Seawater was collected into acid washed, ultra-pure water and sample rinsed, clear polycarbonate incubation bottles. PO4 assimilation rates were measured in triplicate 70-mL samples with ~259 kBq of added 33P-PO4 (Perkin-Elmer #NEZ08000; carrier free), incubated for 30 min to 1h. CO2 fixation rates were measured in duplicate 70-mL samples with ~17 MBq of added 14C-sodium bicarbonate (Perkin Elmer #NEC086H000, 1.6 GBq/mmol), incubated from dawn to dusk. Samples were incubated under simulated light and temperature conditions measured at the sampling site. A killed control sample was also prepared by adding paraformaldehyde (PFA, 2 % final concentration prepared with electron microscopy grade 16 % aqueous solution, Electron Microscopy Sciences) at least 15 minutes before introducing the radioisotope, in order to account for unincorporated radioactivity. At the end of incubation, samples were fixed with PFA (2% final, for 15-min in the dark), and triplicate 20-microliters aliquots were sampled to measure the total radioactivity added (with beta-phenylethylamine for 14C samples). The total microbial activity was determined by filtering a 3-mL aliquot through a 0.2-micron, pore-size polycarbonate membrane filter (Nuclepore). To reduce unincorporated 33P-PO4, the membrane filter was placed onto a filter type HA soaked in 100 mM KH2PO4, then rinsed three times with ~1 mL of 0.2-micron filtered seawater. To remove unincorporated 14C-sodium bicarbonate, the filter was acidified with 0.5 mL of 1N HCl for 24 h. To determine plankton groups specific uptake rates, a 20-mL aliquot was passed through a 0.2-micron polycarbonate membrane filter under gentle vacuum filtration, and the remaining volume from the 70-mL incubation bottle was passed through a 0.8-micron polycarbonate membrane filter. The 0.2-micron and 0.8-micron filters were stored in separate cryovials with 2 mL and 4 mL of the corresponding radiolabeled sample, respectively, vortexed to detach the cells from the filter, then flash frozen for later flow cytometric sorting (see below). The added radioactivity and total microbial activity were assayed by liquid scintillation counting in 7-mL plastic scintillation vials (Simport) with 4 mL of scintillation cocktail (Ultima Gold LLT, Perkin Elmer) added.

Turnover times (TPO4, h) were calculated by dividing the total radioactivity added (Bq L–1) by the rate of radiolabel uptake into the particulate fraction (Bq L–1 h–1). PO4 assimilation rates (nmol P L–1 h–1) were calculated by dividing PO4 concentration by TPO4. We used PO4 concentration estimated from a concentration series bioassays following the method of Wright and Hobbie (1966). Briefly, seawater samples were amended with non-radioactive PO4 to target additions of 0, 5, 10, 25, 50, 75, and 150 nmol PO4 L–1, spiked with 33P-PO4, incubated and sampled as described above. The resulting TPO4 values were plotted against a corresponding concentration of PO4, and extrapolated using linear regression (TPO4 = a x PO4 + b) to estimate the ambient concentration (Sn = b/a), which represents an upper estimate of ambient concentrations as detailed in Zubkov and Tarran (2005). Results from these bioassays were also used to calculate the Michaelis-Menten kinetic parameters for PO4 assimilation rates (Vmax, the maximum rate at saturating substrate concentration and Km, the half-saturation constant).

For cell sorting of Prochlorococcus, Synechococcus, pigmented and non-pigmented protists, the Influx flow cytometer was set at the highest sorting purity (1.0 drop single mode) and potential attached cells were discarded using a pulse width vs. forward scatter plot. The drop delay was calibrated using Accudrop Beads (BD Biosciences, USA) and verified manually by sorting a specified number of reference beads onto a glass slide and counting the beads under an epifluorescence microscope. Performance was validated as described in Duhamel et al. (2018). Three to four proportional numbers of cells from the same incubation sample were sorted for each target population. Sorted cells were assessed by liquid scintillation analysis following previously published protocols (Talarmin et al. 2011; Duhamel et al. 2012; Rii et al. 2016). The 14C-labeled samples were acidified with 1 mL of 2 mol L-1 HCl for 24 h to remove any unincorporated 14C-sodium bicarbonate.

For each group, at least three samples were sorted and regression analysis between the number of cells sorted and the radioactivity taken up by the sorted cells was used to calculate the per cell activity (dpm cell−1). Radioactivity in sorted cells from the PFA-killed control samples (dpm cell−1) was deduced from radioactivity in the sorted cells from the respective samples (dpm cell−1). The cell-specific assimilation rate (nmol cell-1 h-1) was calculated by dividing the radioactivity per cell (dpm cell−1) by the total microbial activity (dpm L−1) measured in the same sample, and then multiplied by the total microbial assimilation rate at ambient substrate concentration (nmol L−1 h−1).

Michaelis– Menten kinetic parameters were determined using the Michaelis–Menten model in Prism 6.