1 Introduction

The ocean is one of the largest carbon sinks, absorbing about 30% of anthropogenic CO₂ since pre-industrial times (Sabine et al., 2004; Cai et al., 2010). However, the CO₂ absorbed by the ocean reacts with seawater, resulting in a decrease in pH (Byrne et al., 2010), which is called the ocean acidification (Caldeira and Wickett, 2003; Feely et al., 2004). The average surface pH of seawater, which was about 8.2 before the industrial revolution, has dropped by 0.1 (Orr et al., 2005; Feely et al., 2009). The pH is expected to decline by a further 0.3 – 0.5 units by the end of this century and global surface ocean pH would become increasingly more homogeneous with time (Brewer, 1997; Jiang et al., 2019). The pH can directly reflect the amount of H⁺ in seawater, which is an important measure of ocean acidification. Dore et al. (2009) reported timeseries measurements of seawater pH and related parameters at the ALOHA observatory near the central North Pacific over the past 20 years, and found a significant long-term downward trend of about −0.0019 ± 0.0002 y⁻¹ in the pH of surface seawater, in response to the increase of the global atmospheric CO₂ concentration. In addition, aragonite has a higher solubility than calcite, resulting in organisms with aragonite calcium carbonate as their structure more susceptible to ocean acidification (Mucci, 1983; Fabry et al., 2009). Therefore, the aragonite saturation state (Ω_arag) is also commonly used to assess the impact of ocean acidification on calcareous organisms. The oceanic uptake of anthropogenic CO₂ would result in concomitant changes in seawater chemistry and adverse consequences for many organisms (Gattuso et al., 1999; Langdon and Atkinson, 2005; Iglesias-Rodriguez et al., 2008). Although pH and Ω_arag are two commonly used parameters in assessing the impact of ocean acidification, factors (e.g., temperature, CO₂ gas exchange) affecting the distributions of pH and Ω_arag could be quite different (Cai et al., 2020).

Previous work in the West Pacific Ocean has focused on the study of air-sea CO₂ exchange (Dore et al., 2003; McKinley et al., 2004), and the carbonate system as well as its related influence factors (Murata et al., 2009; Wakita et al., 2010). For example, Murata et al. (2009) studied the interdecadal variability of anthropogenic CO₂ along the 149˚E transect of the Northwest Pacific Ocean and Wakita et al. (2010) analyzed the interdecadal variation of DIC in the Northwest Pacific Ocean from 1992 to 2008. Both studies revealed the decadal variation of a single carbonate system parameter. However, the investigated data on the carbonate system are still scarce in this region. In this study, we investigated the whole carbonate chemistry of the surface seawater, including DIC, TA, pCO₂, pH and Ω_arag along the 150˚E transect in the Northwest Pacific Ocean. We also predicted the changes in pH and Ω_arag by the end of this century. This study aimed to reveal factors controlling the meridional distributions of the carbonate chemistry, with a special interest in the distributions of surface seawater pH and Ω_arag.

2 Materials and Methods 2.1 Study Area

The survey area (40˚ – 13˚N, 150˚E) is located northwest of the Pacific Ocean, whose hydrological conditions are affected by ocean currents and different water masses (Fig.1). From north to south, the survey area is affected by the Subarctic Current, Kuroshio Extension, Subtropical Countercurrent and North Equatorial Current. The Subarctic Current passing through the survey area is around 40˚N, which originates from the subarctic North Pacific Ocean dominated by upwelling. The Kuroshio Extension is around 34˚N, which is formed by the eastward turning of the Kuroshio Current near the Japan coast. The Subtropical Countercurrent is found in the latitudinal band of 22˚ – 25˚N, and the North Equatorial Current exists at the southern end of the survey area (Qiu, 2001).

2.2 Sampling

Surface seawater samples were collected at each latitude (at an interval of 1 degree) from 40˚N to 13˚N along the 150˚E transect (P1) aboard R/V 'Dongfanghong 3' in November 2019, except for stations P1-4, P1-17 and P1-18 (Fig.1). Water samples were collected using Niskin bottles mounted on a Seabird CTD system (911-plus, Seabird Corporation), which also measured the temperature and salinity of seawater. Duplicate pH samples were collected into 40 mL brown borosilicate glass vials after overflowing with at least twice their volume to minimize exposure with air. TA samples were collected into 250 mL high-density polyethylene (HDPE) bottles, immediately poisoned with 10 μL of saturated HgCl₂, then stored in the dark at room temperature and brought back to the land laboratory for analysis within two months.

