December 24, 2013

Europeans = Neolithic farmers, Mesolithic hunter-gatherers and "Ancient North Eurasians" (etc.)

A new preprint on the bioRxiv reports ancient DNA from a Mesolithic European hunter-gatherer from Luxembourg whose mtDNA was published a few years ago and a Neolithic European LBK farmer from Germany, as well as several Mesolithic hunter-gatherers from Sweden.

The Luxembourg sample is similar to the Iberian La Brana samples and the Swedish Mesolithic samples are similar to Swedish Neolithic hunter-gatherers. The LBK farmer is similar to Oetzi and a Swedish TRB farmer and to Sardinians. The authors also study the recently published Mal'ta Upper Paleolithic sample from Lake Baikal and find that it is part of an "Ancient North Eurasian" population that also admixed into West Eurasians on top of the Neolithic/Mesolithic mix.

The authors' proposed model and admixture estimates:



It seems that the estimates go all the way to "almost pure" Early European farmer ancestry but "West European Hunter-Gatherer" and "Ancient North Eurasian" ancestry isn't found unmixed in any modern populations. The model seems to agree with Raghavan et al. that Karitiana are "Mal'ta"-admixed but also finds the most basal Eurasian ancestry in the European Neolithic farmer. The authors write:
The successful model (Fig. 2A) also suggests 44 ± 10% “Basal Eurasian” admixture into the ancestors of Stuttgart: gene flow into their Near Eastern ancestors from a lineage that diverged prior to the separation of the ancestors of Loschbour and Onge. Such a scenario, while never suggested previously, is plausible given the early presence of modern humans in the Levant25, African-related tools made by modern humans in Arabia26, 27, and the geographic opportunity for continuous gene flow between the Near East and Africa28
The Swedish/Luxembourg Mesolithic hunter-gatherers are all mtDNA-haplogroup U and Y-chromosome haplogroup I, so again no R1a/R1b in early European samples.

An interesting finding is that the Luxembourg hunter-gatherer probably had blue eyes (like a Mesolithic La Brana Iberian, a paper on which seems to be in the works) but darker skin than the LBK farmer who had brown eyes but lighter skin. Raghavan et al. did not find light pigmentation in Mal'ta (but that was a very old sample), so with the exception of light eyes that seem established for Western European hunter-gatherers (and may have been "darker" in European steppe populations, but "lighter" in Bronze Age South Siberians?), the origin of depigmentation of many recent Europeans remains a mystery. Ancient DNA continues to surprise at every turn.

UPDATE (4/4/2014): a new version of the preprint.

bioRxiv doi: 10.1101/001552

Ancient human genomes suggest three ancestral populations for present-day Europeans

Iosif Lazaridis et al.

Analysis of ancient DNA can reveal historical events that are difficult to discern through study of present-day individuals. To investigate European population history around the time of the agricultural transition, we sequenced complete genomes from a ~7,500 year old early farmer from the Linearbandkeramik (LBK) culture from Stuttgart in Germany and an ~8,000 year old hunter-gatherer from the Loschbour rock shelter in Luxembourg. We also generated data from seven ~8,000 year old hunter-gatherers from Motala in Sweden. We compared these genomes and published ancient DNA to new data from 2,196 samples from 185 diverse populations to show that at least three ancestral groups contributed to present-day Europeans. The first are Ancient North Eurasians (ANE), who are more closely related to Upper Paleolithic Siberians than to any present-day population. The second are West European Hunter-Gatherers (WHG), related to the Loschbour individual, who contributed to all Europeans but not to Near Easterners. The third are Early European Farmers (EEF), related to the Stuttgart individual, who were mainly of Near Eastern origin but also harbored WHG-related ancestry. We model the deep relationships of these populations and show that about ~44% of the ancestry of EEF derived from a basal Eurasian lineage that split prior to the separation of other non-Africans.

Link

169 comments:

Valikhan said...

Just wonder if this supports an idea of Uralic populations being not an intermediate rather a source for Caucasians and Mongoloids.

Hamar Fox said...

Valikhan,

Just wonder if this supports an idea of Uralic populations being not an intermediate rather a source for Caucasians and Mongoloids.

Why would it?

whatdid themonke said...

Does haplogroup G represent this basal Eurasian lineage?

Kurti said...

They should have added a Bronze/Iron age Near Eastern component to the Neolithic farmers, mesolithic H&G and ancient North Eurasians, since there is a relatively strong "Caucasus-Gedrosia" impact on Europe which was obviously not present during the Neolithic.

Eastern View said...
This comment has been removed by the author.
Roy King said...

What an seminal study! The easiest, most parsimonious interpretation comes from the k=16 ADMIXTURE analysis. If you explore the three colors: orchid, light blue and light brown (sand colored) then the light blue component tracks the Mesolithic samples, the orchid, the Neolithic component from the Near East--most purely in the Bedouin samples and admixed with the blue in Stuttgart Neolithic and in Sardinia. The light brown seems to be most concentrated in the Caucasus samples and to some degree in MA1 and not in either the European Mesolithic or the European Neolithic. So the European Neolithic is a hybrid between a Near East and an indigeneous Mesolithic, perhaps represented via Y haplogroups: I/I2 versus G. R1a and R1b spread and diversification is later. J has yet to be found in an ancient DNA context.

Eastern View said...
This comment has been removed by the author.
About Time said...

Too bad they didn't list f3 statistics for South Asia or Central Asia.

Would be good to see things like:

f3(Gujarati; Stuttgart, Onge)
f3(Gujarati; Mal'ta, Onge)

f3(Uzbek; Stuttgart, She)
f3(Uzbek; Mal'ta, She)

Eastern View said...
This comment has been removed by the author.
Kristiina said...

It seems that now we have Neolithic yDNA from Europe (Sweden and Germany)!
Loschbour U5b1a I2a1b
Motala 1 U5a1 I*
Motala 2 U2e1 I
Motala 3 U5a1 I2
Motala 4 U5a2d
Motala 6 U5a2d Q1a2 (?)
Motala 9 U5a2 I
Motala 12 U2e1 I2a1b
Stuttgart T2c1d1
Motala dates are 7-6,7 BP.

Gihanga Rwanda said...

I wonder how close this "basel Eurasian" population is to NE Africans?

Kurti said...

@Eastern View

I think you missed something.

The paper says ancient North Eurasians. And there is an important reason why mentions that, because the modern Siberian admixture was totally absent in the Mal'ta individual. So there is no doubt that Uygurs and other Central-North Asian populations are surely the result of more recent, East and West Eurasian mixture.

Kurti said...

What the Mal'ta individual proved us, was something that we already knew, that some West Eurasians (R1*) and some East Eurasians (Q*) share a common ancestry. But we didn't expect that this common origin was still shared even when Q and R split.

The biggest suprise was that R* was not already exclusively West Eurasian (though predominantly!) but had strong Amerindian and Southeast Asian/ ASI component.

But again this doesn't say much about R1*. Since R1* obviously appears to e of West Eurasian origin. So if R1* already was almost exclusively West Eurasian. It is very unlikely that R1a* in Uygurs, and other North Eurasians is not the result of Iranic and Tocharian admixture.

Hamar Fox said...

Eastern View,

So, Hamar Fox, since you've willing to go along with it so far, you'd have to support this statement: Europeans and SW Asians are basically admixtures of Basal Eurasians and Derived Eurasians, while ANE, Amerinds, and East Asians are pure Derived Eurasians. Basically, people in SE Europe and the Near East are admixtures of half or more Basal Eurasian and Derived Eurasians.

Not related to anything I've said in this thread so far. If you're referring to my statement in another thread (which I know you are, but I should point this out to other readers), then, no, the earliest division in Eurasia is between basal Eurasians and wider Eurasians, then between West Eurasians and East Eurasians. One of the three major West Eurasian populations received major geneflow from the basal Eurasian population. East Eurasians are unrelated to the formation of any of the three W. Eurasian components, which is obvious to me, but obviously much harder for you to understand.

Just let it sit in your minds for a few days, weeks, months. See if it makes sense.

Even though it may seem to you that I have many minds, I have but one.

See if it parallels the example I brought up, of making Uighurs and other Central Asians the pure population and the implications of that to its east and west.

Your point was, and remains, poor. It's obvious that basal admixture wasn't the cause of West/East Eurasian divergence, since, as is clear, not all populations that split from East Eurasians had basal admixture. In fact, initially, none did.

MfA said...

MA1 - K19
Lithuanian (N. European) 32,85%
Mbuti (Pygmy) 0,85%
Chipewyan (Amerindian) 16,73%
Mala (ASI) 11,03%
Pima (Amerindian) 6,06%
Papuan (Ocenian) 0,97%
Kalash (Caucasus+Gedrosia) 26,91%
Karitiana (Amerindian) 2,91%
Onge (Onge) 1,70%

http://abload.de/img/malboy_k19qyzbt.png

Grey said...

So modern Europeans are ancient Europeans plus neolithic farmers in varying proportions.

Hamar Fox said...

Kurti,

I think you missed something.

Yes. Yes, indeed he did.

The paper says ancient North Eurasians. And there is an important reason why mentions that, because the modern Siberian admixture was totally absent in the Mal'ta individual. So there is no doubt that Uygurs and other Central-North Asian populations are surely the result of more recent, East and West Eurasian mixture.

Absolutely correct. The 'Eastern admixture' shown in MA-1 in ADMIXTURE analyses is composed of two things: 1) a degree of real Papuan-related admixture (as mentioned by Raghavan et al.) and 2) a misinterpretation by ADMIXTURE, mistaking the material donated from MA-1 to Amerindians as being admixture in MA-1. It seems that the case is further complicated by the fact that some Amerindians appear fully 'East Eurasian' at k=3, so the Amerindian-like material in MA-1, and to a lesser extent the other HG populations, is misinterpreted as East Eurasian in the same way the same W. Eurasian material in Karitiana is misinterpreted as East Eurasian.

This is logically deducible from the fact that both labs now maintain that all Amrrindians are an admixture of W. Eurasian and E. Eurasian, therefore the ADMIXTURE analysis must be incorrect. Only a sophist would recommend a focus on the ADMIXTURE evidence at the expense of the other analyses. A sophist with yawn-inspiring agenda.

Eastern View said...
This comment has been removed by the author.
Eastern View said...
This comment has been removed by the author.
Hamar Fox said...

Eastern View,

Hamar, again, if you go Reich's interpretation, Paleolithic and Mesolithic Europeans were more related to Amerinds and East Asians.

As I've already said, in this and the other thread, this is because of two things: 1) Mesolithic and Paleolithic W. Eurasians carried along the common 'general Eurasian' line after the basal Eurasian population split away. Therefore, they are closer than the third W. Eurasian population with basal admixture. 2) Amerindians received admixture from ANE; therefore that population is especially close. There's no conundrum here.

Some modern Europeans (say Russians) are mostly Derived Eurasian while other Europeans (say Italians) are 40% Basal Eurasian

This is an incorrect figure, since Italians are not 100% Near Eastern/Neolithic (not even Sardinians).

and what about some SW Asians? Are they 60% or more basal Eurasian?

Based on what, other than whatever scheme you wish to create for the purpose of convincing yourself that your favourite populations are this and your least favourite populations are that? I've seen it all before and it bores me. It's the internet afterall. Stick to the facts.

So if we follow this logic, there can be no West Eurasians but rather 2 unrelated populations and admixtures of the 2.

See above. Also see the actual figure. The relationship is laid bare for us by the authors, so the additional filter of your interpretive distortions is unnecessary.

That's the major issue, which has shifted, while you are still on some issue that has become insignificant.

It hasn't shifted. You explicitly mentioned today (or last night, ICR) that all unadmixed Eurasians were either basal Eurasians or East Eurasians. Maybe you've abandoned that point (as is understandable), but how am I to know?

Also, East Eurasians weren't studied in detail in this paper, so they may well, and likely do, have some form of 'basal Eurasian' affinity also, maybe from a different basal population, maybe from the same. In fact, all a population needs to be admixed on some level are two diverse haplogroups at reasonable frequencies.

Kurti said...

@MfA

Thanks for the results. I knew it from the beginning. It was obvious that Caucasus and Gedrosia were eaten up by the green and dark blue components labeled as "South Asian" and "European" in the Mal'ta paper.

So a major part of this R* individuals ancestry is actually Caucasus-Gedrosia.

This is another sign for my theory that North European and Caucasus-Gedrosia had one origin somewhere between West and Central Asia.

And again, it is no surprise that this R* individual showed some traces of East Eurasian "admixture" (there is a reason why I put this in quotation marks because it is not really admixture but part of an older component).
The closer you get to yDNA P* the more likely the autosomal dna will be West and East Eurasian alike.
R1*/R2* were likely already pred. Caucasus-Gedrosia and North Euro.

And also I highly doubt that the original Q* carriers were predominantly East Eurasian. I assume that the first Q* individuals were like Amerindians and likely even more West Eurasian "admixed" than modern Q* carriers.

Eastern View said...
This comment has been removed by the author.
Hamar Fox said...

Eastern View,

Sigh.. Harma, I'm simply stating the tree. You keep assuming West Eurasian=modern European in the tree when modern European in fact is the terminal node at the very bottom, a mixture of Basal Eurasian and Derived Eurasian.

Nope, you've simply changed your position to a more reasonable one and are now trollishly acting as though I were arguing against this position. Did you or did you not previous call what you now call 'derived Eurasian' 'East Eurasian'?

A clue: the correct answer is yes.

You also made this statement very recently:

So if we follow this logic, there can be no West Eurasians but rather 2 unrelated populations and admixtures of the 2.

How are they 'unrelated' if their ancestors had already passed through the point they eventually re-pooled their genes with? They may have undergone a great deal of evolution between splitting and admixing, but unrelated?

And, of course, deep splits of the same nature exists also in East Eurasia, as I've already noted.

Obviously, diverse lineages re-pooled at various points in history to create modern populations, and this study gives us a great example of reconciliation between autosomal genetics and haplogroup phylogeny. Don't pretend I was arguing against this truism because I disagreed with unrelated pieces of your trolling.

Is George Bush more related to Barack Obama (a halfie) or is he more related to Fancois Holland, president of France?

A great philosophical question, eloquently expressed.

Are ancient Europeans more related to East Asians or are they more related to modern Europeans, who are 40% Basal Eurasian?

Another inaccurate figure, and another question I've already adequately responded to.

I'm just having fun because this is where "logic" has led to.

Well, just keep on living life to the fullest.

Hamar Fox said...

Eastern View,

Sigh.. Harma, I'm simply stating the tree. You keep assuming West Eurasian=modern European in the tree when modern European in fact is the terminal node at the very bottom, a mixture of Basal Eurasian and Derived Eurasian.

So I'm going to take a serious look at what you seem to be claiming: that if basal Eurasian admixture had never occurred in any Eurasian population, then there would be no East vs West Eurasian division, and it's precisely this basal admixture that created the divergence we see in modern Eurasians, and that it only seems not to be so on the plane of ancient DNA by mistakenly judging the (majority) ancestors of modern 'West Eurasians' as West Eurasian simply because of who their descendants are. Is this seriously your position?

Lithuanians only have around 15% basal admixture, yet they cluster right in the zone with your supposedly 60% basal SW Asians in global PCAs. Account for the recent minor East Asian in Lithuanians and SSA in SW Asians, and they'll be even closer. 'Basal' proportions in general correspond only weakly with an East Eurasian-pull in populations with differing basal proportions but no actual Eastern admixture.

As I say, a weak association is clearly there, but what you're proposing is unrealistic and better explained by one population splitting off, another one splitting off a bit later, pooling at some point, while another population split off somewhere in the middle and went its own way relatively undisturbed, and becoming much more divergent and dwarfing the differences between the 'basal' and 'W. Eurasian' populations, thus making the differences between the latter two fairly non-crucial on a global scale.

My take. Now let's make peace; it's Christmas :P

Andrew Lancaster said...

Wondering if pigmentation was selected for in some way linked to farm-food. Maybe something lacking in it?

Gihanga Rwanda said...

So essentially Europeans are predominantly western Eurasian (indigenous European + "Near Eastern"), in addition to varying degrees of "northern Eurasian" and "basel Eurasian" admixture. I am particularly interested in the latter; the only exact populations who fit the definition of "basel to derived Eurasians" are non-hunter-gatherer populations in Africa (i.e. West African farmers, and Nilotic and NE African pastoralists). Nevertheless, something tells me that the contribution of this "basel Eurasian" ancestry into Europe and western Eurasia in general is being overestimated.

Is the relationship between western Eurasian and northern Eurasian similar to the relationship between eastern Eurasian (i.e. "Mongoloid") and southern Eurasian (ex. ASI, Andamanese, indigenous SE Asians, Papuans and Melanesians, indigenous Australians and Tasmanians)? Correct me if I am wrong.

mooreisbetter said...

