During boreal winter (December through February), the NEM winds prevail in the WTIO driving cyclonic and downwelling-favorable circulations (Schott et al., 2009; Hood et al., 2017). The WTIO features the Seychelles–Chagos Thermocline Ridge (SCTR) that lies in the band of 5°–10°S, 55°–75°E ( Figure 1 ; Vinayachandran et al., 2021). The SCTR is an important upwelling region that affects sea surface temperature, nutrients, and primary production (George et al., 2013; Vinayachandran et al., 2021). The South Equatorial Countercurrent and South Equatorial Current flow horizontally around the SCTR (Wiggert et al., 2006; Hood et al., 2017). The eastward-flowing South Equatorial Countercurrent is supplied by the southward-flowing Somali Current and the northward-flowing East African Coastal Current in a confluence zone from 2°–4°S, while the westward-flowing South Equatorial Current is largely supplied by the Indonesian Throughflow (Schott et al., 2009). Additionally, the WTIO is perennially subject to nutrient constraints, particularly nitrogen limitation during the NEM (Wiggert et al., 2006).

The data used in this study were collected under the major Global Change and Air–Sea Interaction Program. The survey of the study area (3°N–10°S, 61°–65°E; Figure 1 ) was conducted onboard the R/V Xiang Yang Hong NO.01 between December 21, 2021 and January 11, 2022. For dimethylated sulfur compounds, samples from the surface and the depth of subsurface chlorophyll maxima (SCM, 25–75 m) were collected at all stations (43 stations, red dots in Figure 1 ), and samples from the top 300 m (5, 30, 75, 100, 150, and 300 m) were collected at representative stations (red squares in Figure 1 ), for a total of 150 samples (for single compound). For nutrients, samples from the top 300 m (including SCM layer) with a total of 255 were collected at 37 stations (not collected at the DM2-2, DM2-4, DM2-6, DM2-8, DM2-10 and DM2-13). For Chl a, samples from the top 200 m (5, 30, 50, 75, 100, 150, and 200 m and SCM layer) were collected at 42 stations (not collected at the DM3-12), and totaled up to 327 samples. Hydrographic variables (including temperature, salinity, and depth) were recorded using a Sea-Bird Scientific CTD sensor. A General Oceanics rosette sampler attached to the CTD sensor and equipped with 10-L Niskin bottles was used for all water sampling.

Water samples for analysis of DMS, DMSP, and DMSO were decanted into 125 mL polyethylene bottles from the Niskin bottles, which were allowed to overflow to prevent bubble entrainment in filling the bottles. In situ DMSPd samples were collected onboard using the nonperturbing small-volume gravity drip filtration procedure (Kiene and Slezak, 2006) and kept in 15 mL centrifuge tubes amended with 120 µL of 50% HCl. For total DMSP (DMSPt), 10 mL of the unfiltered water samples were transferred directly via pipette into 15 mL centrifuge tubes amended with 300 µL of 50% HCl. The difference between DMSPt and DMSPd concentrations was calculated to estimate the particulate DMSP (DMSPp) concentration. Other than for the addition of HCl, dissolved DMSO (DMSOd) was sampled using a procedure similar to that adopted for DMSPd, and then stored at −20°C.

Seawater concentration of DMS was measured onboard within 2 h of collection, and determined through procedures using purging, cryotrapping, and gas chromatography with a flame photometric detector (Kiene and Service, 1991). A subsample (>2 mL) was withdrawn from each sampling bottle via a Teflon^® tube attached to a glass-barreled syringe and filtered by a 25 mm glass fiber filter (Whatman: 0.7 µm). The volume in the syringe was carefully adjusted to 2 or 5 mL and then the sample was introduced via a gray septum into the gas-stripping system. DMSPd and DMSPt in the 2 mL subsamples were converted into DMS using 600 µL of 5 M NaOH (sodium hydroxide) after sparging for endogenous DMS with dry nitrogen. A 2 mL subsample pipetted from the DMSOd samples after sparging was added to 600 µL of 15–20% w/w TiCl₃ (titanium trichloride) and then heated at 50°C for 1 h in a water bath, thereby converting to DMS (Kiene and Gerard, 1994).

