Sample Collection:
This work was funded by NSF and conducted in collaboration with the NASA EXPORTS campaign at Porcupine Abyssal Plain-Sustained Observatory (PAP-SO), which sampled the decline of the North Atlantic Spring Bloom (NASB) in May 2021 on board the RRS Discovery (Cruise ID: DY131). All sampling occurred within a retentive eddy named "A2", with a deep anticyclonic structure. Field samples for dissolved (<0.4 or <0.2 micrometer (µm)) trace metals (Mn, Fe, Co, Ni, Cu, Zn, Cd, and Pb), and dissolved (<0.4 or <0.2 µm) macronutrients (P, N, Si) from surface and subsurface waters were collected as environmental context for the incubation experiments also conducted as part of this study.

All sampling was performed following trace metal clean procedures. Surface (~2 meters (m)) water was collected using a trace metal clean towfish that was sampled while underway at a speed of 4-6 knots (Mellett and Buck 2020). Surface dissolved trace metal and macronutrient samples were collected from the towfish while filling the 'Long-term' and 'Short-term' experiment incubation bottles, as follows: prior to filling the first set of carboys corresponding to the first replicate of the treatments (three replicates per treatment total), after filling the second set of carboys, after the third set, and after filling all replicate incubation carboys. These samples were filtered inline through a 0.2 µm (Pall Acropak) capsule filter. These samples are labelled "Survey Samples".

Subsurface trace metal and macronutrient samples were collected using a trace metal clean rosette (TMC CTD) deployed on synthetic line and outfitted with modified trace metal clean x-Niskin bottles (12-liters (L), OceanTestEquipment, Inc.; Cutter et al. 2017). Samples for dissolved trace metal and macronutrient samples were filtered either inline through a <0.2 µm capsule filter (Pall Acropak) or filtered offline through sequential acid-cleaned polycarbonate track-etch (PCTE) filters (5 µm, 0.4 µm) in a dual stage filtration system (Savillex) on a custom acrylic base. For a subset of profiles, additional unfiltered (total dissolvable, TD) samples were collected and filtered offline after eighteen months of storage and prior to analysis (TD Savillex). All dissolved and total dissolvable trace metal samples were acidified to 0.024 M hydrochloric acid (Optima, Fisher) and stored at room temperature until analyzed at USF. All macronutrient samples were stored frozen (-20 degrees Celsius (ºC)) until analyzed in the lab at USF. Seven depth profiles were deployed during the cruise, and water from casts #1, #3, and #6 were used to collect seawater for the Grow-Out incubation experiments. All TMC CTD casts occurred between 11 am and 12 pm local ship time.

Analysis of dissolved and total dissolvable trace metals:
Sample analysis for dissolved trace metals was conducted at the University of South Florida (USF) College of Marine Science (CMS), and Florida State University (FSU) National High Magnetic Field Laboratory (MagLab). To prepare for analysis, dissolved trace metal samples were transferred to 30- or 15-milliliter (mL) perfluoroalkoxy (PFA) vials with caps containing a quartz-window to enable UV-oxidization. Samples were UV-oxidized at USF for 90 minutes at ~20 milliwatts per square centimeter (mW cm-2) with an UVO-Cleaner (Jelight; Model No. 5144AX) to liberate organically complexed dissolved Co and Cu (Milne et al. 2010). For cobalt, nickel, and copper, the UV-oxidized and non-UV-oxidized sample results are presented separately.

