A growth-induced dispersal model of non-motile cells calibrated against time-lapse microscopy

Abstract

One of the main techniques used in Synthetic Ecology is to perform experiments on simple synthetic communities under controlled environmental conditions to isolate and hence infer the mechanisms underlying the community dynamics. We commonly use isogenic strains grown in agar medium of known composition. For non-motile strains that interact with each other through diffusible molecules in the environment, spatial structure plays an important role in predicting community dynamics. Continuum models have proven to be one of the best choices of spatially explicit models to study spatiotemporal dynamics at the scale of diffusing molecules. However, there have not been many efforts towards calibrating such models in order to gain statistically reliable predictions. In this thesis, I extend a continuum model of density-dependent diffusion by augmenting it with terms that capture growth-induced dispersal. Using a reproducible protocol of growing bacterial cultures on agarose pads for imaging under a microscope, I gather time-series data of an E. coli monoculture. Following a pipeline of dynamic model calibration, I estimate the model parameters and infer avenues for improving the model formulation and the experimental design. Furthermore, I present a multispecies decomposition of the model and apply it to compare model predictions against preliminary data from a co-culture of antibiotic-resistant strains displaying cross-protection mutualism.