Just because the conditions I alluded to are man-made doesn’t mean other similar conditions don’t occur in nature.
I tend to think mathematically. I can conceive of local fitness maxima that are not global. For example, you might have traits tailored to a particular local environment which tend to evolve at the expense of some others which might more often be useful. “More often useful”is what I think of as globally optimized. Do you look at it differently?
I can see two things happening here. One is that a species cut down to a small area is vastly more susceptible to accidental extinction due to floods, hard winters, disease or whatever. Two is the fact that the first time ordinary dogs, cats, and rats get introduced into one of Darwin’s island paradises, many of the exotic creatures get wiped out.