search this blog

Saturday, March 16, 2019

Let's try a formal heuristic approach


I created a massive outgroup f3-statistics matrix, featuring almost 300 ancient and present-day populations and individuals, for the purpose of running unsupervised, or at least semi-supervised, fine scale mixture tests with nMonte. Most of the stats were computed with 400-900K SNPs, which is a lot and should provide plenty of power. The matrix is available in a zip file here.

The results I'm getting with this new setup are very similar to those obtained with the Global25. The main differences, as far as I can see for now, are that the f3 data produce more stable results when modeling very deep ancestry, while the Global25 provides more accuracy when modeling fine scale recent ancestry (probably because it's better at picking up more recent genetic drift).

Let's investigate some pertinent issues with the new data using nMonte and PAST. How about we start with these?

- where did Bell Beakers get their steppe ancestry from?

- which Steppe_MLBA group did Indians get their steppe ancestry from?

- do the present-day Irish have any Hallstatt ancestry?

- what is the origin of present-day Basques?

- what is the precise ancestry of Armenia_ChL?

- do the Swat Iron Age samples really lack BMAC ancestry?

- does Anatolia_MLBA really lack steppe ancestry?

Note that the f3 matrix includes the ancients from the new Olalde et al. paper on the genomic history of Iberia (see here). I've also updated the Global25 datasheets with most of these samples.

Global25 datasheet (scaled)

Global25 pop averages (scaled)

Global25 datasheet

Global25 pop averages

By the way, Hajji_Firuz_ChL I2327, from Narasimhan et al. 2018, is now labeled Hajji_Firuz_IA in the above datasheets, because my understanding is that he's actually from the Iron Age rather than the Chalcolithic period. For background reading about this controversial sample see here and here. I don't have any more info on this topic; we'll just have to wait for the formal publication of the Narasimhan et al. manuscript to get all the details. Apparently it's coming very soon.

See also...

An exceptional burial indeed, but not that of an Indo-European

Maykop: a multi-ethnic layer cake?

Late PIE ground zero now obvious; location of PIE homeland still uncertain, but...

77 comments:

a said...

"By the way, Hajji_Firuz_ChL I2327, from Narasimhan et al. 2018, is now labeled Hajji_Firuz_IA in the above datasheets, because my understanding is that he's actually from the Iron Age rather than the Chalcolithic period..."
If that is true, how did I2327 from Narasimhan et al make it through the initial peer review, without questions being raised about its dated origins by the authors and those involved in reviewing the findings?

Richard Rocca said...

Here are the nMonte results for the oldest Iberian Bell Beaker P312+ sample EHU002, who's dating co-author Carmen Alonso-Fernández told me is averaged out to 2490-2335 calBC and not "before 2500 BC" as Diego keeps saying. It is made up of what is already universally known, a mix of Corded Ware (or like), a French population that the Corded Ware population would have mixed with before arriving in Iberia, and the Copper Age Iberian population.

Iberia_Central_CA_Stp:EHU002
distance%=2.0464

CWC_Germany,41.9
Iberia_ChL,39.2
France_MLN,17
Blatterhole_HG,1.9

I'm sure Diego will talk some rubbish about untested samples, etc. and provide not an ounce of actual ancient DNA data data to back up his claims.

Slumbery said...

Some funny G25 nMonte results tangentially connected to the Irish question:

"sample": "Irish:Average",
"fit": 1.2664,
"England_Anglo-Saxon": 100,
"Ireland_EBA": 0,

This could be entirely a local admixture into Anglo-Saxons, but:

"sample": "Irish:Average",
"fit": 1.9077,
"Beaker_Britain": 77.5,
"Ireland_EBA": 22.5,

So it appears that things were on the move in Ireland well after the dissolution of Bell Beaker. But it is possible that this is very recent overall (there are no samples from the last 3500 years or so).

Hallstatt ancestry is very difficult to estimate, because similar ancestry could have arrived later. But I think they have some.

"sample": "Irish:Average",
"fit": 1.0526,
"England_Anglo-Saxon": 80,
"Beaker_Britain": 10.83,
"Hallstatt_Bylany": 9.17,
"Ireland_EBA": 0,

And sorry for not using the shiny new database above, I can't do it now.

Rafs said...

Apart from Iberomaurusian, are there any other ancient North African samples in the file?

J.S. said...

Can you do the same with French samples and the R-P312 Bylanny?

Slumbery said...

@J. S.

Do you mean modelling modern Irish with ancient samples from France or modelling modern French with Hallstatt?

BTW, testing modern French for Hallstatt ancestry is more difficult and has more possible perils than testing the Irish, because ancient France is very under-sampled for its size and complex connections. So it is difficult to choose the right references.

Matt said...

Couple of brutally simple PAST3 functions on this dataset.

1: Using as Principle Coordinates similarity matrix (exploiting that this is a similarity matrix): https://imgur.com/a/OLfG575

2: Correspondence Analysis with moderns as columns and ancients as rows: https://imgur.com/a/eehAOyo

3. As limit with limits to modern West Eurasians as column: https://imgur.com/a/CQnT6Dj

I'm not sure about self similarity (diagonal) as 1; is there any quick way to run a matrix of the equivalent "distance" stat as you've run for me before (where diagonal would be 0 and increasing distance with higher number, not increasing similarity), like the stats you generated for me on 03/01/2019?

Matt said...

Judging by ancient to modern correspondence values, Mycenaean in this dataset seems to share more drift with Caucasus populations that suggested by linear combination of Sintashta/CWG/Yamnaya + Minoan_Lasithi: https://imgur.com/a/jLvL9f0

If that's inconsistent with other data may need a look?

Samuel Andrews said...

Just like that discussion on Iberia ends...

J.S. said...

@Slumberry
I mean testing Modern French for Hallstatt and, if possible, French Beaker ancestry. Modern French cluster close to French and Swiss Beakers on Dave's Boscombe PCA plot, but I would like to know, for instance, how French Beakers are ancestral to Modern French, at least, the ones in the available dataset.

Gaska said...

