The Interplay Between Dimethyl Sulfide (DMS) and Methane (CH4) in a Coral Reef Ecosystem

Earth’s Radiation Budget is partly dictated by the fragile and complex balance between biogenic volatile organic compounds (BVOCs) and greenhouse gases (GHGs), which have the potential to impose cooling or warming once emitted to the atmosphere. Whilst methane (CH4) is strictly associated with global warming due to its solar-radiation absorbing properties, dimethyl sulfide (DMS) is generally considered a cooling gas through the light scattering properties of its atmospheric oxidation products. However, DMS may also partially contribute to the Earth’s warming through a small portion of it being degraded to CH4 in the water column. Coral reefs emit both DMS and CH4 but they have not previously been simultaneously measured. Here, we report DMS and CH4 fluxes as well as aerosol particle counts at Heron Island, southern Great Barrier Reef, during the austral summer of 2016. Sea-to-air DMS and CH4 fluxes were on average 24.9 ± 1.81 and 1.36 ± 0.11 µmol m-2 d-1, whilst intermediate (< 0.5-2.5 um) and large (> 2.5 um) particle number concentrations averaged 5.51 x 106 ± 1.73 x 105 m-3 and 1.15 x 106 ± 4.63 x 104 m-3, respectively. Positive correlations were found between DMS emissions and the abundance of intermediate (R2 = 0.1669, p < 0.001, n = 93) and large (R2 = 0.0869, p = 0.004, n = 93) aerosol particles, suggesting that DMS sea-to-air emissions significantly contribute to the growth of existing particles to the measured size ranges at the Heron Island lagoon. Additionally, a strong positive correlation was found between DMS and CH4 fluxes (R2 = 0.7526, p < 0.00001, n = 93), suggesting that the emission of these volatile compounds from coral reefs is closely linked. The slope of the regression between DMS and CH4 suggests that CH4 emissions at the Heron Island lagoon represent 5% of that of DMS, which is consistent with the average sea-to-air fluxes reported in this study (i.e. 24.9 ± 1.81 µmol m-2 d-1 for DMS and 1.36 ± 0.11 for CH4). These findings provide new insights on the complexity of BVOC and GHG emissions in coral reef systems and their potential role in climate regulation.


