A researcher who studies melanin hits a lot of data nuggets in one post. NuSapiens offers some related speculations (sort of). Also, you might be interested that Heather Norton has a paper in press, Worldwide polymorphism at the MC1R locus and normal pigmentation variation in humans. Of course, I have no access to Peptides, though Heather gave us the gist I think if not the details....

Update: If you are curious, here is an article that attests to some polymorphism among Sub-Saharan Africans on the MC1R locus (one of the three nonsynonomous mutations was found in a Khoisan individual). The standard assumption is that MC1R has been under strong functional constraint amongst dark skinned peoples, and far less (or least different selective forces have become important) among light skinned peoples (see here for an extensive review article). Henry and Greg have suggested that some variants of MC1R that result in reduced eumelanin production could be the result of introgression from non-African hominids ("archaic" H. sapiens, like Neandertal), though here you see that there is at least some variation for selection to work upon even in the African genetic background. Some older hair color related posts here and here. For more about MC1R go here.