Real and Reciprocal Constraints

During evolution, vertebrates can develop some morphological features (e.g. lose digits, reassign function to building blocks) but not others (grow a second head, or a third arm). Re-dimension and re-shape is ok, but not to add. This however is not true for DNA and protein, where bits and regions can be changed, added, multiplied. Nevertheless, I wonder if the morphological limitations can be translated into constraints for DNA/protein evolution. A bit like solvent flattening for crystallographic phases. Yes, maybe this is a circular argument.