The following methods are provided in Mellett & Buck (2020).

Sample Collection
Surface water samples were collected from the West Florida Shelf in the eastern Gulf of Mexico 18-21 June 2015 aboard the R/V Weatherbird II and 27 February to 2 March 2018 aboard the R/V Hogarth. Underway hydrographic data for salinity and temperature were collected on both ships via flow-through systems (SeaBird); flow-through chlorophyll a data were also collected in summer on the R/V Weatherbird II via a calibrated fluorometer. The flow-through system on the R/V Hogarth did not include a fluorescence sensor and discrete samples were collected for chlorophyll a analysis at the University of South Florida (USF). Samples for trace metals, macronutrients, and chlorophyll a were collected from ~2 m depth using a custom surface pump "towfish" system (Bruland et al. 2005) towed alongside the vessels. Our "towfish" sampling system consisted of an all PTFE Teflon air-operated double-diaphragm pump (Jabsco, Cole-Parmer), acid-cleaned tubing (Bev-A-Line IV, ½", Cole-Parmer), and a 35 kg steel and lead "towfish" coated in non-metallic epoxy paint (EPO coat VA, DeCara Inc.). For deployment, the tubing was fed through the towfish such that the tubing opening was located ~10 cm forward of the towfish body, which was suspended in the water from the starboard A-frame using a ~2 m length umbilical of ½" diameter non-metallic line (Vectran) attached to the end of the vessel hydrowire. To collect surface water samples underway, the package was towed at 5-10 knots, which allowed the towfish to sample at the edge of the ship wake with the tubing intake facing into the water ahead of the towfish. The surface water from the towfish was pumped into a positive pressure bubble in the main lab of each ship for sampling.

Dissolved Trace Metals
Samples for dissolved ('d') trace metals were filtered inline through 0.2 μm Acropak capsule filters (Pall) into 125 mL low-density polyethylene (LDPE; Nalgene) bottles. The Acropak capsule filters were cleaned by leaching with 10% TraceMetal Grade (TMG) HCl, rinsing with MQ water (Milli-Q; >18.2 MΩ cm), and then flushing with at least 1 L of surface seawater prior to use. The 125 mL sample bottles used for dissolved and total dissolvable trace metals were cleaned with TMG nitric acid (HNO₃; Fisher) in accordance with GEOTRACES protocols (Cutter et al. 2017), and stored filled with dilute (0.024 M) TMG HCl (Fisher) until use. Total dissolvable ('TD') trace metal samples were collected immediately after collection of dissolved samples by removing the inline filter capsule and rinsing and filling a 125 mL LDPE bottle with unfiltered water. Both sample types were acidified with the equivalent of 4 mL 6 M high-purity (Fisher Optima) HCl per liter of seawater to 0.024M (~pH 1.7-1.8, NBS scale; Johnson et al. (2007)), and stored for at least 6 months. Prior to preconcentration and analysis, total dissolvable metal samples were filtered through 0.4 μm polycarbonate track etched (PCTE; Whatman) filters on a Teflon filtration rig (Savillex) into 125 mL LDPE bottles. The PCTE filters were cleaned prior to use by soaking in 10% HCl (TMG; Fisher) for at least a week, then rinsed 5 times with MQ water, and were stored in 0.024 M HCl (TMG, Fisher) until use.

Dissolved and dissolvable metals were processed using an automated seaFAST-pico (ESI) preconcentration system (Lagerström et al. 2013). Samples were first UV-oxidized to ensure recovery of any organic-bound Cu and Co (Milne et al. 2010; Biller and Bruland 2012). The metals of interest were then preconcentrated by seaFAST onto a Nobias PA1 chelating resin at pH 6.2 and eluted from the column with 1N triple quartz-distilled HNO₃ with internal reference standards of indium and rhodium. The eluent was then analyzed via a method of standard addition on a Thermo Scientific Element XR Inductively Coupled Plasma Mass Spectrometer (ICP-MS) in medium resolution and counting mode at the USF College of Marine Science (Hollister et al. 2020). The blanks and limit of detection of the seaFAST-pico system were calculated using "air blanks" that encompass both the reagent blank and the manifold blank of the system (Table S1). The accuracy of metal concentrations measured using this method was established by comparison of NASS-7 and SAFe (Johnson et al. 2007) seawater reference material results with consensus values (Table S1). All reference material was within the consensus range for each metal measured with the exception of dCd in SAFe S1, which was quantified as 2.4 ± 3.7 pM (n=4) against a consensus of 1.0 ± 0.3 pM. The average relative standard deviation of samples measured in replicate across the dataset were 7.9% (Fe), 5.7% (Cu), 3.2% (Mn), 10.3% (Zn), 7.0% (Co), 5.9% (Ni), 15.4% (Cd), and 8.7% (Pb), respectively. The limit of detection for each metal was 56.7 pM (Fe), 11.6 pM (Cu), 2.4 pM (Mn), 18.9 pM (Zn), 0.9 pM (Co), 41.1 pM (Ni), 0.7 pM (Cd), and 1.6 pM (Pb), respectively (3s of air blanks; Table S1).

