And the men of Europe shall be as he was!
Were Scandinavians the original people of Europe? Such a headline is very suggestive of a press release gone wrong. But no, you just need to see what Eske Willerslev actually said to see the source of the headline. It was his lab which published the recent paper in Science, Genomic structure in Europeans dating back at least 36,200 years. Willerslev states:
”Genetically, he is European and is more closely related to current Europeans than any other people in the world. And that means that some of the earliest people in Europe were actually our forefathers,” Willerslev told the science website Videskab.dk. ”He is actually more closely related to Danes, Swedes, Finns and Russians than he is to the French, Spanish and Germans, so one could argue we are more originally European.”
I don’t know if the initial exchange was in English, or translated from Danish. And, journalists have been known to make “mistakes” in their quotations from scientists. But, taken at face value I’d have to say that this quote, and many of the inferences being made from this paper, strike me as “not even wrong.” Using the whole genome of this ancient man the authors generated fascinating results, but I am not quite so sure about the confidence presented in their interpretations. Some sentences jump out at me as anachronistic and incongruous. Consider:
Altogether, these results suggest that contemporary Siberian populations from the Yenisei basin derive part of their gene pool from a Eurasian HG population that shares ancestry with K14, but is more closely related to Scandinavian MHGs than to either MA1 or western European MHGs, indicating gene flow between their ancestors and Scandinavian Europe after K14 but prior to the Mesolithic.
Europe during the last glaciation
It illustrates both the startling results which demand explanation , and the head-scratching assertions which are strewn about like ticking time bombs, due to the incongruity of their implications. Kostenki 14, K14, lived ~37,000 years ago. For most of the period between then and now Scandinavia, and much of Northern Europe was glaciated, and uninhabited. Much of the Yensei basin was also subject to glaciation. Therefore you can eliminate the possibility gene flow from the geographic region of Scandinavian Europe for much of this period because no humans lived in Scandinavian Europe, and likely in much of the Yensei basin as well. This is not a trivial point because the authors make assertions about the nature of migration, or lack thereof, in prehistory, so they need to be very clear and precise on these issues in regards to geographic provenance. The results and statistical genetic methods are complex enough as it is. Quotes like the one above make it very unclear whether the term Scandinavian is a semantic shorthand, likely, since glaciation is evident in a figure in their paper, or literal description, as the lay public and even non-close scientific readers are likely to infer. The major topline finding that is hard to dispute is that nearly 40,000 years ago on the plains of modern Ukraine an individual lived whose genetic makeup exhibited strong affinities to modern Europeans, in particular, Northern Europeans. This is not a trivial result because it adds more evidence to the model that West Eurasians and East Eurasians diverged before 40,000 years ago (earlier statistical genetic models utilizing computational techniques arrived at dates closer to ~20,000 years ago). Recently the genome of the man who lived about ~45,000 years further east in Siberia was analyzed, and found to exhibit genetic affinities with bothEast and West Eurasians. This implies that the differentiation of West and East Eurasians occurred in the interval between ~35 to 50 thousand years ago, aligning well with certain archaeological and paleontological findings. In addition, the tract length of Neandertal ancestry was longer in this individual than in moderns, just as it was in the Siberian genome, as one would expect. The admixture date was inferred to be on the order of ~50,000 years ago, again, in good alignment with expectations. The issues that I have rather are about the nature of the emergence of anatomically modern humans across Eurasia inferred from these results (and further). First, I can’t speak to the archaeology and ancient DNA analysis. I assume they had paleontologists look at the dating and what not, and it checked out. The dates and descriptions look plausible from what I know, but then I don’t know that much. Additionally, the ancient DNA looks good. Willerslev’s team is top notch, probably the only group within spitting distance of Svante Paabo’s in this area. With 2.4x coverage on the whole genome that’s good enough for reasonable genotype calling with ANGSD, and to compare with the HGDP data set and what not (~500,000 markers merged). This doesn’t mean that the archaeology and ancient DNA quality have no issues, but they aren’t obvious.
Citation: Seguin-Orlando, Andaine, et al. “Genomic structure in Europeans dating back at least 36,200 years.” Science (2014): aaa0114.
