October 21, 2012

Ancient European DNA assessment with 'globe4'

In a previous experiment, I showed that ADMIXTURE at K=4 tracks the same signal of Amerindian-like admixture detected with f-statistics. I encapsulated that analysis in the globe4 calculator over at the Dodecad Project blog, and decided to use it to assess a few ancient European autosomal samples:


Please note that a very variable number of SNPs was extracted from these various samples. These results should be viewed as indicative of possible patterns that might be confirmed by a more thorough analysis. Also, please consult the globe4 post for more details on the methodology behind it, and the interpretation of the 4 components.

With these various caveats, I would say that these results seem to make some sense and to be fairly consistent with the scenario of Patterson et al. (2012):

  • Oetzi and Gok4, the "farmers" seem to lack the Amerindian component
  • Ajv52, and Ajv70, the northern hunter-gatherers seem to possess it
  • Bra1, the Mesolithic Iberian seems to lack it as well
Bra1 also happens to be the most limited sample in terms of available SNPs. Nonetheless, this would appear broadly consistent with the idea that the "Amerindian"-like admixture in Europeans emanated from north-eastern Europe. Today, all continental Europeans seem to possess some of it, but this can be explained by migration of Ajv-like individuals and their mixtures into Western and Southern Europe from central or northern Europe for which there is ample historical and archaeological evidence (e.g., Italo-Celts, Germans, and Slavs, in addition to other, earlier phenomena).

A broader context

The absence of the Amerindian-like admixture in South Indian Brahmins and Armenians, and its paucity Kurds and Iranians might indicate that this type of ancestry was not represented in ancient Armenians and Indo-Iranians. Indeed, all these populations possess less of this admixture than those of the North Caucasus. Cypriots possess none of it as well, where the Greek_D sample, a small 2.5% portion. In a previous analysis, I estimated a historical-era estimate of North European admixture in Greeks, and this admixture presumably incorporates the signal of Amerindian-like admixture. Additionally, an Iron Age individual from Bulgaria will soon be announced as being Sardinian-like.

The sum of these factors leads me to believe that the signal of Amerindian-like admixture did not play an important role in the formation of the Graeco-Phrygians (and their Armenian relatives) and the Indo-Iranians, or at least did so to an insignificant degree. As the former expanded westward from the PIE homeland, and the latter eastward, they would have had little opportunity to encounter this type of admixture; rather, they would have admixed with Sardinian-like individuals in the west, and Ancestral South Indian (ASI)-like or East Asian individuals in the east.

On the other hand, as Indo-European groups expanded into eastern Europe, setting off a chain of events that would eventually transform most of the northern part of the continent, and, in historical times, much of the rest of it, they would have met with Ajv-like individuals carrying the signal of Amerindian-like admixture, as well as the Oetzi/Sardinian-like farmers that had spread all the way to Scandinavia by the late Neolithic. The population formed by this mixture would have carried with it the signal of Amerindian-like ancestry, and would then transpose it across the continent. The signal would become increasingly muted westward and southward, and indeed this is what we observe.

UPDATE: It is interesting to see that South Indian Brahmins (both the Metspalu et al. sample, and my Iyer_D and Iyengar_D samples) lack this admixture, while Uttar Pradesh Brahmins do not, given the rolloff evidence for a more recent admixture of the latter. This is consistent with a historical admixture event, after the migration of Brahmin groups southwards, as described in that post.

31 comments:

Sisophon said...

I am southern Irish (8GGP from the blackwater river catchment) and get 8.12% Amerindian using globe4.par. Looking at the figures from Ancient DNA above, you would think that Ireland had very high levels of Northern Hunter Gatherer (70%?). This does not seem realistic.

Garvan

Davidski said...

Congratulations, this Amerindian-like component looks like a very solid signal of the early Indo-European dispersals.

It also proves that Armenians are language shifters who didn't absorb much Indo-European genetic influence. They're basically Indo-European speaking Assyrians.

