The Unz Review - Mobile
A Collection of Interesting, Important, and Controversial Perspectives Largely Excluded from the American Mainstream Media
 Gene Expression BlogTeasers
Plows of the Gods
Vendela Kirsebom, Norwegian mother, Turkish father
Vendela Kirsebom, Norwegian mother, Turkish father

How Europeans became Europeans is a big question, in large part because Europeans (i.e., “whites”) are still what an ideology in disrepute would term the herrenvolk of the world. But this reality, the truth of which sows discord in any discourse, does not need to negate the fact that the question itself is of interest, and today is eminently answerable. Europe has a long history of archaeology and its climate is mild-to-frigid in a manner which might aid in preservation of subfossils. For decades archaeologists have debated whether the ancestors of modern Europeans were farmers or hunters. It seems quite likely that the real answer is both, and, it’s complicated.

But the Gordian knot of history’s inscrutable veil is now be shredded by Thor’s hammer of Truth. More literally Pontus Skoglund has another paper out in in Science (how many times will I type that?), Genomic Diversity and Admixture Differs for Stone-Age Scandinavian Foragers and Farmers. If you don’t have academic access, the supplements are quite rich.

ancientDNA The panel to the left shows the where and when of the samples. The key is that some of these are farmers and some of these are hunter-gatherers (as inferred by their cultural association). They’ve been examined before with more primitive techniques, but this latest paper ups the ante, even granting genomic coverage as relatively modest by current standards. The authors reiterated that there was a massive genetic difference between the first farmers who arrived in Sweden ~5,000 years ago, and a native hunter-gatherer tradition. The F st was on the order of 0.05. To get a sense of scale the maximum F st for modern Europeans is between Finns (a highly drifted group) and southern Italians (a highly admixed group with significant Near Eastern ancestry), at ~0.01. The F st between modern Europeans and modern East Asians is on the order of ~0.10. In other words, two contemporaneous ancient populations in Sweden which were in near proximity for many generations had a genetic distance on the order of half the distance of Eurasia today. In fact this F st is familiar to me. A few years ago at ASHG an Indian group had a poster which talked about the fact that coexistent South Asian groups (castes, jatis, etc.) within a region might have an F st as incredibly high as…0.05! Obviously I’m not saying that ancient Sweden was characterized by caste, but I am asserting that genetic distinctiveness in Europe on the cusp of full agriculture within a local region probably mirrored modern India, where occupation and community identity are as informative as geography.

treemix So why the huge genetic distance? The figure to the left gives a clue as to why. The farmers from Sweden, represented by Gokhem2, have more “hunter-gatherer” admixture than Otzi the Iceman. Ajvide58 has significant “Ancestral North Eurasian,” which is also found in the New World (Anzick1 is a Clovis individual from Montana). What is of note is that both the European farmers have a component with a red shading that has previously been termed “Basal Eurasian,” in that it forms an outgroup in the non-African phylogenetic tree to populations from West Eurasia, East Eurasia, and Australasia. Because it is such an outgroup it no doubt had a large genetic difference from the others, and a high F st value. Even ~25% contribution, as in the case of Gokhem2, might have been enough to introduce enough diverged drift to generate a great deal of allele frequency difference.

pc Many of the results in this paper were prefigured earlier. For example, it turns out that modern Swedes are closer to the hunter-gatherers than they are to the farmers. The PC plot to the right has northern Europeans at the top and Middle Eastern populations at the bottom. To the right are Finnic groups, with the Sami off the chart (I reedited it). All the ancient hunter-gatherers are outside of the contemporary distribution. One thing to remember is that the ancient individuals were projected upon the variation of modern populations. Therefore I suspect this plot understates how distinctive the ancient groups were. Second, it seems likely that modern Finnic populations are relatively intrusive latecomers to the Nordic scene, as the ancient hunter-gatherers have no particular affinity for them. Rather, if there is one population which resembles the hunter-gatherers from Gotland it is Lithuanians. This is not surprising. I have always held that the last region of Europe to be touched by the farmers from the Near East was the eastern Baltic. It is somewhat ironic that this area, and the Lithuanians precisely, were the last continental Europeans to be Christianized in the 14th century.*

So we’ve established that modern Swedes, and therefore modern Scandinavians, are descended more from the hunter-gatherers than the farmers who brought agriculture to the north. But it is in the functional genome where there’s a twist on the story: the farmers may have looked physically more like modern Swedes than the hunters. That’s because at two SNPs which are fixed (in SLC24A5) or nearly fixed (in SLC45A5) in modern Europeans yield matches to the farmers and not the hunters. These two SNPs are among the strongest quantitative trait loci for pigmentation in Europeans. Without them it seems unlikely that the hunter-gatherers would have been recognized as what we’d term “white.” An immediate objection to this is that the ancient hunter-gatherers had a different genetic architecture for pigmentation, so that their lightening alleles were different. Perhaps, but observe that we’ve already stated that the preponderance of the ancestry of modern Swedes is from these hunter-gatherers, and from what we know the genetic architecture in this population is not particularly surprising. Substituting the ancestral allelic variant at this loci tends to make these individuals darker (that is, through mixed marriage with non-Europeans). One could construct more complex scenarios of gene-gene interactions, but I think at this point we know where the parsimony lay.

With the large genetic distance, as well as the fact that the hunter-gatherers exhibited minimal gene flow from the farmers, and, likely a very salient physical difference, it seems plausible that we have the ingredients for inter-group conflict or at least an uneasy coexistence. Though it seems unlikely that the story of Grendel is an allegory recollecting this far distant time, it might not be far off from the truth in terms of how the farmers and hunter-gatherers interacted. Eventually the two groups congealed, but it took thousands of years.

