search this blog

Sunday, September 11, 2016

Eurasians: an idiot's guide

Below are a couple of images from two unsupervised analyses featuring a variety of ancient and present-day Eurasians (and one ancient American). The results should be self-explanatory. Click on the images to download high resolution versions. The Principal Component Analysis (PCA) datasheet is available here.

I'm using fully homozygous sequences. Most of the ancient samples are composites to maximize the number of markers (over 400K SNPs). All of the ancient samples and almost all of the present-day samples are freely available at the Reich Lab website here, but in some cases with fewer markers than what I'm using.

Let me just reiterate, this is an unsupervised analysis; TreeMix is simply told to find the best fitting migration edge in the data, and it does the rest, showing the ancient American, Clovis, as a mixture of East and West Eurasians. To be honest, I'm blown away by the accuracy of these results and the strong correlation between the PCA and TreeMix graph.

See also...

East and West Eurasians seperated at least 45,000 years ago, but...


Davidski said...

The tree with three edges is interesting, and again in line with the PCA.

Anonymous said...


Does that Treemix suggest that ANE is a mixture of something like EHG and some basal East-Asian?

Davidski said...

I don't think so, because AG3-MA1 is a lot older than EHG, so it's probably showing the development of the ANE clade from something ancestral in Siberia, also related to East Asians, and then also trying to fit in the fact that EHG is a mixture of ANE and WHG.

Having two ANE samples would probably be useful in this case, but AG3 is missing too many markers for this analysis to be run separately.

Anonymous said...

Is AG2 (the one that the original Mal'ta paper discussed) online? It is apparently not a very good one but maybe combine them?

Anonymous said...

It was probably nothing. But I do find it very odd that none of the unparental markers of MA1/AG3 pop up with American Indians. Mind you, for such an influential population we have very little samples.

Davidski said...

AG2 is contaminated, so it'd be hard to know what the results mean.

Ir Pegasus said...

"in the fact that EHG is a mixture of ANE and WHG."
It is not a fact. WHG is drifting towards EHG.

"AG3-MA1 is a lot older than EHG"
AG3-MA1 is older found samples of Karelia and Samara HG. But the EHG origin time is not defined, because there is no earlier samples.

Matt said...

@ Davidski, is it possible to add any of: Kostenki14, GoyetQ113, El_Miron or Ust Ishim to a version of the same PCA?

I'd be interested to see where they fit.

I'd think on the normal modern World PCA, the weight of samples leads the PC2 to load heavily on EEF-East Asian (or WHG-East Asian), so the ancient European Upper Paleolithics would then tend to be displaced centrally.

The loadings and positions here seem different because of the different relative sample sizes. Particularly, would the EUP be further out on the "Crown Eurasian cline".

Also adding Boncuklu / Anatolia Neolithic in if possible...

FrankN said...

@Dave: I suggest to use the Label "Anzick1" instead of Clovis, since it isn't clear whether Anzick1 actually genetically represents Clovis, or the contemporary Western Stemmed Tradition (WST) of the Pacific Coast.
WST seems to be the older of the two. To the extent Mesa Verde/ Chile can be linked to WST as well, in now seems to have appeared in the Americas as early as 19 kya. [This, btw, calls for a re-calibration of TMRCA estimates that have so far been based on a assumed Settlement of the Americas some 15.5 kya.]

WST expanded eastwards from the Pacific Coast, to encounter Clovis somewhere east of the Rocky Mts. After that encounter, Clovis retreated and transformed into the Folsom Culture, while WST continued eastwards expansion across the Rockies into Eastern Colorado. This "clash of cultures" may not always have been peaceful.

Anzick1, a 1-2 year-old infant, was found in Western Montana, in the a/m contact zone that was later given up by Clovis People. The burial was richly equipped with hundreds of stone and antler tools, and projectile points, several of which, as the infant's remains, were covered in red ochre. One might assume a "princely grave", if that wouldn't imply a degree of social differentiation that can hardly be expected for epipaleolothic HGs. So, most likely we are dealing with a ritual offering, which is supported by the fact that the remains were partly dismembered and de-fleshed.

Did the Clovis people sacrifice their own offspring? Or did they get hold of an WST infant for that purpose? We don*t know - both is possible. Against that background I think it is premature to assign an ethno-cultural affiliation to Anzick1.

While I am at it: There is increasing archeological evidence for pre-Clovis settlement of the American East Coast.
"At Page-Ladson [FL], hunter-gatherers, possibly accompanied by dogs, butchered or scavenged a mastodon carcass at the sinkhole’s edge next to a small pond at ~14,550 cal yr B.P. These people had successfully adapted to their environment; they knew where to find freshwater, game, plants, raw materials for making tools, and other critical resources for survival. (..)
Page-Ladson shows that much of the earliest record of human habitation of the American Southeast lies submerged and buried in unique depositional settings like those found along the Aucilla River. This record can only be accessed through underwater investigation, which, if undertaken with intensity and focus, should reveal a rich and abundant pre-Clovis record for the American Southeast.
Interesting here is a/o the reference to dogs. There are a number of dog aDNA studies unanimously pointing at Paleoindian dogs originating in Europe (refererences on request). Moreover, American bottlegourds, w. oldest findings around the Gulf of Mexico, have been demonstrated to belong to the African, not the Asian clade. UP colonisation from Europe could explain Amerindian mtDNA X2, possibly even R1b, which both concentrate in NE America.

Karl_K said...


"UP colonisation from Europe could explain Amerindian mtDNA X2, possibly even R1b, which both concentrate in NE America."

Now this is a reasonable suggestion! I can't see any more reasonable alternatives to this. Especially with those African gourds and all...

capra internetensis said...


