The MISO (Multidisciplinary Investigations of the Southern Ocean) campaign occurred between 5 January and 5 March 2024 aboard the RV Investigator. The voyage departed from Hobart, Australia covering the Southern Ocean from ∼ 43° S to the Mertz Glacier (67° S), east along the coast of Antarctica (∼ 150°–113° E) and north along the I9S oceanographic repeat line (115° E) to Fremantle, Australia (31° S). We will focus on the data collected along the Antarctic coastal leg (62–67° S, 113–151° E) here called “Antarctic Ice-Edge” and the northward I9S transect (32–62°, 115° E; ) here called “Open Ocean” (see Fig. S4 in the Supplement for the MISO voyage track). From an atmospheric perspective, the MISO voyage aimed to investigate how ocean-ice-biosphere-atmosphere interactions influence aerosols, clouds, precipitation and radiation over the Southern Ocean .

A Proton Transfer Reaction Time-of-Flight Mass Spectrometer (PTR-ToF-MS 4000, Ionicon Analytik) was deployed to measure volatile organic compounds including DMS, MeSH and DMSO. The DMS and MeSH measurements by PTR-ToF-MS during the MISO voyage have been described previously and more information is provided in Sect. S1 (Figs. S1 and S2). Briefly, ambient air was sampled from the bow mast 18 m above sea level via a heated (40 °C) polyfluoroalkyl (PFA) sample line (42 m, 9.53 mm I.D.) at a flow rate of ∼ 7 L min⁻¹. A flow of ∼ 1 L min⁻¹ was drawn off the main inlet via a sample valve switchboard and the PTR-ToF-MS sampled 0.37 L min⁻¹ from this flow.

The PTR-ToF-MS was operated with an applied drift tube voltage of 460 V, pressure of 2.6 mbar and temperature of 110 °C (E/N= 132 Td). A mass scan was collected every 10 s and a mass axis calibration was performed every 100 s using a diodobenzene permeation source (PerMaSCal, Ionicon Analytik GmbH, Innsbruck, Austria). Measurements of instrument background were attained by directing ambient air via a zero air generator (Parker Balston 75-83, Haverhill, MA) and the calibration and ambient data were background-corrected. Detection limits were 5, 2 and 2 ppt for DMS, MeSH, and DMSO respectively.

The sensitivity of the system to DMS was determined approximately every 5 d during the voyage from measurements with a ∼ 1 ppm certified gas standard (Apel-Riemer Environmental Inc, CO) diluted with zero air yielding a sensitivity of 518 ± 43 cps (counts per second) ppb⁻¹. MeSH and DMSO were not present in the calibration standard used during the voyage. Instead, MeSH and DMSO sensitivity was derived from post-voyage measurements of certified gas standards containing DMS, MeSH and DMSO (Apel Riemer Environmental Inc, CO). MeSH and DMSO sensitivities of 160 ± 13 and 88 ± 7 cps ppb⁻¹, respectively, were determined with consistent ratios to DMS of 1:3 (similar to ) and 1:6 (similar to ), respectively. These ratios were then applied to the voyage DMS calibrations to derive scaled sensitivities for MeSH and DMSO. Data were processed and analysed using Ionicon Data Analyzer 2.2.0.7 (IDA, Ionicon Analytik, Innsbruck, Austria, ). Air mass contamination by ship exhaust emissions was taken into account by identifying periods of enhanced hydrocarbons (C₄H₈H⁺, m/z = 57) and aromatics (benzene, toluene, and xylene) associated with combustion and such affected data were removed (see more detail in ). We compared this PTR-ToF-MS-specific exhaust influence to a rolling-window median absolute deviation-based outlier method applied to particle number concentrations and then refined through temporal aggregation and detailed manual inspection of particle size distributions, black carbon, radon, air mass history, and other combustion tracers to produce a final 1 min exhaust and air mass classification for the entire voyage; the complete methodology of which is presented in . This filtering method yields minor differences from the concentrations reported in , which used a more conservative approach and therefore reported fewer 1 min DMS and MeSH data (e.g., : I9S Transect VMS n∼ 12 000 1 min, this study: I9S Transect VMS n∼ 15 000 1 min).

