Likelihood-free Inference Reveals the Mechanisms and Parameters Governing Root System Architecture

Plasticity of the root system is essential to a plant’s continued acquisition of nutrients in a heterogeneous and temporally unstable soil environment, with no two specimens having the same root system architecture. Complex root systems can be computationally simulated with relatively few generative parameters, and many models exist to generate a wide variety of systems in varied environments. This talk will detail the use of Approximate Bayesian Computation (ABC) methods to solve the ‘inverse problem’ – inferring the growth parameters of existing root systems from data describing root system architecture. ABC is a highly adaptable statistical technique which approximates the likelihood function when it is not easily computed. This method is used to investigate generative parameters for lab-grown Arabidopsis Thaliana root systems, and for model selection for models including root elongation and lateral root initiation points.