The surveying area was divided into three zones from north to south by the Kuroshio front (~35°N) and the subtropical front (~27°N), such as the KE (35°−39°N), KR (27°−35°N) and subtropical (21°−27°N) zones (Kobashi et al., 2006; Oka et al., 2018; Figure 1A ). The western North Pacific is stired by two strong Pacific western boundary currents: the Kuroshio and Oyashio. The southward-flowing Oyashio Current carries the low-salinity/cold subarctic water and meets the high-salinity/warm Kuroshio water (Yasuda et al., 1996) to form the interfrontal zone with a relatively high Chl a concentration in spring ( Figures 1C-H ). The northeastward-flowing Kuroshio Current separates from the coast of Japan at approximately 35°N, 140°E and turns eastward (Qiu and Chen, 2011), forming the KE ( Figure 1A ). South of the KE, the KR has a deepened mixed layer in winter owing to oceanic heat loss to the atmosphere (Kitamura et al., 2016). Even in late spring, the mixed layer depth (MLD) in the KR and KE zones was deeper than that in the subtropical zone ( Figure 1B ). In winter and spring, sea surface heat loss causes surface pCO₂ to be significantly lower than the air-equilibrated value in the KE and KR zones; thus, the region is a strong net annual CO₂ sink (Takahashi et al., 2009; Li et al., 2022a). Further south of the KR, an eastward-flowing Subtropical Countercurrent (STCC) occupies the area (Yoshida and Kidokoro, 1967).

The survey was conducted in the western North Pacific onboard R/V Xiangyanghong 3 from 10 May to 7 June 2018 ( Figures 1A, B ). We conducted pumping analyses of sea surface temperature (SST), salinity, fluorescence, pH and pCO₂. The seawater sample was pumped approximately ~5 m below the sea surface. The underway SST was recorded every 5 s along the cruise path using an onboard SeaBird flow-through thermosalinograph (SBE 38, Sea-Bird Scientific, USA). An automated flow measuring system (AS-P2, Apollo SciTech, Inc., USA) was used for sea surface salinity (SSS), fluorescence, pH, and pCO₂ analyses, which were recorded continuously every 29 s along the cruise track.

A Honeywell Durafet^® pH sensor was used to measure the pH of the flowing water. The Durafet pH electrode features an integral automatic temperature compensator in a one-piece construction and is suitable for varying pH and temperature ranges. The Durafet pH electrode was calibrated using three standard buffers (pH_NBS = 4.01, 7.00 and 10.01 at 25.0°C, Thermo Fisher Scientific Inc., USA). NBS stands for the National Bureau of Standards, which is now the National Institute of Standards and Technology of the U.S. Department of Commerce. The Durafet pH sensor operates with a short-term precision of ±0.0005 pH over periods of several hours and exhibits stability better than 0.005 pH over periods of weeks to months (Martz et al., 2010). Considering that the Durafet-electrode temperature was slightly different from the SST due to the effects of water pumping and room temperature, a temperature-dependent coefficient of 0.0128 pH °C⁻¹ was derived from the surface waters in the cruise by varying the temperature from 15.0 to 30.0°C at a salinity of 34.6, TA of 2,268 μmol kg⁻¹, and DIC of 1,966 μmol kg⁻¹. Thus, the coefficient was used to calculate the sea surface pH (pH ^in_situ) as,

where pH^Durafet and T^Durafet are Durafet pH and temperature in the system, respectively.

Chl a was translated from underway water fluorescence and validated against field-measured Chl a (Li et al., 2022a). The underway pCO₂ measurement and calibration were described in detail by Li et al. (2022a), and the overall uncertainty of pCO₂ was less than 1%.

Discrete water samples for dissolved oxygen (DO), dissolved inorganic carbon (DIC), and total alkalinity (TA) were collected using 10-L Niskin bottles at a surface layer of ~2 m. Depth profiles of temperature and salinity (Practical Salinity Scale of 1978) were obtained using calibrated conductivity-temperature-depth/pressure (CTD) probes (SBE911 plus, Sea-Bird Scientific, USA).

Water samples for DO analyses were collected, fixed, and titrated onboard the vessel following the classic Winkler procedure (Knap et al., 1996). Any potential nitrite interference in DO titration was removed by adding 0.01% NaN₃ during subsample fixation (Wong, 2012). To quantify the effects of net community metabolism, the apparent oxygen utilization (AOU) was calculated by subtracting the field-measured DO concentration from the air-equilibrated DO concentration. Assuming that DO was initially in equilibrium with the atmosphere, an AOU >0 implies net community respiration, whereas an AOU<0 implies net community production.

