Fieldwork was conducted in the northern coast of the island of Mo’orea, French Polynesia (17°29’00.0”S 149°50’00.0”W), between 12th and 20th of April 2018. Surface water was collected at two sampling sites: one located 4 km offshore in the open ocean (OO) over a bathymetric depth of 1200 m; and the other located in the back-reef (BR) lagoon, over a depth of ∼2.5 m. The northern reefs of Moorea have shallow depths (0.5-3 m; Leichter et al., 2013) and are characterized by patches of Pocillopora spp., Acropora spp. and other corals on a sandy bottom, and partly covered by the brown seaweed Turbinaria ornata (Masdeu-Navarro et al., 2022).

At each of the two sites, seawater samples were collected every 6 h over a period of 34 h on April 12-13 (BR) and April 19-20 (OO) from the surface (0-50 cm) using a small boat. Sampling times were 04:00, 10:00, 16:00, 22:00, 04:00, 10:00, and 14:00 (13:00 at OO) local time. Logistical and personnel limitations prevented extending the sampling to another 24-h cycle, which would have been desirable to ensure that the observed patterns were repeated. For VOCs, DMSPCs, and dissolved organic carbon and nitrogen (DOC, DON), samples were collected using acid-cleaned 0.5 L glass bottles rinsed three times and filled to the brim by hand. For other biogeochemical variables, cell counts and 16S and 18S genes rRNA metabarcoding, water was withdrawn with a peristaltic pump into an acid-cleaned and pre-rinsed Teflon-lined plastic carboy (20 L), while filtered through a 200 μm mesh to remove large particles and organisms. Samples were maintained in a water bath continuously flushed with surface seawater, in dimmed light, until processing in the UC Berkeley Gump laboratory on the island ca. 1 h after collection.

Dissolved VOCs were quantified within 3 h of sample collection by purge and trap gas chromatography-mass spectrometry as described previously (Masdeu-Navarro et al., 2022). Target VOCs were COS, DMS, CS₂, DMDS, isoprene, CH₃I, CH₂ClI, CH₂Br₂, CHBr₃. Each sample was analyzed in duplicate. Limits of detection for a 25 ml sample, including correction for blanks, were 2 pM COS, CS₂, DMDS and isoprene, 15 pM DMS, 0.6 pM CH₃I, 0.2 pM CH₂ClI, 1.5 pM CH₂Br₂ and CHBr₃. For total DMSP (DMSP_t) 35% HCl (10 mL per mL of sample) was added to unfiltered samples followed by storage at room temperature in the dark. To determine dissolved concentrations of DMSP (DMSP_d), acrylate and DMSO, 15 ml sample aliquots were gravity filtered using precombusted 25 mm diameter Whatman glass-fiber (GF/F) filters into 20 ml scintillation vials. Filtrates were microwaved to boiling, bubbled with high-purity nitrogen gas to remove DMS, and acidified (Kinsey and Kieber, 2016). Back in our home lab, DMSP_t and DMSP_d were determined as evolved DMS by purge and trap gas chromatography with flame photometric detection, after alkaline hydrolysis. DMSO was determined as evolved DMS after reduction with TiCl₃. Acrylate concentrations were determined using a pre-column derivatization HPLC method (Tyssebotn et al., 2017). The particulate forms of the three compounds (DMSP_p, DMSO_p and acrylate_p) were determined by subtracting the dissolved from the total form. All analyses were run in duplicate.

The sea surface temperature (SST) was recorded on the boat with an SBE56 sensor (Sea-Bird Sci.) continuously flushed with pumped surface seawater.

For determination of chlorophyll a (Chla) concentrations, duplicate 250 mL aliquots of unfiltered seawater were taken from the sample carboys, filtered through 25 mm diameter GF/F filters and stored at -20°C until analysis. Chla extraction was performed in 90% acetone at 4°C for 24 h. The fluorescence of the extracts was measured with a calibrated Turner Designs fluorometer (model 10-AU-005 field fluorometer) equipped with an excitation and emission filters at 340-500 nm and above 665 nm, respectively.

