Cited references to original works should always be preferable to wiki links or quotes. The wiki links/quotes may, of course, be correct, but they don't have the appropriate weight of evidence behind them:
- A wiki page can be vandalized (or even just changed/deleted/merged) thus rendering a link invalid.
- A quote from a wiki page may be rendered invalid by a subsequent edit that changes the wiki page.
I don't think you can draw an exact line based on number or extent of changes; there will always be blurry areas around such a line. You'll need to use your own judgement. That said:
- If the wiki-quoting answer is not an accepted answer then it seems OK to me to add a new answer incorporating cited references.
- If the wiki-quoting answer is an accepted answer, then you need to tread a little more carefully.
- If the accepted answer is incorrect then by all means add your own answer. You can't really edit an incorrect answer to make it correct.
- If the accepted answer is correct then all you're really doing is adding evidence.
- In which case: is it even necessary to go near this question or answer?
I personally prefer to err on the side of respecting the work someone else put in to answering a question. Editing an answer (aside from obvious cases such as spelling/grammar/formatting) just presses all the wrong buttons with me, so adding my own answer would be my preferred course of action in any event. Prefacing it with some text to the effect of "the other answer is correct but here's some more evidence" is an approach I'd favour. Adding the additional supporting evidence as a comment seems even better, but of course you then need to deal with the comment-limit.