Instantaneous DMS sea–air flux calculation obeys the Henry law, and is the product of the DMS surface concentration and the transfer velocity (k_L). The calculation from Nightingale et al. (2000) was chosen, because of its best fit. k_L has a relationship with wind speed (U) and the temperature-dependent Schmidt number (Sc) given by Saltzman et al. (1993).

Where T = sea surface temperature (5 m). U (1 min) were obtained from the meteorological observing equipment onboard.

Chl a collections were determined by filtering seawater through glass fiber filters (Whatman: 0.7 µm). The pigments collected on the filter were extracted with 90% acetone and quantified with a Trilogy fluorimeter. Nutrient analysis was conducted using a QuAAtro continuous flow analyzer for ammonium ( NH 4 + ), nitrate + nitrite ( NO 3 − + NO 2 − ), silicate ( SiO 3 2 − ), and phosphate ( PO 4 3 − ). The mixed layer depth (MLD) was defined as increase in density of 0.125 kg m⁻³ with respect to that at surface. A few density data from the CTD were loss at the stations, where MLD was determined by a temperature drop of 0.5 °C (Shenoy et al., 2002).

In this study, surface and vertical profiles were produced using Ocean Data View 4 with the DIVA gridded field data display style selected. Sigma Plot 12.5 was used to complete line/scatter charts and histograms. The Spearman correlation method (IBM SPSS Statistics 26) was used to describe the correlations between DMS, DMSP (DMSPd + DMSPp), DMSOd, and environmental variables.

The physical and biochemical characteristics of the water column during the sampling period were typical of boreal winter conditions in the WTIO. Sea surface temperature was high (>29.0°C) in the south but low (<29.0°C) in the north, whereas surface salinity was low (<35.4) in the south but high (>35.4) in the north ( Figures 2A, B ). The low-salinity region (<35.1) to the south of 5°S was attributable to the inflow of the westward-flowing South Equatorial Current and precipitation ( Figure 2B ; Schott et al., 2009; Hood et al., 2017). Surface Chl a concentration was low (<0.15 μg L⁻¹) in the central area but high (>0.15 μg L⁻¹) in the four corners ( Figure 2C ). Surface nitrate concentration was very low (<0.5 µmol L⁻¹) in the studied area except at a very few stations ( Figure 2D ). Surface DMS concentration in the WTIO ranged from 0.5–9.5 nM (average: 4.1 ± 2.8 nM) during the NEM. Higher DMS concentrations (5.0–9.5 nM) were observed in the SCTR region, but much lower values were observed in other areas ( Figure 2E ). DMSPd concentrations varied from 0.7–6.3 (average: 1.7 ± 1.1 nM) with patches of high values (>4.0 nM) in the regions of 5°S, 61°E and 1°N, 65°E ( Figure 2F ). Surface DMSPp concentration ranged from 2.8–22.7 nM with an average of 12.1 ± 4.5 nM, i.e., an order of magnitude higher than that of DMSPd. DMSPp maxima distributed at the northeast of the study area, corresponding to elevated DMS and DMSPd ( Figure 2G ). DMSOd concentrations ranged from 0.5–3.6 nM (average: 1.4 ± 0.7 nM) at the surface with relatively high values in the SCTR region ( Figure 2H ).