Following the procedure recommended by Dickson et al. (2007), water samples for DIC and TA analyses were collected and stored in 250 mL borosilicate glass bottles. Prior to sealing with greased (Apiezon-L) ground-glass stoppers, 1 mL of seawater was removed from each sample bottle to allow for thermal expansion and 100 µL of saturated HgCl₂ was added to the water samples to halt biological activity. The samples were then stored at room temperature until further analysis. DIC was measured using an infrared CO₂ detector-based DIC analyzer (AS-C3, Apollo SciTech Inc., USA), and TA was determined at 25.0 °C by Gran’s acidimetric titration using a semi-automated titrator (AS-ALK2, Apollo SciTech Inc., USA). The reproducibility of the DIC and TA measurements was within 0.1% (Cai et al., 2004). DIC and TA measurements were referenced to certified reference materials from the laboratory of Andrew G. Dickson (Scripps Institute of Oceanography, USA), with a precision of ±2 µmol kg⁻¹.

For discrete samples, pCO₂, Ω_arag, and NBS scale pH (pH_NBS; for simplicity, ‘pH’ in the following text refers to the NBS scale) were calculated from DIC, TA, seawater temperature, salinity, and pressure values using CO2SYS.xls (version 24) (Pelletier et al., 2015), an updated version of the original CO2SYS.EXE (Lewis and Wallace, 1998). Here, Ω_arag is defined as the product of calcium (Ca²⁺) and carbonate (CO₃ ²⁻) ion concentrations divided by the apparent solubility product for aragonite (K _sp*_arag), i.e., Ω_arag = [Ca²⁺] × [CO₃ ²⁻]/K _sp*_arag. The DIC and Ω_arag values were calculated using the program from the underway pCO₂, salinity-based TA ( Figure 2A ), and other auxiliary parameters. The carbonic acid dissociation constants from Millero et al. (2006), total boron/salinity (B_T/S) from Uppström (1974), and dissociation constant of HSO₄ ⁻ from Dickson (1990) were used to calculate the carbonate system parameters. Although the carbonic acid dissociation constants of Luecker et al. (2000) and the B_T/S of Lee et al. (2010) are recommended (Woosley, 2021; Jiang et al., 2022), Li et al. (2022a) found that the measured pCO₂ in the western North Pacific agreed with the values calculated from the combination of Millero et al. (2006) and Uppström (1974). The measured pCO₂ was approximately 8 μatm lower than the calculated values when using the combination of Lueker et al. (2000) and Lee et al. (2010). During the calculation, the surface phosphate and silicate concentrations required by the program were replaced with zero. Given that surface phosphate and silicate concentrations in the surveying area are typically very low (<0.1 μmol kg⁻¹ and<1 μmol kg⁻¹, respectively) (Li et al., 2022b), ignoring these nutrients results in minor errors in pCO₂ (0.2 μatm), pH (0.0002), and Ω_arag (0.0005) values.

To eliminate the dilution and concentration effects of precipitation and evaporation on the seawater carbonate system, we normalized the water TA (NTA) and DIC (NDIC) to a uniform salinity of 35. Salinity-normalized parameters were calculated as NTA = TA × 35/salinity and NDIC = DIC × 35/salinity. The temperature normalized pCO₂ at 28.0°C was calculated by pCO₂ ^28.0°C (μatm) = pCO₂ × e^{[0.0423 × (28.0−SST)]} (Takahashi et al., 2002).

Durafet pH has a constrained uncertainty of 0.005 (Martz et al., 2010) and is therefore used as a reference to assess the quality of the underway and calculate the pH data. To further assess the quality of the underway pH/pCO₂ and discrete DIC/TA datasets, the underway pH/pCO₂ (pH^UW/pCO₂ ^UW) data collected at the stations were compared to the calculated pH/pCO₂ from the measured DIC and TA (pH^DIC-TA/pCO₂ ^DIC-TA). The Ω_arag values calculated from DIC and TA (Ω_arag ^DIC-TA) were compared with those calculated from pCO₂ ^UW and salinity-based TA (Ω_arag ^{p CO2UW-TAsss}). The results showed that pCO₂ ^UW versus pCO₂ ^DIC-TA and Ω_arag ^DIC-TA versus Ω_arag ^{p CO2UW-TAsss} agreed with each other, following a 1:1 line within ±6 μatm for pCO₂ and ±0.03 for Ω_arag (n = 66), respectively ( Figures 2B, C ). These comparisons indicate that the measured and calculated pCO₂, DIC, TA and Ω_arag results were reliable.

