Disease prevalence and benthic composition: Stony coral tissue loss disease (SCTLD) prevalence was calculated by dividing the number of coral colonies showing active disease by the total number of colonies on the reef. Benthic composition was analyzed by using 2,500 stratified random points that were placed over the high-resolution photomosaic and classified to generate reef-wide cover estimates of calcified macroalgae, crustose coralline algae, cyanobacteria, fleshy macroalgae, hard coral, invertebrates, non-biological substrate, other, other invertebrates, seagrass, soft coral, sponge, and turf algae. Full methods for benthic coverage estimates are in Fox et al. (2019).

Flow cytometry, organic nutrient, inorganic nutrient, and chlorophyll analyses for water quality: Flow cytometry samples were processed and analyzed by the University of Hawaii SOEST Flow Cytometry Facility as described previously (Becker et al 2020). Briefly, each sample was stained with Hoechst 33342 DNA stain and excited with both 488 nanometers (nm) (1W) and UV (~350 nm, 200 mW) lasers co-linearly on a Beckman-Coulter Altra flow cytometer (Beckman Coulter Life Sciences). Signals of forward and side scatter and fluorescence were analyzed to distinguish populations and abundances (cells ml−1) of four cell types: Prochlorococcus, Synechococcus, eukaryotic picophytoplankton (picoeukaryotes), and non-pigmented bacteria. Non-pigmented prokaryotes were used as a proxy for heterotrophic bacterial and archaeal cells (Monger and Landry 1993, Marie et al. 1997).

Non-purgeable total organic carbon (TOC) samples and total nitrogen (TN) samples were run with a Shimadzu TOC-VCSH TOC analyzer (Hansell and Carlson, 2001) using a TNM-1 module. We shipped inorganic nutrient samples to Oregon State University for analysis of phosphate, ammonium, silicate, nitrite, and nitrate, as in Apprill and Rappé (2011). Briefly, samples were run on a Technicon AutoAnalyzer II (SEAL Analytical) and an Alpkem RFA 300 Rapid Flow Analyzer to generate nutrient concentrations (μM). We generated total organic nitrogen (TON) concentrations by subtracting concentrations of inorganic nitrogen (ammonium and nitrite plus nitrate) from total nitrogen.

Chlorophyll was extracted with acetone using standard methods (JGOFS, 1996). Filters were thawed individually and immediately placed in a glass test tube with 5 mL or 10 ml of 90% acetone, with 10 ml used in the case the filter appeared particularly dark, and capped. The filters were left to extract for 24 hours in the dark at 4°C. After the extraction, the tubes were vortexed and centrifuged to concentrate any particulate matter at the bottom of the tube. Prior to analysis, blanks including air, 90% acetone, and a black standard were run on an AquaFluor fluorometer (Turner Designs handheld 800446) fitted with a red-sensitive photomultiplier. Approximately 3 mL of solvent was analyzed on the fluorometer at wavelength of 664 nm, followed by acidifying the sample with two drops of 10% hydrochloric acid, then measuring again to assess phaeopigment concentration. Readings were corrected for the volume filtered and concentration of chlorophyll was measured by referencing a standard curve.

