Seawater collection

Seawater was collected using 10 L Teflon-coated GO-Flo bottles (General Oceanics) with the U.S. GEOTRACES trace metal rosette sampling system (Cutter and Bruland, 2012). The rosette was additionally outfitted with a miniature autonomous plume recorder (MAPR; (Baker, 1997) which included an oxidation reduction potential (ORP) and light scattering sensors, and an ultra-short baseline (USBL) transponder to ensure precise site locations on repeat casts. Two bottles were tripped at every depth to ensure adequate volumes of seawater were obtained for sample analysis. Samples were taken for dissolved gases, total dissolvable Fe (dtFe) and total dissolvable Mn (tdMn), dissolved Fe (dFe) and dissolved Mn (dMn), and total suspended particle analysis, all of which were sampled from even numbered bottles. All cleaning and sampling procedures were done according to GEOTRACES cookbook protocols (Cutter et al., 2017; Cutter and Bruland, 2012).

Dissolved and total dissolvable iron and manganese analyses

All samples for tdFe, dFe, tdMn, and dMn were measured shipboard using flow injection analysis (FIA) and analyzed within 6-8 hours of sample collection. The dFe, tdFe, dMn, and tdMn samples were measured using direct injection FIA with spectrophotometric detection modified from (Measures et al., 1995) and detailed in (Sedwick et al., 2008) for Fe and from (Resing and Mottl, 1992) for Mn. Detection limits for Mn and Fe were 1 nM and 1.5 nM, respectively.

Organic iron-binding ligand analyses

Samples for dFe-binding organic ligands were measured using competitive ligand exchange adsorptive cathodic stripping voltammetry (CLE-ACSV) (Abualhaija and van den Berg, 2014; Rue and Bruland, 1995). All titrations were performed on a controlled growth mercury electrode (CGME, Bioanalytical Systems Incorporated) equipped with an Ag/AgCl reference electrode and platinum auxiliary electrode, with a mercury drop size of 14 and acid-cleaned Teflon analytical cell (Bioanalytical Systems Incorporated). Samples for which dFe concentrations were less than 10 nM were first analyzed using the forward titration method using a 5 µM salicyladoxime analytical window (Abualhaija and van den Berg, 2014). Briefly, samples were thawed, and 10 mL were aliquoted into each of 15 acid-cleaned and conditioned Teflon vials (Savillex Corporation). Then 10 µL of 1.5 M boric acid buffer (boric acid, Alfa Aesar 99.99% metals basis, in 0.4 N Optima NH4OH, Fisher Scientific) was added to each vial for a final concentration of 5 µM to achieve a pH of 8.2. Next 12.5 µL of 4M salicylaldoxime (Fluka > 98% assay in Optima MeOH, Fisher Scientific) was then added to each vial for a final concentration of 5µM. The dFe standards (Diluted from SpexCertiPrep in pH 2 HCl) were then added to each vial, ranging 0-10 nM for final concentrations. Aliquots were then equilibrated overnight prior to electrochemical analysis using differential pulse stripping voltammetry (0 to -800 mV), with a 120-180s deposition period with stirring at 0 mV. After the final aliquot was analyzed, 5 nM of dFe standard was added and the aliquot was re-measured to ensure complete titration of the ligands. Peak heights were obtained using ECD-Soft and ligand concentrations and binding strengths were calculated using ProMCC (Omanović et al., 2015) with an inorganic side reaction coefficient of logα_Fe′ = 10 (Abualhaija and van den Berg, 2014).

For samples where forward titrations showed no curvature, indicating excess ligands were present, or in the case where dFe in the sample was > 10 nM, reverse titrations were completed, employing 1-nitroso-2-naphthol (NN) as the competing ligand. The procedure and theory are described in detail elsewhere (Hawkes et al., 2013). Briefly, samples were thawed and 10 mL were aliquoted into 10 acid-cleaned and conditioned Teflon vials (Savillex Corporation). Then 10 µL of boric acid buffer was added to each vial for a final concentration of 5 µM and pH of 8.2. The NN standard (Sigma Aldrich) was prepared in methanol (Optima, Fisher Scientific) and was then added to each vial to achieve final concentrations ranging 0-40 µM, and aliquots were left to equilibrate overnight. After equilibration, samples were analyzed electrochemically using linear sweep voltammetry (-150 to -650 mV) after a 5-minute nitrogen purge (ultra high purity, Airgas) and a 120 s deposition time at -50 mV. Three standard additions of dFe standard were added to the final aliquot and analyzed to calculate the amount of dFe that was exchangeable with NN under the analytical conditions. The estimates of exchangeable dFe often exceeded the ambient dFe concentrations in the samples. In these cases, all ambient dFe was assumed to be exchangeable. Peak heights were obtained using ECD-Soft, and ligand concentrations and binding strengths were calculated using publicly available R code (Hawkes et al., 2013) with the unsaturated Fe fit with logα_Fe′ = 9.8 as the inorganic side reaction coefficient.

