A recent report in PNAS shows the inhabitants of the Ukraine and its environs had much darker hair, skin, and eyes back in the Bronze Age and earlier.
Assuming population continuity, the selective advantage of the alleles they examined must have been very high. In order to see if there had been population continuity, they looked at the mtDNA frequencies of the ancient populations and compared them with mtDNA frequencies of modern populations in the same areas. Since they’re different, but not wildly different, they conclude that there has been population continuity, which was their null assumption.
That null assumption might have been reasonable, if someone had burned every history book ever written, while at the same time suppressing all the ancient DNA evidence.
Since that has not yet happened, I think their assumption is downright embarrassing. People have been moving in and out of this area for all of recorded history (as Razib Khan has also pointed out) : Cimmerians, Scythians, Goths, Khazars, Kievian Rus, the Golden Horde, eventually Russians.
There is no logical reason for geneticists to ignore information outside their field. Ignorance is no excuse. I could say the same for every other discipline. Cross the streams – it would be good.
Back on the original issue: there really has been a lot of change in the European frequencies of alleles that influence skin, hair, and eye color over the past few thousand years. Two population turnovers are a big part of the story, but surely selection is as well. I don’t think we can be sure of the underlying reasons – Vitamin D may not be all of the story. Maybe not even most of the story.