BTW, the Amerindian-like admixture is indeed present in Indo-Europeans of South and South Central Asia, and I'd say it was introduced by the same groups who brought R1a-Z93.

Unknown said...

for otzi ,maybe the so called Amerindian component is shown as asian and european because of the few snps analysed.

Dienekes said...

Congratulations, this Amerindian-like component looks like a very solid signal of the early Indo-European dispersals.

Sure, because as we all know, Indo-European preserves in its vocabulary the language of Mesolithic hunter-gatherers from the Baltic.

It's more likely that this is Comb Ceramic-related. Pottery-using non-farmers seem to have drifted across north Eurasia from the east, where pottery also preceded agriculture.

Matt said...

Thoughts-

1. It's kind of interesting that the Amerind levels in Ajv52 and Ajv70 seem not *that* different from present day Lithuanians and Poles compared to (8.2 and 9.1 vs 9.4 and 12.9), because if we were thinking about massive replacement where around 10% of the ancestry of Lithuanians and Poles is from hunter-gatherers represented by Ajv52 and Ajv70 (vs 90% from local hg), then that would seem like it should not be the case.

Perhaps either:

- Ajv52 and Ajv70 are just at very low levels of Amerindian relative to their population by chance.
- Ajv52 and Ajv70 are already mixed with incoming Sardinian-like farmer populations.
- Subsequent population inflows from an Amerindian-like, but not Asian-like population(ruling out populations like most present day Siberians) also occured.

2. Yes, it seems like this supports an expansion across Europe of the Amerind component more than mixing with local hunter gatherers as an explanation for its presence in Europe, since Brana doesn't have it.

3. Brana's not Asian like at all here, which is interesting given the earlier PC analysis that showed them as Asian shifted.

Dienekes said...

@Matt, according to Skoglund et al. there is only like 10-15% Gok4-related ancestry in Finns/Russians and the like, so it should not be very surprising that these ancient samples would have Amerindian-like levels similar to modern day populations.

But, it's also possible that the present-day percentage represents "a balance of influences": East Eurasian-influenced groups flowed into Northeastern Europe until the Bronze Age at least, from the east, and perhaps beyond, and, at the same time, largely uninfluenced groups flowed from the south.

Slumbery said...

I find two parts of the Globe4 Amerindian admixture levels surprising. One of them what Matt said, that Ajv52 and Ajv70 are not significantly more admixed that the modern populations. Possibly the component was much stronger in the East and had a lower level in West Scandinavia.

The other interesting result is that modern Siberian populations do not have a much higher Amerindian. They have higher of course, but only maximum twice as much as East Europe. I'd except a bigger difference. This is probably partial population replacement in Siberia since people moved into America.

Dienekes said...

The other interesting result is that modern Siberian populations do not have a much higher Amerindian. They have higher of course, but only maximum twice as much as East Europe. I'd except a bigger difference. This is probably partial population replacement in Siberia since people moved into America.


This is consistent with what Patterson et al. observed regarding Karitiana and northern populations.

What seems to have happened was that the "boreal hunter-gatherer population" spawned "Amerindian"-like Comb Ceramic in the west at a time when the expansion of the "Asian" element into Siberia had not yet occurred. This expansion acted like a wedge between Europeans and Amerindians.

A few European groups (such as Finns) also received a little gene flow from the "Asian" element, but most of them only from the earlier "Amerindian" element.

mooreisbetter said...

You're clearly a pioneer, Dienekes. Crafting good high-level theories, and trying to back them up with hard evidence.

A few things:

(1) Your posts could be written a little clearer. For those who drop by and are familiar with many scientific concepts, these admixture/roloff themed posts are a bit hard to understand. Perhaps include a standard explanation of what software you are using at the beginning of each post.

(2) There ARE an awful lot of conjectures and assumptions built into this one.

(3) And this is key, and you've never addressed it: your continuing description of Sardinians as like a post-Neolithic lockdown / isolate is incorrect.