Maybe I can shift gears a bit and ask some questions to the experts and fellow commenters out there.

This study focused on mtDNA, right?

Because it seems so hard to reconcile with Y-DNA haplogroups.

For example, Sardinians are most I-M26. And they cluster almost exactly with Early European Farmers. Does this mean it is time to rethink the notion that Hg J and E brought farming to Europe? Was it perhaps I-M26, traveling along the Anatolia (I2*) to the Balkans (I-P37.2) to Southern Europe?

Also, all of the heavy R-M269 (R1b) populations on the graph (Scottish, Basque, etc.) are all clustered over towards the early West European Hunter Gatherers. Time to go back to the original theory that R1b is Paleolithic?

After all, the other pole, Ancient North Eurasians DOES match a pretty non-controversial theory, that Hg N, common in Finns and other north Uralic speakers, is Eurasian.

Chad Rohlfsen said...

So, where does pale skin come into play? Is it from Indo-Europeans, from say the Black Sea (or more east; or does it arise from the drastic change in diet, leading to a lighter complexion to get vitamin D from the sun? Combination of both, or am I way off here? I'm just wondering if people with some farming Near East ancestry were lighter than Northern Europeans, perhaps diet is the cause for it. How do Northern Europeans all of the sudden get lighter than farmers?
Sorry for rambling on. Please weigh in; all of you. This is very interesting.

German Dziebel said...

@Hamar Fox

"a misinterpretation by ADMIXTURE, mistaking the material donated from MA-1 to Amerindians as being admixture in MA-1. It seems that the case is further complicated by the fact that some Amerindians appear fully 'East Eurasian' at k=3, so the Amerindian-like material in MA-1, and to a lesser extent the other HG populations, is misinterpreted as East Eurasian in the same way the same W. Eurasian material in Karitiana is misinterpreted as East Eurasian. "

The Amerindian component in Mal'ta is the Amerindian component. No mistake here. Not the other way around. (Not sure why keep arguing against all the evidence.) I agree that there's a misinterpretation going on here, though.

And BTW, what's this "basal Eurasian" component? What's the best modern representative thereof?Lazaridis et al. 2013 just made it up, I think, to make their biases come true again.

Chad Rohlfsen said...

Sorry to double post, but I have another question about the data. One graph shows the basal as separate from the WHG and ANE. Is the basal group possibly the back-flow into the Near East; and the WHG and ANE then moving north from say, somewhere in the sub-continent and around to their respective locations? Thanks in advance for your input!

terryt said...

"It seems that the case is further complicated by the fact that some Amerindians appear fully 'East Eurasian' at k=3, so the Amerindian-like material in MA-1, and to a lesser extent the other HG populations, is misinterpreted as East Eurasian in the same way the same W. Eurasian material in Karitiana is misinterpreted as East Eurasian".

I feel sure this is the aspect that German has such trouble getting his head around.

"the fact that both labs now maintain that all Amrrindians are an admixture of W. Eurasian and E. Eurasian"

To me that is so obvious I find it incredible that some still don't accept it.

"if you go Reich's interpretation, Paleolithic and Mesolithic Europeans were more related to Amerinds and East Asians".

No, to the non-East Asian element of Amerinds. Any other interpretation suggests a wish to fit the data to a pre-existing theory rather than taking an objective look at the data.

"all a population needs to be admixed on some level are two diverse haplogroups at reasonable frequencies".

Ahh. But German and Rokus don't agree with the widely accepted phylogenies, so you're wasting your time coming from this perspective.

"This is another sign for my theory that North European and Caucasus-Gedrosia had one origin somewhere between West and Central Asia".

I have come to think there is no convincing alternative explanation. And so it follows that:

"I highly doubt that the original Q* carriers were predominantly East Eurasian. I assume that the first Q* individuals were like Amerindians and likely even more West Eurasian 'admixed' than modern Q* carriers".

Possibly even 'pure' West Eurasian.

"Obviously, diverse lineages re-pooled at various points in history to create modern populations"

Yes. That is why I've just put pure in inverted commas.

Slumbery said...

I am possibly reading too much into a single PCA chart, but it tells a possible story about Bronze Age (or late Neolithic, depending on the region) migrations.

Basically the entire modern population of North and Central Europe is significantly in the ANE direction compared to both WE and NE HGs and first farmers. So a significant part of the ANE admixture happened after the time represented by these samples.

This does not mean this element was not in Europe much earlier though. The entire Eastern half of Europe is a genetical terra incognita for the Mezolithic/Neolithic.

Matt said...

Eastern View, you could certainly describe the split as Siberians, Amerinds and East Asians being "pure derived Eurasians", but it would be remiss to note in this context that Onge are also pure derived Eurasians and that Papuans are pure derived Eurasians plus 5-6% Denisovan.

Whatever the characteristics of "pure derived Eurasians" they are probably closer to Onge / Papuan / East Asian composite (probably particularly Onge or Papuan) than any one population of these.

Eastern View - Some modern Europeans (say Russians) are mostly Derived Eurasian while other Europeans (say Italians) are 40% Basal Eurasian, and what about some SW Asians? Are they 60% or more basal Eurasian?

If you use their estimate that the only basal Eurasian ancestry in West Eurasians is via 44% of Early European Farmers, all Europeans are over 65% "derived Eurasian" (and under "35% basal Eurasian") with the minimum in Greeks up to a maximum of 85% DE in Estonians.

Hamar Fox does seem to very slightly be misreporting the model here - the West Eurasian component in this analysis isn't modeled as receiving any input from basal Eurasian, but as being a) a component that shares about the same amount of drift with East Eurasians (Onge, etc) as Ancient North Eurasians does, b) which contributed to European Hunter Gatherer Ancestry and c) which contributed to Early European Farmers, along with basal Eurasian, in around 56:44 proportions (i.e. Early European Farmers are slightly more "West Eurasian" than "basal Eurasian").

Are ancient Europeans more related to East Asians or are they more related to modern Europeans, who are 40% Basal Eurasian?

Modern Europeans are do not fit as 40% basal Eurasian, but around 25% on average.

Also although the West Eurasian and Ancient North Eurasian components (and thus ancient Europeans) share more drift with the East Eurasian group than basal Eurasian, they are still intermediate between BE and EE (perhaps closer to BE).

So for both these reasons, ancient Europeans are more related to modern Europeans than ancient Europeans are to East Asians.

Hamar Fox said...

Matt,

Hamar Fox does seem to very slightly be misreporting the model here - the West Eurasian component in this analysis isn't modeled as receiving any input from basal Eurasian, but as being a) a component that shares about the same amount of drift with East Eurasians (Onge, etc) as Ancient North Eurasians does, b) which contributed to European Hunter Gatherer Ancestry and c) which contributed to Early European Farmers, along with basal Eurasian, in around 56:44 proportions (i.e. Early European Farmers are slightly more "West Eurasian" than "basal Eurasian").

I completely agree with you. I was about to ask where you thought I'd summed it up differently, but I realise I've been using the term West Eurasian in the sense I normally use it (to refer to non-Eastern Eurasians), and not in the way it was used in this paper (i.e. only one of the two components that I consider 'W. Eurasian'), so I apologise for any confusion.

Lank said...

Asking whether ancient Europeans were closer to East Eurasians or modern Europeans, is similar to asking whether Mal'ta should be considered closest to West Eurasians or Amerindians. Mal'ta forms a common branch with West Eurasians, but has the highest overall genetic similarity to modern Amerindians, because their genetic relationship is more recent. Yet, Ancient North Eurasians contributed just a minority of the ancestry in modern Amerindians. It looks like there's a similar story going on with ancient Europeans.

Rokus said...

And BTW, what's this "basal Eurasian" component? What's the best modern representative thereof?Lazaridis et al. 2013 just made it up, I think, to make their biases come true again.
"Basal" probably means that "(Oceanians, East Asians, Native Americans, and Onge, indigenous Andaman islanders) are genetically closer to ancient Eurasian hunter-gatherers (Loschbour, Motala12 and MA1) than to Stuttgart"
Instead:
"In the Near East, no population has its lowest f3-statistic with Loschbour or Motala12, but all have their lowest f3-statistic with Stuttgart (Table 1), suggesting that most of the ancestry of this sample may be directly inherited from populations of the ancient Near East, while modern Near Easterners have additional influences related to Africa, North Eurasia, or South Asia (Table 1)."
Unfortunately, without a time-line for these African/North Eurasian/ South Asiatic influences in the Near East, this does not suffice to claim that "basal" was not already indigenous in Europe before the Neolithic.

By the way, a welcome study on the northern European variety. Already long before Khrunin et al. (2013) found yet another pole of genetic diversity in northern Russian populations, I objected against that single pooled northern European component referred to in virtually all admixture analyses.

Truthteller said...

Hamarfox, the Northern Eurasian ancestry is East Eurasian. Here is why. If you replace Karitiana(which is used to calculate North Eurasian ancestry) with Dai and calculate admixture in Russians, the difference is only 2, that is 22% vs 20%.

Eastern View said...

The only reason I was willing to play along with the "logics" is to highlight the point that "Basal Eurasian" is, so far, a fictitious dummy population that is necessary for Mal'ta to be a pure population in Reich's tree. It is as fictitious as "Population X" in Raghavan's tree. Reich has decided that Raghavan's Mal'ta "discovery" is a cool idea to "further discover" on, so he is positing ANE, which makes it necessary to posit another Basal Eurasian.

The mental gymnastics that goes both on here and in the tree of Reich is incredible. There are so many logical fallacies and then dislogic to further support that that I don't even wanna point to all of it.

In reality, there is no evidence of Basal Eurasian, not in the archaeology, nor the ancient DNA, nor the uni-parental markers, nor the archaeology. The Gravettian, the last Paleolithic culture of Europe, seems to have its kick-start in the Levant.

Eastern View said...

Notice how also in ADMIXTURE the Mesolithic European samples and Mal'ta behave very differently, and therefore are not the results of the same process. Mesolithics stay intact and become more and more solid as K increases. Mal'ta, on the other hand, breaks down into several components. Mal'ta is an admixed individual in the eyes of ADMIXTURE while the Mesolithic Europeans at K20 is a panmictic population, atleast.

Lank said...

Regarding the "basal Eurasians", the current model puts them in a basal position relative to modern Eurasians (other than contributing to modern West Eurasians), but they share a common branch with Eurasian group relative to the Mbuti Africans. We already know of one likely group that would fit this description; the people responsible for the spread of Y-DNA E lineages from NE Africa into the Middle East.

The authors mention this possibility, but also acknowledge the possibility that "basal Eurasian" traces ancestry from early Eurasian anatomically modern humans, perhaps people related to the early AMH found in the Levant. But it seems unlikely that this population would fail to influence European populations until the Neolithic, while their impact just happens to correspond to the timing of Y-DNA E dispersals (possible linked to the African influence in the Levantine Natufian, just prior to the Neolithic). And what is less commonly known is the presence of African mtDNA in West Eurasia, apparently since Neolithic times. mtDNA L2a1 has been found in Neolithic Syria, and this lineage is also found as a distinct sublineage in modern Europeans.

Eastern View said...

I think a general point I can make is that Hamar and the others who are arguing against me is arguing that ADMIXTURE is random while Lazaridis (sorry not Reich) is not random.

Well, the actual truth is that ADMIXTURE is not random, otherwise it wouldn't be predicting population events that happened over 50K ago (split of Africans from Eurasians and split of Khoisan from non-Khoisan Africans) accurately. It's actually Lazaridis that is making an arbitrary decision for his tree so Mal'ta can be a pure "ANE".

About Time said...

Vitamin D deficiency from eating low meat diet in cloudy climate would be a possibility. Hard to imagine why mesos would get light skinned if pale is like Mal'ta boy were dark. Neolithic makes a lot more sense in terms of environmental selection. Maybe amplified by sexual selection at some point.

About Time said...

Mal'ta boy was R. Ancient north Eurasian peaks in Scottish and Estonians if I remember. That means R1b is conceivably ANE at this point.

German Dziebel said...

@Rokus

""Basal" probably means that "(Oceanians, East Asians, Native Americans, and Onge, indigenous Andaman islanders) are genetically closer to ancient Eurasian hunter-gatherers (Loschbour, Motala12 and MA1) than to Stuttgart"

If this is so, then Amerindians are "basal Eurasians" because they are the closest to Loschbour, Motala, Mal'ta and Stuttgart than any other modern population.

@terryT

"I feel sure this is the aspect that German has such trouble getting his head around."

What Hamar Fox wrote makes no sense. ADMIXTURE is pretty straightforward in that there's AMERINDIAN 1 (BLUE) shared with all non-Africans and some Africans, AMERINDIAN 2 (YELLOW) shared with all East Asians and some non-East Asians and AMERINDIAN 3 (GREEN) shared with some Siberians and some West Eurasians. For comparison the African components remain largely African-specific, hence there was very little out-of-Africa contribution to the non-African gene pool and it definitely didn't go anywhere close to America.

German Dziebel said...

@Lank

"Mal'ta forms a common branch with West Eurasians, but has the highest overall genetic similarity to modern Amerindians, because their genetic relationship is more recent. "

No. The overall greater genetic similarity between Mal'ta and Amerindians is due to shared common descent diluted by substantial West Eurasian admixture. That's why in PCAs Mal'ta is drawn toward West Eurasians.

Kristiina said...

When I look at that successful model and try to fit it to the existing yDNA tree, I get the idea that Onge is yDNA D and other "Eastern non African" that contributed to Karitiana is yDNA C3. Ancient North Eurasian and West Eurasian go back to KIJ, and ANE is yDNA P and West Eurasian is yDNA I. Basal Eurasian could then be yDNA G which is important in Near Eastern farmers and Mediterranean people.

Nick Patterson (Broad) said...


About time asked for some
f_3 stats. As a Christmas present:

## Sources Target f_3 Z
LBK Onge Vishwabrahmin 0.000734 0.641
LBK Onge Gujarati1 -0.006114 -4.582
LBK Onge Gujarati2 -0.006442 -4.955
LBK Onge Gujarati3 -0.001871 -1.325
LBK Onge Gujarati4 0.002510 1.814
MA1 Onge Vishwabrahmin 0.005237 4.136
MA1 Onge Gujarati1 0.002404 1.641
MA1 Onge Gujarati2 0.003153 2.124
MA1 Onge Gujarati3 0.004297 2.923
MA1 Onge Gujarati4 0.009156 6.084
LBK She Uzbek -0.024589 -26.829
MA1 She Uzbek -0.013565 -13.372

(There are 4 Gujarati groups in the Human Origins data, genetically distinct).

Implications for ancient DNA to the history of India and Asia are very interesting but
will need much more work.

matt said...

Have these authors been able to report any ydna short tandem repeats which might give clues as well as the SNPs they were able to recover. For that matter what is the success rate at recovering STRs?

Hamar Fox said...

Truthteller,

Hamarfox, the Northern Eurasian ancestry is East Eurasian. Here is why. If you replace Karitiana(which is used to calculate North Eurasian ancestry) with Dai and calculate admixture in Russians, the difference is only 2, that is 22% vs 20%.

*bangs head repeatedly on desk*

This is so far from what we've learned in the past few months I'm beginning to think if people aren't getting it by now I may as well give up.

Btw, even though your figures are incorrect (or at least come nowhere near any figures I've seen), it still wouldn't matter too much, since Russians obviously have real East Asian admixture unrelated to their ANE ancestry.

Eastern View,

I think a general point I can make is that Hamar and the others who are arguing against me is arguing that ADMIXTURE is random while Lazaridis (sorry not Reich) is not random.

Well, the actual truth is that ADMIXTURE is not random, otherwise it wouldn't be predicting population events that happened over 50K ago (split of Africans from Eurasians and split of Khoisan from non-Khoisan Africans) accurately. It's actually Lazaridis that is making an arbitrary decision for his tree so Mal'ta can be a pure "ANE".


I'm not saying it's random. I'm saying it requires a sober-headed interpretation. ADMIXTURE is a fine form of analysis for relatively unadmixed groups, especially at low k-levels, but it frequently errs when it takes admixed populations to be unmixed.

Lazaridis' ADMIXTURE analysis at very low k-levels shows W. Europeans to be basically all blue (with minuscule noise) and Karitiana and some other Amerindians to be basically all yellow with minuscule noise. Yet we know this can't be the case. Therefore, the ADMIXTURE analysis is not ideal. Otherwise, is it your position that W. Eurasian hunter-gathers have real East Eurasian admixture, but their W. European descendants don't, even when these HG populations constitute at least half, and often a fair bit more of their ancestry? If you're an ADMIXTURE literalist, then that's the only conclusion you can draw. My and Terry's point is that it's an interpretive error on ADMIXTURE's part and not an act of real-world magic.