Finally, let’s get back to the truly exotic fact lodged within these new papers: that there was a group of Basal Eurasians. The Basal Eurasians were further from the hunter-gatherers of Europe than the hunter-gatherers of Europe were from the ancestors of Australians or East Asians. Like the “Ancient North Eurasians” it seems likely that this is a “ghost population,” with no modern exemplars. No doubt as I write this there are papers which are being written or conjectured as to their relationship to populations outside of Europe, as it seems that the Middle East is one area where a deeper probe of possibilities is necessary. But perhaps it is to this group of Basal Eurasians that we owe the innovation of agriculture? Like shadows in our past their cultural impact is strong, but their identity as a distinct people has been eroded away by thousands of years of admixture and the flux and flow of peoples.

Citation: Genomic Diversity and Admixture Differs for Stone-Age Scandinavian Foragers and Farmer, Pontus Skoglund, Helena Malmström, Ayça Omrak, Maanasa Raghavan, Cristina Valdiosera, Torsten Günther, Per Hall, Kristiina Tambets, Jüri Parik, Karl-Göran Sjögren, Jan Apel, Eske Willerslev, Jan Storå, Anders Götherström, and Mattias Jakobsson, Science, DOI:10.1126/science.1253448

* The Sami were Christianized later, but the Sami were not at all part of the European system, as the Lithuanians actually were. The Sami are analogous to various indigenous groups, or Jews. In Europe, but not of it.

• Category: Science • Tags: Ancient DNA 
Email This Page to Someone

 Remember My Information

36 Comments to "Plows of the Gods"
Commenters to FollowEndorsed Only
[Filtered by Reply Thread]
  1. The one thing which leaps out to me that you didn’t mention is the absence of any detectable Ancient North Eurasian admixture in the Swedish farmer. This seems to say that the Western Hunter Gatherer component of their genetic background was not something picked up relatively recently in Scandinavia, but something they already had upon settlement. Judging by the proportions from the Iceman, around half of it was already within Early European farmers when they were in the Mediterranean, and the other half was picked up somewhere else along the way (Northwestern France/British isles perhaps).

    I’m still not convinced that the “Basal Eurasian” population is the end of the story. The Middle East does seem like a plausible place to look for it, it’s true. But I wonder if it could actually be a compound itself, where a second out-of-Africa migration mixed with non-Eurasians, rather than some ancient branch of the out-of-Africa migration which just stuck around the Near East unchanged for tens of thousands of years.

    I’m also considering how the Iceman had a higher Neandertal percentage than any modern humans. We obviously need to check this with more data points, but I do wonder if the distinctiveness of Basal Eurasians was caused by higher Neandertal percentages, as this would certainly make them genetically seem an outgroup compared to other non-Eurasians. Of course, this begs the question of why modern-day Near Easterners don’t seem to show elevated levels of Neandertal, but admixture from elsewhere in the world (particularly Africa) could explain much of this.

    Read More
    These buttons register your public Agreement, Disagreement, Troll, or LOL with the selected comment. They are only available to recent, frequent commenters who have saved their Name+Email using the 'Remember My Information' checkbox, and may also only be used once per hour.
    Sharing Comment via Twitter
    More... This Commenter Display All Comments
  2. As a Finn I admit to having some biases on this, but I have to comment on the idea that there is “no particular affinity to Finnic groups among the hunter-gatherers”. :)

    There in fact is just as much as in Swedes or Balto-Slavs, in the PCA (though as you said the fact that ancients are projections does make it harder to interpret it) in the dimension with most variation ( PC1) it’s the Finns who are closest to the level of hunter-gatherer samples.

    Similarly, in the shared drift graphs LSFIN (East Finnish) group and Estonians show a level of affinity to Ajv58 that is not different from that of Lithuanians or the Swedish groups closest to Ajv58. The Helsinki Finnish sample is as close to Ajv58 as some Swedish groups like Gotlanders, or Belorussians. All these groups are of course closer to Ajv58 in shared drift than Ukrainians or Northern Russians.

    Y-DNA N1c1 might well be recent in the Eastern Baltic or Scandinavian region, but autosomally only Saami show reduced affinity to ancient hunter-gatherers among the current Finnic peoples of the area.

    Read More
  3. Is it really “ironic” that the eastern Baltics would be the last Europeans to be Christianized?

    Read More
  4. “But perhaps it is to this group of Basal Eurasians that we owe the innovation of agriculture? ”

    Difficulty: wouldn’t the first population to hit upon agriculture be the least likely to become extinct?…

    Of course they may just have intermixed with the real first farmers of the middle-east, with their genes surviving only in the admixed populations.

    Read More
  5. Judging by the proportions from the Iceman, around half of it was already within Early European farmers when they were in the Mediterranean, and the other half was picked up somewhere else along the way (Northwestern France/British isles perhaps).

    probably germany. LBK & all.

    But I wonder if it could actually be a compound itself, where a second out-of-Africa migration mixed with non-Eurasians, rather than some ancient branch of the out-of-Africa migration which just stuck around the Near East unchanged for tens of thousands of years.

    you mean mixed with Eurasians, right? i think this isn’t a crazy idea at all.

    Horb, your point is generally right. i basically meant the sami are not some ancient substrate. finnic peoples do have some recent east asian admixture, which is elevated in the sami. suomi seem to be a lot like east baltic people with more of the east asian at low amounts…or east baltic are like finns without (estonians in the middle). scandinavians tend to have more of the ‘middle eastern farmer’ at low levels….

    Is it really “ironic” that the eastern Baltics would be the last Europeans to be Christianized?

    meant that they were the last to adopt a middle eastern innovation both times.

    Of course they may just have intermixed with the real first farmers of the middle-east, with their genes surviving only in the admixed populations.

    probably. not sure ‘basal eurasian’ will last a real construct. there’s something really weird going on here….

    Read More
  6. Yeah, I meant to say Eurasian, or non-African.

    A German affinity would make geographic sense. My prior understanding was that the First Farmers tended to hug the coastline in much of Western Europe initially, leapfrogging large distances and only penetrating the “primeval” forest where the remnant hunter-gatherers were more slowly. But the LBK was the nearest neolithic cultural complex to Sweden, and as likely an origin for the first farmers of this region as anywhere else.

