search this blog

Friday, February 23, 2018

A swarm of locusts?


The dam has truly broken. Below is my usual Principal Component Analysis (PCA) of ancient West Eurasian genetic variation, except now also featuring the new samples from Mathieson et al. 2018 and Olalde et al. 2018. Incredibly, there are almost a thousand ancient individuals on this plot. The relevant datasheet is available here.


My imagination is probably running wild from all of this excitement, and I apologize if it is, but I reckon that the "Post-Kurgan expansion Europe" cluster actually looks like it's beginning to swarm all over "Old Europe", much like a swarm of locusts. These are, of course, our Bronze Age ancestors, rich in steppe ancestry and Y-haplogroups R1a and R1b. I reserve judgment on whether that's a good or bad thing.

In any case, note that I highlighted three samples in this analysis. The reason I did this is because I believe that at least two of them might be crucial to understating the Proto-Indo-European (PIE) expansion. I've given hints as to why on the plot. Am I on the right track? Feel free to let me know in the comments below.

See also...

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

Who's your (proto) daddy Western Europeans?

Migration of the Bell Beakers—but not from Iberia (Olalde et al. 2018)

191 comments:

Eric said...

Davidski, think you could mark where all the Varna, Balkan Bronze/Iron Age samples(including Malak Preslavets, etc) are? The South East Europe paper's PCA had obvious projection bias.

Would be much appreciated. Tbh, just a description of where the samples roughly are would be good enough if you're too busy.

Anonymous said...

The sample I5884 has a very strange dating of 2890-2696 calBCE.
A very strange time - Eneolithic, while everywhere already the Bronze age in Pontic area.
Its culture is not defined.
Obviously, the haplogoup is ancestral to the East of Yamnaya people, but he is not Indo-European.
This suggests once again that R1b(-Z2103) were initially non-Indo-Europeans, but Indo-Europeanized in the process.

Ryan said...

Thanks for doing this David. Pretty clearly an excess of WHG ancestry in this outlier. What's your best model of his ancestry? (I'll finally re-download R this weekend and start playing around with it myself. This stuff is too interesting to watch strictly from the sidelines).

I agree with supernord's take here. I think R1b-Z2103 intruded into IE here. I think R1b-M269 in general intruded into IE groups and then hitched a ride from with them in several waves, and that R1b-Z2103 was one of the first to do so. Hence why only Z2103 has a distribution that spans all IE languages (for example, R1b in Pakistan is 100% Z2103 or close to it). I think the Unetice culture was the next batch of assimilating R1b-M269 peoples into an IE.

Janko Raven Johnson said...

Cousins who became brothers-in-law?

Chad said...

It's obvious that no matter what the data says, some will always keep their heads in the sand on L51. "No, it can't be IE because -insert stupid comment here-"

AWood said...

Huh? Suddenly R1b-M269 is not PIE speaking because one sample has a smaller CHG pull? If you go back to Ukraine_Mesolithic, there is little to no CHG at all, and there are 2x R1a samples. So which is it? I5884 actually looks like a good proxy for the R1b central European BB males before acquiring a lot of the additional EEF admix (2800 BC). We already knew that BB had lower "Steppe" or CHG than the CW counterparts anyhow.

Ryan said...

@Chad - "It's obvious that no matter what the data says, some will always keep their heads in the sand on L51. "No, it can't be IE because -insert stupid comment here-"

How do you explain this Z2103 sample with 0 Yamnaya ancestry then? Do you place the origins of IE elsewhere? Honestly curious.

Archaelog said...

The PCAs are getting more and more cluttered by the day. I hope future papers would move towards more "interactive" PCA plots

Richard Rocca said...

@Chad Rohlfsen said... The oldest known L51 is from Hungary at this point

Which sample are you referring to? That I know of, the two oldest thus far are:

Bell Beaker RISE563 = U152+
2572-2512 calBCE (3955±35 BP, Poz-84553)
Osterhofen-Altenmarkt, Germany

Bell Beaker I5748 = P312+
2579–2233 calBCE (3945±55 BP, GrN-6650C)
De Tuithoorn, Oostwoud, Noord-Holland, The Netherlands

Rob said...

nice plot dave, I get what it’s suggesting, although id add a couple of details.

a said...

We have not even seen 1 sample from the Yamnaya Kurgans from Tisza River-region, Hungary. Various estimates place numbers as high as tens of thousands of Kurgans. I would not bet against finding 1 upstream L51 among those unsampled Kurgans. Is it possible to pca red/blue all BB and CW samples. Also can you point out Szigetszentmiklós on Csepel R1b sample on PCA as it might be connected with the Sarmatian and Jasz/Hungary and Ossetians and or Swat-Pakistan region, all regions with Z2109+ and or Z2110+ and downstream markers.

mike said...

One thing to note is Yamnaya expanded from the Caspian steppe and is extremely associated with Z2103. So whatever the deeper time movements of lineages, Z2103 and textbook Yamnaya had developed c3300BC at the eastern end of the Euro steppe and expanded from there west. Now, we also must recall that most archaeologists see the cuktural roots of IE in pre-yamnaya times in Sredny Stog which was centred significant further west in the steppes that the Yamnaya origin point. In fact the Volga was it's most easterly outpost.

mike said...

It's also important not to confuse two processes into one. The language and culture of the archaic indo-Europeans appears to have been coming to a recognisable entity long before Yamnaya and probably before the mix of autosomal strands we see in Yamnaya had stabilised into its classic form. The old Sredny Stog zone which archaeology sees as the best fit for the earliest indo-Europeans was spread over a vast area from the Dnieper to the Volga. They may have varied greatly in their percentages of the autosomal strands but we're united by a language of the elite who controlled the network who carried Balkans metals sectarian asctge Volga

Anonymous said...

Can't we just say that MA1 and the Afontova Gora peoples spoke a language that would become Indo-Uralic later, and that everyone in that Eastern Europe to Steppe to Siberia region spoke some sort of diallect of it?

How is this unfeasable? Anatolian diverged too early?

Davidski said...

@Rob

To be fair, Ukraine_Eneolithic I5884 is from a burial on the North Pontic steppe, and can be mostly modeled as Ukaine_N, which are samples from the North Pontic steppe.

So it's difficult to divorce him from the steppe narrative.

mike said...

I have a hunch Anatolian didn't leave the steppes early but was just an isolate that remained in or bordering the steppes after the Yamnaya related wave had washed over and it had survived somewhere. It was not necessarily early to exit he stepped or enter Turkey. You get isolated languages surviving off the beaten track, up mountains etc all over the world. It could have survived somewhere like Crimea or the Caucasus after 3000BC and remains there for centuries before expanding into Anatolia

mike said...

David - it's important to note that IE didn't just appear out of nowhere with Yamnaya. Most archaeologists attribute Sredny Stog as it's archaic precursor. Its elite was also in charge of the only huge pre-yamnaya trade network - which stretched from the Dnieper to the volga-Urals and they settled acres a similar zone. Now the roots of Sredny Stog are interesting as it seems to have aspects that came from the Volga end but yet it's classic form car together towards the Dnieper-Don and it's trade was distinctly west to east as it was passing along Balkans metal as far as the VolgaUrals. So that archaic IE group had a very complex criss rossing

Chad said...

No Yamnaya? That's not what I'm seeing with that sample.

Richard,
I'm talking specifically about L51 in Hungary, 2500-2200BCE.

Chad said...

The guy is a few hundred years after Yamnaya samples too. You guys are being anachronistic.

agiering said...

It's interesting that Samara EHG is r1b, while Karelia (which is a millennium or so earlier) is r1a and J. Iron Gates HGs have lots of r1b, and they are even older than Karelia. R1b shows up in Latvian HG earlier than Samara, but Latvia Hg is a WHG/EHG blend.

Is there any indication that Samara EHG is closer to Iron_Gates than Karelia is?

Davidski said...

@Chad

No Yamnaya? That's not what I'm seeing with that sample.

I can model him without any Yamnaya ancestry both in the Global 25 and with qpAdm.

The guy is a few hundred years after Yamnaya samples too. You guys are being anachronistic.

But he comes from Dereivka, thus from an archaeological context that precedes Yamnaya.

So considering that Sredny Stog (Dereivka) people are generally accepted to have taken part in Yamnaya ethnogenesis, it's a valid question as to whether Ukraine_Eneolithic I5884 got his Z2103 from Yamnaya, or whether his people contributed to Yamnaya.

More samples will settle this question, but right now it's an open question.

MomOfZoha said...

@Chetan:
"The PCAs are getting more and more cluttered by the day. I hope future papers would move towards more "interactive" PCA plots"

Here you go:
https://drive.google.com/open?id=1y00csTnGWyoipDx_Xn8esLZaMXvAKxKB

All the GML files may be viewed interactively via Gephi:
https://gephi.org/

But, I also included this PDF of the min-connectivity KNN graph for those ancients
https://drive.google.com/open?id=1iNl13puqg0EeAzky-2aqfzdcQwHNGxxk

In experimenting with the geometric graphs (with *-Geo* in filename), the graphs were way too connected with some other samples -- and *still* are at an average degree above 40. Nonetheless, it was useful to see the effects of my clustering algorithm at this stage, as I believe it reflects an "earlier split" between population groups. You can see the clustering via the "clusterID" column in the GML files (and I also colored the clusters in the pdf). You can also color and resize nodes as you wish using a variety of existing as well as to-be-computed attributes in Gephi, though I don't have time to go into that now as I need to go to my mini VIP now.

I might do an explanatory blog entry myself about all this later. It only uses distance statistics, and is hence consistent with PCA type information. It does not directly yield phylo tree info though it is an open question the extent to which some such info may be inferred...

Cheers, and also thanks a lot to David whose PCA 25 makes all this possible. Dzieki, Davidski.

Adios!

Chad said...

Did you try qpGraph? Better than qpAdm.

Davidski said...

I can model him as part Yamnaya with Global 25 and qpAdm, and probably with qpGraph. But I don't need to do this to explain his ancestry.

So there is a chance that he has no Yamnaya input at all, and the fact that he shows some Yamnaya-related ancestry in Mathieson's supervised ADMIXTURE run might be explained by the possibility that Sredny Stog contributed ancestry to Yamnaya.

Davidski said...

This guy is mostly Ukraine_N and Trypilla I reckon, with some other minor stuff, which may or may not be from Yamnaya.

Actually, using the Global 25 I can model Yamnaya_Samara as CHG, EHG and Ukraine_Eneolithic I5884.

mike said...

There is a possibility that what we are seeing in Z2103's dominance of the Yamnaya kurgans is a wagon riding royal clan and the bulk of the pop was simply not buried in that prominent way. If L51 derivatives and the R1a were basically the invisible rest of the population, that would explain why they seem so yamnaya derived despite not turning up in yamnaya kurgans. It may only be when they arrived in old Europe that they emerged to do their own thing

Davidski said...

@All

Global25/nMonte

[1] distance%=2.8944 / distance=0.028944

Ukraine_Eneolithic:I5884

Ukraine_N 71.4
Trypillia 28.6
CHG 0
EHG 0
Yamnaya_Samara 0

qpAdm...

https://drive.google.com/file/d/17IVNHf0rIVGyl0286FzcDHUluHmWJuAX/view?usp=sharing

Coincidence?

@mike

R1a-M417 isn't invisible. It's found on the Eneolithic Ukrainian steppe in the first male sampled there, as well as in the oldest Corded Ware graves, including the one with the "Yamnaya" pin.

If R1b-Z2103 was an important lineage in this population that expanded from the Ukrainian steppe across Northern Europe, it would've shown up in Corded Ware burials by now.

It may be that R1a-M417 and R1b-M269 initially came from Sredny Stog, but then took different trajectories of expansions due to patrilocality and patrilineality on the steppe.

mike said...

David just meant that on the steppes prior to 2800BC Z2103 may have been a royal dynasty but they were wedded to a system that only worked on steppe type land. As soon as they crossed out of that zone they lost their power and prestige and it seems clear they failed to adapt by the way Z2103 grinds to a halt

Alberto said...

But what's the deal with this I5884 sample (mis)labeled as Eneolithic? If he was from 3800 BCE then it would be quite relevant. But at 2800 BCE he's just an outlier without further significance.

@eric

Davidski, think you could mark where all the Varna, Balkan Bronze/Iron Age samples(including Malak Preslavets, etc) are? The South East Europe paper's PCA had obvious projection bias.

As Davidski explained before, the datasheet is available for anyone to get it and put it through a program called PAST (https://folk.uio.no/ohammer/past/).

Here I put the Neolithic/Chalcolithic samples from the Balkans (Balkans_N, Balkans_Chl, Malak_Preslavets, Krepost_N, Trypillia, Varna, Vinca_MN) in Olive Green. The Neolithic Greeks are in lime green. And Mycenaeans in fuchsia.

https://drive.google.com/file/d/1zpXnHABkslmcgrY-oW13R7ROlvhqZJAU/view?usp=sharing

The one that clusters with Mycenaeans is Krepost_N.

mike said...

Rob the answers won't be simple. Ethnogenesis is a very complex mix of multiple often not simultaneous aspects of genes, languages, culture, subsistance, technology and religion fusing from different sources. It happened in a zone of interaction not a single y lineage. It was also a long process and it's widely agreed that much earlier Sredny Stog (which was centred much further west in the steppes than early Yamnaya) was the archaic phase of IE ethnogenisis.that culture stretched from the Dnieoer bordering the farmers to the Volga and it is lasted a large chunk of a millenium

Rob said...

@ David

"So it's difficult to divorce him from the steppe narrative."

That's not the point, as the discussion was genetic, not geographic

Anyhow, the individual seems unimportant for BB, as it really needs something from further east to fulfill steppe ancestry.