About Time said...

Wow, thank you! So it looks like LBK (but possibly not Mal'ta) is a source for some Gujaratis but not Vishwabrahmins.

LBK and Mal'ta are both definitely sources for Uzbeks. Am I reading this output correctly?

If so, LBK were part of a very, very important movement in prehistory. Just, wow. Thank you.

Wausar said...

Gihanga Rwanda, Eastern View,

The "basal Eurasians" are probably those who continued to receive African gene flow, splitting them off from of the rest of the Eurasians. The authors hint at this possibility, citing "African-related tools made by modern humans in Arabia" and "the geographic opportunity for continuous gene flow between the Near East and Africa."

As for the skeletal record, C. Loring Brace found that the Natufians that ushered in the Neolithic had a "clear link" to Sub-Saharan Africa. Lawrence Angel wrote, "one can identify Negroid (Ethiopic or Bushmanoid) traits of nose and prognathism appearing in Natufian latest hunters (McCown, 1939) and in Anatolian and Macedonian first farmers (Angel,1972), probably from Nubia (Anderson,1969)."

The presence of E1b1b and Afro-Asiatic languages in SW Asia is further evidence of this migration.

German Dziebel said...

@Hamar Fox

" My and Terry's point is that it's an interpretive error on ADMIXTURE's part and not an act of real-world magic. "

Eastern View is absolutely correct in his critique of Lazaridis and Raghavan. And he and I come from completely different angles and don't share the same interpretive framework. You and Terry just need to show some support for your conspiracy theory that ADMIXTURE flips the components around and somehow the rest of the evidence "knows" that.

Grognard said...

But Mal'ta has to be a mix, and a dead end since there's no australoid running around.

I mean the guy has dark skin and freckles! you can't unscramble an egg and keep only parts of it around!

To me this implies waves radiating out of india, and being partially or wholly replaced over time. That is the big interest of Mal'ta, aside from the fact he has some R lineage in NW siberia that far back which seems to dbunk the idea it's recently invasive.

Hamar Fox said...

German Dziebel,

Eastern View is absolutely correct in his critique of Lazaridis and Raghavan.

My thoughts were cobbled together inelegantly in my last post, and what I intended to be two separate points sort of blurred into one. Keeping strictly to EV's arguments, since you acknowledge the difference of perspective between him and you, but since you still praised him, here I'll lay down some observations as to why such praise is unwarranted:

1. We know that there is a relationship between Amerindians and W. Eurasians. EV doesn't seem to contend that either. ADMIXTURE (via Lazaridis and Raghavan), even at the most basic levels, does not show all W. Eurasians to have E. Eurasian admixture or all Amerindians to have W. Eurasian admixture. It suggests that both Amerindians and W. Eurasians can both easily lack any connection to the other on any level. But this can't be true. At least one of the populations should show input from the other, since at least one has to be admixed or there can be no connection (again, not from your perspective, but from EV's). And this connection is well demonstrated through other analyses in both papers.

2. The descendants of the hunter-gather populations, which all show East Eurasian admixture, do not all have East Eurasian admixture at the same k-levels it shows in their ancestors, even when in some cases these ancestors constitute the majority of their ancestry. This again suggests ADMIXTURE's assignment of unreal admixture in populations it can 'get away with', so to speak (based on low representation, fewer SNPs etc.), but when ADMIXTURE can't 'get away with it', as with their much better represented descendants, the illusion/misinterpretation disappears.

And he and I come from completely different angles and don't share the same interpretive framework. You and Terry just need to show some support for your conspiracy theory that ADMIXTURE flips the components around and somehow the rest of the evidence "knows" that.

The two recent papers from two eminent labs are 'some support' for our 'conspiracy theory'. If MA-1 is an ancestor (or something very close) to Amerindians, and he is, according to ADMIXTURE, 2/3rds W. Eurasian, then it's impossible without some kind of unprecedented 'racial selection/deselection' for his descendants to be 0% W. Eurasian. And, swapping it around, if he or any of the other HG are ancestral to W. Eurasians, it's equally impossible/ridiculously unlikely for that specific East Eurasian material in them to be deselected. Yet, still, that's what ADMIXTURE suggests: that MA-1 isn't ancestral to Amerindians and also that the hunter-gatherers aren't ancestral to W. Europeans (even though, of course, they actually are). It's certainly not a whacky theory Terry and I have cooked up. It's a logical observation.

Truthteller said...

@Hamarfox,
It doesn't matter which European population you choose. Replacing Karitiana with Dai still produces similar results. I only used Russian as an example.

terryt said...

@ Hamar Fox:

"This is so far from what we've learned in the past few months I'm beginning to think if people aren't getting it by now I may as well give up.

I have given up. Some people seem so committed to some strange belief that no matter how much evidence appears that contradicts their position they just make up some reason why it might not be so. And they then convince themselves it is not so. However I will add my support this statement from Hamar Fox:

"The two recent papers from two eminent labs are 'some support' for our 'conspiracy theory'. If MA-1 is an ancestor (or something very close) to Amerindians, and he is, according to ADMIXTURE, 2/3rds W. Eurasian, then it's impossible without some kind of unprecedented 'racial selection/deselection' for his descendants to be 0% W. Eurasian. And, swapping it around, if he or any of the other HG are ancestral to W. Eurasians, it's equally impossible/ridiculously unlikely for that specific East Eurasian material in them to be deselected. Yet, still, that's what ADMIXTURE suggests: that MA-1 isn't ancestral to Amerindians and also that the hunter-gatherers aren't ancestral to W. Europeans (even though, of course, they actually are). It's certainly not a whacky theory Terry and I have cooked up. It's a logical observation".

And we have haplogroup and EDAR370A evidence to back the whacky theory up even more.

German Dziebel said...

@Hamar Fox

Thanks for expounding on this.

"The two recent papers from two eminent labs are 'some support' for our 'conspiracy theory'. "

The two ADMIXTURE runs you're criticizing come from the same two Great Labs. So, you decided to "fix" the data to make it more consistent with your beliefs and since you share them with Lazaridis and Raghavan to help them find a "glitch" in their own data. (smiley face)

"ADMIXTURE (via Lazaridis and Raghavan), even at the most basic levels, does not show all W. Eurasians to have E. Eurasian admixture or all Amerindians to have W. Eurasian admixture."

But ADMIXTURE (via Raghavan and Lazaridis, respectively ) does show an Amerindian component in Mal'ta (ORANGE) and in Mal'ta and European HG (GREEN). At K=3 and K=4, Lazaridis's run shows YELLOW which, again, is East Eurasian/Amerindian, as Eastern View noted. So, ADMIXTURE is perfectly consistent in showing a link between extreme West (Loschbour) and extreme East (Amerindians, with or without East Asians). Modern West Eurasians (French, etc.) have the same GREEN and YELLOW components until K=10. Mordivians, Russians, Finns and Estonians have more of it but not as much as Mal'ta. At K=10 these "eastern" components are replaced by MAGENTA(?), which is West Asian and, quite predictably, it's high in Stuttgart.

"Yet, still, that's what ADMIXTURE suggests: that MA-1 isn't ancestral to Amerindians and also that the hunter-gatherers aren't ancestral to W. Europeans (even though, of course, they actually are). "

See above. May be I'm missing something, but populations such as French have "eastern admixture" at lower K levels (consistently with Mal'ta, Motala and Loschbour) and "west Asian" admixture at higher K level (consistently with Shuttgart). This indicates population replacement in Europe, with agriculturalists "replacing" HG.

Eestern View and I agree that Mal'ta is an admixed population. Yes, ADMIXTURE is very clear about it. But everything else is perfectly consistent with it (with or without a broader out-of-America implication). You yourself admit the presence of "real Papuan admixture" in Mal'ta. What neither Eastern View nor myself buys is "basal Eurasian" which looks like a made up component, as there's no support for it in ADMIXTURE.

My beef with ADMIXTURE a la Lazaridis and Raghavan is that it misinterprets an Amerindian component as first West Eurasian=non-African, then East Asian, then its own, while at all of these base levels Karitiana, Surui and Piapoco stay perfectly unadmixed. So, YELLOW in Loschbour, Motala and Mal'ta at K=3 and K=4 should be GREEN just like K=5 because other data shows that all of them are closer to Karitiana than to any other modern population.

Same with Khosans vs. rest of Africans. Juhoan are first pure Africans and then their own. Yoruba are first a mix of Africans and non-Africans and then pure Africans.

Rosenberg 2002 did the best job so far parsing the base levels out.

Locrian said...

I’d like to come into this debate on the side of Harma Fox and terry_t. It seems to me that the problem here is one that arises often, and has to do with the interpretation of admixture as though it were an indication of causality: that if an individual A shows admixture of two modern populations X and Y that this shows that A is composed of elements that have come from X and Y. But all it really shows is degree of overlap — and degree of overlap does not suggest a temporal priority of X and Y.

If we were doing admixture properly we would take a lot of ancient dna (call the groups X, Y and Z) and then see how a more modern individual might be an admixture of these — then causality would be the right way round. Of course we don’t have enough ancient dna to do that so instead we look at degrees of overlap between an ancient sample with the many samples we have from modern populations. But this has to be interpreted very sensitively — and sometimes that does not occur.

This logical problem seems to foul-up many attempts to trace origins. For example it has mislead many people to think that because R1a and R1b are common in modern European populations that that must mean it is deeply ancestral. We now now - definitively I’d say - that this is not true. Europe was once mostly I, then up to the middle ages it was predominantly G2a, and now it is R1b and a.

Rokus said...

""Basal" probably means that "(Oceanians, East Asians, Native Americans, and Onge, indigenous Andaman islanders) are genetically closer to ancient Eurasian hunter-gatherers (Loschbour, Motala12 and MA1) than to Stuttgart"

If this is so, then Amerindians are "basal Eurasians" because they are the closest to Loschbour, Motala, Mal'ta and Stuttgart than any other modern population.

Instead, Lazaridis et al. figured the Stuttgart specimen contained more of the basal component "because" virtually all the rest of the OOA world populations are more similar one to the other. I agree with you this construct is an abomination, it's just another expression of the belief that strongly divergent populations should fit the baseline in an exclusive Out of Africa scenario. Orwell would call this Newspeak, a phrasing method meant to reduce the capability of our minds to see things in a different way. Even worse, Lazardis et al. actively suggest the "basal" component unifies non-similar populations in a single basal pool that e.g. ties Neolithic basal populations togehter with Near Eastern basal populations - while actually such "basal" populations should be reminiscent of a much older layer.

Hamar Fox said...

Truthteller,

It doesn't matter which European population you choose. Replacing Karitiana with Dai still produces similar results. I only used Russian as an example.

Please read Lazaridis et al. and Raghavan et al.'s recent papers, because what you say here

If you replace Karitiana(which is used to calculate North Eurasian ancestry)

suggests that you haven't and are maintaining some of the older assumptions of the Reich lab. MA-1 and Afontova Gora are known to be the best representatives of Ancient North Eurasian ancestry now, and they were not East Eurasian (the position of both papers). ANE and WE in this latest paper are simply geographical denominations, not genetic ones as such (while in earlier Reich lab papers, this was less clear), and we see in the second (leftmost) figure Dienekes includes above that the ANE and WE components share the same descent entirely.

The phenomenon you describe about any Western population being almost as close to Dai as to Karitiana could result from a few things. But first of all I should add I've never seen an estimate of 20% Dai (or East Asian) admixture in Russians, and I've seen lots of estimates of admixture in Russians. Second of all, I've seen estimates of 'East Eurasian' admixture (relative to Sardinians) more than double when Karitiana are used as a representative of East Eurasians instead of Dai.

Both recent papers find that Amerindians are an admixed population (of ANE and East Asian), so it's no surprise that estimates of 'East Eurasian' ancestry are higher when Karitiana are used, because in addition to any real admixture (represented by Dai, with exceptions I'll note below), it also reflects back a portion of their own Western ancestry that is present in Karitiana.

As for why some Western populations can still show some affinity (usually relative to Sardinians) to East Eurasians without admixture, it's explained in this current paper, and clear from the same figure I referred to, that this reflects additional shared drift between East Eurasians and West Eurasians (i.e. the ANE and WE components in Lazaridis et al.) in relation to basal Eurasians, who seemingly admixed disproportionately into modern Western populations. There's currently no information on who these basal Eurasians were, but they had an anti-East Eurasian genetic effect on the populations they admixed with to varying degrees because of this relative lack of shared drift.

Terryt,

And we have haplogroup and EDAR370A evidence to back the whacky theory up even more.

We do. I linked you to a new source in the other thread, but it was buried at the bottom of a post to someone else, so I'm not sure if you saw it:

http://alfred.med.yale.edu/alfred/mvograph.asp?siteuid=SI663326A

Rokus said...

@Wausar,

Thank you for reminding me. The Neolithic "basal component" may have a significant Holocene East African admixture. However, Lazaridis et al. only suggested African admixture in the Near East, not in the European counterpart (EEF) of this presumably Eurasian basal component.

Hamar Fox said...


See above. May be I'm missing something, but populations such as French have "eastern admixture" at lower K levels (consistently with Mal'ta, Motala and Loschbour) and "west Asian" admixture at higher K level (consistently with Shuttgart).

Even in reference to the European HG populations, we see that in many cases (English, French etc.) have less than half the yellow at k=3 than the ancients, even though they're only around 50% Neolithic (49.5 and 55.4%, respectively). If we take into account ANE ancestry (~30% 'East Eurasian' at k=3), French Basques (11.4% ANE) and Pais Vasco (16.3% ANE) should be 3.42% E. Eurasian and 4.89% East Eurasian respectively. Yet they're both 0% (except for one blip in F. Basques).

Hamar Fox said...
This comment has been removed by the author.
Hamar Fox said...

Apologies, I sent the end bit of a really long post and have now lost an incredibly long reply to the remainder of German's points. I'll attempt to recreate it at some point.

About Time said...

What people seem to be missing wrt the eastern % in ANE and WHG can be resolved by looking at the tree model:

After the initial African/Non-african split, Basal Eurasians (Pre-Natufians?)isolated themselves early from all other Non-africans.

The other Non-africans continued to mix and only later split into: A. a branch of Eastern non-Africans ancestral to Onge and Karitiana and B. two branches: B1. ANE and B2. West Eurasians (upstream of WHG and EEF).

So ANE, WGH, and Onge/Karitiana were still mixing when Basal Europeans separated. That's possibly why in admixture they show eastern % that Stuttgart LBK lacks. Not "admixture" per se, but common descent from an as-yet undifferentiated Non-african population.

This also doesn't mean the Basal Eurasians were more African. From the tree diagram at least, it just means they separated from everyone else.

We still don't know where these groups were at the time of this early split. It could have been in a fairly compressed area, with later migrations radiating out.

Basal Eurasians could have even been in border lands between the Africans and Non-africans while maintaining endogamy. Or some Basals stayed endogamous, and others mixed with West Eurasians according to tree model. Some people mentioned the Basals might be pre-Nubians/Natufians.

Hamar Fox said...

About Time,

What people seem to be missing wrt the eastern % in ANE and WHG can be resolved by looking at the tree model:

After the initial African/Non-african split, Basal Eurasians (Pre-Natufians?)isolated themselves early from all other Non-africans.

The other Non-africans continued to mix and only later split into: A. a branch of Eastern non-Africans ancestral to Onge and Karitiana and B. two branches: B1. ANE and B2. West Eurasians (upstream of WHG and EEF).

So ANE, WGH, and Onge/Karitiana were still mixing when Basal Europeans separated. That's possibly why in admixture they show eastern % that Stuttgart LBK lacks. Not "admixture" per se, but common descent from an as-yet undifferentiated Non-african population.


This type of interpretation is what I'd favour if it weren't for a few observations:

1) The Eastern admixture resolves itself in many cases completely into Amerindian at higher k-levels. If it were a general relatedness to East Eurasians, we'd surely see more diversity in the signal. We actually only see a bit of Australasian in MA-1 (which Raghavan et al. suggested may be real) and pretty much nothing but W. Eurasian and Amerindian in the other HGs.

2) We know that Karitiana should appear as admixed at k=3, if we accept the model presented. Therefore, a knock-on effect on other populations is expected, and, as mentioned above, it explains why Amerindian (along with all of what Amerindian entails at k=3 (wrongly)) appears in other populations.

3) The hunter gatherers' East Eurasian proportions are not proportionately inherited in their descendants.

4) Modern populations that differ in EEF-like and HG ancestry (such as Mozabite and Basques) can still score 0% East Eurasian.