@Richard Rocca

EHU002/UE 450: 2562–2306 cal BCE (3933±32 BP, CSIC-1896)- (2.434 BC)- HapY- R1b-P312- Haplogrupo Mitocondrial- K1a4/a1

That's what I said. which on the other hand is the same as Olalde says. I have also said that both this dating and the cave of la Paloma must be checked because the first is a collective chalcolithic grave with two intrusive BB burials, and the second is a cave of the Magdalenian culture with unclear stratigraphy, and also does Not belong to the BB culture because in Asturias there are hardly any deposits related to this culture. Then we just have to check the data.

Even an unintelligent person like me knows that R1b-P312 was in the peninsula before 2,500 BC, you know why? Following the Kurganist reasoning, only the haplogroup R1b-M269 is responsible for the entry into Iberia of the famous autosomal steppe signal. In his first paper Olalde only found 8 samples in Iberia with that autosomal marker to a greater or lesser extent. The oldest case is the following;

Arroyal-I0462-Female- Mit-K1a+195-2.476 BC- That mitochondrial haplogroup is absolutely western, it has only been found in the British islands, in four Spanish BB sites, and Jinonice Czech Republic (BB culture), 500 years later. It is not very difficult to deduce that if the Kurganists are right that steppe inheritance had to pass to that woman from her father, then you have P312 before 2,500 BC in Iberia.

Obviously that is a terrible problem for the supporters of Gimbutas because the current chronology of P312 in Europe is:

I5021/Osterhofen (Bavaria)-2.456 BC- HapY-R1b-U152-L2. Mit-K1c1
EHU002/el Hundido (Spain)- 2.434 BC- HapY-R1b-P312- Mit K1a4/a1
I5478/Oostwoud (Frisia) 2.431 BC- Hap Y-R1b-P312. Mit-X2b4
I5379/Canada Farm 2.380 BC –Hap Y-R1b-L21. Mit-HV0+195
I1390/Sierentz (Alsace)- 2.373 BC- Hap Y- R1b-P312. Mit-X2b4/a
I7205/Radosevice (2.350 AC). HapY-R1b-L2. Mit- H10+16093
I7044/Szigetzenmiklós -2.350 BC. Hap Y- R1b-L2. Mit-U5b1/d1b
I3238/TDPAD- La Paloma (Spain)-2.350 BC Hap Y-R1b1a/1a2a-L49- Mit-H3+152

Taking into account this chronology and that Osterhofen is not P312 but L2, technically the oldest P312 * of Europe is in Iberia, and as everyone knows Spain is in Western Europe, not in the steppes.

Maykop was a disaster for the theory of the steppes, but this paper of Olalde has been devastating.

1-The abundance of data on Iberian mitochondrial haplogroups, has given us much more evidence that shows the Iberian migrations (at least of women) to the rest of European BB regions. I will not bore you with more data.This genetic evidence joins the abundant archaeological and anthropological evidence (Desideri & Besse)

2- It is currently impossible to link R1b-P312 with the expansion of the Indo-European language, and it is impossible to determine the language spoken by the BB culture. And say that Non-IE languages survived in pockets in Italy and in Spain means that you do not know the Spanish and Italian geography well,unless you consider that Navarra, Aragón, Catalonia, Valencia, Murcia, Western and Eastern Andalucia and the western part of Castilla la Mancha is a pocket, when in reality it is approximately half of the Iberian Peninsula.

3-Now it turns out that the Iberians, Basques and Tartesians absolutely R1b, spoke non-Indo-European languages ​​and I2 Celtiberians spoke Indo-European languages. To say that R1b-p312 introduced IE in Iberia is amazing, especially when you have no proof to prove otherwise

Gaska said...

As it seems, now, our friend P312-K1a4/a1-El Hundido (Monasterio de Rodilla, Burgos, Castilla y Leon), is a mixture of the CWC, French Chalcolithic and Iberian Chalcolitic. It is also amazing, because his mitochondrial haplogroup has been found in 7 Spanish sites, 2 of them-neolithic and 5 of the BB culture. Of course, so far it has never been found in France or in the CWC, or in the steppes. It has also been found in 5 other BB sites-

1-Kehf el Baroud Cave, Casablanca, BB culture-2.600 BC
2-Radovesice Czech Republic- BB culture-2.350 BC
3-Amesbury-I2418-England- BB culture-2.320 BC
4-Germany- Haunstetten, Bronze age 2.030 BC
5-Poland-RISE154- Bronze age- 1.845 BC.

Anyone can understand that this data is more than a mere coincidence, and I do not think anyone imagines CWC who had not seen a ship in their life, sailing to Morocco or England to transport the famous steppe ancestry

And with respect to his haplogroup Y- R1b-P312 has never been found in the CWC, and the samples that have been found in France are more recent than the Spanish ones, so it is at least amazing that autosomal composition that you have sent.

Un saludo

Davidski said...

@Diego

Are you suggesting that CWC groups weren't capable sailors? Have you ever heard of the Boat Axe culture that colonized the islands around southern Scandinavia? Obviously not.

Also, you fail constantly when you try to argue that R1b-M269 did not first enter Iberia with populations rich in steppe ancestry.

It's now the consensus that this did happen, and for a very good reason too: the data clearly show that it happened. And you'll never be able to produce any evidence to the contrary, because such evidence doesn't exist.

Just about the only fair point you've ever made in the comments here is to highlight the fact that R1b-M269 and steppe ancestry are actually very strongly associated with non-Indo-European populations (and not just Basques) in ancient Iberia.

This is indeed a problem, and it will require a plausible explanation.

Routine language swapping from Indo-European to non-Indo-European languages in Iberia by the R1b-M269-rich groups doesn't appear to be a sensible explanation considering the scale of this correlation.

However, judging by your performance here to date, I don't expect that this plausible explanation will come from you. Buy by all means, try and prove me wrong.

Gabriel said...

Maykop was a disaster for the theory of the steppes, but this paper of Olalde has been devastating.

I didn’t know it was opposites day.

AWood said...

