The genomic history of the Iberian Peninsula over the past 8000 years – Olalde et al. 2019

This Olalde et al. 2019 was a long awaited paper of which we had some glimpses over the past few months, and now it’s finally here with the full details. In fact, too many details to address in this quick post, so I’ll comment on just a few of them here and will continue with follow ups in the near future. Let me just say for now that this is a monumental paper, bringing 271 new ancient samples spanning from the Mesolithic to the early modern era. An amazing effort that brings the Iberian Peninsula to the highest level when it comes to ancient DNA. I can just hope that other countries will follow soon. As a Spaniard myself, big thanks to all the people involved in this project.

The Bell Beaker Culture(s) – Two cultures, one name?

This is really old news and relates to the previous Olalde et al. 2018 paper, but I still think it needs to be emphasized once more: I wrote “cultures” (plural) above because the name Bell Beaker Culture is still applied to two completely different and unrelated phenomenons, which keeps causing a lot of confusion. So very briefly:

The first one is a late Chalcolithic culture that developed as a continuation of the previous Iberian Chalcolithic ones (VNSP and Los Millares being the most important ones), without any significant change. This culture continues to be characterised by the collective burials seen for very long. There are burials that suggest some sort of warrior ethos, with lithic arrowheads, and of course there’s the bell-shaped pottery that gave it its name. (The “warrior” ethos mentioned is something more related to the zeitgeist than something that defines any specific culture. Just like the rise of patriarchy, the stratification of society, etc… Let’s not forget Gimbutas’ non-existent 1st and 2nd waves of invading steppe warriors that devastated “Old Europe”, the first one reaching as far as Ireland c. 3500 BC and the second one being a more sophisticated one referred to as the “Majkop people”, who formed the Globular Amphora culture, Baalberge, etc…).

The second one seems to have an origin in Central Europe, and its bearers were recently arrived migrants from the Western part of the Eurasian steppe. These people were genetically distinct from Neolithic Western Europeans, and brought with them a dominant male lineages in Western Europe since their arrival (R1b-L51). The culture itself can be distinguished by the single male burials (following the “steppe” kurgan tradition), and there inventory has significant differences, the main one in my opinion being the weapons made of arsenical copper/pseudo-bronze/arsenical bronze (tanged daggers, Palmela points,…).

So when in an Iberian archaeological site you see something like this:

Burial pit at Fuente Olmedo (Valladolid, Spain), c. 2000 BC. From López-Sáez et al. 2015.

You know that we’re talking about the second type of Bell Beaker culture (the R1b-L51 guys with steppe admixture). And you won’t find those things before ~2400 BC (while you can find the first type of Bell Beaker Culture with the eponymous pottery from c. 2700-2600 in Western Iberia). Two different cultures, different people, different origin. More importantly, this is the biggest discontinuity in the Iberian population history since the advent of the Neolithic (which was a big rupture itself), and the last one, since after it the basic genetic structure of Iberians was already met (with smaller impact events affecting it later). So with what we know today, to call the preceding and the succeeding people and cultures by the same name is clearly a bad idea and needs to be revised by historians quite urgently.

As for the samples in this paper, we got the earliest R1b-P312 sample so far in Iberia:

  • EHU002, R1b1a1a2a1a2, 2562–2306 cal BCE (El Hundido, Monasterio de Rodilla, Burgos).

This looks pretty much contemporary with the earliest R1b-P312 from Central Europe (or anywhere else), which shows how fast these guys spread throughout Western Europe once they first appeared (even if in Iberia it took a while for them to take over and was more gradual than the abrupt replacement that happened in the British Isles). The archaeological details of the sample were previously described in Szécsényi-Nagy et al. 2018:

The site is integrated by a collective burial of the Late Neolithic-Chalcolithic of nearly a hundred individuals. The first funerary ritual ended with the deliberated destruction of the monument by fire (Alonso, 2015). About 500 years after, the site had new funerary use within Bell Beaker context, with the construction of three individual graves made of stone (Alonso, 2013). One of the tombs studied was constructed taking advantage of the small corridor from the collective burial. In the tomb a man (older than 55 years), was buried in fetal position with one Ciempozuelos style Bell Beaker, one Palmela point, one bone point and one sphere of pyrite as grave goods. A radiocarbon date from a human bone, places the grave at the transition between Copper Age and Ancient Bronze Age.

The sample in question is modelled in Table S15 as 37.1% Iberia_CA + 62.9% Germany_Beaker, more or less average of those early Bell Beakers labelled as Iberia_CA_Stp (the average of the 13 induvuduals reported in that table is 66.74% Germany_Beaker).

To pinpoint the origin of these Bell Beakers is still quite difficult. From a genetic point of view they do look like a continuation of the Corded Ware Culture people, with the exception of their Y chromosome (R1a-M417 in CWC vs. R1b-L23 in BBC). There are some cultural similarities too, but also notable differences. For example, the typical prestige item found in the male burials of the CWC is a stone axe (Battle Axe). That’s quite more primitive than the tanged daggers mentioned above, and the latter cannot be derived from the former (especially when CWC didn’t work arsenical copper, something that is not a trivial technique that can be learned by copying, but requires significant cultural exchange). The earliest tanged dagger made of arsenical copper with a modern C14 dating (3951-3759 cal BCE) can be found in the Leila Tepe Culture:

Kurgan 1, cemetery of Soyuq Bulaq, Azerbaijan. From Lyonnet et al. 2008.

These type of daggers also appear shortly after in the NW part of the Black Sea, in the Usatovo Culture (though some up to date paper on this culture would be very welcome to clarify a few things. In any case, Rob’s previous post discusses in more detail what was going on in that area at that time), but how and why it appears later with the BBC in Western Europe is something that I honestly don’t know.

A glimpse into the Iron Age in Iberia

I’ll skip the Bronze Age proper and go directly to the Iron Age. The short story is that there are no big surprises. The details seen in the paper basically confirm the bits of information that we already had from abstracts, talks, etc… The Iron Age is from a genetic point of view an continuation of the mixing/replacement of the Neolithic/Chalcolithic populations by the incoming Bell Beakers from Central Europe/Steppe. The mixing was with the females almost exclusively, while the replacement was of the males (almost completely). By the Iron Age we don’t see any sample that still clusters with the Neo/Chalcolithic ones (all have 25% or more Steppe_EMBA admixture), and all the male samples belong to R1b-M269 except one that belongs to I2a.

So let’s take a look at those whose cultural affiliation is known:

The Iberians

There are 16 samples labelled as Iberian Culture and 1 as Pre-Iberian Culture. This one Pre-Iberian sample does belong to an important Iberian center (Font de la Canya, Barcelona, Spain), and the label should mean something like proto-Iberian, in the sense that its date (700-500 BCE, no C14) is in the initial stages of what is usually considered the Iberian Culture, which formed under the influences of Eastern Mediterranean cultures. The sample in question (I4556) is a female (mtDNA K1a12a) and it’s modelled in Table S15 as 41.2% Iberia_CA + 58.8% Germany_Beaker.

Nevertheless, the oldest Iberian Culture sample with a C14 date (I12641) is also from nearby (Can Roqueta, Barcelona, Spain) is actually older (791–540 cal BCE), so I’m not too sure why the former is labelled as Pre-Iberian. This latter sample is an R1b male, modelled as 57% Germany_Beaker. This same site of Can Roqueta has also provided 4 Bronze Age samples (2000-1400 BCE), two males and two females, with only one male yielding enough coverage to get this Y HG (R1b1a1a2a1a2). The average of this male and the two females in that same Table S15 is 38.63% Germany_Beaker, while the average of 3 of the 4 individuals from the Iberian period (790-400 BC) is 47.7% Germany_Beaker, with the 2 males being R1b. So it shows a clear continuity from the Bronze to the Iron Age (the date from the other male -I12640- is actually taken from equus bones buried alongside the human skeleton. Iberians had this thing about horses).

From the 16 samples 13 of them (plus the Pre-Iberian) are from the NE (Catalonia), while 3 are from the SE (Castellón). No difference in these last 3, with the 2 males being R1b (in fact all the 11 Iberian Culture Y chromosomes sequenced are R1b-M269) and an average of 51.46% Germany_Beaker admixture.

The Tartessians

Tartessian is a more or less mythical name that is used to refer to the inhabitants of the South Western part of the Iberian Peninsula (mostly the valleys of Guadiana and Guadalquivir rivers) who lived during the Iron Age (predating the Roman period, when we have the Turdetani people in that area, considered descendants of the Tartessians). The Tartessian language is pretty much unknown. While Iberian can now be quite safely related to Basque, Tartessian remains more obscure. Generally speaking, it’s a non-IE language, more likely than not related to Iberian, but still uncertain.

The site from where the 4 samples come from is in Seville, and it’s been excavated quite recently revealing new information about this civilization (an “Amazon” archer woman and a big male warrior made in into the news), but the paper does not have a detailed description of the burials in question.

Genetically speaking, they are in line with the Iberians, though the low sample count (only one male, R1b-M269) does not allow for further conclusions. They are dated to 700-500 BCE, and in Table S15 the average of the 2 samples with enough coverage to be analysed is 45.65% Germany_Beaker.

The Celtiberians

Here again we have a low sample size of 3, dated to ~400-200 BCE and coming from La Hoya, a site in the south of the Basque Country that was an important Celtiberian center until it got violently devastated by their enemies (Ibero-Vasconic speakers? Nah, it could be anyone and there’s no need to invent legends based on no real facts).

The name Celtiberian refers to the Celtic tribes that lived in the Central part of the peninsula (the Meseta, roughly speaking), and the Iberian suffix refers to the cultural influence they got from their Iberian neighbours, and more importantly their borrowing of the Iberian alphabet to write their Celtic inscriptions (something that didn’t reach further west to the Lusitanians, whose language is only recorded at a later time already in the Roman alphabet).

Celts and Celtiberians (plus Lusitanians) occupied a large part of the Iberian peninsula (maybe 70%?), but it was the least populated part. It’s difficult to estimate the populations and I don’t know of any recent serious attempt at it, but I guess that most people would agree that the non-IE speakers represented more than 50% of the Iberian population, in turn the largest one in Western Europe at the time.

On to the samples: We have two females and one male. This one male is, curiously, the only non-R1b one among the IA samples, belonging to HG I2a1a1a. This is probably a matter of luck, since most Celtiberians should be R1b too, but it’s still funny how elusive R1 seems to be among early Indo-Europeans. At an autosomal level, the samples average 60.63% Germany_Beaker. A slightly higher percentage possibly due to the geographical location and maybe the low sample count (though it could be due to real Central European admixture from Celtic tribes too).

The Greek colony of Empúries

One of the important events that occurred during the Iron Age in Iberia was the strong cultural connections with the Eastern Mediterranean. The contacts existed since the Chalcolithic (the presence of Ivory of Asian elephant in Chalcolithic Iberia leaves no doubt about it), but in the Iron Age they reached a different level with the foundation of colonies by Phoenicians and Greeks. They brought to Mediterranean Iberia things like Iron metalworking, the use of the alphabet and of coins, and introduced the vine and the olive tree.

This paper brings the first samples from one of such colonies: Empúries (Girona, Spain), the most important Greek colony in Iberia. And the samples are really interesting, showing clearly two clusters of local Iberians (labelled as Empuries1) and Greek colonists (Empuries2). The samples go from early after the foundation the foundation to the Roman period (500 BCE – 200 CE). The locals are just like the other samples labelled as Iberian Culture: the males are R1b-M269, and they’re all a mix of Iberia_CA and Germany_Beaker. The colonists from the Greek and Hellenistic period “cluster with Mycenaeans” according to the paper (I didn’t find specific models), while one of the Hellenistic period seems admixed between both populations (there are a few other admixed ones from the Roman period). The two Hellenistic period males in the Empuries2 cluster belong to Y chromosome haplogroup J, while the other in that cluster is a female from the Greek period. There is another male from the Roman period (also HG J) and a female (see Aegean_BA PCA below).

From Olalde et al. 2019.

UPDATE: Davidski, from Eurogenes, has updated his Global 25 datasheets with the new samples. Great job, and big thanks from here. I’ve done a quick run on the Empúries samples:

And that’s it for now. There are still a lot of things to comment about this paper, but it can’t be done in just one post. Early next week probably Robert will write about the Mesolithic and Neolithic part, and when Frank get back on track we’ll see about the rest. I’ll also start to prepare a post about linguistic substrates that I planned to write long ago but postponed until the publishing of this paper. That one will take a bit longer, but I hope not too long. So many things will be coming from now on. Stay tuned!

96 thoughts on “The genomic history of the Iberian Peninsula over the past 8000 years – Olalde et al. 2019

  1. Places like Iberia and Brittany have attracted attention for the volumes of Beaker ceramics. Yet much of these are just deposits in often previously abandoned enclosure or reused sites and megaliths. Often, there is no associated burial; or the ceramics are placed at the entrance of a communal burial place. So it’s a very different use of the BB pot

  2. I am yet unsure what to make of this paper….

    Lets start with the positive:
    1. The Mesolithic (UP?) (partial) replacement of Magdaénien by Epi-Gravettian populations is interesting. While I think that the concurrent MPI Jena paper on this issue has more depth, the Olalde e.a. paper nevertheless adds further data and analysis.

    2. I also in general like the Iron Age part.
    The Empúries case study is absolutely great and demonstrates what diligent sampling and analysis is able to uncover (a diligence the paper is otherwise unfortunately sometimes lacking, see below). An important take-away is that last year’s Mykenaean aDNA is apparently very well capturing speakers of early Hellenic, which so far had been doubted by some bloggers.
    The Iberian-speaker sample is also very robust, numerically and in terms of geographical spread, to demonstrate that their isn’t any automatism between language and external admixture and/or aDNA. Well, we knew that before – the Levante is heavily CHG-admixed, yet has been Semitic-speaking since at least a good 4,000 years, and yDNA J is apparently not helping much in finding out whether an IA settler on the E. Iberian coast arrived there from Greece or Carthage..
    Btw: Does anybody recall the steppe admix in Mykeneaens? In Iberian-speakers, Olalde e.a. have established it as 24.5%.

  3. IMO, we cannot disregard the fact that non-IE speaking Iberians and Tartessians carry R1b-L51, as modern non-IE Basques .

    It is interesting to note that the probable IE speaking Celt-Iberian carried I2a1a1a and the probable Greek-speaking colonists carried J.

    In any case, the Chalcolithic I2 haplotype in Iberia seems to be typically I2a2 or I2a1b and not I2a1a. I have recorded the following I2a1a1a samples:
    LCA Hungary Baden Balatonlelle-Felső-Gamász I2753 I2a1a1a-CTS575/PF3952
    Sintashta Kamennyi Ambar 5 Cemetery Trans Urals I1003 I2a1a1a-CTS575/PF3952
    CA Iberia El Mirador Burgos I1303/MIR25 I2a1a1a-CTS575/PF3952

    By the way, what do you think about this:
    “Finding a Bell Beaker-related group as a plausible source for the introduction of steppe ancestry into Iberia is consistent with the fact that some of the individuals in the Iberia_CA_Stp group were excavated in Bell Beaker associated contexts (9). Models with Iberia_CA and other Bell Beaker groups such as France_Beaker (P-value=7.31E-06), Netherlands_Beaker (P-value=1.03E-03) and England_Beaker (P-value=4.86E-02) failed, probably because they have slightly higher proportions of steppe ancestry than the true source population.”

  4. @Rob

    Yes, the differences in the two Bell Beaker cultures have been much easier to spot once we knew that they were different. I made the exercise of going through the Bell Beaker literature about Iberia in preparation for the Olalde et al. 2018 paper and could already predict that the early Iberian Bell Beaker Culture was a different thing from the one carried by the R1b, steppe admixed Bell Beakers, which I could only identify in a few burials of the 2400-2200 period, and not earlier.

    But of course, I had an unfair advantage, knowing in advance that it was not really possible for the R1b-L51 guys to have arrived to Western Iberia by 2700-2600 BCE. Archaeologists didn’t know this.

    As I said it’s a bit of old news by now (referring to this aDNA world that moves so fast), but I still wanted to mention it clearly because it’s a very important fact for Iberian prehistory. The biggest event since the arrival of the EEFs has been confounded into a false continuity due to a few similarities (the bell-shaped vessels most of all) that does not do justice to our real history. So it’s time to rectify it.

  5. @Frank

    It’s a huge paper in terms of sheer amount of samples and the period covered. I still didn’t have time to digest all carefully, but I expect to find more shortcomings as I dive deeper. It’s normal in sch an ambitious project. But I’m still glad they went on and publish all instead of just the more complete subsets each at a time. It seems that labs work faster sequencing samples than the people writing the papers can process all the data, so let them publish and we can help out with figuring out the details 😉

    Re: Mycenaeans, Lazaridis et al. 2017 have them at 13.2% Steppe_EMBA in a two way model with Minoans_Lashithi as the other source. I think that’s a very underfitted model. Adding more proximate sources for a population that is complex enough as Mycenaeans, the Steppe_EMBA amount is about half of that.

  6. @Kristiina

    Yes, facts cannot be disregarded in this or in any other case. I’m glad that at least now I don’t have to argue much and can just present the data so it can speak for itself. Reasonable people move away from trying to fit the data into a preconceived idea soon enough, because it’s a tiring and frustrating exercise. So then they just start to look at the data for what it says.

    Let’s not forget too that the IE speakers in Iberia had very recently shifted to Celtic around the VI century BC (except Lusitanians, who probably shifted a bit earlier to a para-Celtic language). So they must have spoken some other language before (like all other Western Europeans). I’ll talk further about this in another post.

    Re: the part you’re quoting, it refers to models with Iberia_CA and different European Beaker groups to model Iberia_BA in Table S11. The only 2 successful models are with Germany_Beaker and Iberia_CA_Stp (i.e, the early R1b, steppe admixed Iberian Bell Beakers), while for some reason the other European Beaker groups fail. I’m not too sure about the effect of including those very related Bell Beaker sources in the outgroups. Might be a good strategy to get very precise discrimination. But it could also cause some problems. Since the samples are out, let’s see what further tests show regarding the best admixing source into Iberia_CA.