To the right you see an edited version of a figure from the paper. The barplot is something you’ll recognize as an admixture analysis. You can see that the European hunter-gatherer frequencies from 5,000 to 10,000 years before the present exhibit a very strong modal cluster affiliation, while the Neolithic farmers (to the right) are mixed. To the left you see the two ancient samples of K14 and MA1 (Mal’ta). They are hard to tease apart, though you see that K14 has a lot of the components altogether, like many modern Europeans. The map with the heated circles show genetic affinities from the f3 test, which basically takes a phylogenetic tree, with an outgroup, Mbuti pygmies, and the K14 individual, and another population, X. The outcome of the statistic can indicate the affinity and closeness between K14 and X (basically, the two populations posited to be a clade). The authors use several of these sorts of statistics, but the basic idea is the same, where allele frequencies across topologies should reflect shared (or lack thereof) drift history conditional on the topology being correct or not. In some cases the topology is not totally correct because of gene flow across the clades, so you infer admixture. It does seem rather evident that the K14 individual shares a lot of drift with Northern Europeans. Or more specifically Northern Europeans descend in large part from a population which also contributed to K14. The admixture plot confirms this.
Citation: Seguin-Orlando, Andaine, et al. “Genomic structure in Europeans dating back at least 36,200 years.” Science (2014): aaa0114.
All good so far. My main results qualm though can be found in the supplemental (which is free to anyone without access to the paper itself). In Science Willerslev tells Anne Gibbons that “What is surprising is this guy represents one of the earliest Europeans, but at the same time he basically contains all the genetic components that you find in contemporary Europeans—at 37,000 years ago.” The admixture analysis sort of confirms this, though admixture without a lot of scaffold in terms of what we know can be kind of confusing and like reading tea leaves. That’s why PCA is often useful since it summarizes the variation in a more straightforward manner, plotting out the largest independent components of variance within the data. Take a look at where K14 shakes out. I’ve placed a red pointer toward the K14 sample. In the first plot with all non-Africans K14 is smack the middle of the Central Asian cluster, clearly shifted toward East Asians. All the Europeans aside from Uralic and Turkic populations from Russia are a tight cluster, off to the right of the plot. In the second plot it is constrained more straightforwardly to a sample of Europeans and Middle Eastern populations. Previous PC1 had separated East Asians and Europeans, while PC2 had separeated Oceanieans and Amerindians. Now PC1 seems to separate ancient European hunter-gatherers from Middle Easterners and PC2 Early European Farmers (EFF) from Ma’lta and the Turkic groups with residual East Asian ancestral. Since K14 has the “Ancestral North Eurasian” affinity, it is shifted down toward Mal’ta, but since it is mixed it is not particularly close to the European populations. I point this out to suggest that summaries of the form that “someone just like modern Europeans existed 37,000 thousand years ago” are probably not helpful or illuminating, but that’s what we’re seeing in the press, and not just from the press itself. K14 was an ancient human being. Attempting to understand him as a combination of modern genetic variation is going to have shortcomings. This is obvious. Using “ancestral allele frequencies” derived from model-based admixture analysis to figure out his affinities is useful, but has limits, because those allele frequencies are informative of modern populations. Similarly, PCA is projecting him upon modern population genetic variation. Useful, but again, one must be careful. What’s the big takeway? Let me quote the last paragraph of the Science paper:
Our results further suggest that the early stages of the western Eurasian lineage were already complex (see also Fig. 2). Besides its core affinities with subsequent European groups, K14 also shares alleles with European Neolithic farmers and contemporary people from the Middle East/Caucasus, which are not found in MA1 and western European MHGs, indicating genetic exchange between K14 and a Basal Eurasian Lineage (which eventually contributed to Neolithic groups) after the ancestors of MA1 and subsequent European MHGs had diverged. This implies that early AMH populations became structured early in their history, but already in the UP contained the major genetic components found in Europeans today. As such our findings show the existence of a meta-population structure in Europe from the Upper Paleolithic onwards, remnants of which are still found today, despite migrations to and from Europe since the UP. The early UP contribution is greater among northern than southern Europeans, in agreement with the southeast to west and north gene flow cline resulting from the expansion of Neolithic famers 9-6 ka cal BP (20, 45). However, descendants of the early UP population represented by K14 likely also contributed genes to western Siberian groups living around the mouth of the Yenisei River. Therefore, our findings support the view that these Uralic-speaking populations represent an ancient admixture between European and East Asian lineages. The recently proposed Holocene gene flow from East Asians into northern Europeans (21) can, in our view, be equally well explained by population structure of the hunter-gatherer meta-population within Europe. As such our results paint an increasingly complex picture of colonization history of Europe from the UP to today. Instead of inferring a few discrete migration events from Asia into Europe, we now see evidence that humans in Western Eurasia formed a large meta-population with gene flow in multiple directions occurring repeatedly and perhaps continuously.