The automated seaFAST-pico (Elemental Scientific) with a Nobias Chelate-PA1 resin was connected to a High Resolution-Inductively Coupled Plasma-Mass Spectrometer (HR-ICP-MS; Thermo Scientific Element 2 at FSU, Element XR at USF), to preconcentrate and extract trace metals from the seawater samples inline (Lagerström et al. 2013). The reagents and input flow rates for seaFAST were followed from Hollister et al. (2020) and Burns et al. (2023). The elution acid in this study was a solution of 10% HNO3 (Optima) with 1 ppb Indium as the internal standard. Upon loading the buffered sample (pH ~ 6.2 ± 0.2; Burns et al. 2023) in the resin column, the column was rinsed with Milli-Q (≥18.2 MΩ·cm) to wash out seawater salts, and the chelated trace metals were eluted with the elution acid containing the internal standard. The eluent from seaFAST was then drawn by the ICP-MS and analyzed for the natural abundance of Mn, Fe, Co, Ni, Cu, Zn, Cd, and Pb. Samples were analyzed twice, on the Element 2 at FSU, and on the Element XR at USF.

All analytical runs were comprised of seawater samples, instrument-manifold air blanks, quality control (QC) samples, a matrix-matched multielement calibration curve containing all target metals, a matrix-matched molybdenum (Mo) calibration curve, and certified reference materials, including SAFe D2, GEOTRACES GSP, and NASS-7. All seawater calibration curves and reference samples were prepared and treated the same as the seawater samples throughout the analytical runs.

The in-house QC consisted of 2-L aliquots of filtered (<0.2 µm) PAP-SO surface seawater (49.20469 °N, -14.78163 °E) that was collected in bulk on the 2021 EXPORTS cruise and then subsampled and acidified to 0.024 M HCl (Optima) for each QC aliquot. The matrix-matched curves, including the multielement and Mo calibration curves, were made using the same seawater as the QC to match the matrix of the seawater samples. The matrix-matched multielement calibration curve was made by dilution from a set of working stocks containing the target metals. These working stocks were diluted from ICP primary standards of 1000 micrograms per milliliter (µg/mL) in 2% HNO3 of Mn (SPEX CertiPrep), Fe, Co, Ni, Cu, Cd (ULTRA Scientific), Zn and Pb (RICCA) diluted with 5% HNO3 (Optima, Fisher) elution acid matrix. The Mo ICP primary standard consisted of 1000 µg/mL Mo in H2O (SPEX CertiPrep), from which a Mo working stock was made by dilution with 5% HNO3 (Optima, Fisher).

The analytical runs usually began by conditioning the inline seaFAST-ICP-MS with a few samples of filtered seawater, followed by several air and MQ blanks, calibration curves including the multielement and Mo curves, certified reference materials (also included in the middle of each analytical run to allow replicate measurements), seawater samples, several QC samples spread throughout the analytical run (every ~15th sample), several additional air blanks, and finalized by air blank samples. Sample trace metal concentrations were determined from calibration curves of the intensity counts against standards of known concentrations and corrected for average air blanks in the sequence as described previously by Hollister et al. (2020) and Burns et al. (2023). All dissolved and total dissolvable trace metal analyses were performed by PhD student Delfina Navarro-Estrada with assistance from senior researcher Salvatore Caprara in the Buck lab at the University of South Florida.

Analysis of dissolved macronutrients:
Macronutrient samples were thawed at room temperature and analyzed following standard colorimetric methods (Parsons et al. 1984; Becker et al. 2020) using a QuAAtro39 AutoAnalyzer (SEAL Analytical) at USF. The analytical runs included seawater samples, calibration curves made in artificial seawater, reagent blanks consisting of the artificial seawater used to make the calibration curves, reference materials including CK, CL, and CO (KANSO TECHNOS), and QC samples, including standards with known concentrations of nitrite and nitrate to check the efficiency of the instrument's cadmium column, and the lowest and middle-high standards of the calibration curve to check and correct for drift. QC samples were added approximately every 12th sample to assess the quality of the analytical run. The range in LODs for macronutrients were 0.006-0.048 micromolar (µM) for nitrate+nitrite, 0.011-0.042 µM for phosphate, 0.016-0.108 µM for silicic acid, and 0.004-0.027 µM for nitrite. Values below these limits of detection in each run are reported as 0 µM with accompanying QC Flag 6. Sample analyses for macronutrients were performed by senior researcher Salvatore Caprara in the Buck lab at the University of South Florida.