To reconcile the presence of a high biogenic iron pool prior to the development of the diatom bloom when
TGX-221 concentrations were low (Fig. 9), we developed a simple 3-box model (Fig. 10) that incorporates the exchange fluxes between the three pools (lithogenic (M1), dissolved (M2) and biogenic (M3)) obtained for the budget presented in Fig. 9 (see Supplementary material for model details). Initially the model was run to steady-state, and then perturbed by increasing the biological uptake of iron from 2.5 μmol m− 2 d− 1 to match the rate reported by Boyd et al. (2012) of 7.8 μmol m− 2 d− 1 at the onset of the bloom. In the first model simulation, we set the other input flux (lithogenic dissolution (F2), iron regeneration (F4) and iron diffusion (F7)) and output flux (scavenging (F5)) rates to zero to mimic what might occur at the start of the bloom with respect to iron loss from the dissolved pool as biogenic iron is formed during increasing phytoplankton
biomass production. Note that in the model mixed layer depth was held constant during our simulations.