Genetic TMRCAs aren't based on assumed times of colonization, they are based directly on ancient genomes.

X2a has highest frequency in NE North America, but that seems to be down to an Algonquian founder effect involving the X2a1 subclade, so it doesn't tell us anything about the likely point of arrival. X2a root haplotype was present in western North America 9000 years ago (Kennewick Man), whereas out of around 40 Palaeolithic European ancient mtDNAs, and another 15 Mesolithic ones, none at all have X; it appears with the Neolithic in West Asia and then Europe.

If there *is* an American R1b branch, how could we tell if it centred in the Northeast? - seeing as we don't even know if it exists, much less what proportion of present day - mostly colonial - R1b it makes up in different areas, or where it is most diverse.

Colin Welling said...

@david, can you help clear something up for me. Ive seen these pca's for a long time and for global populations its always the same picture. Africans at a corner, then europeans (especially west med.) at another corner, and finally east asians. I dont understand why europe is connected first, then asia.

Is the chart telling us that africans are closer to europeans than asians? I know this can be supported by admixture and ancient dna studies that suggest an input of west eurasian (probably near east) into africa. Why are western meds the corner? What does the corner signify. Western meds arent "pure" in the sense of mesolithic populations. Why wouldnt WHG sit at a corner? Why does MA and thus EHG lean towards Asia?

Colin Welling said...

"so it's probably showing the development of the ANE clade from something ancestral in Siberia, also related to East Asians"

I guess this could explain why EHG leans towards asia on the europe asian axis. But is that right? I never really heard people even caring to explain why european populations shift eastward, even prior knowledge of ANE.

FrankN said...

@Capra: "Genetic TMRCAs aren't based on assumed times of colonization, they are based directly on ancient genomes."
I still owe you thx for a few days ago drawing my attention to recent developments in TMRCA estimation, in particular the Y-Full paper on "Defining a new rate constant for Y-chromosome SNPs based on full sequencing data."
Now, please note that the a/m paper concludes (p.81): "Our analysis of the BigY and FGC data collected in the YFull database allowed us to fine tune the probable date of the arrival of humans in the Western Hemisphere (14,800 b.p.)."
Apparently, that estimate is off by ca. 5,000 years, which indicates quite some scope for methodological review and improvement.

As concerns mtDNA X2, I agree that even when assuming an UP European (or NW African?) colonisation of the American East Coast, it is challenging to connect American X2a to W.Eurasia (earliest X2 attestation at Ganj Dareh, ca. 10 kya, followed by Anat_Neol - Barcin, Fikirtepe). However, for the dearth of X2 in East Eurasia, a transatlantic path to N. America is more parsimonious than a Beringian one. X1 is mostly restricted to North Africa, suggesting that X originated there, which implies the possibility of X2a also having split off in North Africa. Let's hope and wait for some North African UP aDNA to come in and possibly provide a clearer picture.

"If there *is* an American R1b branch, how could we tell if it centred in the Northeast?"
I commented extensively on it under
(somewhere around comment 150, search the text for my name):
"In short: I have little doubt that a good part, maybe 2/3, of Amerindian R1b relates to recent European admixture. But there is quite some indication of an older admixture layer that appears to have received little attention in research so far."
In fact, all recent studies have immediately discarded any Amerindian R1b as signature of recent European admix, so your question is valid and still hardly addressed scientifically. There was, however, an older (mid-2000s) paper, which I unfortunately fail to recower, that drew attention to the goegraphical overlap of Amerindian mtDNA X2 and yDNA R1(b). For a general impression on the frequency of R1 in native Americans, consult the following:

Awale Abdi said...

@ Colin

Yes, David did a good job by noticing that "ANE" is what causes the Eastern-shift but, last I checked, we don't really know why ANEs themselves shift eastward on PCAs. Any ideas, David?

Denis Anderthal said...

"UP colonisation from Europe could explain Amerindian mtDNA X2, possibly even R1b, which both concentrate in NE America."

Maybe the R1b is from Vikings who made it up the St. Lawrence river -- and possibly even beyond. R1b might have a slight bias towards producing boys (although it didn't for me). And maybe hypergamy could have also helped with its spread.

Seinundzeit said...

Looking at some very interesting TreeMix experiments David tried a few months ago, I think the eastward shift of ANE might be explicable via complex lines of gene-flow from the center of Eurasia to its periphery.

For example, there is definitely a cline among ENA populations, on the PCA plot here. The Native American sample is closest to AG3, and the Australasians are the furthest, with Han/Onge being intermediate.

The distinctive nature of Australasians can't be chalked up to just Denisovan ancestry, since that dimension of variance is accounted for, on this plot. Rather, it seems that they have the least West Eurasian ancestry of all ENA populations, and Native Americans have the most.

Basically, I think the eastward shift of ANE could be a function of widespread ANE ancestry in East Eurasia.

FrankN said...

@Denis: "Maybe the R1b is from Vikings.."
In that case, one would expect it to be joined by quite some I1. However, citing from my comment under
"Hammer e.a. (Fig. 1) provide the following ratios for the USA (Native American share, European Ancestry share, quotient), which disprove that expectation (and also speak against Vikings/Icelanders as a putative source of pre-Columbian R1b ingression):

R1b1a2: 21.9%/58.3% = 0.38

R1a1a: 1.5%/7.2% = 0.21
I1(P30): 2.8%/11.7% = 0.24
I2a1 (P37-2): 0.8%/2.7% = 0.30
I-P19* (I2a2?): 1.0%/4.9% = 0.20
G2a (P15): 0.3%/3.6% = 0.08
Subtotal: 6.4%/30.1% = 0.21 "

I am willing to consider (proto-)Basque whalers. Their 16th/17th ct. presence off Newfoundland is well attested, and it gave rise to the

That whaling history reaches quite back in time, evidenced, e.g., by Roman period official price settings for whale blubber on the Biscayan coast. Hence, (proto-)Basque interaction with NE Americans may date back some 2000 years or more.