First, they are clearly unique genetically and isolated. Luigi Cavalli-Sforza noted that two decades ago. He said the best way to think of Sardinians is almost as a their own race! Almost like a sister group of Europeans. That is how unique they are.

Cavalli-Sforza refused to use them in most of his Principal Component analyses, because they were so off the charts unique.

But you keep describing them as S. Europe Neolithics. Remember, the most unique thing about Sardinians from a sex-linked DNA perspective is their I-M26, which is approximately 40% of the males on the island.

M26 is a purely western phenomenon, reaching appreciable percentages only in Castille, Basque Country, seaboard-France (Normandy, Southern France), Channel Islands, Ireland, England seaboard, etc. If you were to draw a "C" around the west coast of Europe (where the top touches the UK and the bottom touches Sardinia), there's your distribution.

I am not aware of ANYONE who would describe I-M26 as a Neolithic component, yet you essentially ignore the most unique feature of Sardinians when you describe them as if they are the first farmers from the east frozen in time. They do have notable presence of Hg G, but remember, it is I-M26 that defines them...

Dienekes said...

(3) And this is key, and you've never addressed it: your continuing description of Sardinians as like a post-Neolithic lockdown / isolate is incorrect.

All the evidence suggests that they are very much alike Neolithic individuals from Europe. That does not mean they are identical, but they're the best proxy for them that we have. And, indeed, Patterson et al. showed some very good evidence, that at least where the Amerindian-like admixture is concerned, Sardinians seem to be quite similar to Oetzi.

Cavalli-Sforza refused to use them in most of his Principal Component analyses, because they were so off the charts unique.

There are two ideas regarding the lack of polymorphism in Sardinians: according to one, the Sardinians had been just like Europeans who had lost diversity in their island environment. According to another, they never had that "extra diversity"; it was added to mainland Europeans by admixture.

It is now pretty clear that the second idea is correct. Sardinians form an island of their own in PCA space, and stand apart from mainland Europeans because of admixture in the latter. They lack alleles present in mainland Europe, not because they (Sardinians) lost them, but because mainland Europeans gained them by admixture. And, their genetic peculiarity was not the result of isolation in a small island, because people just like them existed in the Alps, and Sweden and Bulgaria.

I am not aware of ANYONE who would describe I-M26 as a Neolithic component, yet you essentially ignore the most unique feature of Sardinians when you describe them as if they are the first farmers from the east frozen in time. They do have notable presence of Hg G, but remember, it is I-M26 that defines them...

You are conflating two different phenomena: Sardinians appear to be like Neolithic _Europeans_. As such they are descended from a mixture of incoming Near Eastern farmers and pre-Neolithic Europeans. This admixture must have happened in the Balkans, during the entry of Neolithic farmers into Europe, because the same Sardinian-like population seems to have existed there and gone to both the Alps and Sweden.

Once you realize that Sardinians are Neolithic Europeans, you will see no discrepancy between what I'm saying and what you think I'm saying.

/HW said...

Underlying logic for the phenomenom is actuallu fairly simple. Admixture identifies these snips with modern Amazonians (Surui, Karitiana et al).

These all are small populations with not that much genetic variation.They are decended from people who crossed
the Bering Straits maybe 15.000 years ago (something like that anyway). Those migrating people were few and didnt represent
the full genetic variation of the Ice Age Northern Eurasia.

They were from the easternmost part of the "Mammoth Steppe".

Mammoth Steppe biotype spanned from what is now Atlantic France all the way through Siberia to Alaska and even slightly further.
It is very likely that novel insitu mutations and certainly the genetic drift have altered the Amazonians, so that they are
not 1:1 carbon copies of those ancient Eurasians. Even those ancient Boreal Eurasian pioneers didnt represent the full genomic
diversity of Mammoth Steppe.