For particulate organic carbon (POC) and particulate organic nitrogen (PON) analyses, 1.5 L seawater aliquots were taken from the carboys, filtered through precombusted (450°C for 4h) GF/F filters and stored frozen at -20°C until analysis. Filters for POC determination were decarbonated with acid vapor (Yamamuro and Kayanne, 1995). No replicates were analyzed. Carbon and nitrogen were determined with an elemental analyzer (Perkin-Elmer 2400 CHN). For the determination of total organic carbon (TOC) and total nitrogen (TN), unfiltered seawater aliquots (30 mL) were collected in acid-cleaned polycarbonate bottles and immediately stored at -20°C until analysis. Carbon and nitrogen were determined after inorganic C removal through acidification using a Shimadzu TOC VCSH instrument. Analytical quadruplicates were run. The equipment was calibrated with potassium hydrogen phthalate. High-purity laboratory water obtained from a MilliporeSigma Milli Q system (MilliQ water) was used as a blank and the reference material used was deep Sargasso Sea water (MRC Batch-15 Lot//11-15, measured TOC: 43.2 ± 1.1 µM, Dr. Dennis Hansell Laboratory, University of Miami, RSMAS). Dissolved organic carbon (DOC) concentrations were calculated by subtracting POC concentrations from TOC. Samples (10 mL) for inorganic nutrient determination were collected unfiltered and stored at -20°C until analysis. Concentrations of nitrate (NO₃ ^-), nitrite (NO₂ ^-), ammonia (NH₄ ⁺), phosphate (PO₄ ^3-) and silicate (SiO₄ ^2-) were determined with an auto-analyzer (Bran Luebbe AA3) with spectrophotometric detection (Grasshoff, 1978). No replicates were analyzed. Dissolved organic nitrogen (DON) concentrations were calculated by subtracting inorganic nitrogen (nitrate, nitrite and ammonia) and PON concentrations from TN.

Characterization of fluorescent dissolved organic matter (FDOM) was performed with a Horiba Aqualog spectrofluorometer. Briefly, ca. 4 mL seawater aliquots were filtered through pre-combusted (450°C for 4 hours) GF/F filters. Fluorescence was recorded as square excitation-emission matrices in the 240-600 nm range. One to four analytical replicates were run. The intensity of peak T (hereafter referred to as FDOM-T), as defined by Coble (1996), was extracted as the fluorescence at excitation/emission wavelengths 275/340 nm using the staRdom package (Pucher et al., 2019). Fluorescence intensity is presented in Raman Units (RU) after normalization to the Raman scatter measured in MilliQ water blanks (Lawaetz and Stedmon, 2009).

For the enumeration of heterotrophic prokaryotes (including bacteria and archaea) and pico- and nano-phytoplankton, 2-5 mL sample aliquots were fixed with glutaraldehyde (0.5%) and stored at −80°C until analysis based on size and fluorescence on a flow cytometer capable of true volumetric absolute counting (CyFlow Cube 8, Sysmex Partec). Heterotrophic prokaryotes were stained with SYBRgreen I (∼20 µM final concentration) prior to quantification using green fluorescence. Prokaryotic (i.e., Synechococcus and Prochlorococcus) and eukaryotic pico- and nanophytoplankton were counted based on red and orange autofluorescence (Gasol and del Giorgio, 2000). No replicates were analyzed.