Beaker_Central_Europe
Globular_Amphora 38 %
Yamnaya_Samara:I0370 37.8 %
Vucedol:I3499 9.35 %
Ukraine_Eneolithic:I5884 8.65 %

BTW can you add Krepost to your G25 ?

Rob said...

@ Mike

"Rob the answers won't be simple"
Preaching to the choir, bud.

mike said...

Does 15884 date as late as 2800BC? Kind of removes his significance given that their are other Z2103 classic Steele Yamnaya guys centuries older than him? No combination of y DNA and autosomal signal is impossible if the y lineage originated 1000 years earlier. It only takes 200 years living among another people to reduced your original autosomal signal to 1%. He may just be an oddity

mike said...

Yfull puts the MRCA of R1a and b at 20800BC. That is the height of the ice age about 4000 years into the lGM. Whatever refugium it was, all R1 was trapped in it together for many thousands of years, probably from 25000BC for 10000 years or more. So originally there must have been a shared R1 autosomal signal.

Davidski said...

@mike

Just meant that on the steppes prior to 2800BC Z2103 may have been a royal dynasty but they were wedded to a system that only worked on steppe type land. As soon as they crossed out of that zone they lost their power and prestige and it seems clear they failed to adapt by the way Z2103 grinds to a halt.

But Z2103 did make it out of the steppe, and did very well in Southeast Europe and West Asia. So based on this, and on the M417 in the earliest Corded Ware graves, it's clear I think that clans that carried Z2103 didn't migrate to Northern Europe.

@Rob

Anyhow, the individual seems unimportant for BB, as it really needs something from further east to fulfill steppe ancestry.

The eastern admixture first spread across the steppe, probably via both migrations and female exogamy between Sredny Stog/Repin/Yamnaya clans, and then it moved out of the steppe into the BB network.

BTW can you add Krepost to your G25?

Not enough markers.

@Alberto

If he was from 3800 BCE then it would be quite relevant. But at 2800 BCE he's just an outlier without further significance.

He might be very significant indeed, if it turns out that his ancestors mixed with, say, Khvalynsk to form Yamnaya.

Rob said...

theres just something which doesn’t gel with the clasificatjon/ dating/ context / genotype of that individual
I agree it’ll be big news if he’s 3800 BC, eg.

Davidski said...

I've e-mailed Iain to ask what's up with this sample, because the dating and archaeological classification do contradict, as far as I can see; 2890-2696 calBC should be the Early Bronze Age even on the North Pontic steppe.

a said...

Poland/Hungary Bell Beakers-Genetiker's results-we will have to get further confirmation from S.M.

Some nice variance-

I4178 Hungary Bell Beaker 2500–2200 R1b1a1a2a1~L51
I2787 Hungary Bell Beaker 2458–2202 R1b1a1a2a2-Z2103
I4251 Poland Bell Beaker 2837–2672 R1b1a1a2~M269
I4253 Poland Bell Beaker 2571–2208 R1b1a1a2a2-Z2103
I4252 Poland Bell Beaker 2463–2142 R1b1a1a2a1a~L151

Samuel Andrews said...

Thanks, a. The fact Z2103 and L51 coexisted in eastern Beaker folk means Z2103 clans also went pretty far into Europe. Probably with Vucedol. Vucedol and whatever Steppe group L51 comes from must have mixed with each other in central-east Europe.

Maybe, both L51 and Z2103 clans were side by side as both they moved deep into Europe around 2800 BC.

Samuel Andrews said...

@Everyone,

Does anyone have an explanation for the extreme Y DNA founder effects in Bronze age Steppe people(s)?

It looks like most eastern Beaker folk belonged to R1b U152>L2. And it looks like a high percentage of British Beaker folk belonged to R1b L21. yfull estimates DF13 dates to 4,300 and L2 to 4,500.

The trend exists in Yamnaya and Sintashta. A large percentage of men in each group traced their paternal line to the same guy who lived maybe only a few hundred years earlier.

This isn't normal. Founder effects are normal in human uniparental lineages. But such young founder effects are not. IMO, some kind of unique cultural practice caused this to happen in Bronze age Steppe folk.

Did they record people's paternal ancestry? Did kings have lots of wives, then their sons had lots of wives, etc. While, everyone else had one wife or none.

Nirjhar007 said...

I've e-mailed Iain to ask what's up with this sample, because the dating and archaeological classification do contradict, as far as I can see; 2890-2696 calBC should be the Early Bronze Age even on the North Pontic steppe.

R1b is from ITALY bud ....

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

(cont'd)

I just found a k-nearest neighbor graph package in R by googling:
https://www.rdocumentation.org/packages/GeNetIt/versions/0.1-1/topics/knn.graph
Geometric graphs are also easy to construct: Given parameter r, connect any two nodes iff they are within distance r to each other. Very easy, with high dependence on r, just as kNN depends on k.

One kind of default parameter to use is the minimum k or r that yields a connected graph. That works well if your data should be at least distantly related, which is already the assumption of any DAG or phylo tree from same data. Nonetheless, there is a sensitivity to k and r, respectively, as well as very different properties of the resulting kNN versus geometric graphs.

Finally, on the subject of clustering, let me emphasize that phylogenetic trees are also often formed by clustering via pairwise distance information alone. So, a hierarchical clustering of the graphs (iterated to the lowest level) also can yield that information, with the advantage that this clustering need not be bi-branching only. I did not attempt this as admixture events are far more important and common impetus for major changes in the human genome at micro-levels considered here, and I don't consider admixture yet.

A good admixture model requires genetic information distinct from distance information, except only in cases where we are considering adjacent nodes that are so distant in time that the "directionality" must go from older-to-younger. That information is not captured by the PCA coordinates. The undirected graphs generatable from the distance information alone are alternative visualizations to PCA, with their own important advantages even if they do not yield phylo-admix-DAGs directly.

I also like Gephi as a graph visualization platform for a lot of reasons, including all the stats and even clustering it computes automatically with the "Statistic" right side bar. Network diameter computes major centrality measures, and modularity is a fast randomized clustering, e.g.. All the node-level information can then be seen in the "Data Laboratory" tab in the middle. From "Preview" you can generate and save the *labelled* graphs as pdf/png/svg by clicking "Show Labels" under "Node Labels".

The graph layout algorithm I prefer is "Force Atlas" or "Force Atlas 2" though Fruchterman-Reingold can be used in combination sometimes when the clusters get too sticky.

On the left "Appearance" sidebar, you can color and size nodes (and edges) based on any attribute. I used my precomputed clusterID attribute with a "Partition" based coloring. I recommend to size nodes based on betweenness centrality (after computing the Network Diameter statistic).

I have not used any of the pop gen graph packages that require the genome information (not just distance info), though I am familiar with some packages that create graphs from gene expression data (microarray or RNAseq). You can convert the adjacency matrix output to a GML easily, and it will still be useful to visualize and analyze in Gephi. Any graph that you get from any of the packages you are using -- all kinds of directed, weighted, attributed graphs -- can be beautifully visualized and analyzed in Gephi. Period.

https://gephi.org/

Rob said...

Here's the R1a Ukraine Eneolithic

Ukraine_Eneolithic:I6561
Samara_Eneolithic:I0122 55.75 %
Balkans_ChL 27.05 % + Trypillia 3.3 %
Armenia_ChL 13.3 %

So R1a-M417 came from the east, or north, R1b-M269 and I2a2a1b from west ?
Also some Q's were appearing in Caspian region too, adding EHG/ ANE type ancestry.

Davidski said...

@Nirjhar

I'm pretty sure that both R1a-M417 and R1b-M269 came from the steppe. No way around that now, and from now on you'll be seeing more of the same of what was said in Olalde et al. 2018.

@MoZ

The thing about qpGraph is that it doesn't just visualize data but analyzes it.

@Rob

I think the potential pitfall with your model is that Balkans_ChL and Armenia_ChL probably have some admixture from the Eneolithic steppe, and if so, this will make it look like Ukraine_Eneolithic:I6561 has ancestry from the Balkans and Armenia.

Rob said...

@ Dave

"I think the potential pitfall with your model is that Balkans_ChL and Armenia_ChL probably have some admixture from the Eneolithic steppe, and if so, this will make it look like Ukraine_Eneolithic:I6561 has ancestry from the Balkans and Armenia.'

Yes but that's not a pitfall . All populations by the Chalcolithic around the Black Sea had started to co-mingle. So what't the result suggests is extra/ ongoing admixture as time went on toward 2800s BC. Unless you find the idea of admixture from those 2 regions unrealistic, which would be unrealistic of you.

Rob said...

But here is a more 'basic', zoomed out look, to avoid the potential cross-overs

Ukraine_Eneolithic:I6561
EHG 45.2 %
LBK_EN 33.6 %
CHG 17.5 %
Ukraine_N 2.9 %

Again, fair whack of LBK.

Archaelog said...

@David The Ukraine Neolithic has a much higher percentage of EHG though (as per Fig 1d in Mathieson). The Eneolithic Ukraine's EHG is much reduced at the cost of CHG and EEF. But since the y haplogroups show continuity from Neo to ENeo, we can assume the components came via the female line.

Is that what you are suggesting in your plot


@mike Khvalynsk was the archaeological precursor of Yamna but with contributions from Sredny and Repin

Samuel Andrews said...

I keep getting Globular Amphora as the farmer source in northern Bell Beaker. This would make modern-day Poland the place where Steppe R1b P312 and farmer mixed. Also, all northern Beaker get about 7% Narva. Maybe the admixture happened somewhere in northern Poland?

Archaelog said...

@Mom of Zoha Ah excellent! I hope more future researchers follow this precedent and come up with better readable graphs

Anonymous said...

@Samuel

"Does anyone have an explanation for the extreme Y DNA founder effects in Bronze age Steppe people(s)?"

Upper class polygamy? Just a wild shot, but Tacitus mentions that rarely very high born Germanics had more that the usual one wife. Might be a left-over from a more common practice.

Also read this: http://www.hinduwebsite.com/hinduism/h_polygamy.asp

"here are four castes (varna, viz.) Brâhmanas, Kshatriyas, Vaisyas, and Sûdras. Males belonging to them may take wives according to the order of the castes, viz. a Brâhmana four, a Kshatriya three, a Vaisya two and a Sûdra one."

Archaelog said...

@epoch2013 Almost every Hindu text that mentions caste related rules was written by the Brahmanas themselves and it's unclear to what degree the rest of society was even aware of them. For example, they never recognized any or very few as Kshatriyas after the Vedic age. Overall the Indian caste system with its exaltation of the priest class is an anomaly wrt to other cultures and I think it is unique to India. The other IE societies had priests, but they were primarily warrior based cultures.

Anonymous said...

@Chetan

OK. I meant it just as an example, mind you, I am hardly known with the finer details of Indian culture.

Samuel Andrews said...

I6580, Poland, 2300–2150 BCE. R1b1a1a2a1a2b1

Yamnaya_Samara,56.7
Globular_Amphora,36.7
Narva_Lithuania,6

Netherlands Beaker.

Yamnaya_Samara,55.7
Globular_Amphora,37.5
Narva_Estonia,7.1

Now, look at the most Steppe rich Iberian Beaker individual.

I6623
Yamnaya_Samara,34.4
Iberia_Southwest_CA, 37.4
Globular_Amphora,22.5
Narva_Estonia,5.2

I'm thinking Bell BEaker Netherlands is the best representative of "pure" Beaker folk who emerged somewhere in central-east Europe.

Looks like I6623 in Iberia got its Steppe admixture from people with both Globular Amphora and Narva-like admixture.

So, to estimate R1b-P312 Steppe-rich admixture in Iberia today I think Beaker Netherlands or Beaker England would be the best proxy.

Anonymous said...

I am curious when Srkz will have a Bell Beaker IDB map.

Aram said...

Well that Yamna outlier shows probably that the southern admixture was not a pure CHG.


Rob
I answered to Your question in the previous thread.

Samuel Andrews said...

Anyone notice Wales_MBA scores 66% Yamnaya? That makes it the best proxy for the Steppe ancestor of R1b L151. Also, 13% WHG and 20% Anatolia Neo. That's more confirmation that R1b L151 folk either mixed with farmers who lots of WHG or have heavy UkraineHG ancestry.

Aram said...

The Cucuteni Trypllian outlier that was present initially and then removed also had inconsistencies. His SNP was too young.

Ryan said...

@David - The fact that this guy doesn't cluster with neighbouring samples though suggests he received an input from somewhere else, no?

Rob said...

@ Aram

Thanks.

@ Sam
Your results are very similar to what I got (see above). Its cool the algorithm 'choses' Iberia Chalc for Iberian BB.

Davidski said...

@Ryan

The fact that this guy doesn't cluster with neighbouring samples though suggests he received an input from somewhere else, no?

I reckon he's the most local one out of all of the Ukrainian_Eneolithic samples, with the most Ukraine_N ancestry.

And indeed one of the closest dots to him on the plot is Ukraine_Eneolithic I5882. On the other hand, the Ukraine_Eneolithic R1a-M417 guy has the most "foreign" ancestry, from another part of the steppe, probably from Khvalynsk. But of course his Y-chromosome might be local too in any case.

If the dating of I5884 was different, say, 3500 BC, like that of the Ukraine_Eneolithic R1a-M417 guy, then that would basically seal the PIE debate for me.

[1] distance%=3.5977 / distance=0.035977

Ukraine_Eneolithic:I6561 (R1a-M417)
Yamnaya_Samara 63.05
Trypillia 33.4
EHG 3.55
CHG 0
Ukraine_N 0

[1] distance%=4.8248 / distance=0.048248

Ukraine_Eneolithic:I4110
Trypillia 38.1
Ukraine_N 32.45
Yamnaya_Samara 25.2
EHG 4.25

[1] distance%=2.8119 / distance=0.028119

Ukraine_Eneolithic:I5882
Ukraine_N 66.85
Trypillia 23.2
CHG 8.6
Yamnaya_Samara 1.35
EHG 0

[1] distance%=2.8944 / distance=0.028944

Ukraine_Eneolithic:I5884 (R1b-Z2103)
Ukraine_N 71.4
Trypillia 28.6
CHG 0
EHG 0
Yamnaya_Samara 0

Balaji said...