Still, such timeframe would be insufficient to explain those five Amerindian R1b sub-haplotypes reported in Bolnick e.a. (Link, Fig. 6) as being 4-8 mutations removed from the nearest European haplotype, measured across 10 microsatellites.

While I am having issues with TMRCA estimates in general, 8 mutations across 10 microsatellites rather seem to point to a divergence time of more than 10,000 years than towards historical timeframes. But I am anything but an expert in this respect - I just wished experts would seriously take up the matter instead of labelling any Amerindian R1b as "recent European admixture" withour further investigation.

Davidski said...


Individuals from dead branches and especially with inflated archaic ancestry are still a problem. I don't have any good ideas how to fit them into a decent PCA. But the Barcin farmers cluster exactly where I'd expect them to.


Western Meds aren't really in a corner; they're shifted towards Africa relative to North Euros, European hunter-gatherers and East Asians. So the corner is really in no man's land.

But the Neolithic farmers and Sardinians do stick out a bit further than anyone else, probably because of shared drift with each other and less shared drift with East Asians.

And like you say, there's a cline more or less linking Europe to Africa caused by relatively recent pulses of admixture mostly from West Eurasia to Africa, but also the opposite way. East Asia hasn't had the same sort of relationship with Africa.

Basal West Eurasian admixture also pulls West Eurasians closer to Africans on PCA plots, although not in formal tests.


The ANE and EHG samples are pulled east because they share relatively more drift with East Eurasians and less with western Neolithic farmers and their descendants. They might also have East Asian ancestry at some level, depending on how we define East Asian.

Mbuti Han Villabruna MA1 0.0203 4.39
Mbuti Dai Villabruna MA1 0.0225 4.901
Mbuti Onge Villabruna MA1 0.0203 4.182

Mbuti Han Villabruna Karelia_HG 0.0255 5.944
Mbuti Dai Villabruna Karelia_HG 0.0245 5.641
Mbuti Onge Villabruna Karelia_HG 0.0136 2.969


It is a fact. EHG are somewhere around 40-50% WHG. But all of the WHG samples that we have are also part ANE, probably usually around 15%. This might just be a feature of the WHG clade because Villabruna has it too. So EHG are probably something like 75% ANE and 25% proto-WHG.

Balaji said...

Davidski explained his Treemix graph with three edges which shows a 41% migration edge from a node before Onge, Han and other ENA as follows, “it's probably showing the development of the ANE clade from something ancestral in Siberia, also related to East Asians”. This seems unlikely to me. The Onge are about as far from Siberia as it is possible to be in the Northern hemisphere. They could not have been affected by any developments in Siberia. Instead it is more likely that both ANE and the Onge originated in the Indian subcontinent and this accounts for their shared ancestry. The ANE went North and the Onge went south.

Shaikorth said...

Wong et al's D-stats with high coverage sets showed Dai etc were slightly Dinka-shifted and Mbuti-shifted compared to French, so not all African groups have West Eurasian. This should be repeatable with SGDP to investigate further

Davidski said...


The Onge are about as far from Siberia as it is possible to be in the Northern hemisphere.

But people travel.

Think about it. Ust_Ishim clusters near the Onge on trees like that.

Davidski said...

Btw, Matt, here's the PCA datasheet with K14 and Barcin_N.

Denis Anderthal said...

@FrankN: Thanks for the informative post. I suspect West-Central Norwegian R1b from 1100 years ago would be somewhat different from modern West Germanic / Celtic R1b in terms of STR values. I'm not sure which markers were used, but some are faster-mutating than others -- although 8 mutations away might be pushing it. Maybe glancing at some Norwegian dna projects and comparing the STR values (and SNPs) of R1b men there to their British counterparts would be helpful.

Regarding dog dna and Kennewick Man's X2, that also fits with Q-L804, the vast majority of which is found in NorthWestern Europe, and which is the closest branch to the Amerindian Q-M3. Maybe the Q-L54 precursor stretched all the way from Siberia to Spain 20,000 ybp, and maybe that could explain some of the 15% ANE found in Villabruna and Motala.

Karl_K said...


The dog DNA studies can easily be explained without African Basque Whalers.

Number 1, most 'pure' Native American dogs are hybrids with European dogs.

Number 2, it was recently shown that Paleolithic Eurasian dogs were seperated into 2 major populations. These have been called 'Eastern' and 'Western' but there is no information on their distribution 15-20,000 years ago. They might better be described as Northern and South-Eastern, or cold and warm.

The ANE-like ancestors of Native Americans could have had Western/Northern/colder dogs, that then went along to the Americas.

And as for all the other things, you are acting as if the least likely scenerio is true, with zero real evidence that this is the case.

Ir Pegasus said...

"It is a fact." Nope.
"EHG are somewhere around 40-50% WHG. But all of the WHG samples that we have are also part ANE, probably usually around 15%. This might just be a feature of the WHG clade because Villabruna has it too."
WHG increase closeness to EHG in the PCA. Switzerland_HG (early WHG) is initially far from EHG (Karelia and Samara HG), and then closer and closer over time. Consequently, it was the influence of EHG to the WHG, which can be reflected either as the impact of EHG or seeming ANE, depending on chance. But not vice versa.
How it happened can be seen at SHG, which is a mixture of WHG and EHG as a result of migration EHG to WHG. WHG we have a lot of different times and a large spread, and EHG is small and only one time and a small spread, so the calculation is incorrect. But in Lazardis et al. 2016, he is independent component already. Now while WHG modelling as 93% Siwitzerland_HG + 7% EHG.