Gas-phase Methanesulfonic Acid (MSA) and Sulfuric Acid (SA) were measured by a high resolution Chemical Ionisation Mass Spectrometer (Aerodyne Research, MA) employing nitrate reagent ions (NO₃-CIMS). A similar instrument setup has been previously described in . Ambient air was sampled from a separate inlet and included a straight 1.6 m stainless steel 25.4 mm O.D. line at a flow rate of 60 L min⁻¹ followed by a flow of 10 L min⁻¹ drawn through 1.2 m of PFA 12.7 mm O.D. tubing to NO₃-CIMS. Diffusional losses in the sampling line were determined following the procedure described in , using the diffusion coefficient of SA, which was applied to both SA and MSA. SA and MSA were quantified from the sum of their respective product ions, normalised to the sum of the reagent ions, and then multiplied by a calibration factor of 8.75 × 10⁹ molecule cm⁻³ ncps⁻¹(normalized counts per second) derived as a median of 21 values reported across studies; which is the same calibration factor determined from .

Sulfate aerosol (SO42-) was measured using a Time-of-Flight Aerosol Chemical Speciation Monitor (ToF-ACSM, Aerodyne Research, MA) fitted with a PM₁ aerodynamic lens and a capture vaporiser. The ToF-ACSM was located in the Aerosol Laboratory in the bow of the ship. Aerosols were sampled from the above bow mast at 180 L min⁻¹ to a sample manifold (∼ 8 m from mast to Aerosol Laboratory sampling manifold, see ), and then via ∼ 3 m, 9.53 mm O.D. copper line at with a flow of 3 L min⁻¹ fitted with a Nafion dryer. After which the ToF-ACSM sub-sampled ∼ 84 mL min⁻¹. The instrument was calibrated with 5 mM ammonium nitrate and ammonium sulfate solutions prior to the voyage, yielding an ionization efficiency for nitrate of 84.8 ions pg⁻¹, and relative ionisation efficiency for sulfate of 1.64. A collection efficiency of 1 was applied to the data, which was analysed in Tofware v3.3.0 (TOFWARE AG, Thun, Switzerland) utilising standard fragmentation tables to determine mass loading (µg m⁻³) of sulfate. Minimum detectable limits (MDL) for sulfate of 5 ng m⁻³ were calculated as three times the standard deviation from measurements sampling with a high efficiency particulate air (HEPA) filter connected at the ToF-ACSM inlet.

Ozone (O₃) was measured using two standard Thermo Scientific Model 49i UV photometric O₃ analysers (Thermo Fisher Scientific Inc, Franklin, MA). Both analysers were situated in the Aerosol Laboratory and sampled via 6.35 mm O.D. PTFE line from the same manifold and inlet described in Sect. . The meteorological instrumentation has been described previously in , including measurement of humidity, wind speed and direction (vane and ultrasonic), sea-surface temperature (radiometer), solar radiation (pyranometer), photosynthetically active radiation (PAR), air temperature, air pressure, rain, ozone, and trace gases from port and starboard sensors. The RV Investigator has two drop keels that can be lowered 4 m below the hull (about 7–8 m below the waterline) to mount acoustic sensors and transducers; the port keel also contains an inlet for the pumped seawater sampling system. A thermosalinograph (TSG) continuously measures ocean surface temperature, salinity, flow rate, fluorescence, pCO₂, and oxygen (optode). All RV Investigator MISO (IN2024_V01) voyage data can be accessed at .

Seawater DMS, DMSO, and DMSP (DMS/O/P) concentrations were determined via the Organic Sulfur Sequential Chemical Analysis Robot (OSSCAR), a custom-built, semi-autonomous purge-and-trap system . DMS/O/P concentrations were quantified using a custom chemiluminescence detector coupled to OSSCAR, following a similar approach to . Full instrument configuration, operating conditions, and calibration procedures are provided in Sect. S2. Briefly, the system first sparges and traps DMS for quantification via chemiluminescence, followed by converting subsequent sub-samples of DMSP and DMSO to DMS via fast NaOH hydrolysis and electrochemical reduction , respectively. The 30 min electrochemical reduction protocol was adapted to use a NiSO₄ catalyst. Nitrogen carrier gas delivered the analyte from OSSCAR to the reaction chamber, where an ozone–air mixture initiated the chemiluminescence reaction, and the amplified signal was recorded using a dedicated DAQ system.