2.3 Analytical Methods

TA was determined by Gran titration using an open-cell with an automatic potentiometric titration system (T960, Hanon), and the concentration of HCl was calibrated using seawater certified reference materials (CRMs, Batch 178) from A. G. Dickson lab. Each sample was measured three times, using aged Pacific surface seawater with known TA values for quality control, with an accuracy of 0.1%. The pH samples were measured on board on the total hydrogen ion concentration scale (pH_{T, 25}) at 25 ± 0.05℃ in a thermal bath using two independent benchtop pH meters (star A211, Thermo Fisher Scientific), each equipped with a combination electrode (8157BNUMD). The pH values were calibrated against the Tris buffer (pH = 8.094 at 25℃) from A. G. Dickson lab and converted to in situ temperatures (referred to as pH_T hereafter) using CO2SYS v2.1 program (Pierrot et al., 2006), with an overall precision of ± 0.005 pH units.

Knowing any two of the four carbonate system parameters (pH, TA, DIC and pCO₂), the others of the carbonate chemistry can be calculated based on the thermodynamic properties (Millero, 2007). Thus, pH_T, DIC, pCO₂, Ω_arag and Revelle factor (RF) were calculated from the measured pH_{T, 25} and TA data, using CO2SYS v2.1 program (Pierrot et al., 2006), together with the in situ temperature and salinity, and with the equilibrium constants of the carbonate acid K₁ and K₂ from Mehrbach et al. (1973) refit by Dickson and Millero (1987), the K_HSO4 was from (Dickson, 1990), and the [B]_T value from Uppström (1974). In order to assess the quality of the calculated data, DIC of an aged Pacific surface seawater (measured by a DIC analyzer (AS-C5, Apollo SciTech)) was compared with the one calculated from the pH_T and TA values which were measured by the same protocol as mentioned above. The difference between the measured and calculated DIC was within ± 2 μmol kg⁻¹, suggesting the calculated DIC data were reliable.

2.4 Predictions of pH_T and Ω_arag by the End of This Century

Current predictions suggest that atmospheric pCO₂ will rise to (900 ± 50) ppm by 2100 and the rise in global mean surface temperature is likely to be 1.4℃ to 3.1℃ higher than present (IPCC RCP 6.0 projections). To predict the distributions of in situ surface seawater pH_T and Ω_arag by 2100, we assumed a CO₂ concentration of 900 ppm in the atmosphere and a temperature 2℃ higher. We further assumed that surface ocean pCO₂ changes at the same rate as the atmospheric pCO₂ and the sea surface temperature (SST) will also rise by 2℃ by 2100 without TA changing over time. Surface seawater pH_T and Ω_arag at each sampling station along the P1 transect in 2100 were calculated from the surface seawater pCO₂ and TA using CO2SYS program.

2.5 Statistical Analysis

The distributions of surface seawater temperature, salinity, DIC, TA, pH_T, Ω_arag, pCO₂ and RF along 150˚E transect from 40˚N to 13˚N were plotted using Ocean data view (Schlitzer, 2018). Correlation analysis in this study was performed by the Pearson correlation test using GraphPad Prism 8, with a significance level of 99%.

3 Results 3.1 Hydrographic Conditions

SST was in the range of 14.5 – 29.4℃ with a mean value of 26.0℃ and exhibited a clear latitudinal distribution (Fig.2a), increasing with decreasing latitude with abnormally low values at two northernmost stations (40˚N and 39˚N). The salinity was in the range of 33.46 – 35.04, with a mean value of 34.56, and the lowest values were also found at the two northernmost stations (Fig.2b), which were 33.61 and 33.46, respectively. Except for those two northernmost stations, the salinity of the rest stations was above 34 and generally followed bimodal distribution. The distributions of SST and salinity along the P1 transect revealed that the two northernmost stations were likely under the impact of the Subarctic Current, which is low in temperature and salinity.

