Role of oceanic iron biogeochemistry in atmospheric CO2 variability between the last glacial maximum and the holocene
During the last glacial cycle, atmospheric CO2 varied by about 80–100 ppmv between glacial and interglacial periods. A substantial fraction of this change is thought to reflect variations in the strength of the marine biological pump, driven in part by changes in iron (Fe) supply from mineral dust to highnutrient, low-chlorophyll (HNLC) regions. In this PhD thesis, I use the Earth system model of intermediate complexity cGENIE to investigate how aeolian dust deposition and the solubility of dust-borne Fe have influenced the ocean carbon cycle and atmospheric CO2 between the Last Glacial Maximum (LGM) and the Holocene. This thesis is organized around two main modelling studies. In the first part, I prescribe time-varying dust deposition both globally and in key HNLC regions (Southern Ocean sectors, North Pacific and equatorial Pacific) along the last deglaciation, while keeping the remaining regions at Holocene conditions. Over the full glacial–interglacial contrast, these experiments show that changes in dust–borne iron supply can explain a global atmospheric CO2 drawdown of up to ∼30 ppmv between typical LGM and Holocene states. This represents roughly one third of the observed 80–100 ppmv glacial–interglacial pCO2 difference. For each HNLC region, I derive empirical relationships linking iron input, export production and atmospheric CO2. When the deglacial dust changes in all HNLC regions are combined, they lead to a pCO2 increase of about 18–19 ppmv, with the Southern Ocean contributing roughly half of this signal and the North Pacific almost one third. The timing of the response is not uniform: Southern Hemisphere dust changes influence CO2 throughout the termination, whereas the impact of Northern Hemisphere dust is concentrated in the later stages. These experiments confirm the dominant role of the Southern Ocean, but also show that high-latitude Northern Hemisphere dust sources make a non-negligible contribution to the deglacial pCO2 rise. In the second part, I investigate how uncertainties in Fe solubility modify the dust–iron–pCO2 connection. Using the same family of global dust fields for the pre-industrial period and the LGM, I perform a large ensemble of equilibrium simulations in which dust-borne Fe solubility is varied over a wide range, together with an updated ligand scheme for the internal Fe cycle. The results indicate that the pCO2 response to increasing Fe solubility is strongly non-linear: for spatially heterogeneous solubility, the additional drawdown of LGM pCO2 saturates at about 30 ppmv relative to Holocene conditions, again close to one third of the full glacial–interglacial pCO2 difference. The strongest sensitivity occurs in the Southern Ocean band between ∼35◦S and the Subantarctic–Polar fronts, although other HNLC regions also contribute. Across the different dust reconstructions, some robust features emerge, in particular the saturation of the pCO2 response at high solubility and the central role of polar oceans. Taken together, these two studies provide a consistent picture of how dust fluxes and Fe solubility control the efficiency of the soft-tissue pump at the global scale. They demonstrate that dust–iron fertilization can explain a significant but limited share of glacial–interglacial pCO2 changes, and they quantify how model assumptions about Fe solubility and ligand complexation modify this contribution. The thesis also identifies important missing processes such as temporally variable solubility, interactive ligands, sediment and hydrothermal Fe sources, and circulation changes—that should be explored in future work. In this sense, the results presented here offer both a quantitative estimate of the dust–Fe effect and a framework for more comprehensive simulations that combine dust, solubility and ocean circulation in a unified modeling approach