The long awaited aDNA results from the Indian Subcontinent must soon be coming. Here is a forerunner paper on the archaeological aspects of the Rakhigarhi cemetery.

http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0192299

https://video.nationalgeographic.com/video/news/180221-ancient-indus-civilization-cemetery-unearthed-india-vin-spd?utm_source=Facebook&utm_medium=Social&utm_content=link_fb20180221video-indiacivilization&utm_campaign=Content&sf182705908=1

Samuel Andrews said...

Getting interesting results modelling western Europeans as part Beaker_Netherlands. Spanish & French both score in Anatlolia_BA. Some kind of east med people came into western Europe after the Beaker period. Romans? I have no idea where it is coming from.

Samuel Andrews said...

The east med reference I'm using is Anatolia BA. Tuscans come out basically ANatolia BA+Beaker. So obviously, Anatolia BA is too rich in EEF to be a good east med reference.

None the less, it looks like Spanish and French have a very watered down version of the same East Med influence that Italians have.

Archaelog said...

@Balaji

"The long awaited aDNA results from the Indian Subcontinent must soon be coming. Here is a forerunner paper on the archaeological aspects of the Rakhigarhi cemetery."

Interesting. You know if any of the authors of this one is on the DNA project?

Slumbery said...

Samuel Andrews

Yes, if there is a "watered down" version of the Italian extra SW Asian in Spain and France, then the obvious guess for the source is Roman colonization. We know it happened in both regions.
Unless it is very old.

Alberto said...

@Davidski

He might be very significant indeed, if it turns out that his ancestors mixed with, say, Khvalynsk to form Yamnaya.

So you really prefer the option of that samples representing the early Z2103 population and becoming Yamnaya by female exogamy? Something like this?

Yamnaya_Samara
Samara_Eneolithic:I0122 40.6%
Samara_Eneolithic:I0433 30.6%
CHG:KK1 23.6%
Barcin_N 5.2%
Ukraine_Eneolithic:I5884 0%

Distance 3.9894%

And based on one single sample from 2800 BCE? Not sure why you would prefer such solution instead of accepting that this sample is a rare outlier and that's it.

Davidski said...

@Alberto

Not sure why you would prefer such solution instead of accepting that this sample is a rare outlier and that's it.

Because archeology suggests that Yamnaya was an amalgamation of Sredny Stog (ie. Dereivka) and Khvalynsk.

And R1b has been in Europe, deep in Europe, since the Upper Paleolithic.

So we'll see if this sample is a rare outlier or not when more samples come in. But my bet is that he's not, and I think that with more sampling we'll learn that R1b-M269 moved from Sredny Stog to Yamnaya.

Rob said...

Well it does seem that there is no M269 yet before 3000 BC, if we ignore any poor quality samples. Wouldn't that make Afansievo the earliest group in which M269 is from ?

Davidski said...

Afanasievo is not dated earlier than R1b-M269 Yamnaya_Samara I0429 3339-2918 calBCE.

http://eurogenes.blogspot.com/2015/10/essential-reading-paleoecology.html

a said...

Elite power/weapon scepters- signs of power, and weapons designed as zoomorphic symbols of the Steppe warfare raiding culture. R1b-Z2103+ Indra's Vajra 3335-2881_BC+/- Kutuluk R1b-Z2103+
http://www.academia.edu/3836804/An_Indo-Iranian_Symbol_of_Power_in_the_Earliest_Steppe_Kurgans
compared with bar copper celts from India.Ind
http://i61.tinypic.com/1zcpteb.png

https://www.youtube.com/watch?v=Z0HCs6PVnzI
@ 22:42 was not in Skelya-Sredny Stog culture of Elites. @21:30

a said...

@Samuel Andrews said...
"Thanks, a. The fact Z2103 and L51 coexisted in eastern Beaker folk means Z2103 clans also went pretty far into Europe. Probably with Vucedol. Vucedol and whatever Steppe group L51 comes from must have mixed with each other in central-east Europe.

Maybe, both L51 and Z2103 clans were side by side as both they moved deep into Europe around 2800 BC."

You might also be interested in the other place they are found together that has promise of showing Horse Husbandry around 2500 B.C.[onset of expansion phase] in related site -BB Hungary sites we also find L51+ and Z2103+. A well built man around 6' [with weaponry] on a tamed large stallion might have been quite an imposing figure back then. Parsing the horse data from BB-CW and Sintashta/Swat Pakistan might hold some [trail]clues.

Skeletal indicators of domestication-

"Some researchers do not consider an animal to be "domesticated" until it exhibits physical changes consistent with selective breeding, or at least having been born and raised entirely in captivity. Until that point, they classify captive animals as merely "tamed". Those who hold to this theory of domestication point to a change in skeletal measurements was detected among horse bones recovered from middens dated about 2500 BCE in eastern Hungary in Bell-Beaker sites, and in later Bronze Age sites in the Russian steppes, Spain, and eastern Europe.[6][30] Horse bones from these contexts exhibited an increase in variability, thought to reflect the survival under human care of both larger and smaller individuals than appeared in the wild; and a decrease in average size, thought to reflect penning and restriction in diet. Horse populations that showed this combination of skeletal changes probably were domesticated. Most evidence suggests that horses were increasingly controlled by humans after about 2500 BCE. However, more recently there have been skeletal remains found at a site in Kazakhstan which display the smaller, more slender limbs characteristic of corralled animals, dated to 3500 BCE.[3]"


"As we look to the Olalde paper, the Szigetszentmiklós individuals are in a genetic sense, a society of mixed ancestries. The individual pictured above (I2787) has the highest concentration of the Steppe-like ancestry of any individual within the Beaker world, and probably Western Europe for that matter. At the same rate, Szigetszentmiklós has an individual (I2741) who exhibits nearly zero Steppe-like ancestry."


http://bellbeakerblogger.blogspot.ca/2017/07/szigetszentmiklos-cemetery-santas-six.html



Anonymous said...

@a

Your information about horses in Kazakhstan is outdated.

Richard Rocca said...

Can someone check to see where Italian P312+ Bell Beaker sample I2478 attained its non-steppe component? Should be Globular Amphora and/or Remedello, but one never knows.

Jijnasu said...

@ balaji "The long awaited aDNA results from the Indian Subcontinent must soon be coming. Here is a forerunner paper on the archaeological aspects of the Rakhigarhi cemetery."
Thanks for the link. This seems to be the same team working on the rakhigarhi DNA. Guess the Rakhigarhi DNA is on its way to publication after a long wait.

Alberto said...

@Richard Rocca

This is what I get with Global 25:

Beaker_Northern_Italy:I2478
Iberia_Southwest_CA 51.05%
Yamnaya_Kalmykia 33%
Tisza_LN 12.95%
Armenia_ChL 2.9%
Ireland_MN 0.1%

Full output

So probably something local best represented by Iberia_SW_CA (with the steppe part coming from Hungary?). Let me know if you'd like me to add/remove any population to the sources list (but I'll have to check it later).

a said...

@supernord said...
@a-"Your information about horses in Kazakhstan is outdated"
We have to wait for the new papers to be peer reviewed-
FYI something that might be relevant.
Sarmatian found by the ruins of the Sinashta settlements was under
R1b- Z2108/9+
Sample Culture Date BC Haplogroup
I0575 Early Sarmatian 500–100 R1b1a1a2a2c1-Y21707* calls
I0577 Aldy-Bel 700–500 R1a1a1b2a2-YP1456* calls
I0563 Pazyryk 400–200 R1a1a1b2a2-S23201* calls
IS2 Zevakino-Chilikta 900–600 Q1a2a1c1-L332* calls

The same branch as sample I7044 the region with alot of horse,remains and possible signs of horse husbandry.

I7044 HUNG498, grave 13 Petrous M U5b1d1b R1b1a1a2a2c1 784178 1240k capture This study
I7045 HUNG499, grave 1 Petrous F T2c1d+152 755438 1240k capture This study
I2741 GEN_20, Grave 49 Petrous (cochlea, left) M H1+16189 I2a1a1 871912 1240k capture This study
I2786 GEN_56, Grave 133 Petrous (cochlea, right) M I1 I2a2a 795354 1240k capture This study
I2787 GEN_59, Grave 688 Petrous (cochlea, right) M T2b R1b1a1a2a2? 669314 1240k capture This study
I4178 GEN_58, Grave 552 Petrous (cochlea, left) M J1c1b1a R1b1a1a2? 587361 1240k capture This study


Modern day Ossetians-Z2110+>5587>5586 downstream from R-Y5592*
We will soon find out more- Hungarian/Polish BBsamples.
Another point R-Y5592*nodal position-is found in Sweden.In theory all these samples should be related.
R-Y5592 Y5592 formed 4800 ybp, TMRCA 4800 ybpinfo
R-Y5592*---id:YFXxXxXX

Anonymous said...

mike:

Feel I should point out, it would be stretching to say it was clear that Sredny Stog was Indo-European. There is no documented evidence of Indo-European until thousands of years later.

Now, it's true that it seems obvious that PIE (or "Late PIE" or "Proto-Indo-Hittite") was spoken on the steppe, and then colonised Europe. Sredny Stog is one of the cultures that appears culturally linked to the presumed speakers of PIE (let's say the Yamnaya). But we do know that language, genes and material culture are not always linked, particularly when there are significant, known cultural exchanges. Hungarian, for instance, is a Uralic language brought to central Europe by a broadly 'Turkic' culture - either Uralic speakers adopted Turkic steppe culture, or Turkic speakers adopted a Uralic language.

So if we're saying that "late Yamnaya" was the product of both Khvalynsk and Sredny Stog, its language, PIE, could easily have been the language of EITHER parent culture, regardless of which culture the child most took after. No? Particularly if, as I understand it, it's believed that it was a Khvalynsk-derived culture that expanded into Sredny Stog territory? The invaders could easily have kept their language while adopting some of the material culture of their neighbours.


If Sredny Stog and Khvalynsk spoke different languages, and both cultures contributed to late Yamnaya, I don't think we can discern which (if either!) language spoke pre-PIE.


[AIUI Sredny Stog had greater CHG ancestry than Khvalynsk? Is that true? If so, it would be nice if SS were the origin of PIE, because it would be nice if we could derive PIE from somewhere in the Caucasus/pre-Maikop area. This would explain the superficial similarities to modern Caucasian languages, and it would give Anatolian a route into Anatolia (we could imagine that PIH was spoken by pre-Maikop, PIE by Sredny Stog, and Anatolian by Maikop or its descendents, who then flooded into Anatolia much as the Hurrians did. However, this would mean that PIE was transfered into Sredny Stog by female migrants, which isn't impossible but isn't exactly the most likely scenario, so... (it may also not line up with the timeline properly - I'm not sure Anatolian is THAT divergent)]

Anonymous said...

Regarding large-scale polygamy: certainly if we could assume a harem-holding culture, that would be hugely useful!

- harems would indeed explain the huge and rapid founder effects that seem to be needed

- harems would be associated with the divergence between social classes apparently seen in the development of Yamnaya (societies that have harems only have them for the ruling class)

- harems would help explain the massive population explosion that would seem to be necessary to explain the PIE expansions

- polygamy is often associated with land-rich societies, precisely because it leads to more rapid population growth (and specifically because in a land-rich society, competition is through the control of labour, which promotes competitive fertility between families). That's exactly what you'd have on the steppe once farming was introduced in a fertile period.

- such cultures often operate an 'open system' of slavery: women would be enslaved from neighbours, but eventually they and their children would be integrated into the family of their owners. So you'd see 'conservative' Y-DNA, because the husbands are all from the slave-taking society, but cosmopolitan DNA overall, because much of the population would descend from imported slaves.

- once invaders reached more populated areas of Europe and India, harem-taking would magnify the genetic effect of even a relatively small migration, further explaining how a small tribe from the steppe could subjugate an entire continent...

- such a society would also be militaristic, which the PIEs seemingly were.




This seems to make a lot of sense. The downsides?

- the PIE vocabulary does not scream 'harems'. There's no clear word for 'sister-wife', for instance, or 'sister-wife's son'. However, this lack of vocabulary could be explained by saying that harem polygamy was a relatively new phenomenon on the steppe at the time of the expansion, not fully reflected in the language, and that any traces of such vocabulary that may have developed in different dialects were later lost when migrants found themselves in more populated environments and settled back down into monogamy. We could also say that harems only occured in the ruling class and didn't penetrate the language of the ordinary people in any detail. Nonetheless, the lack of any clear relevent vocabulary is certainly a blemish on the theory.

[although there are a few slight details that could be analysed as reflecting echoes of wife-stealing and/or polygamy. But only very speculatively.


- is there archaeological evidence of harems? I don't know.

Ryan said...

@David - what happens if you add KO-1 or Iron Gates as an option?

"If the dating of I5884 was different, say, 3500 BC, like that of the Ukraine_Eneolithic R1a-M417 guy, then that would basically seal the PIE debate for me."

Is there anything left to seal really? R1b-Z2103 and R1a-M417 span the entirety of the Indo-European language family. All that's left is understanding what came before and after IMHO.

Anonymous said...



Does anyone know the exact date of the sample SZ1 Szólád (bronze age) R1a1a1b2a2a-Z2123?

Archaelog said...