    Read More
  7. Regarding Saami, in my opinion (that needs ancient autosomal DNA to support it) they did not move to Scandinavia wholesale, but are mostly of typical hunter-gatherer ancestry and absorbed a considerable siberian-like population in the Kola Peninsula or thereabouts during the Bronze Age. Remains from Bolshoy Oleni Ostrov site in the peninsula show mainly East Eurasian mtDNA which can disappear through drift so less visible in modern Saami, if they are autosomally Siberian my idea should get some backing.

    Or maybe it happened as far back as in the Mesolithic, the 7500 years old Uznyi Oleni Ostrov site in Northwestern Russia has mesolithic mtDNA C too, but no autosomals so nothing certain. In any case the Saami are indeed outliers among surviving circum-Baltic populations when comparing them to these peninsular Scandinavian hunter-gatherers.

    Read More
  8. “* The Sami were Christianized later,”

    Probably earlier. Much of the Christian vocabulary in Saami and Finnish is Slavic (ie Greek) and the archaeologists have dug up a whole lot of Orthodox stuff. Before you accept any Scandinavian claim of having Christianized anyone you need to remember that Orthodox faith did not count as Christianity for these claims. Forcibly converting Orthodox Finnic groups to Lutheranism and then declaring that Sweden had Christianized another pagan group of Finn witches was standard practice until Sweden started failing to conquer Orthodox territory in the 18th century.

    This also mattered politically since if Saamis and Finns had been Orthodox first that would mean Russian states would have a claim to them. We don’t know the extent of Christianization before the Swedish conquests but we do know that Finland and Lapland were at least partially Christian before them and that Christianity first came through the Kievan Rus, not Scandinavia.

    The vast majority of claims that people X were Christianized at date Y are politically motivated medieval propaganda. Who brought Christianity to a people and when used to be the central question for who has the claim to rule over that people so this stuff is full of myths and forgeries. The claim that Lithuanians were the last pagan people is only true in the sense that they were the last place where pagan rulers held territories large enough to be worth a whole Teutonic crusade or two or three.

    Read More
  9. Could the Basal Eurasian element come from admixture with a population of archaic humans?

    Read More
  10. Could the Basal Eurasian element come from admixture with a population of archaic humans?

    yes. but what does that mean then? (as in, the term ‘archaic’)

    Read More
  11. The Basal Eurasian element, whatever it is, is quite interesting. It does seem to be associated to the area in and around the Fertile Crescent and farmers from there apparently took agriculture to Europe. However the Basal Eurasian element is virtually absent in the Indian Subcontinent. This would imply that agriculture in the Subcontinent was not brought there by migrants from the Fertile Crescent.

    Read More
  12. “But the LBK was the nearest neolithic cultural complex to Sweden, and as likely an origin for the first farmers of this region as anywhere else.”

    Megalithism was adjacent

    LBK was much further south

    Read More
  13. “Substituting the ancestral allelic variant at this loci tends to make these individuals darker (that is, through mixed marriage with non-Europeans).”

    So if there was de-pigmentation associated with MC1R and/or IRF4 that is disguised by the fixation of derived SLC genes then mixed children who ended up with the MC1R from one parent and the ancestral SLC from the other ought to be lighter (on average) than those who got the ancestral SLC genes but without MC1R?

    Read More
  14. Interestingly, it seems ancient European hunter gatherers were in the same “big” clade as ENA/East Eurasians. In fact, they are technically closer to the indigenous Australian genome than they are to ancient European farmers (in terms of lineage), which is pretty amazing.

    From the supplements:
    “In other words, early European hunter-gatherers are consistent with forming a clade with East Asian groups and the Australian groups to the exclusion of a population lineage that contributed genes to the early farmers.” (p. 29)

    I guess what really differentiates living West Eurasians from ENA is “Basal Eurasian” admixture. Basically, “Basal Eurasian” admixture is the main connective tissue between living West Eurasian populations from Ireland to India, and the main differentiating factor between Eurasia west and east of the Ural+Himalaya. Without “Basal Eurasian” admixture, it would be extremely difficult to separate living West Eurasians from ENA, just like how it’s so tough to separate ENA admixture from ANE or WHG/SHG admixture in living populations. ENA, ANE, and WHG/SHG are extremely close lineages within a huge “Derived Eurasian” clade. I’m not very good with precision, nor with ontologically robust description, but I think the following would describe the current state of knowledge and theory:

    ENA, ANE, and WHG/SHG are all purely “Derived Eurasians”, and constitute a large clade versus “Basal Eurasians”. ANE and WHG/SHG are western lineages within the “Derived Eurasian” clade. ENA is the eastern lineage within the “Derived Eurasian” clade. Native Americans are mixtures between “western Derived Eurasians”, and “eastern Derived Eurasians”. But living West Eurasian populations are complex mixtures between “Derived” and “Basal” Eurasians, but predominantly “Derived Eurasian”, and mostly “western Derived Eurasian”. Taking a wide ranging view at all of contemporary West Eurasia, we find minor “eastern Derived Eurasian” admixture in South Central Asia, and to a lesser extent, the West Asian highlands+Caucasus. We find significant Sub-Saharan African admixture in North Africa, and minor Sub-Saharan African admixture in Southwest Asia. We find minor “eastern Derived Eurasian admixture” in Eastern/Northeastern Europe. Finally, we find very minor Sub-Saharan African admixture in Southern Europe.

    This also explains why Avjide58 has a greater affinity to Anzick-1 (who is approximately 51% ENA, 49% ANE) than to MA-1. It’s because ENA, WHG/SHG, and ANE are just so closely related.

    Naturally, I wanted to see how this relates to the HGDP Pashtun samples. The results are pretty much as expected. The HGDP Pashtuns share the most drift with Avjide58 out of all South Central Asians, and share more drift with this ancient individual than Iranians do. In fact, the Pashtuns aren’t too far behind Northern Caucasians. This is despite the fact that they probably don’t share any ancestry with this ancient individual (no evidence of any European hunter gatherer ancestry in Afghanistan and northwestern Pakistan). I guess Avjide58′s 15% ANE admixture must play a role in this, and heightens the affinity, since the HGDP Pashtuns are very rich in ANE admixture.

