search this blog

Monday, June 29, 2015

K8 results for selected Allentoft et al. genomes

Running the ancient genomes from Allentoft et al. in ADMIXTURE is proving a challenge. The reasons for this are covered in the paper; see section 6 in the supp info. Fortunately, I managed to obtain robust outcomes using my K8 model for ten samples representing almost all of the main archeological groups studied by the authors.

The spreadsheet with the new results is here. The full K8 spreadsheet is here. Below you can see a Principal Component Analysis (PCA) produced with the data. Click on the image to go to my drive, where you can download a high quality copy.

Some of the outcomes might look unusual, but as I say, they are solid. I double checked each one with an unsupervised ADMIXTURE test and genotype-based PCA. For instance, the genotype-based PCA featuring RISE247 and RISE479, respectively, look like this:


Allentoft et al., Bronze Age population dynamics, selection, and the formation of Eurasian genetic structure, Nature 522, 167–172 (11 June 2015) doi:10.1038/nature14507

See also...

Bell Beaker, Corded Ware, EHG and Yamnaya genomes in the fateful triangle


Krefter said...

4mix results using new Allentoft results.

Tuscan using Remedello Italian, LN/BA Central European, and East Mediterranean.

26% Remedello + 32% Bell_Beaker_LN + 33% Cypriot + 9% Cypriot @ D = 0.0073
31% Rem + 29% Bell_Beaker_LN + 3% Syrian_Jewish + 37% Syrian_Jewish @ D = 0.0166

21% Rem + 32% Unetice_EBA + 22% Cypriot + 25% Cypriot @ D = 0.0038
27% Rem + 29% Unetice_EBA + 14% Syrian_Jewish + 30% Syrian_Jewish @ D = 0.0147

27% Rem + 30% Halberstadt_LBA + 14% Cypriot + 29% Cypriot @ D = 0.0052
33% Rem + 27% Halberstadt_LBA + 39% Syrian_Jewish + 1% Syrian_Jewish @ D = 0.0148

4% Rem + 49% HungaryGamba_BA + 0% Cypriot + 47% Cypriot @ D = 0.0054
12% Rem + 44% HungaryGamba_BA + 2.99999999999999% Syrian_Jewish + 41% Syrian_Jewish @ D = 0.0161

Lithuanian using Corded Ware Estonia, WHG/SHG, and MNs.

74% CWC_Est + 18% SHG + 0% WHG + 8% Gokhem_MN @ D = 0.0146
74% CWC_Est + 18% SHG + 0% WHG + 8% Baalberge_MN @ D = 0.0144
73% CWC_Est + 19% SHG + 0% WHG + 8% Esperstedt_MN @ D = 0.0145

Annie Mouse said...

Well from my biased point of view.

EN, MN and Remedello (Italy) look like Sardinians stretched out with a bit of Western Hunter Gatherer or something similar. Fits with an early Sardinian-like population in Europe. The Sardinians do not look related to Turks or the Near East at all and neither do EN, MN or Remedello.

The Bronze age folk have had a massive tug to the right by something Yamnaya-like. Both the Corded Folk and Bell beaker folk look like they are responsible for this. I favour the BBs coming up from Spain but I am not seeing this here.

Modern Southern Europeans (Italy?) and to a less extent SE Europeans (Greece?) have had a big tug towards some part of the Near East that is NOT Turkey or Armenia. My best guess is this is folk flowing across the mediterranean from Tunisia or possibly around the mediterranean bipassing Turkey. It looks recent (well after the Bronze Age anyhow) as there is no evidence of it in Remedello.

I am conscious that other than REmedello we do not have any southern European samples, and a big part of the story could be missing.

Ariele Iacopo Maggi said...

"It looks recent(well after the Bronze Age anyhow)"..... if recent is 4000y ago, yes it is recent. J2 and E1b1 have been found in early neolithic samples, and we can speculate that those people carried east-med type of admixture (Yamma also had west asian and that would complete the puzzle for southern europe); And yes some later influx from west asia and north africa is obvious(roman slaves, phoenicians, sea people, arabs for sicily and spain)but the extent of the influx is not as large as the neolithic one. (Jews resemble pre-arab ancient levant in my opinion)

Alberto said...

Thanks, David, quite interesting results.

The Afanasievo sample is obviously similar to Yamnaya, but slightly more "pure ANE" admixed. The D stats put it closer than BaYam to BaArm, even though Yamnaya has 5% more ENF and 5% less ANE. No luck with the BA Armenian sample to get any clue about how it looks like?

Sintashta also very interesting. More WHG and less ENF than CW. Very European to be in that place.

The BA Hungarians very different. One a farmer, the other one basically like BR1. Any guess as to where could these types (BR1, RISE479) come from?

Davidski said...

The RISE479 Hungarian has even more WHG than BR1. I haven't a clue how people like this reemerged in the Carpathian Basin during the Bronze Age. I suspect they might have come from western Ukraine.

I'm getting garbage results with the K8 for the BA Armenians. The SNP overlap is just too small.

Alberto said...

Yes, Western Ukraine is a good suspect. If HGs from that area were more WHG/SHG than the ones from Samara/Karelia.

Mike, but with that Lepenski Vir option you are suggesting that these people were in the area from Mesolithic times and simply didn't show up in a more visible area/culture till the middle Bronze Age?

I do think this is possible too. It's basically what I was saying in a previous post. At this time (LN/BA) it appears that a lot of WHG ancestry comes out from everywhere. It could not come from farmers and it could not come from Yamnaya (both had less). So these people had to be there around, just "hidden" from the more visible sedentary cultures.

Mike Thomas said...


I modified what I said, because I realize Lepinski Vir is true for the Neolithic, but not late neolithic and early Bronze Age (becuase it had disappearaed by then). I suspect this 'hidden WHG' or 'HG resurgance' is perhaps more related to incomplete sampling of Neolithic groups. But yes, surely some 'cryptic foragers' mat have existed. In fact, several Polish scholars (e.g.) point to existence of Forager groups into the Bronze Age !

Mike Thomas said...

Alberto/ Dave:

To me, what the results show is three-fold:

1) Hungary: a significant continuity of LN populations admixing with some ~ 30% new eastern populations.

2) Germany: a MN population mixing with 50-65% CWC -like population.

