All data were collected from a modified procedure as described in Van Mooy et al (2015).

Sampling - Sampling was conducted aboard the R/V Atlantic Explorer during a cruise in May 2018. Water samples for whole community analyses were collected from Niskin bottles deployed on a rosette with a CTD. Subsamples (1-4 L) for incubations were dispensed from the Niskin bottle into acid-washed polyethylene bottles and promptly taken to a laboratory van for incubation setup and processing. At two stations Trichodesmium colonies were also acquired for uptake and reduction experiments. Briefly, colonies were collected near the surface with a handheld 130 µm net. 6 to 20 colonies were washed twice with freshly filtered (0.2 µm pore size polycarbonate membrane) surface seawater before being transferred into 50 mL of filtered seawater for incubation as described below. At three stations sinking particles were collected using 1.25 m diameter free-floating net traps for 24-hour deployments (Peterson et al. 2005). Once recovered, the particle slurry was further split into 12 equal fractions using an electric splitter (Lamborg et al. 2008). One split was used for total phosphate uptake and reduction measurements as described below where particle slurries were incubated in the dark in 50 to 125 mL of seawater. [C.H. Lamborg, K.O. Bruesseler, J. Valdes, C.H. Bertrand, R. Bidigare, S. Manganini, etc, The flux of bio- and lithogenic material associated with sinking particles in the mesopelagic “twilight zone” of the northwest and North Central Pacific Ocean. Deep-Sea Res II 55, 1540 (2008). M.L. Peterson, S.G. Wakeham, C. Lee, M.A. Askea, J.C. Miquel, Novel techniques for collection of sinking particles in the ocean and determining their settling rates. Limnol Oceanogr Methods 3, 520 (2005).]

Phosphate uptake rates – 50 mL samples of seawater were added to acid-washed polycarbonate incubation bottles. Each incubation bottle was spiked with approximately 2 µCi of 33P-phosphoric acid. The final concentration of 33P-phosphate in the incubations was less than 10 pmol L-1, which was likely two orders of magnitude smaller than ambient phosphate concentrations. The bottles were capped and mixed by gently inverting. To account for any abiotic adsorption of the radioactive tracer, additional 50 mL subsamples were spiked with 10% paraformaldehyde prior to the addition of the 33P-phosphoric acid. These “killed controls” were used for blank subtractions in uptake and reduction rate calculations. All bottles were placed in a flow-through on-deck incubator that was maintained at surface seawater temperatures by continually flushing it with the surface seawater from the ship’s pumping system. Temperature in the incubators was occasionally monitored with a waterproof temperature logger (Onset). The incubators used blue transparent film to achieve a light intensity to mimic 30% PAR. About half of the surface water samples were placed in a dark incubator to determine the affect light had on the incubations. For depth profiles, the incubators used a combination of neutral density screening and blue transparent film to achieve a light intensity to mimic PAR throughout the water column while samples with less than 1% PAR were placed in a dark incubator. After an appropriate amount of time, the incubations were terminated and 5 mL of sample was vacuum filtered (approximately 200 mbar) onto 25 mm diameter 0.2 µm pore size polycarbonate membranes (Millipore). The membranes were quickly rinsed three times with freshly filtered (0.2 µm pore size polycarbonate membrane) surface seawater. The membranes were then immediately placed in a liquid scintillation vial containing 10 mL of UltimaGold liquid (Perkin Elmer) scintillation cocktail, which was then shaken vigorously. The 33P-radioacitivity in the vials was determined using a liquid scintillation counter (Perkin Elmer).

Phosphate reduction to intracellular P(III) compounds – The remaining 45 mL of sample was vacuum filtered as described above. Next, the membranes were immersed in 1.0 mL of ultra-high purity (UHP) deionized water (18 MΩ*cm) in a cryovial (Fisher). The vials were immediately capped and flash frozen for storage and transport back to the on-shore laboratory. For further analysis, the samples were subject to three freeze/thaw cycles where the cryovial was immersed in liquid nitrogen for approximately 10 min, before they were immersed in boiling-hot water for 10 min, and then vigorously shaken. Next, 100 µL aliquots of the samples were injected onto an IC system (Dionex) which pumped an eluent gradient of 23 mmol L-1 to 90 mmol L-1 sodium hydroxide through an IonPac AS18 (Dionex) column at a rate of 1.0 mL min-1. An ion suppressor using UHP water as a regenerant removed sodium hydroxide from the eluent. Three fractions were collected in 60 second intervals at retention times where pure standards of (1) hypophosphorus acid (2) methyl-phosphonate, 2-hydroxethyl-phosphonate, and (3) phosphorus acid elute and the 33P-radioactivity determined as described above. The 33P-radioactivity of the three fractions was summed, corrected for dilution, and then divided by the 33P-radioactivity from the parallel 33P-phosphate uptake subsamples to determine the fraction (%) of 33P uptake that was incorporated into P (III) compounds. All uptake samples were processed at sea in May 2018 and all reduction samples were processed onshore in July 2018. Radioactive decay was accounted for in the final counts per minute (cpm) values.