Extracting phytoplankton physiological traits from batch and chemostat culture data

As the role of phytoplankton diversity in ocean biogeochemistry becomes widely recognized, the description of plankton in ocean ecological models is becoming more sophisticated. This means that a growing number of plankton physiological traits need to be determined for various species and under various growth conditions. We investigate how these traits can be estimated efficiently from common batch culture and chemostat experiments. We use the Metropolis algorithm, a random-walk Monte Carlo method, to estimate phytoplankton parameter values, along with the uncertainties in these values. First, we fit plankton physiological models to high-resolution batch culture and chemostat data sets to obtain parameter sets that are as accurate as possible. Then, we subsample these data sets and assess to which extent the accuracy is sacrificed when fewer measurements are taken. Two measurement points within the exponential growth stage of the batch culture data set are sufficient to constrain the maximum protein synthesis rate, the maximum photosynthesis rate, and the chlorophyll-to-nitrogen ratio. Two measurements during the stationary phase of the batch culture experiment are then enough to constrain the parameters related to carbon excretion and the photoacclimation time. From the chemostat experiment, only four measured points are needed to constrain the parameters connected with the internal reserve dynamics of phytoplankton. Thus, we demonstrate that traits related to key biogeochemical and physiological processes can be determined with only a few batch culture and chemostat measurements, as long as the measurement points are selected appropriately.