Classical chemostat models such as the Monod, Marr-Pirt and Droop models are formulated at the population level. These models are unstructured, which means that all individuals in the population are treated as being identical. Such models fail to describe the experimental data of Dent et al. (1976, Arch. Microbiol. 109, 187-194) in detail. They grew vegetative myxamoebae of the cellular slime mould Dictyostelium discoideum in continuous culture with a bacterial food source, Escherichia coli (\mathrm{B} / \mathrm{r}) fed glucose.
In this paper a new structured model is proposed, based on dynamic energy budgets (DEB) for conspecific individuals which only interact via a common resource. The model fit for the time-course data of glucose, bacteria and myxamoebae is very good; it covers variations in mean cell volumes of both bacteria and myxamoebae. Comparison with curve fitting results for the classical models reveals the mechanisms that are responsible for the better performance of the DEB model. We show that elements from different models, specifically maintenance (Marr-Pirt, gives stability) and energy reserves (Droop, gives oscillations) must be combined to produce acceptable fits. Therefore we reject the assumption made by Bazin & Saunders (1978, Nature, Lond. 275, 52-54) that additional intra-specific interactions must be postulated to explain the data