Planktonic ecosystems provide a key mechanism for the transfer of CO 2 from the atmosphere to the deep ocean via the so-called "biological pump".Mathematical models of these ecosystems have been used to predict CO 2 uptake in surface waters, and more recently have been embedded in global climate models. While the equilibrium properties of these models are well studied, less attention has been paid to their response to external perturbations, despite the fact that, as a result of the variability of environmental forcing, such ecosystems are rarely, if ever, in equilibrium. Human induced perturbations to these ecosystems, namely the addition of limiting nutrients (e.g. iron) to areas where nitrate is plentiful to accelerate the biological pump, have been proposed as a solution to reduce atmospheric CO 2 . In this study, linear theory is used to determine the structure of initially "unit-norm" perturbations to state variables of a five state variable ecosystem model in steady state, describing Ocean Station P (50 • N 145 • W) in summer, that optimize either instantaneous export flux of organic matter at fixed times or integrated export as the ecosystem relaxes towards equilibrium. A common feature in the optimization experiments for both instantaneous and integrated flux is the synchronization of the oscillatory behavior between two state variables. Because of these oscillations, there is an indirect contribution to the export flux that is non-intuitive. For all perturbations, it is found that the flux to higher trophic levels is the primary contributor to export flux, and, contrary to expectations, the contribution of aggregation is negligible.