For bacterial biomass production estimates, 1 L of seawater was collected in acid cleaned amber bottles and stored at ambient temperature until further processing. The biomass production of the bacterial community was determined as described in Fadeev et al. (2023), based on the single-cell incorporation of L-homopropargylglycine (HPG) into newly synthesized bacterial proteins (Samo et al., 2014). Briefly, seawater samples were amended with 50 μM HPG (final concentration of 20 nM) and incubated at in situ temperatures in the dark for 6 h. Afterwards, samples were fixed with formaldehyde (2-4% final concentration) at 4˚C for at least 1 h in the dark and stored at -20˚C until further processing. Prior to the microscopic analysis, the samples were stained with 2 µg mL-1 4′,6-diamidino-2-phenylindole (DAPI) in Vectashield (Vector Laboratories, Newark, CA, USA). The cells were enumerated using a Zeiss Axio Imager M2 epifluorescence microscope (Carl Zeiss AG, Oberkochen, Germany) at 1250× magnification and the DAPI (Ex/Em = 358/461 nm) and the FITC (Ex/Em = 495/519 nm) filter sets. The bacterial abundance was calculated based on the average number of cells from at least 20 counting fields with 20-200 cells enumerated per counting field. At least 20 fields were counted for each filter slice using the Automated Cell Measuring and Enumeration Tool (ACMETool2, M. Zeder, Technobiology GmbH, Buchrain, Switzerland). The total abundance of biomass producing cells was determined as simultaneous signal of DAPI and FITC channels.

Global solar radiation (W/m²) data were provided by the meteorological station at the Gump Research Station (Washburn and Brooks, 2022), located 1.5 km and 5 km away from the sampling sites BR and OO, respectively.

Single seawater aliquots (2 L, no replicates) for DNA collection were taken from the sampled carboys and filtered through 47 mm diameter polycarbonate filters of 0.2 μm pore size using a peristaltic pump. DNA filters were flash-frozen in liquid nitrogen and subsequently stored at −80°C. DNA extractions were performed using a standard phenol chloroform protocol (Massana et al., 1997) with a final step of purification using ultrafiltration in Amicon units (Millipore). Prokaryotic and eukaryotic diversity was determined by amplicon sequencing of the V4/V5 and V4 regions of the 16S and 18S rRNA genes, respectively, using the Illumina MiSeq platform and paired-end reads (2 × 250 bp). PCR amplifications were performed using (1) the prokaryotic universal primers 515F-Y (5’-GTGYCAGCMGCCGCGGTAA-3’) and 926R (5’-CCGYCAATTYMTTTRAGTTT-3’) (Parada et al., 2016) and (2) the eukaryotic universal primers V4F (5’-CCA GCA SCY GCG GTAATT CC-3’) and V4R (5’-ACTTTC GTT CTT GAT YRR-3’) (Balzano et al., 2015). All samples were sequenced at Research and Testing Laboratories (RTL, Lubbock, TX, USA) using the Illumina MiSeq platform (2 × 250 bp paired‐end sequencing).

Illumina reads were processed as described in Masdeu-Navarro et al. (2022). Briefly, raw reads from both 16S and 18S rRNA gene sets were trimmed to remove amplification primers and spurious sequences using cutadapt v2.3 (Martin, 2011) and subsequently processed with DADA2 v1.4 (Callahan et al., 2016) to differentiate the 16S/18S rRNA gene amplicon sequence variants (ASVs) and to remove chimeras. ASVs were taxonomically assigned using the Ribosomal Database Project naïve Bayesian classifier (Wang et al., 2007), as implemented in DADA2, and an 80% minimum bootstrap confidence threshold using SILVA (v132; (Pruesse et al., 2007)) and PR2 (v4.11.1; (Guillou et al., 2013)) as reference databases for the 16S and 18S rRNA sets, respectively. Singletons and sequences affiliated to eukaryotes, organelles, or chloroplasts (for the 16S rRNA set) or to metazoans, Embryophyceae, Rhodophyta, Ulveophyceae or Phaeophyceae (for the 18S rRNA set) were removed.