@Diego

So where's the evidence that P312+ existed in Iberia significantly earlier than 2500? Where is the P312+ in aDNA from Iberia without associated German Beaker ancestry, and by extension Yamnaya ancestry? Yes, you can cherry pick the 1 male Celtiberian who happens to be I2-M26, but we know this lineage also existed in the earlier Chl period.

Drago said...
This comment has been removed by the author.
Richard Rocca said...

I was recently told that the Kromsdorf Bell Beaker sample which has already been positive for R-M343 is dated to a full 2 sigma date of 2850–2488 calBC. That is a midpoint of 2669 calBC. There is no R1b in Iberia that is going to come anywhere near that date, no matter how hard you close your eyes and try to imagine it.

Gaska said...

@Davidski

You know that the CWC was not a homogeneous culture, there were different groups, some of which never got to see the sea. If they had been good sailors or they would have known the neolithic routes of navigation, instead of finding the archeological evidence of the BBC throughout Europe, the British Isles, Ireland, Morocco, Sicily, Sardinia etc, we would find signs there of the CWC, even R1a, and that has never been found.

I'll be honest with you, I do not think that I2a-M26, which is actually a haplogroup found in Iberia since Chalcolithic is a haplogroup typical of the Berones (it was found in the village of La Hoya, which although it is in current Basque territory, belonged in antiquity to the Berones, who were one of the Celtiberian peoples of the peninsula). I have always criticized that geneticists sometimes draw hasty conclusions with few samples studied, and in this case I think the same. The Berones, like the rest of the peninsular Celtic peoples had to be mostly R1b, although they undoubtedly had to be mixed with some older local populations.

That is the most reasonable explanation, although this certainly does not solve the problem because in reality all the peninsular peoples, both those who spoke Indo-European languages ​​and those who did not, were mostly R1b, and this does not make any sense, unless these languages ​​evolved in situ after the arrival of R1b, that is, they were more modern than we all think (Basque, Iberian, Tartessian, Lusitanian and Celtiberian). But if this were so, why some are Indo-European and others are not? this explanation does not make sense either.

I believe that the error is to link the expansion of the Indo-European to an age as old as the chalcolithic and to processes of invasion or conquest. I mean, the migrations from the east, both from Anatolia and from the steppes had to be in small family groups, which undoubtedly mixed over millennia with the local populations. Only when the different European peoples had a sufficient military and demographic capacity to carry out large-scale wars and conquests of territories could they impose their languages ​​on the different conquered peoples, that is, the truly expansion of the Indo-European languages ​​could only have taken place since the middle of the Bronze Age and during the Iron Age.

In the case of Iberia, any haplogroup arrived from outside, it did it in small groups that adapted to the local cultures, also to the different non-Indo-European languages ​​that were spoken in the peninsula (Basque, Tartesian and Iberian). During the Iron Age, Celtic peoples entered Iberia, they have left written testimonies, but were not able to impose their language because they were not demographically hard enough or simply occupied territories that had not previously been occupied.

The arrival of the Romans meant the end of both Indo-European and non-Indo-European languages, because they did conquer the whole of Iberia, although they took 200 years to do so. Only the Basque survived, which has remained as a testimony of the old non-IndoEuropean Europe.



Davidski said...

@Diego

You know that the CWC was not a homogeneous culture, there were different groups, some of which never got to see the sea. If they had been good sailors or they would have known the neolithic routes of navigation, instead of finding the archeological evidence of the BBC throughout Europe, the British Isles, Ireland, Morocco, Sicily, Sardinia etc, we would find signs there of the CWC, even R1a, and that has never been found.

You just totally contradicted yourself by first claiming that CWC wasn't homogeneous and then expecting to find signs of eastern CWC and its R1a in Western Europe.

Well, if CWC wasn't homogeneous, then I guess that explains why northern Beakers are so similar to CWC and yet so different.

More here...

Single Grave > Bell Beakers

Davidski said...

@JuanRivera

There are BA-age R1a clades that have been found in Beaker territory (even Iberia, whose R1a clade is R1a-Y86945), and obviously has to come from CWC.

Yep, probably, but it's impossible to be certain when they arrived in Iberia because this isn't ancient DNA.

They may have come with Goths or Alans instead of any Copper or Bronze Age group.


Davidski said...

@JuanRivera

You're only inferring that those mutations happened in Iberia based on modern samples.

But what if the direct paternal ancestors of these modern people lived in Asia at the time, and by sheer coincidence their Medieval descendants ended up in Scotland, Spain etc.?

It's impossible to be certain about these things without corroboration from ancient DNA, because people love to travel.

Ric Hern said...

Firstly. I know that modern population distribution can be misleading, however when you take R1b and R1a out of the equation and try to link Indo-European with other Y-DNA distributions it looks very uneven. In a lot of Southern European countries Y-DNA Haplogroup E outperforms J2 and J1, and J2 in many areas hardly reaches a 5% Mark while in most areas R1a and R1b reach a minimal of 20% combined. Compared to R1a and R1b most other Y-DNA Haplogroups looks Pockety.

So for me at least the Spread of J2 and most others in Europe does not particularly connect well as vectors for the Spread of Indo-European.

Davidski said...

@JuanRivera

I based that of Iberia because there's one Spaniard in FTDNA that's R1b-M73.

That's not a very safe methodology though, because you don't know the paternal history of this one sample, and it's unlikely that even he knows it going back 500 years or more.

Samuel Andrews said...

X-chromosome data really confirms that extreme sex-bias admixture happened between R1b P312 newcomers & Iberian 'farmers.'

It's safe to assume, the same kind of sex-bias admixture happened between R1b L151 clans straight-out-of the Steppe and northern European 'farmers' to create the standard Kurgan Bell beaker makeup around the Rhine river.

How, Kurgan cultures were able to 'cuck' the men of the lands they 'conquered' is a question for leftist European archaeologists to ponder.

Time, for them to accept lots of violence & killing happened even if it isn't obvious in archaeological remains. The R1b P312 clans 'conquered' western Europe & created new hybrid populations who were a mixture between P312 men & native women.