The IA migration into India was by no means a small group of warriors. The idea of a small warrior band taking over the rest of the Indian population is quite outdated IMHO. Instead there must have been a numerically significant incoming steppe+mixed people who assimilated steadily with the already appreciable Indian population. In India even millennia after Indo-Aryan cultural dominance, the native male lineages still survive substantially even in the north. That ought to give you an idea of what the process would have been like. But the Indo-Aryans must at least have been numerically significant for R1a-z93 to form 40+% of the current Indian male population.

Rob said...

Some of you guys seem to be just making things up.
There hasn’t been found any clear social class in Yamnaya. In fact,compared to earlier periods, there was downtrend in social distinction- they were essentially internally egalitarian bands moving about, moving out and periodically replacing each other

Palacista said...


@vacuouswastrel

If Sredny Stog and Khvalynsk spoke different languages, and both cultures contributed to late Yamnaya, I don't think we can discern which (if either!) language spoke pre-PIE.

Unless they were closely related and SS was the precursor to IA etc and the rest lead to proto IE

bellbeakerblogger said...

@Andrew
"I keep getting Globular Amphora as the farmer source in northern Bell Beaker. This would make modern-day Poland the place where Steppe R1b P312 and farmer mixed. Also, all northern Beaker get about 7% Narva. Maybe the admixture happened somewhere in northern Poland?"

I'd bet a lot of that type is ancestry from the Western CWC, but it would be interesting to see the excess WHG and R1b coming from a place like Latvia after being CW-ized. But it's hard to see how that works in any reasonable timeframe for early Beakers. I'd estimate that 40-50% of Northern Beaker ancestry is CWC, so for now that's where I'd put GAC and Narva related ancestry.

Al Bundy said...

@Rob Personal preferences aside I would go with number 2, then 3, and then 1 in your list of PIE scenarios.Of course, as you've said,if future Myceneans have loads of R1A or something it would be 3, pending Anatolian and no big surprises from India.The issue with 2 is that J would presumably have to be in the PIE community along with M417 and I don't keep up with this but no J has been found yet on the Pontic Caspian steppe?Or M417 came from the North Caucasus? M417 spread IE to Europe in classic kurgan and J spread IE to to Anatolia and Greece.I'm surprised you think that number 1 is more likely than 3, I thought Anatolia was over with.

Samuel Andrews said...

Looks, like Globular Amphora might be an important source for the blonde hair in northern Europe. So far, Globular Amphora has the same allele frequencies as Latvia BA, Andronovo and modern northern Europeans. Wilde 2014 also tested pigment SNPs in Globular Amphora.

https://genetiker.wordpress.com/2018/02/24/pigmentation-of-ancient-southeastern-europeans/

That's sort of a twist in the story. "Nordic" pigment traits might derive from "Meditreaen"-like ancestors.

Rob said...

@ Al
Yep makes sense
Definitely the Anatolian hypothesis is over; but that’s not what my scenario 3 is.
Rather is west Pontic vs Volga Pontic vs Caucasus - Pontic models. If you see Palacista’s comment, you can see it’d be difficult and somewhat artificial to favour either one
But give the genetic distinctness, they’d be separate languages; exerting substrata effects on each other etc

Unknown said...

@Davidski
Thanks for your PC files! As you know the devil is in the details. Focusing on the Eastern Mediterranean and Aegean areas: The late Neolithic Peloponnese samples are shifted toward BA Anatolia and Chalcolithic Anatolia with presumptive CHG input. The earlier Neolithic sample from the Peloponnese aligns with the early Greek Neolithic samples. The later samples are about 4000 BCE in dating and also cluster with Minoan Crete samples. The one Minoan--I9130--who is G2a in Y chromosome looks like the Early Greek Neolithic samples; the rest cluster with the late Peloponnese and the late Anatolian (Chalcolthic/BA) samples. The data strongly suggest a movement circa 4000 BCE from Anatolia to mainland Greece, perhaps associated with J2a1 and the pre-Greek substrate languages (-ss- and -nth1 toponyms)

A said...

Samuel Andrews, blond hair was apparently already present in Scandinavian hunter gatherers and EHG: http://eurogenes.blogspot.co.uk/2017/05/european-blond-hair-may-have-originated.html

EastPole said...

@ Samuel Andrews
“Looks, like Globular Amphora might be an important source for the blonde hair in northern Europe. So far, Globular Amphora has the same allele frequencies as Latvia BA, Andronovo and modern northern Europeans.”


Light pigmentation is also present in Dereivka Late Sredny Stog:

https://s17.postimg.org/fqymqp7tb/screenshot_341.png

So maybe Andronovo and Eastern Europe got their light pigmentation, R1a and Indo-Slavonic languages from Dereivka.

Rob said...

Dave & Chad (& Roy) - what were you hoping to see in Greek Pelo ?

There's nothing much special about them - they're just West Anatolian Neolithics** - apart from I3920 - a female 3800 BC mtDNA H - which has ~ 12% Iran / CHG.


** = Boncukulu + 15-25 % extra Levant Neolithic (assoc. with hg T1 and E1b no doubt)
Full outputs here: https://imgur.com/a/xAlej

The ones which do stand out are Krepost, some of the Tepeciks (5000s BC) and Anatolia Chalcolithic, which show the Iran/CHG.


Also Roy, what about the J2a1 in LBK Austria, Sopot Croat MN, and Anatolia Neolithic I0708.
They all precede the Minoan.

Rob said...

Oh and everyone note the Iron Gates/ Ukraine Neol admixture in Preslavets. Sweet !

postneo said...

@Rob"Definitely the Anatolian hypothesis is over"

Initial Anatolian spread with EEF might be too old to be relevant but what about Anatolia after 5000 BC? Its a region adjoining Greece, Balkans and caucasus.

Davidski said...

@Rob

Dave & Chad (& Roy) - what were you hoping to see in Greek Pelo?

There's nothing much special about them - they're just West Anatolian Neolithics** - apart from I3920 - a female 3800 BC mtDNA H - which has ~ 12% Iran / CHG.


I was expecting the youngest Peloponnese_N individual, I3920, to show substantial CHG ancestry and cluster with Minoans. And I was right.

Chad said...

Best fit for Ukraine_Eneolithic I5884 requires Yamnaya and Globular Amphora. Trypillia makes a much worse fit. No Yamnaya does as well. 58K transversion sites here.

left pops:
Ukraine_Eneolithic_I5884
Ukraine_Neolithic
Globular_Amphora
Steppe_EMBA

right pops:
Mbuti_DG
Kostenki14
Ust_Ishim
WHG
Tianyuan
EHG
Levant_N
CHG
ElMiron
Iran_N
Anatolia_Neolithic
Steppe_Eneolithic
Armenia_EBA
Ukraine_Mesolithic

best coefficients: 0.563 0.304 0.133

std. errors: 0.059 0.047 0.078

fixed pat wt dof chisq tail prob
000 0 11 7.326 0.772118 0.563 0.304 0.133

Chad said...

Running transition and transversion sites, with the same outgroups, required me to add Narva to the run.

left pops:
Ukraine_Eneolithic_I5884
Ukraine_Neolithic
Globular_Amphora
Steppe_EMBA
Narva

right pops:
Mbuti_DG
Kostenki14
Ust_Ishim
WHG
Tianyuan
EHG
Levant_N
CHG
ElMiron
Iran_N
Anatolia_Neolithic
Steppe_Eneolithic
Armenia_EBA
Ukraine_Mesolithic

best coefficients: 0.337 0.238 0.216 0.208

std. errors: 0.099 0.033 0.055 0.079

fixed pat wt dof chisq tail prob
0000 0 10 10.442 0.402587 0.337 0.238 0.216 0.208

Chad said...

The last one is pretty much identical to the ADMIXTURE run I had. qpGraph is looking pretty similar.

Davidski said...

I wouldn't just look for the best statistical fit by using all sorts of nearby reference pops that may or may not be somewhat relevant.

I'd focus on the most plausible solutions, like the foragers from Dereivka and farmers from Trypillia.

Only if this doesn't work, I'd reach for something extra, but not Globular Amphora or Narva, and probably not Yamnaya, which isn't archaeologically attested in that area at that time.

Chad said...

Globular Amphora did reach into Ukraine. Globular Amphora came from around Narva. Yamnaya was already in Ukraine and even Hungary by this time. Not really any issues with the choices here. They're just the best fits. Narva wasn't necessary with transversions.

Samuel Andrews said...

Only one Beaker person with a red hair mutation. It's I2787, the one with Z2103 and 75% Yamnaya ancestry. He had three red hair mutations, so the oldest documented ginger.

Only two out of over 70 Beaker/Copper British indviduals have a red hair mutation. It looks like red hair frequencies more than quadrupled in frequency after the Beaker folk arrived in the British Isles. This makes it a uniquely British development, however, it could have happened simultaneously all over northern Europe including the northeast.

Samuel Andrews said...

@East Pole,
"So maybe Andronovo and Eastern Europe got their light pigmentation, R1a and Indo-Slavonic languages from Dereivka."

It's hard to say. This is all really interesting stuff. Ancient DNA keeps throwing out surprises. An important detail in the data in my opinon is that pigmentation traits can not be attributed to every member of a genetic group. WIthin, lots of genetic groups there was considerable variation.

For example, the difference in pigmentation between Globular Amphora and Iberian farmers was big. And the difference between Poltvaka and Srubnaya was big.

Maybe, both Globular Amphora and Corded Ware were depigmented.

Rob said...

@ Dave

"I was expecting the youngest Peloponnese_N individual, I3920, to show substantial CHG ancestry and cluster with Minoans. And I was right."

Aj, but a contemporary Peloponessian shows much less, and the older Krepost shows more. So what's the take home - They're all "Minoans" ? LOL


@ Postneo

Don't think so. 5000 BC is a period of change in Anatolia. I don;t think much was coming out of it, but rather moving in.

Balaji said...

@Chetan

Prof. Dong Hoon Shin and his team will surely by among the authors as well as Prof. Shinde and his team.

http://shinpaleopathology.blogspot.com/

We can also expect Prof. Reich and his team.

Davidski said...

@Rob

Aj, but a contemporary Peloponessian shows much less, and the older Krepost shows more. So what's the take home - They're all "Minoans" ? LOL

The take home is that Caucasus-related and other types of post-Barcin_Neolithic Near Eastern admixtures were already moving into the Aegean during the Neolithic.

That's how the Minoans came to be, and that's also what the pre-Greek population of the Aegean was.

Rob said...

Iron Age Bulgarian has fair whack of steppe ancestry

Balkans_IA:I5769
Balkans_ChL 41.35 % + Greece_Peloponnese_N:I3709 20.3 %
Anatolia_ChL 19.55 %
Yamnaya_Samara:I0370 15.2 %

Rob said...

@ Dave

"The take home is that Caucasus-related and other types of post-Barcin_Neolithic Near Eastern admixtures were already moving into the Aegean during the Neolithic.

That's how the Minoans came to be, and that's also what the pre-Greek population of the Aegean was."

Yes i agree with the first, because I suggested it years ago
As for the latter, Im not sure how generalisable that result is, for example to Epirus or Thessaly or Macedonia; although Im sure it's true for the Aegean islands

Anonymous said...

@Chad

How well does Blätterhöhle for Globular Amphora fit? GAC is derived from TRB, it is suggested and Blätterhöhle is close to TRB.

Secondly, is it simply the amount of WHG which makes te fit or is there something shared uniquely?

Chad said...

I haven't looked. I didn't include them, because they weren't in Ukraine. GAC is probably more Boleraz/Baden influenced than TRB cousins to the west.

It is hard to say anything about the WHG thing. In the first model it was needed, but the last model, with transversion sites, did not want WHG. So, I don't feel all that comfortable adding it. The safest assumption is with the transversion results.

I'll do some more playing around with qpGraph on this, but it does look like GAC admixture in Eneolithic Ukraine. That's where the strongest Z's are.

a said...

Pretty cool!

Some of the Hungarian[Horseman?] Yamnaya/Bell Beaker clones R1b-Z2109-R1b-Z2110 are different pigment than their Eastern Steppe cousins -coming out with red and blonde hair and or blue eyes like Stredny Stog sample.
I2787 GEN_59, Grave 688 Petrous (cochlea, right) M T2b R1b1a1a2a2 669314 1240k capture This study

I7044 HUNG498, grave 13 Petrous M U5b1d1b R1b1a1a2a2c1 784178 1240k capture This study

I2787 Bell Beaker Hungary 2457–2201 Light Red Brown
I7044 Bell Beaker Hungary 2500–2200 Light Blond/ D-blond Blue

I cant help but think that Bell Beaker I7044 from Hungary sounds like Tocharian man.

http://www.independent.co.uk/news/world/asia/a-meeting-of-civilisations-the-mystery-of-chinas-celtic-mummies-5330366.html
A meeting of civilisations: The mystery of China's celtic mummies
The discovery of European corpses thousands of miles away suggests a hitherto unknown connection between East and West in the Bronze Age. Clifford Coonan reports from Urumqi sounds like Bell Beakers description of Bell Beakers from Christmas island Santa's six foot helpers.

Gioiello said...

Samuel Andrews writes: “Actually, I2787 is a Czech Beaker who had R1b U152>L2”.

Ahahahah. The last reliables SNPs of I2787 are these:
R1b-M343-L754-L388-P297-M269-L23-L23/S141/PF6534
R1b-M343-L754-L388-P297-M269-L23-CTS7340.1-CTS1078/GG419/Z2103
R1b-M343-L754-L388-P297-M269-L23-CTS7340.1-PF7575/Z2104
R1b-M343-L754-L388-P297-M269-L23-CTS7340.1-Z2105

Do you understand of the Y how dow you understand of the mt, i.e. nothing?

You are only a moron with red hair!

Samuel Andrews said...

@a,

Linking mummies to Celts in any way was really pushing it. Plus, the "high frequency" in modern Isles Celts probably was caused by local natural selection after the Beaker period.