INTRODUCTION
Dimethyl sulfide (DMS) and methane (CH 4 ) are key biogenic compounds in climate change processes (Carpenter et al., 2012). DMS is often associated with a cooling effect through contributing to aerosol formation that increase Earth's radiative properties (Charlson et al., 1987). In contrast, CH 4 is responsible for a warming effect with a short-term greenhouse potential that is 23 to 69 times greater than that of CO 2 (Shine et al., 2005). The bulk of DMS is produced through the algal and bacterial enzymatic cleavage of dimethylsulfoniopropionate (DMSP) (Simo, 2001), a biogenic sulfur compound that is synthesised by a wide range of marine algae (Stefels, 2000), bacteria (Curson et al., 2017) and invertebrate corals (Raina et al., 2013). In contrast, the bulk of biogenic CH 4 is mainly produced by anaerobic methanogenic bacterial activity through the reduction of either CO 2 , acetate or methyl-group containing compounds (Liu and Whitman, 2008), although recent studies show that CH 4 can also be produced by plants, fungi, algae and cyanobacteria from methylated nitrogen-and sulfur-containing compounds in the presence of oxygen (Ernst et al., 2022). Thus, a small portion of CH 4 can originate from the hydrolysis of DMS according to the following equation (Kiene et al., 1986;Liu and Whitman, 2008): The link between DMS degradation and CH 4 production has now been reported across several studies on anoxic marine sediments (Kiene and Visscher, 1987;Kiene, 1988;Wang et al., 2009), where methanogenic bacteria are particularly abundant (Mechalas, 1974;Barnes and Goldberg, 1976). Indeed, it seems that the DMS-to-CH 4 conversion can also be mediated by methanogenic bacteria (Kiene and Visscher, 1987) but also by m e t h y l o t r o p h i c b a c t e r i a b e l o n g i n g t o t h e o r d e r Methanosarcinales and Methanobacteriales (Liu and Whitman, 2008). More recently, the role of DMS as a precursor of CH 4 was clearly demonstrated in upwelling waters as addition of 13 C enriched-DMS led to a significant increase in 13 C enriched-CH 4 (Florez-Leiva et al., 2013). Interestingly, DMS was estimated to contribute to about 28% of CH 4 production in both anoxic sediments (Kiene, 1988) and upwelling waters (Florez-Leiva et al., 2013), although these two marine habitats are likely to host very different bacterial communities. DMS as a potential precursor of CH 4 across various marine ecosystems adds a level of complexity to the role of DMS as a climate cooling agent (Charlson et al., 1987;Quinn and Bates, 2011;Jones, 2013).
DMS is expected to be particularly concentrated in coral reef ecosystems due to the high DMSP content in corals, coralassociated symbionts and a wide range of coral reef invertebrates (Deschaseaux et al., 2016;Haydon et al., 2018). However, although coral reef sea-to-air DMS fluxes contribute to atmospheric sulfur emissions (Swan et al., 2017), a recent modelling study suggested that coral-reef-derived DMS emissions most likely have a negligible effect on the local climate of the Great Barrier Reef (Fiddes et al., 2021). Coral reef frameworks are also sites of active anoxic and suboxic organic matter oxidation, making coral reef pore waters particularly rich in CH 4 and coral reef ecosystems ideal platforms for the release of CH 4 to the water column and atmosphere (Sansone et al., 1993;O'Reilly et al., 2015), which could be counteracting the cooling effect of DMS emissions. A recent study showed that permeable coral reef carbonate sediments were a source of DMS and CH 4 into the water column and that CH 4 production could be a sink for DMS in coral reef systems (Deschaseaux et al., 2019). Since sea-to-air DMS and CH 4 fluxes from coral reef systems have been independently reported by previous studies, the focus of this study was to simultaneously quantify and report sea-to-air DMS and CH 4 fluxes from the Heron Island reef lagoon, southern Great Barrier Reef, to assess the interplay of atmospheric DMS and CH 4 emissions.
We hypothesised that coral-reef DMS emissions would contribute to the growth of aerosol nanoparticles and that coral-reef DMS and CH 4 emissions are linked due to a small portion of DMS being hydrolysed to CH 4 in coral reef waters.

Study Site and Sampling
A field campaign was conducted on Heron Island, southern Great Barrier Reef (23.44°S, 151.91°E), in the austral summer of 2016 (4 th to the 17 th of February) where dissolved CH 4 and CO 2 were measured alongside dissolved DMS (DMS w ) using a cavityring-down spectrometer (Picarro G2201-i) (Maher et al., 2013) and a Vapor Generation -Chemiluminescence (VG-CL) device (Nagahata et al., 2013), respectively. A Gas Chromatograph (Varian CP3800 GC) equipped with a pulsed flame photometric detector (GC-PFPD) was used to measure atmospheric DMS (DMS a ) (Swan et al., 2015;Swan et al., 2017).
A water pump and a 60 m suction-rated pipe were used to pump seawater from the Heron Island reef flat to the Heron Island Research Station (HIRS), where all instruments were operated. The seawater inlet was attached to a cinder block 50 cm above the sediment bed and about 50 cm below the low-tide mark. A non-return valve and a 1 mm mesh net were placed around the suction pipe inlet to prevent large pieces of sediment and seaweed getting through the suction line. The mesh was cleaned every 2-3 days to prevent biofouling. Part of the pumped seawater was diverted into a showerhead exchanger that was connected in line with the cavity-ring-down spectrometer for CH 4 and CO 2 measurements, while the remainder was diverted to the HIRS' flow-through seawater recycling system. The recycling water outlet was used to manually sample seawater at a low flow rate using a 50 mL syringe for DMS w analysis. A HYDROLAB HL4 sonde was placed within 1 m of the water intake to record seawater temperature, salinity, dissolved oxygen (DO), and depth, every 15 min. Tide predictions were sourced from the Bureau of Meteorology (BoM) with low tide times locally adjusted +1.25 h for the Heron Island reef flat according to Swan et al. (2017).
An air intake consisting of~10 m Teflon ™ tubing was fixed to the roof of the HIRS, within~100 m in line of sight to the reef flat location where seawater was continuously pumped from. The air intake was shielded from rain. A wireless automated weather station (AWS, model XC0348, Electus Distribution) mounted within 1 m of the air intake provided data for wind speed (WS, ± 1 m s -1 for WS < 10 m s -1 and ± 10% for WS > 10 m s -1 ), wind direction (WD), rainfall, air temperature (± 1°C), humidity (± 5%) and barometric pressure (± 3hPa), at 5-min intervals. Meteorological data at 5 min intervals was used to match to the 15 min interval chemical measurements. Solar irradiance at the proximity of the air intake was recorded using a HOBO (Onset Co., USA) light logger (upper limit~6000 mE m -2 s -1 ). Light intensity HOBO Lux units were adjusted to a maximum light intensity of 2000 mEm −2 s −1 . First and last daytime lights were at 5:14 and 18:57 respectively on average over the sampling period. An air quality monitor laser particle counter (Dylos DC1700, Dylos Corp, CA, USA) was used to record 1-min averaged particle number concentrations at ambient humidity in the two size range fractions 0.5-2.5 μm and > 2.5 μm every 15 min. The air quality monitor was placed in a shielded location near the air intake.

