Dienekes points me to this piece in The American Journal of Physical Anthropology which concludes that there isn’t greater sexual dimorphism in skin color in areas where social/sexual selection could presumably operate because of relaxation of natural selection. Remember that sexual dimorphism tends to evolve very slowly (you need to have sex-linked or developmentally modified loci since men and women share almost all the same genes [except for the Y]), perhaps an order of magnitude more slowly than standard phenotypic evolution given the same selection pressure.