The vertical profiles of temperature, salinity, Chl a, NO 3 − , and dimethylated sulfur compounds along the 65°E transect during the NEM are shown in Figure 3 . In the upper 300 m, temperature and salinity ranged from 10.5–32.3°C (average: 16.5 ± 5.3°C) and 34.6–35.6 (average: 35.1 ± 0.1), respectively. The thermocline was comparatively shallow in the SCTR region (5°–10°S) indicating local upwelling ( Figure 3A ; Schott et al., 2009; Vinayachandran et al., 2021). The low-salinity water mass extended to depth of ~50 m and became weak northward ( Figure 3B ). In contrast to the area of the SCTR, the region between 2°–4°S showed a weak dowelling current with a deepening thermocline ( Figure 3A ; Schott et al., 2009). The calculated MLD ranged from 6–58 m with the shallowest value in the SCTR region, corresponding to the shallower thermocline there. The SCM layer occurred at depths of 30–75 m with Chl a concentrations in the range of 0.15–0.99 μg L⁻¹ ( Figure 3C ). The band of SCM extended vertically to surface at the northernmost and southernmost stations probably due to vertical mixing, leading to increases in surface Chl a ( Figures 2C , 3C ). Nitrate concentrations were close to or lower than the detectable limit in the upper mixed layer but increased sharply in the thermocline ( Figure 3D ). DMS concentrations were high in the top 50 m of water but decreased sharply to a value of<1.0 nM below 75 m ( Figure 3E ). Surface DMS concentrations were >5 nM at most stations except those in the deep thermocline region. Most DMSP occurred as particulates, accounting for 87% of DMSPt, and the DMSPp maxima were nearly 6 times higher than those of DMSPd. Generally, the concentrations of both DMSPd and DMSPp were relatively low in the surface water, reached their highest levels in the subsurface (30–75 m), and then decreased to a level of less than or around the detection limits at depths below 100 m ( Figures 3F, G ). Similar to DMS, both DMSPd and DMSPp showed higher concentrations in the SCTR than in the downwelling region. Different from DMS and DMSP, high concentrations of DMSOd were observed in the surface (subsurface) water in the south and north (middle) of the transect ( Figure 3H ). High values of DMSOd were also observed below the depth of 100 m at the southernmost end of the transect.

In this study, bacterial and photochemical degradation incubations of DMS were completed at three representative stations: DM3-4 (southernmost), DM3-8 (middle), and DM3-14 (northernmost), which represented the SCTR, the downwelling area (2°–4°S), and region north of the equator, respectively ( Figure 1 ).

The microbial DMS consumption rate (8.7 nmol L⁻¹ d⁻¹) at DM3-14 was 4 times higher than that at the other two stations (3.4 and 1.1 nmol L⁻¹ d⁻¹ at DM3-4 and DM3-8, respectively; Figures 4A–C ). The highest photolysis rate at DM3-4 (2.8 nmol L⁻¹ d⁻¹) was much higher than that at both DM3-8 (0.03 nmol L⁻¹ d⁻¹) and DM3-14 (1.5 nmol L⁻¹ d⁻¹) ( Figures 4D–F ). Measured DMSOd concentrations with time during incubation were not significantly different between the natural light (experimental set) and dark (control set) conditions at DM3-8 and DM3-14, whereas a higher DMSOd production rate (1.9 nmol L⁻¹ d⁻¹) was observed at DM3-4 ( Figures 4D–F ). The highest ventilation rate at DM3-4 (1.0 nmol L⁻¹ d⁻¹) was twice that at DM3-14 (0.5 nmol L⁻¹ d⁻¹) and 5 times higher than that at DM3-8 (0.2 nmol L⁻¹ d⁻¹).

The DMS consumption rate constants of the three removal pathways, independent of DMS concentration, are listed in Table 1 . The microbial DMS consumption rate constant at DM3-14 was 5 times higher than that at both DM3-4 and DM3-8. The DMS photolysis rate constants were approximately the same at DM3-4 and DM3-14, while DM3-8 had a minor rate constant. A characteristic feature of DMS removal at DM3-4 was that microbial DMS consumption (47.1%) was comparable with photo-oxidation, accounting for 38.7% of the total. The rate constants of ventilation were the same at the three stations with a mean value of 0.1 d⁻¹. Overall, microbial consumption was the dominant pathway of DMS removal, accounting for 74.4% of the total, while photooxidation and ventilation merely contributed 17.7% and 7.9%, respectively.

