The Indian Ocean, surrounded by India, Africa, Antarctica, and Australia, is the third largest ocean in the world. The total area of the ocean is 7.056 × 10⁷ km², and the average water depth is 3,872.4 m. The study area covers the region from 80°E to 95°E and from 15°S to 15°N in the EIO ( Figures 1A, B ). The northernmost border of the study area is the BoB, which is the world’s largest bay and is substantially discharged by rivers from India, Bangladesh, and Myanmar, including the Ganges River, Brahmaputra Rivers, Irrawaddy River, and Krishna River. The study area is affected by complicated water currents, including the Wyrtki Jets (WJ), the South Equatorial Current (SEC), the South Equatorial Counter Current (SECC), the South Java Current (SJC), the Northeast Monsoon Current (NMC) in winter, and the Southwest Monsoon Current (SMC) in summer, as shown in Figure 1C .

Two cruises were conducted aboard the R/V “Shi Yan 3” in the EIO from 27 September to 6 November 2020 (autumn) and from 28 April to 5 June 2021 (spring). The autumn cruise covered 29 grid stations and two transects, and the spring cruise comprised 31 grid stations and two transects ( Figures 1A, B ). Seawater samples were collected using 12-L Niskin bottles equipped with an SBE 911 plus a conductivity–temperature–depth (CTD) sensor (Sea-Bird Electronics, Inc., USA). Temperature and salinity data were obtained from the CTD sensor. The sampling procedures were performed immediately to minimize gas escape. Dissolved oxygen (DO), pH, DIC, and Alk were sampled sequentially using the same gas-sampling method; samples used for DIC and Alk measurements were then poisoned with 100 µL of a saturated HgCl₂ solution. DO and pH were measured on-board, while DIC and Alk samples were stored at 4°C and measured within 30 days after returning to the lab. Wind speed was measured at a height 10 m above the sea surface using the AWS 2700 Weather Station (Aanderaa Data Instruments, Norway).

DO was measured following the Winkler method (Dickson, 1994). pH was determined using a SevenCompact S210-K pH meter equipped with an Inlab Expert Pro pH electrode (Mettler Toledo, Switzerland), calibrated according to the NBS scale. Precision levels were about ±0.01. DIC was measured by an AS-D1 DIC Analyzer (Apollo SciTech, USA), and Alk was determined by the Gran titration method using a Titrator Excellence T5 equipped with a DGi111-SC combined glass pH electrode (Mettler Toledo, Switzerland). The precisions of DIC and Alk measurements were higher than 1‰ and certified reference materials (Batch #187 with DIC of 2002.85 ± 0.40 µmol kg⁻¹ and Alk of 2,204.98 ± 0.37 µmol kg⁻¹) produced by the laboratory of A.G. Dickson (Scripps Institution of Oceanography, San Diego, CA, USA) were used for calibration. pCO₂ was calculated from pH and Alk by using CO₂SYS software (version 2.1, Lewis and Wallace, 1998) with equilibrium constants proposed by Mehrbach et al. (1973) and refitted by Dickson and Millero (1987). Quality tests were run by comparing the calculated pH from Alk and DIC with its measured NBS value.

For chlorophyll a (Chl-a) analysis, 1 L of seawater was successively vacuum filtered (<0.05 MPa) through a 20-µm pore-size mesh, a 2-µm pore-size mixed cellulose ester membrane, and Whatman GF/F filters (all diameters are 25 mm), which represent microplankton, nanoplankton, and picoplankton fractions, respectively. The filters were soaked in 5 mL of 90% acetone and stored in the dark at −20°C. After 24 h, Chl-a concentration was measured using a Trilogy Laboratory Fluorometer (Turner Designs Inc., USA) according to the procedure described by Parsons et al. (1984). The seawater samples for nutrients (PO₄ ³⁻, SiO₃ ²⁻, NO₃ ⁻, NO₂ ⁻, and NH₄ ⁺) were measured using a Technicon AutoAnalyzer 3 (Bran+Luebbe, Germany) according to the method of Strickland and Parsons (1972).

