In this study, biogeochemical regimes of Marguerite Bay and the adjacent part of the West Antarctic Peninsula (WAP) continental shelf were delineated through integration of nutrient, hydrographic, and biological measurements obtained during the LTER and SO GLOBEC studies during austral summer, autumn, and winter of 2001 and 2002. Marguerite Bay biogeochemical regime was found to differ from those of the adjacent WAP continental shelf. In terms of Treguer and Jacques (1992), Marguerite Bay is a combination of Coastal Continental Shelf Zone (CCSZ) and Seasonal Ice Zone (SIZ) distinguished by shallow mixing regime, high primary production and export production. At the end of the growing season (autumn) in both years, waters in Marguerite Bay were strongly depleted in nutrients (the deficits of total inorganic nitrogen (NO3-+NO2-+NH4+) and silica were >0.6 mol m-2 and >2.5 mol m-2, respectively). Observed ΔN/ΔP removal ratios of 10-12.5, lower than that of Redfield et al. (1963), and ΔSi/ΔN removal ratios as high as 4-5 indicated the dominance of diatoms. High autumnal ammonium stocks (>0.25 mol m-2) were observed in Marguerite Bay and were co-located with the areas of the highest nutrient deficits suggesting spatial coupling between primary and heterotrophic production during both years. Consistency of this feature was not disrupted by significant interannual variability of biological production in Marguerite Bay that resulted in ~30-50% reduction in nutrient deficits and ammonium stocks from the first year to the next.The other two biogeochemical regimes were at the central part of the continental shelf characterized by mixed phytoplankton community and at the outer shelf dominated by diatoms. Both regimes were characterized by considerably lower depletion of nutrients compare to those of the Marguerite Bay regime and were consistent between the two years. Interannual variability of biological production and possible sources of high ammonium stocks in Marguerite Bay were studied with a one-dimensional model, a modification of that of Walsh et al. (2001). The model attributed the decline in nutrient deficits to the difference in sea ice cover dynamics between two years. The greater sea ice presence led to the somewhat lower primary production during the second year compare to the fist one. Moreover, model's tight coupling between primary and bacterial production resulted in a decline of bacterial ammonification between the two years. Bacteria were found to be the primary source of ammonium in the Marguerite Bay model. Yet, 3-4-fold fluctuations in macro- and mesozooplankton biomass might have led to 15-25% variability in model's autumnal ammonium stocks.