DMS w and DMS a Measurements
DMS w concentrations were determined in triplicate every 15 min by placing 10 mL of seawater into a 50 mL sample tube that was manually shaken for 1 min, then pressurised with 30 mL of air and injected onto the chemiluminescence device (VG-CL) (Nagahata et al., 2013). When DMS w mixes with ozone in the VG-CL, it generates an instant chemiluminescent emission where the light intensity is converted into a quantifiable electrical signal. The injection tube was rinsed with deionised water in between each injection to prevent analyte carry-over. A 6-point calibration was run at the beginning of the field campaign, and either a 4 nM or 10 nM DMSP standard was randomly run each day to monitor the reproducibility and stability of the system over time. Because these measurements are labour-intensive and could not be automated, sampling occurred at random times of the day and night over the 2week field campaign.
DMS a concentrations were determined on the GC-PFPD using an automated cryogenic trapping system that collected 4L of air for analysis, providing a 0.1 nmol m -3 (0.002 ppb) limit of detection. The expanded relative measurement uncertainty of the automated GC-PFPD was 13% (k = 2, for a 95% CI). A complete description of the configuration, operation, calibration, and uncertainty analysis of the automated GC-PFPD is described by Swan et al. (2015).

Flux Calculations
Sea-to-air DMS fluxes were estimated based on the different parameterisations proposed by Liss and Merlivat (1986) Table 1). The sea-to-air DMS fluxes and uncertainties presented here correspond to the median and standard error of these combined fluxes.
Briefly, sea-to-air fluxes of DMS (F DMS ) (in μmol m -2 d -1 ) were estimated based on the following equation: Where K T is the gas transfer velocity constant (in m d -1 ), a is the dimensionless Henry's Law constant and C w and C g are DMS concentrations (in μmol m -3 ) in the water and gas phase, respectively, with each parameterisation using a different approach to estimate K T .
The dimensionless Henry's Law Constant (a) for DMS solubility in seawater was calculated using the following equation: Where H K (atm L mol -1 ) is the Henry's Law Constant for DMS, R is the universal gas constant (0.082 L atm K −1 mol −1 ) and T is the seawater temperature in Kelvin. H K was calculated based on the following equation by Dacey et al. (1984): Sea-to-air CH 4 , CO 2 and O 2 fluxes were calculated based on the parameterisation approaches proposed by Ho et al. (2006) and Wanninkhof (2014). Atmospheric concentrations were assumed to be constant (CH 4 1.8 ppm, CO 2 400 ppm and O 2 21000 ppm). Solubility coefficients for CH 4 (Wiesenburg and Guinasso Jr, 1979), CO 2 (Weiss, 1974) and O 2 (Benson and Krause Jr, 1984) were calculated based on temperature and salinity.

Reef Production
Reef DMS, CH 4 , CO 2 and O 2 production (RP, mol m -2 h -1 ) were calculated around each low tide using the following equation: Where DC W is the difference in dissolved concentrations between the highest point following low tide and the actual low tide, DT is the time difference between these 2 points, D is the average depth for that time period and MeanFlux is the average sea-to-air flux for that period.