The microbial DMSPd consumption rates determined through measurement of the control and GBT-added groups, for the three representative stations with the same locations as those considered in the incubation of DMS removal, are listed in Table 2 . The average DMSPd consumption rate was 3.1 ± 1.1 nmol L⁻¹ d⁻¹ during the NEM. The maximum consumption rate occurred at DM3-4, which was almost twice that at both DM3-8 and DM3-14. Correspondingly, DM3-4 exhibited a moderate initial DMSPd concentration and the highest temperature. The lowest microbial DMSPd degradation rate was found at DM3-8 due to the lowest DMSPd concentration. However, the DMS gross production rates were larger than the rates of DMSPd consumption, especially at DM3-14. This implies that surface DMS might reflect the involvement of sources via other pathways, more than that associated with DMSPd turnover.

The calculated DMS flux varied from 0.5–34.9 μmol m⁻² d⁻¹ (average: 6.4 ± 8.0 μmol m⁻² d⁻¹) during sampling. Lower sea-to-air flux of DMS was observed in the north of 5°S, along with lower DMS surface concentration, although the wind speed was comparable to that in the SCTR region with the high average DMS flux ( Figure 5 ). The significantly positive correlation between DMS seawater concentration and sea-air flux (R = 0.557, n = 43, p< 0.01) was found in this study, demonstrating that DMS concentration was an important cause of DMS sea-to-air emission. Despite high DMS concentrations appearing in the SCTR region, DMS fluxes were lower than 8 μmol m⁻² d⁻¹ at the stations of DM3-1–DM3-4, where wind speeds were weaker than those at other stations (DM1-3–DM1-5 and DM2-2–DM2-3). Thus, the impact of wind speeds on DMS ventilation from ocean to atmosphere could not be negligible. In the whole study region, DMS sea-air fluxes were positively correlated with wind speeds (R = 0.651, n = 43, p< 0.01), suggesting that wind speed may serve as a primary factor controlling the differences in DMS sea-air exchange. The high values (>29 μmol m⁻² d⁻¹) occurring in the SCTR region were due to the coupled high wind speeds (>7 m s^-1) and DMS concentrations (>6 nM) ( Figure 5 ). Therefore, the variations of DMS flux in the WTIO were attributed to the combined influence of wind speed and DMS concentration during winter monsoon.

The WTIO is a highly dynamic region where upwelling, downwelling and perennial equatorial currents coexist, which are influenced by the monsoons and southeasterly trade winds. The upwelling of the SCTR, which is driven by upward Ekman pumping and results in a shallow thermocline and nitracline, is active in both boreal winter and boreal summer (Yokoi et al., 2008; Schott et al., 2009). The low-salinity South Equatorial Current, centered at 10°S, is associated with highly stable stratification of the SCTR during the NEM (George et al., 2013). The properties of water masses might influence the marine chemical and biological processes, with implications for DMS production and emission.

Surface Chl a concentration (<0.1 μg L⁻¹) was low because the depth of the nitracline (14–82 m) was deeper than the MLD (6–24 m) at most stations, including those in the SCTR region, where stratification impeded the supply of nutrients to the surface (George et al., 2013; Lee et al., 2017). Consequently, phytoplankton in the oligotrophic water tended to live near or in the thermocline and nitracline (phytoplankton adaptive strategies; Kaiser et al., 2011; George et al., 2013), where nutrients are sufficient and natural light can penetrate ( Figures 3A, C, D ). The upwelling in the SCTR region uplifted the thermocline and nitracline, leading to a shallower SCM depth centered at ~45 m, whereas downwelling deepened the SCM band to ~75 m in the region of 2°–4°S ( Figure 3C ).

Because DMSPp production was derived directly from phytoplankton growth, the variation of the DMSPp maximum band was consistent with that of the SCM depth. The dominance of small (nano- and pico-sized) phytoplankton (55–93%; Tripathy et al., 2020) with high DMSPp/Chl a ratios (average: 175.6 ± 264.1) implied that high DMSP-content prymnesiophytes (Phaeocystis sp.) might be the dominant DMSP-producer in the WTIO (Keller et al., 1989; Herr et al., 2019; Tripathy et al., 2020). Phytoplankton release of intracellular DMSP into seawater through excretion, senescence, zooplankton grazing, and viral attack, formed the DMSPd-maximum bands at the SCM depth (Kiene et al., 2000; Boden and Hutt, 2019).