3) Poland & other CWC territory: we have no Neolithic population group to compare with, so hard to say anything definite. But either, 1) there was an almost whole scale replacement of LN groups (perhaps extinction then 'pull' migration), or, (2) the LN had a large component of 'acculturated' Mesolithic groups with pre-existing EHG admixture.

4) Italy: essentially a continuation of Neolithic groups into Late Copper Age. (late M3), which is pretty significant .

5) Spain ? Britain? central-south Balkans ?

June 29, 2015 at 5:23 AM Delete

Nirjhar007 said...

I guess the Andronovo have too much WHG.....

Alberto said...

4mix of the very WHG Hungary BA sample as EN + Motala:

baHu = 52% Motala12 + 0% Spain_EN + 29% LBK_EN + 19% HungaryGamba_EN @ D = 0.0131

A bit worse as WHG + Caucasus:

baHu = 57% HungaryGamba_HG + 12% Armenian + 31% Georgian_Imer + 0% Lezgin @ D = 0.0151

Much worse as EHG + MN:

baHu = 35% Samara_HG + 65% Spain_MN + 0% LBK_EN + 0% HungaryGamba_EN @ D = 0.1077

Best fit as EN + Motala + bits of West Asian:

baHu = 42% LBK_EN + 53% Motala12 + 4% Armenian + 1% Bedouin @ D = 0.0054

andrew said...

I suspect that the R1b is from the early Nordic Bronze Age, and that the R1a is from the late Nordic Bronze Age, dividing around 1200 BCE. The former was probably Vasconic. The latter was probably Germanic.

Grey said...

If the Hungarian plain was particularly suited to raising cattle and horses (?) then perhaps HGs hiding up in the mountains were recruited to help herd them (assuming HGs can be turned into herders much easier than farmers).

If so the HG percentage might vary with local geography and proportion of cattle/horse raising.

Krefter said...

"I suspect that the R1b is from the early Nordic Bronze Age, and that the R1a is from the late Nordic Bronze Age"

The R1as are older and Corded Ware-like autosomally. The R1bs are younger and North Sea-like autosomally.

Romulus said...

Wow I thought EHG was similar to WHG/SHG but I guess not.

Davidski said...


Yamnaya is definitely a clade with Afanasievo to the exclusion of BA Armenians.

Here are the tests that Allentoft et al. ran to check whether Yamnaya was a clade with Corded Ware to the exclusion of the BA Armenians. It was, and I get very similar results, even though they used the 1000 Genomes dataset and I'm using the Human Origins.

Yoruba RISE_baArm RISE_baYam RISE_baCw 0.0008 0.258
Yoruba RISE_baCw RISE_baYam RISE_baArm -0.0197 -6.083
Yoruba RISE_baYam RISE_baArm RISE_baCw 0.0205 7.003

I ran the same set of tests to check whether Yamnaya formed a clade with Afanasievo in respect to the BA Armenians. This looks like a resounding yes to me.

Yoruba RISE_baArm RISE_baYam RISE_baAfan 0.0045 1.424
Yoruba RISE_baAfan RISE_baYam RISE_baArm -0.0393 -12.684
Yoruba RISE_baYam RISE_baArm RISE_baAfan 0.0439 14.016

To further check whether these results are legit, we can test whether Yamnaya form a clade with Afanasievo to the exclusion of Corded Ware. It does, and the results are more convincing with the younger Corded Ware samples from Allentoft et al., which makes sense.

Yoruba RISE_baCw RISE_baYam RISE_baAfan 0.0052 1.759
Yoruba RISE_baAfan RISE_baYam RISE_baCw -0.017 -5.832
Yoruba RISE_baYam RISE_baCw RISE_baAfan 0.0222 7.753

Yoruba Corded_Ware_LN RISE_baYam RISE_baAfan 0.0002 0.07
Yoruba RISE_baAfan RISE_baYam Corded_Ware_LN -0.0096 -2.96
Yoruba RISE_baYam Corded_Ware_LN RISE_baAfan 0.0098 3.198

In fact, Extended Data Table 1 in the study says "Yamnaya and Afanasievo form a clade to the exclusion of other ancient groups". This fits the ADMIXTURE results in the study and here.

Tobus said...

EHG have lots of ANE and are much closer to MA1/Karitiana than WHG/SHG are.

Alexandros said...

@Ariele Iacopo Maggi
J2 and E1b1 have been found in early neolithic samples

In which early Neolithic sample(s) was J2 found? Can you please provde any evidence?

Alberto said...


Yes, sure, Yamnaya and Afanasievo are obviously very similar, and pretty different from the BA Armenian. Maybe I expressed myself incorrectly. What I meant is this:

Yoruba RISE_baArm RISE_baYam RISE_baAfan 0.0045 1.424

BA Armenians (and maybe more so if they just used the most ancient samples and not mix all of them) are closer to Afanasievo than to Yamnaya. But Afanasievo has 5% less ENF and 5% more ANE.

Even if the difference is small and the stat hardly significant, it makes me wonder a bit, but not to extent and repeat myself about the Central Asian origin, etc... I'll wait for more samples and tests.

Simon_W said...

The strange input in BA Hungary is most definitely from the Yamnaya there and from the western Ukraine, as it had ANE, which neither KO1 from Gamba nor the new Vinca sample with inflated WHG had.

Modern NW Europeans in this K8 PCA are on a cline between Yamnaya/Corded Ware and MN Farmers with a little bit of extra WHG. They are not on a cline between Yamnaya/Corded and BA_Hungary RISE479. Thus what distinguishes modern NW Euros from Yamnaya/Corded Ware is most of all more farmer ancestry. It seems reasonable that MN farmers in Britain had a bit more WHG than those in Germany, as Britain is on the northwesternmost end of the Neolithic expansion and was reached by farmers relatively late. (Though of course the NW Europeans in the PCA are not exclusively British.) And the Bell Beakers from Germany are exactly on the same cline, so the explanation for their DNA is the same. If you draw a line from Yamnaya/Corded Ware to BA_Hungary RISE479, the Bell Beakers are clearly off that line. So it's dubious that the Hungarian Yamnaya played a significant role in the formation of the German Bell Beakers. Though IMO their influence is present in the Czech Bell Beakers.