  7. I updated the post with a quick run on the Eumpúries samples (and included Mycenaean average for reference). One of the two “Greeks” gets a poor fit, so I’ll have to check that one more carefully to see if it’s a quality issue or I’m missing some needed reference.

    UPDATE: Nothing really seems to improve that fit. It just takes Minoan but it’s mostly the same, so I’ll leave it as it is.

    Minoan_Lasithi 30.8%
    Protoboleraz_LCA 28.45%
    Kura-Araxes_Kalavan 25.1%
    Vinca_MN 13.05%
    Hajji_Firuz_ChL 2.6%
    Iberia_Central_CA 0%
    Yamnaya_Samara 0%
    Peloponnese_N 0%
    Anatolia_EBA_Isparta 0%
    Anatolia_EBA_Ovaoren 0%
    Romania_HG 0%
    Balkans_ChL 0%
    Balkans_N 0%
    CHG 0%
    Levant_N 0%

    Distance 3.4387%

  8. @ Alberto
    Indeed kudos, you had pointed out to me what to look for, and digging deeper.

    @ Kristiina

    I suspect the best source for Iberian steppe admixture during the BB would be southern France
    The estimates would therefor be higher

    Beaker_France_South 57.8%
    Iberia_ChL 34.6%
    Beaker_France_no_steppe 7.6%
    Armenia_EBA 0%
    Anatolia_BA 0%
    Remedello_BA 0%

    Distance 1.5585%

    A 60% turnover in the north is probably consistent

  9. @ Frank

    I haven’t looked into Mycenenas closely lately , but in terms of raw Yamnaya ancestry, 10% doesn;t seem unreasonable for my brief look.

    Aegean_N 69.3%
    Maykop_Novosvobodnaya 15.2%
    Yamnaya_Samara:I0370 10.1%
    Baden_LCA 5.4%

    Mediated via the Balkans, that figure can rise to ~15%

    Aegean_N 67.5%
    Yamnaya_Bulgaria:Bul4 16.7%
    Maykop_Novosvobodnaya 12.7%
    Anatolia_BA 2.5%

    Distance 1.9297%

    But mabe Alberto can look at it more closely ?

  10. My hunch on the preference for Germany_BB is that it may be because the chronological phases were set wrongly. As already pointed out in the opener, we are dealing with two very different CA phases. However, by the end of the CA, i.e. 2,200 BC, the yDNA replacement/ NW European Beaker admixing process wasn’t yet finished. Fig. 2B shows clearly that by that time, something like 30% of all males had yet other aDNA than R1b, and “locals” and “newcomers” were still fairly unadmixed – either pure “local” vs. “Beaker” ancestry, or first, at best second admixed. The real shift happened during the EBA, i.e. 2,200-1,800 BC, when the population homogenised. And even afterwards, there seems a trend towards a gradual increase in the Beaker share. So, for a realistic modelling, we need to consider the time slice 2,500-1,200 BC.

    Another problem becomes apparent when you look at Fig. 2a/ Tab. S8:

    – In SW_Iberia, WHG increases from the MLN to the CA, which is what would be expected from NW Beakers coming in. However, most of that WHG increase stems from El Miron, not from KO2 ancestry. Actually, if assuming the shift is just caused by a single source, the incoming WHG can be modelled as 65% El Miron, which means pretty close to Canes, and much more Magdalénien than Loschbour. Either quite a lot of Canes-like Mesolithic pops were hiding out somewhere in Estremadura, to only become neolithicised during the CA, or we had another, quite Magdalénien-heavy migration preceding the Beaker influx. I tend towards the second option – Megalithic inflow from Brittany, for low sampling depth incompletely captured in the MLN, and only becoming visible during the CA.

    – Same tendencies for NE and N Iberia – here actually KO1 ancestry even goes down from MLN to CA;

    – For SE and C Iberia , we have a quite different pattern. ANF ancestry increases slightly, KO1 somewhat more, both on the expense of El Miron. This can’t just have been beakers, It was also some, maybe even quite a lot inflow from the Mediterranean that can plausibly be linked to the emergence of gold, silver and copper mining (probably Remedello via Sardinia, but maybe also some guys from much further East). [C_Iberia, at quite low ANF ancestry during the MLN, may also have seen some “return to the mean”.]

    In short – they are partly comparing apples with oranges, i.e. ignoring regional differences and archeological well documented changes prior to BB (I mean – no consideration of Los Millares in such a study, how can that be!), averaging all out over poorly chosen time transsects, and do some simple three-source admixture models in order to confirm something that was essentially already clear from last year’s study. And the really interesting and new question isn’t even posed: What the hell happened during the EBA, when BB was essentially already over, and El Argar took the stage?

    That takes me to the next disaster – the BA. All those archeologists on the ground seem to have done a pretty good job in putting together samples for two major BA phenomena, namely
    1. El Argar (including sampling some of the pithoi burials – otherwise they seem to rather have captured outward farmers than the elite),
    2. Las Cogotas on the Meseta, assumed to display BB continuity, and the prime suspect when it comes to having provided the IE layer from which Lusitanian and Celtiberian are by some believed to have emerged.

    As natural for such an exercise, you increase sampling depth of specific phenomena at the expense of comprehensive coverage. From what I have seen, there isn’t a single Portuguese BA sample, I think also nothing for the N Iberian BA. I think there is a bit of BA samples from Valencia, maybe connected to El Argar, maybe Independent of it. About Catalonia, I’m not sure.

    As I have said, this is perfectly fine if such selective sampling is handled adequately, they way they have done it for Empuries and, except for some shortcomings, also post-Roman Catalonia and Arabic Andalucia.
    However, if such a highly unrepresentative sample, with predominance of Cogotas known to be the “last BB holdout”, is taken to represent BA Iberia as a whole, things really gets unscientific. Their findings are credible for the Meseta, and also as concerns El Argar’s working class. For the Meseta, they were somewhat expectable, El Argar working class being all R1b comes as a sort of surprise – let’s wait and see what the elite graves will yield, once they are sampled. But beyond that – especially Portugal (Lusitanian, Celtiberian), and N. Iberia including the Basque Country – they don’t seem to have provided anything that helps us to better understand the dynamic between the BB and Roman periods.

    If I were you, Alberto, I would still postpone linguistic considerations a bit, and for the time being focus on
    (a) turning the wealth of raw data provided into potentially meaningful subsets (adequate time slices and cultural assignations),
    (b) subject those subsets to sensible, archeologically informed modelling, and
    (c) try to figure out more about what kinds of population changes occured aside from BBs.

    E.g., I find it intriguing that their best model for SW Iberia_IA is 60,5% Iberia_CA plus 39.5% Unetice. Alberto, do you remember our old E-mail discussion about Unetice-El Argar parallels (halberds, plain pottery) and the obvious break with BB traditions they signify? The a/m model adds an intriguing new perspective to these observations I hadn’t even dared to consider so far…

  11. thanks Alberto:
    I could not quite understand what you write near/after

    “Let’s not forget Gimbutas’ non-existent 1st and 2nd waves of invading steppe warriors that devastated “Old Europe””

    Are you refuting Gimbutas, refining her conjectures or corroborating her views?

  12. @ Frank

    I suspect the choice of German BB is mostly technical – they are the most sampled population, with a combined best coverage, for optimal statistical results. They outline in the paper that in fact none of the choices were quite optimal (Dutch, German BB, etc). More sampling from France might in the future approximate the data more closely.

    Can you outline more about the El Agar set ?
    Which were the males from ? What kind of burials ? How do you know they were just ”workers”

  13. @Frank

    Thanks for all the pointers. I certainly still have to look much more closely at the specific samples, areas, periods,… I left the Bronze Age out of the post because after my first quick reading to put out this post I didn’t have a clear idea about what to say about it. We’ll look into all the details soon enough. Let me know when you get your email working so we can plan about the pending posts.

    Interesting about Unetice showing up there, though I’m still not too sure about the modelling strategy. As mentioned above in a reply to Kristiina, it might be working wonders, but I don’t really know for now. It could also be giving slightly random results. I’d also point out that it’s SW Iberia IA, not SE which is the area of El Argar influence. And IA is a bit late too discriminate such small genetic details about what happened c. 2200 BC. I have to check better, but I didn’t see any of the samples from La Bastida (the origin of El Argar Culture) that were mentioned in an interview about a year ago. They might be coming on a later publication.

  14. @postneo

    What I meant is that Marija Gimbutas saw 3 successive waves of steppe invaders into Europe, based on observations about increasingly hierarchical, patriarcal and warlike societies. However, aDNA has refuted her first two waves, which never happened. What she was seeing as steppe invasions were only internal transformations of the own societies at those times, which were becoming increasingly complex. This increase in complexity brings those natural changes, and cannot be taken as something specific to one culture.

    Throughout the Chalcolithic, Iberia (specially the southern part) became quite a complex society (by Western European standards), which led to changes that resemble those that came with the R1b folk subsequently. But they were still completely unrelated cultures and people, even if both were given the same name.

  15. The Iberian Bell Beaker yDNA record looks like this at the moment:

    Cerdanyola Barcelona Spain I0261 2850-2250 BCE R1b1a2-V88
    Cerdanyola Barcelona Spain 2833-2480 calBCE 0826 I2a2-M436
    Verdelha dos Ruivos Lisbon Portugal 2700-2300 BCE I1970 I-M170
    Cerdanyola Barcelona Spain 2571-2350 calBCE I0257 R1b1a2-V88
    Cerdanyola Barcelona Spain 2474-2300 calBCE I0825 G2
    Arroyal Burgos Spain 2458-2206 calBCE I0458/Roy1 I2a2a-M223
    Arroyal Burgos Spain Roy3/I0460 I2a2a-M223
    South Meseta Camino del las Yeseras I6622/RISE693 P1(xR1b1a1a2)
    Cova da Moura Portugal 2336-2063 calBCE I4229 I2a1a1-L158
    Humanejos Madrid Spain 2500-2000 BCE I6587/Hume 15B I2a2a- M223

    Humanejos Madrid Spain 2500-2000 BCE I6588/Hume 10A R1b1a1a2a1a-L151
    Humanejos Madrid Spain 2456-2046 BCE I6539/Hume 5A R1b1a1a2a1a2-P312(xR1b1a1a2a1a2c)
    La Magdalena Madrid Spain 2500-2000 BCE I6472/RISE701 R1b1a1a2-M269
    Virgazal Burgos Spain 2281-1985 calBCE I5665/RISE911 R1b1a1a2a1a2-P312

    Non R-L51 = 10! R-L51 = 4.

    Instead, Valdescusa is a Bell Beaker associated site with Early Bronze Age chronology and all yDNA is R1b-L51
    BA Iberia Valdescusa La Rioja 1867–1616 cal BCE VAD001 R1b1a1a2a1a2a5-F1343/S225
    BA Iberia Valdescusa La Rioja 1689–1528 cal BCE VAD002 R1b1a1a-P297/PF6398
    BA Iberia Valdescusa La Rioja 1673–1255 BCE VAD004 R1b1a1a2a1a-L151/PF6542
    BA Iberia Valdescusa La Rioja 1742–1546 cal BCE VAD005 R1b1a1a2a1a-L151/PF6542

    The Iberian Bell Beakers are far from being all R1b-L51. However, it is during the Bronze Age that Iberia becomes a R1b-L51 land.

    Many non-R1b-L51 Bell Beaker burials contain vessels in Marittime style while the Bell Beaker burials from Humanejos with R1b-L51 are described as follows: Nine Bell Beaker tombs were excavated with extremely rich grave goods (decorated pottery, coppery weapons as dagger, Palmela type point and even an axe and an Atlantic halbeard, gold jewelry, ivory V-perforated buttons and necklace beads and stone wristguards).

  16. @ Frank

    ”However, by the end of the CA, i.e. 2,200 BC, the yDNA replacement/ NW European Beaker admixing process wasn’t yet finished. Fig. 2B shows clearly that by that time, something like 30% of all males had yet other aDNA than R1b, and “locals” and “newcomers” were still fairly unadmixed – either pure “local” vs. “Beaker” ancestry, or first, at best second admixed. The real shift happened during the EBA, i.e. 2,200-1,800 BC,”’

    As Alberto points out, the shift begins c. 2500 BC, not 2200 BC., which is the tail end of the process. So it occured over ~500 years,.

    ”Another problem becomes apparent when you look at Fig. 2a/ Tab. S8:
    – In SW_Iberia, WHG increases from the MLN to the CA, …”

    I did not catch anything odd in what they reported, or at least what their data shows.
    Usual pattern as expected – rise in WHG ancestry during the MN, which is ~same for Chalcolithic and ‘non-steppe Beaker’ period individuals.
    There is a north -> south cline in WHG, with ~20% for S.E. Iberia and 30% in Northern Iberia. This is mostly Central European WHG, not the El Miron-rich stuff. The first sporadic appearance of North African ancestry in southern Copper Age.

    The Beaker -set individuals with steppe ancestry (as shown above) are predominantly foreign (from north of the Alps), despite being already significantly MNE -admixed., and having acquired further MNE ancestry in Iberia within a generation or two.
    Whatever the exact mechanisms of takeover, these then admix into residual native Chalcolithics (mostly only females), again with a north -> south gradient.
    Again in the Iron Age there are further mirgations from central Europe, permeating from north to south farily diffusely, and the appearance of Levantine/ N.A.- ancestry in the south from 10th c BC.

    So we have at least 4 language avenues to explain the 3 or so groupings of Languages in Iberia (depends on if Acuitanian-Basque can be considered as part of ”Iberian”, which Alberto will go into at some point into future).

  17. I checked the archaeological context of the steppe-rich individuals of this new paper. None is from a Bell Beaker context, although one of them was looted so we may miss material that would identify it with a specific culture.

    CA Iberia Castillejo del Bonete Castilla-La Mancha 2271–1984 cal BCE I3484 R1b1

    Castillejo del Bonete was a ceremonial site used for more than one thousand years during the Copper and Bronze Ages. Rites performed at this site were related to death and resurrection of the sun, human death, and veneration of ancestors. Some examples are feasting rites, offerings to the dead, and architecture oriented towards the winter solstice (70, 71). A natural cave was monumentalized and used as funerary chamber, building a large tumulus and creating cave art. Another good example is Tumba 4, the only multiple burial in this site with a 40-50-year-old male and a 30-40-year-old female who was buried with two ivory buttons and who had a marine diet, suggesting a non-local origin (74). We sampled both individuals from Tumba 4.

    CA Iberia Steppe El Hundido Castilla y León 2287–2044 cal BCE EHU001 R1b1a1a2a1-L51/M412(xR1b1a1a2a1a2a5-F1343/S225)
    CA Iberia Steppe El Hundido Castilla y León 2562–2306 cal BCE EHU002 R1b1a1a2a1a2-P312/PF6547(xR1b1a1a2a1a2a5)

    The Late Neolithic and Chalcolithic burials documented at the site of El Hundido provide interesting new information concerning funerary rituals at the beginning of the Age of Metals in Iberia and the cultural connections between the Meseta and the upper and middle Ebro Valley. The collective burial of nearly a hundred individuals is an exceptional funerary monument in terms of both its structure and its complex ritual practices. These culminated in the deliberate destruction of the monument by fire. El Hundido es en esencia una tumba colectiva calcolítica donde más tarde, a mediados del III milenio cal BC, cuando todavía era reconocible, fueron practicados tres enterramientos intrusivos campaniformes.

    CA Iberia Steppe NW Cueva de la Paloma Asturias 2500–2200 BCE I3243 P1

    Bone samples from 4 adult right tibias are analyzed in this study, out of the 91 anatomically modern human remains originallyrecovered at the site. The new dating of the analyzed human remains, however, does not correspond to an Azilian archeological and chronological context as originally published (40–42, 46), but instead to a Late Neolithic-Chalcolithic chronology. In fact, it was already noted that the superficial levels of the site were removed by looters (40), mixing the sediments and altering the stratigraphic units.

  18. @Kristiina

    Yes, exactly. You’ve shown the problem there. In Iberia we have a late Chalcolithic culture that evolved from the previous ones like VNSP and Los Millares. One of the characteristics of this culture was the bell-shaped beakers of maritime style (I believe the earliest are from near Lisbon c. 2700-2600 BC, and probably it was an evolution from the previous copos).

    Meanwhile, in Central Europe, some people who came from the steppe started to expand throughout Western Europe. One of the characteristics of this culture was a bell-shaped beaker with different decorative styles (cord impressions, etc…).

    This latter culture started to arrive to Iberia c. 2400 BC, and it meant the most disruptive event on a population level since the start of the Neolithic until today. Roughly between 2400 and 2000 BC the population of Iberia was dramatically transformed.

    The detail I didn’t mention above is that the former culture was called the Bell Beaker Culture, and the latter one was called the Bell Beaker Culture.

    While both shared some similarities, they also had significant differences (single burials and metal weapons -tanged daggers, palmela points- in the latter vs. collective burials and flint weapons in the former). Though it’s the DNA what really confirms that they were completely unrelated people, one being of local origin and the other originating in the far east of Europe.

  19. @Rob

    I haven’t looked into Mycenenas closely lately , but in terms of raw Yamnaya ancestry, 10% doesn;t seem unreasonable for my brief look.
    But mabe Alberto can look at it more closely ?

    The quick run a posted in the article had them at 5.2% Yamanya, which was a bit lower than I remembered. I’ve added Minoan and Anatolia_EBA, which improved the fit and pushes Yamnaya up to 6.3%. No big deal, but for the sake of greater accuracy I updated it in the post.

  20. @Alberto, on the Mycenaeans and Lazaridis 2017’s two way model, Davidski’s put a new datasheet on his site, with a massive matrix of ancient and modern outgroup f3 statistics (think using Mbuti).

    Using 292 outgroup f3 statistics (229 ancient, 63 modern), it looks like some Caucasus related input is needed as Mycenaean is too rich in outgroup f3 to present day and ancient Caucasus/West Asian/Iranian/South Asian and too poor in f3 to present day and ancient Europe to fit between Minoan and Yamnaya. (Minoan+Steppe or Minoan+Europe_LNBA is too Anatolian and too European HG). Doesn’t look like error to me.

