The French Bronze Age is what matters

A new preprint on ancient DNA, Ancient genomes from present-day France unveil 7,000 years of its demographic history. It goes from the late Pleistocene to the Iron Age, and has a lot of Neolithic samples, as well as Mesolithic and Bronze Age samples

Major takeaways:

– The Magdalenian populations, as represented by Goyet2, seem to have contributed ancestry to groups in substantial numbers down to the Mesolithic period. Earlier work showed the persistence of this group mostly in Iberia, but these data suggest they were present in France, and perhaps even Central Europe.

– The Neolithic populations are what you’d expect. The transition is what you’d expect (little initial admixture, later increase in Mesolithic hunter-gatherer ancestry). That being said, they seem to not establish whether the Neolithic farmers in France were mostly Cardial [Southewest European] or LBK [Central European]. It seems they lean to the proposition that they were more Cardial. Certainly for the southern samples.

– The arrival of the Beaker Culture heralded major genetic change as in Britain, though perhaps not as much. R1b became common, and steppe ancestry as well was ubiquitous. But, there was lots of variation. One of their samples is only about 25% steppe (the balance Neolithic-farmer), while another is 100% steppe. Southwest France, which had many non-Indo-European speakers until relatively late, had more Mesolithic hunter-gatherer and less steppe.

– Unlike in Iberia, there was significant mtDNA turnover. What this means is that the Indo-European expansion into Iberia was very male-mediated, but it France it wasn’t. Though the Neolithic impact seems higher than in Britain, on the whole there seem broad similarities here.

The shift from the Bronze Age to the Iron Age didn’t result in a change in the average ancestry, but the variance seems to have decreased. The reason for this is that prehistoric France seems to have been undergoing genetic mixing across reagions.

– Finally, strong very recent selection on lactse persistence and pigmentation.


20 thoughts on “The French Bronze Age is what matters

  1. Re the two new Beaker samples, not totally sure about their qpAdm for them; PCA position suggests the “100% Yamnaya” sample may have some MN ancestry (25% ish) while the collective burial G2a sample may not have any. Needs a cross check.

    Should be another pretty soon with 100 French meso->3000BCE samples mainly Neolithic era). – “Ancient genome-wide DNA from France highlights the complexity of interactions between Mesolithic hunter-gatherers and Neolithic farmers”

    Authors: “Rivollat, Maïté, Jeong, Choongwon, Schiffels, Stephan, Küçükkalıpçı, İşil, Pemonge, Marie-Hélène, Alt, Kurt W., Binder, Didier, Friederich, Susanne, Ghesquière, Emmanuel, Gronenborn, Detlef, Laporte, Luc, Lefranc, Philippe, Meller, Harald, Réveillas, Hélène, Rosenstock, Eva, Rottier, Stéphane, Scarre, Chris, Soler, Ludovic, Wahl, Joachim, Krause, Johannes, Deguilloux, Marie-France & Haak, Wolfgang”

    some more info on project –

    In general it seems to me like finding structure among the Neolithic populations is relatively apparent, once shotgun Vs capture confound is corrected for, using f statistics. Using only shotgun / capture samples together it’s relatively clear.

    Also think using different Mesolithic hg groups as way to detect structure this may not be needed, compared to just using Neolithic samples themselves, directly. Still some questions I think in working out the right Neolithic populations to give rise to each LNBA population, around whether happened in one place or little contributions in a few.

  2. There was also a thesis published with 93 ancient genomes from Ireland covering Mesolithic to Iron age. But no DNA was made public and the the thesis PDF has been removed.

    Cassidy 2020
    A Genomic Compendium of an Island: Documenting Continuity and Change across Irish Human Prehistory

    >Two Mesolithic samples. 20+ Neolithic, Bronze, Iron age samples. So, good pretty good spread.

    >The Mesolithic hunter gatherers were extremely inbred. They didn’t have Magdalonian ancestry.

    >Survival of Neolithic ancestry in Bell Beaker/CA in Southern Ireland. 50% Neolithic Ireland, 50% Bell Beaker. This is the one exception to population replacement in Bell Beaker British Isles.

  3. What are all those Bell Beaker samples with Y-DNA haplogroups like R1b1a1a2a1a and R1b1a1a2a1a2 on the above analysis? I suspect, unlike the other R1b samples in the analysis, they are labeled according to an outdated nomenclature, and should be labeled as R1b1a1b1a1a and R1b1a1b1a1a2 respectively and are thus R-L151 and R-P312 respectively.

