As for question 1 there is a lot of great discussion here which has folks tackling this question in different ways
particularly jscolar comments.
Priors can get difficult quickly. However I tend to view them as just additional data you are adding into your model. So if you know some about how all your parameters interact you can code that up.
Like in my area fire is bad for wetland plants in arid system. If my model increase something for fire I might have a prior normal(-0.5, 0.5).