Dataset: NMR FID files

Locations:

Arctic Ocean: P-1-94-AR P21, 84°5' N, 174°58' W
California margin: W-2-98-NC TF1, 41°5' N, 125°1' W
Equatorial Pacific: TT013-06MC, 12°00' S, 134°56' W

These procedures apply to all NMR spectra files.
Prior to the extraction, we freeze-dried, ground and sieved sediment samples to less than 125 μm (Ruttenberg 1992). For a given sample, we weighed four sample replicates (2 g) and placed each in 250 mL HDPE bottles. Sodium dithionite (F.W. 147.12 g/mol; 7.4 g) was added to each sample split, followed by 200 mL of citrate-bicarbonate solution (pH 7.6). This step produces effervescence, so the solution should be added slowly to the sample. We shook samples for 8 h and then centrifuged them at 3,700 rpm for 15 min. We filtered the supernatants with a 0.4 μm polycarbonate filter. We took 20 mL aliquots from the filtrate for each sample split for MRP and total P analyses, and kept them refrigerated until analysis within 24 h. We added 200 mL of ultrapure water to the solid residue for each sample split as a wash step after the above reductive step, shook samples for 2 h, and then centrifuged them at 3,700 rpm for 15 min. We filtered the supernatants with 0.4 μm polycarbonate filters and set aside 20 mL of filtrate from each sample split for MRP and total P analyses. We then extracted the solid sample residues in 200 mL of sodium acetate buffer (pH 4.0) for 6 h. At the end of this extraction step, we centrifuged the bottles at 3,700 rpm for 15 min, filtered the supernatants with 0.4 μm polycarbonate filters and took a 20 mL aliquot of filtrate from each sample split for MRP and total P analyses. We added 200 mL of ultrapure water to the solid residue for each sample split as a wash step, shook samples for 2 h, and then centrifuged them at 3,700 rpm for 15 min. We filtered the supernatants with 0.4 μm polycarbonate filters and set aside 20 mL of filtrate from each sample split for MRP and total P analyses. We repeated the water rinse step, and collected aliquots for MRP and total P analyses as in the previous steps. The concentrations of TP were determined as described below.

Solid sediment sample residues following the pretreatment described above were transferred to two 50 mL centrifuge tubes (2 sample replicates combined per tube). We added 20 mL of 0.25 M NaOH + 0.05 M Na2EDTA solution to each tube, vortexed until all sediment was resuspended and then shook samples for 6 h at room temperature (Cade-Menun et al. 2005). We used a solid to solution ratio of 1:5 for this step to minimize the amount of freeze-dried material that will need to be dissolved for the 31P NMR experiments. Large amounts of salts from the NaOH-EDTA concentrated in NMR samples lead to higher viscosity and increase line broadening on NMR spectra (Cade-Menun and Liu 2014). We chose an extraction time of 6 h to improve total P recovery while limiting the degradation of natural P compounds in the sample. At the end of the extraction, samples were centrifuged at 3,700 rpm for 15 min and supernatants decanted into 50 mL centrifuge tubes. We collected a 500 μL aliquot from each sample, which we diluted with 4.5 mL of ultrapure water. These were refrigerated until analysis for total P content on the ICP-OES. The sample residues and supernatants were frozen on a slant to maximize the exposed surface area during the lyophilization step; this was done immediately after the removal of the 500 μL aliquot. Once completely frozen, the uncapped tubes containing supernatants and residues were freeze-dried over the course of 48 h. Each tube was covered with parafilm with small holes from a tack to minimize contamination. Freeze-dried supernatants from identical sample splits were combined and dissolved in 500 μL each of ultrapure water, D2O, NaOH-EDTA and 10 M NaOH prior to 31P NMR analysis. The D2O is required as signal lock in the spectrometer (Cade-Menun and Liu 2014). Sample pH was maintained at a pH > 12 to optimize peak separation (Cade-Menun 2005; Cade-Menun and Liu 2014). Sample pH was assessed with a glass electrode, and verified with pH paper to account for the alkaline error caused by the high salt content of our samples (Covington 1985).

Spectra were acquired immediately following sample preparation on a Varian Unity INOVA 600 MHz spectrometer equipped with a 10-mm broadband probe [operated by the Stanford Magnetic Resonance Laboratory at Stanford University]. We used a 10-mm rather than a 5-mm probe because larger tubes contain a greater concentration of P and thus require fewer scans to achieve similar signal to noise ratios (Cade-Menun and Liu 2014). The analytical parameters used were: 20oC, 90o pulse, 0.48 s acquisition time, 4.52 s delay time, 5600 scans (8 h experiments), no spin and an external H3PO4 standard. We maintained samples at a temperature of 20oC during experiments to achieve optimal spectral resolution (Crouse et al. 2000) and to minimize sample degradation. No proton decoupling was used out of concern for sample degradation (Cade-Menun and Liu 2014). The ratio of P to Fe and manganese (Mn) was used as a proxy for spin-lattice relaxation times (T1) to ensure adequate delays between pulses and thus quantitative spectra (McDowell et al. 2006). We used 5 s recycle delays, which correspond to three to five times the calculated T1 values, as recommended by McDowell et al. (2006). Peak identification was based on literature (Turner et al. 2003; Cade-Menun 2015).

31P NMR data were processed using the NMR Utility Transform software (NUTS, Acorn NMR). Peak areas were calculated by integration of spectra processed with a 7 Hz line broadening following baseline correction, peak picking and phasing. We accepted peaks that (1) represented at least 1% of the tallest peak in the total integrated area, (2) were identified by the NUTS software and (3) were confirmed as signal by visual inspection.

Freeze-dried sample residues were ashed in crucibles at 550°C for 2 h and then extracted in 25 mL of 0.5 M sulfuric acid for 16 h (Olsen and Sommers 1982; Cade-Menun and Lavkulich 1997). We centrifuged samples at 3,700 rpm for 15 min, filtered supernatants with 0.4 μm polycarbonate filters, and measured P content on an ICP-OES.

Total P concentrations in sediment extracts were measured using inductively coupled plasma optical emission spectroscopy (ICP-OES). Standards were prepared with the same solutions as those used for the extraction procedure in order to minimize matrix effects on P measurements. Sediment extracts and standards (0 μM, 3.2 μM, 32 μM and 320 μM) were diluted to lower salt content to prevent salt buildup on the nebulizer (1:20 dilution for step 1, 1:10 for steps 2 4). Concentration data from both wavelengths (213 nm and 214 nm) were averaged to obtain extract concentrations for each sample. The detection limit for P on this instrument for both wavelengths is 0.4 μM. The MRP concentrations were measured on a QuikChem 8000 automated ion analyzer. Standards were prepared with the same solutions used for the extraction step to minimize matrix effects on P measurements. Sediment extracts and standards (0 30 μM PO4) were diluted ten-fold to prevent matrix interference with color development. The detection limit for P on this instrument is 0.2 μM. We derived MUP concentrations by subtracting MRP from total P concentrations.