Click to enlarge
The authors also present a modified schematic of Lazaridis et al. There are two aspects of the conclusion: 1) Many of the assertions are totally uncontroversial (e.g., “it’s complicated”). 2) Many of them seem to be challenging the model posited by publications coming out of David Reich’s lab, where they partner with Svante Paabo’s groups for the ancient DNA work. Let me quote Willerslev and Reich from the Gibbon’s piece in Science:
“There was a really large met-population that probably stretched all the way from the Middle East into Europe and into Eurasia,” Willerslev says. These people interbred at the edges of their separate populations, keeping the entire complex network interconnected—and so giving the ancient Kostenki man genes from three different groups. “In principle, you just have sex with your neighbor and they have it with their next neighbor—you don’t need to have these armies of people moving around to spread the genes.” [Willerslev] … Other researchers say that this new genome is important because “it is the first paper to document some degree of continuity among the first people to get to Europe and the people living there today,” says population geneticist David Reich of Harvard University, one of the authors on the triple migration model. It also is “a striking finding that the Kostenki 14 genome already has the three major European components present that we detect in modern Europeans,” says Johannes Krause of the University of Tübingen in Germany. But even if the man from Kostenki in Russia had all these elements 36,000 years ago, that doesn’t mean that other Europeans did, Reich says. His team’s DNA data and models suggest that Europeans in the west and north did not pick up DNA from the steppes until much later…. [Reich]
I’ve read the Seguin-Orlando et al. paper (and supplements) several times to try and be fair. What I don’t understand is why they can’t acknowledge the possibility that K14 did not leave modern descendants, and was part of an early population which did not end up flourishing. That is consistent with all their results after all, and, consistent with ancient DNA which seems to show a lot less admixture in Mesolithic groups than K14. The fact that Willerslev talks about meta-populations makes it even more confusing, since one of the aspects of meta-population dynamics is the likelihood of repeated population extinctions and re-colonizations. This seems entirely plausible across the European and Northern Eurasian plain during the last Ice Age, as humans retrenched and expanded multiple times. It’s been a while so I checked Wikipedia, and here’s a representative sentence reflecting what I recall learning: “Kritzer & Sale have argued against strict application of the metapopulation definitional criteria that extinction risks to local populations must be non-negligible.” So there’s an argument about whether extinctions are necessary or not. But it shows how critical they’ve been historically to model metapopulation (I think it makes the math easier). As for the idea that gene flows occurred through diffusion, obviously there’s a lot of that. But the punctuated turnover of mtDNA lineages in ancient DNA transects and the ability to infer admixture events from pulse fusion events seems to suggest a great deal of ancient demography could not be modeled in such a fashion. This paper focuses on Northern Eurasia, but in South Asia, Southeast Asia, and Africa, we see clear instances where genetically distinct populations fused rapidly due to demographic expansion enabled by cultural change. Perhaps Northern Eurasia is different, but cases in other parts of the world should scaffold our expectations. Overall I’m left somewhat more confused and interested. Addendum: One thing I want to emphasize. Willersleve seems to be implying that all the variation of West/North Eurasians in its basic algebraic constituents was present about ~10,000 years after the differentiation of non-Africans. Is it really plausible that the last 35,000 years, a time frame 3.5 times as long, is a coda to that? Perhaps. But color me a little skeptical. I suspect that we’ll get more clarity when we stop thinking of prehistoric populations simply as repositories of the history of extant populations. To see what I’m getting at, there are ~10,000 years separating K14 from East Asian populations alive at that time. There are ~35,000 years separating modern West/North Eurasians and East Asians from their putative ancestry populations. Even if you double the value for K14 vs. Paleolithic East Asians by two because there are two paths of drift, that’s still 20,000 years. The peculiarity of these ancient remains is always clear when you visualize them on TreeMix; they’re often very long branches awkwardly slotted into contemporary trees.