Modelling mixotrophy in harmful algal blooms: More or less the sum of the parts?

Abstract

Mixotrophs are significant components of planktonic food webs, are frequently associated with harmful algal bloom events, and thus warrant inclusion in coastal ecosystem models. There are, however, insufficient quantitative data to support the construction and testing of simple empirical descriptions of mixotrophs. Here, a complex mixotroph model based upon phenomenological understanding (Flynn and Mitra, 2009) was used to generate control “realities” against which to compare contrasting simple descriptions of mixotrophy using a Turing Test approach. The simplest description, adding together phototrophic and heterotrophic functions gave the worst output. The best model tested, in keeping with the evolution of these organisms, used phototrophy as a nutritional supplement mechanism for heterotrophy. However, none of the simple models described kleptochloroplasty — an important process in some harmful bloom species. None of the simple models correctly matched the balance of phototrophy and heterotrophy (grazing); while fits to bulk parameters (biomass, nutrients) could be acceptable, rate processes were often completely in error. This is of particular concern because of the difficulty in determining rate processes. A generalised implication is that a fit to bulk data gives no assurance that the model structure is not dangerously dysfunctional; determining model skill should include locating and removing structural dysfunctionality.