Total Chl a and phytoplankton community structure as a proportion of Chl a (µg Chl a L⁻¹) were assessed using high-performance liquid chromatography (HPLC) pigment analysis on samples collected from Conductivity, Temperature, Depth instrument (CTD) vertical profile deployments across the entire voyage. Full methodological details are provided in Sect. S2. Briefly, analyses were conducted following , with chemotaxonomic interpretation performed using the phytoclass R package . Pigment profiles were grouped using hierarchical clustering to account for variability in pigment : Chl a ratios, and phytoplankton classes were then derived via simulated annealing, resolving eight taxa commonly reported for the Southern Ocean namely: diatoms, haptophytes, dinoflagellates, prasinophytes, chlorophytes, cryptophytes, pelagophytes, and cyanobacteria (Figs. S3 and S4). The ship’s underway seawater supply, which had an intake in the ship’s drop keel (water depth ∼ 7 m), was equipped with a WETStar fluorometer (Wetlabs, Inc., Philomath, Oregon, USA) that measured Chl a fluorescence at 695 nm (excitation at 460 nm) continuously throughout the voyage. Travel time through the ship’s plumbing from intake to the fluorometer was ∼ 1–2 min. Underway Chl a fluorescence measurements were scaled using a correction factor of 3.39, derived from 64 paired in-situ fluorescence and HPLC pigment measurements. These paired observations were obtained from CTD water samples collected between 3.5 and 450 m depth and compared with corresponding fluorescence measurements. Restricting the analysis to samples collected within the underway seawater intake depth (≤ 10 m) produced a similar correction factor of 3.1 (n= 10). As there was little evidence of near-surface non-photochemical quenching during the voyage, we adopted the correction factor derived from the full dataset (3.39), providing a more robust estimate based on the larger number of paired observations.

To understand the origin of the boundary layer air masses the RV Investigator encountered during the voyage, hourly back trajectories were produced using the Lagrangian analysis tool (LAGRANTO version 2.0; ), using hourly 0.25° ERA5 single level and pressure level data. The hourly back trajectories ran for 120 h and were initialized at 975 hPa. For the purposes of categorising air mass histories encountered during MISO, only the final 12 h prior to arrival were retained, representing the most recent atmospheric transport history influencing surface observations on the time scales of VMS emission and oxidation processes. Air masses were categorised according to latitude and exposure to biological productivity. The ERA5 land-sea mask was used as a proxy for surface type (< 0.1 is oceanic, > 0.1 is land/ice). Trajectories were classified into three broad spatial categories: Antarctic, Australian, and oceanic. Trajectories with an average latitude south of 62° S and over regions with high land/ice fraction (> 0.1) were defined as Antarctic, indicating continental/sea ice influence . Trajectories with an average latitude north of 45° S with high land fraction (> 0.1) were defined as Australian, indicating anthropogenic influence. All remaining trajectories were classified as oceanic and spanned 34–67° S.

Back trajectories were combined with monthly satellite products of chlorophyll a (Chl a, in µg m⁻³) and sea surface temperature (SST, in °C) at 0.1° × 0.1° spatial resolution from the Moderate Resolution Imaging Spectroradiometer (MODIS) Aqua sensor . This was repeated using daily ocean mixed layer depth (MLD, in m) fields at 0.125° × 0.125° resolution from the ARMOR3D NRT–TSHUVMld Global Ocean Observation-based product . The cumulative boundary layer Chl a exposure was then calculated for each back trajectory by summing the Chl a value along each point when the trajectory was most recently below the ERA5 atmospheric boundary layer height. A sensitivity test of multiple backward trajectory durations (12–120 h) confirmed that restricting air mass back trajectories to 12 h demonstrated the strongest relationship with VMS species, while distinguishing air mass spatial category consistent with the time scales of VMS emission and oxidation processes (; see Fig. S5). This cumulative Chl a was used as a proxy for biological productivity, divided into high or low productivity based on the median cumulative Chl a (1.23 µg m⁻³). This approach generated six trajectory categories based on 12 h cumulative boundary layer Chl a (cBLChl a) used for later analyses: Antarctic high productivity, Antarctic low productivity, Australian high productivity, Australian low productivity, Oceanic high productivity, and Oceanic low productivity. For the purposes of the current study, only Oceanic and Antarctic air mass influences were considered within the scope of the following analysis.

Air masses sampled along the Antarctic Ice-Edge (62–67° S) and the Open Ocean (32–62° S) legs traversed during the MISO voyage exhibited unique characteristics in their exposure to marine biologically productive regions, and their associated primary VMS concentrations and oxidation product regimes (Fig. ). The transit down to the Antarctic coast (5–10 January; Fig. ) was excluded from analyses here as data were significantly impacted by ship exhaust contamination, compared to the extensively measured I9S Transect. The latitudinal gradients of atmospheric DMS and MeSH from this voyage have been previously described in .