    It looks like Armenia MLBA 16%, Yamnaya_Samara 12%, Minoan_Lasithi 72% fits fairly well. Different proximate groups could change proportions, not necessarily literally true.

    Three way models with Balkans_Chl / Sintashta_MLBA viable, however this would need to increase Anatolian and WHG f3 poor CHG related population as well. 2-way model with Balkans_Chl might work if mixed with a population with no Steppe ancestry like Kura_Araxes.

    Oddly looks like the opposite of what Lazaridis’s paper suggested, that there was a very slight offset of shared drift with European LNBA compared to Minoan+Steppe.

    This said, there’s some odd stuff up with the Mycenaean f3 stats being relatively high to African outgroups, hopefully this shouldn’t affect this, I have mentioned in his comments and poss. Davidski would have another look at it. I don’t know about G25 fits for the same populations as well – as I recall trying to extend clines from Minoans and Mycenaeans there pointed to some kind of West Asian ancestry related enrichment as well.

    (Examples on imgur link – Basically I am using correspondence analysis in PAST3 with a limited set of columns and rows as (wholeset-columns). In the second image, the populations labelled Sim are with decreasing levels of Minoan and increasing levels of the population they are “pointing” at.)

  21. @Alberto:
    I’d also point out that it’s SW Iberia IA, not SE which is the area of El Argar influence.
    I know. But SW Iberia IA seem to be the “Tartessian” samples that should originate from not too far away from the El Argar zone. SW Iberia IA hasn’t been sampled (would have been interesting to learn more about Phoenician/ Punic influence there, but even with more than 170 samples. you can’t cover everything).

  22. Postneo:
    One of the “markers” used by Gimbutas is the spread of lanceotic daggers and arrowheads (her Fig. 10.4, 10.7, otherwise google “leaf-shaped arrowheads” for further detail in addition to the sources below), a reflection of which are BB daggers and Palmela points. She correctly traced back the European roots to Khvalynsk, with spread via Skela, Cucuteni, up the Danube (e.g. Bödrogkeresztur, Lengyel, Mondsee, Cham) and across the Alps (“Iceman”) to Remedello. They are furthermore typical of Western Michelsberg and the British Neolithic (“leaf-shaped arrowheads”), where they may have arrived via a Northern, Combed Ceramics route leading from Khvalynsk up to the Norwegian coast.

  23. For a contrast, see below arowheads from Baalberge, Elbe-Havel Culture (GAC West), and an Elbe-Saale BB assemblage. An idea of Rössen, TRB and Single Gave Culture forms from E. Westphalia is provided by Fig. 2 in the final link. Differences to the “leaf-shaped” forms are obvious.

  24. Now we know that the spread of lanceotic forms wasn’t associated with significant Steppe ancestry. Furthermore, the earliest N. Iberian examples, still lithic, date to ca. 3,500 BC and are associated with “a period of violence”. I tentatively link that, as the contemporary increase of El-Miron related ancestry documented in Tab S.8 of the new Olalde e.a. paper, to incoming populations from Brittany or England .
    Turning early lithic forms into arsenic bronze Palmela points appears to have been a local Iberian_CA, pre-BB innovation.

    If these forms are traced backwards further, it becomes apparent that their European origin was not Khvalynsk, but the introgressive Pre-Caspian culture from which Khvalynsk originated. (see Fig. 4)

  25. Contemporary to the European “expansion”, leaf-shaped points also spread eastwards (Atbasar, Botai, Lokomotiv/Balkai, Taram Bulak), and along the Silk road (Altyn Depe, Gonur, Sarazm, Tarim Basin etc.) and into S. Asia (Mundiglak, pre-Harappa). Their origin is yet unclear, but triangulation suggests an E. Caspian origin, possibly Djebel near the Turkmene coast, the not yet precisely dated “Neolithic” layers of which held such forms.
    “Kelteminar Points” are related, but have a distinct, assymetrical shape. A few leaf-shaped points, however, have been found in Kelteminar II (early 5th to mid 4th mBC) layers.

    A “Palmela Point” from Tepe Hissar, Iran, ca. 2,500 BC can be seen here

    A prototype, not yet from arsenic copper but from a copper-nickel alloy was found in Kül Tepe I, Nakhichevan, ca. 4763-4506 cal B.C. Copper-Nickel ores a/o occur in E. Georgia. Fig. 4

    [Might be interesting additions to your thread opener, Alberto].

  26. I want to offer an alternative approach to the Steppe admixture of Mycenaean and Greek Empúries colonists that replaces Yamnaya related Steppe sources.

    The main reason is to explain the lack of R1b/R1a thus far and the exclusive findings of J2a.

    The second reason is the geographical distance to those Yamnaya Steppe populations.

    Third reason is the presence of samples with relatively high EHG amounts in the relevant region between Bulgaria and Greece. These are not Yamnaya/Steppe related populations but ~30% EGH, ~70% EEF/ANF mixes that lack significant CHG.

    Fourth reason is that Greek IE language may be more related to non-European IE branches such as Armenian and Iranian. Hence the Indoeuropeanization there might differ to the concept of a spread of late-PIE expansion via Yamnaya derived populations.

    aDNA sample I6423 is a representative of an increased EHG population in Greece, already by the neolithic which would not be possible to be related to the Yamnaya-IE expansions (too early). It was in the Wang et al. pre-paper but unfortunately not released.

    Hence it is a path to explain a non-Yamnaya Steppe-like admixture for mainland-origin Mycenaeans. Certainly a valid one in terms of geographic distance and time.

    This, indirectly, also means that middle age Slavic expansions into Greece are not exclusively the reason why today mainland Greek populations are not like ancient Greek aDNA (incl. mainland) found so far.
    We don’t know exactly from where I6423 was but I expect it to have been from a northern region. Maybe a population like that, was the reason Macedonians were regarded as Barbarians/non-Greek by proper ancient Greeks. Hence that kind of population would have it’s fair share into shifting today mainland Greeks towards more Steppe-rich populations, without being ultimately Yamnaya-Steppe related.

    In this scenario I6423-like population was due to a mixture event of a originally EHG-EEF(ANF) mix population, with a CHG/Iran heavy population when these expanded towards Greece. Steppe/Yamnaya would have no role in this, early event.

    That event can be approximated by a two way mixture of Balkans_Chalolithic sample I2181 (4550–4450 bc and Y-DNA R) or Varna outlier ANI163,
    with a Iran_ChL like population such as Hajji_Firuz_C or Tepe_Hissar_C.
    I2181 and ANI163 samples are from the Mathieson et al. 2017 paper.

    Hence the ideal model for the non-Yamnaya-Steppe population responsible for a Steppe-like admixture would look like this for the Greek colonists from Empúries Spain of ultimately mainland Greece origin:

    Greece_N : Oldest farmers G2a, pure EEF/ANF
    Levant_N : Among or shortly after G2a farmers, Levantine farmer population related to Y-DNA E and Natufian admix.
    Bulgaria I2181/ANI163: a northern farmer/hunter-gatherer population potentially bringing as newcomers R1b/R1a and EHG (minor Y-DNA I would be already there from early WHG)
    Iran_ChL Hajji_Firuz_C: early CHG/Iran admixture related expansion, potentially Kura-Araxes/Hattic/early-IE (Hittite like?) related, bringing J2/J1 as newcomers
    Maykop_Novosvobodnaya : later Mycenaean “warrior culture” IE related expansion from the east, bringing J2a and potentially first IE dialects and culture

    Excluded are potentially earlier populations like WHG origin YDNA I or H, T and L, as these had negligible genetic impact.

    I would be thankful if someone could run an admixture run with those source populations. This is also relevant for the Myceanean samples and for a comparison with Minoans (old ones from 2700 BC and late ones from 2000 BC).

  27. Matt: “there’s some odd stuff up with the Mycenaean f3 stats being relatively high to African Outgroups

    Could relate to Mycenaean trade with Egypt (e.g. Baltic amber), and Ethiopia’s post-Mota integration into the Near Eastern trade system.
    Try out Mota, or Ari, as additional sources.

    Patameres: The reason I was banned from Eurogenes was suggesting exactly such a route as envisaged by you. I am certainly not through with that issue – e.g., tholoi (amphora) burials first appear in the Caucasus (Areni 1, etc.), but I will leave a detailed write-up to another occasion.

  28. Kristiina:
    1. El Hundido is BB. If my rudimentary Spanish doesn’t deceive me “enterramientos intrusivos campaniformes” translates as “intrusive BB burials”, starting from the mid 3rd mBC.

    2. I recommend to treat most AMS dates with some caution. In the SI 2 description of AMS dating methods, there is no mention of any assessment of and eventual correction for reservoir effects, not even for the Castillejo del Bonete Castilla-La Mancha female with maritime diet.
    I think it is fair to assume that most coastal dwellers had a substantial maritime element in their food that may require some downward correction of the original AMS dates.
    Another alarm sign goes on whenever I read “Carstic cave”, which seems to have been the prevailing source of samples before the IA. Carstic means carbonised, “hard” groundwater. In regions with little rainfall, at least during the summer, humans (as most animals) will strongly rely on groundwater. For SE Ukraine, “hard groundwater” effects have been shown to lead to Neolithic/ CA bones being regularly dated some 250 years older than organic substance (pollen, grass etc.) from the corresponding archeological layer.
    As such, I recommend to rather orient on the lower chronological boundary given, or to reduce the study’s AMS median age by around 300 years, as long as the dating is just relying on AMS-dated bones, and a maritime diet and/or reliance on carbonised ground water can’t be excluded. The latter might e.g. apply to some sites in the Ebro Valley.

  29. @ Frank It also occurred to me, because Bell Beaker culture in Spanish is Cultura del vaso campaniforme.

    However, if you look at fig. 2, it shows the picture of a “cista campaniforme”, which I interpret as referring to an “enterramiento campaniforme”. It is clearly not about pottery in this context. In Chapter 3.2. Céramica, I did not find the word “campaniforme”, and in addition to this, the description of pottery does not refer to “marittime” or “Ciempzuelos” styles which are characteristic of Iberian Bell Beaker pottery.

  30. Kris: “campaniforme” is the standard term for BB in Romance languages, c.f.

    Rob: “Can you outline more about the El Agar set ?
    There was actually some kind of mis-reading on my behalf when I initially glanced over the SuppMat. I was especially referring to this section:

    Cabezo Redondo (Villena, Alacant/Alicante, Valencian Community, Spain):
    Contact: Gabriel García Atienzar, Mauro Hernández, Virginia Barciela González, Domingo C. Salazar-García

    A preliminary review of the human remains of the Cabezo Redondo shows the different conservation of the remains according to the burial space. The burials deposited in pithos, cists, and individual graves show a good state of conservation, while those found in caves are more disturbed. Demographically, at least 61 individuals are identified, with children being the best represented age range at the site. The number of juveniles is low, as is the number of adults, with a predominance of those who died between 30 and 39 years old. The abundance of children is interpreted as evidence of a high birth rate and a higher
    number of deaths in the early stages of life.
    (..) In the adult population, there are signs that indicate intense physical activity in the arms and legs, typical of a population dedicated to cultivating the land and caring for animals. (..) Evidence of the use of teething for non-food activities is noteworthy. The presence of grooves in the anterior teeth of some individuals shows one of the few documented cases during recent prehistory related to textile activities (35).
    The evidence related to social prestige in Cabezo Redondo is abundant and appears to be associated with habitation areas and funerary deposits. Interestingly, the presence of ornamental objects in only a few burials of adult individuals and, in particular, of a few children, reveals social differences between the inhabitants of the village and the hereditary nature of some privileges. Some ornaments are exceptional beyond their raw material. The gold and silver truncated cones, as well as the ivory combs and glass beads, reveals connections with the El Argar culture. ”

    So, there is extreme social stratification present. However, there is no Information available about the burial context of the individuals analysed, whether acc. to kind (pithos, cist, cave), or relating to grave content.

    Another El Argar site sampled is Cerro de la Encina. The description in the SuppMat is rather Brief, w/o any characterisation of the sampled burial. A good description is provided in
    “The funerary record shows a significant concentration of wealth in burials corresponding to the family groups of the highest social status. Dramatic social differences can also be found in the internal organization of the settlement. The locations of burials within the settlement area, under the floors of dwellings, allow us to establish that the settlement space was closely related to the social identity of the families. The high number of burials with double and triple inhumations, in contrast to other Argaric necropolis, also stands out as an important feature of Cerro de la Encina, suggesting that familial relationships seem to be more marked here than at other Argaric sites.

    Also sampled was Cerro de la Virgen, Andalusia. While appearing in all listings of El Argar sites, I couldn’t figure out anything specific – apparently just a countryside settlement.

    Finally, they sampled La Navilla, “part of a group of collective burials (megaliths) located in the Alhama region in the southwest of the Granada province (83). It is a corridor tomb with a trapezoidal chamber located in the right bank of the Cacín river and containing 34 burials. It is surrounded by a group of orthostats.
    As per
    Culturally it is quite different from the Castellón Alto and Fuente Amarga settlements since it belongs to the Copper Age. There are no burials under the cottages or in the little caves, and they were stock-breeders but not farmers; however the pottery found in the burials is an Argaric typology

  31. @Matt

    Thanks. I have to check that f3 matrix, but I guess it’ll require some extra work before you can feed it into a script to get some model based on lowest distance (nMonte or similar). Maybe some PCA as you once did with the Fst matrix? Or maybe selecting columns carefully. I’ll get to it at some point if Davidski decides to keep maintaining it.

    As for the steppe in Mycenaeans, I won’t really argue much a specific number because we have few samples with varying amounts. I think that one of them does have some 10-12% Yamnaya, while 2 others have very low (one <5%, other traces if at all). And the high class female is too noisy and can be modelled with very few or very high steppe, so I don't trust it.

    I think the picture might be similar to what we see with those 4 and these two ones from Empúries: varying steppe admixture between 0-15%, but difficult to predict the exact average, and paucity of R1 haplotypes. More important will be to see the context of each sample and if there is some pattern regarding the steppe admixture or not (hopefully the early and clearly elite shaft graves can yield some usable DNA at some point, and that would be very valuable).

  32. @Patarames

    Thanks for the suggestion. Tomorrow I’ll look at that. The kind of admixture you mention should be easily found in Ukraine before the arrival of the Progress-Vonyuchka types (i.e., in Sredy Stog I, c. 5500-4500/4300 BC). The Varna outlier is available in Global 25, so I’ll see how that works for Mycenaean samples.

  33. @Frank

    I think I couldn’t find Palmela points before 2400 BC or associated with collective burials. If I remember correctly they were only present in single male burials, so probably only with R1b guys. But I’d need to double check, since I might not remember correctly now.

    @Kristiina & Frank

    Re: El Hundido, I quoted a previous paper (that brought the mtDNA) in the post. The EHU002 sample comes from an intrusive single male burial with Cimpozuelos style Bell Beaker (and a Palmela point).

  34. @Frank

    Yes, the site Cabezo Redondo seems like the best bet to get a glimpse of El Argar elites. But as you point out, there is no information regarding the context of the specific samples analysed. They have lower than average Germany_Beaker admixture, but 3 samples without context cannot tell us much. I might try to contact the authors to get further information if possible.

  35. Here follow the respective sample designations:

    Cabezo Redondo (urban, high social differentiation, multiple burial cultures)
    I3486/S-EVA 26078: 1700–1500 BCE
    I3488/S-EVA 22926: 1700–1500 BCE
    I3487/S-EVA 26688: 1734–1617 cal BCE (3365±20 BP, PSUAMS-2161)

    Cerro de la Encina (socially differentiated)
    I8140_d/13: 2117–1779 cal BCE (3590±40 BP, Beta-230003)

    Cerro de la Virgen (countryside settlement)
    I8144/8: 1877–1636 cal BCE (3426±34 BP, Ua-39403)
    I8136/19: 1606–1418 cal BCE (3216±33 BP, Ua-39408)

    La Navilla (stockbreeding, transitional)
    I8048/13: 2200–2000 BCE
    I8141/7: 2200–2000 BCE
    I8142/8: 2200–2000 BCE

  36. @ Alberto
    How do you think this effects PIE hypotheses (the findings that correlate R1b-M269 with an obviously non-IE context). Sorry, probably too broad & controversial a question, but what’s your brief perspective ?

  37. C. Alonso Fernandez is the author of the report “La tumba colectiva El Hundido (Monasterio de Rodilla, Burgos) y su ritual funerario durante el Neolítico Final y el Calcolitico”. She is also the reference mentioned in supplementary material of this new paper for El Hundido. It is possible that the archaeolocial report, which is very detailed, is only meant to cover the older burials and omit the more recent burials. However, I would still have expected that these Bell Beaker items were mentioned in the report, BUT as they are clearly mentioned in the supplementary material I presume that they are true.

  38. Alberto: Palmela points weren’t part of the classical Central European BB package – here, other symbols such as artefacts made from boar tusks, as already present in GAC and even the Ukrainian Neolithic, played a much more prominent role. I was actually surprised to see a Palmela point in the linked photo above of an Elbe-Saale BB burial assembly. The form is alien to TRB/GAC/CWC/Single Grave traditions and speaks for a more south-western origin, and bronze-making was certainly higher developed in Iberia during the time in question.
    If Palmela points weren’t an Iberian innovation, the roots should lie in the Mondsee-Altheim tradition that already during the 4th mBC produced bronzes, albeit with only some 3-4% tin content.

  39. @FrankN

    I’m looking forward to your work on this topic. The concept I described likely have been inspired by your writings.
    Those EEF-EHG mixes of various ratios existed at the right time and right place.

    The possibility of a IE Steppe branch of Yamnaya origin is actually still a good candidate for me, based on Iranian and Scythian mythology (latter via Greek sources).
    However I mainly favor a coarse Iran/CHG association with at least southern IE branches (among non-IE languages), which is supported by such findings as the Greek colonists and their Y-DNA.

