Thank you. Thinking speculatively about a middle ground, I wonder what is minimally necessary in a model to accurately predict population dynamics - this intervention will tend to grow your colony, this other intervention will tend to decrease it. I think of higher fidelity models as supporting prediction of degree of population change - by how much up or down, and by when. But, I think a thing about "systems" is that you really can't separate dynamics from degree.
The thing is, I would expect every individual "intervention" to be modelled by its own model or equation (whether or not it's literally an equation is irrelevant, at the end all system dynamics are governed by equations), each of which has presumably been backed by some scientific research.
For instance, if bees are getting afflicted by some fungus, then there is some research that says that applying fungicide X will allow the colony to prevent y% decline in the ideal case, or whatever.
At that point, even though the model will then incorporate this "sub-model," you don't need the system dynamics model to say what will happen when you apply the fungicide, you only need that well-researched sub-model.
I feel (and I'm getting a little out of my depth because of my limited research on real-world uses of these models) that the system dynamics model is only useful for attempting to determine what happens when you have a conjunction of various interventions, or interventions with various environmental parameters that were outside the original research of the sub-model. But that's precisely were the limitations of the system dynamics model will come into play, because the non-linear or chaotic effects may mean that the model's prediction that the colony will grow or fail may have little to do with reality, because of the complexity of accurately accounting for so many parameters, and guessing how they interact without specific research on how they interact.
