Samples were collect on the USGS R/V Mary Landsteiner and pumped directly from the surface (1 m deep) with a pump and clean tycoon tubing connected to an inline 0.2 um Whatman Polycap filter.

All filters were pumped and field filtered through 0.7 um Whatman glass fiber filters (GF/F, precombusted at 550 degrees C) using a peristaltic pump after purging the line.

Samples for DOC concentration were acidified to pH 2 and stored in a refrigerator (4 degrees C) until analysis by high-temperature combustion on a Shimadzu TOC-L CPH within two weeks following collection. DOC was calculated as the mean of between three and five injections using a six-point standard curve using established protocols (Mann et al., 2012) and the coefficient of variance was always <2%.

Samples for CDOM absorbance were analyzed in a 1 cm cuvette on a Horiba Aqualog-UV-800-C. Absorbance spectra were measured from 230-800 nm, and corrected for a small offset either due to long-term baseline drift or derived from glass fiber particles during filtration (Blough et al., 1993), by subtracting the mean absorbance measured between 750-800 nm. Two spectral slopes were calculated at 275-295 nm and 350-400 nm (S275-295 and S350-400, respectively), and the spectral slope ratio (SR) was then calculated by dividing the former by the latter (Helms et al., 2008). The CDOM absorption ratio at 250 nm to 365 nm was calculated (a250:a365) and specific ultraviolet absorbance (SUVA254) was calculated by dividing the decadic absorption coefficient at 254 nm by DOC concentration (Weishaar et al., 2003; Fellman et al., 2009).

Fluorescence properties of FDOM were determined using a Horiba Aqualog-UV-800-C. The excitation emission matrices (EEMs) were generated in a 1 cm cuvette at varying integration times (1-10 seconds) to maximize the signal-to-noise ratio based on absorbance values. The EEMs were obtained at excitation (ex) 250-600 nm and at emission (em) 250-600 nm with 5 nm and 2 nm intervals respectively, and the EEMs were corrected for lamp intensity (Cory et al., 2010), inner filter effects (Kothawala et al., 2013), and normalized to Raman units (R.U.) (Stedmon et al., 2003). All corrections were performed using the FDOMcorr toolbox version 1.6 (Murphy, 2011). EEMs were analyzed with parallel factor analysis (PARAFAC) using the procedure described in Murphy et al. (2013). Furthermore, the fluorescence index (FI) (Cory et al., 2010), humification index (HIX) (Ohno, 2002; Zsolnay et al., 1999), and autotrophic productivity index (BIX) (Huguet et al., 2009) were calculated. FI was calculated from the emission wavelengths at 470 nm and 520 nm, obtained at excitation 370 nm (Cory and McKnight, 2005). HIX was calculated using the area under the emission sepctra 435-480 nm divided by the peak area 300-345 + 435-480 nm, at excitation 254 nm (Ohno, 2002). BIX was calculated from the emission intensity of 380 nm and 430 nm, obtained at excitation 310 nm (Wang et al., 2014).

Samples for FT-ICR MS analysis were solid-phase extracted using the procedure described in Dittmar et al., 2008. Filtered samples were acidified to pH 2 before solid phase extraction on 500 mg Agilent Bond Elut PPL cartridges. Each 1 L sample was extracted by eluting 2 mL of of methanol and then diluted to a DOC target concentration of 50 ug C mL-1. Extracted samples were stored at -20 degrees C prior to analysis on a 21 T (Bruker Daltonics, Billerica, MA, USA) FT-ICR MS located at the National High Magnetic Field Laboratory (NHMFL) (Tallahassee, Florida). Direct infusion electrospray ionization (ESI) generates negative ions at a flow rate of 700 nL min-1, and 100 time domain acquisitions were coadded for each mass spectrum.

Molecular formulas were assigned to signals >6RMS baseline noise with EnviroOrg ©,TM software (Koch et al., 2007; Stubbins et al., 2010). Elemental combinations of C1–45H1–92N0–4O1–25S0–2 with a mass accuracy of ≤300 ppb were considered for assignment. Classification of formulas were based on their elemental ratios (Corilo, 2015). The modified aromaticity index (Almod) of each formula was calculated and Almod values of 0.5-0.67 and ≥0.67 were classified as aromatic and condensed aromatic structures (Koch and Dittmar, 2006; Koch and Dittmar, 2016). Other compound classes were unsaturated low oxygen=Almod<0.5, H/C<1.5, O/C<0.5; unsaturated high oxygen=Almod<0.5, H/ C<1.5, O/C>0.5; aliphatics=H/C 1.5-2.0, O/C<0.9, N=0; peptide-like=H/C 1.5-2.0, O/C<0.9, N>0, and sugar-like= O/C>0.9. Sugar-like compounds provide a very minor contribution to %RA (mean = 0.05, ± 0.06 %RA) and so were combined with peptide-like compounds throughout. Although FT-ICR MS allows for the precise assignment of molecular formulas to signals that may represent multiple isomers, they describe the underlying molecular compounds comprising DOM, thus the term compound may be used when describing the signals detected by FT-ICR MS.

Lignin derived phenols were isolated from the dried solid phase extracts followed by cupric oxide oxidation and liquid-liquid extraction modified from Spencer et al., (2010). Briefly, PPL extracts were redissolved in O2 free 2 M NaOH in a 6 mL Teflon vial (Savillex Corp) containing 500 mg CuO, and amended with 100 mg ferrous ammonium sulfate and 50 mg glucose and reacted in a 155 degree C oven for 3 hours. Following oxidation, the samples were centrifuged and supernatants were decanted into 40 mL vials. Oxidation products were acidified to pH 1 with H3PO4 and t-cinnamic acid was added as an internal standard. Liquid-liquid extractions of the oxidation products were undertaken by addition of 4 mL ethyl acetate, vortexing, and centrifugation prior to removal of the ethyl acetate. Extracts were pipetted through drying columns containing sodium sulfate into a 4 mL vial. Samples were dried under ultra-high purity argon between each extraction for a total of three extractions, following the last extraction the sodium sulfate was rinsed with 1 mL of ethyl acetate into the extract vial. Dried ethyl acetate extracts were dissolved in pyridine and derivatized with N/O bis-trimethylsilyltrifluoromethylacetamide (BSTFA) at 60 degrees C for ten minutes. Lignin phenol monomers were measured as trimethylsilane derivatives using an Agilent 6890N GC/5975 MS and were quantified as the relative response factors of each compound compared to the response of t- cinnamic acid and a five-point calibration curve bracketing the concentration range. Eight lignin phenols from three phenol groups were quantified; vanillyl (vanillin, acetovanillone, vanillic acid), syringyl (syringaldehyde, acetosyringone, syringic acid), and coumaryl (coumaric acid, ferulic acid).

Seven neutral sugars (fucose, rhamnose, arabinose, galactose, glucose, mannose, xylose) were analyzed according to Skoog and Benner (1997) with modifications. Briefly, samples were hydrolyzed in 1.2 mol L−1 sulfuric acid and neutralized with a self-absorbed ion retardation resin (Kaiser and Benner, 2000). After desalting with a mixture of cation and anion exchange resins, neutral sugars were isocratically separated with 25 mM NaOH on a PA 1 column in a Dionex 500 system with a pulsed amperiometric detector (PAD).

The following amino acids were analyzed using the method of Kaiser and Benner, 2005: histidine, serine, arginine, glycine, aspartic acid, glutamic acid, threonine, alanine, lysine, tyrosine, methionine, valine, norvaline, isoleucine, leucine, phenylalanine.