Simon_W said...

@ David

Is there anything meaningful that can be said about the LBA Lithuanian sample or is its quality hopless? Its origin and date would suggest that it was probably a Sudovian, and therefore belonged to the Western Balts, unlike modern Lithuanians who are eastern Balts.

Richard Rocca said...

@ David,

Can you add the Bell Beaker RISE563 sample to your PCA? The sample is P312+U152+ and is from a site that, with isotope testing, showed a 37.5% immigration rate. It will be interesting to see if he is pulled away from the rest of the Bell Beaker samples.

Davidski said...

Yes, he looks surprisingly eastern on the genotype PCA.

I'll try the K8 on him later today.

Davidski said...

Simon, I can't get anything coherent out of that Lithuanian sample. I'll try the K8 on him later today, but I'm not sure how that will go.

Richard Rocca said...

@ Davidski said...
Yes, he looks surprisingly eastern on the genotype PCA.

I'll try the K8 on him later today.

Holly crap, I was thinking he would be shifted more west than the other BB samples, not that far east! Very, very interesting to say the least.

Chad Rohlfsen said...

Yeah, that's one of the two that look like Corded Ware. There's another too. I can't remember the number. Both were about 65% Yamnaya in an unsupervised run.

Krefter said...


Are any of the Beakers very EEF?

Chad Rohlfsen said...

There was a Beaker and a Corded sample that were very French like. It might be a quality issue though. I'm trying to add more samples to my K10. They're all pretty low quality though, as many of the Rise samples pull 1-3% SSA noise. I've pretty much had to scrap all of their samples for that reason. I've only got about 15 of the 101 left. It could become less, as they make my Yamnaya component disappear.

Krefter said...

Do the Bronze age Montenegrons come out Spanish-like in all tests? Are they for sure distinct from modern SouthEast Euros despite low coverage?


What are the implications of Spanish-like people so far SouthEast in Europe in the Bronze age(I think they're late Bronze)? Could it be there was more diversity in that region back then: Some being Greek-like, some Spanish-like, and later everyone became more Greek-like. Could it be there was a lot of gene flow from East Mediterranean/Near East into SouthEast Europe?

That's the case with Bronze age Hungarians. If we didn't have Bell Beaker/Unetice samples we could argue *All* of Central Europe was French-like.

Mike Thomas said...


My 2 cents is Im not sure yet what to make of it. The Balkans is a diverse place, and perhaps always has been, so id like to see samples from all 4 of its corners, and over different periods. *But if BaMOn lady is representative* of Bronze to pre-Roman Balkaners, then it merely confirms what we already suspect from historical sources : there was a rather significant population shift between Bronze Age and pre-modern Era. Clearly this was eastern. Dave suggests it was more southeastern (ie West Asian) than Northeastern. However, I'd like to know how he defined his modern SEE cluster (ie did it exclude Greeks? ). He attributes it movement between West Asia and the balkans during Iron Age times, and during the Roman/ Byzantine empire period. Im sure this played a role. But the most significant impact must have been from the Slavs, who were clearly a northern (relatively) population, and not a West Asian one.

What was your point about Ba Hung?

Krefter said...

"What was your point about Ba Hung?"

If all we had from Central Europe between 2000-1000 BC were from Hungary, we could argue all of Central Europe was like them. Hungarians today cluster in the middle of the thick forest European samples create on PCAs. We could argue, "If Hungary was French-like from 2000-1000 BC, that means all regions closely related to Hungary today were French-like back then."

The French-like(note exactly) and people with crazy amounts of EEF or WHG from Bronze age Hungary were mostly replaced by Unetice/Bell Beaker-related people. Maybe in Monetnegro there were Spanish-like people, and in nearby regions Greek-like people. After 1000 BC the Spanish-like people could have been replaced by Greek-like people. So, today everyone is Greek-like, just like how in Hungary everyone is Unetice/BBC-like.

Mike Thomas said...

yes I see
I'm not sure , but i don't see modern Balkans as "Greek like". They also look Unetician; but simply with more Palaeo-Balkan and some west asian admixture. I'd picture Bronze Age greece being very near eastern, with modern Greeks (esp from northern mainland) bein ECE admixed. But im speculating in absence of hard data.

Certainly Id think the "replacement" in Hungary was a gradual process, with arrival of Cimmerian types (IR1) from Iron Age to various other steppe people's to Middle Ages..

Frank said...


if you remember we had an uncontaminated iron age thracian that was tested some time ago. He came out as quite near eastern in all tests, both by eurogenes, geneticker and also on forumbiodiversity. He was more southeastern than modern greeks every way one looked at it.

The montenegro sample might not completely even represent the balkans in that era. Just previously, a clearly mesolithic-like sample from serbia was tested here and then, from around the same time frame, a neolithic-like one from the same general area.

The northern balkans are full of highlands and hard terrain. Montenegro and northern albania to it's immediate east are basically highlands with little to no valleys. It could be pretty easy for such a group to survive there

Simon_W said...

I've always been of the opinion that Bronze Age migrations from Cyprus and Anatolia westwards had an important effect on Greece, Sicily and southern parts of Italy, but now I'm also open to ascribing the West Asian impact in these areas to the Hellenistic and Roman period.

It's interesting that according to

Northern Italy received some considerable (33%) Cypriot-like admixture around 66 BC. The Seleucid empire was crushed by the Romans in 63 BC, after which lots of slaves were carried to Italy. And the Romans had lamented about the lack of manpower on the fertile soils of the Po plain after they had conquered it from the Gauls. Originally I had thought Cypriot-like means most of all Roman citizens from southern Italy, but actually the above admixture map sees stronger mixing coefficients with Syrians and Jordanians, South Italians are not even mentioned as an admixture source.

I've always been sceptical about the claim that slavery sort of changed Italy in any particular way, because slaves were from everywhere, not from one particular part of the empire. But then again, slaves from Sardinia or Gaul wouldn't have added anything new, while slaves from the Seleucid empire and Northern Africa would.

Mike Thomas said...

@ Frank

You're of course correct. that's why I said we need samples from everywhere in the Balkans, and from all periods :)

And I certainly remember the thracian sample. But despite its purity, it had low amount of SNPs. If the earlier analysis of it are correct, it certainly adds much complexity to the overall history of Europe.