Siderophore analyses

At sea, four liters of 0.2 µm filtered seawater (Acropak 200, Pall Corporation) were collected for siderophore analyses. Prior to solid phase extraction, columns were activated with 2 column volumes of ultrapure methanol (Fisher, Optima grade) and rinsed with 2 column volumes of ultrapure Milli-Q water. Filtered seawater samples were then pumped continuously (~15-18 mL min-1) onto a Bond Elut solid phase extraction column (1g ENV, 6 mL, Agilent Technologies). After solid phase extraction, columns were again rinsed with at least 2 column volumes of Milli-Q water to flush remaining sea salts then stored frozen at -20ºC prior to analysis. In the laboratory, columns were thawed in the dark at room temperature and then rinsed with 2 column volumes of Milli-Q water to remove any additional salts. Columns were then eluted with 12 mL ultrapure methanol into 15 mL acid-washed and methanol-rinsed falcon tubes. Eluent was then concentrated to ~0.5 mL on a vacuum concentrator with a refrigerated vapor trap (SpeedVac, Thermo Scientific) and transferred to clean 2 mL low density polyethylene vials. Extracts were weighed to determine exact volumes then frozen at -20ºC until analysis.

Samples were analyzed using liquid chromatography (LC) coupled to an inductively coupled plasma mass spectrometer (ICP-MS) and an electrospray ionization mass spectrometer (ESI-MS). For each analysis, 100 µL of sample was combined with a spike of 15 µL of 5 µM cyanocobalamin internal standard then injected and separated using a Dionex 3000 LC system equipped with a ZORBAX-SB C18 trap column (0.5x35 mm, 3.5µm, Agilent technologies) and a ZORBBAX-SB C18 working column (0.5x150 mm, 5 µm, Agilent technologies) after the method of (Li et al., 2021). For each sample, a 62.5 µL injection was loaded onto the trap column at 25 µL/min with a 5-minute isocratic elution of solvent C (5% Optima methanol in 95% Milli-Q water, with 5 mM ammonium formate). Samples were then separated on working column using a flow rate of 40 µL/min at 30ºC beginning with a 5-minute isocratic elution of 95% solvent A (Milli-Q water with 5 mM ammonium formate) and 5% solvent B (Optima methanol with 5 mM ammonium formate buffer) followed by a 20 minute gradient from 95% solvent A to 90% solvent B, then a 10-minute isocratic elution at 90% solvent B, followed by a 5 minute gradient from 90% solvent B to 95% solvent B, then a 5-minute isocratic elution at 95% solvent B and a 13 minute conditioning step at 5% solvent B prior to the next injection. The same chromatography structure was used for both LC-ICP-MS and LC-ESI-MS (Boiteau et al., 2016; Bundy et al., 2018; Park et al., 2023).

Samples were introduced from the LC to the ICP-MS (iCAP-RQ; Thermo Scientific) via a PFA-ST nebulizer (Elemental Scientific) and spray chamber cooled to 2.7ºC. The ICP-MS was equipped with platinum sample and skimmer cones and a 10% oxygen flow was added to the sample chamber to prevent organic matter precipitation on the cones. Analyses were made in kinetic energy discrimination (KED) mode with a helium collision flow rate of 4.0-4.2 mL/min. ⁵⁶Fe peaks were identified using an in-house R code with a peak threshold of 700 counts above the large background of non-chromatographically resolved dFe-binding organic matter. All identified 56Fe peaks were considered putative siderophores. Putative siderophore concentrations were calculated using a standard curve of ferrioxamine E (25-200 nM) (Boiteau et al., 2016; Bundy et al., 2018; Park et al., 2023). A 50 nM ferrioxamine E standard followed by a MilliQ blank were analyzed every six samples to adjust for instrument drift and ensure minimal carryover between samples.