warming-induced reductions in CO 2 solubility and ocean ventilation would suppress oceanic CO 2 uptake and mitigate the decrease in ocean pH. Total influence of CO 2 -induced warming on seawater pH mainly depends on the relative importance of these two effects, i.e., thermodynamic effects of temperature increasing and warming-induced reductions in CO 2 solubility and ocean ventilation. For ocean in the East China, relative to ocean depths, changes of pH in the ocean surface depend more on thermodynamic effects of rising temperature, resulting in decreased surface pH due to CO 2 -induced warming (Fig. 7a). At ocean depths, at year 2100, pH changes depend more on warming-induced reductions in CO 2 solubility and ocean ventilation, leading to increased ocean pH due to CO 2 -induced warming (Fig. 7b). For instance, at year 2100, in simulation RCP4.5, for ocean in the East China, surface pH decreases by about 0.005 due to CO 2 -induced warming, whereas ocean mean pH increases by 0.005 due to CO 2 -induced warming (Fig. 7 and Table S2). For  (Fig. 7c-f). Compared to the individual effect of oceanic CO 2 uptake, the effect of CO 2 -induced warming on ocean acidification in the East China is relatively small (see Supplementary Fig. S9). For instance, at year 2300, under RCP8.5, for ocean in the East China, CO 2 -induced surface  (Fig. 7, and Supplementary Table S2, Fig. S9).

Discussion
In this study, we conduct model simulations to assess ocean acidification in the East China on the timescale of centuries. During preindustrial time-year 2017, under the atmospheric CO 2 scenario that is based on observations, as atmospheric CO 2 concentration increases, the global ocean experiences acidification, and the ocean surface in the East China is one of the most vulnerable areas to ocean acidification. By year 2017, sea surface pH in the East China drops from the preindustrial level of 8.20 to 8.06, corresponding to a 35% rise in [H + ]. In addition, the decrease rate of surface pH becomes faster in the last two decades. The changes in surface seawater acidity largely result from CO 2 -induced changes in surface DIC, ALK, salinity and temperature, among which Moreover, our simulated results show that the relationship between atmospheric CO 2 scenario used and ocean acidification is nonlinear. This is important because if we want to mitigate ocean acidification in the East China under a scenario of high CO 2 concentration, a deeper reduction of anthropogenic CO 2 emission may be needed. Furthermore, our simulations show that with the continuous oceanic CO 2 uptake, changes in deep ocean chemistry exhibits time lag relative to the surface ocean, due to the long time scale associated with the slow penetration of excess CO 2 to the deep ocean. The long time scale for changes in deep ocean chemistry also indicates the urgent of deep reductions in anthropogenic CO 2 emissions, to avoid continuous accumulation of CO 2 at ocean depths. In addition, CO 2 -induced warming acts to mitigate the reductions in seawater [CO ] 3 2− and Ω A in the East China, and the individual effect of oceanic CO 2 uptake is much greater than the effect of CO 2 -induced warming on ocean acidification.
Under RCP8.5, for ocean in the East China, by year 2300, with the decrease of ocean [CO ] 3 2− , aragonite saturation state (Ω A ) would drop from its initial value of 1.6 to 0.7, and seawater Ω A would become undersaturated (Ω A < 1) at nearly all ocean depths. Ocean in the East China is the habitat of numerous amounts of corals, fish, shellfish, and other calcifying organisms, such as crustaceans (e.g., penaeus, scylla serrata), gastropods (e.g., cypraea tigris, cassis cornuta), coccolithophorids (e.g., chrysophyta), including rare species. These calcifying organisms may not be able to acclimate to the reduction in seawater Ω A . By carrying out short-term pCO 2 /pH perturbation experiments, Wu and Gao concluded that, the combined impacts of seawater acidification and solar UV changes could also inhibit photosynthesis in the China seas 54 . Morphology, physiology and behavior of some other marine organisms (e.g., molluscs, cnidarians) could also be impacted by ocean acidification 55 . Therefore, ocean acidification in the East China could have adverse effects on fundamental biochemical processes and marine ecosystems. Since ocean in the East China plays an important role in global fishery and marine aquaculture industries, its trend in ocean acidification would have far-reaching consequences for the millions of people that depend on the food and other resources in the ocean for their livelihoods 56 .
Our study has investigated the CO 2 -induced ocean acidification conditions in the East China on timescales of centuries by using an Earth system model. Some processes or feedbacks that are not considered in this study may also have effects on ocean acidification 16,17,57 . For example, ocean acidification tends to suppress the calcification rate of some marine calcifying organisms, increasing surface ocean alkalinity and reducing seawater acidity 16 . This study also does not include the interactive feedbacks between ocean acidification and CaCO 3 in the sediments, which is considered to reduce the chemistry change extent in the deep ocean on timescales longer than a millennium [58][59][60] .
In this study, based on model-simulated results, we diagnose ocean acidification in the East China during preindustrial time-year 2017, and highlight the potential future ocean acidification condition over timescales of centuries. Meanwhile, this study tries to provide useful information about the changes in future marine biogeochemical environment. Further observational and modeling studies would be required to develop a better understanding of the ocean carbon cycle and marine biogeochemistry, which is crucial for more reliable projections of future ocean acidification and its impacts on marine ecosystems.