Several Beaker, Copper age (Beaker-descended people) had red hair mutations confirming it had some presence but nonetheless was extremly rare. I2787, someone of 75% Yamnaya decent, with 3 red hair mutation so for sure red hair is indeed really facisnating. That made my day.

The oldest confirmed red hair carriers are actually both EF(s).

I2423, 4520–4356, Bulgaria. rs1805008 3/4
I4303, 4778–4586, France. rs1805009 23/41

Steppe folk have not been extensively tested for red hair mutations. After a couple hundred EF(s) were tested literally only two carriers of a red hair mutation appeared. Five Afanasievo, Yamnaya, and Eneolithic individuals appear to carry a red hair mutation but need higher coverage to know for sure.

Aram said...

Can someone model the Bulgaria EBA and Kum4. It can give us some hints how Cernavoda was looking.

Rob said...

@ Chad

GAC is very interesting could be TRB or related derived, with West Balkan / Carpathian admixture.

Globular_Amphora
Germany_MN 46.5 %
Tiszapolgar_ECA 39.9 %
Iron_Gates_HG 9.15 %
Villabruna 2.4 %

This is intuitively correct, and it seems like it needs extra HG from Balkans.
Can you check with qp at some point ?

Chad said...

Yes, but I think German MN, Narva, and Protoboleraz will be the winners. Maybe even Blatterhohle and Boleraz. I'll try a couple quick before bed, then try again tomorrow, including qpGraph.

Rob said...

@ Aram

Here are a couple:

Balkans_BA:Bul6 (y Hg I2a2)
Balkans_N 41.7 %
Tepecik_Ciftlik_N 35.95 %
Ukraine_N 10.85 %
Iron_Gates_HG 9.4 %

Balkans_BA:I2510 (G2a2a)
Balkans_N 87.3 %
Trypillia 5.1 %
CHG 3.75 %
Tepecik_Ciftlik_N 1.6 %
EHG 1.3 %


Not much steppe there at all. The east shift is due to Anatolian impact.

Aram said...

Rob
Wow. This was the thing that I was suspecting. They didn't have Yamna. So maybe they were pushed there by proto Yamnayans. We could see there even that Ukraine Eneolithic Z2103 alongside I2a2a1b1b-s.

Chad said...

Rob,

Qp is saying basically no needed admixture from the Carpathians. I've tried Baden, Boleraz, and Tiszapolgar. Tiszapolgar didn't fail, but was at 5%+/-20%, so that is no good. Best model is Germany MN plus Narva 93/7, and a good fit. That's with transversion sites, so not a good sign for Carpathian admixture. I'll try qpGraph on this tommorow.

Rob said...

Yep. I also suspected the Caucasus - Anatolia - East balkans Road coming Into play ; with I2a2 from west Ukraine (although originally from Balkans / Hungary)

Rob said...

@ Chad

Ok but then maybe QP isn’t too good :)
Because the cattle burial style must have come from Hungary

Chad said...

Rob,

TRB had cattle burial too. Salzmunde was the big one. Burials aren't too different from GAC. Also not seeing Carpathian yDNA in GAC yet, if I'm not missing something.

Chad said...

qpGraph is the big dog. I'll run stuff through the ringer tomorrow.

Rob said...

@ Chad
"Also not seeing Carpathian yDNA in GAC yet"

GAC is I2a2a1a. The earliest such lineage to date is from Lepenski Vir.
And its not only cattle burials, but also imports from the Carpathian basin to Baltic shores (copper), and vice versa (Amber).
But I look forward to your graph

Unknown said...

@Rob,
The sites in Central Europe are J2a(xJ2a1) ie ancestral for L26, I believe. L26 is the major J2a lineage in the Mediterranean. The process likely is related to the first Late Neolithic settlements in the Cyclades, like at Saliagos or Iziktepe in Northern Anatolia.

a said...

Blogger Samuel Andrews said...
@a,

"Linking mummies to Celts in any way was really pushing it. Plus, the "high frequency" in modern Isles Celts probably was caused by local natural selection after the Beaker period.

Several Beaker, Copper age (Beaker-descended people) had red hair mutations confirming it had some presence but nonetheless was extremly rare. I2787, someone of 75% Yamnaya decent, with 3 red hair mutation so for sure red hair is indeed really facisnating. That made my day.

The oldest confirmed red hair carriers are actually both EF(s).

I2423, 4520–4356, Bulgaria. rs1805008 3/4
I4303, 4778–4586, France. rs1805009 23/41

Steppe folk have not been extensively tested for red hair mutations. After a couple hundred EF(s) were tested literally only two carriers of a red hair mutation appeared. Five Afanasievo, Yamnaya, and Eneolithic individuals appear to carry a red hair mutation but need higher coverage to know for sure. "

No need to evoke introgression of Basal Eurasian and or EEF and or Levant/African evolution in Yamnaya R1b-Z2103 samples.Take a close look at the R1b-Z2103 skulls- typical Cromagnum/Neanderthal mix, with strong mastoid process. The same with genes for positive selection for blonde hair and blue eyes, Samarra HG R1b--M73 7500YBP+/- .

Evolutionary forces of red hair date back to the archaic humans of
Europe-perhaps tens of thousands of years older than EEF. Nature has many examples of parallel evolution on selection. Archaic humans in Europe predate EEF component. Fragment of dna melanocortin 1 receptor(MRC1) from [Spain and Italy] from El Sidron 1252 and Monte Lessini the same modern humans with two copies of the mutation show the same pale skin and red hair.

You don't see red hair very common in sub Saharan and or Australian/Asian.

Slumbery said...

@a + Samuel

I do not have any strong opinion on the ultimate source of gingers, but I note that those late Neolithic Balkan farmers had WHG ancestry probably in two layers. Anatolian Neolithic itself seems to have WHG-like ancestry to begin with. (My guess is WHG related population in the South Pontic during LGM that got absorbed by a population expansion from the Levant.) + Mathieson et al. found non-sex-biased mixing between local HG and EEF in the Balkan.
So those Balkan red hair alleles might came from WHG regardless of the farmer context.
Or not, but so far I do not know of any population with gingers among them that lacks European HG ancestry. (Although the modern frequencies are almost certainly the results of much more recent local selections.)

Alberto said...

@Eren

Off topic, but I looked at the scaled files posted by Davidski and I wonder if you'd consider rounding the numbers to keep things a bit cleaner and easier to manage. Basically something like in line 9:

else {tableSc <- data.frame(round(mapply("*", table, eVs), 6), row.names = popNames)}

I've always done it this way and it shouldn't change the results (the original files are rounded to just 4 decimals anyway).

Eren said...

@Alberto:

Sure, no problem. Since we are at it, what is the preferred output format: txt or csv?

Matt said...

Been waiting since Tues to begin having a look (went on holiday on exactly the day the deluge was released! lots of reading to do).

Some basic reprocessing through PAST3 PCA for the West Eurasian datasheet, after identifying clusters through k-means, then removing clusters and populations which can be considered (subjectively) non-European*:

Ancients alone: https://imgur.com/a/6cwC1
Ancient+Moderns: https://imgur.com/a/qh97u

This basically just allows us to visualise with a *bit* less clutter (that sheer bulk of samples useful for a general view but maybe not that meaningful considering that only a subset of populations are plausible recent European ancestors).

Consider the Bell Beaker populations, a dimension opens up in this reanalysis at PC3 which splits Iberia/Atlantic from Danubian/Hungarian.

It looks to me like (from purely visual analysis):

a) Globular Amphora is intermediate on Iberia/Atlantic to Danubian/Hungarian (possibly related to European HG being intermediate on the Iberia/Atlantic to Danubian/Hungarian dimension).

b) Globular Amphora fits better for the Beaker samples considering Yamnaya_Samara as the non-EEF (steppe) reference, but! considering CWC_Baltic_Early or the single Yamnaya_Ukraine sample, the best fit may instead be Atlantic European.

This is a bit more apparent applying Linear Discriminant Analysis to the data: https://imgur.com/a/MCZU9 (to maximise visualization of between group variation)

There are very small offsets between Yamnaya_Samara, CWC_Baltic_Early and Yamnaya_Ukraine which mostly look to me to relate to CWC_Baltic_Early and Yamnaya_Ukraine swapping a little CHG related admixture / Yamnaya_Ukraine_Outlier admixture for specifically Hungarian/Danubian EEF (CWC_Baltic_Early and Yamnaya_Ukraine and Yamnaya_Samara look to have about the same amount of EHG ancestry?).

These may end up being decisive in which European EEF population is *chosen* whether you pick Yamnaya_Samara, CWC_Baltic_Early or Yamnaya_Ukraine as the steppe ancestor. (Same for if you use the CWC_Germany centroid) :/. GAC and the British/French Neolithic are about as separated on the dimension splitting Iberia/Atlantic to Danubian/Hungarian as Yamnaya_Samara, CWC_Baltic_Early or Yamnaya_Ukraine (looks clearly there was a meeting of Iberian/Atlantic and Danubian/Hungarian ancesties in North-Central Europe?). If CWC_Baltic_Early were an actual better proxy for the steppe people entering into Europe, then GAC may not be the best fit for Bell Beaker ancestry.

This also looks apparent to me passing the same data through a Linear Discriminant Analysis: https://imgur.com/a/MCZU9

(Above is all almost purely visual cluster and vector analysis, so will need some actual model confirmation!).

*my "subjective" non-European set (which means more "A different population history than most Europeans" rather than trying to define a geographical border) essentially removed k-means clusters for Caucasus-Iran, Levant-Middle East, South Central Asian, Volga-Ural with East Asian related ancestry, although I did make an exception on for Yamnaya_Outlier, which I left in despite fitting with Caucasus-Iran samples.

Matt said...

@Chad, in the other thread you said: "What's really tripping me out is that English Beakers look to average higher Yamnaya that CE Beakers. I'll definitely look that over and scan the paper for any stats regarding that."

Yes, for Britain / NW Europe and the Bronze Age in general as well, particularly Wales_EBA, but also Beaker_The_Netherlands.

This is not actually too surprising though is it, since this is exactly what the Rathlin_EBA folk were like (more Steppe rich the Bell_Beaker_CE)?

This could have arisen from a few processes - either:

a) Bell_Beaker_CE group is more of a "core" Beaker group than the Bell_Beaker_Netherlands and the Bell_Beaker_Netherlands / Britain / Ireland arise from a low level absorption of Corded_Ware_Germany type ancestry. (In this model Bell_Beaker_CE is, I guess, a mix Yamnaya_Bulgaria with Balkan/Hungarian farmers, and then ancestry from Northern EEF is required to get them to the NW European Beakers, which I may have been related to part of Jean Manco's model for the formation of the Bell Beaker Culture?).

b) Bell_Beaker_Netherlands / Britain / Ireland are more of a "core" Beaker group than the CE, and CE arise from them picking up ancestry from various groups present in CE (e.g. possibly Vucedol or Balkans ancestry is filtering into Bell_Beaker_CE).

c) Bell_Beaker_CE also seems more diverse in ancestry than the northern groups, so possibly there is some cryptic population structure in Bell Beaker populations where the "genetically northmost" CE beakers migrated to NW Europe.

This will take some archaeological detail to sort out. Certainly beyond me.

Alberto said...

@Eren

I don't personally care, but maybe it's easier for people opening the files if they open by default in a text editor instead of a spreadsheet program? Maybe someone else can say if there's a preference, especially someone using Windows.

Anonymous said...

Slumberry: some of the 'blond' hair in Melanesians and Polynesians is in practice ginger or auburn. However, this is entirely unrelated to European red hair, genetically speaking.

Slumbery said...

acuouswastrel: Thanks.
OK, so more precisely: I do not know any population that have gingers among them and do not have European HG ancestry, except Melanesian, who - by my best knowledge - have different alleles causing this, so cannot possibly be related. :)
I knew about Melanesian (it is actually interesting, especially the possibility of the Denisovan origin of the light hair colors), just didn't consider them in this context.

Also, just to avoid any misunderstanding that could come from the fact that I reacted partly to Samuel post: I think the presence of gingers on both population is not specific enough to assume a specific connection between BB and Tocharians.

Rob said...

@ postneo
I revise what I said- there is suggestion of movement of a Tepecik type population to SEE, peaking in Bulgaria and decreasing toward Greece and Crete. ; but also probably toward Caucasus

Eren said...

@Alberto:

Yeah, it's just a minor detail. As a Windows user I prefer txt, as it's quicker to open with a text editor. And since David uses txt format, I'll go with txt. It's just a difference in the file ending anyway.

Btw. I've seen that you've shared your Xmix and ancestorsMix scripts, cool stuff. I haven't yet understood in detail how they work, but will take a closer look.


@David:

I've implemented the rounding to 6 decimals for the scaling script: https://drive.google.com/open?id=1M-MLj2TPyGvSmSXxRxfS0vMt_JcjAWLd

Btw. have you tried projecting the Basal-rich cluster to the Global 25 yet?

Matt said...

Had a go at trying to use a Europe subset of ancients and moderns on Global_25 to create a visualisation and try to understand some patterns.

I must admit that the diversity of these clusters makes it very challenging to try and visualise them, but in case anyone is interested, here are a few graphics: https://imgur.com/a/RqYqx. These are reprocessing of the Global25 dimensions through PAST3 PCA again.

Same kind of patterns as in the West Eurasia datasheet, where the modern Balkans is outside the general Bronze Age clines, but where drift distinguishes Baltic_BA related populations specifically, several of the Balkans outliers, etc. do overlap the present day Balkans.