3.2 Distributions of Carbonate Chemistry

DIC values ranged from 1915 to 2014 μmol kg⁻¹, decreasing with the decrease of latitude, with a mean value of 1954 μmol kg⁻¹ (Fig.3a). TA was in the range of 2243 – 2291 μmol kg⁻¹, with a mean value of 2270 μmol kg⁻¹; it increased gradually from 40˚N to 31˚N and high values appeared near 31˚N. South of 31˚N, there was no latitudinal gradient in TA distribution (Fig.3b). The salinity-normalized DIC and TA (nDIC = DIC/S*35, nTA = TA/S*35) ranged from 1949 to 2107 μmol kg⁻¹ and 2288 to 2352 μmol kg⁻¹ along the P1 transect, respectively, with abnormally high values both at the two northernmost stations (Figs.3c and d). In general, nDIC decreased with decreasing latitude. But for nTA, except for the two northernmost stations, nTA of other stations were obviously lower and relatively homogeneous. The surface seawater pH_T ranged from 8.044 to 8.110, displaying a clear decrease pattern from north to south along the P1 transect (Fig.3e). Ω_arag also exhibited a significantly latitudinal pattern, but in contrast to pH_T, it increased with decreasing latitude, ranging from 2.61 to 3.88 (Fig.3f). The surface seawater pCO₂ was in the range of 332 – 387 μatm, increasing with the decrease of latitude (Fig.3g); all of them were well below the atmospheric pCO₂ which was around 412 μatm during the survey period (Dlugo-kencky and Tans, 2019), indicating that the whole survey area was a CO₂ sink. The difference in pCO₂ between the atmosphere and surface seawater (ΔpCO₂ = pCO_{2, air} – pCO_{2, seawater}) decreased from north to south, ranging from 80 to 25 μatm, indicating that the northern area was a stronger CO₂ sink. The spatial distribution of RF showed a clear meridional gradient along the P1 transect (Fig.3h), decreasing with the decrease of latitude from 11.02 to 8.98, with abnormally high values at the two northernmost stations.

4 Discussion 4.1 Factors Affecting the Distributions of DIC, TA, pCO₂ and RF

In order to better understand the processes controlling the distributions of carbonate chemistry, the survey regions along the P1 transect were divided into two subregions, namely, region 1 (stations from 38˚ to 13˚N) and region 2 (the Subarctic Current dominant regions: stations 40˚ and 39˚N), according to the SST-latitude and salinity-latitude diagrams (Figs.2c and d).

In region 1, DIC decreased from north to south along the P1 transect, while no clear pattern was found for TA. The variation of TA was closely related to salinity change (r = 0.96, P < 0.0001) (Fig.4a), indicating that TA was mainly controlled by water mixing. In regard to DIC, it is not only affected by the salinity change but also greatly affected by the air-sea CO₂ exchange and biological activities (Cai et al., 2020). In this study, there was a certain correlation between DIC and ΔpCO₂ (r = 0.63, P = 0.0012) (Fig.4b). Since surface seawater pCO₂ values along the P1 transect were all below the atmospheric pCO₂ level, the larger ΔpCO₂ leads to more CO₂ being absorbed, resulting in higher DIC values in the northern parts of P1 transect. The Subarctic Current that originates from the subarctic North Pacific Ocean dominated by upwelling causes surface water around region 2 to have abnormally high TA and DIC, compared to the values predicted by the regression lines from region 1 (red dots in Figs.4a and b).

The strong positive correlation between temperature and pCO₂ (r = 0.93, P < 0.0001, excluding the two northmost stations) (Fig.4c) indicated that temperature plays an important role in latitudinal surface seawater pCO₂ distribution. The higher the temperature, the lower the solubility of CO₂, which results in an increase in seawater pCO₂ and a decrease in carbon sink intensity. The pCO₂ values of the two northernmost stations (red dots in Fig.4c) were significantly higher than those predicted by the regression line from region 1, possibly due to the impact of the Subarctic Current. Nevertheless, the surface seawater pCO₂ remained substantially lower than the atmospheric pCO₂, reflecting the important role of low SST in keeping pCO₂ low via thermodynamic equilibrium shift and in maintaining a significant air-sea CO₂ disequilibrium (Cai et al., 2020).