FrankN said...

@Karl_K: As you implicitly asked for it, here is a bit on dogs in America:
1. "Basal" E.Asian mtDNA A29 (Alask. Malamute, Dingo, Korean Pungsan etc.) is estimated to have formed shortly after the dog-wolf divergence. The linked Skoglund e.a. study estimates the latter at 13.4 kya, the former at 12.8 kya, noting however substantial uncertainty with respect to the mutation rate to apply.

2. In a recent paper (2015), Skoglund has revised the dog-wolf divergence to 27-40 kya, which implies a split of the "basal" East Asian A29 prior or around the LGM. The paper furthermore demonstrates Taimyr wolf admixture in East Siberian and Arctic American dogs, which means that at a yet unknown time, dogs entered Arctic America via Beringia.

3. However, those Taimyr-wolf admixed haplotypes have remained unique to Arctic breeds, and neither made their way further down into the Americas, nor westwards into Central Siberia and beyond. The same applies to "basal" East Eurasian A 29 that has been found with the Alaskan Malamute, but is otherwise restricted to East Asia and Oceania.

4. While the Californian Channel Islands were settled 13kya, the presence of dogs is securely attested only from some 7kya. A dog mandible was found in a 10-8.6 kya cultural layer, but hasn't been directly dated and may stem from a younger site component. Apparently, early coastal migrants didn't yet possess domesticated dogs.

There also haven't any dog remains been found in the well-researched Texas Clovis and Folsom sites.

In contrast, early canid finds are quite frequent in the SE USA. Aside from Page-Ladson/FL mentionned above, they are a/o documented from four more sites in Florida, Coats-Hines/TN (mastodon hunters, so prior to 10 kya), and the Savannah River site (SC, 11-9,5 kya).

5. Three pre-Columbian dogs, including an 8.5 kya sample from the Koster site (IL), fall into mtDNA clade A, which forms a sister clade to a 14.5 kya wolf from the Kesslerloch Cave, Switzerland.

6. While poorly sampled, dog yDNA in non-arctic America turned out to be exclusively HG1, which is of presumed European origin.

In short - presumptive ANE/ENA dogs made it to Alaska and Greenland, but never beyond, while those "hybrids with European dogs" appear as early as 8.5 kya. In this respect, note also from Link (3): "Haplotype A185 was found exclusively in ancient Mexican samples (D25) and in modern samples of the Mexican Chihuahua, strongly indicating direct ancestry of Chihuahua from Mexican pre-Columbian dogs."

Karl_K said...


You are mixing up early and later migrations. We know that people moved down the west coast by water at least 18,000 years ago, if not earlier. Well before the inland path was hospitable.

The arctic dogs clearly came much later, and have nothing to do with South American, Mexican, Carolina Dogs, or other early breeds.

Most of your links are already old news. Until we have lots of ancient genomes that disprove it, I will stick to the most likely scenerio.

All early American dogs came with the first inhabitants via water along the Pacific coast. Later, dogs admixed with arctic wolves arrived in the far north with later human populations.

Gioiello said...

Many of you have questioned about the presence of hg. R1b in Native Americans before Columbus, or Vikings if you prefer, or Thor Hayerdhal etc.
We need proofs. As I wrote many times, I checked all the Native American haplotypes, an the unique old R1b, very likley an R1b1*, was of the types found in the British Isles and linked to Italian and Sardinian ones, i.e., very likely, introgressed from some colonist from Albion.

Karl_K said...


Finally! You are the most reasonable person speaking!

Anonymous said...

Onge's position is really confusing me now and making me think.....

Could Onge itself be a composite of two lineages? I'm thinking, roughly, one coastal lineage, 70,000 kya, and a more inland lineage from closer to 50,000 kya.

How possible would it be? Also, could it even be detected considering the minimal divergence at that time, and almost 40,000 years of isolation on the Andaman islands?

Karl_K said...


There is almost zero chance that Onge have been on those islands for 40,000 years.

Probably much much less.

Samuel Andrews said...

"There is almost zero chance that Onge have been on those islands for 40,000 years.

Probably much much less."

Amen!! This is a lesson for everyone, never believe theories about time scales of population continuum made by genetics based on modern DNA, especially if said time scale is stated as proven fact when it isn't proven.

Karl_K said...


The time of genetic divergence from other populations might be correct, but it has nothing to do with the time spent at the current location.

jparada said...

Yes, David did a good job by noticing that "ANE" is what causes the Eastern-shift but, last I checked, we don't really know why ANEs themselves shift eastward on PCAs. Any ideas, David?

My own guess is that ANE is actually a composite between a population basal to East Eurasians, from which they got their Y-dna, and another one basal to Palaeoeuropeans, from which they got their mt-dna.

Anonymous said...


"My own guess is that ANE is actually a composite between a population basal to East Eurasians, from which they got their Y-dna, and another one basal to Palaeoeuropeans, from which they got their mt-dna."

We know surprisingly little about ANE's unparental markers. One Y-DNA that is extinct but in the R tree. Two mtDNA's: R3 and U.

However, if we consider ANE ancestral to American Indians and EHG, and that seems undisputable, we may consider Y-DNA Q and mtDNA C1 ANE's as well. Currently no Y-DNA R1b is consider native to American Indians and No (?) mtDNA U has been found among them.

We might recall that both "farmers" and "European HG's" showed large, surprising structure when they were investigated in detail.

Ariele Iacopo Maggi said...