So what we are actually looking here is small part of mesolithic ancestry in modern Europeans.
Just that fraction which resembles the Amazonians, it by no means represent the full spectrum of mesolithic ancestry.
I very much doubt that Comb Ceramic Culture has anything to do with this "component" appearing at Europe (but surely it was present in CCC are aswell),
as the Irish score allmost same amount as Poles do.

Fact that some modern North Europeans score very similar amounts of "Amerindia" as the neolithic Gotlanders is because there is very similar
amount of it floating around. Situation has not changed.

There hasnt been any migration from Amazon to Northern Europe to increase it!

This is also indirect evidence that modern North Europeans are still very similar to the Hunter-Gatherers
living at Northern Europe during neolithic age. If there would have been large scale mass migrations after that,
lets say from Southern Europe (where the component reaches 0 at Cyprus forexample), it would have lowering effect to the "Amerindian" score.
If the Gotlanders had 9-12% of this component, and they were allready somewhat mixed with farmers,
then the most neolithic North Euros should score into that range and everything below that implies more farmer-admix.

I'm very surprised about Bra1 showing no connect with the more northern neolithic HGs via the "Amerindian"-component.
This might be because the small amount of available snips in those samples, just like Dienekes wrote in the introduction.

Matt said...

This is probably partial population replacement in Siberia since people moved into America.

What seems to have happened was that the "boreal hunter-gatherer population" spawned "Amerindian"-like Comb Ceramic in the west at a time when the expansion of the "Asian" element into Siberia had not yet occurred. This expansion acted like a wedge between Europeans and Amerindians.

Slumbery/Dienekes - Yes, that's what I would think as well. It feels odd to think of Native Americans as an unadmixed population rather than a bottlenecked one, but that may be the case.

Although on the other hand, when I ran f3 statistics with the publicly available datasets I could get, to check if North Eurasian populations came up as a mix of Dai/Han and Karitiana, the results did not indicate admixture.

Maybe one possibility would be that a population cline existed in North East Asia between Amerind type people, who spread West to Europe and East to the Americas, and East Eurasian type people to the south. Then Amerind type people in the north might have basically either died out or completely migrated elsewhere to be replaced by people who were more of a south-north mix on the Amerind-East Asian cline, without actually being an admixture of Amerind-East Asian groups. Thus explaining why the present day north eurasians don't show as Amerind-East Asian admixtures by f3 stats.

But then it seems fanciful to me that these North Eurasian Amerind type people could "die out" like this... or that they could all migrate out either to Europe or America...!

AWood said...

The biggest surprise for me is 0% Amerind among the ancient Spaniard. I was nearly certain this was going to show up.

AWood said...

@mooreisbetter

I wouldn't really define Sardinians by I-M26, since levels of R1b1b2 are right behind it, followed way behind by G2a(3b1a?). It's some type of bottleneck effect that I-M26 happened to win out, where everywhere else it didn't, including in Spain where it also has appreciable frequencies.

jackson_montgomery_devoni said...

@Matt,

I do not really understand what you mean by this?

'' Then Amerind type people in the north might have basically either died out or completely migrated elsewhere to be replaced by people who were more of a south-north mix on the Amerind-East Asian cline, without actually being an admixture of Amerind-East Asian groups. Thus explaining why the present day north eurasians don't show as Amerind-East Asian admixtures by f3 stats.''

Why does ADMIXTURE show Siberian populations as being mixed between Siberian and Amerindian components then?

Matt said...

@j_m_d

Basically, the idea I was trying that, since present day Siberians show as mixtures of East Asian and Amerind on Admixture, but don't show up as admixed on f3 statistic comparisons of East Asian and Amerindian, then they're probably a populations which branched off between East Asian and Amerind (in some tree topology), and then replaced an Amerind-like population living in Siberia, rather than a mix of an East Asian-like and Amerind-like population.

In other words, I'm just saying that I would think on the basis of f3 (presuming I'm using the software and samples correctly) that the Siberian demographic change is from Amerind-like to populations more intermediate between Amerind-East Asian without any mixing between Amerind-like and East Asian-like populations - more through total replacement than partial replacement with mixture.