    But there is one odd detail. Namely, the shared drift between the HGDP Pashtuns, and Gökhem2. In this case, Pashtuns are again the closest population in all of South Central Asia to this ancient individual, as close as Iranians. They are even closer than populations from Balochistan, which all have very strong Southwest Asian admixture. Pashtuns are also ahead of some populations in the Levant, ahead of Saudis, and far ahead of Egyptians+Yemenis. I’m not really sure what this could possibly imply. Tentatively, I think it must be related to the fact that Gökhem2 is around 77.2% “western Derived Eurasian”, so this makes him closer to the Pashtuns.


    I wouldn’t be so hasty when it comes to “Basal Eurasian” admixture. Based on some analyses, the HGDP Pashtuns seem to be around 21.3% “Basal Eurasian”. Of course, this isn’t a precise estimate, but it’s likely to be in the ballpark. Obviously, Pashtuns aren’t South Asians, but they do have a very deep and ancient relationship to the sub-continent. Although I don’t have numbers for actual South Asian populations, but around 10%-5% “Basal Eurasian” doesn’t feel too far off the mark? We’ll find out soon enough.

    Agriculture in South Asia is a very complicated matter. There is definitely an indigenous component at play with the domestication of some crops. And rice ultimately originates from East Asia.

    Read More
  15. I’d say the gods came later, during the Copper Age from somewhere in the east, and they had horses and wagons, and were related to MA1.


    What makes you think there’s none of the so called Basal Eurasian admixture in South Asia? There has to be, because there’s plenty of Near Eastern admixture in South Asia. You can’t have that without the Basal Eurasian.

    Note that South Asians have plenty of the so called West Asian/Caucasus stuff in ADMIXTURE runs. This looks like a signal of EEF (including supposed Basal Eurasian admixture) mixing with MA1′s relatives (aka. ANE).

    But in any case, I have a feeling that the Basal Eurasian will just turn out to be ancient Northeast African admixture when all is said and done.


    Ajvide58 doesn’t show greater genetic affinity to Anzick-1 than to MA1.

    Also, ANE and WHG might well be part of the same broad Eurasian clade as ENA, but they’re quite distinct. The reason it’s difficult to split them apart in some analyses is because of recent genetic drift affecting modern samples.

    This is especially true for South Asians, who readily produce very robust South Asian-specific clusters in ADMIXTURE runs, which are actually complex composites from several different mixture events.

    Read More
  16. Also, ANE and WHG might well be part of the same broad Eurasian clade as ENA, but they’re quite distinct. The reason it’s difficult to split them apart in some analyses is because of recent genetic drift affecting modern samples.

    i suspect LD decay analysis will be applied in the near future. to see if there was a recent admixture event.

    Read More
  17. @ Commentator – Calling ENA and the WHG/ANE clade extremely close is different to call.

    Still, with the caveat that the following method is imperfect:

    In Eurogenes’ (Davidski’s) analysis, if we estimate FST between East Asian and “La Brana” or MA-1 using the summed fractions of La Brana or MA-1′s components, then the FST is around 0.1, while if we apply this to a present day European population, it is around 0.115.

    So assuming present day European is around half La Brana/MA-1 type and half Basal Eurasian type, we’d expect Basal Eurasian-East Asian FST to be around 0.13 in the terms of that analysis.

    It seems from this La Brana’s people were probably closer to East Asians (and the rest of the ENA clade) than present day Europeans are (and quite a bit closer than the Basal Eurasians would have been). However their distance from ENA was likely still within the realms of the present day East-West Eurasian continent scale FST difference that Razib refers to.

    Similarly, if you estimate FST between West African and “La Brana” or MA-1 by the same method is around 0.155-0.16 (slightly lower than the project’s East Asian-West African FST (0.16)), where present day Europeans-West African is 0.145, so Basal Eurasian-West African might be around 0.13.

    That seems to converge on a component that separated from ENA or Derived Eurasian a little later than West Africans, then didn’t admix with either (this may have been a Northeast African population, as others have said), and didn’t drift as much as the ENA (or DEu) branch.

    This isn’t a million miles away from Lazaridis, except that paper placed less drift in the Basal Eurasian branch (which isn’t impossible still and was best for parsimony in their models as the BEu ghost was the minority contributor to the final populations).

    Without “Basal Eurasian” admixture, it would be extremely difficult to separate living West Eurasians from ENA, just like how it’s so tough to separate ENA admixture from ANE or WHG/SHG admixture in living populations

    It’s not tough to separate the admixture in the present day per se, just relies on estimating relatedness to Native Americans net of quantified East Asian / Oceanian admixture (shared drift) and their strong drift. (splitting ANE and WHG is harder here, now).

    If West Eurasians somehow didn’t have Basal Eurasian ancestry, it might have been *easier* to see ANE ancestry in Native Americans, not harder, because we’d literally just get a nice ANE/WHG vs ENA lower level dimension in our PCAs and cluster analyses, assuming similar weights of populations, of only slightly smaller size than the present day West-East Eurasian dimension. Rather than getting some of this scattered between European vs Middle East and European vs East Asia and East Asia vs American higher order dimensions and clusters, making it hard to identify.

    Read More
  18. @commentator
    “I guess what really differentiates living West Eurasians from ENA is “Basal Eurasian” admixture.”

    Sumer and Akkad might be an example of WHG/SHG first farmers being over-run by Afroasiatic peoples. (And possibly the first of many examples of early farmers getting conquered by the herders they themselves created on their borders.) The newly combined EEF farmer population spreading out in various directions might then constitute a common thread over a wide area although I wouldn’t call it basal personally – but some word that sums up a very wide diaspora. So that would mean

    EEF = WHG/SHG first farmers (from the fertile crescent or thereabouts) + Afroasiatic imo

    with the SSA admixture having a dramatic weighting effect on admixture plots.

    nb I also wonder if the SSA population in the north-east has changed a lot since the Bantu expansion and that might be complicating matters.


