I agree with you that carefully defined boundaries are critical to system modelling.
I am wondering if you consider the time frame (short term, longer term) to be part of system boundaries or if it is merely implied by the data selected for use in the system dynamics model?
I think it would be useful to make it explicit so that the modeler is consistent in the model design.
Also, have you used Elinor Ostrom's work? I have found her work on the Commons to be very useful.