I'm in applied math myself.
I agree that this is a problem with all sciences; but I feel that in most fields (certainly in anything physics based) one starts with equations of motion or whatever for some system and the question is how do I implement this computationally. That might be really hard like in the case of geology or climate modeling, but at least you have a fairly good idea of what the system is you're trying to simulate (granted modulo some issues like cloud formation in climate science etc.)
In mathematical biology a lot of the time the process is the inverse of this. Some phenomenon is observed (in this case menopause), and then people try to cook up some model that reproduces it. The problem is that this inversion procedure is not well posed. There are many different models you can formulate that will give you the phenomenon in question.
It might be the case that the above procedure is the only way to try to study phenomena in complex biological systems. But in this case I think one has to be really aware of the fact that reproducing the phenomenon does not mean one captured some real truth in one's model.
And my other problem with a lot of the work in this field is what I stated before. Presumably the models in geology you're referring to make some quantitative predictions? They may be hard do verify experimentally but still. The authors of agent based models in biology like this often don't even claim any quantitative predictions can be made with the model; the game is simply lets see if these rules for the agents make them do something interesting. Again, if a tight connection to the real world dynamics is not made, and one can't actually get quantitative predictions from the model than what's the point? If they analyzed the model analytically maybe at least there would be interesting math, but it seems like the model was just simulated. I really dont understand who benefits from stuff like this.
That was really long. Sorry.