@ SImon W;

"The Seleucid empire was crushed by the Romans in 63 BC, after which lots of slaves were carried to Italy"

Sorry, but I find this idea a massive stretch. Apart from being tired and overused (the idea), how can slaves make a genetic impact on the genetic profile of 1st century Italy, and indeed much of EUrope ? Apart from needing hundreds of thousands in number, you'd also have to assume that males slaves were allowed free and frequent intercourse with female Roman superiors. Barring the odd rendezvous between slave and master, i think it not possible.

Frank said...

The thracian showed very strong near eastern affinities in all tests. Low SNPs cannot explain that.

Furthermore, I maintain that southeastern europe was quite near-eastern like since a late neolithic-late bronze/iron timeframe.

That green "southwest asian" component that we used to see back then in K=6 analysis is what it probably refers to. (the one that the thracian also had in a good amount) Geography wise, greece and bulgaria are far more connected to anatolia rather than the balkans.

Simon_W said...

@ Mike

I didn't speak of Europe in general, just of Italy, because it was the main importer of slaves. Naturally, the Romans took most of the cheap labour for themselves, their economy was dependent on that. According to historical estimates at times about 1/4 to 1/3 of the population of Italy were slaves. And we don't need to assume that it had to be male slaves approaching rich Roman women. Male slaves also had daughters, and eventually many slaves were freed.

Simon_W said...

Of course there were also women sold into slavery...

Simon_W said...

Some Italian sources claim that the majority of Slaves were from Italy. But that's most certainly at best true for the early republican Rome, when Rome's expansion was confined to Italy. It's a stretch of mind to think it was still that way in the early Imperial period.

Simon_W said...

But admittedly, we need to see Bronze Age and Iron Age autosomal DNA from all parts of Italy to arrive at a definite conclusion about the effect of slavery, and I can't wait to see it. I've heard rumors that there's something on the way, hopefully it won't take too long.

Mike Thomas said...


"Furthermore, I maintain that southeastern europe was quite near-eastern like since a late neolithic-late bronze/iron timeframe."

Yeah, I think its possible. Im not sure about Iron Age..

" Geography wise, greece and bulgaria are far more connected to anatolia rather than the balkans."

I don;t understand. Greece and Bulgaria are in the Balkans- they're the southernmost and eastern aspects, resp. And how separate and isolated due you suspect them to be , mountains or not ?

Contrary to your straightforward division, I'd consider more complex divisions. Eg northern Bulgaria - the Danube plain is likely to show close links to the steppe. Thrace propper - south of the haemus - will be similar to Anatolia. The plains of Macedonia and Thessaly will surely have their own idiosyncracies, perhaps having links to Thrace and Anatolia, Central Europe- Carpathian basin, and the Adriatic. Island Greece very Near Eastern. ..

Frank said...

but how separate and isolated due you suspect them to be , mountains or not ?

In tbe neolithic and bronze age? Quite a bit. Even during the medieval era, the byzantines declared it impossible to settle thematic soldier-farmers there due to the nature of the harsh terrain and the lack of arable land.

Meanwhile, we have a ton of examples of armies or migrations crossing the hellespont or the aegean. It is far easier to get to thessaly from troy than from thessaly to montenegro with a large group.

Mike Thomas said...


"In tbe neolithic and bronze age? Quite a bit. Even during the medieval era, the byzantines declared it impossible to settle thematic soldier-farmers there due to the nature of the harsh terrain and the lack of arable land.'

Having unruly neighbours politically isn;t a function of the propensity for genetic admixture. Sure, terrain played crucial factor, but this was exingent on military strategy and sheer boldness of the northern zhupania. If you don;t believe me, check the DNA of medieval Byzantine period, and it;'ll be almost similar to today: from the outset (9th century).

But basically I agree with you - I expect the Balkans to be diverse in the Bronze and Iron Ages.

Ariele Iacopo Maggi said...


1/3 of the population of Rome, not certainly of Italy

Simon_W said...

@ Frank

Are you sure you mean one of the Iron Age Thracians? As far as I remember there were four samples from Bulgaria, in chronological order:

V2 from a late Bronze Age flat grave: He's perhaps the most complete of the lot and he had about 44% West_Asian in Eurogenes K15, which is almost at Caucasus-level.

P192-1 from an early Iron Age flat grave: He's rather noisy, with 17% West_Asian, with unusually high 21% Red Sea, but also with 3.5% Sub-Saharan (probably noise).

T2G2 from an early Iron Age kurgan, very noisy.

K8 from a late Iron Age kurgan, came out rather Russian-like, Dave suggested he might be contaminated.

Simon_W said...

@ Ariele

What makes you so certain? According to wikipedia (drawing from three different sources): "Estimates for the prevalence of slavery in the Roman Empire vary. Estimates of the percentage of the population of Italy who were slaves range from 30 to 40 percent in the 1st century BC, upwards of two to three million slaves in Italy by the end of the 1st century BCE, about 35% to 40% of Italy's population."

I'd say presumably there wasn't everywhere the same density of slaves, but I can't see why they should have been concentrated only in the city of Rome. Presumably they were most of all to be found where there was a lot of work to be done. And Roman citizens owned land in various parts of Italy, and in the first century BC all free Italians became Roman citizens anyway.

Simon_W said...

@ Andrew

RISE94 from the Swedish Battle Axe culture had R1a before the Nordic Bronze Age even had started... It's strange you suggest that R1a arrived with the late Nordic Bronze Age after 1200 BC.

Davidski said...


Some of these BB samples don't have much to offer in terms of data, but...

RISE559 - could be Iberian, North Italian or French
RISE560 - very Scandinavian/Baltic
RISE562 - looks British
RISE563 - very eastern, could pass for Corded Ware
RISE564 - looks Iberian or significantly EEF
RISE566 - looks British
RISE567 - could pass for Corded Ware
RISE569 - shows a clear eastern vibe, but not quite in the Corded Ware range

Do you have dates for all the Bell Beakers from Haak and Allentoft? It might be interesting to see how they move around on a plot when sorted chronologically.


The Lithuanian BA sample is very similar to Lithuanians, except it's a little more northern, suggesting more HG ancestry.


