Antarctic Ice Sheet fertilises the Southern Ocean

Southern Ocean (SO) marine primary productivity (PP) is limited by the availability of iron in surface waters, such that variations in iron supply to the SO are thought to exert a major control upon atmospheric CO₂ concentrations on glacial/interglacial timescales. The zone bordering the Antarctic Ice Sheet exhibits high PP, exhibiting seasonal plankton blooms in response to elevated dissolved iron concentrations. The source of iron stimulating these PP increases is in debate, traditionally ascribed contributors being aeolian dust, coastal sediments/upwelling and sea ice. More recently, icebergs and glacial meltwater have been suggested as sources. Data from glacial meltwaters worldwide indicate that sub-Antarctic meltwaters are likely to be anoxic, as a result of long flow paths and little surface input of oxygenated meltwaters. Hence, it is probable that they are rich in dissolved iron (as Fe(II)), acquired via the oxidation of sulphide minerals in sediments. In contrast, iron in iceberg rafted debris is dominated by iron oxyhydroxides, generated in oxic sectors of the ice sheet bed by regelation processes or entrained in icebergs as they pass over shelf sediments. The potential for iron from both these ice sheet sources to impact PP has not yet been quantified. Here we apply the MIT marine ecosystem model to determine the potential impact of ice sheet iron export on SO PP. Fluxes of iceberg and meltwater-derived iron are focused along major ice stream corridors, and enhance iron concentrations in surface ocean waters. The impact on SO PP is greatest in coastal regions, including the Ross Sea, Weddell Sea and Amundsen Sea, all of which are areas of high observed marine PP. Inclusion of ice sheet iron sources in modelled scenarios raises SO PP by 10-30%, and provides a plausible explanation for very high seasonally observed PP around the coastal zone. These results highlight Antarctic runoff and icebergs as previously neglected sources of bioavailable iron to the SO, and suggest consideration of both these sources in future biogeochemical modelling.