    “What makes you think there’s none of the so called Basal Eurasian admixture in South Asia? There has to be, because there’s plenty of Near Eastern admixture in South Asia. You can’t have that without the Basal Eurasian.”

    Well if the above point is correct – that the first farmers were a segment of WHG/SHG in the fertile crescent (or thereabouts) who were later conquered by and mixed with an Afroasiatic population out of Arabia – then some of those first farmers may have traveled to India before this event happened.

    Read More
  19. Grey,

    I’m still confused why anyone would claim that South Asians lack the so called Basal Eurasian influence. Is it because of the lack of Y-HG E in India? That’s just a paternal marker.

    In formal mixture tests South Asians can be fitted as Samaritan/Karitiana, Georgian/Dai, etc., which is evidence enough that they’re part Basal Eurasian.

    Also, keep in mind that in the last Reich lab paper on the genetic history of India the best proxies for the Ancient North Indian (ANI) component were Georgians and Abkhazians. That doesn’t mean there was a recent migration from the Caucasus to India, it’s just another way of saying that ANI is mostly a Near Eastern component (WHG-like plus partly Basal Eurasian) but with considerable ANE admixture, which is heavier in the Caucasus than in the Near East.

    Read More
  20. @Davidski,

    Refer to Tables S13 and S14 to see what I mean. By “genetic affinity”, I was not referring to Ajvide58′s cladistic position in relation to the other ancient samples. As you well know, genetic distance is somewhat separate from the question of how samples/populations relate to each other as lineages.

    Also, your point about genetic drift has merit, but it just isn’t exhaustive. There is an irreducible similarity at work. You can use the Dai, Papuan, or Onge to infer ANE/WHG percentages for Europeans. The very fact that this is even possible speaks volumes. The software is freely available from the Harvard guys, and you have the samples, so you can easily verify this for yourself.

    The idea that “Basal Eurasian” is cryptic Northeast African admixture makes sense, and is surely very plausible, but there is the fact that “Basal Eurasian” admixture is supposed to be equidistant to all Eurasians, not closer to “western Derived Eurasians” (since all living West Eurasian populations have some “Basal Eurasian” admixture, the comparison wouldn’t be fair). Also, the fact that Skoglund et al. and Lazaridis et al. arrive at the same conceptual abstraction via different routes does boost one’s confidence. And the idea behind “Basal Eurasian” admixture is slightly older than these papers. Please take a look at this:

    Dienekes stumbled on the same concept using TreeMix on ADMIXTURE components. He finds a migration edge from a “Proto-Eurasian” source into his “Southwest Asian” component. The strength of the migration edge is 18%. In addition, he finds a contribution of around 7% “East African” into his “Southwest Asian/Caucasus/Atlantic-Med” components. His “Proto-Eurasian” group is almost identical in concept to the “Basal Eurasian” construct. Namely, equidistant to all living Eurasians, rather distinct from all living Eurasians and Sub-Saharan Africans, but closer to all living Eurasians than to Sub-Saharan Africans. In short, we have a very nice correspondence.

    But at the end of the day, Near Eastern aDNA is what we need. I wouldn’t be surprised if it really boils down to extremely ancient African ancestry. The 7% “East African” migration edge into the three West Eurasian components is definitely suggestive. But going by what we currently know, “Basal Eurasian” seems like a versatile concept.

    Read More
  21. @Davidski

    “I’m still confused why anyone would claim that South Asians lack the so called Basal Eurasian influence. Is it because of the lack of Y-HG E in India? That’s just a paternal marker.”

    No idea. Your comment made me think what if the first farmers were the segment of WHG who lived in the fertile crescent who spread in a first wave and the “EEF” were a later result of a Sumer-Akkad type situation followed by a second wave of Afroasiatic admixed farmers. It was just me thinking aloud really.

    Read More
  22. Here’s a PCA featuring ANE proxy MA1, along with West Eurasians, North Africans and South Central Asians. The last group is limited to Pathans and one Kalash, to prevent South Asian admixture and recent genetic drift in the region from skewing the results too much.

    It’s obvious that, in this analysis, Northeast Europeans share maximum levels of drift with MA1, which can be seen mostly in PC1, but the vast majority of this is mediated via WHG-like ancestry. So in fact it’s the Lezgins, Pathans and Kalash who harbor the highest levels of direct ANE ancestry.

    The Lezgins are know to have around 29% ANE. We don’t really know yet what the levels of ANE are like in South Central Asia yet, because we need a few different people to tackle that issue, including at academic level. But even just eyeballing this PCA I think it’s fairly clear that the Pathans and Kalash have a couple per cent more ANE than the Lezgins and Chechens.

    So the question is, what depresses the shared drift between the Lezgins, Chechens, Pathans and Kalash and MA1? As per above, anything related to WHG would very likely increase the affinity between these groups and MA1. So it has to be something else.

    Indeed, it’s the Northwest Africans and selected Bedouin who show the least affinity to MA1 on this PCA. Both have Sub-Saharan admixture, but the Lezgins, Chechens, Pathans and Kalash don’t have that, and yet they’re pushed away from MA1, especially in PC2. So as far as I can see they must have something African-like from the Near East.

    We can call this admixture Basal Eurasian, or whatever, but it’s certainly not recent Sub-Saharan ancestry, nor anything closely related to WHG.

    Read More
  23. You’ve got better eyes than I’ve got, mate – I’ve got that blown up as far as it will go, and I’m sitting looking at the screen through a magnifying glass, and I’m damned if I can read it.

    Read More
  24. Why don’t you download the PDF onto your computer? It’s a public file.

    Read More
  25. #@%*&@$$#* Chinese Google.

    It’s OK, I’ve got it downloaded now. Thanks Davidski.

    Read More
  26. @ Davidski

    The Pathans have about 25% ASI ancestry which would push them away from MA1.