To compare samples, ASV tables were randomly subsampled down to the minimum number of reads per sample for both rDNA sets (17,694 and 3,203 reads for the 16S and 18S rDNA sets, respectively) using the rarefy function in the vegan v2.5.7 package (Oksanen et al., 2021) in R v4.0.2 (R Development Core Team, 2021). The final ASV tables contained 2,018 prokaryotic ASVs and 1,473 protistan ASVs.

To gain insight into the diversity and temporal variation of the phototrophic eukaryotic assemblage (namely, autotrophs and mixotrophs), protistan ASVs were classified into four major functional groups on the basis of their taxonomic affiliation: autotrophs (obligate phototrophs), heterotrophs (mainly predators), mixotrophs, and parasites (including saprobes), based on previous works relative to the annotation into functional traits of protistan diversity (Genitsaris et al., 2015; 2016; Ramond et al., 2018; 2019; Minicante et al., 2019). ASV assigned at poor taxonomic resolution (e.g., unclassified Dinophyceae) or to a family/order of organisms that include both autotrophs and heterotrophs (e.g., Gymnodiniaceae) were not attributed to any specific functional groups and were annotated as ‘not determined (ND)”. A detailed list of the classification for each protistan taxa is available in Supplementary Table S1 .

Statistical difference between the data series from both sampling sites (BR and OO) was evaluated with a non-parametric Kruskal-Wallis test. Spatial differences between the microbial community structure at each site obtained from the metabarcoding gene analysis were visualized using nonmetric multidimensional scaling (NMDS) based on Bray-Curtis distances and were obtained using the metaMDS function of the vegan R package. To investigate the relationships between VOCs and DMSPCs, and between these compounds and solar radiation, pairwise Spearman’s correlations were computed with the ggcor v0.9–7 package in R (Houyun et al., 2020). Due to the short data series used (n=7), the threshold for correlation significance was set at p<0.1.

A total of 2,018 prokaryotic and 1,473 protistan ASVs were retrieved over the diel cycles at BR and OO ( Figure S1A ). Rarefaction curves for both the 16S and 18S rRNA datasets and sampling time points approached a plateau in most cases, indicating that a substantial portion of microbial diversity was captured in each sample ( Figure S1B ). Clustering of the microbial community by their Bray-Curtis dissimilarities using the combined 16S and 18S ASV datasets revealed a clear separation of the two sampling sites ( Figure 3A ). BR and OO shared ∼25% of the total prokaryotic ASVs and ∼40% of the total eukaryotic ASVs ( Figure S1A ). Furthermore, 70% of the prokaryotic ASVs from BR were specific to the site. In BR and OO, an average of ∼57% and 37% of the prokaryotic ASVs corresponded to cyanobacterial taxa, respectively ( Figure 3B ). Among the cyanobacteria, the most abundant genera were Synechococcus, (80-99% predominant at BR) and Prochlorococcus (35-92% at OO), confirming the results of flow cytometry ( Figures 2 , 3B ). The main heterotrophic prokaryotic components in BR were from the order Flavobacteriales (Bacteroidetes), representing 16-32% of the total ASVs, and the coral-reef characteristic bacteria Candidatus Actinomarina (Apprill et al., 2016), accounting for 13-22% of the prokaryotic ASVs ( Figure 3C ). To a lesser extent, sequences of alphaproteobacterial clades (such as Rhodobacteriales, SAR11 and SAR116) and Planctomycetes (Pirellulaceae) together represented 24-40% of the total prokaryotic diversity. Essentially the same main taxonomic groups were retrieved in OO despite the low share of ASVs with BR ( Figure 3C ). Yet, a change in the dominance was observed, with Candidatus Actinomarina being the most abundant (20-36%), followed by Flavobacteriales (19-26%), SAR11 (7-23%) and Rhodobacteriales (9-16%).