Methods
Model description. In this study, we utilize the University of Victoria Earth System Climate Model (UVic ESCM) version 2.9, an intermediate complexity Earth system model 46 . The UVic model consists of an energy-moisture balance atmospheric model 61 , a 3D ocean general circulation model 62 , a thermodynamic/dynamic sea ice model 63,64 , and ocean and land carbon cycle models 46,[65][66][67] . The horizontal resolution of the UVic model is 1.8° (latitude) × 3.6° (longitude), which is similar to the resolutions of most coupled Atmosphere-Ocean General Circulation Models (AOGCMs) 45  The ocean carbon cycle model of UVic comprises inorganic and organic carbon cycle modules. The inorganic carbon cycle is based on the Ocean Carbon-Cycle Model Intercomparison Project (OCMIP) 68 . The organic carbon cycle is represented by a nutrient-phytoplankton-zooplankton-detritus (NPZD) ocean ecosystem/biogeochemical model 46 72 , and a few thermohaline circulation experiments 73,74 . The UVic model has been widely used in the studies concerning future evolutions of the ocean biogeochemical cycles 46 , interactions between global carbon cycle and climate change 71 , and projections of ocean acidification 75,76 .
The UVic model also has been widely used to investigate spatial distributions of physical and biogeochemical fields in both paleo and contemporary climate studies. For instance, Alexander et al. (2015) Xiao et al. (2012) also suggested that the mean error of air temperature simulated by the UVic model could be even smaller than that by many famous complex models 45,79 . In addition, Meissner 80 , Weaver et al. 81 , Spence and Weaver 82 , and Muglia and Schmittner 83 conducted UVic simulations to investigate variations in Atlantic meridional overturning circulation. Therefore, it is feasible to use the UVic model to quantify ocean acidification in the East China induced by oceanic uptake of anthropogenic CO 2 on timescales of centuries.
Simulation experiments. The UVic model was first spun up for 10,000 model years with a fixed preindustrial CO 2 concentration of 280 ppm to reach a quasi-equilibrium state of carbon cycle and climate system. Then, using this preindustrial state as an initial condition for the calendar year of 1800, two sets of four 500-year transient simulations are performed (i.e., from year 1800 to 2300). In the first set of simulations, rising atmospheric CO 2 concentration affects both the ocean carbon cycle and atmospheric radiation. While in the second set of simulations, rising atmospheric CO 2 concentration is not allowed to affect atmospheric radiation, that is, the ocean carbon cycle would not be impacted by CO 2 -induced warming. Each set of experiments include four simulations. From year 1800 to 2017, atmospheric CO 2 concentration data are taken from observational-based estimates, and after 2017, CO 2 concentrations are taken from the Representation Concentration Pathway scenarios (RCPs) and their extensions up to year 2300 84 (Fig. 1a). The four scenarios used are RCP2.6, RCP4.5, RCP6.0, and RCP8.5, based on different mitigation policies for greenhouse gases 84,85 . The numbers after "RCP" represent that by year 2100, the radiative forcing reaches 2.6, 4.5, 6.0, or 8.5 W m −2 , respectively. Refer to Meinshausen et al. (2011) for detailed descriptions of these RCP scenarios 84 . Analysis of ocean chemistry fields. In this study, we calculate ocean carbonate chemistry fields, including seawater pH, − [CO ] 3 2 and Ω A , based on equations from the OCMIP-3 project (http://ocmip5.ipsl.jussieu.fr/ OCMIP/). We use UVic-simulated ocean temperature, salinity, DIC, ALK, and observational-based estimates of ocean phosphate and silicate concentrations from the Global Ocean Data Analysis Project (GLODAP) 50 3 2 ), and 4 equations (Eqs. (3-6)) of ocean carbonate chemistry, given 2 known variables, we can calculate the rest 4 variables 87 . Therefore, changes in temperature, salinity (temperature and salinity changes would affect ⁎ K 1 and ⁎ K 2 ), DIC, and ALK, would result in changes in pH (pH = −log 10

Data availability
All data generated or analysed during this study are included in this published article (and its Supplementary  Information files).