Location description:
Water samples were collected between 30 May and 6 June, 2018 at various sites around Puerto Rico (blue squares in figure above). Sampling was concentrated in Fajardo at Laguna Grande, a 50 hectare saltwater lagoon located in the Las Cabezas de San Juan Nature Reserve in the northeast corner of Puerto Rico (Detailed map in figure above). Laguna Grande has an average depth of 3 m, a maximum depth of 5 m, and is connected to the Atlantic Ocean at the beach in Las Croabas, Fajardo by a shallow 1.5 km channel (USGS). The lagoon is surrounded by mangrove swamps, tidal flats and brackish lagoons. It has been estimated that Laguna Grande flushes 13% of its water volume during every tidal cycle on average (USGS), corresponding to an average flushing rate of once every 7.7 days. Sargassum wracks were inundating both Laguna Grande and Las Croabas when sampling began, so no samples were collected before the event. However, because Sargassum was beaching in Las Croabas and entering the lagoon through the channel, it was expected that the entire lagoon would be impacted by Sargassum regardless of sampling time during the tidal cycle.
Sampling and analytical procedures:
At various sites throughout the lagoon and Las Croabas and offshore near Salinas (Figure above), samples for DOM characterization were collected in 10 L low density polyethylene cubitainers that had been previously cleaned with 0.1 N NaOH and rinsed several times with ultrapure MilliQ water. All containers were rinsed at least three times with sample before collection. Samples for DOM characterization were filtered through 0.7 µm GF/F filters (Whatman ®) that had been previously combusted for 4 h at 500 °C into 1 or 2 L glass bottles. Subsamples were reserved in 40 mL glass vials and acidified to pH 2 using concentrated HCl for dissolved organic carbon (DOC) and total dissolved nitrogen (TDN) measurements or left untreated for optical property (absorbance and fluorescence) analyses and stored at 4°C. The remaining ~1 L samples were then acidified to pH 2 using concentrated HCl and DOM was isolated using a solid phase extraction (SPE) technique. Briefly, samples were then drawn through clean Teflon tubing (rinsed with pH 2 ultrapure water) and connected to 1 g Bond Elut Priority PolLutant (PPL) cartridges (Agilent). PPL cartridges were previously activated with methanol (LC-MS Chromasolv, Sigma Aldrich) and rinsed with 0.1% formic acid water (LC-MS Chromasolv Sigma Aldrich). After extraction, all cartridges were rinsed with at 0.1% formic acid water and dried in a vacuum manifold. Dried cartridges were wrapped in foil and stored at 4 °C until returning to the Chesapeake Biological Laboratory. Before elution, cartridges were re-rinsed with 0.1% formic acid water and dried again under a hood using a vacuum manifold. DOM samples were then eluted with 10 mL ultrapure methanol into clean amber glass vials and stored at -20 °C until mass spectrometric analysis (described below).
Ultrahigh Resolution Mass Spectrometry (MS) Analysis
We used ultrahigh resolution mass spectrometry to characterize DOM in all samples and the possible production of DBPs formed during the desalination process. PPL extracts were diluted between 1:5 to 1:10 (depending on initial DOC concentrations) with ultrapure methanol prior to analysis with a Bruker Solarix 12 Tesla Fourier transform (FT) ion cyclotron resonance (ICR) mass spectrometer housed at the Helmholtz Zentrum Munich. We used negative ion mode electrospray ionization (ESI) with a spray voltage of –3.6 kV. The flow rate was held constant at 2 µL min-1 and 1,000 scans were averaged. The autosampler was programmed to wash with 600 µL of 80:20 MeOH:water to prevent carryover, and blank methanol samples were injected approximately every 10 samples. Exact molecular formulas (mass error <0.5 ppm) were assigned using proprietary software, which is based on the combinations of the elements 12C1-∞, 1H1-∞, 16O1-∞, 14N0-5, 32S0-2, 79Br0-3, as well as the 13C, 34S, and 81Br isotopologues. Molecular formula assignments with Br were confirmed manually using isotope simulation in the Bruker data analysis software. Isotope simulation allows for confirmation of 81Br isotopologue at 49.31% natural abundance.