Mark D said...

"First, they are clearly unique genetically and isolated. Luigi Cavalli-Sforza noted that two decades ago. He said the best way to think of Sardinians is almost as a their own race! Almost like a sister group of Europeans. That is how unique they are."

I think everyone should read up a little on Sardinian history. Even a cursory reading of Wikipedia shows that the island received "waves of emmigration" from prehistoric times through the settlements by Carthage and Rome. It would appear the island was a destination for many groups both from Iberia and the present-day Italy. Now, does the resultant mixture of emmigrants relate closely to what a Neolithic admixture should look like? Possibly. But that doesn't make Sardinians any more "isolated" or "genetically unique" than any other group of peoples in southern Europe.

terryt said...

@ Matt:

"Although on the other hand, when I ran f3 statistics with the publicly available datasets I could get, to check if North Eurasian populations came up as a mix of Dai/Han and Karitiana, the results did not indicate admixture".

Have you tested the possibility that Karitiana might be a product of admixture between North Eurasian populations and Dai/Han populations? In other words the Amerindians are possibly themselves the product of admixture. That could be an explanation for the present absence of Amerind type people in Northern Eurasia.

jackson_montgomery_devoni said...

@Matt,

Ahhh okay now I see what you are saying thanks for the explanation.

Would any of this apply to some Northeast European groups such as Finns as they are thought to have some ancient Siberian admix? Or is the Amerindian component found in Finns the same as that found in other Europeans just at a higher level?

SB said...

" It is interesting to see that South Indian Brahmins (both the Metspalu et al. sample, and my Iyer_D and Iyengar_D samples) lack this admixture, while Uttar Pradesh Brahmins do not, given the rolloff evidence for a more recent admixture of the latter. This is consistent with a historical admixture event, after the migration of Brahmin groups southwards, as described in that post."

Dienekes, it is also important to remember that Uttarpradesh/Uttaranchal Brahmins possibly have Himalayan/Tibetan like admixture which is not the case with South Indian Brahmins.

eurologist said...

But then it seems fanciful to me that these North Eurasian Amerind type people could "die out" like this... or that they could all migrate out either to Europe or America...!

Matt,
It seems apparent that during the height of LGM much of Siberia (but not the coastal regions and Beringia) were too cold and too dry to support humans. And whatever people from few refugia allowing survival in southern pockets were then simply outnumbered post-LGM by East-Asian people closer to typical extant populations.

I'm very surprised about Bra1 showing no connect with the more northern neolithic HGs via the "Amerindian"-component.

HW,

I have no problem with the notion that N European Mesolithic populations were already somewhat different from those in SW and S Europe. First of all, some cline could have easily been supported even during LGM refugia, and secondly, surviving SW Siberian and south Ural populations would have tried to expand into opening new habitat in the west post-LGM. I see potential signs for this in the Hamburg and Ahrensburg cultures.

Annie Mouse said...

The Amerind component most likely came from the West not the East. It is absent in central and East Asian samples. I think we need to be thinking about the mammoth hunters as occupying a circlet encompassing Asia and North America. Not a torc with a gap across the north Atlantic.

terryt said...

"I think we need to be thinking about the mammoth hunters as occupying a circlet encompassing Asia and North America. Not a torc with a gap across the north Atlantic".

Seems the most likely explanation to me.

Onur said...

The Amerind component most likely came from the West not the East. It is absent in central and East Asian samples. I think we need to be thinking about the mammoth hunters as occupying a circlet encompassing Asia and North America. Not a torc with a gap across the north Atlantic.

Which "Amerindian" component are you talking about? K=4 or K=10?

Slumbery said...

Annie Mouse

"The Amerind component most likely came from the West not the East. It is absent in central and East Asian samples."

