Bayesian emulation and calibration of an individual-based model of microbial communities

•Description of a quantitative framework for inferring and identifying the most relevant parameters for biofilm simulation.•A combination of the dynamic linear model (DLM) and Gaussian process (GP) regression under a Bayesian framework is explored for the emulation.•The univariate dynamic linear model and Gaussian process (DLMGP) emulator and calibration procedure was extended to the multivariate/matrix-variate cases.•The effect of different values of discounting factors on the DLM models is explored.•The incorporation of small-scale variation in the form of nugget effect in our model formulation improved numerical stability.•The uncertainty levels in the maximum specific growth rate for HET and AOB bacteria have been considerably reduced.