Time, for leftist archaeologist to understand 'gendered' Bell beaker male burials defined by weaponry are not just symbols of 'identity.' Bell beaker men actually used those weapons. Native men didn't willingly give away their women.

Samuel Andrews said...

*If I had a dime every time I've heard an archaeologist say the weapons they find were merely 'symbols of power & identity.' The first records in human history record very violent societies: Rome, Assyria, Greece, Old testament, etc, etc, etc. All the way up till 1945 none stop war. But, they want people to believe, humans used to be peaceful.

Ric Hern said...

https://en.m.wikipedia.org/wiki/Droit_du_seigneur

Slumbery said...

@J.S.
"I mean testing Modern French for Hallstatt and, if possible, French Beaker ancestry. Modern French cluster close to French and Swiss Beakers on Dave's Boscombe PCA plot, but I would like to know, for instance, how French Beakers are ancestral to Modern French, at least, the ones in the available dataset."

Ok, let's try this in G25 nMonte. As you requested, I specifically used the P312 Hallstatt sample.

"sample": "Basque_French:Average",
"fit": 1.3334,
"Hallstatt_Bylany--DA111: 27.5,
"Beaker_France_South": 26.67,
"Iberia_ChL": 18.33,
"Beaker_Italy_North": 14.17,
"Beaker_France": 6.67,
"France_MLN": 6.67,

"sample": "French_South:Average",
"fit": 1.2826,
"Beaker_Italy_North": 23.33,
"Hallstatt_Bylany--DA111: 20,
"Beaker_France_South": 19.17,
"Beaker_France": 16.67,
"Iberia_ChL": 16.67,
"France_MLN": 4.17,


"sample": "French_East:Average",
"fit": 1.6099,
"Beaker_France": 32.5,
"Beaker_Italy_North": 25,
"Hallstatt_Bylany--DA111: 23.33,
"Beaker_France_South": 18.33,
"Iberia_ChL": 0.83,
"France_MLN": 0,

"sample": "French:Average",
"fit": 1.2773,
"Beaker_France_South": 30.83,
"Beaker_France": 25,
"Beaker_Italy_North": 25,
"Hallstatt_Bylany--DA111: 17.5,
"France_MLN": 1.67,
"Iberia_ChL": 0,

Now, I would not take this literally. This is of course a brutal over-fitting, because I used a lot of related Beaker sample-sets. Also I tried to represent Roman-era Italian admixture with Italian Beakers and that is a solution I am not happy with. Basques appear to have the most Hallstatt ancestry, but I suspect it is simply because the other references are lacking the most for them.
Of course Hallstatt itself had a ton of Beaker ancestry. I tried to step back and look at simple models with the main Frech sample-set.

"sample": "French:Average",
"fit": 2.6126,
"Beaker_France": 70.83,
"France_MLN": 29.17,

"sample": "French:Average",
"fit": 1.1559,
"Beaker_Bavaria": 97.5,
"France_MLN": 2.5,

As you can see, German Beakers are a much better fit than local Beakers. In fact the fit is better than with the over-complicated original test. Now this is probably Hallstatt Celtic + later Germanic? I am not sure about the interpretation, but I think the Frech have real Hallstatt ancestry and more than the Irish.
Another possibility that the migration that transformed Iberia at the start of the Bronze Age also transformed France on the way there, but the Frech Beaker samples pre-date that.

Nezih Seven said...

Anatolia_MLBA doesn't seem to have any steppe ancestry:

[1] "distance%=1.2881"

Anatolia_MLBA

Anatolia_EBA_Ovaoren,82.2
Kura-Araxes_Kaps,8.4
Balkans_ChL,7.4
Levant_N,2

Tesmos said...

@JuanRivera,

"Found several Iberian clades of R1b-U106. Of these, only R1b-U106* has enough age to be introduced by Beakers, the rest look Celtic."

The U106 sample from the Olade paper is dated to 777-981 AD from medieval times from Sant Julià de Ramis, Girona, Catalonia, right in the north-east corner of Spain, so this was after the Germanic migration period. The U106 sample could have been easily a descendant of Visigoths. Alternatively, this sample might has a connection with France, because Charlesmagne recaptured Girona in 785. I guess we need to take a look at his autosomal data if anything stands out.

Samuel Andrews said...

@Ric Hern,
"https://en.m.wikipedia.org/wiki/Droit_du_seigneur"

Something like that. Nobles/kings with special 'right' to do whatever they want to their subjects' daughters & sisters. That would mean, R1b P312's created a caste system with themselves in upper castes & 'farmers' in lower castes.

Samuel Andrews said...

But...

IMO, it's simple. They just killed the men & took the women. What happened in Britain proves the "R1b beaker folk" were capable of mass violence/conquest. There's no other way to explain 90% population replacement than mass killings. In Iberia, it was a little different. They only killed the men.

I would be happy challenge leftist archaeologist to think of a better explanation. A disease that only kills men? Or, maybe the Neolithic Brits moved to the Atlantic ocean & built the city of Atlantis.

J.S. said...

@ Slumbery
Thank you very much for your help.

Open Genomes said...

Here's the Ward's distance-squared clustering tree for the preliminary F3 matrix:

Preliminary Ward's distance-squared clustering tree

The tree is broken up into 20 larger clusters outlined by boxes, to make it easier to see the clustering.

Because the distance between individuals and populations is not that different between the maximum variation within most clusters, almost all of the taxa have a "long tail". I suspect that combining the data from separate individuals into "populations" in some cases may have created some strong overlap between populations, because the individuals in these population themselves don't actually form clusters (which should exclusively contain closely related samples) compared to samples from basically identical populations.

Open Genomes said...

@Samuel Andrews

That "Kurgan" men actually used their weapons and killed the "Old European" men and took the women by force has been a standard talking point of "feminist" archaeology since Marja Gimbutas. Is that what you're calling "leftist"? If so, you're espousing the traditional "leftist" position.