    At my level of work some (at least subconscious) agendas may exist, but banning your kind of objective work is really needless.


    Thank you, that would be great.
    You are right, there are more than those two mentioned EEF-EHG samples but for now they are quite rare. Most have already a CHG/Iran admixture.
    Hence I’m very interested to know how that extra of EHG works out for the ancient Greek samples.

    These would be some alternative EEF-EHG mixes:
    I4332 Balkans-BronzeAge
    I1927 Trypilla

  40. @Toby_P

    How do you think this effects PIE hypotheses (the findings that correlate R1b-M269 with an obviously non-IE context). Sorry, probably too broad & controversial a question, but what’s your brief perspective ?

    I think this is just one more data point in the whole puzzle. Not really new, but now well confirmed. While I have briefly commented before on this blog on my view about the PIE question, I’m not too keen on extending on it, since it’s mostly about making predictions that are still quite speculative (and bring controversial debates that don’t add much information, but do add a lot of noise).

    I prefer to keep looking at the data as it comes out and once we get the whole picture we’ll see if we can get a clear answer as to the origin and spread of IE languages. To get a relatively clear answer is my only wish, since I’m aware of the possibility that we won’t get it, leaving us still with just hypotheses.

  41. @Patarames

    I’ve run a few tests using Global 25 datasheet. The first tab is with your suggested sources, while the others add Yamnaya and other alternative setups.

    I’m not sure that too much can be said from the results. I’d still prefer Yamnaya as a source of steppe admixture in Mycenaean Greece, if only because the evidence of any widespread admixture of the Varna_outlier type is missing in other Balkans samples while we do have evidence of Yamnaya intrusions in the Balkans. There’s also the Y DNA thing, where R1b-Z2103 is today relatively common in the area, and that one is from Yamnaya and not from early Chalcolithic sources.

    But I’m open to whatever possibility further sampling might show.

  42. @Frank

    I really don’t know whether Palmela points were an Iberian innovation or not. Or if the inspiration came from local Iberian precedents or from outside. But until I have the chance to double check (or someone can point me to pre-R1b intrusion presence of Palmela points), I’m more or less convinced that they are exclusive to the R1b population in Iberia and don’t appear before.

  43. @Matt & Frank

    Re: the African “signal” in Mycenaean, I’m more inclined to think it’s mostly caused by that I9033 noisy sample. If it was included in the stats, it might be generating that noise. In G25 it takes a good 4% Yoruba:

    Kura-Araxes_Kalavan 28.2%
    Tiszapolgar_ECA 24.6%
    Minoan_Lasithi 18.7%
    Protoboleraz_LCA 13.6%
    Varna_o 10.8%
    Yoruba 4%
    Vinca_MN 0.1%
    Yamnaya_Samara 0%
    Peloponnese_N 0%
    Greece_N 0%
    Anatolia_EBA_Isparta 0%

    Distance 3.4217%

    Well, unless that’s not noise but real admixture. But would be strange (and the sample behaves erratically anyway).

  44. @Alberto

    Great, thanks!

    My motivation for replacing Yamanya is the following:

    – Short glimpse of Greece neolithic sample I6423 in the Wang/Caucasus pre-paper. It proved that a EHG shifted population was already present there before 3200 BC. This date is too early to be related with the mobility/warfare related Yamnaya expansion.
    Hence I6423-like Greeks are most likely the result of such EEF-EHG mixes gradually coming from the north but without Yamnaya as driving force or CHG/Iran admixture.
    Yamnaya, proper-Steppe admixture could have very well played a role at later dates and for the formation of Mycenaeans. But this earlier alternative can’t be excluded as it would inflate potential share of Yamnaya/Steppe admixture. I’m aware of the presence of Yamnaya origin populations just north of Greece starting by 2800 BC. But I think Yamnaya influence on Myceaneans was much smaller and is too inflated if we allow Yamnaya but exclude something Varna_o like (the algorithm can’t distinguish how much of which is the truth).

    – Another motivation is due to the quite few aDNA samples we have by now. Y-DNA wise we have just seen J, almost certainly J2a. Up until R1 has shown up in an ancient Greek elite grave, we have to consider a non-Yamnaya Steppe source for the elevated EHG levels.

    – Minoans were a insular population, unlike Mycenaeans that came from the mainland. Hence they might have just skipped contact with a I6423-like population of mainland Greece. The CHG/Iran source or even the earlier ANF(EEF) farmers might have got there via sea, avoiding contact to incoming EHG shifted populations from the north (mainland).

    In total we have to be careful with the Steppe admixture concept because it is basically a certain ratio of EHG and CHG/Iran for current tools.
    This case is and example for the creation of a pseudo-Steppe signal, mimicking it but completely different historically.

    We certainly should not judge a better explanation due to minimally better fits. I’m sure that the different ratios of those 4 sample candidates I mentioned would create better and worse results.

    Your results a very interesting, as well as your offered alternatives. Changing Hajji_Firuz for Kura-Araxes_Kalavan is a good idea.
    Adding Yamanya-Samara does not cause Varna_o to be declined and basically half of the ancient Greek samples select one of the two. None is clearly favored over the other.
    I would favor to keep the sources as low as the relevant auDNA components, as long as the difference in fits is minimal.
    Here my Maykop and Hajji-Firuz idea could be simplified by just Maykop_Novosvobodnaya I guess.

    The results are amazing considering that 4-5 sources are sufficient to avoid the potential Steppe admixture in ancient Greeks, offering an alternative.

    Thanks again.

  45. @ Frank

    “Palmela points weren’t part of the classical Central European BB package – here, other symbols such as artefacts made from boar tusks, as already present in GAC and even the Ukrainian Neolithic, played a much more prominent role. I was actually surprised to see a Palmela point in the linked photo above of an Elbe-Saale BB burial assembly. The form is alien to TRB/GAC/CWC/Single Grave traditions and speaks for a more south-western origin, and bronze-making was certainly higher developed in Iberia during the time in question.”

    It has become axiomatic to view that Palmela points are Iberian. However I’ve never seen convincing proof of them ever being there before BB.
    Much like the eponymous pottery itself ..:)

  46. @Alberto, yeah, I think naively dropping that raw f3 data into nMonte might present some problems because of the sheer number of columns, which would be mostly redundant data (with compression by PCA or some other method) so would need some processing to compress into a nMonte-safe number of variables or selecting and just cutting down on “outgroups”.

    On that tangent I don’t think Principal Coordinates Analysis a la the Fst Matrix really works with this data because it’s a similarity matrix where I believe there is the (possibly dubious IMO) step of putting self similarity as 1, and this causes all sorts of problems with that.

    At the moment I have just been trying to use Correspondence Analysis on it, but generally too lazy for nMonte recently as the numbers of populations have ballooned and many more close references emerge that are hard to distinguish ;).

    Just to clarify, I’m not totally certain or fixed on the level of Steppe ancestry either, the main interesting thing to me about the Mycenaeans in that data sheet was that the either Minoan+Steppe related / Armenia_MLBA+Aegean_N dichotomy from Lazaridis 2017 (one model or the other) did not seem to hold, and Mycenaean would seem need extra CHG in relation to either of those pairs. Not either/or.

    I’ll bear the note around I9033 in mind in case Davidski is interested in trying to the resolve the odd African attraction of the Mycenaean samples in his datasheet.

  47. Rob, Alberto:
    The notion of the Iberian origin of Palmela Points seems to go back to Harrison 1974,
    And while it is true that Palmela Points only appear around 2400, Harrison’s analysis that their metalurgy, a standardised arsenic bronze, sets forth the tradition of pre-BB Vilanova de Sao Pedro I (VLSP I) seems still valid.

    Another Iberian element of the “BB package” may be wrist guards. Several studies have questioned their practicality for said purpose and suggested they might rather have been blade sharpeners fastened to belts. Such sharpeners, albeit unperforated, seem to be known from pre-BB Western Iberia (can’t find the source at the moment).

    As to the Beakers pots themselves – they are so deeply rooted in NC European traditions going back to TRB that I have no doubt about their origin – whereby I would add the Weser-Elbe triangle and the Lower-Middle Elbe that have also provided early “quasi-BB” to the commonplace Rhenish Beakers.

    There is also good reason to assume a Single Grave origin for the typical BB barbed arrowheads, see (paywalled)

    Heyd has in several publications identified two more types of artefacts that are typical especially for Central European BB, namely disk-shaped bone pendants and amber beads (the latter have also occasionally been found in the Basque Country). Both can be credibly linked to GAC traditions, whereby the bone pendants can be traced back further to CT, Skelya, Khvalynsk, Svobodnae etc., and ultimately to Mesolithic NE Europe (e.g. Oleni Ostrov).

    As such, the “Beaker Package”, with its regionally varying composition, can be interpreted as assemblage of Iberian, NW European and Baltic traditions along a developing pan-European trade network, possibly with a strong maritime element. CA Iberia is also connected through and across the Mediterranean, as evidenced by “exotic” finds such as ostrich eggs and ivory from both African and Asian elephants.

    Obviously, the show was partly run by sailors from the Lower Rhine (also the Lower Elbe / Jutland? Baltic amber must have passed through there on its way to the Basque Country..). However, at least culturally, this wasn’t a one-way road.

    More in my next comment…

  48. @Patamares

    Yes, true that there is no strong preference of one source over the other (Yamnaya vs. Varna_o), and that the models with less sources that are almost as good as ones with more are preferable. So there’s that.

    But I think we need to see what’s up with that mysterious I6423 sample from Greece. And we’ll need more samples from that time period to check their real impact. It’s all still too sketchy in the Balkans for such an important and complex area.

  49. @Matt

    Yes, I noted that your main point was about Mycenaean samples showing in that F3 matrix too (like G25) that they required extra CHG-related admixture from the one provided by Minoan + Steppe_EMBA. My thanks was for corroborating that with those formal stats (since formal stats are used in most publications I hope they can publish a better model next time, though with more samples that would come along the models it might become less necessary.

    Still couldn’t look at the f3 matrix itself. Will report at some point if I find a useful way of using it.

  50. @ Alberto & Patamares

    I agree that Italy and the Balkans are still poorly represented at present
    Undoubtedly they’ll follow different trajectories to northern and Western Europe; and putting it all back together will be a must
    So it’ll either confirm things or mean back to the drawing board

  51. Continuing my previous comment:
    An intriguing aspect of the BB phenomenon is that it transformed CWC/Single Grave at least as radical as Iberia. Daggers, as other lanceotic forms such as arrowheads, were essentially unknown in CWC/Single Grave. During the 24th cBC, however, they quickly and massively appear in S. Scandinavia and the Middle-Elbe Saale (MES) region, within a century completely replacing flint (battle) axes as status symbol in male burials. This replacement was accompanied by other radical changes, e.g. a massive increase of average house sizes in Denmark and Schleswig-Holstein from 40-80 m² before ca. 2,400 BC to 160-400 m² by 2,000 BC. That’s the emergence of the typical Germanic/Saxon longhouse that continued to be built in the North European plain up to the mid 20th century AD and is a/o evidenced for the Wielbark Culture (Vistula Goths).
    For the radicality of the change, in Danish nomenclature the Single Grave culture ends at 2,350 BC; the subsequent phase, up to the Nordic BA, is termed Flint Dagger Culture. In Schleswig-Holstein nomenclature, this corresponds to the shift from Late to Final Neolithic. For details, see J. Müller (2014) linked below

    The situation was more or less comparable in the Netherlands. Some imported Grand Pressigny daggers are known from Single Grave burials, but here as well, Battle Axes predominated, and there doesn’t seem to have been any “local” dagger tradition. Here, as well, the shift from battle axes to flint daggers (mostly imported from Denmark and North Germany including Helgoland) as predominant grave good is taken to mark the transition from Single Grave to Bell Beakers. [The pottery itself wasn’t useful in this respect, since BB-like pottery is already present in early Dutch and Danish Single Grave contexts.]

    In MES, the arrival of BB corresponds to the emergence of tin-bronze – some 300 years earlier than in the remainder of Central Europe. J. Müller (see above) regards this as a sign of “early local experimentation”. However, over these 300 years, tin-bronzes are exclusively found in burials, and even the earliest finds already have 3-4% tin content. When tin-bronzes also start to appear in hoards during early Unetice, they are initially of a far lower quality (tin content), and technologically resemble those from contemporary sites in other parts of Central Europe. This indicates to me that the early MES tin-bronzes were imports, probably arriving together with their bearers.

    IOW: Archeology suggests a substantial foreign element in the genesis of Dutch and MES BB, and the North German/ South Scandinavian Final Neolithic (Flint Dagger Culture). We know that Central European BB have more EEF ancestry than CWC. The common (or at least my) assumption so far had been that the additional EEF was acquired locally, e.g. from post-GAC cultures like the Schönfeld Group around Magdeburg, or the Elbe-Havel Culture around Berlin that had hardly been affected by the CWC phenomenon. But what if the additional EEF arrived from much further afar, including possibly Iberia? In this case, the Olalde e.a. findings, which are essentially based on anachronistic comparison of mid 3rd mBC Iberian with late 3rd mBC Central European aDNA, would require re-interpretation.

  52. Against this background, may I ask you, Alberto and Rob, or anybody else with time for it, for some G25 models?

    1. Danish Single Grave – Flint Dagger Culture transition
    We only have one Single Grave Culture sample so far, namely RISE61, 2851-2492 BC from Zealand. Even that sample needs to be taken with some caution, since Zealand is described as a late holdout of TRB-N. With that in mind, could someone run a first comparison of RISE61 to RISE71, 2196-2023 BC (Flint Dagger Culture) from N. Jutland. I initially recommend a distal set-up according to the standard WHG/EEF/Steppe_EBA model (maybe two WHG sources, KO1 and El Miron-like Iberian) for both to find out if there are notable differences. If so, these differences might be explored further along the lines sketched below.

    2. MES BB
    Targets: BB Quedlinburg (I0112, I0113, I0806, I0805), Benzingerode-Heimburg (I0059, I0058, I0171, I1546, I1549). Since both sites lie just some 10km apart and are more-less contemporary, they may probably be pooled together.

    Rothenschirmbach (I0060, I0108, I0111): Sometimes erraneously treated as CWC, but archeologically the site was clearly BB. Some 60 km afar from Quedlinburg. I0111 deserves special attention, as the burial was marked by a Menhir.

    a) MES CWC: Here, it already gets difficult. Some of the samples originally listed as CWC might actually reflect later epochs (BB, Unetice). Clearly CWC are the samples from Esperstedt (I0104, I0103, I0049, I0106, I1532, I1534, I1536, I1538, I1539, I1540, I1541, I1544)
    Seperately, Karsdorf I0550 should be considered – it is the only available sample that pre-dates BB entry into MES, and as such probably better represents the original CWC MES genetic structure than the Esperstedt samples.

    b.) Possible local EEF contribution: Such contribution could have come especially from the Schönfeld Group around Magdeburg that has delivered AMS dates as late as the 24th cBC, and set forth GAC-West/ Bernburg traditions. In the absence of Schönfeld Group aDNA, I propose triangulation from the following sources:
    – Baalberge
    – Esperstedt_MN
    – GAC Poland
    -TRB_N (preferably Saxtorp5164, which is more WHG-shifted than Gökhem and as such possibly better represents TRB-Tiefstich influence)
    – Blätterhöhle (optional)

    c. Possible distant EEF contribution:
    – Iberia early CA (pre 2,500 BC, ideally from the Atlantic coast);
    – France MLN, to at least somewhat emulate the Dutch Single Grave connection to Grand Pressigny;
    – Remedello (daggers! ) – in fact, Mondsee and/or Cham seem a more likely source, but Remedello is virtually the only Chalcolithic aDNA available from around the Alps;
    – the Iceman, as another Alpine CA sample;
    – Vucedol, separating EEF-like I2792 and steppe-admixed I1475/I3492 (Vucedol shows manifold archeological connections to Mondsee)

  53. @FrankN

    Sorry for the late reply. I couldn’t get to it until now.

    The cultural change from CWC to BBC is a very interesting phenomenon. Which together with the strict separation on the male lineages (despite the otherwise very similar autosomal components) makes it rather mysterious.

    Unfortunately several of the samples you mention are not available. Notably:

    1. RISE61 and RISE71, so can’t test this one at all.

    2. – Several of the targets, but at least all groups represented.
    – Sources: Karsdorf I0550 and Iceman.

    Now for the results:


    Will let you comment about them. Let me know if further tests might be needed.

  54. There’s a new pre-print out:

    The Arrival of Steppe and Iranian Related Ancestry in the Islands of the Western Mediterranean

    A series of studies have documented how Steppe pastoralist-related ancestry reached central Europe by at least 2500 BCE, while Iranian farmer-related ancestry was present in Aegean Europe by at least 1900 BCE. However, the spread of these ancestries into the western Mediterranean where they have contributed to many populations living today remains poorly understood. We generated genome-wide ancient DNA from the Balearic Islands, Sicily, and Sardinia, increasing the number of individuals with reported data from these islands from 3 to 52. We obtained data from the oldest skeleton excavated from the Balearic islands (dating to ~2400 BCE), and show that this individual had substantial Steppe pastoralist-derived ancestry; however, later Balearic individuals had less Steppe heritage reflecting geographic heterogeneity or immigration from groups with more European first farmer-related ancestry. In Sicily, Steppe pastoralist ancestry arrived by ~2200 BCE and likely came at least in part from Spain as it was associated with Iberian-specific Y chromosomes. In Sicily, Iranian-related ancestry also arrived by the Middle Bronze Age, thus revealing that this ancestry type, which was ubiquitous in the Aegean by this time, also spread further west prior to the classical period of Greek expansion. In Sardinia, we find no evidence of either eastern ancestry type in the Nuragic Bronze Age, but show that Iranian-related ancestry arrived by at least ~300 BCE and Steppe ancestry arrived by ~300 CE, joined at that time or later by North African ancestry. These results falsify the view that the people of Sardinia are isolated descendants of Europe’s first farmers. Instead, our results show that the island’s admixture history since the Bronze Age is as complex as that in many other parts of Europe.

