Iron enrichment to high-nutrient low-chlorophyll (HNLC) regions is being considered as a possible way of atmospheric CO2 sequestration to the deep sea. Mesoscale iron-enrichment to the HNLC subarctic Pacific induced a massive diatom bloom and led to a large decrease in pCO2. In response to the diatom bloom, the heterotrophic dinoflagellate Gyrodinium sp. increased and phagotrophically fed on the diatoms up to 12 times their length. Mathematical simulations show the carbon fixed by diatoms is mostly respired by Gyrodinium sp. in the sea surface. The emergence of initially rare species and their key biogeochemical roles were unexpected due to our limited understanding of food-web components. This indicates that the prediction of ecosystem responses to natural or anthropogenic perturbation remains a challenging issue. Effective carbon sequestration as a geoengineering technique may not be accomplished by purposeful iron-enrichment, at least in the western subarctic Pacific where rapid-growth diatom grazers stand by.