RF, defined as the ratio between the fractional change in pCO₂ to the fractional change in DIC under the condition of constant temperature, salinity and TA, is a measure of the buffering capacity of seawater (Zeebe and Wolf-Gladrow, 2001). RF can be regarded as a function of temperature, which decreased with increasing temperature (r = −0.98, P < 0.0001, excluding the two northmost stations) (Fig.4d), explaining the latitudinal variation of RF along the P1 transect from north to south. The abnormally high RF values (red dots in Fig.4d) at the northern ends of P1 are mainly related to the abnormally low salinity and TA caused by the Subarctic Current passing through this region.

4.2 Effects of SST and ΔpCO₂ on pH and Ω_arag

From the north (40˚N) to the south (13˚N) along the P1 transect, the SST increased by nearly 15℃ and ΔpCO₂ decreased by 54 μatm, which provides a favorable condition for studying the effects of spatial variations of temperature and ΔpCO₂ on pH and Ω_arag. The spatial variations of pH_T and Ω_arag between the northern end (40˚N) and the southern end (13˚N) along the P1 transect were 0.064 and −1.20, respectively. There were strong correlations between pH_T and SST (Fig.5a, r = −0.96, P < 0.0001), and between Ω_arag and SST (Fig.5b, r = 0.99, P < 0.0001), indicating that SST might play important roles in spatial distributions of pH_T and Ω_arag. According to Jiang et al. (2019) and Xue et al. (2020), the influence of temperature on pH and Ω_arag can be divided into two aspects, namely, the internal temperature effect and the external temperature effect. The internal temperature effect refers to changing the existing form of CO₂ dissolved in water under constant TA and DIC conditions (i.e., closed system), increasing the dissociation of HCO₃⁻ and H₂O and the concentration of H⁺ and CO₃²⁻, thus decreasing pH and increasing Ω_arag. The external temperature effect refers to the gas exchange caused by the change in CO₂ solubility, which decreases with increasing temperature, resulting in a decrease in seawater CO₂ concentration, thus increasing both pH and Ω_arag. By altering the acidbase equilibrium of the carbonate system and CO₂ solubility-driven air-sea exchange, the combined effects of temperature would cancel out for pH resulting in little change of pH with temperature, whereas they reinforce each other for Ω_arag. However, this hypothesis might not be completely true when air-sea CO₂ disequilibrium or other nonthermal components (e.g., water mass mixing and biological processes) have to be taken into consideration. As revealed by the study of Xue et al. (2021), if the seawater is under air-sea disequilibrium or under the influences of water mass mixing or biological processes, pH would be either more controlled by the thermal or nonthermal components, depending on their competing effects, while Ω_arag would be almost always dominated by its nonthermal components. In this study, pH_T decreased with the increasing temperature, suggesting that it is more controlled by the thermal components (i.e., internal temperature effect, Fig.5a, red dashed line) than the nonthermal components. However, the effect of the thermal components on pH was about 0.21, but the actual pH_T decreased only about 0.06, indicating that the effect of the thermal components was partially counteracted by the nonthermal processes. Although Ω_arag was positively correlated with SST, the thermal components only partially contribute to an increase in Ω_arag (Fig.5b, red dashed line).

To further understand the impact of the nonthermal components on pH and Ω_arag, we examined the correlations of ΔpCO₂ (ΔpCO₂ = pCO_{2, air} − pCO_{2, seawater}, and ΔpCO₂ > 0 along the investigated P1 transect) with pH and Ω_arag, which were normalized to the temperature of 25℃ (pH_{T, 25} and Ω_{arag, 25}) to remove the thermal components. As shown in Figs.5c and d, both pH_{T, 25} and Ω_{arag, 25} are negatively correlated with ΔpCO₂ (r = −0.90, P < 0.0001 and r = −0.96, P < 0.0001, respectively), which is due to the fact that ΔpCO₂ to a large extent represents the nonthermal components, while the nonthermal components of both pH and omega are in phase; larger ΔpCO₂ would result in more CO₂ being absorbed, which would reduce the pH and Ω_arag in the same direction (Xue et al., 2021). Thus, from north to south along the P1 transect, although a decrease in ΔpCO₂ would increase pH, the effect is outweighed by the impact of temperature increase, resulting in a decrease in pH_T from high latitude to low latitude, while both an increase in temperature and a decrease in ΔpCO₂ (main controlling factor) increase the Ω_arag.

