Modelling multi-nutrient interactions in phytoplankton; balancing simplicity and realism

Modelling multi-nutrient interactions in phytoplankton growth is considered in the context of balancing the level of complexity with the adequacy of output for ammonium, nitrate, P, Si, Fe, temperature and light limitations. Phytoplankton submodels for placement in ecosystem simulators should be capable of producing a believable output. This requires not only a realistic growth rate response to limiting nutrients but also a realistic consumption of non' or lesser limiting nutrients. The Monod structure, and allies, fails these requirements. The complexity of the Droop version of the quota model is not matched by flexibility, especially for the description of Si assimilation. A normalised version of the Caperon-Meyer quota model is a better (and no more expensive) structure that, by the addition of a feedback controlled uptake equation and a Monod-type Si assimilation control, gives considerable flexibility at reasonable computational cost. Once the step is taken to include one complex mechanistic component there is considerable advantage to be gained from adding further mechanistic structures with little more cost in integration time. This is especially so for light-iron interactions. Overly simplistic models should not be used just because they offer advantages in computational costs at the expense of realism, even if they give satisfactory fits to a particular data set, as output is more likely to be erroneous in 'what-if' scenarios and also during simulation of data-poor periods of extant data series. (C) 2003 Elsevier Science Ltd. All rights reserved.