What's up with selection on olfactory genes in humans?

Several genome scans for selection in humans find strong signals around olfactory-related genes, which is counter-intuitive since humans are relatively unreliant on smell compared to other mammals (primates?). Or are they? ... some explanations include:

• P-ter at GNXP has discussed this issue, and points us to this paper in PLoS Biology arguing that our sense of smell may actually be really good, and to another paper that we can actually be very good scent trackers.

• I also remember reading somewhere a long time ago that the odor that humans are most sensitive to is beta-mercaptol (or something like that) which is found in rotting meat.

• There's also a possible sexual selection argument, such that appropriate/compatible mate choice is more important in humans than in chimps and other primates, and we know that mate choice is mediated by olfactory signals.

• It has also been found that these signatures of selection may be recent and may differ between ethnic groups, as seen in this paper on the dynamics of admixture and selection in Puerto Rican, where they also point out that HLA genes are nearby the olfactiory genes, possibly confounding these findings.

This area is pretty ripe with testable hypotheses, using phenotypic and genotypic data, that could inform us about various aspects of human reproductive and feeding ecology.

There's a new paper (see abstract below) on OR511, one of the top fifty human genes showing evidence of positive selection (based on synonymous vs. non-synonymous substitutions). They look at SNPs in this gene from 39 human populations. They confirm a signature of selection, show that it doesn't vary between populations, and that selection on this gene was likely spurred before the dispersal of humans out of Africa.
They don't really give any human-specific explanation for selection on olfactory genes.

Signatures of Selection in the Human Olfactory Receptor OR5I1 Gene
Andrés Moreno-Estrada, Ferran Casals, Anna Ramírez-Soriano, Baldo Oliva, Francesc Calafell, Jaume Bertranpetit, and Elena Bosch,
Molecular Biology and Evolution 2008 25(1):144-154

Abstract: The human olfactory receptor (OR) repertoire is reduced in comparison to other mammals and to other nonhuman primates. Nonetheless, this olfactory decline opens an opportunity for evolutionary innovation and impirovement. In the present study, we focus on an OR gene, OR5I1, which had previously been shown to present an excess of amino acid replacement substitutions between humans and chimpanzees. We analyze the genetic variation in OR5I1 in a large worldwide human panel and find an excess of derived alleles segregating at relatively high frequencies in all populations. Additional evidence for selection includes departures from neutrality in allele frequency spectra tests but no unusually extended haplotype structure. Moreover, molecular structural inference suggests that one of the nonsynonymous polymorphisms defining the presumably adaptive protein form of OR5I1 may alter the functional binding properties of the OR. These results are compatible with positive selection having modeled the pattern of variation found in the OR5I1 gene and with a relatively ancient, mild selective sweep predating the "Out of Africa" expansion of modern humans.