Statistical Analysis
The significance of the correlations between DMS, CH 4 , CO 2 and O 2 were evaluated using the Pearson correlation method. Given that DMS and CH 4 showed the strongest correlation (R 2 = 0.7526, p <0.00001, n = 93), we specifically assessed what drives DMS and CH 4 fluxes by carrying out stepwise multiple linear regressions (MLRs) against seven potential predictors and their interactions (salinity, pH, depth, wind direction, dewpoint, windchill and solar irradiance). In order to avoid over-fitting and find a balance between model complexity and explanatory power, we followed a backward elimination process based on the Akaike Information Criterion (AIC, details in Supplementary Figure 1) starting with all seven potential measured predictors. Note that we opted not to include temperature and wind speed, which are covariates in flux calculations. Windchill and dewpoint are influenced by air temperature, which is not a covariate in flux calculations, and wind direction is not correlated to wind speed. Calculations were performed using the functions "boxplot", "stepwisefit", and "plotEffects" in MATLAB.

Environmental Data
Sea surface temperature (SST) within the Heron Island reef lagoon fluctuated between 25.0 and 29.2°C over the course of the campaign with an average SST of 27.3 ± 0.94°C. Salinity and pH were on average 37.0 ± 0.25 ppt and 8.18 ± 0.19, respectively. Seawater depth ranged from 0.46 to 1.38 m, 10 m wind speed from 0 to 9.20 m s -1 , windchill from 21.3 to 29.9°C, dewpoint from 20.0 to 23.4°C and light intensity from 0 to 2000 μE m -2 s -1 .
All 24h-integrated sea-to-air fluxes showed a diurnal trend with DMS, CO 2 and CH 4 fluxes being generally greater at night than during the day while O 2 fluxes showed the opposite trend ( Figure 1). Sea-to-air CO 2 fluxes showed the most variability during the first hours of sunlight while DMS and O 2 fluxes showed the most variability between dusk and midnight. CH 4 fluxes showed the most variability both at night and during the first hours of sunlight.

Multiple Linear Regressions
The backward elimination process used in the stepwise Multiple Linear Regressions (MLRs) revealed that salinity, pH, depth, wind direction, dewpoint and wind chill were the six main drivers of seato-air DMS fluxes and that pH, depth, wind direction, dewpoint, wind chill and light were the six main drivers of CH 4 fluxes ( Table 1). The linear fit between measured and predicted DMS and CH 4 fluxes exhibited an R 2 of 0.717 and 0.631 (n = 93, p <0.00001), respectively ( Figures 3A, B). The most negative drivers of DMS emissions were wind chill and pH whereas the most positive driver was wind direction ( Figure 3C). Similarly, the main negative and positive drivers of CH 4 emissions were pH and wind direction, respectively ( Figure 3D). When plotting wind direction (in azimuth degrees) against DMS and CH 4 fluxes (Supplementary Figure 4), it appeared that the wind direction leading to the greatest DMS and CH 4 emissions was predominantly at 180°, which corresponds to the direction of the dominant southerly trade winds at Heron Island.

Reef Production
The DMS reef production exhibited positive values at all times, with a few sporadic spikes around peak hours of sunlight (1 1:00) and in the middle of the night (~23:00) ( Figure 4A). Although an order of magnitude lower, the CH 4 reef production was also positive at all times but with greater values recorded at night ( Figure 4B). The CO 2 reef production was positive at night and for the first half of the day but negative from about 12:00 to 20:00 ( Figure 4C). As expected, the O 2 reef production was  essentially positive during the day until dusk, and negative at night ( Figure 4D).

