An improved model of carbon and nutrient dynamics in the microbial food web in marine enclosures

A description of an improved dynamic simulation model of a marine enclosure 1s given. New features In the model are the inclusion of picoalgae and rnixotrophs; the ability of bacteria to take up dissolved inorganic nutrients directly; and, for the phytoplankton functional groups, the inclusion of luxury uptake and the decoupling of the nutrient uptake dynamics from carbon-assimilat~on dynamics. This last feature implies dynamically variable phosphorus/carbon and nitrogedcarbon ratios. The model was calibrated with experimental results from enclosure experiments carried out in Knebel Vig, a shallow microtidal land-locked fjord in Denmark, and verified with results from enclosure experiments in Hylsfjord, a deep and salinity-stratified Norwegian fjord. Both observations and model simu-lation~ showed dominance of a microbial food web in control enclosures with low productivity. In N- and P-enriched enclosures a classical food web developed, while an intermediate system was found in N-, P-and Si-enriched enclosures. Mixotrophic flagellates were most important in the nutrient-limited control enclosures where they accounted for 49% of the pigmented biomass and about 4 8 % of the primary production. Lumping the mixotrophs in the simulation model with either the autotrophic or the heterotrophic functional groups reduced total primary production by 74 %. Model-derived, timeaveraged phosphorus budgets suggested that bacteria competed with algae for orthophosphate in the control enclosure, but not in the enclosure to which N and P had been added, where bacteria functioned as net mineralisers of phosphate. In the N, P and Si enclosure, bacteria took up only 10% of the amount of orthophosphate taken up by the primary producers, passing most of the organic phosphorus on to their grazers, the heterotrophic nanoflagellates, and mineralising only a small fraction directly. Inclusion of luxury nutrient uptake affected the simulation of the nutrient-enriched enclosures, while the decoupling of carbon and nutrient dynamics affected the simulation of the control enclosure. Without these 2 processes it was not possible to simulate the carbon and nutrient dynamics in the different enclosures adequately with the same parameterisation.