[1] "distance%=8.9032 / distance=0.089032"
"Vestonice16" 71.4
"Han" 28.6
"Yoruba" 0
"Papuan" 0
"Somali" 0
"Israel_Natufian" 0

--Adding Karitana
[1] "distance%=5.7347 / distance=0.057347"
"Vestonice16" 67.15
"Karitiana" 26.6
"Israel_Natufian" 4.95
"Han" 1.3
"Yoruba" 0
"Papuan" 0
"Somali" 0

Could be Han into ANE, or ANE or basal ANE into Han. Notice that the distance is only 5 and 8%. If you try Villabruna with ancients you will have a far worst fit.
[1] "distance%=15.6492 / distance=0.156492"
"Vestonice16" 77.45
"AG3-MA1" 22.55
"Yoruba" 0
"Papuan" 0
"Han" 0
"Somali" 0

I don't see why ANE can't be mostly Vestonice+HAN.
Btw, Han as Karitana+Papuan+AG3-MA1...
[1] "distance%=17.4662 / distance=0.174662"
"Karitiana" 50.85
"Papuan" 29.4
"AG3-MA1" 19.75

Anonymous said...


Also, is there any Y-DNA Q among EHG or its decendants? The decendants of ANE seem evenly split: R1 in the west, Q in the America's

Anonymous said...

@Ariele Iacopo Maggi

"If you try Villabruna with ancients you will have a far worst fit."

Could Villabruna as an admixture of Gravettian, an unkown source and slight Magdalenian and ANE work?

Dude ManBro said...


Yes, Y-DNA Q was found in one of the Khvalynsk men; which were predominately EHG, if I recall correctly.

Shaikorth said...

"Tianyuan from North China at 40,000 BP was already "East Asian" on his chromosome 21."

I wouldn't be so sure.

Tianyuan ASD to
Yoruba 31,059
Dai 23,339
French 23,168

It's plausible it went to the East Eurasian branch in the original study's tree model because it's like Ust-Ishim, undifferentiated Crown Eurasian.

@Ariele Iacopo Maggi
Lazaridis 2016 tried to fit EHG's as WHG+Han, it failed and EHG+Onge for Han looked much better. AG-2 and MA-1 combined with Onge look good for Han too. ANE and Onge may be mixed (perhaps involving even more extreme unknown populations in the ANE-Onge cline than MA-1 and Onge themselves), but Han are a more recent mix.

Open Genomes said...

David and everyone, here again is something quite remarkable:

Eurogenes Eurasians PCA superimposed on the Old World
(David, if you'd like you can post this image at the bottom of the post itself along with a link to the larger image above.)

Look at how the rotated PCA plot more or less corresponds to the geographical locations of the ancient and modern samples!

Open Genomes said...

The differences with the real geographic locations are fascinating:

● Villabruna is not very far from the Near Easterners, compared to the East Asians and Oceanians. Villabruna appears to have had some input from the post-LGM Near East, compared to the earlier Magdalenians/Epigravettian Culture. We know that there was a population in Central Anatolia before 8,300 BCE/10,300 BP that appears to have been essentially identical to Villabruna / WHG. Of course, Villabruna himself was R1b1a-P297*. Yes, in fact Villabruna is derived for not one but two R1b1a-P297 SNPs, and this of course is the main non-R1b-V88 branch. On the other hand, R1b1-L389* was found in the ancient Near East and today in Eastern Anatolia and Georgia. This is an earlier split from the R1b tree than Villabruna from 14,180-13,780 BP. Kura-Araxes I1635 was in an R1b1-L389* branch.
● Motala is even closer to the Near Easterners. We know that a Swedish Hunter-Gatherer, Motala 2, was Y-DNA I2c2, and this was also found in the Early Neolithic farmer I1096/Bar26 from Barcin in Northwest Anatolia.
● CHG plots very closely to the Iranian Chalcolithic. The Eastern Anatolian PPNA population appears to have been very close to the much later Iranian Chalcolithic, but we can see that both Satsurblia and Kotias have non-trivial EHG admixture. Kotias in particular appears to have some Eastern Anatolian input missing in Satsurblia.
● The Levantine Natufians/Pre-Pottery Neolihtic is shifted to the southwest, which may be expected, but in fact this is the result of a shift towards North Africa. If the Natufians were plotted separately, they would be much further towards Northeast Africa than the Levantine PPN.
● The Eastern Hunter-Gatherers are further to the northeast, toward the Caucasus, but like the Western Hunter Gatherers/Villabruna and Scandinavian Hunter-Gatherers/Motala, they are much closer to the Near East than expected. We know that the Eastern Hunter-Gatherers, like the Scandinavian Hunter-Gatherers, could not have lived in the far north during the LGM. Y-DNA J1 (very likely J1b) was found in the Karelian Eastern Hunter Gatherer I0211 (also, mtDNA C1f was found there), and R1b1a1-M478 was found in the Samara Eastern Hunter-Gatherer I0124. J1 of course is dominates in the Near East today, and R1b1a1-M478 is concentrated in Southern Central Asia (Tajikistan). Again, there is a Near Eastern/Central Asian connection to the EHGs, just like the Scandinavian Hunter-Gatherers and the Villabruna WHGs.
● ANE AG3-MA1 is drawn towards the Near East, and particularly the EHGs and Iranian Neolithic. Does this have something to do with the "Mammoth Steppe" and the Gravettian Culture, which extended from Europe to Siberia? (MA1/Mal'ta was Gravettian.)
● Clovis is about halfway between ANE and the Han. From AdMix we see that Clovis is about 40% ANE, and this corresponds perfectly. However, Clovis of course is much further south than his physical "Beringian" origin. Clearly some of the ancestors of the people of Beringia (and the Kets of Central Siberia, who speak a language distantly related to Na-Dene of North America) moved north from somewhere that the Siberian ANE and East Asian population could meet. This would be further south than Mal'ta and Afontova-Gora in Siberia. Those ancestral East Asians would clearly have not been Gravettians.
● The Han and Onge are remarkably close. The Han and other East Asians, and the Onge, were clearly isolated from the very start of the spread of AMH into Southeast Asia. Tianyuan from North China at 40,000 BP was already "East Asian" on his chromosome 21. The East Asians have about 20% more Neanderthal admixture than the West Eurasians, which draws them further away from other Eurasians.
● The Nasoi from Indonesia and the Papuans show evidence of their isolation and drift from the time of the the first settlement of AMA in Oceania at 40,000 BP. This is in addition to the substantial Papuan admixture with Denisovans.