The net sea–air CO₂ exchange flux (F) was calculated based on the formula:

where K _H is the solubility of CO₂ (Weiss, 1974), and ΔpCO₂ is the difference between the pCO₂ of seawater and air. Gas transfer velocity k is calculated according to the Wanninkhof (2014) empirical function of short-term wind speed; u is the field-measured wind speed at 10 m height; Sc is the Schmidt number for CO₂ (a function of water temperature) as calculated by Wanninkhof (2014).

The pCO₂ in the air was calibrated according to the atmospheric pressure and corrected to 100% humidity at in situ seawater surface temperature (T) and salinity (S) through the following equation (Dickson and Goyet, 1994):

where P _air is atmospheric pressure, P _water is saturated water vapor pressure calculated according to the Weiss and Price (1980) formula, and xCO₂(air) is set as 406.30 ppm in October 2020 and 408.13 ppm in May 2021, which are the values of the closest atmospheric CO₂ station (BKT, Bukit Kototabang, Indonesia) to the study area during the study period.

The difference between the measured values of DIC and Alk and the theoretical values of the linear regression of DIC and Alk with salinity ( Figure 2 ), expressed by ΔDIC and ΔAlk, respectively, was calculated according to following formulas:

where positive values indicate processes inputting additional DIC and Alk and negative values indicate processes consuming DIC and Alk. The absolute values of ΔDIC also represent the intensity of mixing processes; if the mixing is stronger, the measured DIC will be closer to the theoretical DIC value and thus the absolute values of ΔDIC will be lower. Large absolute values of ΔDIC indicate the effects of other processes (such as CaCO₃ precipitation–dissolution, biological processes, and sea–air exchange) besides mixing processes.

In the case that the endmember salinity in our study area was not sufficiently low to obtain the regression equations, theoretical data obtained in our study were compared with those calculated using equations from Sarma et al. (2018), in which the study area was taken as the coastal waters off Visakhapatnam (India) in the western BoB and the endmember salinity is 23. This comparison was used to confirm the reliability of the theoretical data obtained in our study.

According to the composition of ΔDIC and ΔAlk, these parameters can be expressed as:

where ΔDIC_p-d, ΔDIC_bio, and ΔDIC_a-s represent DIC changes caused by CaCO₃ precipitation–dissolution, biological, and sea–air exchange processes, respectively; ΔAlk_p-d and ΔAlk_bio represent Alk changes caused by the processes of CaCO₃ precipitation–dissolution and biological processes, respectively; and −17/106 × ΔDIC_bio is the expected Alk change during the biological use or release of DIC assuming the Redfield ratio (Redfield et al., 1963). Considering that the change in Alk from biological processes is relatively minor compared to changes in DIC, the biological term was ignored in practice according to previous studies (Xue et al., 2016; Li et al., 2022; Zhang and Hu, 2023).

Since CaCO₃ precipitation removes DIC and Alk at a 1:2 ratio, the relationship can be expressed as

In addition,

where F is the sea–air fluxes of CO₂, t is the cruise period, ρ is the seawater density, and MLD is the mixed layer depth.

Thus,

The temperature of the surface seawater of the EIO ranged from 26.02°C to 29.46°C with an average value of 28.77°C during autumn 2020, while it varied from 28.04°C to 31.64°C with a mean of 29.99°C during spring 2021. The overall temperatures during autumn 2020 were lower than those recorded during spring 2021, although a trend of decreasing temperature from north to south was detected in both seasons ( Supplementary Figure S1 ). The salinity of the surface seawater presented an opposite increasing trend from north to south during both seasons; these values were in a similar range during autumn 2020 (31.79–35.04) and spring 2021 (31.84–34.87), with averages of 34.29 and 34.04, respectively. Low salinity values emerged in the BoB due to riverine inputs.