I like your theory, and I think it definitely explains the effect we still see in ADMIXTURE runs where Amerindians are expressed as admixtures. For example, at k=3 here (Dodecad):

https://docs.google.com/spreadsheet/ccc?key=0ArJDEoCgzRKedGR2ZWRoQ0VaWTc0dlV1cHh4ZUNJRUE#gid=0

and Harappa, again at k=3:

https://docs.google.com/spreadsheet/ccc?key=0AuW3R0Ys-P4HdHdGXzdLR0prSk1UUWZjNGl5cUEtWlE#gid=0

What you describe explains why, for example, many Western Europeans get ~1% Eastern. But I think there are additional factors going on in both Raghavan et al. and Lazaridis et al.'s analyses. I really wish I'd not deleted by mistake my response to German, since I think I expressed my thoughts better there than elsewhere and now I'll have to do it again.

Truthteller said...

Hamarfox,
The main flaw in these studies based on ancient populations is the lack of samples to begin with. As a result I feel that these new studies aren't, accurate to begin with. We have just one sample tha represents ANE for now, the Mal'ta samples. Not to mention that ADMIXTURE suggests clear South Asian affinity even more so than European in many respects. In fact the South Asian like ancestry peaks in South Indian tribas, although I do believe their runs combined the Gedrosia and South Indian into one.
I have used th admixtools software and I feel that Europeans have an ancestry that is clearly East Eurasianlike. I define East Eurasian as all the non African and non west Eurasian populations like Dai, Han, Onge, Papuan as well as Karitiana. I also include ANE as defined in the mixmapper paper as East Eurasian because replacing Karitiana with Dai, Han and I would bet Onge as well as Papuan would still be able to produce very similar results. Furthermore, the ANE in the Mix Mapper Paper and the ANE in the mal'ta paper may not even be describing the same thing. Also I do not like how far the chart in this current paper is showing ANE to be from East Eurasian. That does not make sense, since like I said even the Dai, Han as well as Onge all are able to produce this East Eurasianlike admixture in Europeans.
For example, the French are 3% ANE relative to Sardinians. In reality the real number will be higher.
We don't have enough ancient samples for this paper to be very accurate. You need at least 10 ANE samples.

About Time said...

Hamar Fox,

Hmm. I am wondering if another factor was at play, like Neanderthal introgression in some of the non-Basal Nonafricans or at a later stage (like Mal'ta).

eurologist said...

From the PC plot I don't see how present Europeans (apart from extreme SE Europe) show much N Eastern admixture. Note that you cannot use PC1, but you first must rotate the plot ~30 degrees clockwise. Then you can see that most extant Europeans, ANE, and EEF all show about the same distance to N Eastern populations (again, with the exception of SE Europeans, as expected).

N and C European are mostly a mix between ANE and European HGs, with just a little S European (=EEF) admixture. I don't see much N Eastern in them from the PC plot.

Scandinavian HGs appear to differ from the W European ones due to a small amount of ANE admixture.

Basque are a little closer to WE HGs.

If indeed EEF did not contain ANE, a possibility is that the Gravettian (ANE) pressure compacted the SE European and N Eastern population, so that at the beginning of the Neolithic, ANE-admixed populations were present just north of the Balkans and Anatolia, and probably also in the Caucasus and the southern Caspian.

In Admixture, the West Asian component keeps growing in Europeans with little or no "Caucasian" with higher K number - until it becomes the majority in EEF, Basques, and Sardinians. This happens in stages when isolates are removed from West Asians (and make their own component), such as Caucasian, Beduins and Mozabite. Those clearly have drifted and/or have admixture that was not present in ancient Europeans and is still not present much in Western and Northern Europeans - especially Sardinian, Basque, and S French. But with it, the poster-boys of that N Eastern magenta W Asian components vanish

As always, quite astonishing how early the specific mtDNA subgroups are (e.g., Loschbour U5b1a with two additional substitutions, and Motala 4 & 5 U5a2d with 1-2 additional substitutions, Motala 9 U5a2 with 6 additional substitutions).

There is a high number of positive HG y-DNA I positives, here - which most of us suspected for a long while.
Loschbour belonged to Y chromosome haplogroup I2a1b*(xI2a1b1, I2a1b2, I2a1b3); Motala12 places in haplogroup I2a1b*(xI2a1b1, I2a1b3).

"In reality, there is no evidence of Basal Eurasian, not in the archaeology, nor the ancient DNA, nor the uni-parental markers, nor the archaeology. The Gravettian, the last Paleolithic culture of Europe, seems to have its kick-start in the Levant."

Eastern View, I agree with you until your last remark. The Gravettian is decidedly northerly, around the northern Pontic. But, to me, it seems clear that one of its effects was a contraction/ compression of existing SE European and Near Asian populations into what is now Greece and the adjacent Anatolia and near east.

Truthteller said...

Anybody notice that Iranians are one of the groups that have a low f3 number. Just think why for a second here. The Iranians have some ASI as well as some East Eurasian Turkic. This causes the Iranians to appear more ANE, which like I said is related to both Han/Siberian and Onge. A lot of the populations( not all) in the ANE list are known to have east Eurasian ancestry.

Hamar Fox said...

German Dziebel,

The two ADMIXTURE runs you're criticizing come from the same two Great Labs. So, you decided to "fix" the data to make it more consistent with your beliefs and since you share them with Lazaridis and Raghavan to help them find a "glitch" in their own data. (smiley face)

Let's not go downhill tonally. While I've been less than pleasant with EV (nor he with me), I've only been courteous with you in our interactions.

I'm not 'fixing' anything. I'm trying to bring to harmony as many pieces of evidence as possible. Where two things are incongruous, I have to make a judgement which thing to sacrifice based on wider evidence. If neither 'incongruent' can be explained away as an error, then the theoretical framework needs to accommodate the seemingly disharmonious facts by coming out of its own 'box' enough so that according to it they no longer contrast. I understand this. But, at present, I don't believe ADMIXTURE's results defy explanation. And this position is supported by the fact that I've seen Amerindians appear admixed at k=3 in other runs, so I know it's far from an impossibility. Whether I'm right ot wrong, my position does not stem from any dishonesty. In fact, it would be more dishonest for me to ignore data that doesn't fit my own explanations than to address and examine it.

So, ADMIXTURE is perfectly consistent in showing a link between extreme West (Loschbour) and extreme East (Amerindians, with or without East Asians).

But my point is that, taken at face value, it shows no relationship between some W. Eurasians (F. Basque, Pais Vasco, S. French, Sardinian -- all blue) and some Amerindians (Karitiana, Surui etc. -- all yellow), yet we know that a special relationship exists. So why is ADMIXTURE not picking anything up? Because it's less than ideal for solving this problem.

Modern West Eurasians (French, etc.) have the same GREEN and YELLOW components until K=10. Mordivians, Russians, Finns and Estonians have more of it but not as much as Mal'ta. At K=10 these "eastern" components are replaced by MAGENTA(?), which is West Asian and, quite predictably, it's high in Stuttgart.

The Amerindian green component at higher levels is slightly more expansive than k=3 E. Eurasian yellow (another inconsistency), so some populations that have no yellow at k=3 can still have Amerindian green in very small doses (e.g. Basques, but not Sardinians). Populations that have a bit of yellow at k=3 tend to have a bit more Amerindian green later on in step with this trend.

But when judging the patterns in extreme detail, some things need to be borne in mind:

A complication is that some yellow at k=3 is an effect of the misidentification of Amerindian admixture in those populations and all that the fact entails (based on the mentioned mistake of finding them 100% East Eurasian, as I believe is mostly/wholly the case with the HGs), while in other populations it is real E. Eurasian admixture + the above phenomenon.

TBC

Hamar Fox said...

An even bigger complication is that, even if it's real, it doesn't necessarily stay yellow as opposed to green at higher k-levels. We see this in Hungarians and Lithuanians, for example. Some of their green should be yellow, since we know these populations have minor (<3%) E. Asian admixture, haplogroup N, and EDAR, but their k=3 yellow turns almost entirely to green eventually. I can best explain this by appealing to the fact that the 'Amerindian' component comprises both Eastern and Western signatures, so if some of the Eastern stuff in those populations resembles the Eastern stuff already contained in the Amerindian component, then under the principle of parsimony it may prefer to group both elements under one banner: Amerindian. We see ADMIXTURE fails to do this completely with populations that have too much E. Asian for it to explain away as 'Amerindian', so in Finns and Russians, for example, we see their k=3 East Eurasian resolve itself into both Amerindian and other-East-Asian, but still likely in slightly confused proportions because of what I explained.

It's extremely tricky business. And, as About Time, rightly mentions, there are even more factors, factors whose relevance to any given calculator need also to be weighed up and assigned explanitory weight individually.

You yourself admit the presence of "real Papuan admixture" in Mal'ta. What neither Eastern View nor myself buys is "basal Eurasian" which looks like a made up component, as there's no support for it in ADMIXTURE.

I'm not sure where I stand on the Papuan connection. Raghavan et al. touch on it briefly. It maybe hints at contact with the ancestors of, say, the Jomon. Papuans are the only East Eurasian population MA-1 had any affinity to (though didn't share much drift with them, apparently), and, AFAIR, had about 6% admixture with them. It may turn out to be nothing. AG seems to have had none.

Simon_W said...


What a wonderful paper! A christmas present, I'm tempted to say.

Looks like I was wrong: The Scottish are among the European peoples with the strongest ANE admixture, and yet they don't seem anyhow closer to the Near Eastern cline than the others; to the contrary, they are rather distant from it. So I guess their ANE admixture came from the northern Pontic area, and their R1b probably as well.

Interestingly, in the admixture analysis some Basques do have a slight presence of the „Kalash“ component (which traces the ANE admixture in Europe), and according to the paper they probably have even more of the ANE admixture, but not nearly as much as the Scots. The Sardinians clearly have the lowest ANE admixture, and they don't have much R1b either.

But it's clear that Near Eastern peoples contributed to the spread of the ANE admixture as well – after they had received it from highland peoples near the Caucasus who have a lot of this admixture. Y-haplogroups J1 and J2, stemming from the northern Near East, were probably involved in the dissemination of this admixture. In the PCA there is a cline leading from Bergamo over Tuscany to Sicily and Greece /Albania (Bulgaria being somewhat pulled away from this cline by Slavic and other admixture) and then on to the Ashkenazi Jews, to Cyprus and finally to the Druzes.

A little bit of this influence is probably present in Spain, France, Hungary and Croatia as well, and there we also find some J2. When analyzing my southern German + Swiss grandparents with the Dodecad K12b oracle (especially with the 4 population version at Gedmatch) I noted some influence that's similar to Ashkenazi Jews and Lebanese, so that secondary Near Eastern influence appears to be relatively strong there too. This may be related to the old non-IE Raetic language which was a relative of Lemnian.

Presumably this southern, J2-related expansion of the ANE admixture wasn't related to the spread of Indo-European, because exactly along the above mentioned cline we find in pre-Greek times on Cyprus the non-IE Eteocypriot language, on Crete non-IE Minoan or Eteocretan, on Lemnos non-IE Lemnian and from there an axis goes to non-IE Raetic in the eastern Alps.

German Dziebel said...

@Hamar Fox

"Let's not go downhill tonally. While I've been less than pleasant with EV (nor he with me), I've only been courteous with you in our interactions."

It was a good-natured joke.

"But my point is that, taken at face value, it shows no relationship between some W. Eurasians (F. Basque, Pais Vasco, S. French, Sardinian -- all blue) and some Amerindians (Karitiana, Surui etc. -- all yellow), yet we know that a special relationship exists. So why is ADMIXTURE not picking anything up? Because it's less than ideal for solving this problem."

It does pick it up at K=3 and K=4 and K=8 and K=9. It does lose it at K=5, K=6, K=7, true. But Karitiana, Surui and Piapoco stay "unadmixed" all the way through. There are many reasons why there's some inconsistency in Europe. For instance, ADMIXTURE may misinterpret the special connection between Europe and the "east" in those middle levels as generic non-Africanness. Generally ADMIXTURE is pretty accurate in documenting the "eastern admixture" across Europe considering the granularity of the data.

"I can best explain this by appealing to the fact that the 'Amerindian' component comprises both Eastern and Western signatures, so if some of the Eastern stuff in those populations resembles the Eastern stuff already contained in the Amerindian component, then under the principle of parsimony it may prefer to group both elements under one banner: Amerindian."

Bingo! Just like with the generic "non-Africanness" above ADMIXTURE sometimes falls back on an "older" genetic component among the two options, unless, like you said, the younger component expands, then it keeps both.

Eastern View said...
This comment has been removed by the author.
Eastern View said...
This comment has been removed by the author.
terryt said...

"I linked you to a new source in the other thread, but it was buried at the bottom of a post to someone else, so I'm not sure if you saw it"

I did see it at the time, thanks.

"I'm not sure where I stand on the Papuan connection"

I'm reasonably convinced that Y-DNA MNOPS and mt-DNA R both originated in Sundaland and expanded from there. I think we can assume the ancient Sundaland population was very similar to what we find in Australia/New Guinea today, taking into consideration later admixture in those regions.

"I’d like to come into this debate on the side of Harma Fox and terry_t"

Thank you Locrian, but I think I'll give up on these people who, as Hamer Fox says, don't agree that 'it would be more dishonest for me to ignore data that doesn't fit my own explanations than to address and examine it'. Those whose ideas are so fixed they concoct all sorts of reasons why the obvious answer is not the correct one.

"Bingo!"

Don't get too excited German. He did say:

"the fact that the 'Amerindian' component comprises both Eastern and Western signatures"

which is the point several of us have been trying to make all along. It is a mixture, not a source.

Eastern View said...
This comment has been removed by the author.
Eastern View said...
This comment has been removed by the author.
Dienekes said...

A simpler scenario would've been that it was 1) admixed, 2) a eastern component of this admixture was atleast partially sharing the same heritage with ancient South Asians, Papuans and yet on their way into developing (partially atleast) into East Asians.

If Mal'ta was admixed with something akin to "Papuans" or "ancient South Asians", then Papuans would be closer to Mal'ta than to Loschbour or Motala12. But:

"Every eastern non-African
population except Native Americans is genetically equally close to Loschbour, Motala12, and
MA1"

Dienekes said...

For example, Australasians seem to be closer to "West Eurasians" than Onge is.

You got it backwards.

"Ancient Eurasians (Europeans and MA1) are genetically closest to Karitiana, intermediately
related to Onge/Atayal, and least related to Papuans" (SI 12).

Eastern View said...
This comment has been removed by the author.
Eastern View said...
This comment has been removed by the author.
Dienekes said...

I don't mean something akin to Papuans or South Asians, not at all. I mean a population sharing a common ancestry on their way into developing into something else.

Make a diagram, because the above means nothing at all. All populations "share a common ancestry" at some time in the past and are "on their way into developing into something else" into the future.

Dienekes said...

@Dienekes
Isn't that the substantial Denisovan component in Papuans? Once you get rid of that..


If you have something to say, say it, because people don't have the time (or patience) to guess what you are talking about.

You stated that "Australasians seem to be closer to "West Eurasians" than Onge is". What's your evidence?

Eastern View said...
This comment has been removed by the author.
German Dziebel said...

@terryT

"I think I'll give up on these people who, as Hamer Fox says, don't agree that 'it would be more dishonest for me to ignore data that doesn't fit my own explanations than to address and examine it'. Those whose ideas are so fixed they concoct all sorts of reasons why the obvious answer is not the correct one."

Please do give up, Terry. You're not bringing anything to the debate.

"Don't get too excited German. He did say:

"the fact that the 'Amerindian' component comprises both Eastern and Western signatures"

That's Hamar Fox's opinion. But since it is not in the data, I ignored it. What's important is that Amerindian GREEN is more expansive than YELLOW. I agree with him on that.

@Eastern view

"In his own words, Lazaridis CHOSE for Mal'ta to be unadmixed. A simpler scenario would've been that it was 1) admixed, 2) a eastern component of this admixture was sharing the same heritage with ancient South Asians, Papuans and yet on their way into developing into East Asians and Amerinds. This would also explain why there is a Papuan component in the Mal'ta sample"

Yes to everything without the Amerinds part. Hamar Fox is tempted to axe the Papuan component although I agree it's systemic to the situation.

"This model probably fits all the data better and also gets rid of the need for ANE and Basal Eurasian."

Yes. ANE was born out the "fear of an Injun" in people's minds. Basal Eurasian out of love for Africa.

@Rokus

"Even worse, Lazardis et al. actively suggest the "basal" component unifies non-similar populations in a single basal pool that e.g. ties Neolithic basal populations togehter with Near Eastern basal populations - while actually such "basal" populations should be reminiscent of a much older layer."

yes. So now we have three people, with completely different big picture theories, equally questioning Lazaridis and Raghavan. And we have a couple of people with the same agenda supporting them.

terryt said...

" it becomes non-sensical unless ANE hopped over East Asians and admixed with Papuans. (I've actually heard terryt argue for this somewhere.)"

What?

libya said...