What hasn't been known to or acknowledged by Marja Gimbutas and her followers are the Late Neolithic "death pits" and the signs of violence across Europe that predate the entry of the Steppe ("Kurgan") people.

There's a dichotomy between "diffusionist" and "migrationist" archaeology, and also there's been "feminist" archaeology, but unless you're talking about a strict classical Marxist interpretation of history, the term "leftist" isn't appropriate here. If anything, the extreme right-wing nationalists try to claim that their particular group today has "primordial roots" in the place that they live now, and they refuse to acknowledge any kind of conquest and migration. This kind of "archaeological primordial ultra-nationalism" is a regular feature of discussions in this group.

Davidski said...

@All

I've updated the f3 matrix and Global25 datasheets.

The f3 matrix now includes all of the Iberian sample sets from Olalde et al. 2019 that I could use.

Same links as above. Let me know what you come up with!

Drago said...

@ Davidski
What would you deem to be “recent” and “more ancient” wrt F3 vs G25 use ?

Davidski said...

@Dragos

The f3 matrix generally does a better job of modeling Paleolithic and Mesolithic ancestry proportions, but only if you choose a sensible set of outgroups (columns). The more the better, but they should be older than or clearly distinct from the test/reference samples.

But the Global25 seems better at picking up fine structure the closer one gets to the present-day.

KVN said...

@Samuel Andrews
Let me preface my comments by saying that I read this blog with great interest but that I am a rank amateur.
I have Y haplotype I2a2a which I am informed came from hunter gatherers. This haplotype had a high frequency at one time and yet no longer does. The Rs came in and took over. Am I somehow deficient? I have wondered why things unfolded the way they did. The I think that the hunter gatherers must have somehow self-regulated their population size to maximize survival. I believe that group selection must have come to operate in hunter gatherer societies. In order to limit population size some males may have come to have "altruistic" tendencies of not participating in reproduction as frequently as was possible. A Darwinian would argue that such altruism would select itself out. However, Trivers argued that kin-selection allows altruistic behaviors to persist if the individual refraining from reproduction has lots of DNA in common with the person that does reproduce. This would arguably be the case with hunter gatherers. If hunter gatherers in some location found new ways to support larger populations by say agriculture or pastoralism, then Darwinian selection would take over and remove the statistical advantage of kinship-based altruism in a hunter gatherer. All throughout the process, females continues to watch the competition among males for her favors. Hunter gatherers would withdraw quickly upon realizing they lost. Now come hunter gatherers from a different area, say Rs, that have abandoned their altruistic inhibitions and reproduce with abandon. They never give up the competition to reproduce. The Is keep giving up and the rs keep persisting. Soon, no more Is. It wouldn't have to always be violent. Rs would only have to keep on competing even after they lost an initial round of trying to impress a female. Of course, I am autosomally about the same as everyone else so it is not a big deal.

Matt said...

Davidski: "But the Global25 seems better at picking up fine structure the closer one gets to the present-day."

I agree though looking at these f3s, they actually do seem surprisingly sensitive to shared drift / population history between recent ethnic groups in f3(Outgroup,Modern1,Modern2).

For example: https://imgur.com/a/8mOvPCd

The residuals and nearest neighbour trees do still work out fairly matching to fine scale relationships.

Critias said...

Interesting idea. But when looking at european y-dna you'd also have to consider that HGs first mixed with Early European Farmers before mixing with Steppe pastoralists. It's generally accepted that EEF were mostly G, which has died out in Europe. So in the 1st mixture, HGs' I took over.

epoch said...

@KVN

At the end of the Middle Neolithic, just before the arrival of steppe peoples, the majority of Y-DNA of *farmers* already was I2.

BTW, the population control of HG's was basically infanticide.
https://en.wikipedia.org/wiki/Infanticide#Paleolithic_and_Neolithic

Matt said...

@Critias: Some of this depends on, when we talk about farmers, which region we're talking about, because the pre-IE populations had substructure.

I may be misremembering but as I can remember G2 was plentiful in SE Europe and related farmers in Hungary, and remained plentiful until quite late in history, when I2 and R1a probably from north and J2, E1b from SE both intrude.

G2 doesn't seem to have ever been that plentiful in North-Central Europe, where I2 from HG became dominant (I think the only exception to this is early LBK in Germany?). In Iberia, I2 and I looks like they are at higher frequencies than G2 from the beginning (4000 BCE), but G2 remains reasonably present through the whole sequence until R1b replacement.

I'm not sure there during the Neolithic->Chalcolithic period ever a parallel of the R1b replacement event, with G2 and I2 instead.

epoch said...

@David

Could you do:

Mbuti GoyetQ116-1 French Dutch
Mbuti GoyetQ116-1 Dutch Finnish
Mbuti El_Miron French Dutch
Mbuti El_Miron Dutch Finnish

?

epoch said...

@Matt

LBK was the initial migration of farmers. It settled mainly on Loess soils because these re among the most fertile in the world. The Middle Neolithic cultures saw an expansion into other soil types. TRB in the Netherlands is found on sandy soils, in Sweden on clays.

I think that is relevant in explaining the abundance of I2 vs G2. Not sure exactly how, but the expansion into other ecosystems and the rise of HG Y-dna which (used to) live there seems connected.

Andrzejewski said...

@epoch "@Matt

LBK was the initial migration of farmers. It settled mainly on Loess soils because these re among the most fertile in the world. The Middle Neolithic cultures saw an expansion into other soil types. TRB in the Netherlands is found on sandy soils, in Sweden on clays.

I think that is relevant in explaining the abundance of I2 vs G2. Not sure exactly how, but the expansion into other ecosystems and the rise of HG Y-dna which (used to) live there seems connected."

So in plain English what you're saying is that the WHG rich I2 population overtook the original LBK to form TRB? So basically and essentially BB (Dutch) owes lots of its heritage and formation to the Western HG rather than farmers?

Matt said...