In the upper 200 m of transect E87, the temperature generally decreased with increasing depth during both seasons ( Supplementary Figure S2 ); however, the temperature during autumn 2020 (13.02–29.44°C) was in a narrower range compared to that measured during spring 2021(12.40–31.64°C). Below 75 m, the temperatures in the northern stations were lower than those in the southern stations for the same depth during autumn 2020, whereas the opposite pattern appeared during spring 2021. Salinity during both seasons was homogeneous in the upper 200 m except at stations in the BoB. Furthermore, the riverine inputs had more extensive penetration in latitudes and depths during spring 2021. Salinity were in the ranges of 31.79–35.43 and 31.84–35.22 during autumn 2020 and spring 2021, respectively. The MLD during autumn 2020 was deeper than that during spring 2021, with averages of 66 and 34 m, respectively. Nevertheless, the MLD of the northern stations (especially in the BoB) was shallower than that of the southern stations during both seasons.

At transect EQ, the distributions of both temperature and salinity presented similar patterns in the two seasons. Specifically, the temperature decreased and the salinity increased with increasing depth in the upper 200 m ( Supplementary Figure S2 ). In addition, salinity in the eastern stations was lower than that in the western stations for the same depth above 100 m. During autumn 2020, temperature and salinity were in the ranges of 12.74–29.23°C and 34.30–35.28, respectively, whereas during spring 2021, they were in the ranges of 14.11–30.40°C and 33.98–35.32, respectively. The average MLD was 76 m for both seasons. In addition, the MLD was relatively stable along transect EQ during both seasons, except for a relatively shallow depth detected in the middle of the transect.

During autumn, the pH of the surface seawater of the EIO ranged from 8.11 to 8.23 with an average value of 8.15. The pH generally decreased from east to west; the maximum pH value occurred at the easternmost station EQ-12 and the minimum was observed near the middle of the study area (the intersection of 87°E and the Equator). The values of DIC and Alk were in the ranges of 1,964–2,197 and 2,205–2,501 µmol kg⁻¹, with averages of 2,066 and 2,350 µmol kg⁻¹, respectively. Both DIC and Alk presented higher values in the northern hemisphere than in the southern hemisphere. The maxima of DIC and Alk both occurred at the northwest station E80-13, while the minimum DIC value was observed in the southernmost station S10-03 and the lowest Alk level was observed at the southwest station E80-2. pCO₂ varied from 385 to 503 µatm, with a mean of 453 µatm. pCO₂ values decreased gradually from west to east, which was opposite to the pattern observed for pH. The highest and lowest pCO₂ occurred at the westernmost station E80-13 and the easternmost station EQ-12, respectively, which were both located near the Equator ( Figure 3 ).

The pH of the surface seawater of the EIO during spring ranged from 8.08 to 8.22 with an average value of 8.14. The pH generally decreased from north to south and presented high values in the BoB (maximum at station E87-31) and low values in the west of the study area (minimum at station E80-09). The values of DIC and Alk were in the ranges of 2,044–2,251 and 2,274–2,460 µmol kg⁻¹, with averages of 2,140 and 2,390 µmol kg⁻¹, respectively. Both DIC and Alk showed low values in the BoB (minima at station E87-30), which was again opposite to the pattern observed for pH. High values were observed in the southernmost of the study area: the maximum DIC at station S10-08 and the maximum Alk at station S10-04. pCO₂ varied from 372 to 572 µatm, with a mean of 478 µatm. pCO₂ increased from north to south with the maximum observed at station E80-09 and the minimum observed at station E87-31, which was again exactly opposite to the pH distribution pattern ( Figure 3 ).

At transect E87, the pH generally decreased from 8.23 to 7.68 with increasing depth in the upper 200 m during autumn. In addition, the pH in the north was lower than that in the south at the same depth below 50 m and the minimum pH occurred at the 200-m depth of station E87-36. Conversely, pCO₂ increased with depth and from south to north with a range of 345–1,424 µatm, which was opposite to the pattern observed for pH. DIC presented low values at the surface layer and high values below 25 m except at the 75- to 125-m depth in the southern hemisphere. DIC concentrations varied from 1,970 to 2,223 µmol kg⁻¹ with the maximum observed at the 100-m depth of station E87-30 and the minimum observed at the 5-m depth of station E87-18. Alk was generally elevated with depth but presented comparatively low values between 5°S and the Equator from the surface to 200- m depth. Alk values ranged from 2,254 to 2,509 µmol kg⁻¹, with a maximum observed at the 50-m depth of station E87-03 and the minima observed at the 5-m depth of station E87-14 and at the 25-m depth of station E87-11 ( Figure 4 , Supplementary Figure S3 ).

