Biogeochemical and physical controls of nitrogen fluxes in a highly dynamic marine ecosystem—model and network flow analysis of the Baltic Sea

Time, space and ecological organisation are key features of an ecosystem that are rarely explored simultaneously. A dynamic coupled physical–biogeochemical model, with a high vertical and temporal resolution that describes the large-scale hydrography and nitrogen dynamics of the Baltic proper has been used as a tool to explore these scales in a flow network analysis of an ecosystem. The ‘sampling’ of nitrogen concentrations and flows in this ecosystem model can be done with a much higher frequency in space and time than is practically possible in the sea. The model can also be used to follow salt and water fluxes. Thus, it was also possible to evaluate the relative importance of biogeochemical and physical processes. Measures of total system throughput (TST), which is intended to be an overall system attribute, suggest that the physical contribution to nitrogen fluxes is minor compared with those caused by biogeochemical processes. However, measures of material cycling estimated as Finn cycling index (FCI), which also is intended to capture overall system performance, show variations that can be explained to a great extent by the physical contribution. This demonstrates the difficulties to separately analyse the contributions of physical versus biogeochemical processes on the same levels of organisation. Input–output analyses show only minor differences when comparing the system in steady state and in dynamical mode. The internal spatially resolved time characteristics of nitrogen and salt, in our case measured as residence time, shows slightly different values for the system, when comparing short and long time scales, that it is questionable to speak about long term means. The turnover time for salt shows, a little surprisingly, the opposite of what was earlier expected, i.e. longer turnover for the surface layers and shorter turnover for bottom layers. The residence time, calculated for a dynamic system, shows that the expected time for a unit of nitrogen to reside in the system is highly dependent on when during the year it is entered. If one unit of nitrogen is enterering in the spring–early summer the residence time is slightly less than a year. If entered in autumn–winter, the unit of nitrogen will reside about a year and a half. The large temporal and spatial variations in ecosystem properties, measured in the model also illustrate the great difficulties to extrapolate observations from short-term field measurements to an ecosystem scale