Regarding the eukaryotes, the relative abundance and diel variation of the protistan assemblages are shown in Supplementary Table S1 . The phototrophic eukaryotic assemblages (i.e., autotrophs and mixotrophs) were similar at both sites, with abundant sequences affiliated to dinoflagellates and haptophytes ( Figure 3D ). The largest difference between the two sites was the presence of the green algae Mamiellales (specifically, Micromonas sp.) at BR, where they represented up to 53% of the eukaryotic phytoplankton sequences, and their virtual absence at OO (<2.5%). Other groups present at BR and absent at OO were Cryptomonadales and diatoms.

The low dispersion of the data clouds in the NMDS ordination ( Figure 3A ) suggests that the microbial diversity was relatively stable over time at both sites, which was confirmed when examining the taxonomic compositions ( Figures 3B–D ). In conclusion, our metabarcoding analysis confirmed the presence of two distinct bacterioplankton and phytoplankton assemblages at BR and OO, each with a rather stable composition over the diel cycle.

For the non-volatile DMSPCs (DMSP, acrylate, DMSO), there were no remarkable differences in the partitioning between the particulate and dissolved pools at the two sampling sites ( Figures 4A–D, F, G ). DMSP was 5–19% dissolved at BR, and 8– 25% at OO. Acrylate was 58–92% dissolved at BR, and 74–95% at OO. DMSO was 88–100% dissolved at BR, and 89-97% at OO. The particulate concentrations of the three compounds were similar at the two sites (Kruskal-Wallis p>0.05), which could be coincidental or indicative of rapid connectivity through the water flow (Hench et al., 2008; Leichter et al., 2013). In contrast, the dissolved concentrations of DMSP and DMSO were slightly higher at OO (Kruskal-Wallis p<0.05), indicative of higher microbial DMSP_d consumption at BR (Xue et al., 2022) and higher photochemical DMSO_d production at OO (see below). The concentrations of the volatile DMSPCs, namely DMS and DMDS, were higher at BR (Kruskal-Wallis p<0.01) by factors of 1.3-3 and 1.2->20, respectively. Note that we interpret the occurrence of DMDS in the Mo’orean seawater chromatograms as a non-quantitative reflection of the presence of methanethiol (MeSH), since the high temperatures and activated carbon of our purge and trap system are expected to partly oxidize MeSH to DMDS during the analysis (Cheng et al., 2007). Indeed, DMDS concentrations in OO (0-10 pM) were much lower than the few existing measurements of MeSH in low-Chla tropical waters of the Atlantic (100-300 pM; Kettle et al., 2001). DMS concentrations in OO (0.6-1 nM) were near the lower end of previous measurements in the tropical oligotrophic oceans (1-5 nM; Dani and Loreto, 2017). The higher concentrations of both compounds in BR could be due to non-planktonic sources of DMS and MeSH within the reef, such as DMSP degradation in coral holobionts (Masdeu-Navarro et al., 2022), seaweeds (Burdett et al., 2013), and sediments (Deschaseaux et al., 2019).

Non-volatile DMSPCs exhibited diel patterns, however these patterns were different between compounds and sites. At OO, particulate DMSP (DMSP_p) increased during the day and decreased at night, following solar radiation ( Figure 4A ). Even though DMSP_p is known to be associated with phytoplankton (Stefels et al., 2007 and references therein), its diel pattern did not follow size and fluorescence-based phytoplankton abundances ( Figure 2 ) or Chla concentrations ( Figure 1B ). The closest pattern of a biogeochemical variable was that of POC ( Figure 1K ); however, the DMSP_p/POC ratio increased during the day, somehow amplifying the diel cycle observed with POC ( Figure 5A ). This suggests that intracellular DMSP in the open ocean varies partly with phytoplankton biomass and partly with photophysiology, so that DMSP biosynthesis increases with light (Simó et al., 2002). In the tropical Atlantic, Archer et al. (2018) measured higher rates (d^-1) of DMSP synthesis than rates of carbon fixation into POC at all irradiances, an expected result since a substantial proportion of POC is detrital and heterotrophic biomass, while most DMSP is contained in phytoplankton. This may explain the amplitude of the DMSP_p/POC ratio over the diel cycle at OO. Another example of a molecule that varies with both biomass and photophysiology is Chla, whose cellular concentration varies over diel cycles (Owens et al., 1980). At OO, Chla varied with an opposite pattern to that of POC, so that the Chla/POC ratio increased at night and decreased during the day ( Figure 5B ). At BR, the DMSP_p diel trend was virtually opposite to that at OO, with a nighttime peak, even though the POC diel pattern was the same at both sites ( Figure 4A ).