Ariele Iacopo Maggi said...

[1] "distance%=7.0841 / distance=0.070841"
"Israel_Natufian" 52.35
"AG3-MA1" 46.1
"Papuan" 1.55
(absolutely shocking, 7% is a good distance for a-dna. Corrupted dataset?)
[1] "distance%=6.3616 / distance=0.063616"
"AG3-MA1" 50.3
"Israel_Natufian" 43.35
"Papuan" 5.3
"Yoruba" 1.05
(Another corrupted sample?)
[1] "distance%=15.1334 / distance=0.151334"
"AG3-MA1" 77.55
"Israel_Natufian" 22.45
(Also Villabruna and Loschbour have high distance if you force ANE+natufians. Wheren't they (WHG) closer to middle eastern?)
[1] "distance%=11.3794 / distance=0.113794"
"AG3-MA1" 67.95
"Israel_Natufian" 32.05
(El Miron still has Vestonice admixture and thus she has a lower distance. I'm puzzled. Deeply flawed methodology? Corrupted data?)
[1] "distance%=10.2688 / distance=0.102688"
"Vestonice16" 86.7
"LaBrana1" 7.1
"Yoruba" 6.2
(Something wrong, I tought that WHG were different from UP europeans beacuse they were closer to modern middle east, maybe natufians are another story?)
[1] "distance%=10.6544 / distance=0.106544"
"GoyetQ116-1" 70.1
"LaBrana1" 22.45
"Yoruba" 7.45
(With Goyet there is more LaBrana)
[1] "distance%=8.8794 / distance=0.088794"
"GoyetQ116-1" 60.6
"AG3-MA1" 34.75
"Yoruba" 4.65
"LaBrana1" 0
(No LaBrana at all in Iran neolithic)
[1] "distance%=10.6634 / distance=0.106634"
"Vestonice16" 65.8
"AG3-MA1" 23
"LaBrana1" 8.85
"Yoruba" 2.35
(Ok, Palestians have 23% more ANE, but they still prefer Vestonice)
[1] "distance%=10.9365 / distance=0.109365"
"LaBrana1" 40.6
"AG3-MA1" 35.85
"Vestonice16" 23.55
"Yoruba" 0
(South Italians start to prefer LaBrana over Vestonice)
[1] "distance%=13.0491 / distance=0.130491"
"GoyetQ116-1" 51.95
"LaBrana1" 44.9
"AG3-MA1" 3.15
(Ana Neolithic has still more UP europeans then WHG)
[1] "distance%=9.1481 / distance=0.091481"
"LaBrana1" 60.65
"AG3-MA1" 39.35
"Vestonice16" 0
"Yoruba" 0
(Kent and other northern/central europeans don't show any Vestonice)

1)Vestonice is higher in more basal groups.
2)Vestonice works better as ANE+Natufians then WHG
3)Also ElMiron and Goyet show similar trend.
4)Could a corrupted sample increase the african affinities and then show up as more basal?

Also. ANE+Han or not, Native americans are a very divergent eurasian branch.
[1] "distance%=25.6749 / distance=0.256749"
"Han" 80.1
"AG3-MA1" 19.9
(Super bad distance)
[1] "distance%=24.2213 / distance=0.242213"
"AG3-MA1" 72.8
"Papuan" 27.2
(Better then Karitana as ANE+Han)
[1] "distance%=8.8076 / distance=0.088076"
"LaBrana1" 79.05
"Han" 20.95
(EHG are more then WHG and Han, but not much more...)

Using ONGE for Han is questionable. They cluster togheter in many PCAs. Look at the last David's plot. They are closer then ANE and WHG.

Davidski said...

@Open Genomes

The second dimension in the PCA does show a very strong correlation with geography for all of the samples except Amerindians. I'd say this is a reflection of so called Isolation by Distance. Amerindians are a relatively recent product of the colonization of the Americas, that's why they're sitting in the middle of dimension two, rather than beyond the Oceanians.

However, the first dimension shows a very poor correlation with geography. This isn't because of each of the European hunter-gatherer meta-populations colonizing Europe from the Near East, but due to the Out of Africa bottleneck, which results in all Eurasians more or less showing the same genetic distance from Mota.

Note that the first dimension is not able to differentiate between the Han/Onge cluster and the Basal-rich cluster, which seems very unusual at first. But this is because of the Out of Africa bottleneck, as well as the slightly inflated archaic admixture in East Asians, which makes them look a little more basal.

Kristiina said...

Dude ManBro, R1a from Khvalynsk was very close to EHG, while Q1a was shifted to Yamnaya and Iron Age Altai in Davids' PC map. Khvalynsk is dated c. 5000-4500 BC.(

On the basis of a 7000 years old sample we cannot conclude what a 30 000 years old Q was like. In general, we cannot use a younger sample to conclude what much older samples would have been like. Moreover, we also have older Q samples: Anzick, c. 13 000 years, and Kennewick, c. 9 000 years. IMO, yDNA samples vary autosomally a lot in space and time, and theoretically, the original autosomal makeup could only have been the makeup that a mutated yDNA had immediately after it split. In order to say what the original Q was like, we need an autosomal analysis of a, let's say, 30 000 - 40 000 old Q.