At transect EQ during autumn, the pH ranged from 7.79 to 8.24 and decreased from the surface to the 200-m depth with the minimum at 200 m of station E80-10. DIC varied from 1,995 to 2,251 µmol kg⁻¹ and presented low values at the 50–100 m depth range; relatively high values were observed below the 100-m depth with the maximum observed at the 150-m depth of station EQ-02. Alk showed relatively low values in the middle layers west of 87°E. The maximum value (2,516 µmol kg⁻¹) was observed at the 25-m depth of station EQ-07 and the minimum value (2,250 µmol kg⁻¹) was observed at the 75-m depth of station EQ-02. pCO₂ was in the range of 349–1,033 µatm and also presented opposite distribution patterns to those of pH, which was consistent with transect E87 ( Figure 5 , Supplementary Figure S4 ).

At transect E87 during spring, the pH generally decreased from 8.26 to 7.70 with increasing depth in the upper 200 m. There were comparatively low pH values at approximately 75- m depth of both end stations. DIC concentrations were lower in the upper 100 m compared to those measured below 100 m. Furthermore, in the upper 100 m, DIC concentrations were lower in the region north of 5°N compared to those measured south of 5°N. The maximum DIC (2,346 µmol kg⁻¹) was observed at the 150-m depth of station E87-08, while the minimum (2,040 µmol kg⁻¹) occurred at a depth of 25 m of station E87-24. Alk showed similar distributions to those of DIC. The highest Alk level (2,534 µmol kg⁻¹) appeared at the 200-m depth of station E87-08, which was close to the region of the maximal DIC, whereas the lowest Alk level (2,274 µmol kg⁻¹) emerged at the surface of station E87-30. pCO₂ increased from 336 to 1,359 µatm with increasing depth ( Figure 4 , Supplementary Figure S3 ).

At transect EQ during spring, pH and pCO₂ were in the ranges of 7.80–8.16 and 449–1,073 µatm, respectively, and presented opposite trends with respect to their vertical distributions. DIC and Alk were in the ranges of 2,048–2,306 and 2,321–2,521 µmol kg⁻¹, respectively. DIC and Alk concentrations in the upper 50-m depth were generally lower than those below 50 m and presented minima in the upper 50 m at station EQ-12 ( Figure 5 , Supplementary Figure S4 ).

The average sea–air fluxes of CO₂ in the EIO during autumn 2020 and spring 2021 were 10.00 ± 10.86 and 16.00 ± 15.60 mmol m⁻² d⁻¹, respectively, indicating that the study area serves as a CO₂ source, albeit with substantial spatial variability during different seasons ( Figure 6 ). During autumn 2020, only station EQ-12 acted as a very weak sink with CO₂ fluxes of −0.20 mmol m⁻² d⁻¹, whereas high values of sea–air CO₂ fluxes over 30 mmol m⁻² d⁻¹ occurred at stations E87-26 and E87-28. During spring 2021, sea–air CO₂ fluxes decreased from the southwest to the northeast, with the maximum value (68.94 mmol m⁻² d⁻¹) observed at station E80-01 and the minimum value (−0.53 mmol m⁻² d⁻¹) observed at station E87-31. Overall, more substantial sea–air exchange of CO₂ was detected in spring 2021 than in autumn 2020. Integrating over the area of the observations (~5 × 10¹² m²), the CO₂ emissions from the EIO were estimated to be approximately 0.22 PgC yr⁻¹ during autumn 2020 and 0.35 PgC yr⁻¹ during spring 2021.

