Log-ratio of silica to aluminium counts (ln(Si/Al)) from ODP site 108-658

Growing evidence suggests that the low atmospheric CO2 concentration of the ice ages resulted from enhanced storage of CO2 in the ocean interior, largely as a result of changes in the Southern Ocean1. Early in the most recent deglaciation, a reduction in North Atlantic overturning circulation seems to have driven CO2 release from the Southern Ocean**2, 3, 4, 5, but the mechanism connecting the North Atlantic and the Southern Ocean remains unclear. Biogenic opal export in the low-latitude ocean relies on silicate from the underlying thermocline, the concentration of which is affected by the circulation of the ocean interior. Here we report a record of biogenic opal export from a coastal upwelling system off the coast of northwest Africa that shows pronounced opal maxima during each glacial termination over the past 550,000 years. These opal peaks are consistent with a strong deglacial reduction in the formation of silicate-poor glacial North Atlantic intermediate water**2 (GNAIW). The loss of GNAIW allowed mixing with underlying silicate-rich deep water to increase the silicate supply to the surface ocean. An increase in westerly-wind-driven upwelling in the Southern Ocean in response to the North Atlantic change has been proposed to drive the deglacial rise in atmospheric CO2 (refs 3, 4). However, such a circulation change would have accelerated the formation of Antarctic intermediate water and sub-Antarctic mode water, which today have as little silicate as North Atlantic Deep Water and would have thus maintained low silicate concentrations in the Atlantic thermocline. The deglacial opal maxima reported here suggest an alternative mechanism for the deglacial CO2 release**5, 6. Just as the reduction in GNAIW led to upward silicate transport, it should also have allowed the downward mixing of warm, low-density surface water to reach into the deep ocean. The resulting decrease in the density of the deep Atlantic relative to the Southern Ocean surface promoted Antarctic overturning, which released CO2 to the atmosphere.

(Tables 2) The bPSi:POC ratio of the total diatom community, and abundance of full, empty, and broken diatom cells compared to heavily silicified diatoms during the EIFEX experiment

During the European Iron Fertilisation Experiment (EIFEX), performed in the Southern Ocean, we investigated the reactions of different phytoplankton size classes to iron fertilization, applying measurements of size fractionated pigments, particulate organic matter, microscopy, and flow cytometry. Chlorophyll a (Chl a) concentrations at 20-m depth increased more than fivefold following fertilization through day 26, while concentrations of particulate organic carbon (POC), nitrogen (PON), and phosphorus (POP) roughly doubled through day 29. Concentrations of Chl a and particulate organic matter decreased toward the end of the experiment, indicating the demise of the iron-induced phytoplankton bloom. Despite a decrease in total diatom biomass at the end of the experiment, biogenic particulate silicate (bPSi) concentrations increased steadily due to a relative increase of heavily silicified diatoms. Although diatoms >10 µm were the main beneficiaries of iron fertilization, the growth of small diatoms (2-8 mm) was also enhanced, leading to a shift from a haptophyte- to a diatom-dominated community in this size fraction. The total biomass had lower than Redfield C : N, N : P, and C : P ratios but did not show significant trends after iron fertilization. This concealed various alterations in the elemental composition of the different size fractions. The microplankton (>20 µm) showed decreasing C : N and increasing N : P and C : P ratios, possibly caused by increased N uptake and the consumption of cellular P pools. The nanoplankton (2-20 µm) showed almost constant C : N and decreasing N : P and C : P ratios. Our results suggest that the latter is caused by a shift in composition of taxonomic groups.