I just returned from Tromø, Norway, from a workshop on “modeling marine systems from first principles”. I should probably concentrate on the scientific part here, but I feel I can’t go there before saying that the venue was stunning – the meeting was organized at Sommarøy, a small island close to Tromø, and for the first days we were incredibly lucky to see both polar light at night and sun at daytime, just great!
The idea of the workshop was to bring people with different backgrounds together to discuss whether it is possible (and useful) to build minimal models of marine systems only from a set of biophysical constraints such as conservation of mass and energy, etc., while randomizing other aspects that are usually modeled, such as functional responses, etc. An example of such a model would be this study of Mullon et al., who was also present at the workshop.
We had a lot of good discussions on that subject, and I think this is an interesting approach that could certainly find some applications in terrestrial ecology as well (Hubbel’s UNTB comes to mind of course). One thought over which I personally pondered is how to use such a minimal (or null?) model in the right way (as did others before). I guess it could both act as a reference for more complicated models, in the sense that a model that includes more informed species-specific processes should lead to better predictions before we decide that these processes need to be included, and for explicitly rejecting randomness as the reason for observed population patterns, so basically model selection vs. rejection. But both can be tricky for dynamic models because appropriate summary statistics usually need to be picked to compare models, which can be connected with surprising statistical subtleties, e.g. for doing model selection. Anyway, an interesting route for ecological modeling I think, I guess time will show where it leads to!