Balancing selection & pigmentation loci

A few days ago I mooted the possibility that balancing selection may be more common than we had assumed, and that much of the recent evolutionary action in our species' history might be characterized by non-fixed allele frequencies which exhibit the signatures of positive selection because of their shallow time depth. I was interested in the idea for an important reason. Below the fold are are a range of data for two loci implicated in skin color variation in human populations; SLC24A5 and SLC45A2.

The numbers are are derived allele frequencies; and these derived alleles resulted in an average effect of lighter skin within the population in an additive manner. That is, if you substitute one derived allele so that will result in a fairer complexion than for an ancestral homozygote, while another substitution so that one is a derived homozygote results in an even paler complexion. The data are from this paper, Genetic evidence for the convergent evolution of light skin in Europeans and East Asians, which also suggest that both these loci have been subject to recent selection. How recent? Evidence for Recent Positive Selection at the Human AIM1 Locus in a European Population:

The maximum-likelihood estimate of t is 10,965 years. Ninety-five percent confidence intervals...produces the values...1,328 an...39,609 years.

Because scientists are very clever SLC45A2 also goes by the appellations MAPT and AM1. A big 95% window you say? Yes, but at least there's a number. What about for SLC24A5? Well, European skin turned pale only recently, gene suggests:

The data suggest that the selective sweep occurred 5300 to 6000 years ago, but given the imprecision of method, the real date could be as far back as 12,000 years ago....

Again, the certitude is less than awe inspiring. But, both selective sweeps seem relatively recent, though SLC24A5 is likely to be more recent than SLC45A2. What's going on here? Talking to a friend we agreed that it seemed likely that there was pleiotropy at work. No doubt there were many side effects of the substitutions in question and one of these was constraining at lower latitudes. Now I wonder, could it be that SLC45A2 is actually balancing? Even at northern European latitudes it does not quite fix. Unlike LCT its phenotypic effect on skin color does not manifest dominantly, homozygotes are paler than heterozygotes. Also recall that SLC45A2 has likely had more time than SLC24A5 to sweep, and yet it is the latter which has totally replaced the ancestral variant so that SLC24A5 is the second most powerful ancestrally informative marker in the human genome.