Retreating a bit to a simpler case, I tried restricting the Global_25 data to a subset of Early Neolithic and European HG groups from across Europe, and then reprocessed in a PCA. See here: https://imgur.com/a/5OVae

When reprocessed here, Dimension 1 distinguishes HG from Neolithic, then Dimension 2 splits different HG from each other, and so also splits apart Neolithic populations with different HG admixture, e.g. Malak Preslavets is between Romania_HG and Barcin, Iberia_Chal between Iberia_EN and WHG, etc.

Dimension3 then looks more specific to Danubian/Hungarian farmers vs Atlantic farmers. Weirdly, Globular Amphora sits mostly with the Atlantic farmers here. GAC doesn't seem any more Hungarian/Danubian than Sweden_MN.

Adding a couple of simulations, both of Iberia_EN+LatviaHG (e.g. to approximate Cardial farmer plus local Polish HG) and Baden_LCA+Rochedane (to contrast to Hungarian Copper Age+WHG): https://imgur.com/a/J3RCE.

In the dimensions 1v2 they are about as good at approximating GAC, but looks like Iberia_EN+LatviaHG may be slightly better in Dimension 3. So GAC more like Sweden_MN (slighly more like Cardial route farmers+NE European HG than Danubian route farmers?). What do people find with nMonte?

Rob said...

@ Matt

Thanks for the visuals. If I understand them right, GAC and Sweden MN are themselves shifted toward Balkans compared to the west Euro/ Scando contemporaries and predecessors.

This is reflected in the run i got:

Globular_Amphora
Sweden_MN 29.1 % + Germany_MN 22.8 %
Tiszapolgar_ECA 32.35 % + Varna_outlier 8 %
Villabruna 6.15 %

IMO there's a definite north Balkan component.

Arza said...

Both, Steppe_EMBA and Yamnaya_Ukraine_outlier, can be modelled as EHG, CHG and Trypillia.

right pops:
Mbuti.DG Ust_Ishim Kostenki14 MA1 WHG Iran_N Natufian Levant_N Armenia_EBA Anatolia_N

left pops:
Steppe_EMBA EHG CHG Trypillia

best coefficients: 0.463 0.437 0.100
std. errors: 0.019 0.024 0.021

chisq: 4.108 tail: 0.767271114
numsnps used: 219103


left pops:
Yamnaya_Ukraine_outlier CHG Trypillia EHG

best coefficients: 0.590 0.212 0.198
std. errors: 0.053 0.048 0.041

chisq: 3.244 tail: 0.86151569
numsnps used: 194836

Arza said...

Trypillia seems to be a mix of Ukraine_Neolithic_outlier and... Ukraine_Eneolithic:I5884

right pops:
Mbuti.DG Ust_Ishim Kostenki14 MA1 WHG Iran_N Natufian Levant_N CHG Anatolia_N Steppe_Eneolithic Armenia_EBA Steppe_EMBA EHG

left pops:
Trypillia Ukraine_Eneolithic_I5884 Ukraine_Neolithic_outlier

best coefficients: 0.218 0.782
std. errors: 0.053 0.053

chisq: 4.074 tail: 0.982061367
numsnps used: 76587

Matt said...

@Rob, it looks to me like in some subtle way* GAC and Sweden_MN are shifted the SE Neolithic relative to at least the Atlantic Neolithic (though with more HG). By the same token they look shifted towards the Atlantic Neolithic relative to what Germany_MN or Czech_MN are like (though there are fewer samples of each of these, Sweden_EN/Sweden_MN/Germany_MN/Czech_MN, compared to the bigger Neolithic sampling for Britain&Ireland, France, Balkans, Hungary, Iberia).
Still experimenting with this stuff.

Btw, I had a go at adding a subsample of a few Beaker "populations" to a subsamples of the different Neolithic and HG samples: https://imgur.com/a/biPIM. The "Central European Beaker" sample is highly diverse in the dimension that emerges and seems to distinguish the two Neolithics, while the other "populations" are much less so.

*that is, to the degree Global_25's dimensions contain specific variation between two different Neolithic waves, rather than just measuring the same Neolithic wave interacting with slightly different HGs!

Arza said...

So Steppe_EMBA can be modelled as:

right pops:
Mbuti.DG Ust_Ishim Kostenki14 MA1 Iran_N Natufian Levant_N Armenia_EBA Anatolia_N WHG

left pops:
Steppe_EMBA EHG CHG Ukraine_Eneolithic_I5884 Ukraine_Neolithic_outlier

best coefficients: 0.524 0.359 0.006 0.111
std. errors: 0.067 0.038 0.079 0.037

chisq: 4.638 tail: 0.590950915
numsnps used: 107640

But the I5884 fraction is tiny-to-nonexistent.

Rob said...

@ Arza
Tripolje as Ukraine Eneolithic combo isn’t a historically viable model

Rob said...

@ Matt

As i see it, GAC and MN Sweden are right shifted compared to EN Sweden, Iberia, British Neolithics, etc. Germany MN is there with Polgar etc, & therefore might be the intermediary for the other 2.
Nicely depicted in your plot, i think

@ Roy
Thanks for clarifying about the J2a's.

Henry said...

@davidski is there tocharians in this pca and if not where would they fit for you ?

Arza said...

@ Rob

I'm intentionally ignorant here. Just like in the case of modelling of Slavs as part Bell_Beaker_Czech:RISE568 (BTW that ghost that I've made in times when no one heard about Baltic_BA sits inside of the cluster).

Everyone here takes care to use only the right samples, from right cultures, times and of good quality. So I chose to do the opposite. The idea is to check things that no one else here will ever check. Additionally I treat this rather as a sort of geometric puzzles or n-dimensional Rubik's cube (I'm very rarely watching plots with less than 3 dimensions) so I'm rather interested in finding geometrically working solutions. Confronting them with archaeology and history is another separate step for me.

And actually this model wasn't my idea. qpAdm told me to do so when I was trying to do the opposite.

left pops:
Ukraine_Eneolithic_I5884
Ukraine_Neolithic
Trypillia
Ukraine_Neolithic_outlier
Lepenski_Vir

chisq: 4.498 tail: 0.922094757

Awesome result, isn't it?

best coefficients: 0.125 4.070 -3.181 -0.014

Yep, 407% of Trypillia and -318.1% of UNo.

Imagine that this is ADMIXTURE:

[OXXX] - Trypillia
[XXXX] - UNo

Algorithm exaggerated Trypillia and UNo to form this:

[OOOOXXXXXXXXXXXX]
[XXXXXXXXXXXX]

And then subtracted one form another - [OOOO]. qpAdm, using UNo, extracted something from Trypillia, that is present in I5884. In other words there is a genetic cline here, a mixture of 2 pops and UNo, Trypillia and I5884 are sitting on this cline, no matter how young I5884 is.

PS Who in his right mind would try to model Steppe_EMBA as part Minoan?

Steppe_EMBA EHG CHG Ukraine_Neolithic Minoan_Lasithi
best coefficients: 0.486 0.411 0.010 0.093
std. errors: 0.047 0.030 0.043 0.024
numsnps used: 276809
chisq: 6.226 tail: 0.717115642

Not as good as Trypillia, but it works too.

Arza said...

@ Rob
Of course you are right, but this result may be showing something we don't yet know.

Alberto said...

What I find interesting is that Eneolithic Ukraine had so much EEF admixture even in Eastern Ukraine as 4000 BCE (with EHG and CHG being reduced with time and moving west, instead getting more Ukraine_N). And then there's again a big shift towards EHG/CHG when Yamnaya arrives.

Matt said...

Tried a few more things to help with identifying outliers from general clusters using Global25 data.

1: Putting population averages for ancients and moderns on a Neighbour Joining Tree with Beaker and Balkans samples: https://imgur.com/a/GPQjC

(Note that the overall set of samples is now so large that PAST3 can't even plot all the ancients together on a single neighbour joining tree!)

Anything here which neighbour joins with an atypical population or has a long branch from neighbours may be an outlier.

Examples of Beaker_CE that neighbour joing far from others in their population are:

I4253, I4889, E09569 (closer on the tree to West Asia than others)
I2787 (steppe MLBA like Beaker)
I5025 (Slav_Bohemia like Beaker)
I2741, I1392, I7283 (Neolithic like)
E09538, I5524 (modern Iberian like)
I7044, I7045, I2364, I3594 (ancient Balkans like)

I3528, I7286 and I5017 look like examples of Beaker_CE that have a long branch and are distant from all others.

There's also what looks like a subgroup of Beaker_CE that are all close together but seem more French Austrian-like than most samples (which sit closer to NW Europe). These are: I4946, I5834, I3529, I7275, I4945, I6582, I3588, I7214, I7289, I4886, I6591, I4896, I5529.

With Balkans_BA -
Bul6, I2175, I2176 and I2520
and then I4331, I4332, I4331, I3313
look to form two decently close clusters while the other two are quite different from each other. I2163 is the steppe MLBA like Balkans BA and I2509 is the EEF like Balkans_BA.

2: Using a restricted set of samples (European, ancients only, avoiding full HGs) are reprocessed through PCA: https://imgur.com/a/ha6cX

This identifies a good amount of variation between samples and reconfirms the samples which are outlying within Bell Beaker CE. This also shows the following other outlier: CWC_Czech - I7272 (EEF like)

While there is scarcely any unity within Hungary_BA, I7042 is also a especially an EEF (SE European) like outlier.

For Hungary_BA, the dimension that separates out Baltic_BA from others also shows that I1520 and I1504 are fairly present day Eastern European like, while I7043 and I5665 look more present day Central European like.

Among Beaker_CE, dimensions that break out Baltic_BA also show that I7044, I5025, I3528 are especially Baltic_BA like Beakers, and this is also true to a lesser extent of I5524, I4891, I7286, I5531.

Among Beaker_Iberia, the outliers are of course I6623, I6539, I1388.

I'm sure these have all been identified by the papers, but will have to check again; maybe the Global25 has allowed some subtle shift to be found which Olalde or Mathieson's papers missed!

Chad said...

Rob,

Still working here. I'm under a deadline for a final, so it may be a day or two. As it looks, there isn't any reason for Hungarian admixture from Lengyel or Baden. Tiszapolgar was contemporary for an early part of GAC, but from SE and never mixed, as it looks, so that is out. The ones that are suspected, will not go into GAC.

Chad said...

I'll keep trying, but it is looking like the Beaker situation.

Matt said...

Looking closely at those outliers I've mentioned above (and changing some ids I got wrong), archaeological information from supplements:

Examples of Beaker_CE that neighbour join far from others in their population are:
I4253, I4889, E09569 (closer on the tree to West Asia than others)


I4253/RISE1124/grave no. 13: 2457–2208 calBCE - Male inhumation burial (25-30 years), with N-S orientation, located on the left side. The only element of equipment was a ceramic bowl, posed in the northern part of the grave. Samborzec (Małopolska, Poland). R1b1a1a2

I4889/Grave 11, feature 544: 2281-2062 calBCE - Left sided crouched burial, head toward the north-east. Sex: orientation – M, anthropology – M, DNA – M. Age: mature individual. Grave goods: three arrowheads, stone wrist-guard, antler flint-knapping tool, bow-covering made of animal bones. Prague-Kobylisy (Czech Republic). R1b1a1a2a1a2b1.

E09569/Grave I/3: 2397–2149 calBCE - male individual in contracted position with an arrowhead and several pieces of flint as grave goods. Unterer Talweg 85 (Augsburg, Bavaria, Germany). R1b1a1a2a1a2b1.

I2787 (steppe MLBA like Beaker)

I2787/GEN_59/Grave 688: 2458–2202 calBCE - Male individual lying of his left side, in contracted position. The rectangular shaped grave pit, oriented northeast–southwest, was enclosed by a round ditch. Grave good include a small jar. Szigetszentmiklós, Felső Ürge-hegyi dűlő (Hungary). R1b1a1a2a2.

I5025 (Slav_Bohemia like Beaker)

Female, Kněževes, Czech Republic, 2500–1900 BCE. Also known as RISE567, F0523, A0766, gr. 8.

I2741, I1392, I7283 (Neolithic like)

I2741/GEN_20, Grave 49: 2458–2154 calBCE. Male individual lying of his left side, in contracted position. The rectangular shaped grave pit, oriented northeast–southwest, was enclosed by a round ditch. Grave goods include a Bell Beaker, a bowl, a stone wrist-guard and a dagger. Szigetszentmiklós, Felső Ürge-hegyi dűlő (Hungary). I2a1a1

I1392/13-Grave9: 2832–2476 calBCE. adult mature individual who is genetically female. The spatial orientation and the grave goods are consistent with a female Bell Beaker burial. Hégenheim (Haut-Rhin, France). The decoration can be attributed to a mixed maritime style, considered to be an early stage of the Bell Beaker tradition.

I7283 - I7283/Grave 117/78: 2500–2200 BCE. Right-sided crouched burial, head towards the north-east. Sex: orientation – atypical (right side = female, orientation and grave goods = male), anthropology – F, DNA – F. Age: adult (over 50 years). Grave goods: two vessels (beaker, ornamented cup), gold and silver jewelry, copper dagger, copper awl, flint industry. First-degree relative of I7282. Radovesice (Czech Republic).

E09538, I5524 (modern Iberian like)

E09538/Feature 68 Skeleton 1: 2471–2300 calBCE. Close to the back of the male individual, a bowl was placed in the burial pit and a wristguard was placed on one of his lower arms. Seen from a stylistic and relative chronological perspective, the beaker and the wrist-guard point to a rather early time within the Beaker Complex. Unterer Talweg 58-62 (Augsburg, Bavaria, Germany). G2a2a1a2a1a.

I5524/F0224/grave 5: 2500–2000 BCE. Grave 5 was of an adult female, lying in right hocker position with the head to the SE. At her feet was a single undecorated handled vessel. Mother or sister of I5526 (female). Landau an der Isar (Ldkr Dingolfing-Landau, Bavaria, Germany). Migrant to region (Sr isotopes).

