The Ross Sea, the southernmost sea on Earth, presents several iconic features of polar seas: sites of deep water formation, high summer primary production, floating ice shelves, the annual cycle of advance and retreat of sea ice, polynyas and katabatic winds. Furthermore, sea ice in McMurdo sound (western Ross Sea) is one of the most productive marine environments. However, sea ice inorganic carbon dynamics and related air-ice CO2 fluxes have never been documented in the Ross Sea.Two surveys were carried out in the western Ross Sea to bridge over a critical gap in the current understanding of sea ice: autumn and winter processes. The land-based YROSIAE project was a temporal survey from late winter to summer within landfast sea ice. The ship-based PIPERS project was an unique opportunity to study the early stages of sea ice formation (in polynyas) and more common consolidated pack ice in autumn. Based on these two consistent surveys, this work aims to (i) examine the bulk ice pCO2 dynamics in landfast sea ice from late winter to summer (ii) investigate the seasonal pattern (net source vs net sink) and diurnal pattern of air-ice CO2 fluxes (iii) analyse the depth-dependent physical and biogeochemical processes involved in inorganic carbon dynamics (iv) assess the precipitation of calcium carbonate in autumn and during a full bloom season.CO2 fluxes were measured using the chamber technique in autumn, late winter and spring, over open surface water, frazil ice patch, grey unconsolidated ice and consolidated first-year ice. These new autumn and winter data provide a first step to set up the budget of air-ice CO2 fluxes over the year and evaluate the large-scale influence of these fluxes on the annual uptake of CO2 by ice-covered oceans. Our results confirm that sea ice acts as a CO2 source for the atmosphere during ice growth, with enhanced fluxes reported at the early stages of sea ice formation, and shifts to a sink in spring. In late spring, diel pattern superimposed upon this seasonal pattern and was potentially assigned to either ice skin freeze-thaw cycles or diel changes in net community production. The snowpack plays a complex role in CO2 exchanges and can no longer be considered as an inert reservoir lying at the sea ice surface.The main features of the normalized DIC distribution (DIC35) through the ice column were: (i) a marked depletion at the surface from autumn to spring induced by the CO2 releases to the atmosphere (ii) bubble-driven gas enrichment below or within impermeable layers and (iii) an initial DIC35 enrichment in the bottom layer disappearing in spring when the seasonal peak in biomass occurs. At the bottom of landfast ice, in spring, a particular assemblage of microorganisms, the biofilm, led to a massive biomass build-up counterintuitively associated with nutrients accumulation. This biofilm formation may also promote calcium carbonate precipitation. However, in young pack ice or in cold landfast ice in early spring, limited calcium carbonate precipitation was reported. This suggests that calcium carbonate precipitation is not an ubiquitous process, especially in winter and autumn Antarctic sea ice. Comparison of calcium carbonate precipitation and pCO2 measurements advocates that the calcium carbonate precipitation is rather controlled by pCO2 than temperature. / Doctorat en Sciences / info:eu-repo/semantics/nonPublished