This study confirms (once again) genetically what already has been pointed out linguistically that neolithic "EEF" newcomers to Europe are the ones that introduced IE languages to Europe upon macro-vasconic speaking ancient populations ("WHG") of Europe

Eastern View said...
This comment has been removed by the author.
Kurti said...

In this debate I take side of Hamar Fox and terry_T because I know that "admixture" is often based on the definition of one component, and how you take it. If you unconsciously (but wrongly) take a modern component as pure (for example Amerindian) without actually knowing how and out of which more ancient genes this component emerged, it is very likely you will interpret genes which are actually West Eurasian as East Eurasian. The same was the case in Dodecad v3. The "South Asia" component was made up of two pieces of Eurasian genes. ANI (West Eurasian) and ASI(rather East Eurasian). So as a result you had some populations appearing more "Southeast" Eurasian appearing as they in reality were.

When I saw the Green (South Asian) and Dark Blue (European) Component in the Mal'ta paper in some West Asian populations, I was sure that these "components" must have eaten up other components of Dodecad K12b which we know as Caucasus and Gedrosia. And when looking at Eurogenes K19 results I seem to have been correct.


Admixture must always be taken with a grain of salt.

Slumbery said...

Eastern View
"Mal'ta is .04% "Papuan-Melanesian" and 0% East Asian."

These are equally zero. .04% is noise level.

Eastern View said...
This comment has been removed by the author.
GailT said...

As always, quite astonishing how early the specific mtDNA subgroups are (e.g., Loschbour U5b1a with two additional substitutions, and Motala 4 & 5 U5a2d with 1-2 additional substitutions, Motala 9 U5a2 with 6 additional substitutions).

It is also interesting that none of the U samples in this study appear to have survived to modern times (or at least have not yet been detected in modern populations). I posted in more detail on the extra mutations in each sample at Anthrogenica. It is interesting how often this seems to be the cases, e.g., the Kostenki U2* and Mal'ta U* appear to have been lost also.

German Dziebel said...

@Kurti

"If you unconsciously (but wrongly) take a modern component as pure (for example Amerindian) without actually knowing how and out of which more ancient genes this component emerged, it is very likely you will interpret genes which are actually West Eurasian as East Eurasian."

Will you be willing to apply your logic to the African component? We have no ancient DNA to support its antiquity. But your argument doesn't apply to the Amerindian component because it's ascertained as such in a 24,000 year old specimen, hence confirming its antiquity. Once again, there's nothing in Raghavan et al. that shows YELLOW (East Asian) + BLUE (West Eurasian) admixed in Karitiana. Instead, it's Amerindian ORANGE that's found in Mal'ta. The presence of a Papuan component confirms the admixed status of Mal'ta.

Also, on the conceptual level, you and others mistakenly equate a modern population with a recently derived population, while an ancient population with an ancestral population.

German Dziebel said...

@GailT

Same for B4'5 in Tianyuan. It got lost. I agree there's a pattern here.

About Time said...

@German

The Basal Eurasian is not African. Look again at the tree model. If anything, Basal Eurasian is less drifted away from Africans than the other Non-Africans due to an early split. But that's actually not shown on the tree.

What differentiates Basal Eurasian from all other Non-africans is that it separated early (forming its own unique group) and only later fed into WHG and EEF.

If anything, Basal Eurasian would be an isolate with its own evolutionary trajectory, that was later apparently either close enough (North Africa / Levant if Natufian) or beneficial enough (due to independent internal selection during period of isolation) that it became part of the makeup of Europeans and apparently Arabs, Uzbeks, and some Gujaratis (according to Stuttgart f3 statistics).

So Basal Eurasian is not really "African." It's sui generis that became a sort of "elder sibling" (and teacher?) to several major populations.

terryt said...

"In this debate I take side of Hamar Fox and terry_T because I know that 'admixture' is often based on the definition of one component, and how you take it".

Thank you Kurti. Unfortunately your comment will have not the slightest influence on how German sees the world.

"your argument doesn't apply to the Amerindian component because it's ascertained as such in a 24,000 year old specimen"

Which 24,000 year old speciment is 'Amerindian'? Come on. Stop making things up. I presume you're talking about the Mal'ta individual but he has other elements as well as Amerindian so cannot be used as an Amerindian sample.

"hence confirming its antiquity".

It confirms the antiquity of the specimen, and the component it carries that appear in Amerindians. But it certainly does not confirm the antiquity of 'Amerindians'.

"on the conceptual level, you and others mistakenly equate a modern population with a recently derived population"

And you're mistaking an ancient population with an admixed more recent population.

Locrian said...

Eastern View said : “I’ve already mentioned the evidence. It's in the ADMIXTURE results for Mal'ta in Raghavan et al. Mal'ta is .04% "Papuan-Melanesian" and 0% East Asian.”

“0.4%” is how we write four hundredths of a percent. It is not 4%. And it is usual to dismiss this as noise.

Locrian said...

Apologies, my own decimal point was in the wrong place. I meant to say:

“.04%” is how we write four hundredths of a percent. It is not 4%.

Annie Mouse said...

Basal European might be North African. It is possible that the first "Out of Africa" was actually the movement of folk up the Nile and onto the southern Mediterranean coast.

Locrian said...

German Dziebel wrote : “Also, on the conceptual level, you and others mistakenly equate a modern population with a recently derived population, while an ancient population with an ancestral population.”

A modern population IS a recently derived population. Tautologically! Whether it is a mixed population is another matter. But the Second Law of Thermodynamics says that mixing increases over time. So a modern population is overwhelmingly likely to be a mixture of earlier populations. (There may be extraordinarily rare cases of small inbreeding populations isolated for a long period of time, but they have to be regarded as of low probability tending to zero the longer the time scales). It is a fundamental absurdity to think of an ancient population as being mixed, in a causal sense, from later populations. Overrlap yes, but that is the only sense of ‘mixture’ that makes sense in this case.

Also it is a real logical possibility that two populations (call them X and Y) may show admixture and yet neither be causally ancestral to the other. This is because there may be a third population (call it Z) such that Z is causally ancestral to X and also to Y.

German Dziebel said...

@About Time


"Basal Eurasian is less drifted away from Africans than the other Non-Africans due to an early split."

That's what I mean: it supposedly split away from Africans prior to all others. Too bad we don't have any evidence for this split.

What does Basal Eurasian correspond to in ADMIXTURE?

Matt said...

The Basal Eurasian is not African. Look again at the tree model. If anything, Basal Eurasian is less drifted away from Africans than the other Non-Africans due to an early split. But that's actually not shown on the tree.
One way to define whether the "basal Eurasian" component here branched off in Africa or Eurasia, would be to look at the Neanderthal associated SNPs.
If a modelled "basal Eurasian" reaches the normal pan-Outside Africa sharing excess over Africans, then we can describe it as a component belonging to people from Eurasia, as the admixture event with Neanderthals affected all Out of Africa populations equally and happened in Eurasia.
If not, then it's a component that branched off in Africa.
If Eurasian populations with high levels of "basal Eurasian" (Near East and then Caucasus-European) have the same levels of Neanderthal ancestry, as ones that do not, then basal Eurasian probably has the same level of Neanderthal ancestry and should be viewed as a Eurasian component.

Also, further, on models, being discussed here, please note the paper's supplement has summed up all the models they have attempted to use, all the one admixture edge models and two admixture edge models, and explained why they do not fit drift statistics.

This is Figure S12.2 and the reasons why all other two edge admixture models were dismissed are given by Table S12.6.

Could I suggest that if you think there is a specific model which would be better than the one they have ultimately given, please look through this section, find the closest fitting model, and see why it has been dismissed before commenting further? And in any case refer to the model when you're describing your model, so people can visualize it?

Models lacking ghost populations have been dismissed because they do not fit the drift statistics, not through arbitrary choice. The ghost populations are only summoned up to fit the drift data, not because the study authors "wanted" the ghost populations.

If there isn't a model that fits a theory you hold, maybe draw a diagram and post it up as Dienekes has suggested, and we can compare to the existing models in the supplement to see if it can be dismissed. Referring to the model diagrams will help us avoid further empty talk / babbling about ADMIXTURE.

(ADMIXTURE attempts to recreate allele frequencies in individuals by combining components with different allele frequencies, but there is no guarantee that these components fit a sensible tree with admixture edges at all, which is what is needed to actually recreate population history.)

Simon_W said...

@ Mooreisbetter

No, the study's focus lies on autosomal DNA.

For example, Sardinians are most I-M26. And they cluster almost exactly with Early European Farmers. Does this mean it is time to rethink the notion that Hg J and E brought farming to Europe? Was it perhaps I-M26, traveling along the Anatolia (I2*) to the Balkans (I-P37.2) to Southern Europe?

Sardinians are also rich in G2a, the haplogroup associated with the Cardium Pottery people, judging from ancient DNA. In contrast, the notion that Hg J and E brought farming to Europe is in no way supported by ancient DNA. There is so far just one E-V13 vs. 26 G2a found in Cardium or Cardium derived samples. And precisely this study found several instances of I2 and I2a in mesolithic hunter-gatherers, even from Scandinavia.

Also, all of the heavy R-M269 (R1b) populations on the graph (Scottish, Basque, etc.) are all clustered over towards the early West European Hunter Gatherers. Time to go back to the original theory that R1b is Paleolithic? 

But so far no R1b was found in pre-neolithic context. All the European hunter-gatherers tested so far were I. Furthermore, somehow this ANE component must have spread in Europe, and although some of it may stem from Scandinavian hunter-gatherers, these don't suffice as an explanation. Apparently, a lot of the ANE component spread later, from a source other than West- and North European hunter-gatherers. And it would make sense if this was accomplished by R people, given their relationship with Amerindian Q, and even more so given the fact that the Mal'ta sample, the ANE par excellence, was R. Accordingly, the Scottish, who are among the most strongly ANE admixed Europeans, are also strongly dominated by R1b.

Simon_W said...

Oops, my apologies, I forgot to reply to this:

Mooreisbetter said:
After all, the other pole, Ancient North Eurasians DOES match a pretty non-controversial theory, that Hg N, common in Finns and other north Uralic speakers, is Eurasian.

But as the paper demonstrates, Uralic speakers and peoples with a lot of Hg N do have additional Siberian, East Eurasian admixture that is something different than the ANE admixture.

Simon_W said...

Eurologist:

From the PC plot I don't see how present Europeans (apart from extreme SE Europe) show much N Eastern admixture. Note that you cannot use PC1, but you first must rotate the plot ~30 degrees clockwise. Then you can see that most extant Europeans, ANE, and EEF all show about the same distance to N Eastern populations

But in my view the point is rather how, in what direction, the EEF deviate from the Western European Hunter-Gatherers, and it is in the direction of the Bedouins. And accordingly, in the ADMIXTURE analysis the component that differentiates EEF from W.E. Hunter-Gatherers peaks in Bedouins. The rest of the Near East is a cline that goes into a similar direction as the ANE, with the Caucasus peoples, who have a lot of ANE ancestry, at its top, and we may speculate that in early neolithic times the northern Near East was still closer to modern Bedouins. On the other hand its conceivable that already the SE European hunter-gatherers deviated from W European hunter-gatherers in the same way as EEF did.

N and C European are mostly a mix between ANE and European HGs, with just a little S European (=EEF) admixture. I don't see much N Eastern in them from the PC plot.

But that's just the first two principle components – it doesn't show the relationships fully. According to the authors' model even in the Baltic there is 30% EEF ancestry. (But EEF themselves were mixed, according to the paper.)

Simon_W said...

Libya said:

This study confirms (once again) genetically what already has been pointed out linguistically that neolithic "EEF" newcomers to Europe are the ones that introduced IE languages to Europe upon macro-vasconic speaking ancient populations ("WHG") of Europe

Or it was the ANE admixed folks who did it...

Eastern View said...

Matt, you are missing the point and not looking at the study with a dispassionate, critical eye.

On page 76 of the supplementary material, it is stated clearly that they made an active choice to assume Mal'ta is unadmixed (They never considered an alternative model of Mal'ta being admixed in their 2-edged scenarios). They excluded that whole line of alternatives under the assumption that if Mal'ta is admixed, it should be closer modern East Asians than Mssoltihic Europeans are. In doing so, they a priori set up a scenario that there were modern southern East Asians (non-Altaic; non-Siberian) in western Siberia 22 thousand years ago.

There are many reason why these are huge, problematic assumptions.

I've made a blog post with hopefully a more understandable explanation:
http://2012-1221.blogspot.com/2013/12/in-looking-at-lazaridis-et-al.html

Eastern View said...

At the other guy, it's 4%, not .04%. Please look at the data yourself instead of arguing with me.

About Time said...

I don't imagine a different tree model (deferring to experts working with actual data and computations). I just pointed out that Mbuti split from Nonafricans prior to the Basals splitting off. Which means Basals were an isolated type of Nonafrican (prior to joining back with WHG to become parental to EEF).

Some posters here suggested Basals are instead a type of African, which contradicts the tree. Whether Basals were ancestral to some Africans today is a different question not addressed in tree model. But it looks like Basals or EEF were ancestral to Arabs, Uzbeks, some Gujaratis. Maybe Afar and others in NE Africa too.

Eastern View said...

About Time, my estimates for Near Easterners are 50.06 to 67.92% Basal Eurasian.

If Basal Eurasian is closer to Africans, then we should expect a SE-to-NW pattern in Europe. To contrary, Spaniards are more Basal Eurasian then Greeks:
Greek - 65.3% non-Basal Eurasian
Spanish - 64.4%
Sardinian - 64.1%
Sicilian - 60.3%
Maltese - 59%

How did you get your Arab, Uzbek, and Gujarati estimates?

German Dziebel said...

@matt

Judging by their content, I assume you wanted to address your comments at me.

"One way to define whether the "basal Eurasian" component here branched off in Africa or Eurasia, would be to look at the Neanderthal associated SNPs."

That's a great idea. I suggested it in another string. I don't know why the authors didn't include "archaic" populations, considering the recurrent evidence for archaic admixture in Eurasia, but instead resorted to "ghost" ones.

"the admixture event with Neanderthals affected all Out of Africa populations equally and happened in Eurasia. If not, then it's a component that branched off in Africa."

There's evidence of Neandertal-enriched gene flow into SS Africa. So we would have factor that in before we talk about would-be Eurasians branching off within in Africa.

"Models lacking ghost populations have been dismissed because they do not fit the drift statistics, not through arbitrary choice. The ghost populations are only summoned up to fit the drift data, not because the study authors "wanted" the ghost populations."

Scientists, who let statistics rule uberalles, turn into "ghost" scientists. We've seen this recently with the application of statistics to the divergence of Indo-European languages. These studies are routinely accepted into journals such as Science and Nature.

"Referring to the model diagrams will help us avoid further empty talk / babbling about ADMIXTURE."

So, here's a simple tree model for testing. A basal Amerindian population (ascertained in Mal'ta, Loschbour, Stuttgart and Tianyuan) splits into East Asian and West Eurasian populations.

"which is what is needed to actually recreate population history."

What's needed in order to recreate a real population history is a combination of divergence and admixture. So we need to understand what populations are pure vs. admixed before we build a phylogeny. Otherwise, you'll end up with a wrong phylogeny.

Dienekes said...

as the admixture event with Neanderthals affected all Out of Africa populations equally and happened in Eurasia

No, West Eurasians have less Neandertal ancestry than East Asians. See the supplement of the recent Prufer et al. paper.

About Time said...

@German, I get f3 statistics showing EEF (Stuttgart LBK) admixture in Arabs from Table 10 of the original paper. F3 stats for Uzbeks and Gujaratis is from Nick Patterson's comment earlier on this thread.

EEF in Afar is just a guess based on k-means admixture tables of the Stuttgart/Sardinian component that shows up there.

@Eastern View, I don't see evidence in paper that Basal Eurasians are "more African." Just that they separated early from other NonAfricans before mixing with West Eurasians at a later time.

Someone posted Basals might be related to Nubians/Natufians that later were the first sedentary agriculturalists. I speculated (pure imagination on my part) that they might have self isolated from other Nonafricans during an aggressive range expansion. The other Nonafricans might have mixed with Neanderthals around that time.

@Dienekes, yes--- the Neanderthal link with east asia, native Americans, and Europe makes me wonder if Malta/ANE was specifically part Neanderthalized. Would be easy to test.

Could explain weird performance the non-European % that routinely show up in far west Northern Europeans like Orcadians and Irish (hard to argue for Turko-Mongols there, but ANE is there).

Dienekes said...

They excluded that whole line of alternatives under the assumption that if Mal'ta is admixed, it should be closer modern East Asians than Mssoltihic Europeans are. In doing so, they a priori set up a scenario that there were modern southern East Asians (non-Altaic; non-Siberian) in western Siberia 22 thousand years ago.

