Air exposure of coral is a significant source of dimethylsulfide (DMS) to the atmosphere

Corals are prolific producers of dimethylsulfoniopropionate (DMSP). High atmospheric concentrations of the DMSP breakdown product dimethylsulfide (DMS) have been linked to coral reefs during low tides. DMS is a potentially key sulfur source to the tropical atmosphere, but DMS emission from corals during tidal exposure is not well quantified. Here we show that gas phase DMS concentrations (DMSgas) increased by an order of magnitude when three Indo-Pacific corals were exposed to air in laboratory experiments. Upon re-submersion, an additional rapid rise in DMSgas was observed, reflecting increased production by the coral and/or dissolution of DMS-rich mucus formed by the coral during air exposure. Depletion in DMS following re-submersion was likely due to biologically-driven conversion of DMS to dimethylsulfoxide (DMSO). Fast Repetition Rate fluorometry showed downregulated photosynthesis during air exposure but rapid recovery upon re-submersion, suggesting that DMS enhances coral tolerance to oxidative stress during a process that can induce photoinhibition. We estimate that DMS emission from exposed coral reefs may be comparable in magnitude to emissions from other marine DMS hotspots. Coral DMS emission likely comprises a regular and significant source of sulfur to the tropical marine atmosphere, which is currently unrecognised in global DMS emission estimates and Earth System Models.

nubbins were attached to 10 mm plastic PVC piping plugs with a non-toxic epoxy resin (Milliput®̀ Standard). Ramets were followed throughout experimental design and selected from different genets to ensure biological replication for each treatment. All nubbins were distributed equally on holding racks receiving either 100 or 400 µmol photons m -2 s -1 respectively. It is worth noting that these light levels are significantly lower than maximum light levels in a natural reef environment, where noon irradiances can range from 700 -1200 µmol photons m -2 s -1 (1) .These two light intensities that were determined to be sub-saturating (100 µmol photons m -2 s -1 ) and saturating (400 µmol photons m -2 s -1 ) for calcification based on previous experiments 1 .
All nubbins were kept on a 12:12 light: dark cycle within an acclimation tank maintained at 26 ± 1 °C 1 . Acclimation tanks were supplied with Tropic Marin®̀ PRO REEF salt-based seawater supplemented with NaHCO 3 and CaCl 2 and circulated via a common biological sump of Fijian live rock (Tropical Marine Centre Ltd., Chorleywood, UK). Inorganic nutrient concentrations monitored every two days were undetectable throughout 1 . Alkalinity, determined from a Titrino titrator (Metrohm, Buckingham, UK), remained constant at 2.7 ± 0.2 µmol kg -1 . Nubbins were used for experimentation after 8-10 weeks of acclimation to these conditions, by which stage all were 3-8cm tall.
Determination of atmospheric DMS by API-CIMS. DMS gas in the outflow of the coral flasks were determined using: where Flow std and Flow coral are gas flow rates of the D3-DMS standard (5 mL min -1 ) and the flow out of the coral incubation vessels (~60 mL min -1 ), and Conc 66 is the D3-DMS gas standard mixing ratio (9.68 ppm, confirmed using an Eco Scientific DMS permeation tube and associated weight loss record). Sig 63 and Sig 66 represent the background-corrected API- For analysis, samples were incubated at 30°C for 24 h, and then microliter volumes (3 -40 µL) were withdrawn from each vial using a gas tight Hamilton syringe, followed by direct injection into the same GC system described above. For these measurements the GC oven was operated at 120 °C and the carrier gas flow was set to 10.56 mL min -1 (linear velocity of 80 cm s -1 ) to give an approximate DMS retention time of 1 min 3 . DMSP content is normalised by the coral surface area (CSA) (  (Table S3).

Determination of DMS concentration in coral mucus and mucus ropes.
One colony each of P. cylindrica and A. cf. horrida were removed from the aquarium and exposed to the air for 15 minutes. After 15 minutes, known volumes of mucus that had collected on the surface of the corals were taken with a pipette and transferred to 4.92 mL headspace vials, and made up to 3 mL with seawater from the aquarium (samples of seawater were used as a control).
Mucus ropes that formed upon re-submersion of A. cf. horrida were also collected and handled in the same way. Vials were vortexed for 1 min before a 200 µL headspace sample was withdrawn and injected directly into the injector of the GC-FPD 3 . The detection limit of DMS in a 200 µL headspace sample was ≤50 nM (the lowest concentration used in headspace calibrations) 3 . Note, these are not the same specimens used in the main experiments, but were sourced from the same supplier and kept under the same conditions in the same aquarium. Thus the data may not be fully representative of the experimental findings.
Additional coral variables. Additional variables were collected at the end of experimentation from all fragments. A Waterpik was used to remove a small area of tissue from each nubbin 10 ; this area was then quantified using ImageJ. Tissue slurries produced were homogenised in a known volume of filtered seawater and a small aliquot removed for cell counts via a haemocytometer. The remaining volume was filtered through Glass Fibre Filters (Whatman) and immediately extracted in 5 mL of methanol for 24h at 4°C for subsequent chlorophyll a quantification 11 . Each nubbin was processed for buoyant weight and surface area (SA) using the paraffin wax technique 12 , and surface area measurements were used to normalised all DMS, DMSP and DMSO data. Chlorophyll a and zooxanthellae density data is summarised in Table S3.  cf. horrida, before, during and after air exposure, and in the dark. Grey shaded areas = coral submersed (Stage I and III), white areas = coral exposed to air (Stage II). Symbols represent the data from four separate flasks. Flasks represented by squares and circles held coral nubbins that had been acclimated to 400 µmol photons m -2 s -1 , whilst flasks represented by triangles and diamonds held coral nubbins that had been acclimated to 100 µmol photons m -2 s -2 . Error bars (generally smaller than symbol size) represent the standard error on 5 minaveraged 0.25 Hz API-CIMS measurements (see Methods). Tables   Table S1. Summary of DMSP content normalised to surface area (nmol cm -2 ) and zooxanthellate cell counts (fmol cell -1 ) of experimental coral nubbins. Light acclimation level indicated by 100 (100 µmol photons m -2 s -1 ) and 400 (400 µmol photons m -2 s -1 ).  Seriatopora hystrix (all) 28.3 11.6 0.9 0.6 6.2 x 10 6 2.9 x 10 6