Shaikorth said...

"Using ONGE for Han is questionable. They cluster togheter in many PCAs. Look at the last David's plot. They are closer then ANE and WHG."

It's not that questionable, Han show relationship to outgroups consistent with that kind of model and the PCA's dimensions are defined by other things than ANE-Onge specific drift which forces non-identical populations such as Iranian ancients and CHG, or Han and Onge, close together. The non-Onge portion (about 8% if EHG is used and 11% if MA-1 is used) in Han is less significant than the ANE divergence from WHG in the models.

With the data most widely used by scientists and genome bloggers (Human Origins dataset) the ANE-Onge hybrid model for East Asians looks good. If in the future it turns out that the Lazaridis model is wrong because stats gained from high coverage sequence data such as SGDP contradict it, other models derived from genotype sets will have to be re-evaluated too.

Anonymous said...

I am probably reading way to much into this abstract, but if different admixtures up to 30% were there the admixtures must have taken place locally. If done elsewhere the migration should have smoothed out the more extreme differences in admixture.

That means EHG-ish HG's were around at the end of the Mesolithic. That may conflict with the idea some have of the Balkans as origin of the Villabruna clade.

Hector said...

"They cluster togheter in many PCAs. Look at the last David's plot. They are closer then ANE and WHG."

Actually ANE and WHG are not that close as Razib mentioned last week. The split is actually not far from the split between ENA and non-Basal "West Eurasians".
They look closer because there was a lot of admixture from the two in the later populations and due to the political and psychological need you guys have to make ANE one of "yours".

Davidski said...

Hector you're a total freak.

Matt said...

@ Hector, just to try and put some numbers on how related ANE is to WHG, Lazaridis 2016 estimates that the EHG is about 75% ANE, 25% Switzerland_HG, where WHG should be 7%, 93% respectively. The Motala HG people in this estimation would be 44%, 66%.

If we go by that, then based on regression equations with some of the D-stats Davidski has posted (albeit this has its limits), it looks to me like ANE should share about approx as much drift with Villabruna (WHG without that 7% ANE) as the Barcin Neolithic Anatolians do (Anatolia Neolithic are estimated by Laz to be approx. 25 Basal Eurasian, 75 WHG).

Assuming the statistics I'm running are comparable, also about as much drift as the Levant Neolithic shares with the Anatolian Neolithic.

Or if we instead use the same D-stats and just go by the shared drift between AG3 and Switzerland_HG (as in theory the unadmixed reference), with the former as maybe an imperfect reference for ANE, then ANE should be closer to WHG than any present day population (closer than Lithuanians, as the most WHG like West Eurasians today).

Seems like it should be probably somewhere in that range.

If ANE were some population that contributed only, like 40% to EHG and 25% to Motala HG, then we could expect it to be really quite distant from WHG, but with the estimated high contributions to EHG and Motala and with them both being relatively close to WHG and Near Eastern farmers there isn't really a way it can be that distant.

Though D-stats are I think in theory more insensitive to drift rather than divergence time of the phylogeny, so possibly any actual ANE (like a different branch that contributed to East Eurasia) could have been more divergent if more heavily drifted.

I think we don't have enough information yet to say whether the divergence between the ancestors of Onge / Han is greater or lower than that. Possible that it could be greater if the population movements in Asia and subdivisions were before the split of ANE from the broader WHG clade.

Matt said...

@ Epoch2013: There's time between the Upper Paleolithic and the Mesolithic in which things can happen. It's possible that the Villabruna group could have moved from the Balkans successfully into Western Europe, but then been displaced in its center of origin.

What feels a bit puzzling for me is how the HG in the Balkans could have been more ANE related than KO1, as SHG is. I would expect that you would have the ANE level fall off neatly from the steppe with latitude, but from this I guess perhaps that's not the case.

Hector said...

Modern populations, even if they are 100 percent either ANE or WHG(even though such don't actually exist), obviously will be likely farther apart from each other than ANE and WHG are from each other at some remote past because there simply was more time to drift apart from each other.

And estimating the shared drift of a mixed population with a source population is nonsense.

If the "ANE" that contributed to East Asians diverged earlier from WHG than the mainstream ANE did... then it is not ANE. That would be a population X basal to WHG and ANE but still within the Western side of "Crown Eurasians"(ie after the split with ENA). I am sure you guys would love that theory since that solves a lot of problems you guys hate to admit. However that will require East Asians to pull toward ANE and WHG to an equal degree instead of just ANE alone(but it solves the problem of other ENAs not being part of this equation).

I know why you guys are trying this. But no matter what you do you will not get rid of Basal Eurasian components among modern West Eurasians. Thus "West Eurasians" will remain paraphyletic and as political contrivance.

Hector said...

One of the most entertaining and amateurish errors I have seen is that you guys equate a statistical ghost with an actual event.

The current statistical estimate of East - West Eurasian split is greater than 40000 even 45000 years old. Yet most samples even younger than that(such as Ust Ishim, Tianyuan or even Oase) don't show clear membership to either. There are ways out of this such as claiming that these men are from populations that diverged just prior to the split(but lingering on until they become extinct later). But these phony excuses become less and less likely.

The proper interpretation would be that the basal components among West Eurasians were not properly accounted for and the East-West Eurasian split is a ghost that lies between Basal-non Basal split and ENA - North Eurasian split.

Hector said...

"Hector you're a total freak."

Man, I hate to hear that from someone who habitually complain about research establishment hating "White folks". I thought you had cleaned up from your wilder days of DNA-forums and fights with Dienekes until I saw another one of your outbursts a few months ago. I thought Genetiker might have hacked your blog.