I might try some formal stats on those Montenegrin BA samples later this week. The K8 doesn't work on them.


The I1 Scandinavians cluster with modern Scandinavians, more or less.

Simon_W said...

@ Ariele

Regarding J2 in the early Neolithic: Yes, it was recently found in Sopot and Lengyel in Hungary. It's not exactly early Neolithic, but still rather long ago. But we don't need to speculate about the autosomal make-up of these people, because NE7 from Gamba et al. is from the Lengyel culture. He doesn't have the K15 West_Asian component, but he does have quite some East_Med (15%) and a little bit of Red_Sea (1%). But on the whole he doesn't differ much from other early/middle Neolithic Europeans. East_Med and Red_Sea were in fact much stronger in the early Neolithic cultures like Starcevo and LBK. The appearance of J2 in Lengyel doesn't seem to have resulted in a significant autosomal change nor to explain the bridge between the clines in the K8 PCA.

Simon_W said...


Thanks, good to know.

Davidski said...

Here's the most Iberian of the Bell Beakers. But this might just be German EEF ancestry talking.

K8 baBb_RISE564
ANE 0.066929
ENF 0.493048
WHG 0.417518
Oceanian 0.022465

Genotype PCA

Simon_W said...

Looking at the K8 PCA, I'm more convinced than ever that the teal component in the Caucasus didn't result from a migration of teal people from south central Asia through Iran. Pashtuns followed by Tadjiks are the most eastern ones of the southern cline, because of strong ANE. Then the Caucasians follow. These populations are also the ones with strongest teal – it's directly related with ANE affinity on the southern cline. Iranians and Kurds however are more to the left. On the whole the impression is one of ANE coming from the north (this doesn't exclude central Asia IMO) and surviving especially in these populations, which resulted in a strong teal component around the Hindukush and the Caucasus – with less teal inbetween.

Also interesting to note that judging from the K8 PCA Yamnaya are clearly a mix of EHG and Caucasians, nothing else.

Nirjhar007 said...

Any News on the Parkhai genome? or is it dead?....

Richard Rocca said...

Davidski said...
Here's the most Iberian of the Bell Beakers. But this might just be German EEF ancestry talking.

Thanks David. I am in the process of sending a communication to some of the collaborators of the paper that no doubt provided the BB samples. The disparity between RISE563 and RISE564 is even more interesting because they are from the same site. If the authors can tell us which is the "local" and which is the "immigrant" as per their prior Sr isotope study, I think that will tell us a lot.

Frank said...


yes, I was talking about P192-1. He was the only one presumed to be a native rather than a IE noble and in all the tests I remember he came out as every near eastern. Something like 70% Tuscan + 30% Georgian. There was another, more detailed breakdown in ABF that suggested it had very clear "middle eastern" ancestry.

Anyway, I suppose we gotta wait for more samples. We'll see.

Grey said...


"What are the implications of Spanish-like people so far SouthEast in Europe in the Bronze age(I think they're late Bronze)?"

Depending on the date sea peoples?

truth said...

I think the fact they looks spanish-like is just because of similar composition (WHG+EEF+ANE proportions being similar to that of iberians, but doesn't mean it came from Iberia to the Balkans.

Chad Rohlfsen said...

It could all just be dreadful overlap on a few specimens. There was even a Corded Ware that looked SW European.

Davidski said...

Lack of markers are a problem for some of the tests, but I'm using all of the available SNPs to run the genotype PCA, so if they create the right shape and aren't too fuzzy the results should be correct.

And one of the Corded Ware samples does look very EEF. You can see how it sticks out in the ADMIXTURE analysis in the paper.

Mike Thomas said...

"And one of the Corded Ware samples does look very EEF."

Perhaps a surviving MN "farmer" in the process of being integrated...
Such Not yet sampled finds mihjt account for the apparent decrease in "pure CWC" derived ancestry from MBA onward ?

Davidski said...

Yes, and I haven't checked this, but it's probably a female.

Mike Thomas said...

It's interesting . Male farmer lineages mostly became extinct. Female lineages continued. Mesolithic female lineages from Central Europe became mostly extinct, but Y lineages continued.

Krefter said...


By the way I sent you an email( with a DNA file I'd like you to find the Ancestral origin of. I'll send more later if that's okay. They're from openSNP and in 23andme format.

Cossue said...

"It could all just be dreadful overlap on a few specimens. There was even a Corded Ware that looked SW European." Why 'dreadful'???

Simon_W said...

Amazing disparity between these Bell Beaker samples even at the same location.

@ Frank

I think the late Bronze Age Bulgarian V2 with his very strong West_Asian component is quite instructive. Really seems to be in line with a significant population wave from Anatolia to southeastern Europe in the Bronze Age. And this had in all likelihood affected Greece, too. It didn't reach Montenegro, which is more secluded and to the west.

But whether these West Asians carried an IE language with them or something else is hard to tell. The BA Montenegrin was probably an Illyrian, speaking a Centum language. The West Asians might have spoken something akin to Eteocypriot or, if they were relatives of the Armenians, Greek. Since late Iron Age K8 looks different and more northern, we cannot say that the incoming West Asians were Thracians.

Simon_W said...

It's a pity we have no dates for the Allentoft Bell Beakers. Those from Haak et al. seem to follow the pattern that the earlier ones (I0108, I0111, I0112) are more Northwest European-like, some of the later ones (I0806, I0058) more eastern, but this might be a fluke.

Ariele Iacopo Maggi said...


