I think the selective value you propose relies on the evolution of the " animal " with respect to the " plant ", but it doesn't seem as if it could really work that way, given the willow bark example in my response above ( humans pre-evolving a reason to eat willow bark before ever encountering willow bark ? ).