    BEA is associated with Y-haplogroup E1b. The lack of E1b in South Asia suggests that ancestry from the Fertile Crescent is less in South Asia than it is in Europe. Even though it is only several hundred miles from the Fertile Crescent to western Pakistan, the intervening deserts and mountains are not conducive to farming.

    Read More
  27. ASI ancestry doesn’t have much of an effect on the Pathans on that plot. I’d say that if they weren’t part ASI they’d be sitting among the two Lezgins closest to MA1, which isn’t far away from where they are now.

    Adding more South Asian samples to the PCA certainly increases the effect of ASI on the positions of the Pathans. But then they’re actually pushed closer to MA1 in PC2. See here…

    By the way, no one knows yet whether E1b was associated with the so called Basal Eurasians. But even if true, it’s hardly an argument for a lack of Basal Eurasian admixture in South Asia.

    This type of ancestry is all over the Near East, and South Asians have Near Eastern ancestry, so how can it be missing from South Asia? It might have been taken there with people mostly belonging to haplogroup G, like Oetzi, or J.

    Read More
  28. Balaji,

    On the contrary, ASI admixture would make the Pashtuns closer to MA1. That’s why more superficial analyses show that the Irula are around 40% ANE, which is obviously completely incorrect, but does tell us something on a phylogenetic level. ENA is substantially closer to ANE than BEA is to ANE.

    The Pashtuns are distant from MA1 on the same dimension as Bedouins, along with Lezgians and Chechens, so “Basal Eurasian” admixture is the most parsimonious explanation.

    Also, this is a very minor point, but 25% ASI for the Pashtuns was probably an overestimate (the Moorjani et al. estimate was actually around 30% ASI). One minor contributing factor was the exclusion of a few samples. Out of 22 HGDP Pashtuns (I’m ignoring the Turkic-Hazara admixed individual, so 22 rather than 23), 15 were used, and that 15 includes 4 outliers that cluster with Punjabis/North Indians. If you use fineSTRUCTURE, 18 HGDP Pashtuns constitute a cohesive population, and many of those samples were excluded, while the 4 Punjabi-like outliers were included. It boils down to what they were trying to accomplish in terms of analysis, but we must always remember this caveat. For what it’s worth, the Pashtun average (using these 18 samples) for the “South Indian” component in “HarappaWorld” is 21%. For some perspective, the “South Indian” component is clearly a composite. If you try Dienekes’ “Zombie” technique, and measure the “South Indian” component in relation to other “HarappaWorld” components, you get 50% “Caucasian”, 42% “Southeast Asian”, and 8% “Papuan”. That makes sense, as most people have predicted that the “South Asian/South Indian” component is basically around 50% West Eurasian. When Zack had an “Onge” component, it was 50% of whatever percentage a population gets for the “South Indian” component, which pretty much confirms this.

    A much greater confounding factor was the relationship between ENA and ANE. Before MA1′s data was analyzed, there was work suggesting that Europeans range between 20% and 40% Native American-related East Eurasian. That wasn’t fully on the mark. Without MA1′s data, most admixture related to his clade registers as ENA. With his data, things become much more clear. Also, he always registers some of the South Asian-specific component in unsupervised ADMIXTURE runs, just like he registers Native American and Northern European-specific components. So, the South Asian/South Indian-specific component we see in ADMIXTURE could actually be more than 50% West Eurasian, if we count ANE as West Eurasian (which isn’t necessarily correct, but it’ll do for our purposes). But that’s beside the point. Short and simple, ASI admixture among Pashtuns is more likely to be around 10%. Specifically, one usually gets a range of around 5%-13%, and some of that ENA is affiliated with Siberia. For comparison, the HGDP Sindhis seem to be around 20% ENA. Specifically, one usually gets a range of around 15%-25% ENA. Little if any of that is going to be related to Siberia.

    Going outside of ADMIXTURE, formal mixture tests are suggestive. The upper bound for Samaritan-related ancestry among the 18 HGDP Pashtuns is 93%. The upper bound is rarely off by much, so around 90% West Eurasian seems reasonable. In an academic context, the “Ancient Admixture in Human History” paper (Patterson et al.) has the upper bound for all 22 HGDP Pashtuns, using Druze. The upper bound for Druze-related ancestry among the 22 HGDP Pashtuns is 92.2%. Please refer to “Table 5 Three Population Test in HGDP”. Again, strong evidence for around 90% West Eurasian ancestry for the HGDP Pashtuns. One can also calculate admixture using ALDER, with only 1 reference population. Using one of the Georgian populations, one gets 91% Georgian-related ancestry for 18 HGDP Pashtuns (for some reason, another Georgian group yields around 35%. Not sure about what accounts for the difference). Another thing of note, haploblock paintings. They usually give people from South Asia around 90% West Eurasian in South Central Asia/northwestern South Asia, and around 60% West Eurasian in South India. In this case, “West Eurasian” is defined by European samples (often northern European samples), so WHG/ANE-related ancestry is still within the purview of “West Eurasian”, and doesn’t get siphoned off into ENA. Do refer to old South Asian “Ancestry Paintings” for 23andMe, try Interpretome, and look at Dr. McDonald’s chromosome paintings for South Asians, assuming only three reference populations (African, East Asian, and European). Just some stuff to remember.

    Again though, even if the Pashtuns are around 25% ASI (which is very unlikely), that wouldn’t pertain to David’s particular PCA plot of MA1 and West Eurasian populations. If “Basal Eurasian” admixture is “real”, there is probably a good amount of it in South Asia.

    Read More
  29. @Commentator

    Thanks for the information on Pathans. They do seem to be pretty heterogeneous, especially compared to Indian castes. This is also evident in PCA plots. I am aware that Moorjani et al. curated their data. Still, I am more inclined to go with their estimate of 30% than with 10%. I would be interested in what others have to say about this.