Flux Estimations
DMS fluxes reported in this study (Min = not-detectable, Max = 69.5 μmol m -2 d -1 , Mean = 24.9 ± 1.81 μmol m -2 d -1 , n = 93) fell within the range of previously reported sea-to-air DMS fluxes for 5 coral reef systems across the Great Barrier Reef, which varied from not-detectable to 153 μmol m -2 d -1 in the austral summer wet season (Jones et al., 2018). However, seasonal DMS fluxes recorded in the current study were about five and four times greater on average than the 2012 summer wet season study conducted on Heron Island (5.0 μmol m -2 d -1 , n = 651) (Swan et al., 2017) and the average DMS fluxes reported by Jones et al. (2018) (6.4 μmol m -2 d -1 , n = 237), respectively. Since DMS a (mean ± SD) at Heron Island in the 2012 and 2016 summers were similar (i.e. 3.9 ± 1.5, n = 651, and 3.7 ± 0.8, n = 761 nmol m -3 , respectively, data not shown), the high fluxes in this study most likely result from temporally elevated DMS production in the Heron Island reef lagoon in the year 2016 or on the section of the Heron Island reef flat where seawater samples were collected. The significant differences in the average DMS fluxes for the summers of 2012 and 2016 at Heron Island might also reflect differences in the employed flux calculation (photochemical ambient mass balance approach used by Swan et al. (2017) as opposed to the gradient method flux calculation used in this study). Similarly, Jones et al. (2018) used the LM86 gradient method flux parameterisation, which was shown to give much lower flux estimates than the other gradient method flux parameterisations used in the current study (see Supplementary  Table 1), hence indicating that the approach employed can lead to significantly different values. The estimated flux difference might also reflect the limitations of the non-automated VG-CL instrument used in this study, which ultimately led to  Table 1).

The Interplay Between DMS and CH 4
The strong positive correlation between water-air DMS and CH 4 fluxes (R 2 = 0.7526) suggests that DMS and CH 4 emissions from the Heron Island reef lagoon are closely linked. Further to this observation, the MLR analysis revealed that sea-to-air DMS and CH 4 fluxes were both driven by pH, depth, wind direction, dewpoint and wind chill. This indicates that DMS and CH 4 fluxes are driven by very similar environmental factors, thus potentially explaining part of the correlation between DMS and CH 4 emissions at the Heron Island reef lagoon. Dissolved concentrations of DMS and CH 4 were also strongly correlated (Supplementary Figure 3), which indicates that the production of DMS and CH 4 in this reef system is also intimately linked. In contrast there was no clear correlation between the RPs of DMS and CH 4 , which suggests that DMS and CH 4 were subject to different accumulation rates and thus to different production and/or degradation processes.
DMS production mainly relies on the enzymatic cleavage of DMSP in the water column (Simo, 2001) while methanogenesis mainly depends on the conversion of CO 2 and acetate into CH 4 under anaerobic conditions (Liu and Whitman, 2008), which most likely occurs in marine sediments in coral reef ecosystems (Deschaseaux et al., 2019;Reading et al., 2021). The main sinks of dissolved DMS are expected to be biological and photochemical oxidation (del Valle et al., 2009) as well as emission fluxes from the water column to the atmosphere (Lana et al., 2011). Similarly, oceanic emissions (Weber et al., 2019) and microbial oxidation to CO 2 through both aerobic and anaerobic pathways (Pain et al., 2019) are considered the main sinks of dissolved CH 4 in marine systems. However, the rate and magnitude of DMS and CH 4 sinks rely on different degradation processes (e.g. microbial communities, vertical distribution in the water column). For instance, the magnitude of the aerobic and anaerobic CH 4 oxidation sinks is dictated by oxygen gradients and groundwater residence time, respectively (Pain et al., 2019). The different sources and sinks of DMS and CH 4 in coral reef systems thus most likely explain the divergence in correlation between DMS and CH 4 sea-to-air emissions on one hand and the RPs of DMS and CH 4 on the other hand.
At the Heron Island reef lagoon, negative correlations were found between dark DMS and CH 4 fluxes from permeable carbonate coral reef sediments to the water column (Deschaseaux et al., 2019), suggesting that part of the DMS produced in coral reef sediments was degraded to CH 4 under dark anoxic conditions, most likely by anaerobic methanogens (Kiene and Visscher, 1987). However, CH 4 production can also occur under aerobic conditions and across all living organisms, including marine phytoplankton (Klintzsch et al., 2019;Ernst et al., 2022). As such, DMS-to-CH 4 hydrolysis may not only occurs in permeable carbonate coral reef sediments, but also in reef waters through phytoplankton activity and possibly within the coral tissue, where DMS plays a major role in structuring coral-associated bacterial communities (Raina et al., 2010).
Growing evidence shows that organic compounds containing sulfur-bonded methyl groups such as DMS are a source of CH 4 under both anaerobic and aerobic conditions (Liu and Whitman, 2008;Ernst et al., 2022), with CH 4 production by certain bacteria being enhanced in the presence of reactive oxygen species under oxidative stress (Ernst et al., 2022). DMS concentrations in marine algae and corals also increase under oxidative stress due to DMS being used as an antioxidant (Sunda et al., 2002;Deschaseaux et al., 2014). This suggests that the increase in CH 4 production might be due to more sulfur-methylated compounds made readily available to methylotrophs in various types of organisms.
It was estimated that DMS contributes up to 28% of CH 4 production in various marine systems (Kiene, 1988;Florez-Leiva et al., 2013), which could well be the case for coral reef ecosystems. At the Heron Island reef lagoon, it was estimated that a small portion (~6.5%) of the CH 4 produced in coral reef sediments under dark anoxic conditions might originate from DMS (Deschaseaux et al., 2019). Here, the slope of the regression between DMS and CH 4 indicates that 0.05 mole of CH 4 is emitted for each mole of DMS being emitted to the atmosphere. This shows that sea-to-air CH 4 emissions from the Heron Island reef lagoon represents 5% of that of DMS, which is consistent with the average sea-to-air DMS (24.9 ± 1.81 μmol m -2 d -1 ) and CH 4 (1.36 ± 0.11 μmol m -2 d -1 ) fluxes reported in this study. However, without using isotopic tracers, we cannot draw conclusions on the actual portion of CH 4 emissions from coral reef systems that actually originate from DMS hydrolysis.