In the surface seawater of the EIO during autumn 2020, both the maximum pH and the minimum pCO₂ appeared at station EQ-12, which was also the station presenting the maximum values of NO₃ ⁻ (5.97 µmol L⁻¹) and NH₄ ⁺ (1.49 µmol L⁻¹). The maxima of Alk and DIC emerged at the same station (EQ-07) with the minimum level of nano-sized Chl-a (0.060 µg L⁻¹). In addition, pH presented positive correlations with NO₃ ⁻ in the surface seawater of autumn 2020. These phenomena indicated that high nutrient availability (especially N nutrients) promoted the productivity and photosynthesis of phytoplankton (especially nano-sized phytoplankton), further lowering the inorganic carbon concentrations. Overall, the carbonate system parameters showed few correlations with physical and biogeochemical parameters in the surface seawater ( Table 1 ), which demonstrated that in addition to biological processes, there may be a comprehensive effect of interactions among several processes influencing the distributions of carbonate system parameters in the surface seawater of the EIO. We found a positive correlation between DIC and Alk and a negative correlation between pH and pCO₂.

In the surface seawater of the EIO during spring 2021, the maximum pH, minimum pCO₂, and maximum SiO₃ ²⁻ levels were all observed at the same station (E87-31), and the minimum pH, maximum pCO₂, and minimum of micro-sized Chl-a were also detected at the same station (E80-09). Over the whole study area, the average concentration of NO₃ ⁻ decreased by 38%, while the average concentration of SiO₃ ²⁻ increased by 36% from autumn 2020 to spring 2021. The average content of nano-sized Chl-a decreased by 24%, while the average content of micro-sized Chl-a increased by 63% from autumn 2020 to spring 2021. In addition, pH presented positive correlations with SiO₃ ²⁻ during spring 2021. These phenomena indicated a shift in corresponding nutrients utilized by the phytoplankton along with the phytoplankton species dominating the inorganic carbon concentrations from autumn 2020 to spring 2021. At the surface seawater, carbonate system parameters correlated with salinity and temperature, which indicated a stronger physical process in spring than in autumn. In addition, all carbonate system parameters showed correlations with each other, except for pH or pCO₂ with DIC ( Table 1 ).

Overall, the surface pH was lower during spring 2021 than during autumn 2020, while surface DIC, Alk, and pCO₂ levels presented higher values during spring 2021 than during autumn 2020. This pattern might reflect a seasonal or interannual fluctuation (Chakraborty et al., 2021). The low values of DIC and Alk detected in the BoB in both seasons were mainly attributed to the river inputs, which is in line with previously reported data (<30 salinity,<2,200 µmol kg⁻¹ Alk, and<1,900 µmol kg⁻¹ DIC) by Bates et al. (2006). Bates et al. (2006) also found relatively lower DIC values between the Equator and ~15°S (<2,000 µmol kg⁻¹), which is in accordance with the results of our study during autumn 2020. Low DIC values might be related to the advection processes by the WJ with relatively high salinity (>34) and temperature (>30°C) along the Equator. Wei et al. (2019) found that the WJ contributed to an increased abundance of picoplankton at the surface seawater of the Equator, which could, in turn, reduce the DIC levels through photosynthesis processes. pCO₂ presented high values near Sumatra and Sri Lanka but was higher during spring 2021 than during autumn 2020, which was driven by coastal upwelling in these areas during monsoon periods, as evidenced by the relatively low temperature and high salinity near Sumatra and Sri Lanka ( Supplementary Figure S1 ). Previous studies reported similar phenomena (Körtzinger et al., 1997; Chakraborty et al., 2018). pCO₂ levels reached over 600 μatm in the northwestern Indian Ocean (i.e., off Somalia and the Oman coast) during peak southwest monsoon periods (June to August). Chakraborty et al. (2018) developed a high-resolution state-of-the-art ocean biogeochemical model (Regional Ocean Modeling System), which showed that upwelling enhanced the pCO₂ level by 34 μatm at the sea east of Sri Lanka during the summer monsoon (May to August). As shown in Figure 2 , the correlations of DIC/Alk and salinity during spring 2021 were more significant than those during autumn 2020, which indicated that the effects of mixing processes on carbonate system parameters were enhanced during spring 2021. The wind speed was in the ranges of 1.9–14.2 m s⁻¹ and 2.3–15.3 m s⁻¹ during autumn 2020 and during spring 2021, respectively, with averages of 7.8 m s⁻¹ and 8.2 m s⁻¹, which was consistent with previous data of 0–18 m s⁻¹ obtained by different methods (Parekh et al., 2007). The average wind speed during spring 2021 was slightly higher than that detected during autumn 2020, which could also be generally conducive to intensifying the mixing processes (Sinha et al., 2020). Furthermore, wind speeds presented a positive correlation with pCO₂ and a negative correlation with pH during spring 2021, demonstrating their relatively stronger regulation effect on carbonate system parameters compared to the corresponding influence during autumn 2020.