Observations were classified into four air mass history categories based on 12 h back trajectories and cumulative boundary layer Chl a exposure (cBLChl a): oceanic or Antarctic origin, each with low or high biological productivity (threshold = 1.23 µg m⁻³; Sect. ). Antarctic Ice-Edge air masses were predominantly oceanic (83 %) and high productivity (78 %; Figs. S6 and S7; marine in-situ underway Chl a up to 5 µg L⁻¹, > 62° S), coinciding with the highest atmospheric concentrations of DMS, with extreme spikes up to 5.7 ppb and median concentrations of 340 ppt. In comparison, MeSH concentrations were up to 200 ppt (median = 23 ppt, Fig. ) over the same spatial regions. Open Ocean air masses were exclusively oceanic and mostly low productivity (81 %; Figs. S6 and S7; marine in-situ underway measured Chl a < 1 µg L⁻¹), where DMS and MeSH were more tightly coupled (R2 = 0.80, slope = 0.173, p< 0.05) ranging from below detection limits (BDL) to 1.4 ppb (median = 271 ppt) and BDL to 253 ppt (median = 28 ppt), respectively. Across both regions, air masses with high biological productivity consistently exhibited higher concentrations of DMS, MeSH and DMSO than low-productivity air masses (Fig. ).

As illustrated by , the MeSH : DMS diurnal cycle observed during the MISO voyage exhibited stronger diurnal variability at lower latitudes (maximum-to-minimum ratio of ∼ 3.5 at ∼ 38° S) than at higher latitudes (maximum-to-minimum ratio of ∼ 1.5–2 at ∼ 55–60° S). This latitudinal difference was attributed to variations in OH radical abundance and small differences in reaction rates of MeSH and DMS with OH. In line with the VMS oxidation schematic illustrated in Fig. , higher concentrations of gas-phase DMSO were observed across the Antarctic Ice-Edge (up to 84 ppt, Fig. ) and showed strong coherence with DMS. Elevated gaseous MSA concentrations (up to 1.5 × 10⁸ molecules cm⁻³; Fig. ) did not coincide with precursor VMS levels and suggest evaporation from MSA-containing aerosols and long-range transport from the Antarctic continent . Sulfuric acid concentrations were highest at warmer low latitudes (up to 1.3 × 10⁶ molecules cm⁻³, < 45° S).

Next, we explore whether air mass history, in-situ observations, and trajectory-integrated metrics can explain and predict variability in VMS composition within Antarctic and oceanic air masses. A consistent co-varying relationship between DMS and cBLChl a occurred over the Antarctic Ice-Edge, which was especially driven by the oceanic high productivity air masses (R2= 0.49, n= 92, slope = 0.21, p< 0.01), as illustrated in Fig. . In fact, 6-hourly back trajectories demonstrated the strongest DMS–cBLChl a relationship in this region (Fig. S5); however, we use 12-hourly because it offers improved differentiation between Antarctic coastal and continental influence (Sect. ). This DMS–cBLChl a coupling was significantly stronger than for DMS and measured in-situ Chl a (R2< 0.05). Likewise, for 12 h back trajectory averaged Chl a (R2< 0.1) and > 12 h back trajectory cumulative Chl a (R2< 0.2) over both transects.

At the Antarctic Ice-Edge, hyper-local and rapidly evolving phytoplakton blooms can develop over timescales of only 1–3 h, resulting in pronounced short-term variability in atmospheric VMS concentrations (Figs. and ). To more appropriately represent the spatial and temporal mismatch between measured atmospheric DMS concentrations and the underlying biogeochemical indicators integrated by the 12 h air mass back trajectories, atmospheric DMS concentrations were averaged from 1 to 5 h resolution (Fig. ). This smoothing reduces the influence of transient local-scale variability while retaining the broader signal associated with air-mass exposure to the highly productive Antarctic coastal waters. Consistent with this interpretation, the relationship between DMS and cBLChl a strengthened following 5 h averaging (R2= 0.49) compared with the unsmoothed 1 h data (R2= 0.20). In contrast, no temporal smoothing was applied to the Open Ocean analysis because biological productivity was more spatially diffuse and less episodic. Under these conditions, 1 h atmospheric measurements combined with 24 h back trajectories provided the most representative relationship between DMS and cBLChl a (Fig. ).

Given the abundance of high DMSP- and DMS-producing phytoplankton species and shallow MLDs encountered along the Antarctic coastal transect , it is unsurprising we observed a close coupling between atmospheric DMS and cBLChl a in the high productivity Antarctic air masses (Sect. ). Perhaps more notably, for the Antarctic coastal regions, cBLChl a, derived from monthly MODIS retrievals and 12 h back trajectories, was able to reproduce the time-varying behaviour of atmospheric DMS occurring on the order of hours (Fig. ). Consistent with rapid DMS oxidation via OH-addition or bromine monoxide radical (BrO) oxidation at the Antarctic Ice-Edge (Fig. ), DMSO strongly co-varied with DMS under this temperature regime (R2= 0.7, slope = 0.07, p< 0.01) and therefore was also tied to cBLChl a (R2= 0.29, n= 443, p< 0.01).