    Thanks for sharing your PCA plots. To a first approximation, in your plot with West Eurasians, North Afticans and Pathans, the position of a person on the PCA plot in the PC1 direction will be proportional to that individuals ANE+WHG+UHG, where UHG is unknown hunter gatherer. According to the supplementary data provided by Lazaridis et al., the Lezgins have ANE of 0.29 and Near East ancestry of 0.71. We can estimate the UHG+ANE of Lezgins as 0.56*0.71 + 0.29 or 0.69 (assuming that 0.44 of Near East ancestry is BEA). Since the Pathans are close to the Lezgins on the PCA plot, we can estimate that the Pathans too have UHG + ANE of 0.69 which leaves 0.31 for ASI + BEA. If we accept Moorjani’s estimate of 30% ASI, the estimate for BEA is 0. Pathans and Lezgins do have a lot of shared ancestry but it is via UHG+ANE and not BEA.

    Read More
  30. Balaji,

    Your calculations are actually a good argument for around 10% ASI. You have 30% left for BEA+ASI. The preliminary calculations I’ve seen show 21% BEA admixture for the Pashtuns. So, 9% ASI! Something around 10% doesn’t sound too inaccurate (of course, I’m not basing 10% on just the percentages you’ve postulated. Supervised ADMIXTURE experiments, formal mixture tests, ALDER, and haploblock paintings, all point to this. Most of these analyses actually suggest around 7%-9% ASI, but that’s close enough to 10%). Of course, these percentages your’e postulating (44%, 69%, 31%, etc) are just educated guesses. Also, “Ancient Near Easterners” surely had more BEA than Stuttgart, more than 44%. This is because despite being overwhelmingly “Ancient Near Eastern” in ancestry, Stuttgart shows solid evidence of European admixture. This is also mentioned in the paper. The closest living population to “Ancient Near Easterners” are the “Bedouin B”, minus their minor Sub-Saharan African admixture (4%-7%). I think it’ll all get much clearer, once we get aDNA of sufficient antiquity from the Near East.

    Nevertheless, I think we don’t really need to guess/assume numbers in this case. The Pashtuns are distant from MA1 in the same direction as Bedouins, and cluster with Lezgians and Chechens. Is this not evidence for substantial BEA admixture among the Pashtuns?

    But, I’m willing to go with you on this. If I’ve learned anything in the past few months, it’s that nothing is really certain, and one should always be open to different theoretical possibilities. In the spirit of intellectual clarity, I’d just like to pose a few questions. In your understanding, why would BEA be missing from South Asia, and what would this imply for South Asian genetic history? Does this relate to the roots of agriculture in South Asia? Is Y-HG E the only uniparental (on the paternal side) marker for BEA ancestry? What about Y-HG G? For what it’s worth, Pashtuns have a lot of G.

    Read More
  31. Commentator,

    I am well aware that the absence of E1b only suggests but does not conclusively establish the absence of BEA. Paternal and maternal lineages were studied in detail by the early years of the 21th Century and based on these, Kivisild et al. wrote in 2003 in AJHG, “Taken together, these results show that Indian tribal and caste populations derive largely from the same genetic heritage of Pleistocene southern and western Asians and have received limited gene flow from external regions since the Holocene.” A consensus was building around this view which was further reinforced by papers in 2006 by Sengupta et al. in AJHG and Sahoo et al. in PNAS. However after the landmark 2009 paper by Reich et al. on reconstructing ancient Indian population history, many people came to believe that there was indeed substantial gene flow into the Subcontient during the Holocene and specifically due to the migration of farmers from the Fertile Crescent who largely constitute the ANI. This view was somewhat challenged by a paper in 2011 in AJHG by Metspalu et al. The authors found a component they called “k5” which is basically what Dienekes calls “West Asian” and Zack calls “Baloch-Caucasian”. This is the predominant ANI component and is also found in Europe and the Middle East. They found, “We found no regional diversity differences associated with k5 at K = 8. Thus, regardless of where this component was from (the Caucasus, Near East, Indus Valley, or Central Asia), its spread to other regions must have occurred well before our detection limits at 12,500 years.” Now that Lazariditis et al. have discovered BEA, if BEA is absent in the Subcontienent, this will be a vindication of Kivisild et al.’s earlier findings.

    Regarding the ASI proportion in Pathans, I tend to trust the numbers from the Reich Lab. I am not aware of any published challenges to their methods or their numbers. Also looking at PCA plots such as the ones kindly provided to us by Davidski, we see that the Pathans are roughly halfway between West Eurasians and South Indians. That South Indians should be about 60% ASI (just as the Swedes are majority ANE+WHG) seems about right and so 30% ASI for Pathans also looks reasonable.

    Read More
  32. Balaji,

    I most certainly appreciate the history lesson, but your logic is still rather opaque.

    Also, for your viewing pleasure, here is a PCA plot with MA1, West Eurasians, and ENA:

    (three 23andMe genotypes are on this PCA plot. You’ll easily make them out, just by the labels)

    The output is rather interesting. As expected, we have differentiation between West Eurasians and ENA, and West Eurasians constitute an exceedingly long cline, from Southwest Asia/Southwestern Europe to South Asia. For me though, it is interesting that the two poles of West Eurasian variation are Bedouins (I believe “Bedouin B”) and Gujaratis. Basically, non-ANE admixed, “Basal Eurasian” admixture-rich Southwest Asians, and predominantly “derived Eurasian”, ANE admixture-rich South Asians.

    And it’s very interesting that MA1 is clustering in South Central Asia. One can’t make an argument here concerning projection bias, nor can an argument be made concerning a lack of good dimensionality. The South Asian pole of genetic variation is fully fleshed out here, and MA1 is still closest to the HGDP Pakistanis. The South Asian-driven eigenvector (PC2) isn’t tied to ASI, because Tuscans, French, and “Bedouin A” are much more “South Asian” here than the “Bedouin B” and Basques, yet they don’t have any ASI admixture. The most “South Asian” West Eurasians, outside of South Central Asia, are the Chechens and Lezgians, who also don’t have any ASI admixture. In short, ASI is definitely not the primary reason for PC2, and MA1 is within Pakistani variation for that eigenvector. It is suggestive that when comparing Chechens and Lezgians, it is the Lezgians who are more “South Asian”. According to Lazaridis et al., Lezgians have more ANE than Chechens. Based on all of this, I think there is a good foundation here for saying that the usual dimension which splits South Asians from other West Eurasians is actually a measure of ANE affinity, and not a measure of ASI admixture (I’m referring to PC2 of most, if not all, Eurasian PCA plots). Possibly, ancient Near Eastern ancestry, specific only to South Asia, might also be contributing to PC2 (for some populations).