Moreover, monsoon currents in different seasons appeared to play important roles in affecting the distributions of carbonate system parameters, although our study periods were not completely consistent with the monsoon periods: autumn 2020 and spring 2021 correspond to the very early stage of the winter and summer monsoon period, respectively. The currents flowed in different directions during different monsoon periods: the NMC and SJC flowed out of our study area during the winter monsoon, whereas the SMC and SJC flowed into our study area during the summer monsoon, with the latter causing upwelling, thereby giving rise to high productivity and phytoplankton blooms, which also brought influxes of nutrients and DIC from the deep waters. Meanwhile, as higher trophic-level species have evolved behavioral responses to these seasonally changing conditions, their respiration process also results in the production of more CO₂ (Hood et al., 2017; Vinayachandran et al., 2021), which can explain the detection of lower pH and higher DIC, Alk, and pCO₂ values during spring 2021 compared to those in autumn 2020.

In the vertical profiles evaluated during autumn, pH presented strong negative correlations with salinity and nutrients (including NO₃ ⁻, PO₄ ³⁻, and SiO₃ ²⁻) and positive correlations with temperature, Chl-a (including nano-sized, pico-sized, and total Chl-a), and DO. However, the correlations of DIC and pCO₂ with the above parameters were in the opposite directions ( Table 1 ). The relationships of carbonate system parameters with temperature, salinity, and nutrients were mainly attributed to physical processes. Specifically, the temperature decreased, while the salinity and nutrient contents increased with depth mainly due to the pronounced stratification ( Supplementary Figure S2 ). Moreover, the correlations between carbonate system parameters and Chl-a demonstrated that the photosynthesis of phytoplankton, especially small-sized phytoplankton, dominated in the EIO. Furthermore, these phenomena validated that despite the high nutrient contents in deeper waters, inorganic carbon could not be removed without the contributions of phytoplankton. All carbonate system parameters showed correlations except for pH with Alk along transect E87 and pH/pCO₂ with Alk along transect EQ.

Carbonate system parameters along vertical profiles showed stronger correlations during spring 2021 than during autumn 2020. In addition to its correlations with pH, DIC, and pCO₂, Alk also exhibited associations with temperature, salinity, nutrients, Chl-a, and DO. All carbonate system parameters showed significant correlations in the vertical profiles during spring 2021 ( Table 1 ).

Overall, the concentrations of carbonate system parameters along transects E87 and EQ were slightly higher during spring 2021 than during autumn 2020. Along transect E87, pCO₂ decreased from north to south during both autumn 2020 and spring 2021. DIC and Alk showed different distribution patterns in different seasons but always presented relatively low values at the depth of chlorophyll maximum (~75 m), reflecting the importance of biological processes in regulating the distributions of carbonate species. At low latitudes south of the Equator, a large volume of Pacific Ocean water penetrates the Indian Ocean via the Indonesian Throughflow in the upper 300-m depth (Goyet and Touratier, 2009). This water, characterized by relatively high concentrations of anthropogenic CO₂, led to high values of DIC in the upper 200 m in the southern hemisphere of our study area near Indonesia. The MLD was consistently below 100 m along transects E87 and EQ during both seasons, and all carbonate system parameters generally showed minimal fluctuation above the MLD. At transect E87, the MLD in the BoB during spring 2021 was shallower than that detected during autumn 2020, which might be due to the increased river inputs during spring 2021 that further affected the distributions of carbonate system parameters. A previous model-based study in the BoB also found a prominent impact of river inputs on the MLD (Jana et al., 2015). The MLD is generally shallower with the addition of river input, which is consistent with the results of our study. The MLD at transect EQ was deeper than that at transect E87 during both seasons, which might be attributed to the influence of currents (especially the development of the WJ during the monsoon transitions) near the Equator (Masson et al., 2002; Chi et al., 2021). In addition, the increase of MLD in the region south of 10°S was determined to be driven by the climatological winds, which is consistent with the findings of Keerthi et al. (2016). The carbonate system parameters fluctuated distinctly in the thermocline below the MLD. However, since our data were only collected in the upper 200-m depth and did not cover the entire thermocline, more samples will be collected at greater depths to comprehensively explore the effects of the thermocline on the carbonate system parameters in the future.