Targeted and untargeted metabolomic laboratory processing and mass spectrometry: Untargeted samples were run 2/5/2020. We eluted dissolved organic matter (DOM) from the SPE cartridges and prepared samples for analyses as outlined by Weber and colleagues (2020). To summarize, 4 bed-volumes of 0.01 M HCl were added to the cartridges to remove salt. The cartridges were then dried for five minutes, and eluted into combusted glass vials using 6 mL of 100% methanol. Extracts were frozen at -20ºC until they were dried down using a vacuum centrifuge. Extracts were then resuspended with a 95:5 (v/v) MilliQ water: acetonitrile (ACN) solution with deuterated biotin (final concentration 0.05 milligrams per milliliter (mg per mL) (200 μL total) and vortexed. Pooled samples were made for all mass spectrometry runs by combining equal volumes of all extracts into one vial. The pooled samples were used for monitoring instrument drift and run quality. After preparation, all extracts were stored at -20◦C until analysis. For targeted metabolomics, 100 μL aliquots of each extract was placed in a separate vial with a combusted glass insert. For the untargeted metabolomics analysis, 600 μL of the deuterated biotin standard and water: ACN solution was used to dilute a 25 µL aliquot of each extract. Untargeted metabolite analysis was performed using an ultrahigh-performance liquid chromatography system (Vanquish UHPLC, Thermo ScientificTM) coupled with an Orbitrap Fusion Lumos Tribid mass spectrometer (Thermo ScientificTM). A Vanguard pre-column and Waters Acquity HSS T3 column (2.1 mm × 100 mm, 1.8 μm), was used for chromatographic separation at 40ºC. The column was eluted at 0.5 mL per minute with the following solvents: A) 0.1% formic acid in water and B) 0.1% formic acid in ACN. The chromatographic gradient was: 1% B for 1 min, 15% B for 1-3 minutes, 50% B for 3-6 minutes, 95% B for 6-9 minutes, and 95% B for 10 minutes. Between injections, the column was washed and re-equilibrated with 1% B for 2 minutes. Individual autosampler injections (5 μL each) were made for negative and positive ion mode analyses. In negative ion mode, the electrospray voltage was set to 2600 volts (V). Settings for source gases were 55 (sheath), 20 (auxiliary), and 1 (sweep) in arbitrary units. The temperatures of the heated capillary and vaporizer were 350ºC and 400ºC, respectively. MS data were collected in the Orbitrap analyzer with a mass resolution of 120,000 FWHM at m/z 200. The automatic gain control (AGC) target was 4e5, with a 50 sec maximum injection time, and a scan range of 100 to 1000 m/z. Data-dependent MS/MS spectra were collected at 7,500 resolution in the Orbitrap analyzer. Parent ions were isolated with a 1 m/z width in the quadrupole, and fragmented with a HCD (higher energy collisional dissociation) energy of 35%. All data were collected in profile mode. Samples were run in random order and after every seven samples, a pooled sample was run. Raw data files from the instrument were converted into mzML files using msConvert and then processed using XCMS (Smith et al., 2006). Peak-picking was performed with the CentWave algorithm and a Gaussian fit with the following parameters: noise = 10000, peak-width = 3-15, ppm = 15, prefilter = c(2,168.600), integrate = 2, mzdiff = -0.005, snthresh = 10. Retention times were then adjusted using Orbiwarp and correspondence between the peaks was conducted. The coefficient of variation across the eight untargeted pooled sample features was 0.044, demonstrating good agreement between the pooled samples, and the pooled samples were removed from further analyses. Untargeted peak intensities were normalized by dividing the peak intensities by the total seawater volume. In the untargeted analysis, only MS1 features were analyzed, and were defined as unique combinations of mass-to-charge ratios (m/z) and retention times (RT). This analysis yielded a table of MS1 features (m/z x RT) and their peak intensities across each sample.

Extracts prepared for targeted metabolomics were run on a triple-stage quadrupole mass spectrometer (TSQ Vantage, Thermo Fisher ScientificTM) using UPLC (Accela Open Autosampler and Accela 1250 Pump, Thermo ScientificTM) coupled to a heated electrospray ionization source (H-ESI) and operated in selective reaction monitoring (SRM) mode. The same chromatography column, conditions, gradient, and flow rates were used for targeted analyses as those described for untargeted analyses. Separate autosampler injections of 5 μL each were made for positive and negative modes. Additionally, as with the untargeted analysis, samples were run in random order and pooled samples were run every seven samples. SRM parameters were optimized for each compound using a standard as described in Kido Soule et al. (2015) and two SRM transitions (precursor-product ion pairs) were monitored for quantification and confirmation. Target metabolites included compounds found in central carbon metabolism and metabolites that are environmentally relevant in marine habitats or are produced by marine microorganisms (Fiore et al. 2015, Fiore et al. 2017, Kido Soule et al. 2015). The resulting XCalibur raw files (MS/MS data) were converted into mzML files using msConvert (Chambers et al., 2012) and processed with the open-source program El-MAVEN (v.774)(Agrawal et al., 2019) . Using El-MAVEN, 8-point calibration curves based on integrated peak area were generated for each compound. Environmental concentrations of metabolites were determined by dividing each concentration by the original sample collection volume. Next, metabolites that passed the limits of detection and quantification for the UPLC-MS/MS analysis (Kido Soule, Longnecker, Swarr, unpublished) were corrected for extraction efficiency based on published data for each metabolite in seawater (Johnson et al., 2017).

DNA extraction and sequencing for 16S rRNA and shotgun metagenomes: We extracted DNA from 25 mm filters used for 2L seawater collections using Qiagen PowerBiofilm kits (Qiagen, Germantown, MD, USA). To begin, we added the filter directly to the bead tube, then proceeded with the extraction following manufacturer protocols. We also included four DNA extraction controls by extracting DNA from unused filters. Resulting DNA from these extractions was used as the template for both 16S rRNA gene sequencing and shotgun metagenomic sequencing.

