I've been meaning to link to this Gene Expression post on melanin variation in humans, and this AP story has given me an excuse.

Duke University researchers on Sunday reported the first direct evidence that those melanin differences indeed may be a culprit. It turns out that redheads' melanin is more vulnerable to a type of DNA-damaging stress from the sun's ultraviolet rays.

The study was a chemical experiment, where the effects of UV radiation on the pigments were viewed under a microscope. The next step would be to investigate epidemiological evidence to see if the apparent UV instability of pheomelanin translates into greater cancer risk.

"There has been speculation for years that pheomelanin could be a key pathway" in skin cancer formation, said Dr. Martin Weinstock of Brown University, a spokesman for the American Cancer Society. "The thought is that eumelanin does a reasonable job of protecting against UV and pheomelanin might in fact aggravate damage."
While more research is needed, Simon said in an interview that his study reinforces some practical advice: Slather on sunscreen that promises to protect against both UVA and UVB rays.

The distribution of pigmentation phenotypes and alleles is a story that has lately developed a few twists. Like this one:

East Asians tend to exhibit very high frequencies of the Arg163gln allele (as high as 80% among the Dai of Yunnan province in China, decreasing to 40% among the Uighers of Xinjiang and less than 10% in South Asia [ie; India]). The authors suggest that this allele might have been subject to strong directional selection in the recent past. They note that "under neutral expectation, the estimated arrival time of the Arg163Gln allele is older than the age of modern humans...."

Worth a raised eyebrow, but mainly just evidence for selection.

But then, most of the other non-African alleles are generally explained by relaxed selection instead of positive selection. This owes to the fact that none of them have risen to very high frequencies yet -- as one of them would if it were positively selected for very long.

Yet this is a rare evolutionary case in which the many ways that something could be messed up present an advantage. Usually, it is the single way that something could be made better that is the only path for selection. I wonder if we really have a good understanding of what alleles should look like in this unusual case. Maybe alleles like Arg163gln are the rare lucky ones that exceed the neutral expectation, while other alleles have as a sum faced a slight advantage, but haven't yet reached frequencies appreciable enough to battle it out for fixation. I wonder how MC1R allele frequencies compare to those of minor hemoglobinopathies around, say, the Mediterranean?