Since the late 1970's increasing interest in the role of bacteria, protozoa and small phytoplankton in pelagic marine ecology has resulted in a new food web paradigm by which nanoplankton (2-20 um) and picoplankton (0.2-2 um) account for a large proportion of primary and secondary production. Bacteria assimilate dissolved organic matter (produced by phytoplankton exudation and zooplankton 'messy feeding') and flagellate protozoa graze upon the bacteria and picophytoplankton. Ciliates consume zoo- and phyto- flagellates and are in turn grazed on by crustacean zooplankton. This bacteria-protozoa food chain has become known as the 'microbial loop'.

The purpose of this study is to asses by modelling, the importance of the microbial loop as a carbon salvage pathway for mesozooplankton relative to direct grazing on net phytoplankton. The modelling process demonstrates the quality of existing data on marine microbial ecology and so the model's description transcends a simple methods section. The model divides organisms trophically and metrically and contains manifold time and depth intervals, allowing simulation of variation in microbial ecology to an unprecedented degree. Both carbon and nitrogen cycles are modelled and a 14C label can be included. The model "standard run" is shown to be stable over many years and to recover from perturbation. Sensitivity analysis has been carried out for spring and summer periods.

In the model, picoplankton, though productive, play a relatively minor role in mesozooplankton nutrition thus supporting the 'sink' hypothesis. However when nutrients are limiting, both bacteria and picophytoplankton out compete other phytoplankton thereby enhancing the relative role of the microbial loop. Nanozooflagellate grazing releases much of the picoplankton nutrient uptake. Nanophytoplankton are an important indirect food source for mesozooplankton via the microzooplankton, except in winter. The model shows considerable seasonal and spatial variation in community structure, and small diurnal changes. When applied to three differing environmental regions of the North Atlantic by varying initial winter nutrient and total irradiance levels the model results are generally in agreement with reported environmental observations.