Despite the moderately strong correlation between atmospheric MeSH and DMS at the Ice-Edge (R2= 0.67, p< 0.001, using 5th–95th percentile data), MeSH did not show as close a relationship with cBLChl a (R2< 0.14; Fig. S8). Moreover, MeSH : DMS correlation was the same in both Antarctic high- and low productivity air masses, indicating that Chl a was not a common predictor of the coupling between MeSH and DMS at the Antarctic Ice-Edge. This is likely attributable to the dominance of heterotrophic processes involved in DMSP conversion to MeSH and not primary production, as is the case for DMS (Sect. , Figs. S4, S9, and S10). These findings indicate that, in the context of summertime ice-edge conditions, even low time resolution remote sensing ocean optical properties, used to construct our cBLChl a metric, may be a sufficiently robust predictor of atmospheric DMS emissions, but not for atmospheric MeSH emissions. This has the potential to significantly improve prediction of extreme atmospheric DMS dynamics, which remain a key challenge for polar sulfur chemistry models .

Over high productivity regimes, other in-situ measurements and back trajectory variables including wind speed, air and sea surface temperature, humidity, total precipitation rate, ozone and boundary layer height, explained less than 20 % of the variance in DMS (Fig. S11). However, removing the biological source influence by analysing residuals after accounting for the cBLChl a resulted in in-situ wind speed explaining up to 36 % of the residual variance in DMS and up to 52 % of the residual variance in MeSH. Although a negative association between Chl a concentration and MLD was generally observed (Fig. S12), no direct linear relationship was identified between these variables within the high productivity regimes at the Antarctic Ice-Edge (R2< 0.07; Fig. S12). Neither was there a correlation between MLD (or Chl a : MLD) and DMS (R2< 0.05; Figs. S11, S12).

In contrast to the Antarctic Ice-Edge where cBLChl a was a reliable predictor of atmospheric DMS variability, cBLChl a was a poor indicator over Open Ocean regimes (Fig. ), where the DMS–cBLChl a relationship was weaker (R2= 0.15, n= 487, slope = 0.22, p< 0.01) and an optimal relationship occurred using 24-hourly back trajectories (Figs. and S5). North of the oceanic Polar Front, the absence of land barriers and the influence of the Antarctic Circumpolar Current create a more heterogeneous ocean surface with weaker, more dispersed, and often deeply mixed biological sources . The dominant taxa are often weaker DMSP producers compared to polar bloom species , especially when compared to those observed at the Ice-Edge (Sect. ). Relative to the shallower bathymetry and shallower MLDs in Antarctic shelf regions, such as the Mertz Polynya Area (MPA; 65.33–67° S, 140–150° E; see later in Sect. ), where total Chl a consistently peaked and thus cBLChl a best predicted DMS (Figs. , S9 and S11; Sect. ), MLDs over the Open Ocean were consistently deeper and phytoplankton biomass occurred at greater depths (Sect. ; Fig. S4; i.e., 80 m along the I9S transect versus 5 m at the MPA). The MODIS Aqua satellite retrieves only near-surface concentrations of Chl a , and therefore may not accurately represent total phytoplankton biomass in the bulk water over the Open Ocean region (34–62° S).

The Open Ocean region was characterised by generally weak and inconsistent coupling between atmospheric VMS concentrations and the range of meteorological, oceanographic, and biological variables examined (Fig. S11). However, consistent with previous studies reporting the effect of open ocean slope transitions on biological sources , Figure a illustrates that measured seafloor depth explained 35 % of the variance of our atmospheric VMS observations. The highest observed VMS concentrations occurred at the shallowest ocean depths (Fig. a), especially at the Subtropical Front (MeSH > 200 ppt and DMS > 1 ppb, < 4000 m; Figs. e and a). MLD exhibited a substantially stronger correlation with Chl a along the Open Ocean transect than at the Antarctic Ice-Edge (R2= 0.35; Fig. S12); however, no direct correlations were observed between MLD and DMS over the Open Ocean (R2< 0.01, p= 0.7; Fig. S12) or the Antarctic Ice-Edge (R2= 0.06, p< 0.01; Fig. S12). While we found that cBLChl a and MLD were moderately correlated over the Open Ocean (R2= 0.37), we did not observe a consistent relationship between these variables and atmospheric VMS (R2< 0.1; Fig. S11) unlike other previous research that has shown a correlation between seawater DMS and MLD as well as DMS and Chl a : MLD relationships . Despite this, we observed a moderately strong inverse relationship between VMS concentrations and measured bathymetry (R2= 0.35, Fig. ), a pattern not observed over the region we define here as the Antarctic Ice-Edge. However, this relationship likely reflects the indirect influence of large-scale circulation and associated air mass history rather than a mechanistic control on VMS production or emission. Atmospheric factors that co-varied with VMS over the Open Ocean included ozone (Fig. b), relative humidity (Fig. c), back trajectory specific humidity, total precipitation rate and air temperature (Fig. S11), which together explained ∼ 25 %–50 % of the variance in VMS.