    Like any form of ENA admixture, ASI primarily pushes samples towards the East Asians. The Pashtuns are only slightly more ENA-shifted than Russians, on this PCA plot. And, if we consider the fact that the Pashtuns have much more ANE than the Russians, it becomes perfectly probable to attribute some of that East Asian-shift to ANE admixture. If one uses f4 ratio estimation, the HGDP Pashtuns range (from sample to sample) between 34% and 37% ANE. Using the same formula for the Lezgians, one gets 28.9% ANE. According to the paper, the Lezgians are 28.8% ANE. In other words, this estimate isn’t going to be too far off. Basically, the HGDP Pashtuns are around 35%-40% ANE. These estimates are probably in the ballpark, unless Lazaridis et al. got the Lezgians wrong. We’ll see if TreeMix validates this.

    Concerning ASI and Pashtuns, you’ll eventually see a paper on this issue (dealing with all of South Asia), very soon. Reich/Moorjani et al’s. estimates don’t need to be challenged. The methods were solid. Instead, they need to be revised, in light of ancient genomic data, and they will be.

    *A side note, but you bring up the “Caucasus_Gedrosia” component, also discovered by Metspalu et al. It’s a fairly standard cluster with ADMIXTURE. The product of any good data-set, coupled with lower K. Do take a look at where it peaks. The “Caucasus”, and “Gedrosia”.

    I’m sure you’ll agree that the Caucasus has a substantial helping of BEA. Why not Gedrosia (Makran/Balochistan, and with it, the rest of South Asia)? Again, I’m not sure how this helps the argument for no BEA in South Asia? I suppose we both need to wait for more technically competent hands to work with the data.

    Read More
  33. The West Asian, Caucasus, Balochi etc. clusters from ADMIXTURE are meaningless when it comes to ancient population movements.

    They’re modern composites of ancient ancestral components, and peak in the Caucasus and Hindu Kush because of recent isolation, endogamy, and genetic drift.

    Academia really screwed up as far as the population history of South Asia is concerned. This was mainly the result of dodgy methodology, but also probably political correctness, with many scientists desperately trying to prove that the Indo-Europeans didn’t come from the Eurasian steppe, and that the Aryans didn’t conquer India.

    So we’ve had to put up with nonsense such as that R1a originated in India, when in fact, the earliest split under R1a-M417 was between the Northwestern European R1a-CTS4385 and Eastern European/Central Asian R1a-Z645. Then the next main split was between the Eastern European R1a-Z282 and Central Asian R1a-Z93.

    And all of the R1a in India falls under R1a-Z93, which, based on complete Y-chromosome sequences, isn’t older than the Copper Age, and probably didn’t arrive in India until the Iron Age.

    The past few years have been like living in the Twilight Zone, with scientists from Stanford etc. attempting to argue that up is down and down is up. Totally bizarre.

    Read More
  34. Davidski,

    These are very good points. Hopefully, the lab working on Farmana will succeed in acquiring some South Asian aDNA. It’ll solve (and also raise) so many questions. For what it’s worth, I think the R1a dates are very strong pieces of evidence for what you’re saying.

    I wanted to check the f4 ratio estimation versus estimates obtained via TreeMix. Using a tweaked formula, one gets 34% to 37% ANE, depending on the HGDP Pashtun sample. To check if this isn’t problematic in any way, one can try the same tweaked formula on the Lezgians. One gets 28.9% ANE for the average Lezgian. Almost identical to Lazaridis et al. My friend, an absolutely brilliant individual, has already downloaded TreeMix, so I asked them to try TreeMix on the HGDP Pashtuns and MA1.

    They were having some trouble in unsupervised mode. So, they inputted a known event, ANE admixture into the Karitiana, in the same proportion shown by Lazaridis et al. This allows them to obtain a migration edge from MA1 into the HGDP Pashtuns. They get 0.37668 MA1-related admixture into the HGDP Pashtuns.

    In short, according to TreeMix, the HGDP Pashtuns are around 38% ANE. That’s basically it folks. Unlike other estimates, one avoids a lot of trouble with TreeMix. I think 38% ANE for the HGDP Pashtuns is a very reasonable estimate, and probably the most accurate one I’ve seen. Here is the graph:

    (Intriguingly, there is also an edge from MA1 to Papuans. It’s extremely minor. Only a meager 0.08%. Very minor, but it’s there. Interestingly, my friend also tried the Paniya, and there is no evidence of any ANE admixture for them)

    *Balaji, this might be pertinent to our discussion concerning BEA. In unsupervised mode, my friend was getting a migration edge from Yoruba to Pashtuns, a contribution of 30% Yoruba-related ancestry. Pashtuns have never shown any evidence of Sub-Saharan African ancestry, in any genetic context. And based on Lazaridis et al’s. global PCA plot, the Yoruba are substantially Eurasian-shifted in comparison to African hunters. Dienekes also found evidence of substantial Eurasian admixture for the Yoruba. So, 30% Yoruba-related ancestry for the Pashtuns seems to be pointing to around 30% admixture with an African-like Eurasian population. Does this not sound suspiciously close to 30% BEA for the Pashtuns?

    Read More
  35. According to TreeMix, I’m 41.1% ANE. When David tried his most accurate PCA-based test (ME percentages matched the paper), both myself and the Karitiana were 36% ANE. So, it’s plausible that I’m as ANE-admixed as the Karitiana, but I’m not really sure. We’ll find out, in the next few days.

    Regardless, I wanted to share this graph, no migration edges. I think it clearly demonstrates the point that ANE, WHG, and ENA are extremely close:

    Read More

Comments are closed.