Drivers of DMS and CH 4 in Coral Reef Systems
In addition to pH, depth, wind direction, dewpoint and wind chill, DMS emissions were also positively driven by salinity. These findings could be counter intuitive given that the most intense plume of DMS measured over the Heron Island reef occurred in the winter of 2013 at low tide when rainfall on the aerially exposed reef apparently caused a combined hypo-salinity osmotic and hypo-thermic temperature shock to the coral (Swan et al., 2017). However, salinity in this study only ranged from 36.4 to 37.4 ppt, and thus DMS emissions under rainfall at low tide was not captured. Nevertheless, it could also be that it was not only low salinity that led to high DMS emissions reported by Swan et al. (2017), but also temperature stress and/or other raininduced environmental conditions. For instance, rain increases transfer velocity (Ho et al., 1997), which could explain the higher emissions in Swan et al. (2017) under rainfall at low tide.
Similarly, in addition to pH, depth, wind direction, dewpoint and wind chill, CH 4 emissions were also negatively affected by light whereas previous studies show that CH 4 production by phytoplankton was accentuated under increased light conditions (Klintzsch et al., 2020). Given that light mainly influences photosynthesis and O 2 production in biological systems, this data suggests that light-driven O 2 production might negatively impact CH 4 production or lead to enhanced CH 4 oxidation in coral reefs. This is consistent with sea-to-air O 2 fluxes being negatively correlated with both CH 4 and CO 2 in this study, with CO 2 being a precursor of CH 4 production (Liu and Whitman, 2008). Although CH 4 can originate from various biogenic sources in the presence of oxygen (Ernst et al., 2022), it could be that the main source of CH 4 in coral reef systems are methanogen-rich permeable carbonate sediments (Deschaseaux et al., 2019), which are productive under dark anoxic conditions.
Maximum dissolved DMS and DMSP concentrations in coral reef systems coincide with low pH, especially over areas dominated by seagrass and macroalgae (Burdett et al., 2013), which is consistent with pH negatively affecting DMS emissions in the current study. Coral-reef DMS emissions are thus expected to increase under low pH conditions, possibly due to macroalgae using DMSP to maintain metabolic functions during periods of low carbonate saturation state (Burdett et al., 2013). Sea-to-air CH 4 fluxes at Heron Island were also negatively correlated with pH, which suggests that ocean acidification may increase CH 4 emissions in coral reef systems, possibly due to low pH enhancing organic matter degradation in coral reefs like it is the case in seagrass sediments (Ravaglioli et al., 2020).
Spikes in DMS emissions at the Heron Island reef lagoon were previously detected at low tide generally under low wind speeds < 2 m s -1 (Swan et al., 2017), which agrees with depth being a negative driver of DMS emissions in this study. On the other hand, it is unknown whether dewpoint and windchill have ever been considered as environmental factors driving DMS and CH 4 emissions in marine systems, although it seems that low dewpoint and air temperatures coincided with high DMS and CH 4 emissions at the Heron Island reef system. This suggests that air temperature also plays a role in the gas transfer velocity of DMS and CH 4 , most likely as it influences water temperature and thus the temperature-dependent gas solubility of these two compounds.
DMS and CH 4 emissions were highest under southerly winds, which travelled over the Heron Island's reef and the adjacent large Wistari reef, directly south of our sampling site. This indicates that Wistari reef is possibly the major source of coral-reef-derived DMS a measured at Heron Island and that coral reefs are a more predominant source of DMS and CH 4 than the surrounding ocean.