Underneath all that D-stats, treemix and f3 you seem to love lately, you are still that loon that I know so well.

Davidski said...

Hey Hector, tell us again how the Andronovo people were indigenous Siberians rather than migrants from Eastern Europe, so I can have yet another laugh at your expense.

Hector said...

Maybe the same way Oase is East European? You are intentionally conflating geography with phylogenetics. That is why you remain an amateur.

Rob said...


"What feels a bit puzzling for me is how the HG in the Balkans could have been more ANE related than KO1, as SHG is"

You have a cryptic writing style Matt. :)
I follow the first part, i think. And that's because Ko1 must be "western Technocomplex" HG. It this makes perfect sense that he doesn't have much ANE

What do you mean by "as SHG is"?
Sorry I don't follow that English . Can your clarify please Matt ;)

Davidski said...


Maybe the same way Oase is East European? You are intentionally conflating geography with phylogenetics.

The Andronovo genomes are freely available, and for all intents and purposes they're Europeans who were living in Siberia.

That was always my assessment and this is now also the academic consensus.

You're such a sore loser. :)

Hector said...

LOL. Andronovo is of Bronze Age. I did not even feel the need to answer your false accusation. The only thing that comes close to it, I can think of, is that I said remains from specialized graves like Kurgans may not be representative of the population at the time and accidental deaths provide more realistic pictures.

You are imagining things. That is why you are a loon. Did Academia stop hating white folks by now?

Davidski said...

You argued very specifically that Andronovo were not of European origin but indigenous Siberians, and only imagined as Europeans by Eurocentrics.

No squirming out of that now.

Clearly, you're too emotional, butthurt and incompetent to be able to make these sorts of calls correctly. Eurocentrics are stalking you, bahaha.

Go lay down for a while on a shrink's couch, you incompetent fool.

Hector said...

It is generally a good sign to be considered incompetent by a loony who imagines things I never said.

Davidski said...

Watch out for those Eurocentrics Hector. They're dangerous and they're out to get you...

Hector said...

Probably because they have such awesome psychokinetic power; they make things happen simply by imagining things. Force be with you. Keep watching pseudo sci fi's like Star Wars.

Gioiello said...

@ Davidski

"E tu onore di pianti, Ettore, avrai
ove fia sacro e lagrimato il sangue
per la patria versato e finché il sole
risplenderà sulle sciagure umane"
(Foscolo, I Sepolcri)
Have I to put also Hector in the Middle Easterner tumulus with Behar and all the others?
Look at another paper next on ISOGG: mt H7j "Ashkenazic" introgressed from Eastern Europe.

Anonymous said...


We also have older samples from Romania: Muieri and such. They don't show special affinity to Villabruna. At least this find limits the place of origin in time.

KO1 is also strange as it's mtDNA (R3) is shared by AG3 and not found anywhere else in old samples. So it might have a link with ANE, albeit not in an autosomal component. R3 is currently found only (?) in Armenians.

Matt said...

@ Hector:

If the "ANE" that contributed to East Asians diverged earlier from WHG than the mainstream ANE did... then it is not ANE. That would be a population X basal to WHG and ANE but still within the Western side of "Crown Eurasians"(ie after the split with ENA).

Definitely no one ITT or academically is saying this though, so this seems like a non-sequitur from you. The model from Lazaridis 2016 has the contributing population to East Asia and Native Americans from the same groups that splits from West Eurasian HG ancestry and contributes 75% to EHG.

@ Rob:

The Mathieson abstract that refers to Balkan HGs: "We document how the dynamics of admixture between (the Balkans') first farmers and its indigenous hunter-gatherers was complex, with evidence of local admixture from hunter-gatherers related to those from Scandinavia and Eastern Europe."

Therefore implication is Balkan HG had ANE ancestry, like SHG and EHG ("hunter gatherers from Scandinavia and Eastern Europe").

Therefore "HG in the Balkans could have been more ANE related than KO1, as SHG is (more ANE related than KO1)".

Rob said...

That's what I thought thanks
So K01 must be related to other west European HGs, with the most east shifted HG ancestry arriving in the Carpathian basin, and further west, quite late

Davidski said...

They couldn't have arrived after KO1, because KO1 dates to the Neolithic.

Rob said...

Yes Dave I know the Neolithic is after the Mesolithic, but my point is the foragers in Hungary, and those admixed in Hungarian and nearby (eg German) farmers were "pure" WHG, whilst there existed in a broad arc, perhaps discontinuous, some kind of "SHG' foragers in the eastern Balkans, trans carpathian and Ukraine (I'm guessing). These appear to have remained separated until the LNBA

Gioiello said...

Davidski, I answered that to Dartraighe on eng.molgen. Am I wrong in something?

We know for certain that Hunter-gatherers in Western Europe were practically the same both in Iberia as in Switzeland as in Italy. The fact is that all the others went extint (in fact the hg. C of Iberia is very rare now) and Western Europe was peopled after the Younger Dryas from Villabruna. After Villabruna migrated also to Yamnaya, beyonf Anatolia and Middle east, (the same Davidski said that a few minutes ago on Eurogenes blog), diluting ANE in Samara, thus the mix were fifty-fifty: you should find there the R1b and perhaps the R1a of Eastern Europe from Villabruna. It is possible that Villabruna came from elesewhere, not only from North, but from the whole European-Siberian corridor, but in the Villabruna time, from LGM to the end of the Younger Dryas, Villabruna differentiated from EHG, otherwise we couldn't understand why Davidski said that he had 15% ANE against the 70% of EHG.
We'll never know what there was in Northern Europe, if they all went extinct. If we'll find some bone, they very likely were similar to Villabruna, with more or less ANE as to they were more or less linked to EHG.