    Fernandes et al. 2019

    It’s starting to get busy. I hope we can keep up.

  55. Alberto: The Sardinian stuff looks interesting – will need to dig further into it. An immediate takeaway, however, is that the Beaker period in Sardinia apparently didn’t substantially change that island’s genetic make-up.

    Now to your models: I found it convenient to sum up the various individual sources by main group – it may also save time-pressed readers a look into your file. I have re-sorted the samples in order to better point out trends.

    I0111 – BB Rothenschirmbach
    57.65% CWC Esperstedt
    11.05% Iberia_ChL:I3276
    9.25% Vucedol:I3499
    9.20% Blätterhöhle I1563 (eastern-shifted)
    9.10% France_MLN:I4303
    3.75% Esperstedt_MN

    Distance 2.3621%

    That’s the “big boss” buried below a Menhir. The guy in the mid-20s, with a golden hair ring (“Levkas ring”, otherwise found between Greece and Croatia), buried right next to him may well have been his son.
    Obviously a quite “international” background. The Iberia_CA/ France_MLN / Blätterhöhle combination might speculatively approximate Eastern France (post-SOM), but we can’t be sure without samples from that area,

    I0108 – BB Rothenschirmbach
    59.25% CWC Esperstedt
    33.10% Iberia_ChL:I1838
    7.65% Blätterhöhle I1593

    Distance 2.0807%

    I1593 is the most “western” Blätterhöhle sample, i.e. WHG is purely Loschbour, no KO1/ Baltic_HG present. As there is nothing Eastern (GAC) in here, this Blätterhöhle admixture looks like picked up on the way from Iberia to the Saale, somewhere in Eastern France, the Low Countries or Westphalia.

    I0113 – BB Quedlinburg
    60.40% CWC Esperstedt
    28.85% Iberia_CA (no Steppe)
    9.70% Remedello_BA:RISE489
    1.05% France_MLN:I4303

    Distance 2.2543%

    Holy shit ! 40% West Mediterranean – that’s far more than I had expected.

    I0112 – BB Quedlinburg
    75.55% CWC Esperstedt
    10.65% Blätterhöhle I1593
    10.45% GAC Poland
    3.95% Vucedol:I3499

    Distance 1.8732%

    The Vucedol component comes as a bit of surprise (or not, when considering I0111). The Blätterhöhle-GAC Poland admix is how a post GAC West population, e.g. the Schönfeld Group (some traces of which have actually been found at Quedlinburg) might be expected to have looked like.

    I0805 – BB Quedlinburg, yDNA R1b1a2 (M269)
    77.10% CWC Esperstedt
    12.95% Esperstedt_MN
    5.35% Blätterhöhle I1563 (eastern-shifted)
    2.55% GAC Poland
    1.45% Vucedol:I3499
    0.60% Iberia_North_CA:I1978

    Distance 2.3415%

    Again, mostly local MN admix. If M269 was from Spain, it must have arrived 7-8 generations earlier, i.e. latest around 2,500 BC. Blätterhöhle is more proximate….

    In summary: Three of the five MES BB samples carry around 40% non-local admix each, most of which seems to be of Franco-Iberian origin. The other two look like 75% CWC, and 25% local surviving GAC/Bernburg, whereby it is uncertain how much of that local MN adstrate was already absorbed before by the comparatively late Esperstedt CWC samples.
    Moreover, as Rothenschirmbach lies just some 5 km W of Esperstedt, Esperstedt CWC may well incorporate some of the non-local ancestry that is clearly visible in Rothenschirmbach BB. As such, the modelling is likely to rather under- than overestimate the foreign, mostly Franco-Iberian element in the genesis of MES BB.

    Roughly, the five MES BB samples contain in average 17% Iberia_CA plus France_MLN ancestry. No wonder that German BB do so well in approximating Iberia_BA. When discussing BA population replacement in Iberia, the Franco-Iberian genetic (and also cultural, see my previous comments) influence on CE BB need to be considered.

    Rob: If you have the time, you may want to conduct a similar exercise on Bavarian BB (of course using Franconian/ Bavarian CWC such as RISE446 as source). I am curious how much Vucedol, but also Remedello influence you ‘ll be able to identify there…

  56. Frank,
    I think it’s worth waiting for the next few papers to clarify BB in central Europe, however a brief look might propose:

    Beaker_Hungary_o 54.7%
    Globular_Amphora 27.4%
    Blatterhole_MN 8.6%
    Maykop_Novosvobodnaya 6.2%
    Baltic_HG:Spiginas4 3.1%
    Baden_LCA 0%
    Iberia_North_CA 0%
    Remedello_BA 0%
    CWC_Germany 0%

    Distance 2.5577%

  57. While the W. Mediterranean influence on MES BB (and by extention, most likely also Dutch_BB and Danish Flint Dagger Culture) isn’t surprising, it can’t explain the early start of tin bronze metalurgy in MES. In Iberia, in spite of sizeable local tin deposits, arsenic copper dominated until well into the LBA. The same applies to Remedello and North Alpine (Mondsee etc.) metalurgists. Occasional finds of tin bronze are known from Vucedol. However, they seem to be imports – also Vucedol predominantly produced arsenic bronzes. Moreover, being famous for its copper axes, Vucedol is a rather unlikely candidate when it comes to having triggered the shift from axes to daggers as male status symbol that sweeped through Middle/ Northern Germany and South Scandinavia during the 24th mBC.

    Tin bronzes are generally believed to have been invented in SE Anatolia, from where they spread through Central Anatolia to the Troad and the North Aegaean. Some early specimens (1st half of 3rd mBC) are also known from the Armenian plateau. However, for scarcity of local tin ressources, tin bronze production in the Caucasus only picked up in the late 3rd and 2nd mBC, when tin from South Iran and Central Asia became available.

    As such, when looking for an external stimulus to MES bronze-making, eyes should turn to the SW Pontic area. Inspired by Patamares’ idea of a „pseudo-Yamnaya“ independent admix of CHG and EHG (see also Rob’s Spiginas/Maykop model), I thus want to ask you, Alberto, for another modelling run on MES BB with the following sources added:

    a.) Narva Lithuania, e.g. Donkalnis6, Spiginas1, motivated by the presence of Baltic amber in the Central European „Beaker package“. Alternatively, or in addition, you may also try Lithuanian CW (Gyvakarai1, Plinkaigalis241/242; but not Spiginas2, which is chronologically too late).

    b.) Anat_CA I1584 – not ideal, but the best approximation of the early 3rd mBC Troad available so far; alternatively, you may give Minoans a try.

    c.) “BA Bulgaria” I2510 (early 3rd mBC, chronologically still CA) – motivation as above, and the Trnovo area the sample is from has delivered golden hair rings similar to the one found in Rothenschirmbach.

    d.) Yamnaya Bulgaria (you never know…), and
    e.) Maykop Novosv.

    Of course, unused sources from the first run can be removed, and pooling all CWC Esperstedt samples might facilitate analysis and interpretation.

    @Rob: „I think it’s worth waiting for the next few papers to clarify BB in central Europe
    Yes and no. CE BB seem to represent a new elite – at least they have substantially altered CWC/ Single Grave status symbols. If that elite has strong Franco-Iberian roots, this would of course add a completely new dimension to the ongoing linguistic debate. We would have to start thinking about the possibility that BBs not spoke IE, as commonly assumed, but instead some kind of EEF-derived para/proto Vasco-Iberian.

    Otherwise, I don’t think it is adequate trying to model BB MES with BB Hungary, since BB’s arrival in Hungary was comparatively late. When trying to trace Pannonian impact on the emergence of the BB phenomenon, Vucedol should be the source to select.

  58. Hhm Frank. IMO it’s pretty clear by now that R1b-L51 isn’t from France ; and BB Hungary has some of the earliest BB dates

  59. Otherwise, I have just realised that BB suffered from a plague epidemy – one BB from Augsburg ( 4.203 cal BP~2.250 cal BC) has been tested positively for Y. pestis bacteria.

    I hope the Reich Lab will make available their Iberian and British BB/EBA samples to the MPI Jena team so they can as well be checked for presence of the plague. In any case, the plague would be a plausible explanation for the massive genetic and cultural shift observed after ca. 2,200 BC. Of course, if BB was associated with a major epidemic, survivors would quickly do away with respective symbols (pots, daggers etc.), which is exactly what can be observed in El Argar and Unetice alike.

    The earliest European case of plague documented in the study linked above is from Yamnaya N. Caucasus (RK1001 sample in Wang e.a. 2018), and steppe ancestry appears to involuntarily but effectively have cleared its path into Central Europe by spreading the disease.

    Susceptibility to the plague is genetically determined. as evidenced by several rodents that are immune to the disease. The mechanisms aren’t yet fully understood – they relate a/o to gene mutations affecting the way how toll-like receptors (TLR) recognise infections and stimulate immune defense, but also to some genes that control metabolism. These mutations prevent plague bacteria from “highjacking” the lymphatic system and turning leucocytes, which in principle should fight diseases, into a vector of their spread.

    mtDNA codes a good part of genes related to metabolism and immune defense, and there is ample evidence for correlation of mtDNA with susceptibility to or immunity against certain diseases, e.g. HIV/ AIDS. Now, there is indication of certain mtDNA, e.g. mtDNA I, being over proportion represented in medieval Black Death burial grounds (London, Lübeck), and having in England and Norway declined in frequency over the last millenium. For mtDNA H, the opposite trend is apparent.

    However, for HIV/AIDS it has been shown that the highest immunity is achieved when mutations are homozygous, i.e also inherited from the male side. Early and regular exposure of steppe populations to the plague implies selective pressure on males in favour of such mutations, an immunity that EEF and WHG males would have lacked. As such, one may speculate that some R1 males had acquired gene mutations that increased their resistance against the plague, a feature that would have become most effective when coupled with certain mtDNA, say mtDNA H.

    Inspired by maps like
    I have been looking a bit into the geographical pattern of Black Death mortality. Except for England, there isn’t any comprehensive analysis available, one needs to go into individual chronicles and historical texts. With that said, certain areas (partly not reflected in the map above) stand out as hardly or under proportion affected:
    – Lombardy (Milan),
    -the Basque Country,
    – Wales/ Western England,
    – W. Flanders (Bruges),
    – Central Europe from Nuremberg via Prague and Cracow to Warszaw

    The Milan and Bruges cases seem somewhat special, but the remainder obviously relates to R1-heavy regions.

    Over proportion hit (50-70% mortality) were
    – Tuscany (Florence)
    – the Lower Rhone (Avignon)
    – the North Sea basin (East Anglia, London, Cologne, Hamburg, Lübeck, S. Norway)
    As concerns the latter, the Hamburg (>50%) – Magdeburg (30%) – Nuremberg (10%) transsect is informative, as it mirrors yDNA I frequencies across Germany.

    In short – alternatively to all those “kill the men, rape the women” scenarios currently en vogue among certain blog commentators, the possibility that the rise of yDNA R1(b) is due to improved resistance against the plague should be given serious consideration.

  60. Rob: Note that in the plague study linked above, the Augsburg BB was found to be infected with the Baltic CWC strain (Kunilal II, Estonia), not the Vucedol strain (GEN72). Plus, as I have said, I can’t see how axe-focused Vucedol could have triggered the shift towards daggers as male status symbol.

    As to anything beyond R1b-M269, I propose to await the Swiss data that is hopefully coming soon. The Alps were neolithicised comparatively late and might have hosted Vilabruna-derived HGs. You are certainly aware that France has its share of the Western Alps….
    A bit of SOM aDNA could IMO also do no harm – might be similar to Blätterhöhle (the burial style certainly was), which has yielded R1b.

  61. @FrankN

    I didn’t know about those early dates of tin bronze in MES BB burials. Otherwise, yes, tin bronzes only start to spread throughout Europe only from 2000 BC. The route seems to be clearly from Anatolia to the Balkans (the Balkans being the only region in Europe I knew about having it before 2000 BCE). For what you say about the quality, I would also think that they were imports.

    Regardless, I can’t see Iberia as a source of any sort of bronze metallurgy. There it arrives with the steppe admixed/R1b-M269 population, so this population must have got it somewhere else before moving into Iberia. I’d also note that it’s arsenical bronze exclusively in the early dates. Tin bronze only appears after 2000 BC (I’d need to check for earliest date, but could be closer to 1800 BC).

    Another thing to consider is that the dates from the earliest R1b/steppe admixed samples in Iberia are pretty much contemporary with the Central European ones, if I’m not mistaken (would need to double check the earliest ones in CE to make sure), which apart from showing a fast spread could also mean that neither were the source, but rather we might need to find a slightly earlier source somewhere from where it could have moved almost simultaneously to MES and Iberia. But so far the sampling didn’t catch something significant in that respect.

    I’m glad that the possible role of disease is being actively researched, because I also see it hard to believe that newcomers from areas with low population densities, pastoralists (pastoralism cannot really sustain a large population. Meat quickly becomes a privilege of the rich as population increases) would be able to replace local and settled populations with violence – not at a time when the numbers made the difference. (I usually don’t like to compare disparate cases, like the arrival of Europeans to America with LN/BA events, but now that it’s confirmed that the Salmonella strain that literally massacred the Aztec population arrived from Europe, the role of disease is something to keep high on the list of possible causes for these replacements in Europe’s LN/BA).

    Anyway, back to your request, I added the suggested sources and, in short, the main change is the very significant presence of CWC_baltic_early in 3 out of 5 of the MES BB samples (27.5% average in them, this at the expense of CWC_Germany). Together with some 3-4% Narva_Lithuania in 4 out of 5 samples, it suggests a northern origin (or at least connections?) of the steppe side of those MES BBs. Other than that, it’s only the 5.85% Maykop_Novosvobodnaya in one sample (I0111) that changes.

    Full output

  62. Alberto: Thx for the new models. Actually, CWC_Baltic isn’t only driving down CWC Esperstedt, but also increases the WestMed admix. This might seem like a technical artefact: North-eastern sources are balanced out by south-western ones. However, as the fit improves, sometimes quite substantially, I think this is more than just an artefact but instead reflects actual movement. The pivotal role of the E. Baltics within CWC has already been demonstrated by Furholt 2014, whose network analysis (Fig. 7) identified two major, interconnected nodes – the Netherlands and Lithuania – each of which assembled its own, regional periphery. Lithuania’s role in this network of course relates to amber.

    I think it makes sense to re-publish here your latest model for the “big boss”, since it comes out quite different from the previous, and also from Rob’s model:

    I0111 – BB Rothenschirmbach
    33.35% CWC_Baltic_early:Plinkaigalis
    24.00% France_MLN:I4303
    19.40% CWC Esperstedt
    7.80% Sweden_N:Saxtorp5164
    6.10% Esperstedt_MN
    5.85% Maykop_Novosvobodnaya
    3.50% Narva_Lithuania

    Distance 1.961%

    Let me conflate the groups somewhat more:
    44.65% Lithuania/Sweden
    5.85% Maykop_Novosvobodnaya
    25.5% Esperstedt (neighbour village)
    24.00% France_MLN:I4303

    So – one Baltic parent that seems to have had one Maykop and one S. Swedish grandparent, and the other parent half local, half Southern French. I bet the family was active in long-distance trade. This forebodes the LBA „Danish princess“, which, according to isotope analysis, originated from the Black Forest and annually migrated there, presumably to ship copper to Denmark, and the La Tene long distance marital alliances involved e.g. in trading Iberian/ S. French wine to Manching.

  63. As to Iberia and tin bronze – this is what Llorens/ Montero Ruiz have to say:

    The earliest tin bronzes of the Iberian Peninsula were found in Bauma del Serrat del Pont (Tortellà, Girona), a settlement that was inhabited for a long time span. An arrowhead and metal waste made of low tin bronze were unearthed in Chalcolithic strata with Bell Beaker ceramics dated to 4200 ± 70 BP (Beta-90622) and 4020 ± 100 BP (Beta-64939), respectively
    2800–2450 (2σ) cal BC. Evidence for tin bronze production (smelting crucibles, metal waste, objects) is more numerous at this site in the following Early Bronze Age.
    Tin bronze technology moved southwards very slowly, reaching the area of the El Argar culture in the Southeast, not earlier than the 19th century cal BC. It is thought that this new technology arrived in Iberia from afar, probably from southern France.

    Since there is some tin found in the Massif Central, Southern France maybe isn’t such a bad guess. Brittany has quite a lot of tin, which was exploited during the BA. However, both regions aren’t exactly famous when it comes to early tin bronze. The best bet IMO is still the E. Mediterranean, which a/o supplied ivory to Iberia during the period in question.

  64. @ Frank
    I think you might be reading the nMonte results a tad too literally. For such a reconstruction, we’d need 2 -3 generations of individuals samples, with isotopic , uniparental and genetic data to even begin to approach a proposal.
    IMO, a Baltic- Pyrenean model cannot account for the ubiquity of R1b-L51 in BB.

  65. Since the discussion about Olalde 2019 is still going on here, I cooked up the following graphics in PAST3 which may be a complement to those in the paper for those who are interested in looking at the individual samples.

    1. – Based on the Table S15 from the supplement. Similar to Figure 2B from the paper, however only a subset of the individuals from that figure have data in Table S15 (I don’t what the reason for this is, whether only a subset of the individuals in Fig 2b were run for individual Germany_Beaker proportions or what). The potentially useful element is that I’ve put on the individual sample IDs.

    2. – Taking the samples from Olalde’s Table S15 present in the Global 25 data (only some of them, prob. due to quality thresholds) and reprocessing them through a PCA with Beaker_Bavaria and Beaker_CA, to cross compare. Position along the main X axis should correlate with Germany_Beaker ancestry from 1. Variation in the Y is mostly residual leftover WHG:Anatolian I think.

    3. – as 2, adding Beaker_Iberia and BA samples not present in Olalde 2019’s Table S19

    4. – as 3, introducing all other Iberian samples in Olalde 2019 from the 2500 – 2000 BCE interval (except I4246) who are present in G25. There’s only about 3 of them.