Matt said...

cont.

I7044, I7045, I2364, I3594 (ancient Balkans like)

Grave 13 (I7044/HUNG498): Inhumation grave in contracted position. The trace of the grave pit was not observable. The S-N oriented male (50-70 years old) individual laid on its right side261. The arms were located in front of the ribs, the legs were bent in knee. The grave goods were next to the pelvis, behind the skeleton. Finds/Bell Beaker period: handled jug with biconical body and a slightly outward-curving rim, handled bowl with short curved neck and with 4 little symmetrical handles. Szigetszentmiklós-Üdülősor (Hungary). R1b1a1a2a2c1

Grave 1 (I7045/HUNG499): A pit with a female (40-50 years old) individual in contracted position that was revealed within the settlement. The filling of this pit was different from the other features of the settlement261; it was yellowish clay from the subsoil. The pit was oval, with curved bottom. The SE-NW oriented skeleton was laid on its right side, close to the SW wall of the pit. The right hand was under the face and the left arm in front of the face. Finds of the pit/Bell Beaker period: handled jug with biconical body and a slightly curving rim situated next to the grave, handled bowl with T-shaped rim located in the northern part of the grave, fragments of a badly preserved pot in front of the skeleton (which was disintegrated during excavation), fragments of grindstones, antler tools in front of the knee and copper wire found close to the cranium. Szigetszentmiklós-Üdülősor (Hungary).

Grave 219/B (I2364, GEN 10a): Double burial excavated in 1966. Two individuals were lying on their right side in contracted position, without grave goods. Individual B is an adult male. 2470–2060 calBCE. Budapest-Békásmegyer, Királyok útja (former Vöröshadsereg útja) (Hungary). H2.

I3594/Grave 9: 2300–2150 BCE. Half-way supine with flexed legs, half-way slightly right-hand side crouched burial, south-north oriented; inventory consists of broad metope-decorated beaker, 10 V-formed perforated bone buttons and 3 flints; anthropologically adult woman. Genetic data show that she is a first-degree relative of I3597/Grave 12 (infant child). Alburg (Lerchenhaid-Spedition Häring, City of Straubing, Bavaria, Germany).

I3528, I7286 and I5017 look like examples of Beaker_CE that have a long branch and are distant from all others.

Grave 276 (I3528, GEN 85): Skeletal male burial with north-south orientation, located on the left side, legs bent at the knees. At the southern part of the grave vessels and two stone artefacts were found. Budakalász, Csajerszke (M0 Site 12) (Hungary). G2a2a1a2a1

I7286/Grave 53/80-I: 2500–2200 BCE. Left-sided crouched burial, head towards the north. Sex: orientation – M, anthropology – M, DNA – M. Age: adult (20–30 years). Grave goods: three vessels (beaker, cup, bowl), stone wristguard. Radovesice (Czech Republic). R1b1a1a2a1a2b1.

I5017/RISE559/F0174/grave 4: 2461–2207 calBCE (3855±35 BP, Poz-84458). Migrant to area. Augsburg Sportgelände (Augsburg, Bavaria, Germany). Female.

I2163 is the steppe MLBA like Balkans BA and I2509 is the EEF like Balkans_BA.

I2163 / Merich 2 (Individual 5) Adult male, found on the periphery of the second heap, buried in a shallow pit. Positioned with the head to North, legs bent at the knees and holding a small cup in its right arm. Bulgaria. 1750-1625 calBCE. R1a1a1b2

I2509 / G2. Adult female. It is located in trench 1, at the bottom of a contemporaneous ditch. The skeleton is not in an anatomical position, its bones scattered over an area of approximately 4 m2. Bulgaria. 4452-4354 calBCE.

Matt said...

cont.

CWC_Czech - I7272 (EEF like)
I7272/Grave 23: 2900–2500 BCE. Right-sided crouched burial, head towards the west. Sex: orientation – M, anthropology – ?, DNA – M. Age: infant (4 - 6 years). Grave goods: vessel (small beaker), pelvis of cattle. Brandýsek (Czech Republic). I2a2a2.

While there is scarcely any unity within Hungary_BA, I7042 is also a especially an EEF (SE European) like outlier.

Grave 244 (I7042/HUNG496): Round pit with cylindrical wall and straight bottom. On the bottom of the pit a child (10-12 years old; male according to genetic sex) individual in unusual, unfolded position was revealed, without any grave goods. The cranium was located on its left side, close to the southern wall of the pit. Orientation: SSW-NNE. The lower limbs were on the right side of the body. This skeleton is a typical “thrown in” body and could be presumably dated to the early phase of the Nagyrév culture. I2a1

For Hungary_BA, the dimension that separates out Baltic_BA from others also shows that I1520 (edit: I1502) and I1504 are fairly present day Eastern European like, while I7043 and I5665 (edit: I7041) look more present day Central European like.

I1502 - HUNG370, BR1. Female. 2190-1980 calBCE.
I1504 - HUNG381, BR2. Male. J2a1. 1270-1110 calBCE

I7043 - HUNG497, grave 119. Male. Grave 119 (I7043/HUNG497): Round pit with cylindrical wall (with a slightly concave part on the eastern side) and with curved bottom. On the bottom of the pit (in the centre of it) a ESE-NNW oriented child (12-14 years old; male according to genetic sex) individual in a left sided contracted position was located. The grave was partly disturbed, without grave goods. There was only one piece of EBA sherd found on the ribs. The left arm situated under the chest, but the right arm is missing. The mandible was not in its right anatomical position. R1b1a1a2a1a(xR1b1a1a2a1a1,xR1b1a1a2a1a2). 2500–2200 BCE.

I7041 - HUNG495t, grave12. Grave 12 (I7041/HUNG495t): Round pit (size: 128×130 cm, depth: -79 cm) with light brown homogeneous filling and hard clay wall. On the straight bottom of the pit a N-S oriented male (20-25 years old) individual was unearthed, laid on its left side but in abnormal prone position. The left arm was under the body, whilst the right one was close to the upper part of the body. Both arms were bent back (the right hand was just under the scapula); both hands were in front of the cranium; both legs were bent in knee. Finds/proto-Nagyrév period: a little handled, profiled cup, upside down in depth - 63 cm, next to the eastern wall of the pit. Some sherds appeared above the skeleton and could be remains of intentionally broken vessels260. This individual is a brother of I7043. Male. R1b1a1a2. 2500–2200 BCE.

Matt said...

cont.

Among Beaker_CE, dimensions that break out Baltic_BA also show that I7044, I5025, I3528 are especially Baltic_BA like Beakers, and this is also true to a lesser extent of I5524, I4891, I7286, I5531.

I7044 / I5025 / I3528 / I5524 / I7286- See above.

I4891/Grave 13, feature 546: 2281-2062 calBCE (3765±20 BP, PSUAMS-2848). Extended position with crouched hands on the shoulders, head toward north-east. Sex: orientation – M, anthropology – M, DNA – M. Age: adult. Grave goods: one vessel – big bowl, two arrowheads. Prague-Kobylisy, Czech Republic. R1b1a1a2a1a2b1.

I5531/F0180, grave D: 2500-2000 BCE - Weichering, Bavaria, Germany). Female.

Among Beaker_Iberia, the outliers are of course I6623, I6539, edit: I0461.

I6623 aka RISE696. A 1–5-year-old child (I6623) was inhumed at the far end of the cave and was covered by the body of a 20– 30-year-old woman (I4245), carefully placed in supine position with the head to the left and flexed legs. In terms of the grave goods, a small decorated cup was found on the child, whereas two bigger decorated inlayed cups had been placed. between the breast and left arm of the woman. The archaeological context strongly suggests that both individuals were interred at the same time.

I6539/Hume 5, UE 455.2: 2500–2000 BCE. Adult male. R1b1a1a2a1a2(xR1b1a1a2a1a2c). I0461/Roy4/SU19, Inhumation 1: 2348–2200 calBCE. Female

Tesmos said...

@Matt,

Could you also check Czech_EBA I7196?

Matt said...

Hi Tesmos,

For I7196, looking at position, he looks like a fairly typical European Bronze Age sample, fairly similar to England_Roman average on overall Steppe-WHG-Anatolian position and close to European Bronze Age samples in position on the Baltic_BA dimension (e.g. sitting with NW Europeans).

Details from supplements:

I7196/Grave 59: 2200–1700 BCE. Right-sided crouched burial, head towards the south. Sex: orientation – ?, anthropology – ?, DNA – M. Age: adult (35–50 years). Grave goods: bronze hair ring, bone awl, flint industry. R1b1a1a2a1a1c1a.

I7197 though is definitely an EEF like outlier.

I7197/Grave 77: 2200–1700 BCE. Right-sided crouched burial, head towards the south. Sex: orientation – ?, anthropology – M ?, DNA – M. Age: adult (20–30 years). Grave goods: three vessels (bowl, two cups). I2a1.

Both of the Prague-Jinonice (“Zahradnictví”, Prague 5 – Jinonice, Czech Republic) site.

Also, slight correction to above, I got the I2509 is the EEF like Balkans_BA wrong and it's actually I2510: Sub-adult male. Excavated from trench 8, exactly underneath the top level (0.30 m), it is disrupted by an Early Bronze age pit. Only the head, several bones of both hands and the upper ribs are preserved. Bulgaria. G2a2a1a2.

Cpk said...

In light of all these findings, what do you guys think about the Arctic home of Aryans theory?

capra internetensis said...

thanks for all that, Matt

does anyone know how I4245 looks in the autosome? she is the one with the African mt hg (L1b1a).

Tesmos said...

Thanks Matt!

Davidski said...

@Cpk

In light of all these findings, what do you guys think about the Arctic home of Aryans theory?

Not much.

Rob said...

@ Matt

Thanks. So are your graphs nMonte dimension results fed back into a PCA ?

@ All

Looking at Minoans and Myceneans with EMBA ancients from SEE and surrounds (BA Anatolia, MBA steppe, the R1b Hungarian BA male, etc), without any pure HGs, and nothing too distant (eg Iran Chalc).

https://imgur.com/a/mODTy

The major difference between Minoans and Myceneans is Minoans have major continuity with local Neolithic, whilst myceneans are more “cosmopolitan”, with a major impact from 3rd millenium Anatolia, but also small & patchy impact of “MBA steppe” in individual Myceneans (mostly by way of BB Central Euro)- which are the J2a1 male I9041 and the I9033 female . This is all as predictable.

Davidski said...

@Rob

The important difference between Minoans and Mycenaeans is that the Mycenaeans clearly have ancestry from the steppe, while Minoans don't.

Both harbor significant Caucasus-related ancestry, which was already streaming heavily into the Aegean during the Neolithic, around 4000 BC, because one of the Peloponnese Neolithic samples clusters with the Minoans.

Matt said...

@Rob, nah, these are just Global25 PCA fed back into PCA, not based on nMonte fits.

Davidski said...

@All

Alberto and me had a bit of chat with Iain Mathieson about sample Ukraine_Eneolithic I5884 (ie. the individual belonging to R1b-Z2103).

There are no indications that the dating for this sample is wrong. However, despite the late dating, and the presence of bronze technology across the steppes at this time, there are apparently solid archaeological grounds to classify the relevant burial site as Eneolihic, as opposed to Early Bronze Age.

So we need more and earlier samples from that region to figure out what was happening there in terms of migrations and direction of gene flow.

a said...

Let's tally-R1b-Z2103 -Z2305
R-Z2103Z2103 formed 6100 ybp, TMRCA 5700 ybpinfo +/-


positive and probable- ratio

1]Yamanaya culture+
2]Afanasievo culture-highly probable
2]Poltavka culture+
3]Sredny Stog culture-highly probable
4]Vucedol culture+
5]Bell Beaker culture+ and highly probable{HUngary-Poland}
6]Sarmatian culture+

How about a PCA with R1a and R1bZ2103 samples parsing Steppe levels.


Rob said...

The steppe MBA -> Minoan model doesn’t pan out . It reduces a complex history into a fictionalised steppe fable . It shows a complete ignorance of historic processes- and they were; at the end of M3 the Yamnaya people were sent packing from SEE.
Everything matters- the degree of admixture per population and individual. So far the elite kurgan conquest theory has no support .
Do I think there’ll be R1a in shaft graves ? Hhm - maybe but they appear to have all died out.
Maybe like CHG male lineages died out on the steppe. ?

Davidski said...

The Mycenaeans have significant steppe-derived ancestry. Generally 15-20% depending on the steppe references used. And over 30% if we use Balkans_BA as a steppe-derived reference.

And we don't even have any high status Mycenaean male samples from royal burials yet.

This is all very important and completely in line with the academic consensus that Indo-European languages were introduced into the Aegean and Anatolia during the Bronze Age via elite dominance.

You should all prepare yourselves for more of the same and much more. Here's a preview of what's to come from Iron Age Greek burials, which includes a male with R1b (14:16 into the clip).

https://www.youtube.com/watch?v=HGKZKoH4yv0

Ric Hern said...

So did Proto-Bell-Beakers migrate via a Land Route to Portugal from the Balkans during the Late Chalcolithic/Early Bronze and from Portugal backtraded towards the East ?

Why did Steppe related migrants adopt Bell Beakers so easily it seems ? Did early Steppe migrants have good relations with Proto-Bell-Beakers in the Balkans and retained some knowledge of this after each went their own way to only meet again +-500 years later in Western/Central Europe ?

Or was this seemingly easy adoption due to a shared Beaker of beverage ?
Or was it all about the bling ?
Or maybe all of these above ?

Ric Hern said...

Somehow Proto-Bell-Beaker and Varna rings a bell but I could be mistaken....?

Rob said...