However, the pH^UW value was lower than that of pH^DIC-TA by 0.015 ± 0.004 (n = 66) ( Figure 2D ), which was higher than the uncertainty of Durafet pH. We found that the differences between pH^DIC-TA and pH^UW were discrete relative to SST, SSS, and Chl a ( Figures 2E–G ), but were significantly related to cruise time (day) ( Figure 2H ), indicating that the Durafet pH electrode had a time-dependent baseline drift. Therefore, we corrected pH^UW according to the linear relationship between the differences and cruise time. The results showed that the corrected pH^UW and pH^DIC-TA were consistent with each other, following a 1:1 line within ±0.004 (n = 66) ( Figure 2I ). Note that ‘pH’ in the following text refers to corrected pH.

In this study, we aimed to quantify the contribution of the controlling factors to ΔpH and ΔΩ_arag in the western North Pacific. We used a systematic approach based on first-order Taylor series deconvolution (Murata and Shu, 2012; Hagens and Middelburg, 2016) to decompose the ΔpH and ΔΩ_arag into the contributions of individual water chemistry parameter changes in temperature (ΔT), salinity (ΔS), NDIC (ΔNDIC), NTA (ΔNTA), and residual (Res). The residual term represents contributions from other acid–base systems, although these may be negligible. Therefore, we used this method to decompose ΔpH and ΔΩ_arag as follows,

where V indicates the pH and Ω_arag. ΔV was calculated from the real-time value relative to the mean value in the subtropical zone (temperature = 28.0°C, salinity = 34.4, TA = 2,260 μmol kg⁻¹ and DIC = 1,945 μmol kg⁻¹), where sea surface pH and Ω_arag were close to the air equilibrium. ΔT, ΔS, ΔNDIC, and ΔNTA were calculated based on the differences in water chemistry parameters between the real-time and mean values in the subtropical zone. On the right-hand side of equation (2), four partial derivative terms were calculated based on the observed data, assuming a 1‰ change in the relative parameters, while the other three parameters were held constant.

The spatial distributions of SST and Chl a are shown in Figures 3A, B , and the variations in SST, SSS, Chl a, and AOU are shown in Figures 4A, B . Overall, SST, Chl a, and AOU showed strong spatial variations from the subtropical zone, through the KR zone, to the KE zone ( Figures 3A, B , 4A, B ). Briefly, SST decreased from 27.93°C ± 0.99°C in the subtropical zone to 22.65°C ± 1.45°C in the KR zone and to 18.27°C ± 2.37°C in the KE zone. This strong gradient may be related to oceanic heat loss due to the northeastward flow of the Kuroshio Current (Qiu and Chen, 2011; Kitamura et al., 2016). However, SSS varied within a limited range of 34−35 in the open ocean of the surveyed area ( Figure 4A ). In the KE zone, the SSS was only slightly lower than that in the subtropical and KR zones, suggesting a limited influence of the low-salinity Oyashio current on the hydrological characteristics of the KE surface waters during the surveying cruise.

The Chl a concentration is relatively low at 0.2−0.4 mg m⁻³ in the subtropical and KR zones ( Figures 3B , 4B ) due to the limited nutrient availability that restricted phytoplankton growth (Wong et al., 2002; Yasunaka et al., 2014). However, the KE zone has a high Chl a concentration of 0.4−0.9 mg m⁻³, suggesting that biological activity may influence OA metrics in this region. Correspondingly, the AOU value of −9 ± 3 μmol kg⁻¹ in the subtropical and KR zones was slightly lower than the air–sea equilibrated value (0 μmol kg⁻¹), probably related to the low primary production. However, the KE zone had a relatively low AOU value of −26 ± 6 μmol kg⁻¹ ( Figure 4B ), indicating the effect of relatively intense biological oxygen production in the KE surface waters.