The absence of a robust predictor suggests that Open Ocean region atmospheric VMS variability emerges from the integration of heterogeneous source regions, transport, and atmospheric processing, rather than being governed by clearly identifiable source-related drivers like those observed at the highly biologically-mediated Antarctic Ice-Edge. Unlike the shallow, highly stratified polar waters, the deep offshore water column of the Open Ocean is shaped by large-scale oceanographic processes rather than local biological production, indicating it was not as strongly source-driven by local surface-ocean processes as what was observed at the Antarctic Ice-Edge. Along the I9S transect (115° E) measured during this voyage, previous work by demonstrated that the water column is influenced by deep potential-vorticity maxima and enhanced mixing where dense western-boundary waters interact with lighter Antarctic Circumpolar Current waters.

Spatial patterns in the atmospheric MeSH : DMS ratio provide insight into how underlying DMSP cycling and atmospheric processing modulate VMS speciation across the Southern Ocean regimes. As first presented in , atmospheric MeSH : DMS ratios showed pronounced spatial variability, with lower and more variable ratios observed along the Ice-Edge (∼ 0.05) than over the Open Ocean (∼ 0.15), and lower than previously reported from other Southern Ocean data sets collected at lower latitudes (∼ 0.13–0.19; ). The low atmospheric MeSH : DMS ratios that we measured at the Antarctic Ice-Edge are more consistent with ratios of ∼ 0.02 reported from measurements of dissolved MeSH and DMS in coastal waters west of the Antarctic Peninsula , and median atmospheric ratios < 0.1 reported from Northern Hemisphere high latitudes . These lower and more variable MeSH : DMS ratios may have implications for modelling studies which have used global seawater DMS climatologies and the relationship between MeSH and DMS to construct estimates of oceanic MeSH . These modelling studies have used MeSH : DMS ratio scaling factors on the order of 0.1–0.2, which likely resulted in an overestimation of MeSH emissions for regions such as the Antarctic Ice-Edge.

In contrast to the air masses encountered at the Antarctic Ice-Edge, the Open Ocean air masses exhibited a consistent coupling between DMS and MeSH (as first described by ), yet their distributions are largely decoupled from surface chlorophyll biomass (Figs. and S11). This is because DMS and MeSH largely share a common biogenic origin from marine DMSP cycling and experience relatively homogeneous atmospheric processing. At the Antarctic Ice-Edge, the MeSH : DMS relationship is more complex, as discussed in the next Sect. .

The atmospheric MeSH : DMS ratios observed over the Open Ocean (∼ 0.15) in this study are consistent with the 0.13–0.19 ratios reported from other Southern Ocean data sets collected at lower latitudes and the ∼ 0.1–0.2 ratios reported from other ocean regions . We estimated sea–air fluxes of DMS and MeSH from their nocturnal accumulation in the marine boundary layer following (see detailed method in Sect. S4). This approach exploits reduced nighttime photochemical loss, allowing DMS and MeSH to accumulate in the boundary layer and providing an estimate of their net oceanic emissions. Consistent with this coupling under low-productivity conditions (where cBLChl a does not dominate variability in this study), nocturnal accumulation periods over the Open Ocean yielded sea–air flux estimates of 15–124 for DMS and 0.1–17 ng m⁻² s⁻¹ for MeSH (Sect. S4; Tables S1 and S3), with a broadly variable MeSH : DMS flux ratio (0.008–0.462) that generally exceeded recent southwest Pacific nocturnal accumulation estimates . However, the increase in atmospheric MeSH : DMS with increased SSTs observed in this study, especially over the Subtropical Front (Fig. S13), is not consistent with the trend determined in a previous global synthesis of seawater MeSH and DMS concentrations , which reported lower MeSH : DMS ratios (∼ 0.1) associated with SSTs above a threshold of 8–12 °C, and higher MeSH : DMS (∼ 0.3) at lower SSTs; nevertheless, that analysis was based on very limited data from these latitudes of the Southern Ocean.

