Plants develop diverse root system architectures (RSAs) to capture soil water and nutrients. Understanding the development of RSAs is difficult because they consist of numerous interconnected roots which develop in parallel in response to many intrinsic and environmental cues. RSA models have proved useful to deal with that complexity, yet they suffer from the lack of quality growth data for roots grown in soils.

We present recent advances in the construction, calibration and validation of RSA models. Models use density distributions and time-delay partial differential equations to describe root developmental mechanisms. We also present methods to estimate root density distribution from experimental data (kernel-based estimators) and test different optimisation algorithms to parameterise the model on such data. Results show that density-based models can be parameterised accurately from experimental density distribution function, paving the way for new approaches to understand growth responses to environmental cues.

This study shows that density-based models provide a simpler link with experimental data, thereby complementing existent RSA models and setting the stage for characterisation of root properties and investigation of root functions.