"Symbolic regression algorithm sheds light on resource dimensionality in microbial ecology"Sun, AnthonyBacterial communities constitute handy experimental systems in studying macroecological patterns. Microbial interactions are mainly mediated by strain-specific resource consumption traits, which determine the interplay between populations and their environment. Although the consumption of a single limiting resource is well-understood, the literature has proposed contradictory models to account for multi-resource environments. Moreover, we lack methods to observe the resource environment in laboratory settings, directly or indirectly. Here, we use theoretical insights to guide the computational search of suitable mathematical models for microbial interactions. Using symbolic regression, we are able to retrieve classical generative models from simulated data. When applied to experimental data, the algorithm provides information on the effective complexity of the resource environment from the point of view of a bacterial population in laboratory experiments. Thus, our study demonstrates how computational methods can be used to enhance the dialogue between mathematical modelling and empirical approaches in ecology. Indeed, their synergy offers both an empirical evaluation of available theory and indirect observation of ecological dimensions previously difficult to access. |
« back