Having identified the spatial patterns in atmospheric DMS and MeSH concentrations and the key predictors of their variability across the Antarctic Ice-Edge and Open Ocean regions, we now investigate selected case studies to gain process-level insights into how Southern Ocean and Antarctic coastal oceanographic and biological features influence atmospheric DMS and MeSH production and emission. Extreme seawater DMS concentrations, in excess of 100 nM, have been reported in Antarctic coastal waters associated with springtime sea ice break up that releases trapped DMS/P and seeds subsequent intense phytoplankton blooms that can persist throughout the summer . The spatial and temporal heterogeneity in VMS emissions introduced by these hotspots is not well represented in models, which instead often implement open ocean derived fluxes . Furthermore, current models do not account for MeSH, which growing evidence suggests also plays an important role in aerosol formation and the oxidative capacity of the atmosphere . Modelling efforts to date have exploited the typically close coupling between DMS and MeSH to derive global MeSH emission fields from DMS climatologies . However, as mentioned above, although the MeSH : DMS ratios over the Southern Ocean are relatively consistent (10 %–35 %), they vary seasonally and are much lower (< 5 %) and more variable at the Antarctic coast .

We observed two distinct episodes at the Antarctic coast with extreme atmospheric VMS concentrations, up to 25 times the MISO voyage medians, which were associated with oceanic and Antarctic high productivity air masses sampled over intense blooms at the Mertz Polynya (MPA), and at a process station (PS2) along the continental slope. In both case studies, atmospheric concentration dynamics were predominantly source driven, rather than significantly controlled by atmospheric loss processes. This is supported by the close coupling between air mass DMS concentrations and exposure to phytoplankton biomass as indicated by monthly MODIS Chl a (Fig. ) along with the high dissolved concentrations of DMS at the MPA (Fig. S3; dissolved DMS measurements were not available at PS2). Wind speeds and water temperatures in both case studies were not outside of normal ice-edge conditions (Figs. and S14), suggesting that high atmospheric VMS concentrations and concentration ratios are therefore considered indicative of water-side processes. A third period of atmospheric VMS enhancements occurred at the biologically productive Subtropical Front region.

Once emitted into the atmosphere, gas-phase oxidation and multiphase chemical processing are important loss mechanisms for VMS (Fig. ), which contribute to the atmospheric sulfur cycle and to Southern Ocean aerosol and cloud processes . In the cloud-free marine boundary layer, DMS is oxidised to DMSO, mainly via oxidation by OH-addition and BrO (Fig. ; ). BrO oxidation has previously been identified as a potentially important mechanism, contributing up to 44 % of global DMSO production in a modeled global tropospheric DMSO budget . MeSH is primarily oxidized via H-abstraction to short-lived radicals and SO₂ (not measured on MISO), and apart from acting as an OH-scavenger, MeSH oxidation would not contribute to the observed atmospheric DMSO abundance . Observed DMS and DMSO concentrations are well correlated at the Ice-Edge (R2= 0.69, p< 0.01) but weakly correlated over the Open Ocean (R2= 0.04, p< 0.01). The DMSO : DMS ratio varied significantly (∼ 0–1) with temperature across spatial regions of the Southern Ocean (Fig. ).

South of the atmospheric Polar Front, subzero air temperatures coincide with elevated DMSO : DMS and a strong coupling (R2= 0.69, Fig. ), consistent with previous work highlighting temperature-driven partitioning of DMS oxidation products near Antarctica . We propose that the strong DMSO : DMS coupling at the Ice-Edge could be evidence of local, rapid OH-addition oxidation of DMS in the photochemically active boundary layer of the summertime Antarctic Ice-Edge. This pathway may be amplified by BrO-mediated oxidation, which has reaction rates an order of magnitude faster than OH-mediated oxidation in the Antarctic environment, leading to estimated DMS lifetimes on the order of minutes to hours which is consistent with the observed DMSO : DMS correlation at the Ice-Edge (Fig. ). The DMSO : DMS correlation is even stronger and an increased slope is observed (R2= 0.76, slope = 0.16, p< 0.01; Fig. S17) when only considering intense sunlight conditions (PAR > 85th percentile: > 700 µmol m⁻² s⁻¹) and low-moderate in-situ wind speeds (1–10 m s⁻¹), suggesting that under such conditions local photochemical production of DMSO from DMS likely occurred at the Ice-Edge. This is further supported by the absence of any significant time lag in the time-varying behaviour of the two species. An autocorrelation analysis shows that they are optimally coupled at the time of measurement, with no improvement in the correlation when time-lagging the data (Fig. S18). Although these are likely significantly influenced by meteorology, the lack of observable lag under stable low wind speed conditions points to the importance of oxidation mechanisms, which occur faster than characterized reactions between DMS and OH. However, co-emission of DMSO and DMS from the surface ocean could, in theory, also explain the lack of lagged time varying behaviour.