    Alberto, as I think you’ve noted before back in March 2018, I4229 I2a1a1 male may have slight enrichment of a very low level of Steppe related ancestry, and he is perhaps joined now by I3485 female.

    The released Fig 2B in the published paper seems to be more like one of the two versions of the same figure Reich presented at presentations before it’s release, the one with more compression around the Iberia_CA set –

  66. Regarding my last post, I would also say to note I3485 (30-40-year-old female) who is assigned C_Iberia_CA (no steppe)in the paper is in the same “tumba” in a multiple burial (the only one found at Castillejo del Bonete site) with I3484, a 40-50 year old male with low quality ancient dna (69982 snps), but who is nonetheless assigned to C_Iberia_CA_Stp (35.6% Germany Beaker in the supplement, lowest of the pre-2000 BCE steppe admixed Iberians, and so probably about 15% steppe ancestry) and R1b1 of some kind.

    This seems to me (and this may be archaeologically naive) to circumstantially support that the signal of low level steppe ancestry she has in G25 is probably real and she was from somewhere within a network of people with steppe ancestry. Simply on eyeballing, I would peg her at about 15% Germany_Beaker given PCA positions, so 7% steppe ancestry and fairly similar to the male relative to some of the other samples about at the time.

    Sentimentally, I wonder if they were husband and wife….?

    edit: This is of course the site that is described in the paper in following terms: “These patterns point to a higher contribution of incoming males than females, also supported by a lower proportion of nonlocal ancestry ancestry on the X-chromosome (table S14 and fig. S7), a paradigm that can be exemplified by a Bronze Age tomb from Castillejo del Bonete containing a male with Steppe ancestry and a female with ancestry similar to Copper Age Iberians.”. (So it is interesting for the interpretation if she has Steppe ancestry too, at a lower clip than him, rather than none at all.) Also, thanks Frank.

  67. Matt: Good stuff!

    Some takeaways (numbering acc. to your graphics):
    1a) BB_Germany ancestry seems to decline from ca 0.65 in CA(steppe) to ca. 0.4 in Iberia_BA. This is of course somewhat comparing apples to oranges, since non-Steppe CA samples need to be taken into account as well. Nevertheless, it suggests an ongoing admixture process prior to and into the BA.

    1b) BB_Germany ancestry appears to have initially followed a N to S cline. being higher in N, NE and C. Iberia than in SW/SE. During the MBA (after ca. 1800 BC), some homogenisation seems to have taken place, w. BB_Germany ancestry rising in SW Iberia, but decreasing in N. Iberia.

    1c.) Against this trend, SE Iberia (El Argar) shows a MBA decrease in BB_Germany ancestry. This might indicate arrival of East Mediterranean populations in connection with the “orientalisation” of El Argar II (pithos burials etc.).

    4 a.) While pretty diverse as a population, BB_Bavaria hardly overlap with Iberia_BB / Iberia_BA. I am uncertain whether they represent the actual source. How do French_BB (mostly from Alsace-Lorraine, IIRC, as such representing the hypothetical Rhine-Rhone axis “epicentre”), and Elbe-Saale_BB (Quedlinburg, Rothenschirmbach, see above) fare in this respect?
    4b.) There seems already some “northeastern pull” to be visible on certain “no steppe” individuals. S. France? Sardinia? Brittany? Neolithic Britain?

  68. @Matt

    Thanks for all those graphs. I still have in my to-do list to run all the Iberian BA samples to check for specific patterns, affinities,… Waiting to compile some extra bits of information about some sites and then I hope we can write about the BA part of the paper.

    Yes, interesting to see that I2a male with some steppe admixture. By the date it’s nothing surprising, but it seems the only one with that Y chromosome until the Celtiberian guy.

    Re: that other I3485 sample (the woman buried with a steppe admixed man), you’re right that it seems to have steppe admixture too:

    Iberia_North_CA 49.4%
    France_MLN 30.8%
    Beaker_Bavaria 18.8%
    Beaker_Mittelelbe-Saale 1%

    Distance 2.7722%

    I’d add this bit from the supps:

    Another good example is Tumba 4, the only multiple burial in this site with a 40-50-year-old male and a 30-40-year-old female who was buried with two ivory buttons and who had a marine diet, suggesting a non-local origin (74).

    Reinforcing that pattern of patrilocality seen in previous studies.

  69. Frank, can’t answer the question at 4. but here is the same graphics with Beaker_Mittelelbe_Saale and Beaker_France in case it is useful:

    Alberto, interesting to connect her non-local origin with patrilocality, I hadn’t thought about that aspect at all.


    Looking at the paper’s qpAdm methodology, to try and understand Fig 2B – where we see a strict dichotomy of no Beaker related ancestry in some samples either female or non-R1b-M269 , and above 16% levels in some individuals (above 30% in before Bronze Age), either combining with R1b-M269 or else female. So if I3485 looks in G25 like she has some German Beaker, why doesn’t it show up here?

    I may be reading it wrong but it seems to be the case that for the underlying data they didn’t actually run qpAdm on all their transect to come to the proportions that underlie the figure.

    Relevant section of supplement is:

    As I read it, it seems more like they used a classification procedure to split Iberia_CA into Iberia_CA and Iberia_CA_Stp, then *only* ran the qpAdm on Iberia_CA_Stp, together with IA and BA, using the samples they classified as Iberia_CA and some others pops (mainly Germany Beaker). (And which classification procedure they do not seem to detail in the paper or supplement – probably D-stats?). I3485 was presumably classed into Iberia_CA.

    I doubt that’s wholly a wrong distinction (likely mostly correct), but I don’t totally understand why they’d do it like that. Why not simply use Iberia_MN (which is pretty much like Iberia_CA and basically a clade with it) and then run qpAdm on all the Iberia_CA without needing to specify Iberia_CA_Stp? The problem with doing it as they did is that it seems like it could end up to some degree simply reflecting whatever cutoff procedure they use for defining Iberia_CA_Stp, which is opaque and I think unspecified in the paper.

    If so, (and hopefully I’m just wrong about their methods), it seems like there is a risk in particular of excluding samples in Iberia_CA that actually would show some steppe/German Beaker ancestry in qpAdm, but beneath their Iberia_CA_Stp threshold, which as far as I can tell from G25 may be potentially be the case for at least I4229 and I3485 (but many other samples in this interval have too few SNPs for G25). That seems like it probably matters because these 2500 BCE to 2000 BCE Copper Age samples are rare and even low levels of steppe/”Northern Beaker” ancestry would tell us something about how populations are interacting…

  70. Thanks for the image Matt

    Frank, there doesn’t seem to be any extra “East Med” in SE Iberia Bronze Age, in current data. Their position lies among the ANF-WHG cline axis; and its variation is a subset of preexisting LNCA gradient. The only difference is shift toward BB Central Europe

    In any case; the earliest tholoi appear c 2900 BC, in the pre-Beaker Copper age of south Iberia; and this seems to be a local development , out of display and competition. These are the guys with wealth, ostrich eggs & ivory; but they ain’t Beaker.

    And there’s the beauty of aDNA; not only does it show migrations previously denied; but they also negate migrations when previously assumed .

  71. @ Alberto & Matt

    If I were to approach the BB question for western Europe, I’d do iterative local modelling, e.g. as following (should be self-explanatory)

    Beaker_Bavaria 73.2%
    France_MLN 23.4%
    Blatterhole_MN 3.4%
    Armenia_EBA 0%
    Balkans_ChL 0%
    Anatolia_BA 0%
    Remedello_BA 0%

    Distance 1.8811%

    Iberia_Northwest_CA_Stp:I3238 >/i>
    Beaker_France 37.2%
    Beaker_France_South 31.6%
    Iberia_North_CA 20.1%
    France_MLN 11.1%
    d 3.48%

    ^NB 1: This individual mentioned above is R1b-M269 & has significant steppe ancestry. I gather he is labelled as ”Iberia_CA north” because the cave from which he was discovered appears to have been looted, removing the BB gifts & disturbing the antrhopological context. If this was a fresh excavation, I’d bet that despite the cave burial, he’d be buried in usual BB fashion
    NB2: poorer fit due to coverage, or we missing something ? Adding BB_Netherlands, BB Germany, etc didnt improve things.

    Moving south;
    Beaker_North_Iberia 61.5%
    Iberia_Southeast_CA 38.5%
    Distance 1.3558%

    And that’s pretty much it.
    What remains to be seen is how German BB emerged.

  72. Rob:
    And that’s pretty much it.
    Not quite. What catches the eye in the new graphics (thx, Matt) is that I3188 from Annecy clusters amidst Iberia_BA. He is especially close to I6618 from the eastern outskirts of Madrid, datewise pre-Cogotas / early Cogotas I.

    Otherwise, we need to include Britain in the picture. Copper mining in Ross Island, Ireland seems to have been initiated from Iberia already during the CA, possibly as early as 2,500 BC. Cornish tin seems to have been exploited from at least 2,100 BC onwards. Since Iberians had little use for tin at that time and still prefered arsenic bronze, the impulse most likely came from Unetice, i.e. MES. The gold on the Nebra sky disc is definitely from Cornwall. It is also interesting that, after some 700 years of hiatus, the rondel (henge) tradition in MES lived up again around 2,300 BC, contemporary with the erection of Stonehenge.
    There are furthermore intriguing parallels between Schönfelder bowls (cremation urns), Irish EBA “Bowl Food Vessels”, and Ciempozuelos pottery from Central Iberia. The Schönfeld (Quedlinburg) – Irish (Rathlin) connection is aDNA-wise attested, but on the Irish-Iberian connection, a bit more modelling couldn’t do harm.. (Fig. 73)

    Last but not least, I believe that we will not be able to fully understand BB (and also Megalithism before it) before we get a decent chunk of aDNA from Brittany.

    Frank, there doesn’t seem to be any extra “East Med” in SE Iberia Bronze Age, in current data. Their position lies among the ANF-WHG cline axis; and its variation is a subset of preexisting LNCA gradient. “
    Where do you take this information from? The samples in question haven’t been included in Dave’s G25 set and therefore also don’t appear in Matt’s graphics 2-4. The only available source is Olalde 2015 Tab S15, which doesn’t check for changes in ANF-WHG, and even less so for CHG/Iran_N, arrival of which in Sicily during the EBA is documented in the new Fernandez e.a. paper.

    the earliest tholoi appear c 2900 BC ..
    Here, you have been mixing up tholoi and pithos. I had been referring to the latter – burials in amphorae, as e.g. known from Areni I, Kura-Araxes, (pre-) Hittite NC Anatolia, Crete, the Middle Helladic, and El Argar II.

  73. @ Frank thanks for your points.

    1) ”Not quite. What catches the eye in the new graphics (thx, Matt) is that I3188 from Annecy clusters amidst Iberia_BA. He is especially close to I6618 from the eastern outskirts of Madrid, datewise pre-Cogotas / early Cogotas I.”

    Do you mean I1388 ? That’s French BB from ~2300 BC, I6618 is from 600 years later.
    What exactly have we uncovered there, in their plotting near each other., and how does it change what I surmised ?

    2) ”Otherwise, we need to include Britain in the picture. Copper mining in Ross Island, Ireland seems to have been initiated from Iberia already during the CA,”

    That might very well be the case, but its clear that British & Irish BB men derive from Dutch BB, and not Iberian Copper Age groups. On the other hand, Iberian BB does not derive from Britain. That has been described in the paper, and is supported by the uniparentals.

    3) ”Last but not least, I believe that we will not be able to fully understand BB (and also Megalithism before it) before we get a decent chunk of aDNA from Brittany.”

    aDNA from Brittany would be great, but I don’t suspect it will impinge on the origins of BB.

    4) ”Where do you take this information from? The samples in question haven’t been included in Dave’s G25 set”

    There are 4 Iberia SE BA in the G25, which date between 22-1700 BC, i.e. during the El Agar civilization.

    5) ”Here, you have been mixing up tholoi and pithos. I had been referring to the latter – burials in amphorae, as e.g. known from Areni I, Kura-Araxes, (pre-) Hittite NC Anatolia, Crete, the Middle Helladic, and El Argar II.”

    Ha ! I got tongue tied. But I think the point remains – tholoi are native elaborations, and the pithoi are native adaptations too.
    The only site with pithoi i found in the suppl. is Cabezo Redondo., and I3486 is modelled as: 65% Iberia CA 35% Germany BB.
    With the recent set of aDNA papers, it seems East Med ancestry had only got to Sardinia, but not to Spain.
    More sampling might clarify things; like La Bastida , should be definitive.

  74. @rob, that sounds like a sensible strategy for G25 and nMonte, though requires some very conscious decisions about time (e.g. do all Beaker Bavaria come before Beaker France? If not should we use the subset that are clearly earlier?), and sampling levels (e.g. if we have 3-4 from some population and they’re all heterogenous in their ancestry, how certain are we that have a good representation of these guys as as source population).

    I guess my concern is more about the steppe ancestry transitional period samples in Olalde 2019 which are not able to be projected on these PCA though, and which we can’t cross check.

    If their modelling strategy has missed that two of the samples from Iberia_CA/Beaker_Iberia which can be put on G25 have a level of German Beaker ancestry that is in possibly equal or higher than at least one of the Iberia_BA samples they do model (I7689), then might they have missed any others with us (and the whole archaeogenetic community) none the wiser (because we can’t see it on these PCA)?

    For ex, even with the samples we know, for another visualisation, for the samples that have been included in G25, I’ll plot time by Distance from Mittelelbe-Salle Beaker (as a proxy for ancestry from Central European Beaker):

    Compare to Fig2B from the paper, where you see really an absence of low level Beaker ancestry individuals before the Bronze Age and a flat distribution within the samples they label as Iberia_CA.

    Here you have a suggestive reduction in distance from distance from Beaker MittelElbe-Salle in the two out of three of terminal Iberia_CA samples between 2300-2000 BCE (including one I2a2a male), at the same time as a trend with the samples they label Iberia_CA_Stp. Would this validate with a wider cut of their samples that don’t make it onto G25? We can’t really know.

    (Note: ATP016 early Copper Age is an outlier to the above graphic as she seems to behave weirdly in G25).

    The only real solution for that would probably be for me to learn to engage qpAdm and then email the authors with any result in a form they’d understand…

    @Alberto, btw, re: “but it seems the only one with that Y chromosome until the Celtiberian guy” which I should have responded to yesterday, yes if we’re talking I2a2a only, but possibly “no, there is another” (in the words of Yoda) if we’re talking non-R1b-M269. The COV20126 sample from the MBA from seems to have G2a and some steppe ancestry, though someone at anthrogenica who looked directly at y-full and the data file seemed to think he was I2.

    Pretty slim pickings and it’s pretty clear that combinations of founder effect and male sex bias in the migration in the first place led to huge expansions of Iberia specific R1b-M269 clades however.

  75. Rob:
    1)I indeed meant I 1388, sorry for the typo. The implication is that (Central) Iberian EBA ancestry may partly have been assembled outside Iberia and migrated there during the EBA.

    2) “its clear that (a) British & Irish BB men derive from Dutch BB, and not Iberian Copper Age groups. On the other hand, Iberian BB does not derive from Britain. That has been described in the paper, and is clear from the uniparentals.”
    Is it? From Olalde 2018 (published version,;
    During the first centuries after the initial contact, between approximately 2450 and 2000 bc, ancestry proportions were variable (Fig. 3), which is consistent with migrant communities just beginning to mix with the previously established British Neolithic population” There is e.g. the I2416 Boscombe Bowman (2470-2285 calBC , Amesbury Down, Wiltshire, R-L151) “low steppe” outlier. The Amesbury Archer obviously didn’t come from the Netherlands, and on other early British BBs, the final verdict is still out. For 14 Britain_BB samples, modelling as 2-way Britain_N – NL_BB admixture failed (negative coefficients, see Table S9).
    Now, note that only 7 of the 33 Britain_BB samples have been directly AMS-dated to before 2,200 BC, while I counted at least ten cases assigned as BB according to archeological context (e.g. I1770, Staxton Beacon, 2400–1600 BCE; I4950, 2500–1800 BCE; I6777 from a highly disturbed context “rendering any artefact associations uncertain“). Note furthermore that the Olalde e.a. 2018 chronological division of BB_Britain reaches up to 2,000 BC, and some of their analyses pool Britain_BB with Britain_EBA, up to ca. 1,500 BC. In the light of the Iberian results, and taking account of the 2,200 BC plague in Central Europe, the British results may require a finer chronological separation and subsequent analysis before final conclusions are drawn.
    As concerns uniparentals – of the 21 Britain_BB males, 13 were tested positively for R1b-L21, one of which (I2565, Amesbury Down) has been AMS-dated to before 2,200 BC. So, your argument is valid for the EBA, but not for Chalcolithic BB.
    Otherwise, we so far lack any aDNA from Cornwall, and the only Irish aDNA available is Rathlin.

    4. “There are 4 Iberia SE BA in the G25, which date between 22-1700 BC, i.e. during the El Agar civilization.
    That’s EBA El Argar A. The “orientalisation” appears in MBA El Argar B. Check out for a start
    In the phase B of this civilization, burial in pithoi (large jars) becomes most frequent (see: Jar-burials). Again this custom (that never reached beyond the Argarian circle) seems to come from Greece, where it was used after. ca 2000 BC.
    I am certain Alberto can provide more background, if required.

  76. @Matt

    Yes, I can only agree with you that it’s a bit unfortunate not to have the full details of all the samples. It does seem a bit strange, since in order to label separately the Iberia_CA and Iberia_CA_Stp groups they must have tested each individual and pooled together as Iberia_CA the ones with no steppe admixture to then proceed to model the individuals with steppe admixture seen in Table S15. Seems odd that they missed at least a couple of samples. Maybe with formal stats those samples don’t show steppe admixture as they do in G25?

