A new paper in PNAS, Direct evidence for positive selection of skin, hair, and eye pigmentation in Europeans during the last 5,000 years, uses ancient DNA to examine the possibility of very recent natural selection in Europeans. In particular, it focuses on eastern Europeans, and roughly a region coterminous with Ukraine ~6000 to ~4000 years ago. The sample seems somewhat biased toward the low end of the age range if you look the supplemental tables. In the paper itself (which is open access) I don’t see a map to get a sense of the distribution of the sites from which the DNA was extracted. So I took the supplemental table and used the latitude and longitude information, as well as the samples from each site, and produced a density map with a bubble plot overlain upon it with specific locations (size of bubble proportional to number of samples at site). Like the earlier ancient DNA from a few European hunter-gatherers one must keep in mind the limitations of the scope of sampling so few to infer about so many. Though the number here is far larger (N >20 or >40 depending on the SNP), the set of markers examined was much smaller, a few pigmentation loci and mtDNA. Nevertheless this is not a trivial geographic example, nor is the time frame, from the Early Eneolithic down to the Bronze Age.
The clearest illustration of the topline result is found in the supplements (I prefer figures to tables obviously). What you see here is that there is a large difference in allele frequencies between ancient samples and modern ones from the equivalent geographic region at specific markers diagnostic for variation in pigmentation in modern Europeans. HERC2 is well known for being one of the two loci which span a long haplotype strongly correlated with blue eyes in Europeans. SLC452 and TRY are part of the standard suite of pigmentation genes which show up as variable across Eurasia. I am confused as to why they did not focus on SLC24A5, a locus which is nearly totally fixed in modern Europeans for the A allele, but may not have been so in hunter-gatherers. But in any case the result is rather clear: the ancient populations sampled here are statistically differentiated from modern populations in the same region, and, seem to have been darkly complected in comparison. The natural inference then is that powerful sweeps of natural selection increased the allele frequencies of lightening alleles in Europeans within the last ~4,000-6,000 years. This is not a crazy proposition; tests for recent natural selection in Europeans are often enriched around pigmentation loci, which are genomically atypical (long homogeneous blocks are common). What this study does is intersect inferences from modern variation with the distribution of variants in an ancient population presumed to be ancestral.
A model like is made more plausible by the fact that many of these individuals were of the Yamna culture, Kurgans. The thesis forwarded by some scholars is that it is these Kurgans, a patriarchal nomadic society, who brought Indo-European languages to central and western Europe ~5,000 years ago (their eastern cousins becoming Tocharians and Indo-Iranians, their southern ones Hittites and perhaps Armenians). Probably the best recent outline of this thesis is by David Anthony in The Horse, the Wheel, and Language: How Bronze-Age Riders from the Eurasian Steppes Shaped the Modern World. I found it so engrossing that I finished it in one sitting in 2008. If these data are correct the Kurgans did not look like blonde Aryan Übermensch, rather, they became that (though to be fair, in this case we are talking about them becoming Slavs, who the Nazis labelled Untermensch). But one of the general assumptions about Kurgans is that they were groups of mobiles males. In that case one wouldn’t be surprised if their mtDNA tended to reflect subject peoples, while the whole genome was more mixed and cosmopolitan, reflecting their migrations.
So the crux then is whether to trust this mtDNA evidence as representative of the whole genome. If I simply had the mtDNA, along with the information about provenance in terms of time and place, I’d probably accept the argument for continuity. But the phenotypic markers are so different, either there’s been population replacement, or, we’ve had a lot of in situ selection. Replacement seems like the more boring hypothesis, especially in light of the fact that many of the sites sampled were not in classically Slav zones of habitation, but were occupied by Iranian or Uralic peoples, or more recently Turks. Though the researchers are using contemporary East Slavs to compare to the ancient samples, across many of these sites Slavs only become dominant in the area with the rollback of the Ottomans in the 18th century.
Ultimately I’m very unsure that the assumption of genetic continuity in this case will hold, so let’s simply take that as a given for now. Then what? You have lots of selection. The question naturally moves to why. What drove the selection? In the discussion the authors the go over many of the hypotheses rather thoroughly. Roughly they fall into two classes, the ecological/environmental and the social/sexual. The former generally has do with a combination of a switch to agriculture and the need to synthesize vitamin D due to the shift away from fish in the far north. The latter focuses on sexual selection, and favoring particular markers due to heightened paternity certainty. In particular the sexual selection hypothesis would seem to be able to explain the rise of HERC2, which is associated with light eyes, as that may be a favored trait. The immediate rejoinder is provided in the text: many of the pigmentation loci have pleiotropic effects. In other words, they tune overall pigmentation, skin, hair, and eyes, though perhaps to different extents. So if the selection was environmental due to skin it would not be totally surprising if hair and eyes changed as a side effect. Of course, as suggested in the comments here one need not posit that there was one singular selection event, as opposed to a sequential composite. Perhaps it was both environmental and sexual selection?
This again is another area where I’ll throw my hands up the air. If selection is the answer, and not population replacement, then it’s very strong. It seems that these loci were subject to sweeps in the same range of power as that around LCT, for lactase persistence, the Tibetan high altitude adaptations, as well as the various malaria resistance alleles (which have different selective dynamics, some of them balancing). One can actually still detect differential fitness at high altitudes based on phenotype, and the same with malaria, at least before modern medicine. The problem I have is that I’m just not aware of studies on the extent of differential fitness in human populations due to sexual selection. In theory sexual selection is very powerful, especially in contexts of hyper-polygyny, but to have it be realized in humans would require very particular social structures. The environmental selection arguments by their nature tend to be simpler, and therefore more attractive. But we’ve reached a point where there’s a lot of confusing stuff coming out of ancient DNA, and we need to go back to first principles, and reexamine everything. This includes sexual selection, as more than simply a deus ex machina to throw out there when we don’t have a better model on hand. That necessitates a serious examination of patterns of variance in reproductive output by phenotype, and plugging these back into models of selective sweeps.
Note: Yulia Tymoshenko has very dark eyes. So I assume she’s not a natural blonde.