Gene Expression this morning is worth some thought, a post about the mtDNA of Andaman Islanders and their connections to mainland Asian populations. "Present genetic variation is a weak guide to past genetic variation". In a nutshell, some anthropologists and geneticists had hoped that Andaman Island people were a kind of "time capsule" of the original migration of people out of Africa. The mtDNA lineages are inconsistent with that hypothesis.
On a final note, if the Andaman Islanders arrived ~20 thousand years before the present from the South Asian mainland they don’t tell us very much about the “Out of Africa” people. They’re not “living fossils,” and it was frankly somewhat stupid probably to think they would be.
I don't have time at the moment to do my own review but definitely there is a deeper issue at play. It is extremely interesting that we're finding the Andaman Island population fits into the genetic landscape of South Asia at the Last Glacial Maximum, and not earlier. Even if the islands were first inhabited at the LGM, we might expect early inhabitants to preserve variation that had later been supplanted within South and Southeast Asia by the spread of agriculturalists. Apparently, they don't. It is likewise extremely interesting that Neolithic European mtDNA is predominated by haplogroups that are rare or absent in earlier Europeans. With a fuller review, I think we could likely come up with several more instances where fairly large pre-agricultural turnover was happening...I have two or three in mind.
These observations show that the present distribution of genetic variation is in some ways completely unrepresentative of the patterns in the past. The thing that strikes me: It takes a pretty massive demographic turnover to make this happen. And what we're looking at in today's populations is many, many instances of such turnovers during the last 20,000 years.
I've spent a good part of my career as a voice in the wilderness, saying that things just aren't simple enough to use genetics and a Wright-Fisher population model to reconstruct events before the Neolithic. But in many ways, mainstream geneticists weren't making an unreasonable assumption that one might reconstruct those events in a straightforward way using mtDNA or the Y chromosome. It's just that reality is stranger than they expected.