Incubation setup:
Note that these four incubations were carried out over the course of six total experiments, but only the long-term incubation data are presented here; hence, the incubations are labeled as 2, 3, 5, and 6.

Seawater for all four incubation experiments was collected near Ocean Station Papa (50 °N, 145 °W) using a trace metal clean towfish system (Mellett and Buck 2020) on the R/V Roger Revelle August 17 to 29 of 2018 (See supplemental file named 1002908_v1_Suppl_Fig_1_Seawater_collection_and_incubation_setup.pdf). Seawater from ~ 2 meters (m) depth was prefiltered inline to remove large grazers using an acid-cleaned 150-micrometer (µm) mesh and collected in acid-cleaned and Milli-Q (MQ; ≥ 18.2 MΩ cm)-conditioned 20-liter (L) polycarbonate (PC) carboys (see Hollister et al. (2020) for carboy cleaning protocol). Each carboy was rinsed three times with seawater prior to filling. Water was collected underway at 5 to 10 knots, and the carboys were filled round-robin style to homogenize any natural surface variability. Discrete samples were also collected during the filling process to check for changes in nutrients and dissolved trace metals during setup. Incubations 2 (Inc 2) and 5 (Inc 5) were conducted in the 20-L carboys; for incubations 3 (Inc 3) and 6 (Inc 6), homogenized seawater from the 20-L carboys was aliquoted into 4-L PC bottles for incubation.

Nutrient amendments for Inc 2 and 5 were selected to induce nutrient stress of a particular nutrient, and all Fe additions were accomplished with a stable isotope tracer in the form of ⁵⁷FeCl₃ (Isoflex). Treatments included multi-nutrient additions of +20 micromolar (μM) nitrate +1.25 μM phosphate +20 μM silicic acid ("AllButFe") and +20 μM nitrate +1.25 μM phosphate +5 nM Fe ("AllButSi") to induce Fe and Si stress, respectively. For Inc 3 and Inc 6, single treatments of +1 nM Cu, +5 nM Fe or +20 µM silicic acid, as well as a combined +5 nM Fe +20 µM silicic acid treatment, were used to assess relief of specific nutrient limitation. For each incubation, triplicates of the controls were sampled for initial conditions (day 0) once all 20-L carboys or 4-L incubation bottles were amended, and remaining carboys and bottles were placed in deck-board incubators. These carboys and bottles were incubated in the deck-board incubators, which were flushed continuously with surface seawater for temperature maintenance and covered with mesh screening to replicate surface water light levels. Incubations 2 and 3 were carried out for 6 days, and Inc 5 and 6 were carried out for 8 days. For the final time points of the incubations, triplicates of each treatment were gently mixed by inversion and sampled in a trace-metal-clean (TMC) bubble. Between each incubation the 20-L carboys and 4-L bottles were rinsed once with MQ, twice with methanol, three times with 10 % hydrochloric acid (HCl; trace metal grade), and then four times with MQ again. They were also rinsed three times with seawater during the setup of subsequent incubations.

Sampling and analyses:

Trace metals:
Incubation samples for dissolved trace metals were collected through sequential acid-cleaned 5 µm and 0.4 μm polycarbonate track-etched (PCTE) filters via a custom-made trace metal clean vacuum filtration apparatus (Burns et al. 2023). Trace metal samples were acidified (0.024 M; Optima HCl) and stored at room temperature until returned to the lab for analysis. Following preconcentration on the seaFAST pico (ESI) offline, a suite of trace metals were analyzed by high-resolution inductively coupled plasma mass spectrometry (HR-ICP-MS) on an Element XR (ThermoScientific) by standard addition (Burns et al. 2023). Prior to preconcentration, samples were ultraviolet (UV) oxidized in a UVO-Cleaner (Model No. 342; Jelight Company Inc.) for 90 minutes (Biller and Bruland 2012; Hollister et al. 2020). Samples were also re-run without UV oxidation and final Ni and cadmium (Cd) concentrations for each sample, which did not significantly change with or without UV oxidation, are presented as the average of all runs (i.e., both UV and no UV).