In both surface seawater and vertical profiles, physical, chemical, and biological factors all affected the carbonate system. The physical characteristics could influence the distributions of carbonate system parameters directly, and they could also change the dominating nutrients and phytoplankton species in different seasons and ultimately lead to variations on carbonate system parameters.

As mentioned above, the distributions of carbonate system parameters might be comprehensively controlled by several interacting processes rather than by only one process. Therefore, we separated the effects of different potential controlling processes of DIC using the method described in Section 2.5. As shown in Figure 7 , the average ΔDIC was −27 µmol kg⁻¹ during autumn 2020, which meant the measured DIC was lower than the theoretical DIC by 27 µmol kg⁻¹. The ΔDIC could be attributed to CaCO₃ precipitation–dissolution, sea–air exchange, and biological processes by −11, −6, and −10 µmol kg⁻¹, respectively; in other words, the effect of CaCO₃ precipitation decreased the DIC by 11 µmol kg⁻¹, sea–air exchange brought 6 µmol kg⁻¹ CO₂ out of the seawater, and photosynthesis consumed 11 µmol kg⁻¹ of DIC. However, during spring 2021, the average ΔDIC, ΔDIC_p-d, ΔDIC_a-s, and ΔDIC_bio values were −14, −1, −11, and −2 µmol kg⁻¹, respectively. The contributions of CaCO₃ precipitation and photosynthesis accounted for 41% and 37% of ΔDIC during autumn 2020, while ΔDIC was mainly attributed to sea–air exchange (79%) during spring 2021. Compared to those measured for autumn 2020, the absolute values of ΔDIC, ΔDIC_p-d, and ΔDIC_bio were lower in spring 2021, although they were still negative, which indicated a strong mixing process and weak CaCO₃ precipitation and photosynthesis processes occurring during spring 2021. Sulpis et al. (2021) reported that the ΔDIC_p-d was −12.5 µmol kg⁻¹ at the surface seawater in the subtropical Atlantic Ocean and that the subtropical gyre regions of the Pacific, Atlantic, and Indian Oceans showed generally similar patterns. Sabine et al. (2002) determined that the ΔDIC_p-d at the surface seawater of our study area ranged from −17.5 to 0 µmol kg⁻¹, which is consistent with the present findings. Sulpis et al. (2021) also pointed out that global ocean CaCO₃ surface production is mainly derived from biotic (planktonic) precipitation (Milliman, 1993; Milliman et al., 1999; Gehlen et al., 2007). Therefore, biological processes were the main contributors to ΔDIC as most of the CaCO₃ precipitation could be related to biological processes. The overall weak photosynthesis processes during spring 2021 were likely due to the heterotrophy of the BoB. Indeed, Wiggert et al. (2013) found that the increased river inputs of the summer monsoon drove the BoB toward heterotrophy and we also observed such increased river inputs in our study during spring 2021 ( Supplementary Figure S2 ). The inefficient CaCO₃ precipitation processes during spring 2021 were likely due to the lower pH and the corresponding low biological activity. Nevertheless, the sea–air exchange process was enhanced due to the higher wind speed and pCO₂ during spring 2021. In addition, the average ΔDIC caused by CaCO₃ precipitation–dissolution and biological processes exhibited a similar magnitude during both autumn 2020 and spring 2021, although the variation ranges were greater in autumn 2020, suggesting more fluctuation during autumn 2020. In general, during both autumn 2020 and spring 2021, the CaCO₃ precipitation process was stronger in the southern hemisphere than in the northern hemisphere, the sea–air exchange process outgassed CO₂ continuously with the maxima found at the westernmost region, and intense respiration typically occurred north of transect E87. Specifically, the maximum positive ΔDIC values occurred in the BoB where the biological (respiration) process dominated, whereas the maximum negative ΔDIC values emerged at the south of transect E80 during both autumn 2020 and spring 2021. The controlling processes at transect S10 changed completely from autumn 2020 to spring 2021; that is, the dominance of photosynthesis processes during autumn 2020 shifted to a dominance of respiration processes during spring 2021. This shift of biological process was due to the lower phytoplankton abundance (<100 cells L⁻¹; Wang, 2022) and weaker solar radiation (according to the collected PAR data: https://hermes.acri.fr/index.php) during spring 2021.