A Spearman’s correlation analysis ( Figure 6 ) confirmed that DMSP_p covaried significantly and positively with solar radiation at OO but not at BR. This discrepancy might be because, in our Mo’orean reef lagoonal waters, DMSP_p sources not only include plankton but also corals, seaweeds and sediment debris. In support of this, the 16S rRNA metabarcoding analysis ( Figure 3C ) revealed abundant sequences of Planctomycetes and Verrucomicrobia, typically associated with macroalgae and sediments. However, DMSP_p concentrations (1.4-15.9 nM) were lower at BR over most of the diel cycle, even though Chla concentrations and phytoplankton abundances were higher, and the aforementioned contribution from debris was expected; therefore, ecophysiological explanations are to be invoked. First, the relative abundances of protistan taxonomic groups, as described by metabarcoding with the 18S rRNA gene ( Figure 3C ), depicted higher abundances of haptophytes and dinoflagellates at OO, especially Gymnodiniales and Prorocentrales that are reported to contain moderate to very high cellular DMSP concentrations (Keller et al., 1989) and contribute to high DMSPC concentrations in the surface ocean (Saint-Macary et al., 2023). Indeed, microscopy counts revealed that the biomasses of coccolithophores and dinoflagellates, two groups that are prolific DMSP producers, were higher at OO than at BR by a factor of 4 (Masdeu-Navarro et al., unpublished). Second, back reef waters contained 5 to 10-fold higher concentrations of nitrate ( Figure 1C ), likely due to microbial nitrification exacerbated by microbes associated with sediments and benthic filter-feeders (Scheffers et al., 2004; O’Neil and Capone, 2008);. Groundwater inputs cannot be rule out either. Relieving acute nitrate limitation has been reported to decrease phytoplankton intracellular DMSP_p concentrations (Bucciarelli and Sunda, 2003).

DMSO_p concentrations were quite low (0.1-0.6 nM) at both sites. This was unexpected, since a previous meta-analysis had shown that the intracellular DMSO pool relative to DMSP typically increases towards the high seawater temperatures observed during our study (Simó and Vila-Costa, 2006). Furthermore, DMSO_p exhibited a diel peak around midnight ( Figure 4F ). The hypothesis of an antioxidant role for DMSP under high irradiances (Sunda et al., 2002) would imply higher intracellular DMSO production rates during the day, but whether this should result in intracellular accumulation is unknown. Correlation analysis ( Figure 6 ) did not show covariation of DMSO_p with any potentially regulatory variables, and too little is still known about this cellular component to provide a unequivocable explanation for the low levels and daily pattern of DMSO_p observed in Mo’orean waters. Particulate acrylate (acrylate_p) followed diel patterns ( Figure 4C ) that somewhat paralleled those of DMSP_p at both sites, which was expected since the latter is precursor of acrylate through intracellular cleavage (Kinsey and Kieber, 2016). Dissolved concentrations of DMSP (DMSP_d) and acrylate (acrylate_d) varied opposite to their particulate pools in OO, with a maximum at night likely attributed to intracellular release through grazing-driven mortality ( Figures 4B, D ). At BR, acrylate_d and DMSO_d were also opposite to their particulate pools, but DMSP_d paralleled DMSP_p. We note that DMSO can also be produced through cleavage of the recently-discovered intracellular component DMSOP (Thume et al., 2018), which unfortunately was not measured. A variety of microorganisms, including the microalgae haptophytes and dinoflagellates and the bacteria SAR11 and Rhodobacteriales, are capable of cleaving DMSOP to DMSO and acrylate using the same lyases for DMSP (Carrión et al., 2023).

