authors: Coralie Fritsch, Jérôme Harmand, Fabien Campillo
Why I like it: This is an easy one to overlook, because the highlights and abstract are not well written and the title is opaque. It is worth persevering, however, because the methods, results and discussion are interesting. For a simple population system, the paper compares a continuous, deterministic model (i.e. a traditional deterministic ODE model) with a discrete, stochastic model (an individual based model or IBM), a discrete, deterministic model (an IDE) and a continuous, stochastic model. They evaluate where the models converge and where they diverge, showing that there are some classes of scenarios where the higher computational performance of the continuous models pays off, while there are other population dynamics cases that the these models are unable to reproduce, which often occur at small population sizes. They advocate for the development of hybrid models to handle situations that may fluctuate between large and small population sizes, or include some large and other small populations. This is an intriguing idea and I guess it is, in a way, the approach taken in the model, Atlantis, that uses traditional mechanistic models to simulate phytoplankton production but agent-based models to represent fish and other animals.









