Recent selection among non-Africans at gene related to circadian rhythms

The gene PER2 is involved in the regulation of the circadian clock.
What are the patterns of variation in this gene across worldwide human populations?
They examine this question by sequencing 20 people, then genotyping 499 people from around the world at five polymorphic sites. They then looked at haplotype diversity, linkage disequilibrium, Fst and other measures of diversity.
They look for a correlation between latitude and measures of genetic distance, and examine differences in genetic diversity between groups with similar latitude...and overall, find no evidence of a relationship between genetic variation at this locus and latitute.
They do find high levels of differentiation between populations and they find a low degree of sequence variation in the associated haplotypes, suggesting recent selection. For the haplogroup with the lowest sequence diversity they fnd a TMRCA of 8.7 +- 8.7 kya.
There's some discussion at the end regarding variation in this gene being associated with a person's preference for morningtime or eveningtime... they didn't look at these specific variants. It remains to be seen what explains the patterns of variation and recent selection that they find.

Genetic diversity patterns at the human clock gene period 2 are suggestive of population-specific positive selection
Fulvio Cruciani, Beniamino Trombetta, Damian Labuda, David Modiano, Antonio Torroni, Rodolfo Costa and Rosaria Scozzari
European Journal of Human Genetics Advance online pub. 25 June 2008

Abstract: Period 2 (PER2) is a key component of the mammalian circadian clock machinery. In humans, genetic variation of clock genes or chronic disturbance of circadian rhythmicity has been implied in the onset of several phenotypes, ranging from periodic insomnias to advanced or delayed sleep phases, to more severe disorders. Peculiar geographic diversity patterns in circadian genes might represent an adaptive response to different light/dark cycles or environmental changes to which different human populations are exposed. To investigate the degree and nature of PER2 gene variation in human populations of different geographic origin, and its possible correlation with different latitudes, we sequenced a 7.7 kb portion of the gene in 20 individuals worldwide. In total, 25 variable sites were identified. The geographic distribution of haplotypes defined by five polymorphic sites was analyzed in 499 individuals from 11 populations from four continents. No evidence for latitude-driven selective effects on PER2 genetic variability was found. However, a high and significant difference in the geographic distribution of PER2 polymorphisms was observed between Africans and non-Africans, suggesting a history of geographically restricted natural selection at this locus. In support of this notion, we found several signals for selection in the sequences. The putative selected haplotype showed a recent coalescent age (8.7 Kyr), and an unusually high frequency in non-African populations. Overall, these findings indicate that a human clock-relevant gene, PER2, might have been influenced by positive selection, and offer preliminary insights into the evolution of this functional class of genes.