    On the other hand, I don’t want to complain much. The details of those transitional samples are important, and the more the better. But it’s also fair to say that the individualized table is in itself an improvement over previous papers with just group models (for example, the Mycenaean 4 samples should have deserved an individual treatment, and probably the Swat Valley ones too due to their variability; in G25 going from 0-20% with an average around 10-12%, but in the paper pooled with a 20% Steppe_MLBA admixture in another underfit model, not separating the high AASI Indus periphery from the other two, and instead including Onge for extra AASI in Swat samples, with the other 3rd source being Steppe_MLBA. Was it too much to include the rather obvious BMAC samples too?).

    So if not perfect, it’s still quite a good job they’ve done with the models in this paper. I’ll see if I get the chance to ask about those possible samples missed as I gather some other details about the BA sites and samples. Which brings me to…


    I was planning to dedicate a different post to the Bronze Age, and wanted to do this with you especially. However, realising that the samples from La Bastida (from which we got hints about the results in an interview over a year ago) are missing from this study I guess that some other one might be coming soon. So we might as well wait for those further samples to comment on the whole Bronze Age with all the data available. Let me know when you’re up with your email for easier communication.

  77. @Alberto: That’s a positive take on it; individualized proportions are certainly the way they should go when there’s indication that the samples aren’t all equally related to outgroups and the coverage is sufficient to run as individuals. Olalde did this in the previous paper on British Beaker and it is quite useful for looking at individual samples there, and this seems to be the approach Lara Cassidy will take with her Irish megaliths paper.

    It’s the bits where they are splitting samples from shared contexts on statistical properties that are not indicated transparently, and seemingly mixing results from qpAdm runs on individuals and groups with the detail hidden in the supplement that I’m being a bit hard on here.

    Maybe with formal stats those samples don’t show steppe admixture as they do in G25?

    Maybe, depending on what the criterion is for showing steppe ancestry (they don’t tell us!). The correlation between the G25 distance from Beaker Mittelelbe-Salle and their modelled ancestry proportion is pretty strong though for those that are present in both:

    But if so, that’s a stronger case for transparency around this. (And it would be good practice to apply the criterion throughout their periodisation; e.g. don’t model I7689 with steppe ancestry if she also fails the same criterion.)

  78. @ Frank
    I understand. I guess we had already known there was structure to the fusing process as well as ongoing migration, with pioneers and later groups moving in, and becoming demographically dominant over 2-300 years.

    I look forward to yours & Alberto’s post on the Bronze Age.

  79. @ Matt
    Im not sure if Im quite following the line of enquiry, but i think I agree.

    Most of the ”Iberian Beaker No-steppe” individuals simply fall with the other Iberia CA samples. Two stand out slightly which show a signal of admixture with northerners – good spot Matt & Alberto !

    I3485 – Iberia CA – Female 23-1900 BC; Castilla, central Spain.
    I4229 – ”BB Iberia no-steppe”. Male, I2a1, Portugal 23-1900 BC

    Then there is (black dotted circle) the ”BB Iberia Steppe” cluster, which includes:

    I5665 – BB Iberia (M) R1b; El Virgazal, Burgos; 22-1900 BC
    I0461 – BB Iberia Female Arroyal 23-2200 BC
    I6539 – BB Iberia (M) R1b Humenajos 25-2000 BC
    I6472 – BB Iberia (M) R1b La Magdalena 25-2000 BC
    I3238 – ”Northwest CA Iberia” (M ) R1b 25-2000 BC
    I6623- B.B. Female C. Yaseras 25-200 BC.

    In order of steppe ancestry, with the highest being the BB from Camino de Yeseras being highest, and the new B.B. male from Asturias (I3238) 2nd highest. These are the immigrants.

    The pink cluster are the younger, Bronze Age Iberians, all shifted, with south – north cline

    Further out, and going back in time, a looser cluster from BB-France & BB-Bavaria, followed by a tighter cluser of BB, Ned, France , Bavaria

    So the picture again suggests a migration via France, arriving in Iberia c. 2500 BC, a shift over 3-400 years, and a new ‘set point’, for the Bronze Age.
    The details then lie explaining how differnt parts of Iberia went different ways at different times.

    NB; Otherwise I don’t think we’re missing anythig huge .THe ”CA Iberisa steppe” are just ”BBs”, and differe to the other “CA” samples in tat they’re non-locals (rel. speaking). Of course, with more analysable individuals, we’d find more on the path to initial admixture, so it would be great if Matt does qpADM on all others ! In addition, despite signs of initial admixture, it seems to suggest that it mostly became one way process socially & patrilinearly.

  80. The Iberian study is really good and the results of it so very nicely and simply laid out that you barely have to do any work to absorb the basics at least. Congrats to the authors. I also have to commend G25 here again since the broad models in the paper were basically replicated in it when trying to model modern Iberians with a reasonable model, even back when we had much fewer samples. Some comments that have been mostly covered I suppose but I couldn’t help adding my two cents:

    – Basques appear as an Iberia_IA population, though a question to be asked here is whether they’re also a slightly more northern population that moved southwards later on, during the LBA-EIA and that complicates the question of when they basically acquired their modern structure. It was clear that Basques needed extra steppe-heavy/central European kind of ancestry since Iberia_BA samples were overall more western/EEF-WHG-heavy. That being said they do overlap the northeastern part of the Iberia_BA range. The kind of reasonable model you’d get in G25 was Iberia_BA + England_Roman/England_IA and the more outlying/extreme of the paper’s individuals are similar to the latter. Not that they represent the admixing individuals necessarily, rather the “northeastern” direction towards which the new admixture generally points.

    – Rest of Iberians having additional Central-Eastern Med-like (though it’s interesting in the paper that the Mycenaeans and classical Greeks don’t seem to provide a good enough proxy for that but modern Italian and Greek populations do. I wonder what that exactly means, a population with extra steppe ancestry compared to the Mycenaeans and classical Greeks? Republic-era Romans?) and North African kind of ancestry. The equivalent model in G25 was Iberia_BA + England_Roman/England_IA + various Central/Eastern Med stuff + Iberomaurusian. Interestingly non-Basques appeared as having less Iberia_BA-related ancestry compared to Basques and more England_Roman/England_IA-related. This might tie in nicely with the results of the paper that show the historically IE-speaking areas to have acquired some further steppe admixture compared to non-IE-speaking ones post-BA.

    It’s interesting but perhaps not surprising (Catharge etc.) that North African ancestry in the south, at least to an extent since there seems to be a slight further increase on average, predates the Muslim conquest and it looks like you can get modern Iberians via varying mixture of the medieval Northeast and Southeast clusters. The apparently Visigothic-related ancestry that the one population shows here seems minimal overall (but some Iberian subpopulations seem slightly drawn towards that cluster) and I also wonder whether this Visigothic ancestry is purely Scandinavian-/Germanic-like. It doesn’t seem so at first glance based on the position of some of them on the PCA but I haven’t taken much of a separate look at them. Possible they have the mixed kind of ancestry some of the previous East Germanic groups showed? [In general it seems that Europeans everywhere come to their absolutely modern likeness (whether the differences are more or less subtle) essentially in medieval times, even if the basic structure of the continent is laid down by the Bronze Age. Probably not too unexpected either.]

    The Central European scenario for the spread of Celtic to Iberia is certainly strengthened with these results, in my view, as people have already at least indirectly alluded to. The early Beaker spread into Iberia would seem more responsible for Lusitanian-related or even earlier western/Italo-Celtic IE related languages in the case where IE-related varieties persisted (if we don’t also ascribe non-IE varieties to them as well per below). Later samples from Ireland will be welcome too since we can see a clear shift towards the EEF side compared with Ireland_EBA but it’s a bit unclear whether that might even mostly be via some process of increased Ireland_MN-related ancestry to simplify things a bit. I hope Koch pays attention to these new results since he seems to have assimilated the genetic data of the previous papers quite well, even though his Atlantic view is weakened by this. One question I have about the eastern Hallstatt samples we have is that, when I took a look at them in G25 a while ago, even though they generally seemed French-like in the first two dimensions (but lower steppe, higher EEF/WHG on average), they looked a bit modern Central-Eastern European like in others. I’m not exactly sure what that implies (even though they’re French-like, they affected mostly Central-Eastern European populations?) but maybe that’s part of the reason why they don’t seem like the best fit as an admixing source for the post-BA Irish right now?

    Also the Iberian Mesolithic samples seem to behave kinda weirdly (not in the sense of unexpected or nonsensical but very outlying and they seem to be pulling some of the Iberian CA samples towards them) in some dimensions but I haven’t looked at them much yet. Hope for some comments on that in the HG post I noticed has just been published.

    A question about the Greek Emporion samples is whether they represent essentially complete continuity with the Mycenaeans (because they really do seem like they basically belong to the same population) or whether they represent mixture between a slightly more northern population (say Balkans_IA-like) and an Anatolia_MLBA-like one considering the ties of these western colonies to the Greek cities of the Anatolian coast who often exhibit important substrata in onomastics, suggesting important local mixture. It might be hard to tell the two scenaria apart considering how similar in some ways these populations are.

    I also notice David has now considered the possibility of Beaker becoming multilingual and carrying with it Vasconic as well. I seem to recall that being a theory brought up by Rob and Alberto on his blog too (as in previous posts here) that he was -very- vehemently against. For my part I’ll keep to the “Y-DNA replacement but linguistic continuity” as handwavey as it sounds for the time being but this situation certainly makes you wonder a bit and it’s an area where one should keep (and have kept) an open mind. Either way, it definitely looks like the early BB spread didn’t cause Indo-Europeanization in a good part of its range as expected. And more generally, all these phenomena were likely quite complex, not unlike contemporary archaeologists and anthropologists insisted rather than just a popular fantasy of Indo-Europeans playing a constant game of whack-a-mole (not that there wasn’t a lot of that) that’s often brought up online.

    The Mycenaean “African” signal that Matt brought up is likely damage/low coverage in I9033 like Alberto said. At least that was my impression a while ago, probably from looking at the paper together with the G25 models, but I don’t remember the specifics right now. At around the same time I noticed one of the low-quality Sintashta samples gets (or got, in case it’s been removed) a similar “African” signal and I do remember that being very low coverage in the original paper. Its weird position in the sort of PCA, relevant to this post too (modern European + paper’s samples in G25 + some North and sub-Saharan African populations), compared to the other Mycenaeans hints at that too: . Without the African samples: .

    Apart from the interesting recent Sardinian and West Med preprints that I haven’t had a chance to take a look at yet, I notice a Jomon preprint has been published as well, in case people weren’t aware.

  81. @Egg

    Thanks for adding your comments. I mostly agree, but this is such a large paper in scope and samples that I could only comment so far on a very small part of it. It will take me time to get into all the details.

    A few comments to add to yours:

    – Basques appear as an Iberia_IA population, though a question to be asked here is whether they’re also a slightly more northern population that moved southwards later on, during the LBA-EIA and that complicates the question of when they basically acquired their modern structure.

    Yes, the basic idea is that Basques come from Aquitaine (SW France). this is based on the ample evidence of Vasconic (Aquitanian, in fact) in that area while the evidence is rather scarce south of the Pyrenees (I mean during antiquity). So it has been argued for a while that modern Basques only moved to Iberia during the Roman period. While I’m personally not sure if that’s totally correct (could be), what is clear is that Basques and Aquitanians were the same people, so regardless of their historical geographical range modern Basques are expected to be more “northern” due to that strong connection with SW France.

    – Interestingly non-Basques appeared as having less Iberia_BA-related ancestry compared to Basques and more England_Roman/England_IA-related. This might tie in nicely with the results of the paper that show the historically IE-speaking areas to have acquired some further steppe admixture compared to non-IE-speaking ones post-BA.

    Could be, at least judging from those 3 Celtiberian samples (though they are from the modern southern part of the Basque country). However, I’d also note the possible role of the middle ages in this.

    When at school we studied the “Reconquista”, it was actually “Reconquista y repoblación”. The process of re-populating the newly conquered lands was an important enterprise to secure them, and that needed some efforts to bring new settlers to the areas. It was highlighted the importance of the Road to Santiago, a pilgrimage route very popular at that time that brought pilgrims from all Western Europe to visit the (according to the legend) burial place of Santiago (St. James, Sveti Jacob,…). It was said that lands were offered by the Christian kingdoms of the north to foreigners (in addition to locals) to settle in the conquered areas, which brought many Western European settlers to Iberia.

    The genetic impact that this may have had is still unknown. From the Y-DNA point of view there is not evidence of much impact, it seems. But I guess it’s the same for Celtic/Para-Celtic migrations too. So this is still an open question.

    – The apparently Visigothic-related ancestry that the one population shows here seems minimal overall (but some Iberian subpopulations seem slightly drawn towards that cluster) and I also wonder whether this Visigothic ancestry is purely Scandinavian-/Germanic-like.

    The Visigoths didn’t arrive to Iberia in any direct way. Their migration started long before, moving south to the shores of the Black Sea, starting to have significant contacts with the Roman empire, duelling in the southern Balkans, then moving to Italy (where the Eastern and Western Goths split into Ostrogoths and Visigoths), then settling in France and finally in Iberia. They had already become Christians during their time in the Eastern Roman Empire, and adopted Latin as their language. So they must have been a pretty mixed population by the time the moved to Iberia.

    But yes, in any case their genetic impact has never been estimated to be large, and DNA seems to confirm this.

    – The early Beaker spread into Iberia would seem more responsible for Lusitanian-related or even earlier western/Italo-Celtic IE related languages in the case where IE-related varieties persisted (if we don’t also ascribe non-IE varieties to them as well per below).

    It seems difficult to connect Lusitanian to the early Bell Beaker spread into Iberia. If it split from Celtic 2000 years earlier it would be impossible for it to keep such resemblance. It’s often been argued if Lusitanian is Celtic or not, and while it’s now mostly considered non-Celtic, it’s still a para-Celtic language (Venetic seems a reasonable sister language). So its arrival cannot be pushed much further back, maybe to around 1200-1000 BC, of not a bit later.

    I still didn’t have time to start writing about the linguistic substrates, which should be an interesting subject with rather broad implications. When I get around it we’ll get a better chance to look into all these linguistic problems.

    – Either way, it definitely looks like the early BB spread didn’t cause Indo-Europeanization in a good part of its range as expected. And more generally, all these phenomena were likely quite complex, not unlike contemporary archaeologists and anthropologists insisted rather than just a popular fantasy of Indo-Europeans playing a constant game of whack-a-mole (not that there wasn’t a lot of that) that’s often brought up online.

    Yes, agreed. But keep in mind that the same people arguing for this latter scenario were quite fast to change the tune when confronted with the Mycenaean and even more with the Hittite samples.

    – I also notice David has now considered the possibility of Beaker becoming multilingual and carrying with it Vasconic as well. I seem to recall that being a theory brought up by Rob and Alberto on his blog too (as in previous posts here) that he was -very- vehemently against.

    Definitely a change in the right direction and attitude. I hope that his own change will have a positive effect on Eurogenes as a whole, since for a year it has been in a really poor state in the comments section (and this was his own decision back then, so it’s good to see things turning around for the better again).

    Re: Mesolithic, yes, I hope to comment about it in the relevant post when I have the time to look into the samples more carefully too.

  82. Also, since I had some time now to run a few models again, to add -and probably ramble- a bit on Matt’s interesting comment about Balkans_BA seemingly preferring a Steppe_MLBA-like source and the Mycenaeans a source between Yamna and the Caucasus that has been discussed a decent bit before on here…

    Clearly you need either a more HG-heavy/”western” steppe source than Yamna or a more HG-heavy ‘local’ population than Balkans_Chl. This can be achieved by either considering a Steppe_MLBA source OR a Balkans_Chl + Tripyllia combo for the local ancestry in which case a Steppe_EBA source works about as well.

    On average, and you can sorta tell with the ADMIXTURE analysis in Mathieson et al. or Rob’s G25 run in a previous post here when looking at HG/steppe ratios, there seems to be slight structure within the Balkans_BA sample as well with (discounting the Z93 late outlier, plus the one sample with no steppe at all) samples from Bulgaria exhibiting lower steppe and higher HG-related ancestry and samples from Croatia the reverse on average and I wonder if that has to do with geography or chronology too. Would have to refresh my mind on the latter.