DMSO is highly soluble in seawater, with a Henry's law constant 5 orders of magnitude lower than that of DMS . Its ventilation from seawater is therefore not thermodynamically favoured. This is supported by observations during on board mesocosm incubation experiments which were conducted along the Antarctic Ice-Edge (Sect. S5). In these experiments, DMS concentrations reached as high as 15 ppb (∼ 2–3 times the median ambient concentrations) within the oxidant-free mesocosm headspace, while DMSO had a maximum atmospheric concentration of 40 ppt, it was rarely observed above detection limits in the headspace, and did not co-vary with DMS. This is also consistent with who studied an induced phytoplankton bloom and found little evidence for ventilation of DMS oxidation products into oxidant-free headspace environments. Therefore, multiple lines of evidence point to gas-phase production of DMSO from DMS oxidation pathways rather than ventilation from the ocean surface as the dominant source of DMSO.

However, in addition to microbial degradation and ventilation to the atmosphere, DMS is also lost from the surface ocean via photolysis . This process occurs in the upper mixed layer when DMS is oxidised by reactive species generated from light absorption by photo-active substances, particularly CDOM and nitrate, with peak photolysis at UV wavelengths ∼ 320–330 nm, yielding DMSO as the main product . The mesocosm setup used in the experiments (Sect. ; Fig. S16) filtered out UV wavelengths < 364 nm. Overall, DMS loss via photo-oxidation has been shown to be particularly important in the Southern Ocean and potentially responsible for the decoupling observed between DMS and Chl a in frontal regions exhibiting high CDOM and nitrate concentrations . It is unclear if DMSO produced at the air-sea interface can be efficiently transferred to the atmosphere; however, it may provide another possible explanation for the tight DMS : DMSO coupling observed in the DMS hotspots encountered along the Ice-Edge in this study.

We found no consistent correlation between any of the VMS species (DMS, MeSH, and DMSO) and any of the later VMS oxidation products measured during the MISO voyage, namely MSA (g), SA (g), or aerosol-phase sulfate (all R2< 0.2; see Fig. S11). This decoupling is likely indicative of the importance of multiphase chemistry, meteorology, and sinks in the atmospheric concentrations of later VMS oxidation products, and is consistent with other studies of atmospheric sulfur in remote marine atmospheres . Persistent low-pressure systems and cyclones around the Antarctic continent promote strong vertical uplift, shifting much of the MSA and SA production into the free-troposphere where low temperatures and low condensation sink favour slower, multistep oxidation pathways and new particle formation. Recent work by suggests that katabatic outflow transports free-tropospheric, highly acidic aerosol particles back into the boundary layer, where low RH and high acidity can drive evaporation of particle-phase MSA, which helps to explain the decoupling between our observed MSA/SA from local VMS emissions.

As illustrated in Fig. , the DMSO : DMS ratio is significantly and consistently higher under the cold conditions of the Ice-Edge below the atmospheric Polar Front (median DMSO : DMS = 12 %) than in Open Ocean conditions between the polar and subtropical fronts (median DMSO : DMS = 5 %). A significant potential contributor to this dynamic is likely the temperature dependence of DMS oxidative pathways, illustrated in Fig. . As described in , at low temperatures (< 285 K) the OH addition pathway is favoured, leading to higher production of DMSO and MSA. Under warmer conditions however, the H-abstraction DMS oxidation pathway is more favoured, which would lead to lower yields of DMSO and higher yields of SO₂ and SA. While the transition from a low to high DMSO : DMS regime occurs at significantly less than 285 K and synoptic scale dynamics related to DMS lofting above the boundary layer have also been identified as important contributors to DMSO production , the temperature dependence of oxidative fate remains an important likely contributor to observed dynamics. As the air temperature increases north of the atmospheric Polar Front and over the Open Ocean, we find that DMSO : DMS generally decreases and they become decoupled (R2= 0.04, Fig. ). This is consistent with austral summer DMSO measurements conducted at a similar latitude at Amsterdam Island (37.83° S, 77.50° E; ), which found that DMSO co-varied with DMS but they were only moderately correlated (R2= 0.21). Our current temperature-dependent conceptual framework for VMS oxidation does not fully capture the observed behaviour of an increased DMSO : DMS ratio at the warmer Subtropical Front (42–52° S, Fig. ). This suggests that latitudinal temperature gradients may not always be the primary factor responsible for DMSO : DMS variability across the broad regions of the Southern Ocean, although the mechanistic processes underlying this remain insufficiently understood and unresolved.