DMS and Climate in Coral Reef Systems
Although there is insufficient information to attribute the particle counts in this study to new particle formation, because the measured size ranges are far above that of new nanoparticles (Clarke et al., 2006), it was interesting to see that sea-to-air DMS fluxes positively correlated with intermediate (0.5-2.5 μm) and large (> 2.5 μm) particles, and that a stronger correlation occurred between DMS emissions and the abundance of particles in the intermediate size range. This likely indicates that DMS oxidation products possibly contributed to the growth of existing particles within the 0.5-2.5 μm measured size range, which are suspected to be predominantly sea spray aerosols (SSA) (Quinn et al., 2015). Given that number concentrations of intermediate size particles were correlated with wind speed (R 2 = 0.3337, Figure 5C) and that wind speed is a covariate in DMS flux, the observed correlation is expected to be linked to wind speed, which exchanges more DMS and SSA from the ocean. Given that coral-reef derived DMS and CH 4 emissions could be linked, with a small portion of DMS being hydrolysed to CH 4 in the water column, and that CH 4 is a GHG with a stronger greenhouse potential than that of CO 2 (Shine et al., 2005), it is important to consider both dissolved DMS and CH 4 when predicting BVOC and GHG emissions from marine systems.

CONCLUSION
To our knowledge, this study is the first to simultaneously measure and report sea-to-air DMS fluxes alongside sea-to-air CH 4 fluxes in a coral reef ecosystem. Depth, pH, wind direction, dewpoint and wind chill were common drivers of DMS and CH 4 emissions at Heron Island, a coral reef system on the southern Great Barrier Reef. Additionally, salinity was a positive driver of DMS emissions while light negatively affected sea-to-air CH 4 fluxes. This research also showed the strong correlation that exists between DMS and CH 4 emissions at Heron Island. Although it is not possible at this stage to estimate the portion of CH 4 that derives from DMS hydrolysis, it is clear that DMS and CH 4 emissions from the Heron Island reef are intimately linked, with potential consequences on ocean warming. DMS emissions were well correlated with the abundance of intermediate size particles (0.5-2.5 μm), which indicates that DMS contributes to the growth of existing aerosol particles, which could eventually form cloud condensation nuclei and induce cloud-mediated cooling on a local scale. This study highlights the complexity of BVOC and GHG co-emissions and the potential impact they may have on the regional climate of the Great Barrier Reef.

DATA AVAILABILITY STATEMENT
The data collected for this study has been made publicly available from the Southern Cross University online repository at: https:// doi.org/https://doi.org/10.4226/47/59c460fbe8322 and https:// doi.org/10.25918/data.193.

AUTHOR CONTRIBUTIONS
ED, BE, GJ, HS and DM conceptualised the experimental design. ED conducted the DMS w analysis, DMS w data processing, all data compiling, and manuscript writing. HS conducted the analysis and processing of DMS a and particle number concentrations. DM conducted the analysis and processing of CH 4 and CO 2 data. ED and DM conducted the flux calculations for DMS and CH 4 , CO 2 and O 2 , respectively. KS conducted the MLR analysis and data processing on MATLAB. EK and KT designed the VG-CL and provided analytical support in the field. BE and GJ funded this research. All authors contributed to the article and approved the submitted version.

FUNDING
This research was funded by the ARC Discovery Project grants DP150102092 (recipient BE), "Dissolution of calcium carbonate in sediments in an acidifying ocean" and DP150101649 (recipient GJ), "The Great Barrier Reef as a significant source of climatically relevant aerosol particles".