From what i understand you want to explain high caucasus and east-med admixture in italy with a massive influx of genes from the roman era slavery. I'll try to counter this theory.
1) Those Remedello samples are too early to say anything really, they were not IE 2) late iron age italy was probably dominated by anatolian (Etruscans like) and Italic (central europe like) groups 3) in the time frame between Remedello and the roman empire there is plenty of time for a migration from the levant and anatolia of a pre arabs like middle eastern population (probably similar to jews and the test done by Krefter supports this recostruction:
Tuscan using Remedello Italian, LN/BA Central European, and East Mediterranean.
26% Remedello + 32% Bell_Beaker_LN + 33% Cypriot + 9% Cypriot @ D = 0.0073
31% Rem + 29% Bell_Beaker_LN + 3% Syrian_Jewish + 37% Syrian_Jewish @ D = 0.01664)

4)When we will put are hands on a late iron age italian sample we will be able to establish the amount of additional middle eastern admixture compared to modern Italy, since then ours ideas are educated guesses or mere speculations. 5) From all the slaves only the caucasus east med ones were having babies? There are accounts of african slaves in Rome but the only significant admixture seems to be in eupedia's maps and in Sicily (you can swimm to Africa from there and once was an islamic califate), and even the north african affinities are not that high, see, it's not obvius that these slaves had family and children since there were a lot a slaves from that regions too (if they were in massive numbers from the middle east why not from those regions?). 7) The biggest issue that I have whit your theory is that you want to explain high east-med and caucasus in Italy with slavery but Greece is very, very, very similar, this a massive incoherency, there is no way that Greece had massive slavery at any point in his history, there is no way in hell!Two similar case, two completely different explanations! How convenient?
8) In concluding this thought I'm asking you this: is it physically impossible that west-med and east-med came in Italy before the roman empire?

Mike Thomas said...


I don't understand why you don;t fall back to the more obvious and correct conclusion that Italy probably shifted 'West Asian' due to Mid-late Bronze, Iron and Greek period movements, rather than Roman slaves.

Simon_W said...

Ariele, you got me slightly wrong. I said I've always been of the opinion that Bronze Age movements from Anatolia and Cyprus, or more generally from the eastern Mediterranean, had an important effect on Sicily and parts of mainland Italy. And that now, since recently, I'm open to also consider other explanations. What's wrong with considering various hypotheses?

At least as regards northern Italy the slavery hypothesis seems promising, for several reasons:
- There are no strong archeological ties between northern Italy and the eastern Mediterranean in the Bronze Age, aside from a few Mycenaean sherds on the Po estuary.
- The last population in the Po plain prior to the Roman conquest were the Gauls, which would suggest a rather northern-like population.
- Slaves were necessary to work on the estates of rich Romans.
- The Roman conquest and the second Punic war had reduced the number of Gauls, so that the population density was low.
- According to North Italians show a signal of admixture between a Welsh-like and a Cypriot-like population around 66 BC. The Cypriot-like population has strong mixing coefficients with Cypriots, Syrians and Jordanians, but not with South Italians or Sicilians, suggesting that it wasn't primarily a Central/South Italian influence.
- The Seleucid empire was conquered by the Romans in 63 BC. The Greek geographer Strabo (1st century CE) recorded how an enormous slave trade resulted from the collapse of the Seleucid Empire (100–63 BCE).

But in other regions of Italy there may be other causes behind the Cypriot-like admixture. There is no reason why the cause has to be the same everywhere. Indeed in Tuscans the admixture signal, between a French-like population and a Cypriot-like one, is dated to around 942 AD, so this must have quite another explanation. And for southern Italy and Sicily there was no admixture signal with Cypriot-like people detected, probably because the admixture is too old.

Now a few comments on what you wrote: The Remedello samples are not all very old, RISE486 dates to approximately 1950 BC, and indeed he has a bit more of the Caucasus component (in the paper's ADMIXTURE run) than the others. But however, I don't conclude much from them, because the West Asian admixture is much stronger in Southern Italy, in parts of central Italy, and on Sicily than in the North. So if the North lagged behind in terms of West Asian admixture already in the Bronze Age, this shouldn't come as a surprise. Indeed Etruscans and Italics were important in Iron Age Italy, the problem's just that Etruscans don't seem to have been a very Anatolian-like population, for whatever reason. The West Asian component in modern Tuscans is below the Italian average, and moreover, according to the admixturemap site, dates to the Middle Ages. Paradoxically it's the old Italic parts of Italy that have more West Asian admixture. As for the Greeks, yes, they didn't have this massive slavery, and I already said in my above post that they were probably affected by the same Bronze Age West Asian admixture that showed up in two of the ancient Bulgarian genomes. And so it's quite possible that this also swapped over to parts of Italy, eventually.

Simon_W said...


Partly I did, at least for Sicily, Southern Italy and parts of central Italy this seems like the best explanation, but on a more general line I prefer to ascribe it to various very different admixture events. So that the West Asian admixture summed up over time, simply as a consequence of the geographical exposure.

Creative said...

I would say Middle-Easterns were more interested in buying “their own” slaves then Romans were. Out of a business stand point, it was probably much easier for Romans to sell ME prisons of war to a regional City or Market within the ME, than towing them back to Rom. The urban culture of the ME was well enough established and in demand for slaves. In contrast too early Gaul or Germania. I would also think that Roman auxiliary units from the ME were also interested in slaves, as a form of payment.

Simon_W said...

After successful war campaigns there were times when there was an oversupply of slaves, which caused a decline in prices. At that time it would habe been reasonable to sell some of the slaves to the ME. But in later centuries of the Imperial Age, when Rome was no longer expanding, slaves became a scarce good, and the Roman economy had problems to adjust to this.

Simon_W said...

On an unrelated note, here a few points to demonstrate how incredibly accurate the ANE K8 works when combined with 4mix:

Some of the best approximations I found for my ancestry:

26% West_Scottish + 47% HungaryGamba_BA + 2% Belorussian + 25% Tuscan @ D = 0.007

9% Corded_Ware_LN + 2% Belorussian + 78% HungaryGamba_BA + 11% Cypriot @ D = 0.0071

According to IBD sharing I have 29% of my East Prussian grandmother.
26% West_Scottish + 2% Belorussian = 28%
Very close, and maybe she has 1% central European ancestry.

According to IBD sharing I have 25.4% of my North Italian grandfather. Allowing for some rounding, 25% Tuscan hits the nail!

According to IBD sharing I have 45.6% from my South German & German Swiss grandparents. 47% HungaryGamba_BA is close.

Also, according to the quick and dirty K6 test I'm 11.7% "Middle Eastern". That's extremely close to the 11% Cypriot in the second approximation.

If I'm 11% Cypriot-like, then my Tuscan-like Italian grandfather must be 44% Cypriot-like.
And according to Tuscans are 41% Cypriot-like, again very close! And the other part of their ancestry is French-like, which would correspond to the other part being like HungaryGamba_BA in the second approximation.

Ariele Iacopo Maggi said...

