A new paper in Science has just been published which in its broad outlines has been described in conference presentations. When examining the autosomal genetic variation of three individuals of the hunter-gatherer Pitted Ware Culture (PWC), and one of the agriculturalist Funnel Beaker Culture (TRB), the authors found that the two groups were sharply differentiated. The number of SNPs was on the order of 10,000 or so if I read the methods correctly. This is rather thin for studying contemporary within European population differences (~100,000 or more seems to be safe), in particular using hypothesis based clustering algorithms (it seems more manageable for PCA). But the findings are strong enough that I think we shouldn’t discount them. The most fascinating aspect of the results is that while the PWC seem to exhibit affinities with Northern and Northeastern Europeans, the TRB individual seems more similar to extant Southern Europeans!
Others have already commented extensively on the results. Keeping in mind the small sample sizes, limitation of comparisons, and the relatively thin marker set, I think the primary result we can take away from these findings is that old models of pure cultural and demographic diffusion are false. By this, I mean that prior debates which culminated in the early aughts on the “Paleolithic vs. Neolithic” contribution to the ancestry of modern Europeans were fundamentally premised on a demographic diffusion dynamic, whereby genes and ideas exhibited a continuous flow across a flat and featureless landscape. On the contrary, the basic outlines we are seeing here is that the human past exhibited spatial and temporal discontinuity. And why should this surprise us? There is no dialect continuum between Spanish and Chinese across Eurasia. Rather, broad language families are sharply differentiated from each other at zones of contact. Though there are theoretical reasons why the variation in genes should be more clinal, the reality remains that cultural parameters are going to shape the outlines of genetic variation, and those parameters are discontinuous.
Most of the information in the paper is well summarized by figure 1, which I’ve reedited considerably to illustrate the primary finding. Please note the relative affinities of the LBK and PWC individuals. The LBK individual in the PCA is placed near the Basque and Sardinian populations of Southwest Europe, while the PWC individuals are outside of the distribution of the HGDP Northern Europeans. In another PCA they seem to cluster with East Baltic populations, such as the Latvians. This is in line with earlier mtDNA work.
Also observe the pattern in ADMIXTURE: again, a close correspondence with the Sardinian and Basque samples for the LBK. Of particular interest is the relative lack of component yellow (K = 4) in all of the ~5,000 year old samples. The difference in proportion of this is what also distinguishes the French from the French Basque, and the Tuscans from the Sardinians (along with K = 2, the blue component in the case of the latter). I have truncated the Finnish samples, but unfortunately I suspect what we’re seeing here is something similar to the Kalash, a very homogeneous and relatively inbred population throwing up its own component. So I do not believe K = 1, the purple, is particular informative. The most likely model for the ethnogenesis of the Finnic peoples in Northeastern Europe seems to involve an exogenous Siberian element. From reading the supplements I do not see this affinity in the PWC. That may then date the expansion of the Finns as a circum-polar group of ethnicities in Western Eurasia.
In any case, as Dienekes has noted, I suspect that what is missing from the model outlined in the paper are subsequent population movements in Europe after the initial contact between the indigenes of the north and the LBK from the south. This may be why the PWC individuals in some PCA analyses seem to be placed outside of contemporary distributions, just like Otzi the Iceman.
Using a two-way admixture model the authors construct a framework whereby Northwest Europeans are about 50 percent farmer and 50 percent hunter-gatherer; that is, 50 percent LBK and 50 percent PWC. But as Maju cautions we need to be careful about substituting these ancient populations for all ancient populations. Additionally, the two-way admixture model estimates that Sardinians are 95 percent farmer, while Swedes are 40 percent farmer, farmer as understood as the descendants of LBK. I think the problem here, as I have argued before, and Dienekes has also suggested, is that there were almost certainly multiple waves of agriculturalists within Europe. The Sardinians seem to be relatively unadulterated descendants of an early farmer demographic expansion, which has been supplemented across much of Europe. I’ve you with their estimates of Neolithic farmer ancestry:
Image credit: Oskar Karlin