It is not absent from the Central Asian samples.
Uzbeks: 8,6%
Hazara 7,9%
Uygur: 8,8%
Tajiks: 6,4%
Turkmens: 3,9%
Kyrgyz: 12,5%
+ Altai: 15,7%

These are all European or even above levels. We do not have anything from Kazakhstan and the Turkmens are the only ones under the typical European, but even they have about the same level as Italy. So I frankly do not see were you getting the idea that it is absent in Central Asian samples.

As for East Asia, the K4 Amerindian is abundant anywhere north from the Han Chinese, and even south from this "line" still well present in various ethnic groups. Even Japanese is 4,9%, that is about the same as Spanish. (Also it is not completely absent even in Han Chinese.)

So your above statement is just not true.

Also you should make some reality check. The Eastern connection is well documented in archaeology and even in linguistic, while the traces of trans-Atlantic connection are quite lacking...

SimonW said...

@Sisophon

It seems reasonable to assume that this Amerindian component was - at least initially - associated with y-Haplogroups N, I1 and R1a.

Of course, R1b completely predominates in the British Isles, but in Ireland, Scotland and Orkney I1 comes right behind it, reaching levels of 6-9%. I would therefore assume a Scandinavian influence, possibly predating the Viking incursions.

terryt said...

"The Eastern connection is well documented in archaeology and even in linguistic, while the traces of trans-Atlantic connection are quite lacking..."

Agree. I misread Annie's comment, 'Not a torc with a gap across the north Atlantic'. I think it is a 'torc' with a gap across the north Atlantic. Any gap in northern eurasia is a product of later movement north of populations containing Y-DNAs N and C3.

"It seems reasonable to assume that this Amerindian component was - at least initially - associated with y-Haplogroups N, I1 and R1a".

Unlikely to be any of those as none have been found in America. Of course there could be some connection with R as it is related to Q, a common American Y-DNA haplogroup.

SimonW said...

@terryt
Sorry, I should have precisised: at least initially and as far as Europe is concerned. Correct me if I'm wrong, but I doubt that R1b-M269, stemming perhaps from somewhere south or west of the Caspian sea could have been high on this boreal HG component.

SimonW said...

I just checked the spreadsheet: Lezgins have got 4,6% Amerindian, so maybe there's something to your idea. They are on the northern side of the Caucasus mountains, though. And the two Ajv individuals from Gotland were hardly R1b.

SimonW said...

I'm terribly sorry for triple posting again, my thoughts came only drop-wise...

One thing I would like to add: there's clearly a north-south distribution of the Amerindian component in western Europe: from >7% in Ireland, Argyll and Orkney, over 6-7% in England, the Netherlands and France, to 5% in Catalans and <5% in the rest of the Iberian peninsula. This pattern doesn't parallel the incidence of R1b.

Looking at the whole picture of the entire Europe the gradient becomes quite clear:
from >10% in the northeast (Finns, Russians, Mordovians, Chuvash)
-> 8-10% in the adjacent north and northeast (Norwegians, Swedes, Lithuanians, Belorussians, Poles, Ukrainians)
-> 7-8% in the adjacent central Europe (Germany, Hungary) and remoter parts of the British Isles (Ireland, Argyll, Orkney)
-> 6-7% in adjacent parts of central and western Europe (Netherlands, Austria, France, England)
-> 5-6% in adjacent parts of southeastern and southwestern Europe (Romania, Serbia, Catalonia)
-> 4-5% in adjacent parts of southern Europe (Spain, Portugal, northern Italy, Bulgaria)
-> 3-4% in adjacent parts of Italy (Tuscany, central Italy)
and so on...

It spreads from the northeast counterclockwise to the south.

Nirjhar007 said...

''it is interesting to see that South Indian Brahmins (both the Metspalu et al. sample, and my Iyer_D and Iyengar_D samples) lack this admixture, while Uttar Pradesh Brahmins do not, given the rolloff evidence for a more recent admixture of the latter. This is consistent with a historical admixture event, after the migration of Brahmin groups southwards, as described in that post.''
It is great to see that a different path is leading towards the same direction.