Tuscany is actually wery west asian(10%), in comparison north italy is like 7% and the isolated abruzzo is 16%, why do you need slavery? If i'm not wrong some Yamma were 25% (just saying). It's a geographic clinate, and the fact that your grandfather shows up as tuscan tells a lot about the precision of that calculator. And then, I agree that from Celtic northern italy to today that was a massive schift towards the mediterranea, but you are taking fro granted that romans were not cypriots like or more southern italian like then today (did the barbaric invasion left no trace?). Another issue that i have whith your war-slaves thoery is that Italy lacks of south asian, and if we are talking about the fertile cresent, Iran is like 10%. Another source could be phoenicians (also explains Spain and souther France east-med, red sea, west asian). One last question, have you heard of Raeti people?

Ariele Iacopo Maggi said...

And then, how in the world do you explain the lack of nord-east african SSA and in nothern italy? (0,07, 0,03). Compare that one to the middle east.....
Syrian 3,2 and 1,6
Iran 0,4 and 1,09
Lebanon 2,3 and 0,8
Saudi 4,1 and 0,3
South italy 0,5 and 0,4
Sweden 0,2 and 0,2
Tuscany 0,2 and 0,08

1) If there were slaves why not african?
2) If there were salves from the middle east, those african compents will be present in those slaves.

Simon_W said...

I meant Tuscany is not particularly West Asian for Italian standards. It's below average. For example, in Dodecad Globe13, Sicily is 23.6% West_Asian, central Italy 20.5%, the two Tuscan samples 17.7% and 18.3%, Dodecad North Italians 14.4% and Bergamo 13%. I understand that people don't want to be descended from slaves, rather from proud warriors :D but to be objective means that it shouldn't be ruled out a priori.

The Eurogenes K15 West_Asian component isn't simply a geographical cline, it's a component that peaks in the Caucasus and that was absent from Europe during most of prehistory. Yes, Yamnaya had it and it expanded with the Yamnaya-like admixture, but in the Mediterranean presumably also with other people.

That my grandfather shows up as Tuscan is probably not a failure. His parents were from the Romagna, thus from southeasternmost Northern Italy, and it's not unlikely that the people there are more similar to Tuscans. There was a paper long ago on the autosomal structure of Italy (Di Gaetano et al. 2012) that suggested the same, i.e. Emilia-Romagna clustered rather with Tuscany than with the North.

I did assume before that the Cypriot-like admixture in the North is from Romans from southern Italy, but according to the admixturemap site South Italians were West Sicilian-like before ca. 886 AD, when they received slight North African input. So I would expect rather something West Sicilian-like than Cypriot-like if it was from southern Italy. But admittedly, it's somewhat tentative.

As for the lack of South Asian, I wasn't suggesting something like Iranian admixture, but something akin to Cypriots, Syrians and Jordanians.

Phoenicians in northern Italy? Sounds far fetched.

Of course I've heard of the Raeti, they were related with the Etruscans and Lemnians. Hypotheses abound, but it's hard to say anything conclusive about their relationship.

But there is a point I didn't mention yet: The time of admixture often doesn't equal the time of migration. And so there is the possibility that a Celtic-like and a Cypriot-like population existed in northern Italy for a longer period of time, and that they just didn't merge before ca. 66 BC. The question's just: Who is this Cypriot-like element supposed to be, and is the idea in line with archeology and history? I could speculate that the Padanian Etruscans were the Cypriot-like element and that they abstained from mixing with the Celts until everyone was Romanized. But doubts remain: Why is the main impact of Cypriot-like admixture in Tuscany dated to the Middle Ages? Either the Etruscans were not very Cypriot-like, or they were a relatively small element in Tuscany.

Simon_W said...

I don't know when the Northeast_African and SSA components in the Levant reached their present-day values. Conceivably the early Medieval Arabic expansion had an important effectin this point. On the whole I'm saying the admixing element was closest to Cypriots, and these have just 0.49 / 0.35. Since they are an insular population, they have conserved the ancient genetics of their mainland neighbours better.

One way or the other there is ancient Eastern Mediterranean admixture in Northern Italy, and if this is more from slaves or from free Etruscans a priori should't make a difference as far as admixture components are concerned.

Sub-Saharan slaves were probably less common in Roman Italy.

Ariele Iacopo Maggi said...

"Why is the main impact of Cypriot-like admixture in Tuscany dated to the Middle Ages?" Do you even belive that? Middle eastern in northern italy in the middle age, it's like saying african in Japan in the sengoku period. It's not real man, those calculator are broken, if they show something crazy and impossible they are wrong by definition. That's eupedia type of science, drawing conclusion from selected dna calculator errors.

Ariele Iacopo Maggi said...

"As for the lack of South Asian, I wasn't suggesting something like Iranian admixture, but something akin to Cypriots, Syrians and Jordanians." Those places were sparsely populated, doesn't fit your milions of people deportaion model (now that i'm thinking about that's the main reason why all of this it's crazy, Rome alone had a milion inhabitants, how many slaves do you need to affect such population?).

Ariele Iacopo Maggi said...

"I understand that people don't want to be descended from slaves, rather from proud warriors" actually romans slaves might have been more warriors then those late neolithic anatolian, levantine farmers.

postneo said...

Economics drove Aslave ancestry to be 10 % in semi industrial US. Slaves had kids who could be sold. In the Roman Empire the demand would be even higher.

Periodic collapse of occupations and Christianity would lead to freeing of slaves and upward mobility.

postneo said...

Even recently Islamics and crusaders swapped huge numbers of POWs and galley slaves.
The occupTion did not last forever and they became parts of their host empires

Ariele Iacopo Maggi said...

"Even recently Islamics and crusaders swapped huge numbers of POWs and galley slaves." Are you suggesting massive population replament in northern europe? Most crusader came from France and England and they were led by their kings. Maybe that will explain the high karitana-hindukush in the U.K. (I'm kidding, I'm simply highlighting the fallacy)

Ariele Iacopo Maggi said...

"Periodic collapse of occupations and Christianity would lead to freeing of slaves and upward mobility." no doubt about that, I doubt that these slaves were levantine like (a very small population even today, and there was a recent baby boom there)

Ariele Iacopo Maggi said...