The variations in the surface NDIC and pCO₂ are shown in Figures 4C, D . The variations of surface NDIC were generally mirrored the SST ( Figure 4C ), and the NDIC value increased from 1975 ± 11 μmol kg⁻¹ in the subtropical zone, to 2,017 ± 10 μmol kg⁻¹ in the KR zone, and to 2,076 ± 7 μmol kg⁻¹ in the KE zone. The surface NDIC was close to the air equilibrium in the subtropical zone and lower than the air-equilibrated NDIC by 17 ± 7 μmol kg⁻¹ in the KR zone and 38 ± 7 μmol kg⁻¹ in the KE zone ( Figure 4C ). In contrast to the distinct variation in surface NDIC, surface TA and SSS generally followed a linear relationship ( Figure 2A ), and surface NTA showed a uniform value of 2,295 ± 4 μmol kg⁻¹ in the study area, which is consistent with the previously reported mean NTA value of 2,297 ± 5 μmol kg⁻¹ in the western North Pacific (Ono et al., 2019). This uniformity of NTA further supports the limited influence of the Oyashio current on the KE surface waters during the survey cruise, as the Oyashio water has a relatively high NTA of ~2,400 μmol kg⁻¹ (Ishii et al., 2014; Takahashi et al., 2014). Figure 4D shows that the distribution of sea surface pCO₂ in late spring is similar to that of SST, as previously described in detail by Li et al. (2022a). In addition, the temperature normalized pCO₂ at 28.0°C mirrored in situ pCO₂ but was similar to that of Chl a ( Figures 4B, D ), suggesting a relatively weak effect of biological CO₂ drawdown.

The variations in surface pH and Ω_arag are shown in Figures 3C, D , 4E, F , and they also show large environmental gradients. Surface pH variations were similar to those of NDIC but mirrored Ω_arag, SST and pCO₂. The pH (Ω_arag) increased (decreased) from 8.18 ± 0.01 (3.57 ± 0.12) in the subtropical zone to 8.21 ± 0.01 (3.25 ± 0.12) in the KR zone and to 8.24 ± 0.02 (2.95 ± 0.16) in the KE zone. Similar to NDIC and pCO₂, the surface pH and Ω_arag were close to the air equilibrium in the subtropical zone, which is consistent with the results of Ono et al. (2019). They found that the surface pH and Ω_arag in the area have been in air equilibrium annually over the past three decades. In contrast, the surface pH and Ω_arag were higher than the air-equilibrated values in the KR and KE zones ( Figures 4E, F ). Especially in the KE zone, the low surface pCO₂ and high pH corresponded to the relatively high Chl a content and negative AOU, indicating the effects of photosynthesis-induced oxygen addition, pCO₂ decrease, and pH increase. However, the surface Ω_arag is the lowest in the KE zone, which is contradictory to the photosynthesis-induced Ω_arag increase. These results indicate that photosynthesis was not the only factor controlling surface pH and Ω_arag distributions.

To reveal the mechanisms governing the distributions of sea surface pH and Ω_arag in late spring, we investigated the relationship between surface pH, Ω_arag, and SST ( Figures 5A, B ). There are significant correlations between surface pH, Ω_arag, and SST with correlation coefficients (r) of 0.84 (p<0.001) and 0.93 (p<0.001), respectively. We further divided the effect of temperature on pH and Ω_arag into internal and external effects, both of which coexist in the contemporary ocean (Jiang et al., 2019; Cai et al., 2020; Wu et al., 2021; Xue et al., 2021). The internal temperature effect is computed based on the assumption that temperature is the only variable, because it is linked to shifts in the species in the CO₂ system as the temperature varies. The external temperature effect was calculated by varying the temperature and assuming a consistent air-sea equilibrium (Cai et al., 2020). Note that, for the internal temperature effect, the magnitude of the pH is initially much larger than that of Ω_arag (Jiang et al., 2019).

The results showed that surface pH generally followed the internal temperature effect but was lower than that predicted by the internal temperature effect and higher than the air–sea equilibrated values in the KR and KE zones. This result indicates that the internal temperature dominated the latitudinal gradient in surface pH, although the air-sea gas exchange induced by temperature partially offsets the internal temperature-driven pH pattern. In contrast, the surface Ω_arag generally followed the external temperature effect (air–sea gas exchange induced by temperature) but was lower than that predicted by the internal temperature effect and higher than the air–sea equilibrated values in the KR and KE zones. This result indicates that the surface Ω_arag is mainly controlled by the external temperature effect, although the internal temperature effect enhances the external temperature-driven Ω_arag gradient.