David

You're conflating Balkan Bronze Age with 'Yamnaya', and that's disingenious of you because they're different phenomena. Moreover, let’s not dance around the bush that some of your estimates are laughable exaggerations, because your datasheet of results had Acermaic period Boncuklu scoring 10% "Yamnaya". ; and a Sicilian woman with no steppe scoring 25%

Your theories imagine evidence where there is none while ignorant the elephant in the room - large migrations from West Asia. A miserly 10-15% steppe influence in only half of Myceneans isn't going effect an indo-Europeanization.
I’m wary of appeals “elite dominance” without backing data, otherwise it seems like special pleasing
On the contrary, my run showing local Greek, then Tizapolgar, B.B., then West Asian is scarily Perfect . You’d know that if you had some semblance of a clue..

And no, I wouldn't be surprised about R1b in Iron Age Greece, who would ?

Davidski said...

@Rob

Here's what I meant...

[1] "distance%=2.7811 / distance=0.027811"

Mycenaean

"Peloponnese_N_outlier" 60
"Balkans_BA" 29
"Tepecik_Ciftlik_N" 11
"Peloponnese_N" 0

OK, not over 30% in this one. No big deal though. Close enough. And it makes perfect sense: gene flow from Anatolia to Greece from the Neolithic to the Bronze Age, then, bam, the pre-Greeks arrive from the north!

Let's see how the Minoans do in the same model, to help strengthen this conclusion.

[1] "distance%=1.6556 / distance=0.016556"

Minoan_Lasithi

"Peloponnese_N_outlier" 68.4
"Peloponnese_N" 31.6
"Balkans_BA" 0
"Tepecik_Ciftlik_N" 0

Conclusion strengthened. Thanks for watching folks.

Davidski said...

@eren

I'm not having much luck in creating Basal-rich synthetics that actually work in the Global 25.

I may have actually forgotten how to do it properly. Need to consult my notes and try again.

Rob said...

@ Dave

Both Myceneans and BA Balkans are heterogeneous in origin, even if the former plot closely on a PCA, so your averaged ghost "BA Balkan' population isn't real
But let's go with it:

Balkans_BA
Balkans_N 15.5 %/ LBK_EN 49.4 %
Ukraine_N 16.25 %
CHG 8.9 % / Iran_N 2.65 %
EHG 7.3 %

What's 7% of 29% ? Something irrelevant

Davidski said...

Well Rob, all populations are heterogeneous, if you go far enough, and this is especially true of post-Neolithic populations in Europe.

The point I was making with my model was that there was a relatively sudden burst of gene flow into Greece form the north that included steppe admixture.

This happened after 4,000 BC, which is the time frame that the proto-Greeks are supposed to arrive in Greece based on archeology and linguistics, according to most archaeologists and historical linguists. And no ever claimed that these people were pure descendants of foragers from Eastern Europe.

So I can't see anything overly complicated here. The data is essentially speaking for itself.

Rob said...

@ Dave

Thanks for confirming what I've been suggesting this whole time. "This happened after 4,000 BC" of a mostly Balkan Chc-BA peoples from the north with a bit of steppe.

Further, your data confirms what you've been attempting to refute - the mostly local Neolithic origin of Minoans, outlier or not, Peloponessus Neol are essentially Minoans.
By contrast, Myceneans need the extra west Asian (here depicted in your Tepecik) in addition to northern impact.
Your error seems to have stemmed from your discomfort in west Asian intrusion, which you attempt to pass off as Minoan, but in actual fact, is more important in Thrace, Mainland Greece and Italy than Crete. Anyhow, lets wait for more data

Rob said...

Btw your models are problematic because you omit EBA Anatolia and use archaic Tepecik

Davidski said...

@Rob

Feel free to use Anatolia_BA in the models I ran. It makes no difference, because this is what I said...

And it makes perfect sense: gene flow from Anatolia to Greece from the Neolithic to the Bronze Age, then, bam, the pre-Greeks arrive from the north!

Matt said...

Quick nMonte fits for the Globular Amphora Culture Average in Global_25 using all preceding and contemporary samples

------------

Globular_Amphora:Average

France_MLN,16.8
Sweden_EN,15.8
Iberia_EN,14.2
Tisza_LN,11.4
Varna_outlier,10.8
Iberia_Central_CA,10.4
Iberia_Southwest_CA,7.4
Koros_HG,4
Lepenski_Vir,3.8
Blatterhole_HG,3
Ukraine_N_outlier,1.4
Czech_MN,0.8
South_Africa_2000BP,0.2

Removing the Varna Outlier and Lepenski Vir

Globular_Amphora:Average

Tisza_LN,16.6
Sweden_EN,14.8
France_MLN,13.6
Iberia_EN,13.4
Iberia_Central_CA,11
Romania_HG,5.4
Sweden_MN,5
Ukraine_N_outlier,4.4
Czech_MN,3.6
Iberia_Southwest_CA,3.2
Iron_Gates_HG,3.2
CWC_Baltic_early,2.6
Blatterhole_HG,2.4
Koros_HG,0.4
Scotland_N,0.2
South_Africa_2000BP,0.2

Restricting the inputs to Early Neolithic population averages (Iberia, Greece, LBK, Koros/Starcevo, Balkans, Anatolia, Ukraine, Krepost), Euro HG samples, ancient CHG, Yamnaya, Levant, Iran and world ancients.

Globular_Amphora:Average

Iberia_EN,53.6
Ukraine_N_outlier,18
Iron_Gates_HG,8.8
Blatterhole_HG,8.2
Ukraine_Eneolithic,5.2
LBK_EN,3.2
Ukraine_Mesolithic,2.8
Koros_HG,0.2

Remove Blatterhole_HG

Globular_Amphora:Average

Iberia_EN,56.8
Ukraine_N_outlier,19
Iron_Gates_HG,16.8
Ukraine_Eneolithic,3
Ukraine_Mesolithic,2.6
Afanasievo,0.8
LBK_EN,0.8
Koros_HG,0.2

-------------

Models allowing all European Early Farmers farmers sensibly like GAC as a mix of Atlantic 41.2% (France_MLN,13.6, Iberia_EN,13.4, Iberia_Central_CA,11, Iberia_Southwest_CA,3.2), Hungarian/Ukraine_N outlier 20.6% (Tisza 16.6%, Ukraine_N_outlier,4.4) and intermediate 23.4% (Sweden_EN,14.8, Sweden_MN,5, Czech_MN,3.6) with 15% balance of mainly East-Central European HG.

(Ukraine_N outlier is mostly like LBK_EN).

Simplest models allowing only Early Neolithic and combinations fit it as 56.8% Iberian_EN, 19% Ukraine_N outlier and remaining from HG between Romania-Ukraine.

-------------

Average Sweden_EN and Sweden_MN (Saxtorp5164 and Gok2) then model with the last set:

------------

Sweden_EN_MN:Average

Iberia_EN,41.6
Koros_HG,22.4
Koros_EN,15.8
Ukraine_N_outlier,13
Boncuklu_N,4.4
Afanasievo,2.4
Lapita_Tonga,0.4

--------

NJ tree clustering shows GAC is very akin to Sweden_EN and Sweden_MN: https://imgur.com/a/cfiui

---------

I was a little puzzled by why, if Beaker farmer ancestry is mostly from GAC like populations, then present day NW Europeans should show slight excess of similarity in haplotype chunks to early Iberians, rather than SE European Neolithics and the LBK.

But maybe this is just explained by the GAC simply fitting more into being like Sweden_MN than I thought they would be, plus the small 10% *real* British/French Neolithic ancestry in NW Europeans.

Chad said...

Iberian is not possible and French is probably impossible too. Nothing substantial at least. Easter TRB and local hunters is the way it's leaning.

Matt said...

Impossible on archaeological grounds then I'd guess is your view (hopefully not impossible on a "Farmers moving into Northern Europe never went West->East for reasons!" type mental wall thing?)

Rob said...

@ Dave

“And it makes perfect sense: gene flow from Anatolia to Greece from the Neolithic to the Bronze Age, then, bam, the pre-Greeks arrive from the north”

Lol bam! Out of your arse
Go work for Disney mate

Eren said...

@David:

No problem, I didn't wanna bother you too much. I don't know what method you tried, I was thinking about creating synthetic individuals from the allele frequencies of each cluster (the F file used by ADMIXTURE calculators).

Ric Hern said...

@ Matt

The Spike in Iron Gates when you removed Blatterhole tells me that Iberian EN are related to a very early migration of Balkan Farmers to Iberia.

Early farmers surely couldn't avoid the Iron Gates on their way to the West...

Koros does not seem to be the route that they traveled so a more Southern route looks to me more plausible..?

Ric Hern said...

So GAC seems to also avoid the Koros route. So maybe a later expansions from Bulgaria or maybe the Western Balkans, from a similar population that were ancestral to EN Iberians ?

Matt said...

@Ric, Iberia_EN does actually favour Koros_EN, Ukraine_N_outlier and Balkans_N as input over Barcin/Greece, along with WHG (Loschbour, Villabruna, no Iron Gates HG).

Fit for GAC without Iberia_EN, but otherwise equivalent to the last model is: Globular_Amphora:Average - LBK_EN,37.8, Ukraine_N_outlier,36, Villabruna,12.6, Koros_HG,10.2, Ukraine_Mesolithic,3.2, South_Africa_2000BP,0.2

Still, I see small differences in these dimensions that suggest that GAC has some similarities to the Western group, without El_Miron affinity.

See here: https://imgur.com/a/d8wL1. Compare position of the two simulations, with and without Iberia_EN as a source, across the dendrogram and the PCA dimensions for this restricted panel (original source data reprocessed is from Global25).

It is tricky though, these divergences are very slight and hard to call.

Matt said...

(Essentially, in graphics above, clines between SE Neolithic and Western European HG and SW Neolithic and East-Central European HG work about equally well for GAC, in dimensions dominated by HG contrasts, first WHG v EHG, then Iron_Gates HG v El_Miron. SE Neolithic and East Central HG doesn't really work. But then in a dimension that's more specific to Western farmer vs Eastern farmer, GAC fits better with Western farmers. Subgrouping GAC.).

Formally, we could possibly test this, but possibly not with qpAdm, since detecting this via outgroups seems not likely - rather use the simulations method from Lazaridis 2017 - create simulated doubles in ratios 1:100 for all pairs (SE Neolithic,SW Neolithic,Western European HG,East-Central European HG) then check which shares more drift with the real GAC via D(Outgroup,RealGac;SimA,SimB). Could be that PCA based on massive sample sets might be doing a better job of finding recent correlations in ancestry than formal stats models based on very few samples do though.

Or other suggestions would be y-analysis (does the y I2 found in GAC link them to the I2 resurgence in the West?) or analyses which split the GAC into the Ukrainian and Polish populations.

Matt said...

Repeat of the graphics/analysis in last post using individual samples and the data labels from Mathieson's supplement: https://imgur.com/a/6nJwP

It is very similar qualitatively. GAC_Poland and GAC_Ukraine look like they could have slightly different HG ancestry, but otherwise very similar.

Perhaps interestingly in clustering it looks for the two "German MN" like the early Funnelbeaker associated Baalberge sample sits closer to the Swedish Funnelbeaker and the broad Isles-France-Sweden-Iberia cluster, while the Esperstedt sample is closer to the Hungarian Late Copper Age remit.

Ric Hern said...

@ Matt

I was looking at your Koros Hunter Gatherer not Koros Early Neolithic....

Chad said...

Matt,

It is impossible on archaeoligical and genetic grounds. GAC is rooted in eastern TRB. TRB is not Iberian.

Not only is qpAdm and qpGraph certain on this, Chromopainter says so too. Minimal sharing between Ballynhatty and Gok2 is from both having some northern German TRB roots. There isn't squat for Iberian haplotypes east of the Rhine because TRB and GAC are not Iberian.

This shouldn't even be a discussion. This just shows the limits of working outside of AdmixTools and other methods like Chromopainter and Rarecoal. That's all that can overturn anything on qpGraph at the moment.

If there was something better than qpAdm/Graph, then Germany and Cambridge would be all over it.

Chad said...

Also, France_MN is a German and Iberian mix that is probably most of Ballynhatty's ancestry. Part of why she is more Loschbour to LaBrana than Iberia MN.

These Megalith people move to Iberia to during the Chalcolithic, creating the shift towards Loschbour there after the MN. Cardial ancestry isn't squat outside of Sardinia. Even Spain now is probably as much Danubian than Cardial. Italy is now more Danubian than Cardial too. It's E to W and not the other way.

Matt said...

@Chad, in Martiniano's supplements, Gok2 does indeed have its most enriched coancestry matrix with CB13 (Early Cardial) rather than LBK, NE6, N7, CO1 and whatnot. (Though this is perhaps biased by their preponderance of Iberian samples). CO1 doesn't and is most enriched with the Southeast Neolithic (NE1, NE7, Rev5 etc.). They also note that Gok2 is placed in the "Atlantic Neolithic" cluster and suggest that this may be "a Scandinavian farmer (Skoglund et al., 2014) (S18 Fig). This might suggest a link between Neolithic samples in Atlantic Europe which coincides with the presence of Megalithic cultures" (although possibly this is just some function of HG introgression). I have not seen any haplotype data for the GAC samples, nor the German Middle Neolithic samples yet, perhaps you have?

"Cambridge and Germany" seem more like they would make statements about their qpGraph such as "We caution, however, that the relationships between Neolithic European populations are likely very complex, including multiple admixture events, and that the available ancient DNA data are still sparse (especially for the Early Neolithic period). As a consequence, the admixture graph models presented here likely represent a simplification of the true history relating these populations" than simply dismissively claim things are impossible.

I am under no illusion that nMonte+PCA based models would get past peer review, hence why "Cambridge and Germany" do no use them (except for using similar models for simulated individuals for model confirmation).