The upwelling effect was constrained by the fresh water from the eastward-flowing South Equatorial Current (George et al., 2013), and the regions with higher productivity and DMSP concentration were mainly concentrated in the subsurface rather than at the surface. However, the volatile DMS in the seawater was readily brought upward by Ekman pumping. Subsurface DMS produced from DMSP degradation via bacteria and/or algae (Kiene et al., 2000; Simó et al., 2000; Bullock et al., 2017) was moved upward by upwelling into the surface mixed layer (Bailey et al., 2008), and therefore the DMS peaks in the upper 30 m were shallower than those of Chl a and DMSP in the SCTR ( Figure 3E ). In contrast to the SCTR, downwelling decreased subsurface DMSP and DMS production, leading to the lowest concentration of surface DMS in the region of 2°–4°S. To the north of 1°S, the concentration of surface DMS increased with elevated levels of DMSP where vertical mixing extended the distribution of the SCM band owing to the deepened MLD (29~58 m) (George et al., 2013). There were also other complex factors linked to the uplift of the DMS maximum layer. For instance, intracellular DMSP content differs between phytoplankton species (Jean et al., 2005), and it is regulated by changes in salinity, temperature, and nutrients (Sunda et al., 2002).

The DMS concentration in the surface mixed layer was elevated by promotion of microbial DMSPd consumption in the SCTR ( Table 2 ). Moreover, the thinnest MLD (6 m) was centered at 6°S, where the DMS yield from DMSP was believed to be high (MLD< 10 m; Simó and Pedrós-Alió, 1999). The average DMSPp/DMS ratio (4.3 ± 4.2) was greater than the DMSPd/DMS ratio (0.7 ± 0.7), suggesting that DMS production was contributed not only by DMSPd degradation but also by intracellular DMSPp cleavage within the highly active DMSP-lyase algae in this study (Keller et al., 1989; Simó et al., 2000; Galí and Simó, 2015). In the northeast of the study area (2°–3°N, 65°E), the DMS gross production rate (9.8 nmol L⁻¹ d⁻¹) was approximately 4 times higher than the rate of DMSPd microbial consumption (2.6 nmol L⁻¹ d⁻¹), implying that DMSPp lysis by phytoplankton was a main pathway of DMS production (Galí and Simó, 2015). Meanwhile, high salinity (>35.4) in this region stimulated phytoplankton to produce intracellular DMSP and DMS against environmental stress (Sunda et al., 2002). Additionally, DMSOd reduction was not a significant process for DMS accumulation, as suggested by the undetectable level of DMS at depths below 100 m at 10°S (Dixon et al., 2020).

Accumulation of DMS in the water column depends on the state of complex dynamic equilibrium between DMS production and consumption (Galí and Simó, 2015). In accord with previous studies (Kiene and Linn, 2000; Vila-Costa et al., 2008; Herr et al., 2019), microbial consumption was found to be the dominant pathway of DMS removal (74.4%) in the WTIO, followed by photodegradation (17.7%), and ventilation (7.9%). This finding demonstrated that photolysis and ventilation were less important and that bacterial degradation was the main sink of surface DMS in the WTIO.

The rates of DMS removal were not distributed uniformly throughout the WTIO. The microbial consumption rate in the area north of the equator was 5 times higher than that in other regions, corresponding with the highest DMS gross production, and indicating that it played a more important role in DMS removal. In the SCTR region, the shallow stratification of the water column led to slower microbial DMS consumption (47.1%) through enhanced inhibition of bacterial activities by ultraviolet radiation, but an increased level of importance of photooxidation (38.7%) owing to the stronger light intensity (Sunda et al., 2002; Simó, 2004; Toole et al., 2006; Vila-Costa et al., 2008). Thus, physical conditions were probably important factors in the spatial differences in the rate of DMS consumption in the WTIO.