In summary, biological processes (including photosynthesis and biotic precipitation) dominated the surface DIC during autumn 2020, whereas sea–air exchange was the main contributor during spring 2021. Compared with autumn 2020, the mixing process intensified, CaCO₃ precipitation reduced, sea–air exchange enhanced, and photosynthesis weakened during spring 2021.

The Indian Ocean north of 18°S is considered to serve as a net source of CO₂, with an estimated annual flux of CO₂ across 14°N–14°S of the Indian Ocean of 0.18 PgC yr⁻¹ (Takahashi et al., 2002; Sarma et al., 2023b). However, the sea–air fluxes measured in our study were slightly higher than those reported previously, which was likely due to the high instantaneous wind speeds (>13 m s⁻¹) at some stations and the slight difference in the location of the study area (15°N–15°S of the Indian Ocean). According to Sarma et al. (2023b), the CO₂ fluxes in the equatorial Indian Ocean displayed seasonality with high fluxes occurring from January to May and low fluxes occurring from June to October. In addition, they found a prominent trend of seasonality in the BoB characterized by higher CO₂ fluxes from May to August associated with monsoon mixing and lower fluxes from October to December due to river discharge and stratification (Sarma et al., 2018), which is in line with our finding of higher fluxes in May 2021 than in October 2020.

Table 2 provides a summary of the key data of surface carbonate system parameters in different areas of the entire Indian Ocean. We were only able to collect limited observational data, whereas the other data were based on model output and interpolation from the limited observational results; notably, the focus of these studies is on the accuracy of the model and the sea–air fluxes of CO₂ rather than all carbonate system parameters. For consistency of comparison, we converted our data to the total hydrogen scale using CO₂SYS software, and the pH in our study (average: 7.97 during autumn 2020) would be a little lower than that reported by Xue et al. (2014), which might be related to the smaller area examined and that pH was spatially averaged and seasonally corrected in their study. The DIC and Alk values measured in our study area were generally in the same ranges as those reported in other areas of the Indian Ocean. The pCO₂ levels in our study were in the same range as those reported in the Arabian Sea but higher than those reported in the south Indian Ocean, which was attributed to the fact that the south Indian Ocean serves as a stronger CO₂ sink than the north Indian Ocean. Even though the EIO served as a CO₂ source during our sampling period, it might transition into a CO₂ sink during monsoon periods (Bates et al., 2006). Furthermore, other areas in the Indian Ocean, especially the south Indian Ocean, play significant roles in the uptake, storage, and transportation of anthropogenic carbon. In addition, since the biological productivity in the Indian Ocean accounts for 15%–20% of global ocean productivity (Behrenfeld and Falkowski, 1997), the effects of a biological pump on the ocean carbon cycle should not be ignored. Therefore, more long-term and continuous observations in the Indian Ocean should be carried out to better explore the interannual variations of the carbonate system, the influences of the El Niño-Southern Oscillation and Indian Ocean dipole, the tendency of ocean acidification, the ability to absorb anthropogenic CO₂, and other related factors.