How does the fact that Mal'ta is modeled as "non-admixed" set up a scenario in which "modern southern East Asians (non-Altaic; non-Siberian) in western Siberia 22 thousand years ago."?

It's consistent with either them being present but not admixing into Mal'ta or with them being absent.

What kind of admixture are you thinking of? Clearly, Mal'ta doesn't have any kind of "Papuan" or "East Asian" admixture because Mal'ta and European hunter-gatherers are equally close to East Asians and Papuans genomewide. So, if Mal'ta has "Papuan" admixture then LOshbour and Motala must also have equal "Papuan" admixture to compensate.

Matt said...

@ Dienekes - No, West Eurasians have less Neandertal ancestry than East Asians. See the supplement of the recent Prufer et al. paper.

Good catch, I wanted to see if anyone would get that. Although unless variation within West Eurasians happens in a manner systematically related to this paper's estimated ancestral components (e.g. higher Neanderthal in more WE and ANE populations, lower in more BE populations), it does not affect the outcome of basal Eurasians being likely to be at least as deep a Eurasian ancestry as the "West Eurasian" "Ancestral North Eurasian" ancestral components.

@ Eastern View, you state on your blog they assume "that 1) there were East Asians 22 thousand years ago and that 2) they were present in western Siberia at that time".

No, I don't think that's what they were doing - their assumption is that a "basal Eastern non-African population" (ENA) which later gave rise to Onge, East Asians, Papuans and Bougainvillians was present at that time, i.e. as you state "paleo-populations that were sister or ancestral to East Asians". This seems pretty clear from their modelling.

(Of course, a unified "basal Eastern non-African" population that subsequently diverged could be questionable, but it is not an assumption that there were East Asians at that time and there is no genetic samples evidence we can currently use as counter evidence.)

Anyway, MA-1 is closer to the above populations than Sardinians and Stuttgart are. So is Loschbauer.

We can question whether there is any role for this for admixture of MA-1 with this "basal Eastern non-African population", after Loschbaur - MA-1 divergence, or because MA-1 drifted with ENA for longer than Loschbaur. If so MA-1 should be closer to all ENA descendents than Loschbaur is.

This isn't the case, MA-1 is no closer to the whole ENA clade than Loschbaur, so no-post Loschbaur divergence admixture for MA-1 and no later divergence between MA-1 and ENA than Loschbaur and ENA.

So any admixture into MA-1 from any the ENA clade as a whole happened prior to divergence between MA-1 and Loschbaur, and prior to the divergence of the ENA clade.

But MA-1 does share more drift with Karitiana than Loschbaur does. We need a special relationship between MA-1 and Karitiana of some kind.

Now, if this were Karitiana to MA-1, then MA-1 *would* be closer to Karitiana than Loschbaur is... but, to re-iterate, MA-1 would also be closer to the whole ENA clade than Loschbaur is, which as we've discussed, it isn't. So the direction of admixture must be MA-1 to Karitiana.

On the same tip, Loschbaur has to have separated from ENA at the same time as MA-1, but also has to be closer to Stuttgart than MA-1, and Stuttgart is further from ENA than MA-1 or Loschbaur, who are both as distant from ENA. This can be reconciled only by admixture from Loschbaur into Stuttgart, along with a more divergent "ghost population".

What seems remaining and questionable to me about their model is whether population X (shared ancestral population to ANE, WE, ENA) ever existed or whether population W (shared ancestral population to ANE and WE) represents an admixture of the ENA and BE clades. But this doesn't affect the final topological outcomes of admixture into the Karitiana of an MA-1 like group, or of a Loschbaur group into Stuttgart, etc. and would add an extra superfluous admixture edge.

Annie Mouse said...

"I just pointed out that Mbuti split from Nonafricans prior to the Basals"

The "Out of Africa"/Non African group includes North Africa for all practical purposes. Perhaps we should be refeering to "Out of Sub-Saharan Africa" to be more accurate..

Annie Mouse said...

"If Basal Eurasian is closer to Africans, then we should expect a SE-to-NW pattern in Europe. To contrary, Spaniards are more Basal Eurasian then Greeks:"

Which is quite reasonable if Basal European entered Europe from North Africa via Gibralter and/or via a Tunisian sea crossing to Italy. In fact these are the most likely entry points for a North African population.

mooreisbetter said...

@Simon_W: Great comments, particularly when you wrote:

"Furthermore, somehow this ANE component must have spread in Europe, and although some of it may stem from Scandinavian hunter-gatherers, these don't suffice as an explanation. Apparently, a lot of the ANE component spread later, from a source other than West- and North European hunter-gatherers. And it would make sense if this was accomplished by R people, given their relationship with Amerindian Q, and even more so given the fact that the Mal'ta sample, the ANE par excellence, was R. Accordingly, the Scottish, who are among the most strongly ANE admixed Europeans, are also strongly dominated by R1b."

Please see the discussion currently going under the http://dienekes.blogspot.com/2013/12/ancient-dna-what-2013-has-brought.html thread.

We stated the exact same thing, except with different mechanisms.

Locrian said...

Eastern View said: “At the other guy, it's 4%, not .04%. Please look at the data yourself instead of arguing with me.”

I suppose this is your cute way of saying: “My apologies, I made a simple mistake. I’ll try to be more careful in future.”

German Dziebel said...

@Matt

"Now, if this were Karitiana to MA-1, then MA-1 *would* be closer to Karitiana than Loschbaur is... but, to re-iterate, MA-1 would also be closer to the whole ENA clade than Loschbaur is, which as we've discussed, it isn't. So the direction of admixture must be MA-1 to Karitiana."

Why in the world should MA-1 be closer to the whole ENA clade than Loschbor for the gene flow to go from Karitiana to MA-1? Karitiana is closer to MA-1 followed by Loschbour and Matala followed by Shuttgart. Karitiana is closer to all of the above than any other ENA. Clearly, "gene flow" "went" from Karitiana westward all the way to Western Europe. Unless you propose that Karitiana split from Western Europeans around Luxembourg.

@Dienekes

"Mal'ta doesn't have any kind of "Papuan" or "East Asian" admixture because Mal'ta and European hunter-gatherers are equally close to East Asians and Papuans genomewide. So, if Mal'ta has "Papuan" admixture then LOshbour and Motala must also have equal "Papuan" admixture to compensate.

Mal'ta doesn't have East Asian admixture but it does have Papuan admixture. Look at ADMIXTURE in Raghavan. Forget about East Asians altogether: they weren't around at the time of Mal'ta. What you need to compare is affinity between Karitiana and West Eurasians vs. Papuan and West Eurasians. Fu et al. report that Papuans are closer to Sardinians than Karitiana is, while Karitiana is closer to French than Papuans. Indirectly (!), this supports the evidence of Papuan admixture in MA-1 that parallels Amerindian admixture. Considering that the Mal'ta site is geographically close to where Denisovans lived and Denisovans have affinity with Sima in Europe and with Papuans, the Papuan-like admixture in MA-1 and the proximity of Papuans and Sardinians in Fu et al. may be related phenomena.

eurologist said...

"Getting back to Gravettian, a new paper states there is a predecessor in the Balkans that supposedly has origins in the Levant. Thus, Gravettian (and atleast I) had Levantine origins."

Eastern View,

Do you have a source for this? I think I missed that paper.

"EEF deviate from the Western European Hunter-Gatherers, and it is in the direction of the Bedouins ...
On the other hand its conceivable that already the SE European hunter-gatherers deviated from W European hunter-gatherers in the same way as EEF did."


Simon,

The way I see it, firstly, EEF to extant Europeans to ANE all are pulled towards all W Asians (but much less so than extant SE Europeans), and then secondly, EEF is low in ANE not because it is in any way closely related to African-admixed Bedouins/ SW Asians, but because its source population wasn't ANE admixed, yet.
And yes, the second statement is my view: EEF is from a SE European source, possibly with admixture from W Anatolia and N Levant - but I don't think there was much of an autosomal difference between those during that time, just a cline.

As to a source region for the "basal Eurasians," I have often argued that the climate isolate and available diversity of E Afghanistan to N Pakistan and extreme NW India (the upper Indus river basin) is a good candidate for a population island. During dry times, it would have been cut off from any other population. Once the climate became better ~55kya, the population could have expanded south along the Indus river. Perhaps even forcefully driven south by peoples moving in from the E/ SE (MPNOS etc.).

Hamar Fox said...

I decided to examine Eastern View's polite and humble suggestion that 'basal Eurasian' and 'derived Eurasian' are not only two qualitatively different 'races' (in his words), but between them account for not only most, but essentially all Eurasian genetic variation. With the aid of Eurogenes' global PCA and Lazaridis' estimates of basal admixture (i.e. 44% of each population's EEF percentage), I created the map below (arrows aim at the approximate population average for the populations whose BE proportions are estimated):

http://s674.photobucket.com/user/camelsloop/media/EurogenesGlobalPCA--derivedvsbasal.png.html

The population with 60% BE is the Druze, based on EV's lowest estimate of BE ancestry in them.

I tried to exclude populations with excessive recent SSA and EA admixture (>5%), but admixtures of this nature still obviously exert an effect on the positioning of some of the populations considered.

I'd sum it up, but I think it speaks for itself.

eurologist said...

I'd sum it up, but I think it speaks for itself.

Please do, since 5 people will spin it in 5 different directions - myself included. ;)

About Time said...

@Hamar Fox, can you post a higher res version of that labeled PCA?

Unless I'm missing something, I can't zoom enough to read any labels. Or, do you have a link to the unlabeled PCA for reference?

Hamar Fox said...

Eurologist,

Please do, since 5 people will spin it in 5 different directions - myself included. ;)

No problem. We need to bear in mind that EV has claimed this paper shows that West Eurasians are clearly East Eurasians plus basal admixture, and that basal admixture is responsible for the W-E Eurasian split. He also seems to confuse names (of his own coinage, to boot) with actually existing distinct historical entities, seemingly undifferentiated in their parts and with no relationship between each other. So he basically thinks that 'derived Eurasian' is one population and that 'basal' is another, completely unrelated population, failing to realise that by his own obscure logic, all one would need to do to make basal Eurasians the same population as derived Eurasians would be to identify (or even just imagine) a population further upstream of both, so that both could be called 'derived X' and (again, through his obscure logic) be considered the same population. I can provide quotes if necessary.

Above I said this: "As I say, a weak association is clearly there, but what you're proposing is unrealistic and better explained by one population splitting off, another one splitting off a bit later, pooling at some point, while another population split off somewhere in the middle and went its own way relatively undisturbed, becoming much more divergent and dwarfing the differences between the 'basal' and 'W. Eurasian' populations (ANE and WE), thus making the differences between the latter two (basal and non-basal non-Easterners) fairly non-crucial on a global scale."

I think my MS paint masterpiece (:P) supports that. There's obviously no even remote evidence for BE creating the divide between W and E Eurasians. In addition, small amounts of recent non W. Eurasian admixture exert much greater pull on populations in any direction than significantly greater differences in BE ancestry, consistent with a significantly greater divergence of modern populations, even within Eurasia, at the point of admixture than between BE and W. Eurasian at the point of admixture.

Variation within East Eurasia (EV's 'pure derived race') equals that within W. Eurasia, when recent admixture is accounted for. While archaic admixture is responsible for at least some of this (e.g. in Papuans), I'd like to see EV's explanation of the rest.

The weak effect observed in my map doesn't even take into account that many of the populations with lower BE have higher modern E. Asian admixture, and many of the populations with higher BE also have higher recent SSA admixture, which amplifies what little trend may have been there to begin with.

Eastern View said...

Harma, since Mesolithic Europeans don't have any Basal Eurasian, they can't be a panmictic population with Basal Eurasian.

Lazaridis estimates Basal Eurasian to account for 34-54% of the Stuttgart sample, which "regresses" back to 88.52%-34.69% in the original Near Eastern population, based on Lazaridis' estimate of Near Eastern admixture in Stuttgart, 61-98% (supplement 10). Lazaridis has adjusted for later African admixture. The upper limit for Basal Eurasian for Stuttgart and Near Easterners actually corresponds nicely to K10 in Lazaridis' ADMIXTURE analysis, which shows a "Basal Eurasian" element appearing that is roughly 54% of Stuttgart and 89% for certain Saudis and Bedouins. This "element" increases dramatically in Stuttgard starting at K17, once Near Eastern Basal Eurasian pulls away to form its own cluster.

To go to a more understandable comparison, if we take an "Armenoid" (excuse my reference to arcane race literature) as an approximate to Saudis without African admixture) and a very northern European "Alpinoid" as an approximate for Mesolithic hunter-gatherer (short face, pudgy nose), they look as different to each other as a Papuan or Mongolian is to the Saudi.

Please take a look at the picture links below for comparison.
Basal Eurasian:
http://www.theapricity.com/snpa/bilder/troe423.jpg

Mesolithic European:
http://4.bp.blogspot.com/-fhcZVIcpF8E/Ur8cftE3i7I/AAAAAAAAACo/sOELoHK2QWs/s1600/Europoid.jpg

Matt said...

@ German Why in the world should MA-1 be closer to the whole ENA clade than Loschbor for the gene flow to go from Karitiana to MA-1?

As I understand it, because under this model Karitiana is descended from an ENA clade.

As an analogy, if, say, I had a niece, and your nephew and my niece had children, they would be closer to me than if your nephew mixed with an unrelated person, even though you had not actually mixed with me. Because my niece would carry genetic similarity to me, which we'd both inherited from our shared ancestry, even though she would also carry her own "uniqueness".

If Karitiana is the descendant of ENA and mixed with another population, W , that split off prior to ENA divergence, to make MA-1, then MA-1 would not just get the drift that separates Karitiana from ENA, but more of the drift that separates the whole ENA clade from W. So MA-1 would be different from other descendants of the W clade (e.g Loschbaur) in its affinity to ENA. Which it isn't.
If I understand correctly, with their data, Onge and Karitiana have to be partially descendants of a shared clade (here labelled ENA) due to their shared drift, but either one also has to be admixed to explain their differential similarity to both MA-1 compared to other ancient populations from West Eurasians and to ancient West Eurasians compared to modern ones and ancient farmers in general.

There could be admixture into Onge as an alternate model, but this would require an extra basal East Eurasian population (which admixes with ENA to give Onge), for which they have no adna (it would be a ghost population much more phantasmal than anything they have conjured here) and would still not explain why MA-1 is closer to Karitiana while further from ENA, which Karitiana->MA-1 admixture cannot do (see above reasons, which would still apply if Onge were admixed between ENA and a "ghost").

@ Hamar Fox - To try and compare with Eurogenes PCA graph, I added up all the branch lengths between the ancestral populations from figure S12.12, put them in a matrix and then generated an MDS (Multidimensional Scaling) plot from this - http://i.imgur.com/ZlsVTQL.png

In this home made MDS, Basal Eurasian seems to fit in a South Asia like position, and the second largest dimension, after the first between Africans and Eurasians, is between the West/North Eurasians and ENA's descendant Z. This reflects that, although the derived ancestral Eurasian modeled populations W and ENA are slightly closer to one another than either is to basal Eurasian, their descendants are more distant to one another.

(Aside from that, it also makes it clear to me that I probably wasn't thinking clearly at all when earlier I suggested that population X (ancestor of W and ENA) and W (ancestor of ANE and WE) may not exist - that doesn't fit the patterns of distances).

Although the vertices roughly fits to Eurogenes PCA, population positions projected on this graph based on admixing Basal Eurasian and West/North Eurasian wouldn't seem to mirror that closely.

I find it hard to understand why this fit would not be the case, if populations harbor substantial amounts of Basal Eurasian.

One reason I could think of is that although Basal Eurasian has a South Asian like position in PC1 and PC2 of the home made MDS graph, it is strongly distinct in PC3, which accounts for about half the variance of PC2, and strongly distinguishes BE from all other world populations. Perhaps some of this PC3 drift blends with PC1-PC2 drift in present day populations, where West Eurasian and Basal Eurasian are supposed to be quite heavily mixed. Another factor might be additional drift in Neolithic populations.

Hamar Fox said...

About time,

@Hamar Fox, can you post a higher res version of that labeled PCA?

Unless I'm missing something, I can't zoom enough to read any labels. Or, do you have a link to the unlabeled PCA for reference?


Yes, sorry. I just wanted to give a general impression. Here is my source:

https://docs.google.com/file/d/0B9o3EYTdM8lQQVZ6bjZxeUR6bzg/edit

You might want to download it rather than zoom in on the page I linked to directly, since you get much better resolution with a dl. Also, because there's heavy overlapping, you can make use of the 'find' function to single out members of the same population (keep in mind it rarely shows all members at once, but often only ten or so at a time, so you'll need to tap a couple of times to see them all).