As for the volatile DMSPCs, the most striking feature was the pronounced diel patterns of DMS and DMDS concentrations at BR ( Figures 4E, J ). Both compounds decreased at night and increased during the day, with a comparatively lesser increase on the second, cloudy day. This resulted in a significant positive correlation of DMS to solar radiation ( Figure 6A ). This lends support to the hypothesis that DMS is produced by and released from planktonic and benthic microalgal cells (including coral-associated Symbiodiniaceae) in response to high irradiance or due to its antioxidant properties (Sunda et al., 2002; Slezak et al., 2007; Galí et al., 2013a, Galí et al., 2013b; Lawson et al., 2021; Masdeu-Navarro et al., 2022). Alternatively, the energy of midday incident radiation can easily exceed the photosynthetic electron transport capacity, conditions under which DMSP can be released, as such or upon transformation into DMS, as a strategy to dissipate excess energy (Stefels, 2000; Kinsey et al., 2023). The hypotheses of oxidative stress and excess energy dissipation are not mutually exclusive, and support to the latter is given by the fact that DMDS (hence MeSH) behaved the same as DMS (positive Spearman’s correlation, Figure 6A ), suggesting DMSP release by phytoplankton as well as corals (Masdeu-Navarro et al., 2022) and rapid bacterial degradation to DMS, acrylate_d and MeSH. Note that the prokaryotic assemblages at the two sites ( Figure 3C ) contained abundant taxa within Rhodobacteriales, SAR11, SAR116 and gamma-proteobacteria that are known to harbor ddd- genes for DMSP cleavage into DMS and acrylate, as well as taxa within SAR11, Rhodobacteriales, SAR116 and other alpha-proteobacteria harboring the dmdA gene that eventually leads to MeSH production (González et al., 2019; Tang and Liu, 2023). Acrylate_d, in turn, would be quickly consumed by bacteria (Xue et al., 2022), and MeSH would be both biologically consumed and (photo)chemically oxidized to DMDS. In line with this, bacterial activity at BR was higher during the day ( Figure 2B ) and so were bacterial DMSP_d and acrylate_d consumption rates (Xue et al., 2022).

In reef lagoonal waters of the Great Barrier Reef, Broadbent and Jones (2006) reported diurnal increases in DMS and DMSP_p concentrations, which they attributed to a physiological response of coral holobionts to increased light and temperature, including the use of these compounds by endosymbiotic Symbiodiniaceae to scavenge reactive oxygen species (Jones and King, 2015), and the diurnal expulsion of endosymbionts by the coral. However, short-term variation of DMSPCs was also largely affected by tides (Broadbent and Jones, 2006; Jones et al., 2018). Higher DMS concentrations coincided with low tides, and the greatest effect occurred when corals were exposed to air at very low tides. DMS production was further exacerbated if a low tide coincided with rainfall. The effect of tides on coral DMS production is explained by the stress that low tides impose on corals, more so if corals are exposed to air. In our Mo’orean reef, though, conditions were very different: tides were semi-daily (low tide around 6:00 and 18:00) and weak, with an amplitude of ca. 30 cm, and corals were not exposed to air. Our results ( Figures 4E , 6A ) suggest that solar radiation was the main driver of diel DMS variability in the reef.

The daytime increase in DMS was not observed at OO, probably because DMS was released at lower rates and concentrations were capped by DMS photolysis (Galí and Simó, 2015). In agreement with this interpretation, the concentrations of DMSO_d, a major product of DMS photooxidation (Hopkins et al., 2023), were higher at OO and sightly increased in the morning ( Figure 4G ).