It doesn’t require any such thing, no.
It obviously is indeed fact.
One need only look at the sequences, infer a tree, and then use one’s brain to figure out what those mutational histories would have to look like. Where the sequences differ is where mutations occurred.
If species A has AAAATTTAA and species B has AAGATTTAA then there was an A<->G transition in either A or B. And a rooted phylogeny with more species could be used to infer which is the ancestral state. Simple principle that can basically applied to any observed difference. Be those duplications, insertions, deletions etc.