Dissolved Cu and Fe speciation
Filtered (<0.2 µm, Pall Acropak) samples were collected for dissolved Fe-binding and Cu-binding organic ligands in 500 mL fluorinated high-density polyethylene (FPE) bottles (Buck et al. 2012). These sample bottles were cleaned by soaking in a soap bath for at least one week, then in a 10% TMG HCl (1.2 M) bath for at least a month. Bottles were then rinsed at least 3-5 times with Milli-Q water (>18.2 MΩ cm) and stored filled with Milli-Q until use. Prior to filling with sample, bottles were rinsed three times with sample and were stored frozen (-20°C) until analysis on land.

The concentrations and conditional stability constants of Cu- and Fe-binding organic ligands were measured using competitive ligand exchange adsorptive cathodic stripping voltammetry (CLE-AdCSV). The competitive ligand salicylaldoxime (SA; Sigma-Aldrich) was used to compete with natural ligands (Campos and van den Berg 1994; Rue and Bruland 1995). Measurements were made on a BioAnalytical Systems (BASi) controlled-growth mercury electrode interfaced with an Epsilon 2 analyzer (BASi). Final SA concentrations of 2 μM and 25 μM were used for Cu and Fe speciation measurements, respectively (Buck et al. 2012; Jacquot and Moffett 2015). Deposition times of 300 s and 90 s were used for Cu and Fe speciation, respectively.

The full details of the Fe and Cu speciation methods and theory may be found elsewhere (Campos and van den Berg 1994; Rue and Bruland 1995; Buck et al. 2012, 2018). Briefly, for each titration, aliquots of 10 mL were distributed into fifteen lidded Teflon vials (Savillex), which had previously been conditioned with Milli-Q, borate buffer, metal additions, and SA prior to use. Each 10 mL aliquot was buffered to pH 8.2 (total scale) with a 1.5 M borate buffer (7.5 mM final concentration) made in 0.4 N ammonium hydroxide (Optima grade; Fisher) and cleaned for trace metals using Chelex resin. For Cu titrations, additions of 0–20 nM Cu were used and for Fe, additions of 0–10 nM Fe were used in separate vials. Metal additions were allowed to equilibrate overnight, and SA additions were equilibrated at least 1 h before analysis (Buck et al. 2012; Abualhaija and van den Berg 2014).

Titration data were processed using ProMCC software (Omanović et al. 2015; Pižeta et al. 2015), which fits the datasets using multiple regression models, both linear and non-linear. A single ligand class was found to be the best fit for the speciation data for both metals. Refer to Mellet & Buck (2020) for full details.

Electroactive Fe-binding humic-like substances
Remaining volume from the speciation samples used for the Fe- and Cu-binding ligand analyses was used for the quantification of naturally occurring electroactive Fe-binding humic-like substances (HS-like ligands). The measurements were made following the method of Laglera et al.(2007) as modified by Bundy et al. (2014). Briefly, a 10 mL sub-sample was amended with Fe (50 nM) to saturate any excess Fe-binding ligands present in the sample, buffered with 1.5 M borate buffer, and allowed to equilibrate for at least 2 hours. After equilibration, 20 mM potassium bromate oxidant was added and the electroactive response of the natural Fe-ligand complexes formed in the sample was measured on a Metrohm model 797 VA in direct current mode. Potential was scanned from -100 mV to -1100 mV with a deposition potential of -100 mV, and a deposition time between 60 s and 180 s depending on peak response. This approach directly detects any Fe-binding ligands in the samples that form electroactive complexes with Fe. Quantification of these electroactive Fe-binding ligands was made by method of standard additions of Suwannee River humic acid standard (SRHA; International Humic Substance Society) ranging from 0–60 mg L⁻¹ or 0–400 mg L⁻¹ (dependent on peak response). The concentrations of the electroactive Fe-binding ligands determined with this approach are thus described as Fe-binding HS-like ligands in the samples.

WFS_GoM_2015_2018.csv