4.3 pH and Ω_arag by the End of This Century

The spatial distribution characteristics of pH_T and Ω_arag along the P1 transect at the end of this century under the IPCC RCP 6.0 projection were similar to the present (Figs.6a and b). pH_T²¹⁰⁰ was in the range of 7.769 – 7.748, slightly decreasing with the decrease of latitude. Ω_arag²¹⁰⁰ increased with decreasing latitude, ranging from 1.40 to 2.40. Both pH_T and Ω_arag showed large declines compared with the present (Figs.6c and d). The declines in pH_T and Ω_arag ranged from −0.343 to −0.296 and −1.17 to −1.50, respectively, in response to the temperature rise and the accumulation of atmospheric CO₂ concentration as projected. The decline in pH is consistent with the modeling study by Jiang et al. (2019), showing that the global average surface seawater pH_T decreased by about 0.33 ± 0.04 units from 2000 to 2100 under the RCP8.5 'business-as-usual' scenario. As discussed above, two effects of temperature mainly cancel out for pH, whereas they reinforce each other for Ω_arag (i.e., Ω_arag would increase with temperature increase). Therefore, the rise of temperature by the end of this century contributes little to the decline in pH and Ω_arag, but rather the accumulation of the atmospheric CO₂, which leads to a continuous increase in atmospheric pCO₂, forcing more atmospheric CO₂ into seawater through the sea-air exchange, resulting in an increase in H⁺ and a decrease in CO₃²⁻, thus leading to the decrease of both pH and Ω_arag.

The decline in pH_T from the present to the year of 2100 (∆pH_T) decreased with decreasing latitude; ∆pH_T was larger at high latitudes than that at low latitudes along the P1 transect (Fig.6c), which is closely related to the distribution of RF. RF is a good indicator of ocean buffer capacity; the higher its value is, the weaker the buffer capacity of seawater will be, and the greater the pH change will be, resulting in a smaller latitudinal difference in pH along the P1 transect. The decline in pH_T is consistent with the result obtained by Jiang et al. (2019) that the pH of global ocean surface water would gradually become homogenous with time. The decline in Ω_arag from the present to the year of 2100 showed a different latitudinal distribution pattern, which was slow in the northern end stations (i.e., the Subarctic Current dominant region), and relatively uniform and fast at low latitudes (Fig.6d). This is consistent with the findings of Feely et al. (2018) that the decreasing rate of Ω_arag is faster in warm and high Ω_arag waters. In a word, the decline in pH and Ω_arag over time can vary differently in cold high latitude vs. warm low latitude waters, which is consistent with the findings from the American ocean margins (Cai et al., 2020).

5 Conclusions

The distribution patterns of carbonate chemistry along the 150˚E from 40˚N to 13˚N and their controlling factors were investigated in November 2019. The Subarctic Current at the northern ends caused abnormally low temperatures and salinities of the surface seawater, and high values of nDIC and nTA. DIC and Revell factor decreased with the decrease of latitude, while pCO₂ increased with the decrease of latitude but all were below atmospheric pCO₂ level, resulting in a decrease in ΔpCO₂ from north to south. The distributions of pH_T and Ω_arag were out of phase; from north to south, pH_T was more controlled by the thermal components resulting in a decrease in pH_T while the increase of temperature and the decrease of ΔpCO₂ both increased the Ω_arag. In addition, pH_T and Ω_arag along 150˚E at the end of this century were predicted to decrease but at different decline rates, with larger declines in higher pH_T and Ω_arag regions, suggesting that meridional gradients of pH_T and Ω_arag would become homogenous with time in future.

Acknowledgements

This study was supported by the Key Research and Development Program of Shandong Province (No. 2020 ZLYS04), the National Key Research and Development Program of China (No. 2017YFA0604300), the Qingdao Pilot National Laboratory for Marine Science and Technology (No. 2018SDKJ0105-1), the Fundamental Research Funds for the Central Universities (No. 202072 001), and the Young Scholars Program of Shandong University (No. 2018WLJH43). We wish to thank crew members of the R/V 'Dongfanghong 3' for their help during the investigation.