I also should have added my own key. In the W. Eurasian cluster, red is Sardinian, yellow is French Basque (I forgot to add Pais Vasco), Orange is Scottish and English (they overlapped in the PCA; English is based on Kent, Cornwall and the general variation pattern of 'British' labelled participants), Red-brown is Norwegian, light green is Lithuanian, pink is French, purple is Druze, black is Ukrainian, dark green is Ashkenazi, dark blue is Bulgarian, light blue is Tuscan and Spanish.

In East Eurasia, I didn't go for population averages as such; I just wanted to show the range of populations that are 0% basal (apparently). But yellow is Papuan and red-brown is Amerindian (Amerindians have a strange distribution, with some plotting much closer to W. Eurasians than others; the arrow points to the most East Eurasian-shifted cluster of Amerindians).

German Dziebel said...

@Eastern View

"The upper limit for Basal Eurasian for Stuttgart and Near Easterners actually corresponds nicely to K10 in Lazaridis' ADMIXTURE analysis, which shows a "Basal Eurasian" element appearing that is roughly 54% of Stuttgart and 89% for certain Saudis and Bedouins. This "element" increases dramatically in Stuttgard starting at K17, once Near Eastern Basal Eurasian pulls away to form its own cluster."

Precisely. A component that emerged very late in the history of Eurasia is declared "basal." Just because Near East geographically close to Africa.

Eastern View said...

@Dienekes, to over-simply, once we take out the Gravettain/European element out, the "indigenous" Mal'ta element could be something intermediate between modern East Asians and archaic South Asians.

This requires no ghost populations and is the reason why I said it's simpler many posts ago.

Alternatively, there could be a Central Asian population that mixed with Gravettians coming from Europea (CA could either be ancestral to or mixed with indigenous Siberians).
1. This requires only one new node/ghost population/
2. This is exactly what we see in ADMIXTURE at K20: Ancestral South Asian+Onge, Mesolithic European, Amerind.
3. It's what archaeologists have been saying about Mal'ta for decades: Gravettian mixed with Central Asian technology.

@German: the signature at K17 is actually a signature of a new EEF population created out of admixture between Basal Eurasian and Mesolithic European. This is the initial admixture with Mesolithics that pre-date the whole-scale absorption of hunter-gatherers.

A new color should have been assigned to this signature but it was assigned to Arabs when they should've kept the Basal Eurasian color that emerged at K10.

Yes, by the rough phylogeny Basal Eurasian should've emerged earlier in clustering.

Eastern View said...

The 2 models I propose above requires no ghost population or only 1, whereas Lazaridis' model requires 2.

Hamar Fox said...

Matt,

@ Hamar Fox - To try and compare with Eurogenes PCA graph, I added up all the branch lengths between the ancestral populations from figure S12.12, put them in a matrix and then generated an MDS (Multidimensional Scaling) plot from this - http://i.imgur.com/ZlsVTQL.png

Thanks, you did a good job.

In this home made MDS, Basal Eurasian seems to fit in a South Asia like position, and the second largest dimension, after the first between Africans and Eurasians, is between the West/North Eurasians and ENA's descendant Z. This reflects that, although the derived ancestral Eurasian modeled populations W and ENA are slightly closer to one another than either is to basal Eurasian, their descendants are more distant to one another.

Yes, this is pretty much what I imagine the situation to be: originally similar populations differentiating over time, so that, frozen at any given time, the relation of the ancestors may belie the ultimate state of the descendants. Also note the additional distance from ENA to Onge. So, from, say, Y to BE we have 133, while from Y to Onge we have 211. If there are comparable distances from ENA to other individual East Eurasian populations (likely, given the paper's conclusions), then this goes some way (though clearly not the whole way) toward explaining the PCA-related phenomena.

As we might infer from the shorter distance between ENA and Z than between ENA and Onge, admixing populations become arrested in their differentiation, while extant populations continue to diverge. Two diverging populations admixing relatively early may stunt a process which continues in a third, originally similarly-related population, to the point of eclipsing the differences between the first two on the world stage.

But while the model explains the modern distribution theoretically, it doesn't explain the observed scale of it. Your PC3 explanation seems reasonable, but I've seen comparable fst distances from East Eurasian populations between two proportionately distinct populations in terms of BE. Moreover, these populations generally have low fst differentiation between each other, suggesting minimal effects of rates of drift or additional admixture.

Although the vertices roughly fits to Eurogenes PCA, population positions projected on this graph based on admixing Basal Eurasian and West/North Eurasian wouldn't seem to mirror that closely.

I agree. There's also a similar phenomenon regarding component fst distances: Southern ('basal'?) related components tend to be markedly further from East Eurasian components than are Northern European and West Asian related components, yet when comparing population-to-population, there's only a minimal effect of 'southernness' on distance from E. Eurasians, while direct admixture is much more significant.

I don't think we'll have an answer until we sequence ancient ME genomes, whereupon I'm sure our concept of BE will be modified, and the picture always tends to become more logical with more information.

Hamar Fox said...

Eastern View,

Harma, since Mesolithic Europeans don't have any Basal Eurasian, they can't be a panmictic population with Basal Eurasian.

It's actually a statistical 'ghost' population at present, so we have no idea of any specifics. That the current model doesn't fit Mesolithic Europeans as having any BE ancestry is a more cautious position.

Lazaridis estimates Basal Eurasian to account for 34-54% of the Stuttgart sample, which "regresses" back to 88.52%-34.69% in the original Near Eastern population, based on Lazaridis' estimate of Near Eastern admixture in Stuttgart, 61-98% (supplement 10). Lazaridis has adjusted for later African admixture.

You haven't worked it out, but the more you inflate the BE ancestry estimates for Near Easterners, the more you're handing my case to me on a silver platter. This is your position: "BE and (ANE + WE) are completely unrelated. Some populations are up to 88.52% BE. Some are as little as 16%. These populations cluster closely together on global PCAs and share low fst distances for no reason."

The upper limit for Basal Eurasian for Stuttgart and Near Easterners actually corresponds nicely to K10 in Lazaridis' ADMIXTURE analysis, which shows a "Basal Eurasian" element appearing that is roughly 54% of Stuttgart and 89% for certain Saudis and Bedouins. This "element" increases dramatically in Stuttgard starting at K17, once Near Eastern Basal Eurasian pulls away to form its own cluster.

Yep, get that % as high as you can.

and a very northern European "Alpinoid"

Central European.

they look as different to each other as a Papuan or Mongolian is to the Saudi.

One problem: Mesolithic/'Neolithic' phenotypes cut across the Northern/Southern (basal) divide. Armenoids are native to the Caucasus, yet Caucasus/Gedrosia/West Asian ADMIXTURE components are just as East Eurasian-shifted as Northern European components. Likewise, Berbers are part of a UP W. Eurasian phenotypic continuity that stretches from N. Africa into Siberia, yet, genetically, Berbers are very Southern.

Simon_W said...

Eurologist, I agree that there is kind of a parallel cline to the Near Eastern cline, leading from the Sardinians to the Mordovians. But it's not quite correct to put all the Near Eastern peoples into a generic West Asian category. All the Dodecad experiments for instance showed that Bedouins and Caucasus peoples are two quite different poles, the former rather belonging to a „Southern“ cluster than to a West Asian one. And concordantly, Ötzi and Gok4 were partly „Southern“ and partly „Atlantic-Baltic“, but not „West Asian“. Also the fact remains, that the EEF in the PCA are almost halfway between WHG and Bedouins. (The Subsaharan admixture in Bedouins undoubtedly increases the difference.)

Simon_W said...

Eastern View, I think it's quite obvious that you cannot take an Armenoid as an approximate for basal Eurasians or Saudis. The latter type is rather characterised by a small, narrow face, a long (!), narrow skull, a weak chin, full lips and a convex, but not hooked nose. People conforming to this are not rarely seen on Sardinia either.

Kurti said...

@Simon_W
But in my view the point is rather how, in what direction, the EEF deviate from the Western European Hunter-Gatherers, and it is in the direction of the Bedouins. And accordingly, in the ADMIXTURE analysis the component that differentiates EEF from W.E. Hunter-Gatherers peaks in Bedouins. The rest of the Near East is a cline that goes into a similar direction as the ANE, with the Caucasus peoples, who have a lot of ANE ancestry, at its top, and we may speculate that in early neolithic times the northern Near East was still closer to modern Bedouins. On the other hand its conceivable that already the SE European hunter-gatherers deviated from W European hunter-gatherers in the same way as EEF did.

Not exactly but that goes into the direction as I speculated earlier.
at least the western parts of the Near East before the expansion of the Caucasus-Gedrosia "components" was most likely predominantly a "Mediterranean like" component.
And I assume this "basal Eurasian" is Mediterranean less H&G admixed as in Europe and less African admixed as "Southwest Asian" is nowadays. The proto farmer component is likely naturally closer to African as other components. Wasn't there some ancient African admixture in Ötzi too?

Kristiina said...

I think that the Eastern non-African component (ENA) is not a real component comparable to WHG and ANE, as the latter components derive from a real ancient sample but ENA does not. There is also a clear difference of scale between WHG/ANE and ENA, as ENA seems to cover India, Southeast Asia, East Asia, Island Southeast Asia, Papua-New Guinea, Melanesia and Australia.

Matt, you say that ”there could be admixture into Onge as an alternate model, but this would require an extra basal East Eurasian population (which admixes with ENA to give Onge), for which they have no adna." People who are high in yDNA D have very different mtDNA. Tibetans and Japanese have similar mtDNA (A10/A5, C4/M8, D4/D4), but Onge have ASI (Ancient South Indian) mtDNA lines, including M3 (if I remember correctly). There are two possiblities: either yDNA D took the northern route and Japanese and Tibetan mtDNA are more original mtDNA lines or yDNA D arrived from India and Indian mtDNA lines are more original. If we take Onge to be unadmixed, then Onge should be close to ASI.

In that Eurogenes Global PCA figure, Native Americans are scattered between East Asians and Europeans with some of them closer to Africans. I suppose that recent admixture does affect the position of at least some groups. It would be interesting to see the position of the ancient Peruvian mummies on this map.

Roy said...

Chad, diet is probably irrelevant, there are different alleles for light pigmentation in European and East Asian populations. It's an important example of natural selection. See Richard Wellers TED talk on the significance of sunlight to cardiovascular functioning at higher latitudes.

Kurti said...

@Roy

There are different alleles for light pigmentation in East Asians and Caucasians, yes. But both groups are relatively light skinned compared to the non farming populations in the far North and South of the world.

It is likely a combination of three things.

environment: fogish-cloudy vs dry-clear

genetics: Some alleles common in West Eurasians

diet: Vitamin D poor vs Vitamin D rich food.

Now if you consider this.

3 of these factors fit to Central, North, Northwest and East Europeans
So they must be logically have the lightest features.

2 and 1/2 of these factors fit into most of South Europe and northern West Asia, Central Asia and parts of North Africa.
So they must be logically slightly darker than the group above.

The third group is similar to the second just that the environment is much more dry and clear. This fits to southern parts of West Asia, South Asia and many parts of North Africa. So logically they must be slightly darker as the second group.

Matt said...

@ Hamar Fox, Thanks

"Your PC3 explanation seems reasonable"

To try and test it out, I generated dummy "Sardinian" and "Iceland" populations drift distances based on inferred BE, WE and ANE proportions (via the EEF, WE and ANE tables and admixing proportions of EEF) and then both ran a new MDS based on these samples and a projected based on proportions on the orginal MDS. I chose the Sardinian and Iceland as the represented close to the European peaks for BE and WE ancestry.

http://i.imgur.com/HVTh9sl.png

You can see how in the rerun MDS relative to the projection, the second PCA takes on more of the factors distinguishing Basal Eurasian from the "East Eurasians" and less of the shared drift between WE+ANE with "East Eurasians".

So this plus extra drift in BE not described by the paper (perhaps because it is post these ancient events, or not distinguishable from extra drift in EEF based on the panel of populations) seems like it might be enough to reconcile their model with present day PCA.

I don't know about whether distances from East Eurasians are explainable by this model or this model plus additional drift in at least certain populations. It seems kind of hard to get a feel for.

"I don't think we'll have an answer until we sequence ancient ME genomes, whereupon I'm sure our concept of BE will be modified, and the picture always tends to become more logical with more information."

Yes, it could be. As more data emerges, models change. The model they have given is only the most parsimonious model for the populations they are inputting and the output they get out, being the only one that works at 2 admixture edges and with one ghost population.

For instance, a more German and Eastern View friendly model with an extra population that branches off between Z/Karitiana and Onge and which contributes to Loschbaur, as well as admixture also happening from Karitiana to ML-1, would also seem to be able to explain the patterns of closeness, but would require an additional admixture edge and ghost population for which there is no current data. If we had that data, then that model would become viable, but we don't.

In a similar way, ancient ME genomes could change our understanding of what BE is like or whether it is even viable.

German Dziebel said...

@Matt

"As I understand it, because under this model Karitiana is descended from an ENA clade."

Well, this is a problem. If ENA is descended from Karitinana, then the proximity of MA-1 to Karitiana is easily explained as gene flow from Karitiana to MA-1. That's the simplest explanation for the pattern.

"either one also has to be admixed to explain their differential similarity to both MA-1 compared to other ancient populations from West Eurasians and to ancient West Eurasians compared to modern ones and ancient farmers in general."

Again, why would Amerindians need to be admixed to explain their similarity to MA-1 vs. ancient Eurasians? MA-1 is geographically closer to the New world than Western Eurasia plus it's older than the Western Eurasian sites. Over time and with geographic distance the signal of the founding "gene flow" weakens due to drift and subsequent admixtures in Western Eurasia in which Amerindians weren't involved due to isolation.

Matt said...

German, I think that if ENA is descended from Karitiana and MA-1 is also partially descended from Karitiana, again MA-1 should be closer to ENA than Loschbaur (my son, mixed with another person, would be closer to a drifted copy of me than a person who was not my descendant). So having ENA descended from Karitiana rather than vice versa, does not get around the conclusion at all.

If MA-1 simply diverged from the ENA clade later than Loschbaur that wouldn't explain why it is relatively closer to Karitiana, relative to its distance from ENA descended clades, whether this ENA clade is descended from Karitiana or not.

Simon_W said...

Hamar Fox said: Berbers are part of a UP W. Eurasian phenotypic continuity that stretches from N. Africa into Siberia, yet, genetically, Berbers are very Southern.

But in the Admixture analysis of the present study, for instance at K=15, Tunisians, Mozabites and Algerians do have a nontrivial share of the Northern/European component, this might explain the „UP“ phenotypes. On the other hand, not all Berbers are of this UP type I think; don't know how common it actually is, it may differ between regions and tribes; and the old physical anthropologists variously also described a „Südmediterranid“ or Saharid type that is more aptly Mediterranean and Southern.

Justin said...

Kristiina,

Hey I am currentlly visiting Helsinki, Finland. Do you have any advice/recommendations? I don't know if I can use blogger to have this conversation not in this thread. :/
kiitos

Hamar Fox said...

Kristiina,

It would be interesting to see the position of the ancient Peruvian mummies on this map.

You can find it in the supplementary information for this paper:

http://www.cell.com/AJHG/fulltext/S0002-9297%2813%2900459-X

I took a screenshot for you:

http://s674.photobucket.com/user/camelsloop/media/ancientperuvian.png.html?filters[user]=94709316&filters[recent]=1&sort=1&o=0

The ancient Peruvian seems to occupy the same position relative to other world populations as the Amerindians I pointed to in my MS Paint map, so it's likely that those Amerindians are unadmixed (in post-Columbian terms), while the modern Amerindians who plot closer to SSA and West Eurasians possibly have minor recent admixture.

Simon_W,

On the other hand, not all Berbers are of this UP type I think; don't know how common it actually is, it may differ between regions and tribes; and the old physical anthropologists variously also described a „Südmediterranid“ or Saharid type that is more aptly Mediterranean and Southern.

Modern N. Africans have considerable SSA and Middle Eastern admixture. Modern Berbers can only approximate the appearance of their less admixed ancestors. Kabyle seem to be the least admixed Berbers that I know of, with SSA characteristics being almost entirely absent from the admittedly few that I've seen. Google searches inevitably lead to anthro-forums and the low intellectual standard associated with them, so I can't guarantee searching 'Kabyle' doesn't lead to cherry-picked examples, but, still, I'd say people like Zinedine Zidane are about as close in appearance to ancient Berbers as is achievable.

