Ability of a “minimum” microbial food web model to reproduce response patterns observed in mesocosms manipulated with N and P, glucose, and Si

We compared an idealised mathematical model of the lower part of the pelagic food web to experimental data from a mesocosm experiment in which the supplies of mineral nutrients (nitrogen and phosphorous), bioavailable dissolved organic carbon (BDOC, as glucose), and silicate were manipulated. The central hypothesis of the experiment was that bacterial consumption of BDOC depends on whether the growth rate of heterotrophic bacteria is limited by organic-C or by mineral nutrients. In previous work, this hypothesis was examined qualitatively using a conceptual food web model. Here we explore the extent to which a “simplest possible” mathematical version of this conceptual model can reproduce the observed dynamics. The model combines algal–bacterial competition for mineral nutrients (phosphorous) and accounts for alternative limitation of bacterial and diatom growth rates by organic carbon and by silicate, respectively. Due to a slower succession in the diatom–copepod, compared to the flagellate–ciliate link, silicate availability increases the magnitude and extends the duration of phytoplankton blooms induced by mineral nutrient addition. As a result, Si interferes negatively with bacterial consumption of BDOC consumption by increasing and prolonging algal–bacterial competition for mineral nutrients. In order to reproduce the difference in primary production between Si and non-Si amended treatments, we had to assume a carbon overflow mechanism in diatom C-fixation. This model satisfactorily reproduced central features observed in the mesocosm experiment, including the dynamics of glucose consumption, algal, bacterial, and mesozooplankton biomass. While the parameter set chosen allows the model to reproduce the pattern seen in bacterial production, we were not able to find a single set of parameters that simultaneously reproduces both the level and the pattern observed for bacterial production. Profound changes in bacterial morphology and stoichiometry were reported in glucose-amended mesocosms. Our “simplest possible” model with one bacterial population with fixed stoichiometry cannot reproduce this, and we suggest that a more elaborate representation of the bacterial community is required for more accurate reproduction of bacterial production.