@epoch, good point, I think it makes a lot of sense that being able to "fall back" on other subsistence types (hunting / fishing) during expansion of farmers out of the most productive zone. Clonal dna like the y from HG groups could have an easier time surviving under those conditions that autosomal, depending on how the processes ran. (Roughly clonally transmitted culture like language might also survive better, though we can't ever really know.)

Davidski said...

@epoch

I don't have any suitable Dutch samples to run formal stats like that.

Andrzejewski said...

I recall @Davidski mentioning that LBK and TRB might have spoken completely different languages. Can it be that the resurgence of HG ancestry responsible for TRB becoming more HG-like culture and language-like? Now, if Dutch Beakers are the "original" BB, can it be that the WHG strong ancestry responsible for BB emergence? Also, GAC is basically Funnel Beaker which is an amalgam of the Southern German LBK with the Northerly Baltic shore Ertebolle Culture. Can it be that the blondism in GAC (Polish?) is a heritage of the SHG-rich Globular Amphora? If Rhenish CWC is the origin of the Bell Beakers, can it be that Basque is actually more of a WHG language subsumed by Steppe migration rather than a local Iberian farmer/HG language (and so, not a continual of El Miron/La Brana or anything local at all?)

Andrzejewski said...

@All There was a strong resurgence of HG ancestry for the last 1,000 years before the Kurgan invasion. Was it attributed to a decrease of farmer ancestry due to a pestilence? Was it because ANF could no longer advance their crops north of a certain latitude, allowing HG to rebound? Was it stemming from the fact that the first pulse of Anatolia_N farmers in the Balkans merged with HG, but then for the next millennia it took slow cohabitation and gradual intermarriage until both groups were genetically and culturally undistinguishable by the time Kurganists imposed their patrilocal and cultural hierarchical structure? We will in all likelihood never know, but it's some fodder for thought.

Andrzejewski said...

@Ric Hern "Firstly. I know that modern population distribution can be misleading, however when you take R1b and R1a out of the equation and try to link Indo-European with other Y-DNA distributions it looks very uneven. In a lot of Southern European countries Y-DNA Haplogroup E outperforms J2 and J1, and J2 in many areas hardly reaches a 5% Mark while in most areas R1a and R1b reach a minimal of 20% combined. Compared to R1a and R1b most other Y-DNA Haplogroups looks Pockety."

Will you attribute these haplogroups to a Natufian element (or later Phoenician migration or Greek back-migration) incursion into the Balkans, in tandem with, earlier and/or later with Anatolian Farmers?

Andrzejewski said...

@Samuel Andrews "But...

IMO, it's simple. They just killed the men & took the women. What happened in Britain proves the "R1b beaker folk" were capable of mass violence/conquest. There's no other way to explain 90% population replacement than mass killings. In Iberia, it was a little different. They only killed the men."

I'm fascinated as to the reason they massacred 90% of British Stonehenge builders but spared the women in Spain.

Gabriel said...

Will you attribute these haplogroups to a Natufian element (or later Phoenician migration or Greek back-migration) incursion into the Balkans, in tandem with, earlier and/or later with Anatolian Farmers?

You’re getting into Semitic > Germanic territory there.

Andrzejewski said...

@All It may be a controversial view, but do you think that over the last 1,000-1,500 since the fall of the Roman Empire, that "exotic" autosomal markers such as East Med, Asia/ANE in Eastern Germanic tribes, North African in Spain and Sicily (predating Moors) and other "non-European" influences (mtDNA c1f/c4/c1e) slowly died out and thus the shift that Samuel mentioned from 25% mtDNA to 40% mtDNA was so pronounced (pan-European homogenization)? If so, would you think that Christianity or some Pan-European conscience played any role in it (for example, refusal to marry people who looked, ie. had exotic, non-standard DNA?

the dude said...

RE whether the beaker migrations were sudden mass slaughters or not. The evidence is not conclusive either way but there is one limiting factor on the 'killed all the men, stole the women' model - the number of generations from the common ancestor of the P312 beaker subclade who took over the area. We know that the isles were probably not invaded by beaker people until some point around 2400BC. We also know that this was likely an L21-DF13 group. So, what length of time is there between the DF13 common ancestor and the archaeologically known date of arrival of beaker people in the isles c. 2400BC? The answer is not a great deal of time. Nowhere near enough for some sword and sorcery type invasion. In fact its likely the isles invasion took place very soon after the DF13 common ancestor. Even if you say it was a general L21 invasion of the isles, it still doesnt leave a huge amount of time between the common ancestor and the archaeologically known date of appearance of beaker in the isles. A clan with a common G G G grandfather are not going to be big enough to topple the entire Neolithic population of the isles. Even at the outside and undergoing exponential growth, a common ancestor would not likely produce more than a couple of thousand of descendants across a century or two. Now consider that they spread across the entire British Isles in a single generation and you get some idea of how thinly spread this L21 group must have been c 2400BC. Its clear that this was not an invasion they way people think. It must be more complex than that or there are indeed major hidden factors in the fate of the Neolithic people.

Andrzejewski said...

@Them meee "You’re getting into Semitic > Germanic territory there."

No, I was referring to what @Matt said about the later non-G2 repopulation of the Balkans, not Germany.

Davidski said...

@Matt

What do you think about Iberia_Northeast_Empuries2 based on the f3 matrix? Are these straight up descendants of Mycenaeans?

Arza said...

@ Davidski

Mycenaeans, when compared with Minoan_Lasithi, share more drift only with South_Africa_2000BP. Bug or feature?

Samuel Andrews said...

@Open Genomes,
"That "Kurgan" men actually used their weapons and killed the "Old European" men and took the women by force has been a standard talking point of "feminist" archaeology since Marja Gimbutas"

I guess I was sort of ranting, I didn't catch that. You're right. I understand nationalist almost always want as you say primordial roots. But, they also support invasion theories in archaeology especially Indo European ones.

This is in part why after WW2, archaeologists switched to a strict anti-migration stance. But now, DNA has shattered that stance.

Furthermore, my issue with archaeologists is how they describe ancient societies from an entrenched leftist perspective & don't try to see it from the perspective of the ancient people.