I mean, think about Petra, Palmira, wouldn't have been easy to ship middle estern slaves there? Simon_W is saying that they shipped to northern Italy, and Tuscany is been invaded by levantin in the middle age, I mean. The easier explanation is that pre indoeuropean southeastern mediterranea was geneticly the same with small variations identifiable even today, Italy and Greece more West_Med, the Levant more Egyptian.

Simon_W said...

@ Ariele

It's not a DNA calculator, it's from a paper by Garrett Hellenthal et al. that was published in Science:
Of course, not everything that gets published in Science is correct, and Davidski also said he doesn't believe in their results. But at least it's not some fringe stuff some crazy blogger pulled out of his arse.

And yes I think it's very possible that the Tuscan population changed in the Middle Ages. The Byzantine Empire was decaying and conceivably many Byzantines didn't want to live under Turkish rule but prefered to emigrate to Tuscany. Historians think their influence was one of the main causes for the start of the Renaissance in Tuscany.

And when I say the mixing coefficients point to Cyprus, Syria and Jordania, it follows that we're dealing here with a generally East Mediterranean influence from the area Cyprus-Levant-Anatolia. Of course not of the Syrian desert which is indeed sparsely populated, lol. But the Levant is part of the fertile crescent and Anatolia the home of some of the oldest civilisations. And anyway, as I said, the existence of such an East Med admixture in Italy cannot be denied, the point of contention is merely how and when it came to Italy.

And how densely or sparsely populated northern Italy was after the second Punic war isn't a foregone conclusion either. And if slaves were 40% of the population they surely affected it.

Simon_W said...

That said, the Romans believed to be of Trojan descent. And the same was told about the Elymi in Sicily. And the Veneti allegedly were descended from allies of the Trojans, hailing from northern Anatolia. And legend had it that the Pelasgian Asili were led to Picenum by their king Aesis. Hence Jesi on the Esino (both Aesis in antiquity). I'm sure my enumeration isn't complete. Historians tend to regard these stories as mere inventions, contrived to endow ancient Italic tribes with a noble descent, because the East Med was considered cool back then. But modern DNA research suggests that there may well be a true core in these stories.

Simon_W said...

Apparently the ancients didn't quite have a clue what had been going on in the Bronze Age. Dionysius of Halicarnassus wrote about Pelasgians settling on the coast of Etruria and in Latium, but that of course the nasty Etruscans had nothing to do with the sympathetic Pelasgians. Pelasgian descent may be at the origin of Roman legends about Trojan roots. The Romans were proud of both, yet didn't figure out that this may actually mean that they're related with Etruscans - which they would have found less enjoyable.

According to Dionysius the Pelasgians were the original inhabitants of Greece, according to Thucydides the Etruscans were Pelasgians, according to Herodotus the Etruscans were from western Anatolia. Sounds like contradictory, but if non-IE Anatolians had populated Greece in the earlier Bronze Age, it could make sense.

Also this allegedly Anatolian influence in the Veneti might explain the presence of the Raeti in the nearby Alps. We're used to think of the Raeti exclusively as relatives of the Etruscans, but what if there was some general non-IE Greek/Anatolian Bronze Age population affecting almost all of Italy to varying degrees, and the Etruscans were only a part of this movement?

Simon_W said...

I think the best argument against the slave theory is still the fact that slaves were from everywhere, there were just as well Celtic and Germanic slaves as there were Anatolian and Levantine ones. But at least it seems possible that by chance one area received the majority of its slaves at a time when slaves of one particular area were being available abundantly.

Creative said...

During the 2 and 3 Punic war the main (source) of slaves (POWs) were Greeks and South Europeans, according to Titus Livius Patavinus.

(Numbers are exaggerated)
2,000 inhabitants of Lilybacum (Livy, 21, 51, 2)
5,000 Hirpini (Livy, 23, 37, 12)
30,000 Tarentines (Livy, 27, 16,7)
3,000 Mandurians (Livy, 27, 15, 4)
5,700 Macedonians (Livy, 33, 11,2)
5,632 Histrians (Livy, 41, 11, 8)
700 Ligurians (Livy, 42, 7, 9)
2,500 Boeotians (Livy, 42, 63, 11)
150,000 Epirotes (Livy, 45, 34, 5)

Ariele Iacopo Maggi said...

Thank you Creative. That "150,000 Epirotes" maybe explains high frequency of E-V13 in Italy (but maybe not). In comparison the north africans E-M81 is farly low in Italy (is 60% in Tunisia) less the 1% in half Italy, is much higher in France if I'm not mistaken, looks geographic admixture and not slavery Is it possible that they were not taking slaves from north africa and milions of slaves from a very particular point in the middle east?. Think about the punic wars? And the hundreds of year of roman domination in north Africa (Leptis Magna ecc..). The impact of slavery in Italy might be massive, I don't tnink that a very small population of levantines could have possible change the genetic of a very large and populated area (such as northern Italy) in any foundamental way. And we already determined that those modern countries are not a good fit as an admixture as they are today (Turkey is very asian, the levant has to much africans, and Iraq, Iran have to much south asian; and armenians, caucausus are not east med enough) then you'll need another theory, another admxiture event in those country that change them substantially, I mean, it's one the most complicated theory I've seen on genetic-historical matters. Last point, East med and Red sea are scattered even in northern europe (at low rates but are there), how do you explain that? The romans never even conquered a lot of those territories, and we are talking about very isolated population, this fact fits a late neolitic, early bronze age migration from the middle east and anatolia (agriculture and metal working), that have both historical and archeological bases (etruscans!!!).

Simon_W said...

I'm tired of this discussion and we can leave it as it is. But I want to add that we don't need „another theory“ to explain Asian admixture in Turkey. Have you never heard how the Turks conquered Anatolia in the Middle Ages and that their deep roots are in central and even eastern Asia? Also the Arabic invasion of the Levant occured in the early Middle Ages together with the spread of Islam. That's all factual, not theory. I never considered Iraq or Iran as an important source of Roman slaves. Iran was never part of the Roman empire, Iraq just for a couple of years.

East Med and Red Sea in northern Europe may go back to various sources, one of them being the early Neolithic farmers.

Archeologically you can't derive the Etruscans from Anatolia. At least not directly and obviously.