In addition to the effect of temperature on the surface pH and Ω_arag, biological activities tend to increase the sea surface pH and Ω_arag in the KE zone (Section 3.2), probably enhancing the surface pH gradient but counteracting the surface Ω_arag gradient. In addition, entrainment of subsurface low-pH and Ω_arag water decreases the surface pH and Ω_arag in winter and spring (Ishii et al., 2001; Takahashi et al., 2002; Kim et al., 2015).

As mentioned above, the changes in surface pH (ΔpH) and Ω_arag (ΔΩ_arag) were affected by internal temperature, external temperature (air–sea gas exchange induced by temperature), biological activities, and vertical mixing. The effects of these processes on the spatial and temporal variations in pH and Ω_arag have also been previously reported by some studies (e.g., Ishii et al., 2011; Takahashi et al., 2014; Jiang et al., 2015; Wu et al., 2021; Xue et al., 2021). Here, we first decomposed ΔpH and ΔΩ_arag into the contributions of changes in individual parameters.

Decomposition provides quantitative constraints on how the various drivers of pH and Ω_arag change spatially ( Figures 6A, B ). Generally, among the different drivers, the ΔpH values were mainly attributed to ΔT, whereas the contribution of ΔNDIC to ΔpH offset approximately 61% ± 16% and 63% ± 8% of the ΔpH caused by ΔT in the KR and KE zones, respectively. With respect to ΔΩ_arag, the values were primarily controlled by the ΔNDIC, whose contributions accounted for 69% ± 9% and 73% ± 3% of ΔΩ_arag in the KR and KE zones, respectively. In addition, the contributions of ΔNDIC to ΔΩ_arag were enhanced by ΔT, with contributions to ΔΩ_arag of 34% ± 9% and 30% ± 3% in the KR and KE zones, respectively. Furthermore, the direct contributions of ΔS to ΔpH and ΔΩ_arag were negligible, while the indirect contributions of ΔS via the variation of TA can also contribute to ΔpH and ΔΩ_arag (Kuchinke et al., 2014; Kwiatkowski and Orr, 2018; Li and Zhai, 2019). However, the contributions of ΔS-induced TA changes to ΔΩ_arag were offset by ΔS-induced DIC changes because the sensitivities of pH and Ω_arag to DIC and TA are approximately equal and opposite, respectively (Cao et al., 2007; Cai et al., 2020; Xue and Cai, 2020).

We further decomposed ΔNDIC into the contributions of fundamental biogeochemical processes to ΔpH and ΔΩ_arag, e.g., the changes in biological activities (ΔNDIC^Bio) and other ΔNDIC, which were mainly associated with air–sea gas exchange and vertical mixing (ΔNDIC^ASM). Therefore, we represent the contributions using the terms: ΔV ^T = (əV/əT)ΔT, ΔV ^ΔNDIC = (əV/əNDIC)ΔNDIC, ΔV ^Bio = (ΔNDIC^Bio/ΔNDIC)ΔV ^ΔNDIC, and ΔV ^ASM = ΔV ^ΔNDIC − ΔV ^Bio. Therefore, we decompose ΔV as follows,

where ΔV ^Bio indicates the effect of biological activities. ΔNDIC^Bio was calculated as ΔNDIC^Bio = ΔAOU × 117/170, where ΔAOU was calculated from the real-time value relative to the mean value in the subtropical zone (−9 μmol kg⁻¹) and 117/170 is the C/O ratio (Anderson and Sarmiento, 1994). The discrete AOU values ( Figure 4B ) were linearly interpolated between the two stations and then adjusted to the underway time. ΔV ^ASM was calculated from the difference between ΔV ^ΔNDIC and ΔV ^Bio, indicating the effects of temperature-induced air–sea gas exchange and vertical mixing, both of which tend to decrease the surface pH and Ω_arag in the KR and KE zones. We have included the minor contributions of ΔS and ΔNTA in the residual component in Figures 6A, B .