Labile particulate trace metals were collected on the 0.4 μm and 5 μm PCTE filters used to filter the dissolved trace metal samples. Filters were folded into eighths and stored frozen (-20 degrees Celsius (ºC)) in 1.5-milliliter (mL) snap-cap polypropylene microcentrifuge tubes that had been acid-cleaned in 10% HCl (TraceMetalGrade, Fisher). Labile particulate trace metals were extracted via the Berger leach method (Berger et al. 2008) to quantify the biologically cycled particulate metals (Rauschenberg and Twining 2015). Briefly, microcentrifuge tubes containing the folded sample filters were thawed at room temperature and placed into a 4-Way Flipper™ Rack (Fisher) capable of holding up to 32 tubes. For each sample, 1 mL of a solution of 25% acetic acid (Fisher, Optima) with 0.02 M hydroxylamine hydrochloride (Fisher, ≥99%) was pipetted into each tube. The sample rack was then placed into a 90 ºC water bath for 10 minutes, and then removed and allowed to cool to room temperature for 1 hour and 50 minutes. For each sample, 1 mL of the leach solution was transferred into a trace-metal-clean 15-mL Teflon flat interior vial (PFA, Savillex) using a pipette, and the filter was rinsed with three 1-mL aliquots of MQ water that were also transferred to the vial. Leach solutions and rinses were spiked with 100 microliters (µL) HNO₃ (Fisher, Optima) and then dried down on a hot plate. This digestion was repeated again with another 100 µL HNO₃ (Fisher, Optima), and the final residue was redissolved in 3 mL of 2% HNO₃ (Fisher, Optima) containing 10 parts per billion (ppb) indium (In) internal standard for analysis on the HR-ICP-MS Element 2 (ThermoScientific) system at the National High Magnetic Field Laboratory. Concentrations of labile particulate trace metals and phosphorus were measured by standard addition.

An internal dissolved QC surface seawater sample was made from the North Pacific EXPORTS cruise in August 2018. Dissolved reference materials (SAFe S, GSP) with consensus values available on the GEOTRACES website were used to assess accuracy. Three replicates of air blanks were measured during each seaFAST run (Hollister et al. 2020; Burns et al. 2023) and subtracted from the calculated dissolved sample concentrations as process blanks. Limits of detection were calculated as triple the standard deviation of the air blanks. For the labile particulate samples, blank PCTE filters were processed on the filter rigs with MQ during each sampling event and treated the same as samples to account for processing blanks.

Dissolved macronutrients:
Samples for the dissolved macronutrients nitrate+nitrite (N+N), silicic acid (Si), and soluble reactive phosphorus ("phosphate," or P) were collected along with dissolved trace metal samples from the 0.4 µm PCTE filtrate described above. Samples were frozen at -20°C and shipped to the analytical facility at the University of California Santa Barbara where they were thawed and analyzed on a Lachat Instruments QuikChem 8500 Series 2 analyzer (Parsons et al. 1984; Brzezinski et al. 2022).

Chlorophyll a:
Chlorophyll a (chl a) samples were collected on 0.6 μm and 5 μm GF/F Whatman glass microfiber filters by vacuum filtration. The filters were leached with 95% ethanol immediately following collection (Jesperson and Christoffersen 1987; Morison and Menden-Deuer 2015) and analyzed on a 10AU fluorometer (Turner Designs, Inc.) at sea.

Biogenic silica:
Biogenic silica on 0.6 µm and 5 µm filters was measured as described in Brzezinski et al. (2022) at the University of California, Santa Barbara (UCSB). Briefly, each filter was folded and stored in a plastic container and immediately frozen at -20°C to prevent opal dissolution. Biogenic silica was measured using a NaOH digestion method using manual colorimetry.

Particulate organic nitrogen and carbon:
Particulate organic carbon (POC) and particulate organic nitrogen (PON) samples were collected on 0.6 µm and 5 µm pre-combusted GF/F filters that were placed in glass scintillation vials and frozen at -20°C and analyzed at UCSB as described by Sharp (1991) using a CHN analyzer (Leeman Labs Inc., CE Model 440).