Using nitrogen as the limiting nutrient, the default version of a microplankton-detritus model linked chlorophyll concentration to the autotroph nitrogen. However, phosphorus dynamics were added to simulate the results of a microcosm experiment. Using standard parameter values with a single value of microheterotroph fraction in the microplankton taken from the observed range, the best simulation successfully captured the main features of the time-courses of chlorophyll and particulate organic carbon, nitrogen and phosphorus, with root-mean-square error equivalent to 29% of particulate concentration. A standard version of microbiological model assumes complete internal cycling of nutrient elements; adding a term for ammonium and phosphate excretion by microheterotrophs did not significantly improve predictions. Relaxing the requirement for constant microheterotroph fraction resulted in an autotroph-heterotroph model AH, with dynamics resembling those of a Lotka-Volterra predator-prey system. AH fitted the microcosm data worse than did MP, justifying the suppression of Lotka-Volterra dynamics in MP. The paper concludes with a discussion of possible reasons for the success of the simple bulk dynamics of MP in simulating microplankton behaviour.