When, they see a rich burial they focus on "social stratification" and "inequality." What, the ancient people cared about was how cool & awesome that guy was which is why they gave him rich burial goods.

or when, they talk about Bell Beaker male-specific burials the talk is often about the "construction of gender identity" and how weapons were symbols of identity & power. I can guarantee you that is not how the ancient people thought about it.

DNA, indicates Bell Beaker men actually used those weapons. Their weapons were not merely symbols of identity.

olga said...

If Bell Beakers came to Iberia looking for metals, gold, silver copper and tin, they used native men to work in the mines. These men became slaves and died of malnutrition and the new diseases brought from east. Besides,if they brought advanced techniques of metal working, beer manufacture, and introduced new ways in social life, they became first class citizens, monopolizing most of the women, without rape or mass slaughter.
Social stratification started and rich men, who have always had concubines all over the world, spread their foreign genes with great speed.
The same happened in South America, in case you need a real model.

Grey said...

the dude said...
"Even at the outside and undergoing exponential growth, a common ancestor would not likely produce more than a couple of thousand of descendants across a century or two. Now consider that they spread across the entire British Isles in a single generation and you get some idea of how thinly spread this L21 group must have been c 2400BC."

the Atlantic coast has its own climate: leaching from heavy rainfall creating acidic soil (which wheat doesn't like), so i wonder if the region was only sparsely populated with farmers and so the newly arrived herders could spread dramatically into relatively unoccupied territory.

(this wouldn't apply to most of Iberia though)

Ric Hern said...

If a huge army is not intent on settling in a specific area it could pass through it, kill the males, impregnate the women and move on without any time to transfer language. Only in areas where some chose to Settle a Language Shift would have been probable. Maybe this army travelled all over the Iberian peninsula trying to find the sources of Metal leaving a DNA trail behind them without any significant Language change in most areas.....

Ric Hern said...

The other thing is that there most probably were military actions going on in Iberia before the Steppe related Males arrived because of the evidence of fortified settlements predating their arrival by +-1000 years. So maybe the Male population was already thinned out significantly before they arrived. Maybe some settlements greeted them with open arms.....

Andrzejewski said...

@Ric Hern Would you think it's possible that Basque may be either a Neolithic language or a WHG language, but in both cases from the origins of the BB area, i.e. the Rhine area?

Ric Hern said...

@ Andrzejewski

Well apparently according to the abstract Basque can be linked to Iron Age tribes. What this suggests to me is that they are related to the Aquatanians. For this then we have to look at how Aquatania evolved since the Neolithic to the Iron Age.

Matt said...

@Davidski, using the correspondence method I'm experimenting with using these f3 stats (and still experimenting) Emp2 and the Mycenaean look relatively similar: https://imgur.com/a/0JhKwg4

(Method basically being take a set of outgroup columns likely to be differentially related to rows, then run correspondence analysis. I quite like using this method because seems each individual set of f3 stats seems quite liable to noise for particular pairs, so directly looking at correlations between two sets of stats is hard to read unless the differences are really obvious. But still experimenting with it. PCA seems a bit less clear here to look at, because it's so dominated by the strong overall correlation between all the non-African stats.)

Though Emp2 looks slightly more Anatolian shifted, not by very much in the scale of the large differences between ancients, but noticeable relative to the scale of differences between present-day populations. I don't know if they look like straight up descendants, but not exactly between Myc and Emp1 or the NW Africa shifted southern Iberians, or anything like that, just slightly more Anatolian.

As Arza mentions, as a separate phenomenon, it seems like South_Africa_2000BP stat is higher for Mycenaean for reasons that don't make sense. Literally this means that more sharing with South_Africa_2000BP *relative* to whatever was used as an outgroup, I'm guessing Mbuti(?), not in an absolute sense. That looks like something odd, and I suspect not real. But it's a separate issue from the general "positioning" that Myc shares with Emp2 in respect to other stats.

Try and have a look at this again later, got to go.

J.S. said...


@Ric Hern
Notably, the Basque-centred cluster has (...) a very large contribution of 91% (88-93) from France.
5...). Specifically the Basque share even more DNA with the French group than predicted by their mixture representation, which might reflect, for example, that the DNA the Basque share with present-day French is only a subset of modern French ancestry. This pattern is seen for other Spanish groups also, but to a much lesser extent.
page 22 and 24
https://static-content.springer.com/esm/art%3A10.1038%2Fs41467-018-08272-w/MediaObjects/41467_2018_8272_MOESM1_ESM.pdf

zardos said...

For small scale mobile war bands slaves could be more of a burden than an asset. Your lineage takes the land, the resources and females by force, the local males will always be your enemies.
Of course they killed them.

As for the pre-IE HG lineages taking over in the North: Most likely the local HG started with raids on and trade with the weakened Neolithic farmers and at some point hijacked communities.
Those new farmer groups with HG yDNA I, increasing HG DNA, a more pastoralist and warlike way of life subsequently captured one farmer group after another.
Thats why the absolute HG ancestry stayed low. It was like a snowball causing an avalanche.
IE were different as they reached higher numbers and a more complete societal system before expanding West.
But throughout Europe similar things happened in Late Neolithic times.
So some carriers of steppe ancestry and living the same way could very well stem from a non-IE HG lineage.
Probably some Bell Beakers from Central Europe were no IE speakers.

Ric Hern said...

@ Andrzejewski

Artenacian Culture ? So basically a France Neolithic Culture and Language.

Grey said...

"as a separate phenomenon, it seems like South_Africa_2000BP stat is higher for Mycenaean for reasons that don't make sense"

wild speculation but could south_africa_2000bp have some pheonician (or equivalent seafarer pop)

iirc there was gold traded along the coasts - maybe depends how south it was

Kristiina said...

@ Olga
I have noticed that women and men tend to see the history very differently in this respect. I fully understand your point-of-view.
Men seem to think that violence and slaughter is the key to success and our children are like insects and reptiles who need no father to protect them.

I am very happy that some male commentators on this blog are not even half as brave as they believe they ancestors were. :)

olga said...