    Running a bunch of models in a basic local farmer + steppe population format, the best results (though very often the distances are very similar) are, after excluding CWC and Beaker-related results to simplify things a bit, in order

    – Balkans_Chl + [steppe]
    Srubnaya_MLBA “distance%=1.5063”
    Sintashta_MLBA “distance%=1.5284”
    Srubnaya-Alakulskaya_MLBA “distance%=1.5565”
    Yamnaya_Samara “distance%=2.1271”
    Yamnaya_Kalmykia “distance%=2.2326”
    Poltavka “distance%=2.2358”
    Yamnaya_Ukraine “distance%=2.2896”
    Ukraine_Eneolithic “distance%=2.3608”
    Catacomb “distance%=2.3941”
    Potapovka_MLBA “distance%=2.4250”
    Yamnaya_Karagash “distance%=2.4637”
    Yamnaya_Bulgaria “distance%=2.4638”
    Varna_o “distance%=2.6172”
    Yamnaya_Caucasus “distance%=2.7410”
    Progress_Eneolithic “distance%=2.9498”
    Vonyuchka_Eneolithic “distance%=3.2806”
    Yamnaya_Ukraine_o “distance%=3.6034”

    – Balkans_Chl + Tripyllia + [steppe]
    Srubnaya_MLBA “distance%=1.1997”
    Srubnaya-Alakulskaya_MLBA “distance%=1.2222”
    Catacomb “distance%=1.2991”
    Poltavka “distance%=1.3214”
    Sintashta_MLBA “distance%=1.3271”
    Yamnaya_Kalmykia “distance%=1.3524”
    Yamnaya_Samara “distance%=1.3759”
    Yamnaya_Ukraine “distance%=1.5231”
    Yamnaya_Karagash “distance%=1.5364”
    Yamnaya_Bulgaria “distance%=1.5421”
    Potapovka_MLBA “distance%=1.6305”
    Yamnaya_Caucasus “distance%=1.6590”
    Progress_Eneolithic “distance%=1.7344”
    Vonyuchka_Eneolithic “distance%=1.9310”
    Yamnaya_Ukraine_o “distance%=2.2949”
    Ukraine_Eneolithic “distance%=2.3608”
    Varna_o “distance%=2.4059”

    Removing the Steppe_MLBA-like Z93 outlier from Balkans_BA and re-averaging doesn’t change things too much:

    – Balkans_Chl + [steppe]

    Srubnaya_MLBA “distance%=1.5875”
    Sintashta_MLBA “distance%=1.6536”
    Srubnaya-Alakulskaya_MLBA “distance%=1.6793”
    Ukraine_Eneolithic “distance%=1.8432”
    Yamnaya_Samara “distance%=2.1407”
    Varna_o “distance%=2.205”
    Yamnaya_Kalmykia “distance%=2.2099”
    Yamnaya_Bulgaria “distance%=2.2103”
    Poltavka “distance%=2.2136”
    Yamnaya_Ukraine “distance%=2.2555”
    Catacomb “distance%=2.3216”
    Potapovka_MLBA “distance%=2.3752”
    Yamnaya_Karagash “distance%=2.4182”
    Yamnaya_Caucasus “distance%=2.5738”
    Progress_Eneolithic “distance%=2.7176”
    Vonyuchka_Eneolithic “distance%=2.9196”
    Yamnaya_Ukraine_o “distance%=3.2174”

    – Balkans_Chl + Tripyllia + [steppe]

    Srubnaya_MLBA “distance%=1.2553”
    Yamnaya_Bulgaria “distance%=1.3299”
    Srubnaya-Alakulskaya_MLBA “distance%=1.3311”
    Catacomb “distance%=1.3821”
    Sintashta_MLBA “distance%=1.4117”
    Poltavka “distance%=1.4156”
    Yamnaya_Kalmykia “distance%=1.4357”
    Yamnaya_Samara “distance%=1.4667”
    Yamnaya_Ukraine “distance%=1.5590”
    Yamnaya_Karagash “distance%=1.6046”
    Yamnaya_Caucasus “distance%=1.6155”
    Progress_Eneolithic “distance%=1.6441”
    Potapovka_MLBA “distance%=1.6552”
    Vonyuchka_Eneolithic “distance%=1.7092”
    Ukraine_Eneolithic “distance%=1.8432”
    Varna_o “distance%=1.9468”
    Yamnaya_Ukraine_o “distance%=2.0487”

    When you add the HG-richer Tripyllia as a second non-steppe source, the differences between the Steppe_EBA and Steppe_MLBA models predictably diminish a lot to the extent that some Steppe_EBA sources furnish even better fits in G25 at least.

    The Mycenaeans are a bit weirder and you can’t really get the distance below 2.5+ or 2.0 with a fair amount of overfitting without some other sources. Though adding the Mycenaean-like Classical and Hellenistic samples and averaging the lot, the average gets a more reasonable distance with the Aegean-Balkan sources we have right now:

    [1] “distance%=1.5911”



    Removing the Ukraine outlier Vucedol

    [1] “distance%=1.6095”



    Even with just Minoan and Peloponnese N, you get a decent distance

    [1] “distance%=1.8281”



    Without the already Balkan-mixed Yamnaya_Bulgaria

    [1] “distance%=1.8743”



    It’s a bit hard to tell the specifics apart, like we’ve discussed in the past, since the differences in distances are extremely small but this broad model holds. The Myceanean sample itself being relatively limited in numbers might be part of it and adding those two new ones seems to help. Some 15% steppe-related ancestry makes sense. About 20-30% in the Balkans_BA sample, and much of that coming from rather northern locations, with more appropriate farmer samples, and an apparent further drop if the Balkans_IA average holds.

    Trying to unite what’s going in the Balkans in general and keeping in mind the I2a and Z2103 lineages that seem to accompany the steppe ancestry, a model of local farmers (Tripyllia-Balkans_Chl for more northern Balkans, Minoan-Peloponnese_N for at least southern Greece) + a late Yamna or Catacomb (very similar to Yamna in the first place, so might be a bit hard to really tell apart in this sort of thing) population seems to work pretty well without potentially overfitting things. At least some of the more northern samples seem to have a similar extra CHG kind of ancestry, as Rob showed too with using Anatolia_EBA as an extra source in a previous post. I think it’s hard to explain at least the Vucedol sample otherwise. I also think this kind of ancestry more reasonably comes from the Aegean-Anatolian area per the Krepost, Minoan and Peloponnese samples without having to consider some less likely imo hypotheticals like Yamnaya_Ukraine_o-like or Kemi-Oba (being plausibly more Caucasus-affected, though no samples yet, to my knowledge) groups having extra/special influence there.

  83. Having had a brief look at some of the comments at eurogenes on the new post, Matt brought up an imo good objection about Western vs central European Celtic. If there’s long-term exchange/homogenization within the Atlantic area, how can we tell that scenario apart from a Central European migration one? Leaving non-genetics aside since most linguists and archaeologists seem to generally favor a Central European LBA/EIA expansion of Celtic to my knowledge (but that’s against what one would consider Koch’s less mainstream view in the first place after all) I’m not sure we fully can with our current samples. My first thought was that “steppe increase = Urnfield/Central European expansion?” and I admit I hadn’t fully considered that possibility. Also on that topic, I had asked Carlos -who didn’t get back to me on that- before, if there were general signs of Urnfield influence throughout Iberia. One reason I had considered an Atlantic possibility as well is that to my very general knowledge that sort of Central European influence tends to peter out towards the northeast of the peninsula, not that this would necessarily be a deal-breaker considering there doesn’t always seem to be an 1:1 in these matters. Same with Ireland where only parts of it seem potentially affected. What are people’s views on that? Am I just totally mistaken about this aspect?

    A first impression is that, looking at most relevant BA-IA western samples I could think of, both Iberia and the Isles are generally (and relatively subtly to be fair) drawn to the center of the range over time. Other than that, I’m not totally sure – only taking autosomal considerations into account, there doesn’t seem to be a total knockout argument either way right now that I can think of. We might need samples from within the Atlantic range to see what sort of changes were going on everywhere to fully answer all objections…

    Perhaps in something more concrete, taking a quick look at the Hallstatt samples again to refresh my memory about the “eastern European”, what I think I had in mind was DA112 which shows a sort of eastern European shift compared to DA111. It doesn’t seem fully explainable either by the amount of steppe ancestry or some weird eastern Lokomotiv_N/Shamanka_N-like (though very minor) stuff it gets.

    Using Baltic_BA, it seems to have a certain attraction towards that side of things:

    When throwing everything in G25 at it together, it picks up on Baltic_BA, Golden_Horde_European etc. in appreciable amounts.

  84. Egg- re Bylany:
    Take a look at the map of BA/IA amber roads:

    Lovosice, from where the Bylany samples originate, lies on the Upper Elbe, just where the Kujavia-Notec-Warta-Havel-Elbe CE route leaves the Elbe in order to via Prague or Pilsen cross into the copper-rich Inn Valley (Kramsach area mines)and ultimately, via the Brenner, reach the N. Adria. As such, substantial Lithuanian (Baltic BA) influence on Bylany aDNA certainly makes archeologic sense.
    Check out this for better understanding:,+Oblast+Kaliningrad,+Russland/Biskupin,+kujawsko-pomorskie,+Polen/Wicina,+68-320,+Polen/Niederkaina,+Bautzen/Lovosice,+Tschechien/Vladař,+Žlutice,+Tschechien/Svržno,+345+25+Hostouň,+Tschechien/Weingarten+1,+94333+Geiselhöring/Kramsach,+Österreich/Este,+Padua,+Italien/@49.9753806,16.241576,5.31z/data=!4m62!4m61!1m5!1m1!1s0x46e33d8d4b7c21a9:0x5050960016126ed3!2m2!1d20.4522144!2d54.7104264!1m5!1m1!1s0x470362e834317f91:0x66990b45e63007be!2m2!1d17.716667!2d52.783333!1m5!1m1!1s0x47062704b1585a65:0x1f14db8a50a63d6!2m2!1d15.0771228!2d51.7678063!1m5!1m1!1s0x4708ff38b7517493:0xcf1c1b056bcbb0fd!2m2!1d14.4783285!2d51.1913116!1m5!1m1!1s0x470980115104caf7:0x400af0f661564a0!2m2!1d14.0510416!2d50.5150429!1m5!1m1!1s0x470a656f5a2d9683:0x7b73547015e1fe50!2m2!1d13.2154191!2d50.079903!1m5!1m1!1s0x470aa0df818db49d:0xa00af11d99ab320!2m2!1d12.7712992!2d49.5749452!1m5!1m1!1s0x479fd635ef9a131d:0x3fdc9c27e41c5ee!2m2!1d12.3496554!2d48.8192482!1m5!1m1!1s0x4777d4a3aed94feb:0xa1871b86b8b5f226!2m2!1d11.875953!2d47.4435993!1m5!1m1!1s0x477f1df880e3f27d:0xdd9d79bc3efc1f2f!2m2!1d11.6603917!2d45.2254915!3e2
    For the “weird eastern Lokomotiv_N/Shamanka_N-like (though very minor) stuff“:

    a) Scythian presence on the Middle Oder around 600 BC is archeologically attested. That’s a good century later than the Bylany sample, but some “scouting” should have been going on before Scyths decided to en masse try their luck in S. Poland.

    b) Trade certainly continued beyond Lithuania. In Varangian times, a northern branch of the Silk Road ended in Old Novgorod – the relation might already have started earlier. Otherwise – how does Karelia BOO fare in modeling that “weird eastern stuff” ? And – hasn’t the Baikal produced some early “Uralic” yDNA N?

    Edit: I have updated the Google map above with some major Lusatian Culture sites, and Hallstatt “Fürstensitze” (princely seats) and trade centers for the route in question.

    Edit2: On the southern end of this road, see Abb.3 inüdwestdeutschen_Verbandes_Altertumsforsch._e._V._Nürnberg_2010._UPA_69_Langenweißbach_2012_89-94

  85. @egg: trying to look at the same concept that motivates your PCA, by reprocessing the G25 data: (PC1 captures EEF vs Steppe_EMBA+BalticBA shift, PC2 Steppe_EMBA vs Baltic_BA, PC3 seems further East vs West Europe dimension). IberiaIA_North are Celtiberians, IberiaIA_East are the Iberians, Iberia_NE_Hel/Greek_Gpop are the Empuries samples minus the Mycenaean-like outliers.

    The shift of DA112 is pretty obvious and off cline as anything that makes much sense as contributing to England_IA / Iberia_IA, while DA111 has minimal enough of that that this sample could contribute. (There aren’t 3x Bylany or IberiaIA_North on here btw, I just added an average on to make the their convex hull visible).

    To me it seems hard to say what the North / East Iberia IA / Greek Era groups are doing (which they’re exactly between), other than that I8206 is an outlier very close to England_IA. I guess as I have suggested in the Eurogenes comments section, it is hard to tell if there are any pulse migrations that have a significant genetic impact, since relatively subtle changes and few samples compared to time and space being considered (and apparent trends with time and geography within Iberia).

  86. @egg, though I don’t think in theory there would be anything clearly wrong with trying to use D-stats based on different EEF affinity to try and distinguish between Iberia_IA_North and other Iberian groups. (Unless in practice data quality / issues impede it).

    If we were thinking of DA111 admixing with Iberia_North_BA or Iberia_East_IA to generate Iberia_North_IA, then we’d probably expect D(Iberia_MN;Some_Central_European_EEF;DA111,Iberia_IA_North) to be about 1/3-1/2 (Iberia_MN;Some_Central_European_EEF;DA111,Iberia_IA_North_ancestor).

    That’s actually something they do seem to have looked at in the paper as well.

    Much of the focus in the paper and discussion is mainly on the rough individual and group Germany_Beaker or Steppe_EBA proportions computed using qpAdm outgroups insensitive to particular EEF groups. See – for the outgroups; minimally sensitive to differences within EEF.

    But they do use an extended outgroup set that is more sensitive to different EEF ancestry. See the relevant sections from the supplement:

    The subgroup sensitive modelling in the higher coverage samples (Iberia_North_IA / Iberia_East_IA) seems to get England_MBA as best fitting combined with Iberia_BA, for whatever that does or does not mean.

    In theory you could do the same modelling with Halstatt_DA111, Halberstadt_LBA, or whichever.

  87. @Rob

    Since you are interested in the Balkans, have you noticed the LBA Sicilians in the recent Mediterranean paper? They carry G2a of the Baden/Thracian variety (Z1903) with a BA TMRCA, one of which is under a young, typically Scandinavian clade.

    I’m pretty sure that 1. they are recent immigrants and 2. IE was spoken in LBA western Sicily.

  88. @egg: on difference between EEF stats, some stats I had run a while back to estimate the continuity of Iberia_CA ancestry in Iberia (as distinct from European MN groups in general):

    Various other samples could well be added to these sort of plots to test degrees of any introduction of specifically Central European EEF ancestry.

  89. @ Marko
    I havent yet collated all the new data in my head in terms of the Bronze Age in Southern Europe. But there are already some interesting lineage links along with the autosomes. There is a type of I2a1a1 seen in LN Tissza, Baden, BB-Csepel, Remedello, Sardinia, Late Copper Age Iberia, S. France (Trier culture), Iberian Celt, etc.
    There are G2a links also (which I’m not yet even broadly updated on…); and the distintive pattern of J2a1 and J2b2. Still too few samples to say anything definitive, and made more complex given that some of these lineages were already flaoting around during the Neolithic,; but eliciting those which expanded during the B.A. will be important with future data. What have you made of it ?

  90. @ Egg
    To address your question – I think at this point the broad picture in Iberia fairly clear, it is for the Iron Age as it is for the BB phenomenon itself.

    The most north-shifted groups are specifically ”northeast Iberia Iron Age” samples. Ideed, the arrival of Urnfield, Halstatt & even La Tene influences were all mediated via the north east corner of Iberia. However, they were not restricted to only I.E. -speaking areas, but permeated through the non-IE Med. coast as well. What is therefore important, is the resulting demic impact and how these immigrant groups & influences were received in the local millieu. And prior to all this, was the Cogatas – El Argar distinction.
    Throwing in modern Iberians, we see an persistent (but more homogenised) distinction between IE-Spanish & Basques.

  91. @Rob

    Very weird archaeological context for these guys, Dated between 1300-900 BCE, yet buried in Dolmens associated with Beaker pottery. Evidence of significant violence. Looks like they have less CHG and steppe than previous inhabitants (Western Sicily was Bell Beaker central).

    Chronologically they should be associated with the hilltop fort builders of LBA/EIA western Sicily.

  92. A more intra-northern European look at DA112, alongside some of the other CWC/BB and post-CWC/BB samples that seem to show this eastern European’Baltic_BA affinity to accompany some of the further comments:

    Frank, thanks for the material. For the slight eastern/ENA stuff (if genuine), I had considered the influence of the Cimmerians at the time which seems to have affected at least eastern Hallstatt so something along the lines of what you suggested. I’m really curious about the Baltic_BA-related stuff though and I’m not sure if specifically ties to Lithuania/the eastern Baltic are the cause necessarily. This affinity seems to be decently widespread in eastern Europe throughout that period (it even seems to appear in some pre-Slavic, likely southeastern European samples to a small extent IIRC) though a bit spotty; then again we don’t have the most extensive sampling from there. One sample I’d compare more directly to DA112 is the Scythian_Ukraine scy009 which appears almost like a mix between something like Hallstatt (and carries R1b-P312, so perhaps a sign of more western influence) and a Baltic_BA-like population. Considering the area and potential date (the context seems Chernoles?), it might not even be inconceivable that it represents a mix between a proto-Slavic (or related) individual and a more western one…

    Matt, thanks for all the analyses. I also considered the possibility that maybe it was just generally elevated WHG-related ancestry in that particular sample that caused that but the DA112 shift appears to be specifically related to this Baltic_BA effect, per the above link too. IIRC Mittnik et al. went with a model of CWC_Baltic + Narva in the Northern European/Baltic preprint for Spiginas2/Baltic_BA but didn’t attempt to exactly argue for the specific source in the published paper since various other more western HG populations seemed to fit even better? I’d need to refresh my memory on that. At least at first blush it does appear to be some sort of HG population in-between WHG and EHG though much closer to the former (so I wonder if what’s causing much of this is elevated local ancestry by eastern European intermediate HG populations) though there seems to be some weird behavior too so I’m not sure if that’s the full story. This general behavior in G25 has been discussed a bit in various places, though not with specific reference to DA112 which was sorta related topic-wise, so I might be harping on this point a bit.

  93. Egg:
    Scythian_Ukraine scy009 indeed seems to come from the heartland of Chernoles on the western bank of the Middle Dniepr (see the map in
    However, scy009 isn’t dated very precisely, so he may well belong to a later, “Scythian” phase (from the late 7th cBC onwards).

    Cernoles Culture isn’t particularly well analysed. It is usually described as reflecting influence of the so-called “Thracian Hallstatt” circle (around modern Moldova) on local ( Trzciniec Culture) substrate. The material base was most likely export of Moldavian salt to Dniepr stockbreeders .

    That relation dates already back to the Neolithic, when Trypillian mega-cities received their salt from Cucuteni. The Neolithic exchange line extended backwards across the Carpathians into Transsylvania, where Bödrogkeresztur extracted and processed copper for exchange against Cucuteni salt. A similar relation, now also including iron, may also have existed during the LBA/IA and have connected “Thracian Hallstatt” to the eastern, Pannonian reaches of Hallstatt proper [further west, of course, the eponymous Hallstatt salt mines and those from the Salzburg area stepped in, to the NE of Hallstatt lies the MES (Halle/Bernburg) salt district that in historical times exported as far as East Prussia.] A few “Hallstatt swords” have been found in Chernoles / early Scythian burials.

    As such, it is in fact “not (..) inconceivable that it [scy009] represents a mix between a proto-Slavic (or related) individual and a more western one….”

    Otherwise, it is intriguing how close scy009 plots to modern Poles.

  94. Marko, they could be Cetina culture expansion ? They expanded though central Mediterranean


Comments are closed.