In addition to DMS concentration and light intensity, the DMS photolysis process might be associated with secondary photosensitizers such as chromophoric dissolved organic matter and NO 3 − (Brimblecombe and Shooter, 1986; Toole et al., 2004; Bouillon and Miller, 2005). Low DMS photochemical consumption seemed to be attributed to low NO 3 − concentration (0.10 ± 0.18 µmol L⁻¹) at the surface (Toole et al., 2004; Bouillon and Miller, 2005). However, the positive correlation between DMS and NO 3 − (R = 0.444, n = 36, p< 0.01) at the surface indicated that nitrogen deficiency tended to limit the growth of DMSP producers and stimulate the antioxidation of DMSP and DMS (Sunda et al., 2002; Herr et al., 2019). Low NO 3 − concentration and strong solar light could probably act as oxidative stresses to promote DMS accumulation instead of removal through photooxidation (Herr et al., 2019). This low DMS photooxidation with limitation of DMS microbial consumption by intensified light led to the low DMSOd pool at the surface (Del Valle et al., 2007; Del Valle et al., 2007; Schafer et al., 2009).

DMS ventilation (14.3%) in the SCTR was an order of magnitude higher than that in the region to the north of the equator (7.9%), suggesting that more DMS in the SCTR should be released to the atmosphere. However, DMS ventilation was weakened in the area to the north of the equator ( Table 1 ), where DMS microbial consumption increased with increased bacterial activities in the thicker mixed layer (Toole et al., 2006). DMS sea–air flux is closely correlated with wind speed and the DMS concentration in seawater that is affected by MLD ( Figure 5 ; Nightingale et al., 2000; Marandino et al., 2009). The fact that DMS sea–air flux is wind-speed dependent suggests that high DMS concentration in seawater alone cannot result in elevated DMS sea–air flux with weak winds. Hence, owing to the strong wind speed (>6 m s⁻¹), the SCTR can release large amounts of DMS to the atmosphere. The lowest DMS flux near the equator resulted from downwelling and low DMS production.

Shenoy et al. (2002) and Zavarsky et al. (2018) also reported surface DMS concentrations and DMS sea–air fluxes in our study area or adjacent regions, as shown in Table 3 . In the NEM of 1998, Shenoy et al. (2002) found the lowest DMS concentration in the region of 7°–15°S, whereas in the following winter, the concentration in this region was higher. Compared with our study, their sampling date was one month later and the region was further south. Their calculated DMS sea–air flux in the 1999 was lower than that of our study, although the wind speed in both cases was similar. This is because the k_L calculation (linear equation) from Shenoy et al. (2002) decreased the DMS flux, whereas our study employed the k_L from Nightingale et al. (2000) for the DMS flux calculation. In a region similar to that of our study, Zavarsky et al. (2018) obtained a similar estimate of DMS sea–air flux via eddy covariance measurements but a lower DMS concentration in seawater during the SWM. However, the sampling sites used by Zavarsky et al. (2018) covered less of the SCTR region than that sampled in our study.

Averaged DMS concentration and DMS sea–air flux in the SCTR showed high levels in the Indian Ocean during the NEM ( Table 3 ) because of the higher DMSP concentration and stronger winds (Shenoy and Kumar, 2007). Seawater DMS in the SCTR showed an upper-level concentration during boreal winter in the tropics (Belviso et al., 2004). DMS sea–air flux of the SCTR was an order of magnitude higher than that of the western tropical Atlantic Ocean, corresponding to the lower DMS concentration in seawater in May and September ( Table 3 ). The eastern tropical Pacific region also showed upwelling currents in January, where the measured DMS flux was higher than that in the SCTR ( Table 3 ), although the sampling process and calculation method used to obtain the former result differed from those used to derive the result for the SCTR. However, the western Pacific region including coral reefs showed lower DMS flux than that of the SCTR during boreal winter owing to low DMS seawater concentration ( Table 3 ).

By combining previous results with those of our study, and through comparison with other tropical regions, we propose that the SCTR region is an active region for DMS during both the NEM and the SWM. However, additional studies on DMS in the SCTR region should be conducted in spring and fall, i.e., the period of intermonsoonal winds, to elucidate the seasonality of DMS and establish the physical–biological drivers.