The contributions of the fundamental biogeochemical processes to ΔpH and ΔΩ_arag are shown in Figures 6C, D . The contribution of ΔNDIC^ASM to ΔpH offsets approximately 55% ± 20% and 77% ± 12% of ΔpH caused by ΔT in the KR and KE zones, respectively. In contrast, the contribution of ΔNDIC^ASM accounts for 62% ± 15% and 87% ± 9% of ΔΩ_arag in the KR and KE zones, respectively. This result indicates that the effect of air–sea gas exchange and vertical mixing increased from the KR zone to the KE zone, probably related to the lower temperature-induced air–sea gas exchange ( Figure 1A ) and deeper mixed layer depth ( Figure 1B ). The effect of biological activities was mainly constrained in the KE zone ( Figures 6C, D ), where waters with low AOU and high Chl a levels were located ( Figure 4B ). The contribution of biological activities accounted for 36% ± 26% of the ΔpH_arag and offsets approximately 14% ± 9% of the ΔΩ in the KE zone. However, the contribution of biological activities offsets only about 17% ± 11% of the ΔNDIC^ASM-induced ΔpH_arag and 15% ± 9% of the ΔNDIC^ASM-induced ΔΩ in the KE zone, respectively. Therefore, biological activities enhanced the temperature-driven pattern in pH but counteracted that in Ω_arag. Moreover, the Chl a concentrations were highest in the late spring of the year ( Figures 1C–H ), indicating that the influence of biological activity on the distribution of pH and Ω_arag in late spring was the most notable and representative in the year. However, compared to the effect of temperature or air–sea gas exchange on surface pH and Ω_arag patterns, the degree of biological influence was relatively small, although primary production was highest in late spring ( Figure 1E ). This result was further supported by the results of Li et al. (2022a), who found that the biological influence on the distribution pattern of surface pCO₂ was relatively minor.

In summary, these quantified results are consistent with the qualitative results, e.g., SST variations dominated the ΔpH, although air–sea gas exchange and vertical mixing counteracted the temperature-driven pH pattern to a comparable extent. In contrast, ΔΩ_arag was mainly controlled by air–sea gas exchange and vertical mixing, the effects of which on the Ω_arag pattern were enhanced by temperature. However, biological activities have a limited influence on the ΔpH and ΔΩ_arag values.

We showed that the surface pH and Ω_arag were out of phase in late spring in the western North Pacific, e.g., the pH (Ω_arag) was gradually decreased (increased) southward across the KE, KR, and subtropical zones. This latitudinal distribution pattern of surface pH and Ω_arag is consistent with the climatological distribution of surface pH and Ω_arag from low to mid-latitudes but is contrary to the distribution pattern from mid to high latitudes, where pH and Ω_arag are generally in phase (Jiang et al., 2015; Jiang et al., 2019; Xue et al., 2021). In addition, we have presented the combined effects of temperature, air–sea gas exchange, biological activities, and mixing on springtime pH and Ω_arag distributions in the western North Pacific. Similarly combined effects have also been shown from East Asia to the Arctic Ocean, where Wu et al. (2021) found that biological activity counteracts the temperature-driven pattern in pH but reinforces that in Ω_arag in the Bering and Chukchi Shelf, where the region has higher temperature and higher primary production relative to the Arctic Basin. This result contradicts with our result in the KE zone, where the region has a lower temperature and higher primary production than the subtropical zone ( Figures 6C, D ). In general, both spatial and seasonal variations in pH and Ω_arag are larger in coastal oceans than in open oceans (Feely et al., 2008; Borges and Gypens, 2010; Feely et al., 2010; Gruber et al., 2012; Xu et al., 2020).

We found that temperature and its induced air–sea gas exchange fundamentally controlled pH and Ω_arag distributions from the subtropical zone to the KE zone. This is further supported by the results of Xue et al. (2021), who found that when pH is mainly controlled by the internal temperature effect (thermal), surface pH and Ω_arag tend to be out of phase because the effects of thermal and non-thermal (e.g., air–sea gas exchange, biological activities, and mixing) on pH are out of phase; however, when pH is mainly controlled by non-thermal effects, surface pH and Ω_arag will be in phase because their non-thermal effects are intrinsically in phase. Similarly, the variations in surface pH and Ω_arag and their controls from low- to mid-latitudes agree with the seasonal variations in pH and Ω_arag at mid-latitudes, which are also primarily driven by temperature and its induced air–sea gas exchange (Ishii et al., 2011; Kim et al., 2015; Fassbender et al., 2018; Kwiatkowski and Orr, 2018). Taken together, these studies suggest that the spatiotemporal changes and drivers of the pH and Ω_arag depend on the particular ocean environment.