Thanks, Kristina. It amazes me the men¨s view in trying to recreate scenaries centered only in slaughters and rapes.
I have read that in Spain, the Celts arrived something like 500 years BCA and settled in the heart of the peninsula, leaving no trace of wars. They called themselves Celtiberians and used the Iberian alphabet to write in some sort of mixed language that was lost with the arrival of the romans.
According to what I have read, Bell Beakers introduced new social ways like beer drinking in communities perhaps belonging to different forms of religion and new cults that could have been very atractive.
All Europe has been covered with fertility rites, mascarades and feasts where dancing, drinking and sexual activities took place, whose origin is lost in old times. and have contributed to the demographic growth trough history.

Open Genomes said...

Several ancient Iberians cluster with Basques:

The larger Basque-related clade on the Global25 Ward's distance-squared clustering tree

The complete Global25 Ward's distance-squared clustering tree (PDF)

First, I3758 clusters closely with Basques, and it's a very good candidate for a Vasconic speaker.

A closely associated clade has I2478, a Beaker Italy North, and I12879 Iberia East Iron Age who cluster with modern Spanish Pais Vaco, French South, and French. Ancient Iberian is not thought to be a Vasconic language but these may be similar to ancient Iberian speakers, since they apparently were present in Southern France. Another possibility here is that they were Ligurian speakers. Ligurian was a divergent centum IE language.

Another group that clusters with Basques is EHU002 Iberia Central Copper Age with Steppe, and I3238 Iberia Northwest Copper Age with Steppe, and significantly, I7238, a Czech Beaker. The Chalcolithic is quite early for Steppe ancestry in in Iberia. Central and Northwest Iberia were not historically Vasconic, but rather Celtiberian. The interesting thing is that this kind of "Vasconic-mix" ancestry was also found far from Iberia too, in Czech Beaker.

I8339, Iberia Northeast Roman Period would seem to be yet another Vasconic speaker, and he clusters with Iberia Northeast 6th century CE.

Probably the most interesting thing in this clade is a group of ancient Iberians, I3324 Iberia East Iron Age, I8203, I8214, and I8209 Iberia Northeast Empuries 1 along with I3759 Iberia North Iron Age who cluster with I6582 and I6561, Beakers from Poland. There is something about these Polish Beakers that makes them cluster away from the other Beakers. This may be higher Globular Amphora ancestry than the others. Poland of course is even further away from Iberia than the Czech Republic.

Open Genomes said...

Here are the Global25 runs for I3758, the Iberia North Iron Age who clusters closely with Basques:

Bronze Age and earlier ancestry composition of sample: I3758 Population: Iberia_North_IA Iron Age Europe

I3758 has quite a bit of Steppe and other Beaker ancestry, including 14.0% Srubnaya. Nothing particularly Vasconic here.

However:

Chalcolithic and earlier ancestry composition of sample: I3758 Population: Iberia_North_IA Iron Age Europe

Here I3758 has 57.6% Iberia Southwest Chalcolithic / Scotland Neolithic with 3.8% Tisza Late Neolithic. His Steppe-related ancestry is actually just 18.6%.

Therefore, he has very high Iberian / British Isles Neolithic ancestry compared to Steppe ancestry.

It would seem that Vasconic speakers differ from regular Steppe-related Beakers in that they have very much elevated ancestry from the Neolithic of Atlantic fringe, the Megalithic cultures. This raises the question as to whether the Megalithic cultures of the Atlantic coast and British Isles were Vasconic-speaking.

Unknown said...

@Olga The culture of a people deeply affects the way they use and react to violence. I have killed small animals and hated t, though I hypocritically eat meat. I am a soft, modern western city dweller. I know my grandmothers would mock me. We have the records of many conquering steppe peoples in historic times, most famously the Mongols, and their delight in annihilating populations (and taking women) is well attested. We only need recall the effectiveness of simple weapons like the machete in Rawanda. Steppe horseman armed able to shoot arrows from fast moving horses have an enormous advantage over farmers. I'd also be very reluctant to compare the tiny, scattered mining operations of prehistoric times with the (by comparison) vastly larger operations the Spanish ran in the Americas for nearly 300 years to serve the demands of vastly more populous societies. And even with the help of devastating old world diseases, they didn't come close to wiping out native y-dna.

Garvan said...

Looking at the f3 data-set, I can see how picking out pairs of ancient populations in the columns and plotting selected rows against them in PAST clarifies their genetic relationship. These graphical results make sense in the examples I looked at.

I tried selecting columns as out-groups and rows as in-groups, and looking at admixture percentages with nMonte but my skill level was not up to it. I did not get results across a range of test populations that I had confidence in. Still I learned how out-groups influence admixture calculations, so I am happy with this, even if I did not get as far as being able to answer the questions David posed in the post above.

As I have time, I will keep working on this, at this point I just wanted to say I appreciated the work in preparing these f3 statistics.

Targamos the Based, son of Kavkasos son of CHG son of said...

@Kristiina

Please save your idealization of the past for children's stories. Maybe you will also claim how Indo-Europeans and their language(s) spread into Eurasia through friendly chats with locals? Or did they play Rock, Paper, Scissors and the loser adopted the language of the winner?

Your view is entirely a modern Western view. Mongols will tell you that Chingiz Khan was a great person and that the conquered peoples even loved him. Persians will tell you the same about Shah Abbas, Turks about their genocidal rulers. These rulers did not live in the very far past, and the farther in the past we go the bloodier the conflicts were. A short glance at the pre-Persian "peace treaties" between winners and losers in the Near East is enough to demonstrate that. Tribal warfare was far bloodier than modern warfare, percentage-wise. If we know that Urartians and Assyrians took 100 boys and 100 girls from the defeated Colchians every year, entirely destroyed their cities and took survivors of those cities as slaves etc., then we don't even need to argue about how bloody the IE expansion would have been.

Andrzejewski said...

@Knowledgeable Geneticist You're castigating the Indo-Europeans ancestors as genocidal maniacs. Too much WWII fantasies?