For 16S rRNA gene sequencing of bacteria and archaea, we included 2 μL of template DNA into a 50 μL (total volume) PCR reaction. We added a PCR negative control by including one PCR reaction with 2 μL of PCR grade H2O instead of DNA. Earth Microbiome Project primers, 515F (Parada et al., 2016) and 806R (Apprill et al., 2015), were used to amplify the V4 region of the small subunit (SSU) rRNA gene in bacteria and archaea and included sample-specific barcodes with an 8 bp barcode, 10 bp pad, and 2 bp link, similar to Kozich et al. (2013). The 50 μL reactions were diluted in UV-sterilized nuclease-free water and contained 2.5 units of GoTaq DNA Polymerase (Promega, Madison, WI, USA), barcoded primers at 0.2 μM, 0.2 mM dNTP mix (Promega), 2.5 mM MgCl2, and 1X colorless GoTaq flexi buffer (Promega). The reactions were run on a Bio-Rad Thermocycler using the following criteria: denaturation at 95°C for 2 minutes; 28 cycles at 95°C for 20 seconds, 55°C for 15 seconds, and 72°C for 5 minutes; and extension at 72°C for 10 minutes. We used gel electrophoresis to verify successful amplification using 5 μL of product on a 1% agarose-Tris-borate-EDTA (TBE) gel stained with SYBR Safe gel stain (Invitrogen, ThermoFisher Scientific). We used the QIAquick 96 PCR Purification Kit (Qiagen) with the QIAvac 96 (Qiagen) and vacuum pressure to purify the remaining 45 μL of PCR products following manufacturer’s protocols. We applied the HS dsDNA assay on the Qubit 2.0 fluorometer (ThermoFisher Scientific) to quantify the DNA concentrations then converted to nM assuming an average library size of 450 bp, and average molar mass of DNA nucleotides of 660 g/mol. We diluted individual barcoded PCR products to 10 nM, pooled all samples, and shipped the pooled, ready-to-run library to the Georgia Genomics and Bioinformatics Core at the University of Georgia for sequencing on an Illumina MiSeq using paired-end 250 bp sequencing.

We prepared a library for shotgun metagenomic sequencing following the Illumina DNA Prep Reference Guide (Illumina, San Diego, CA, USA, Document # 1000000025416 v09 June 2020). DNA input for all samples was between 100-500 ng. Concentration of the four DNA extraction control samples was below detection, so we included 30 μL of each in the procedure and processed in the same way as all seawater samples. We used IDT for Illumina DNA/RNA UD Indexes Set A, Tagmentation (Illumina, 96 samples, Cat # 20027213), to apply sample-specific indices to each sample. Following the procedure, we eluted samples in 30 μL resuspension buffer. To verify successful processing, we used a fluorometric assay (HS dsDNA) on a Qubit 2.0 fluorometer to measure DNA concentrations of a subset of samples. All final concentrations were greater than 4 ng/μL, and therefore deemed sufficient for pooling. All samples were pooled, and the final concentration was 5.30 ng/μL. The final library was run at the Georgia Genomics and Bioinformatics Core at the University of Georgia on an Illumina NextSeq 2000 with the P3 flow cell and paired-end 150 bp sequencing.

On-ship near-coral seawater sample processing and sequencing: To expedite the turnaround time between sample collection and sample processing, we performed on-ship DNA extraction, PCR, and sequencing with the Illumina iSeq 100 System on near-coral seawater samples following methods for in-the-field microbiome preparation described previously (Becker et al., 2021). Seawater samples targeted for on-ship sequencing included those sampled from 5 reefs over two days (June 9-10, 2019).

DNA extraction and sequencing of near-coral seawater and coral microbiomes: All processing of near-coral seawater and coral tissue slurries proceeded as described previously to identify bacteria and archaea within each environment (Becker et al., 2021). Briefly, we extracted DNA from all seawater and tissue samples using the DNeasy PowerBiofilm kit (Qiagen). PCR occurred in a two-stage procedure. In stage one, we used Earth Microbiome Project primers, 515F (Parada et al., 2016) and 806R (Apprill et al., 2015), to amplify the V4 region of the small subunit (SSU) ribosomal RNA gene of bacteria and archaea. In stage two, we attached unique index primers to each sample using the Nextera XT v2 set A kit (Illumina). For PCR that occurred at the Woods Hole Oceanographic Institution, we used larger benchtop centrifuges (Eppendorf 5418) and thermocyclers (Bio-Rad), rather than the small and portable versions used on the M/V Alucia. Following purification of stage two PCR products, we diluted and pooled samples such that we included approximately 40 samples. Seawater and tissue samples were randomized across all library pools. Pooled libraries were diluted to approximately 90 picomolar (pM), and a 10% PhiX Control v3 (Illumina) spike-in was added to increase base diversity. All libraries were run on the Illumina iSeq 100 System (Illumina) with the i1 cartridge pack, over a total of 6 sequencing runs.