Here are some reconstructions of an ancient N. African (Mechta Aflou):

http://s653.photobucket.com/user/Tyranos/media/reconstitution.jpg.html

http://www.city-data.com/forum/attachments/europe/97136d1339853647-finns-hungarians-russians-eastern-slavs-part-sapiens5-1.jpg

http://images.sciencesource.com/preview/BD8726.html

Kabyle and Mechta Aflou both fall into the 'Southern Cro Magnoid' category (when 'Cro Magnon' is used as a 'UP European' synonym, since I know it means different things to different people). Yet, when SSA is accounted for in N. Africans, they equal or exceed Middle Easterners in genetic Southernness:

https://docs.google.com/spreadsheet/ccc?key=0ArAJcY18g2GadHZ6SHpiLTNTa3lsUmZJY2pQblVRR2c#gid=0

Eastern View said...

@Harma, you don't seem to realize that Basal Eurasian needs to exist in order for ANE to exist. Please read and understand the supplements. This is the crux of the matter: a hypocrisy of having the pie and eating it too.

@Matt, I don't see what extra ghost population you are talking about, unless you are still assuming Mal'ta to be pure. If you assume Mal'ta is an admixture, a para-Ancestral South Asian population mixing with Mseolithic European merely takes place of Ancestral North Eurasian. Depending on how you see it, this requires no new node or only 1 new node, where as their model requires 2 new nodes: Pure Mal'ta/Ancestral North Eurasian and Basal Eurasian.

I also do not understand what new data you are talking about. The data has always been Mal'ta, and deciding it to be pure forces much of Western and Amerind DNA to be alligned with Mal'ta, thus forcing a Basal Eurasian necessity. Deciding for Mal'ta to be admixed would generate another set of data. Both models fit the f4 Z-score distances of the populations they tested, which itself problematic.

Eastern View said...

@Eurologist, please check at Maju's blog for the link to the paper about Gravettian origins.

I've had a change of mind, based on the new Russian finds, and many if not all of European paleolithic cultures could have a Central Asian origin.

@Harma: But Caucasus people do have Mesolithic European admixture and they on average do look less Near Eastern than hypothetical pure-Saudis would. The Armenoid type is just documenting an extreme case in the Caucasus. And Berbers do have a Paleolithic European component that is archaeologically even documentable. There is no natural cline in the two ways you propose.

Hamar Fox said...

Eastern View,

@Harma, you don't seem to realize that Basal Eurasian needs to exist in order for ANE to exist. Please read and understand the supplements. This is the crux of the matter: a hypocrisy of having the pie and eating it too.

As I said, it's currently a statistical concept. I never denied that it corresponded to something real, only that that real something is currently unknown. The model aimed at efficiency, so it came up with the best fit. But a similar effect can be produced with less admixture from a more distantly related population than the one modelled, or more admixture from a less distantly related population than presently modelled, for example. More ancient DNA should prove insightful.

But Caucasus people do have Mesolithic European admixture and they on average do look less Near Eastern than hypothetical pure-Saudis would. The Armenoid type is just documenting an extreme case in the Caucasus. And Berbers do have a Paleolithic European component that is archaeologically even documentable.

Yes, but you're not getting the point: Caucasus/Gedrosian/West Asian/ANI populations clearly have less 'UP' influence than Europeans, but, in terms of the components that constitute them, they're typically no further from East Eurasians than Northern European components are, especially when accounting for the fact that late-appearing N. European components are more shifted toward E. Eurasia than expected when they absorb a portion of the E. Asian admixture of Finns and N. Russians, and to a lesser extent all Eastern Europeans.

Likewise, Berbers are much more UP than Caucasus/Gedrosian/West Asian/ANI populations, yet are much further from E. Eurasians in terms of the components that constitute them, even when SSA is taken into account.

So it's not so simple as you wish.

Harma

Even though you think it's annoying to deliberately misspell my name, on about the third or fourth time you did it, it actually made me smile.

German Dziebel said...

@Matt

" I think that if ENA is descended from Karitiana and MA-1 is also partially descended from Karitiana, again MA-1 should be closer to ENA than Loschbaur (my son, mixed with another person, would be closer to a drifted copy of me than a person who was not my descendant). So having ENA descended from Karitiana rather than vice versa, does not get around the conclusion at all."

It does. Under an ancient New World origin model for ENA and a part of the Western European pool, MA-1 is so much closer to the New World geographically and so much older than Loschbour that it's naturally more proximate to Amerindians genetically than Loschbour. You are overthinking it.

Onur said...

Yes, but you're not getting the point: Caucasus/Gedrosian/West Asian/ANI populations clearly have less 'UP' influence than Europeans, but, in terms of the components that constitute them, they're typically no further from East Eurasians than Northern European components are, especially when accounting for the fact that late-appearing N. European components are more shifted toward E. Eurasia than expected when they absorb a portion of the E. Asian admixture of Finns and N. Russians, and to a lesser extent all Eastern Europeans.

Likewise, Berbers are much more UP than Caucasus/Gedrosian/West Asian/ANI populations, yet are much further from E. Eurasians in terms of the components that constitute them, even when SSA is taken into account.


"Caucasus", "Gedrosia" and "West Asian" are not populations, they are just ADMIXTURE components, and ANI is a hypothetical population. I hope your mistakes are not deliberate.

Kristiina said...

Justin, if you are a scientific person, you might find Heureka interesting: http://www.heureka.fi/en . If you need any other information, I may be able to help you (sanava@pp.inet.fi).

Hamar, thank you for the screenshot. The Peruvian mummy is clearly closer to the Japanese and Chinese clusters than to the European clusters. It also looks like the modern Native Americans have recent European and African admixture.

If we take Mal’ta as unadmixed, then we may assume that the European component comes from Siberia and this component also influenced South Asia and may be responsible for ANI. We may also assume that Papuans are in part derived from ancient Siberians. The link between Mal’ta and Arctic & Native American people is somewhat obvious and easy to understand.

Simon_W said...


Hamar Fox,

It would be interesting to know how „southern“ or „northern“ an ancient, pure UP North African like Mechta Aflou would be in an Admixture analysis. We don't know so far. Clearly such Cromagnoid types are also common on Sardinia, and even Ötzi looks rather like a Cromagnoid Sardinian than like a typical Mediterrean. But then again it's clear that both Sardinians and Ötzi do have quite a lot of northern admixture, even though they are very southern for modern European standards. But I suppose UP types are uncommon in Saudis and Bedouins.

Yes, in the Dodecad K7b analysis Bedouins, Saudis, Mozabites and Moroccans are the peaks of the Southern component. The question is, how conclusive this is. In the same analysis Norwegians and Swedes are as northern as Lithuanians and Finns. While at higher K, some southern admixture in the former becomes evident, and only the latter remain the peaks of „northernness“. In the Admixture analysis of the present study, right from the appearance of the split between the Southern/Bedouin and the Northern/European cluster at K=10, there is some northern admixture in Tunisians, Mozabites and Algerians that is virtually absent in Saudis and Bedouins.

Zidane looks rather like a robust Mediterranean type to me, somewhat depigmented, but not predominantly UP. I think the Mediterranoid types of Northern Africa are not exclusively derived from the Medieval Arabian expansion.

Simon_W said...

ANI/Gedrosia seems non-UP, but here the Caucasus type according to Soviet anthropology:
http://img545.imageshack.us/img545/5827/n97k.png

Simon_W said...

To avoid any confusions let me add that although the terms Cromagnoid and UP are often used interchangeably, it would be more appropriate to make a distinction: Cromagnoids in the narrow sense are those with a broad, relatively short, typically squat face, thus having a low facial index. But definitely not all Upper Paleolithic Europeans were of this type. What's common to all of them is just the wide absolute breadth of the face. But some were at the same time very long faced too, having a high facial index. (And I think the mesolithic people from Muge were even already gracilizing their absolute measurements before receiving any genetic input from neolithic newcomers.) Therefore Coon was basically right when he didn't take the relative facial index into consideration when distinguishing between his UP types and the „Mediterraneans“. So in this sense you could say Zidane is somewhat UP, although he isn't Cromagnoid.

Hamar Fox said...
This comment has been removed by the author.
Hamar Fox said...

Simon W,

It would be interesting to know how „southern“ or „northern“ an ancient, pure UP North African like Mechta Aflou would be in an Admixture analysis. We don't know so far.

I agree. It'd be interesting to see if the two major physical types within W. Eurasia (robust CM and gracile Mediterranean) really do correspond with the major genetic divisions within W. Eurasia. If they cut across these divisions, then maybe we should look to drift in which physical features remained relatively unchanged as an explanation.

Clearly such Cromagnoid types are also common on Sardinia, and even Ötzi looks rather like a Cromagnoid Sardinian than like a typical Mediterrean.

Yes, I agree. Also, the recent reconstruction of a Neolithic Briton also shows clear CM features.

Yes, in the Dodecad K7b analysis Bedouins, Saudis, Mozabites and Moroccans are the peaks of the Southern component. The question is, how conclusive this is.

Proportions certainly change between calculators, especially for closely-related components or already-admixed components, so I wouldn't call it conclusive. Saudis seem to be more consistently non-European. But can the physical differences between Berbers and Saudis be explained by ~10% admixture? If I get more time, I'll look into comparing N. Africans with, for example, Druze and Saudis in terms of both features and ADMIXTURE results.

Hamar Fox said...

Onur,

"Caucasus", "Gedrosia" and "West Asian" are not populations, they are just ADMIXTURE components, and ANI is a hypothetical population.

Considering I posted this:

"yet Caucasus/Gedrosia/West Asian ADMIXTURE components are just as East Eurasian-shifted as Northern European components"

on December 30th, then unless I suffered some kind of stroke between December 30th 2013 and January 2nd 2014, I think your trivial nitpick is superfluous.

Quite obviously by 'Caucasus/Gedrosia/West Asian populations' I was referring to populations dominated by those ADMIXTURE components, as should have been clear in the context of the point I was making and the points I'd already made.

I hope your mistakes are not deliberate.

No mistake was made. To clarify by a purely hypothetical example: a mistake would be, say, someone claiming 25% 'ancient Mongoloid' admixture in Europeans. Of course, no one would ever make such a laughably incorrect statement. As I say, it's a purely hypothetical example.

Onur said...

Hamar,

You are mistaken because you compared populations with components. You wrote:

"Caucasus/Gedrosian/West Asian/ANI populations clearly have less 'UP' influence than Europeans, but, in terms of the components that constitute them, they're typically no further from East Eurasians than Northern European components are, especially when accounting for the fact that late-appearing N. European components are more shifted toward E. Eurasia than expected when they absorb a portion of the E. Asian admixture of Finns and N. Russians, and to a lesser extent all Eastern Europeans."

Populations with high levels of the "Caucasus", "Gedrosia" and/or "West Asian" components also have high levels of the "Mediterranean" and/or "Southwest Asian" components, so they cannot be compared with the "North European" or similar components.

As for my mistake, it was due to a misunderstanding of the results and conclusions of David Reich's 2012 papers on my part. By then, I was not familiar with formal admixture tests. Now I am much more familiar. I accept that my mistake was far from trivial. But it is an issue separate from the one you and I are discussing now.

Hamar Fox said...

Onur,

Hamar,

You are mistaken because you compared populations with components.

Populations with high levels of the "Caucasus", "Gedrosia" and/or "West Asian" components also have high levels of the "Mediterranean" and/or "Southwest Asian" components, so they cannot be compared with the "North European" or similar components.


Yes, but I accounted for these things, and found there was still a phenomenon to be described.

The phenomenon that interests me is this: A 'Southern' (i.e. relatively high-scoring in this component) population such as Mozabite has/may have stronger features associated with the East-Eurasian-shifted UP population under discussion than a 'Southern' + 'West Asian/Gedrosian etc.' population.

This is, broadly speaking, the reverse of what we should expect if these features perfectly correspond to originally distinct populations, or if these features are indicative of global affinities.

Of course, the above is entirely debatable. But my point was always that it's a complex matter and deserves more analysis than the seemingly ideological treatment EV is giving it.

I didn't rigorously display my logic because I felt the thread was winding down and I was planning to leave it, since I don't find conflict enjoyable.

As for my mistake, it was due to a misunderstanding of the results and conclusions of David Reich's 2012 papers on my part. By then, I was not familiar with formal admixture tests. Now I am much more familiar. I accept that my mistake was far from trivial. But it is an issue separate from the one you and I are discussing now.

Yes, I respect that. I interpreted your comments in this thread and the other recent thread (regarding ASI) as petty retribution for past disagreements, hence my defensive tone. I'm glad there seems to be no genuine animosity.

About Time said...

Disagree about Gedrosia etc not being populations. K-means clusters are not magically derived ancestry signals. They are simply divisions of a bunch of data points into groups, thus artificial populations.

No more or less real than "countries" as criteria for "what is a population." Especially since many countries (especially big ones) vary genomically.

Admixture then uses these artificial populations (clusters) to assign proportional ancestry for individual data points.

People, look into the math for understanding. Don't treat these things as a black box, it's just machine output ultimately. We have to use our grey matter to actually understand it, ideally from many computational angles.

Onur said...

Yes, but I accounted for these things, and found there was still a phenomenon to be described.

The phenomenon that interests me is this: A 'Southern' (i.e. relatively high-scoring in this component) population such as Mozabite has/may have stronger features associated with the East-Eurasian-shifted UP population under discussion than a 'Southern' + 'West Asian/Gedrosian etc.' population.

This is, broadly speaking, the reverse of what we should expect if these features perfectly correspond to originally distinct populations, or if these features are indicative of global affinities.


I quite did not understand what you are talking about here. Could you give some examples?

Yes, I respect that. I interpreted your comments in this thread and the other recent thread (regarding ASI) as petty retribution for past disagreements, hence my defensive tone. I'm glad there seems to be no genuine animosity.

My ASI discussion is nothing new. I have been to that discussion several times over the past one year.

Onur said...

Disagree about Gedrosia etc not being populations. K-means clusters are not magically derived ancestry signals. They are simply divisions of a bunch of data points into groups, thus artificial populations.

No more or less real than "countries" as criteria for "what is a population." Especially since many countries (especially big ones) vary genomically.


Then you do not understand what ADMIXTURE components are or how ADMIXTURE works.

About Time said...

Then you do not understand what ADMIXTURE components are or how ADMIXTURE works.

It was my impression that ADMIXTURE aggregates data points into clusters using k-means and then assigns proportional ancestry ("admixture") based on those aggregations.

Therefore "Gedrosia" is an artificial population composed of whatever data points were fed into ADMIXTURE initially. Nothing more, nothing less, and nothing different than a population aggregated of whatever Baluchi, Sindhi, etc data points that compose it.

Thus, ADMIXTURE output is completely dependent on what data points are used as input.

When ADMIXTURE says so-and-so data point is x% of Gedrosia, it means that this artificial population (i.e., a bunch of Baluchis etc) parsimoniously accounts for x% of what makes up that genotype. 1-x% of that person's genotype is accounted for by other artificial populations/clusters aggregated from other data points used as input in that ADMIXTURE run.

If I am mistaken, please correct me.

Chad Rohlfsen said...

Something new on light skin, if anyone is interested. What's your take on it Dienekes and everyone else?

http://www.dailymail.co.uk/sciencetech/article-2535288/Light-skin-colour-Europeans-stems-ONE-ancestor-lived-India-Middle-East-10-000-years-ago.html

Onur said...
This comment has been removed by the author.
Onur said...

It was my impression that ADMIXTURE aggregates data points into clusters using k-means and then assigns proportional ancestry ("admixture") based on those aggregations.

Therefore "Gedrosia" is an artificial population composed of whatever data points were fed into ADMIXTURE initially. Nothing more, nothing less, and nothing different than a population aggregated of whatever Baluchi, Sindhi, etc data points that compose it.


It depends on what you mean by "artificial population". Anyway, my objection was not to your use of the phrase "artificial population" but to your apparent treatment of not-so-artificial populations, ethnic groups. Ethnic groups are not as artificial as you seem to imply (though they are much more artificial than are races). An ethnic group is very far from being pure but it has some widespread genetic patterns that are the result of relative homogenization (as statistically marriages are more often with people from the same ethnic group than not). Another point: ethnic groups are usually genetically closest to geographically closest ethnic groups (as statistically inter-ethnic marriages are mostly with people from neighboring ethnic groups). Therefore, multi-ethnic countries are usually not as genetically heterogeneous as you seem to imply (New World countries are an exception to the rule, of course).

David Jacobson said...

I was impressed by the quality of the global admixture data in what appears to be extended data figure 3. At k=20, that data appears to show three major European populations, the original hunter gathers in blue, the early farmers in pink, and a third population in beige. The third population appears to have a history on the whole arc from the Atlantic to India and be particularly prominent in the Middle East and Central Asia. A few Northern European countries also have a fourth population in dark blue that is shared with Siberia and Eskimos. The four ancient European samples do seem to fit well with the hunter gathers and in one case the farmers. By contrast, MA 1 is a single sample of low quality data with 6 different colors that make no geographical sense. One wonders how much can really be reliably concluded from that one sample.