  4. Another error: one of the Neolithic samples is labeled as W5b for both the mtDNA haplogroup and the Y-DNA haplogroup on that analysis. Obviously the Y-DNA haplogroup cannot be W5b, a non-existing Y-DNA haplogroup.

  5. Interesting to see that the old Goyet Magdalenian ancestry persisted visibly till the neolithic at least in some individuals.

  6. To provide a clarification of the initial observation in the OP with a bit of background, the Magdalenian period is a post-Last Glacial Maximum and pre-Holocene (ca. 17000 to 12000 years BP) period in Mesolithic Western Europe. France and all but a few refugia in Europe (in Iberia, Italy and the Caucuses) were completely depopulated at the Last Glacial Maximum ca. 20000 years BP. In particular there is a gap of perhaps 8,000 years in France where there were no humans present. The widespread hypothesis supported by genetic evidence is that Europe was mostly repopulated by people from one or another refugium who were the literal bottleneck population for homins in Europe, reducing genetic diversity in the subsequent Mesolithic population derived from them.

    Per :

    “In 2016, researchers successfully extracted DNA from several ancient human fossils at Goyet (with direct dates): GoyetQ116-1 (35,160-34,430 BP) and GoyetQ376-3 (33,940-33,140 BP) from the Aurignacian; GoyetQ376-19 (27,720-27,310 BP), GoyetQ53-1 (28,230-27,720 BP), GoyetQ55-2 (27,730-27,310 BP), GoyetQ56-16 (26,600-26,040 BP) and Goyet2878-21 (27,060-26,270 BP) from the Gravettian; and GoyetQ-2 (15,230-14,780 BP) from the Magdalenian. GoyetQ376-19, Goyet53-1 and Goyet56-16 were found to cluster genetically with several other Gravettian [i.e. after 34,000 BP but still pre-LGM] individuals from Europe in the Věstonice Cluster, while GoyetQ-2 was found to cluster genetically with several other Magdalenian individuals from Europe in the El Mirón Cluster. All later Europeans after GoyetQ116-1 show some genetic affinity for this individual. GoyetQ116-1 also exhibits more genetic affinity for Tianyuan man than any other ancient individual from West Eurasia.”

    The Magdalenian link in the OP notes that “70-80% of the ancestry of these [Magdalenian] individuals [including GoyetQ-2] was from the population represented by Goyet Q116-1, associated with the Aurignacian culture of about 35,000 BP, from the Goyet Caves in modern Belgium.”

    Despite total depopulation of Europe for thousands of years, there is thus more genetic continuity between the pre-LGM and post-LGM populations of Western Europe than there is between the Mesolithic and the Neolithic era, or between Neolithic Europe and late Bronze Age Europe.

  7. @SamuelIsaacAndrews

    A couple corrections re Cassidy’s paper. First, it was a PhD thesis from October 2017, not 2020, even though it didn’t see the light of day until 2020, elaborating on the results of a published 2016 paper in which Cassidy was the lead author, but the published paper had fewer samples.

    Second, while it was true that the paper found that “The Mesolithic hunter gatherers were extremely inbred.” it did not say that “They didn’t have Magdalonian ancestry.” Indeed, it basically said the opposite. The abstract from the paper notes that:

    “The Mesolithic population shares high genetic drift with contemporaries from France and Luxembourg and shows evidence of a severe inbreeding bottleneck, apparent through runs of homozygosity (ROH).”

    The samples from France and Luxembourg with which the genetic drift is shared would have included the Magdalonian samples released in this paper.

    The Irish Neolithic samples were all Cardial and not LBK in origins.

  8. @Razib,

    I’m not seeing evidence for very recent selection for pigmentation in Europe if by very recent you mean after 0 AD.

    Most of the selection for lactose persistence was also completed by 0 AD.

    Iron age Ireland circa 500 AD, Bronze age Baltic circa 100 BC, Bronze age Germany circa 1000 BC all had 100% frequency in SLC45A2.

    Also, the Unetice samples from Czech and Poland have a frequency of SLC45A2 at close to 90%. They date to 1800-2000 BC, so a lot of selection was completed by the Early Bronze age.

    Iron age Ireland had 50% frequency of LCT, modern Ireland has 80%. Bronze age Baltic had modern LCT frequencies, as does site in Iron age Poland.

    2500-1000 BC, I think when most selection for lactise persistence and pigmentation happened.

  9. @ohwilleke,
    “France and all but a few refugia in Europe (in Iberia, Italy and the Caucuses) were completely depopulated at the Last Glacial Maximum ca. 20000 years BP”

    So, there are only a few choses of where WHG came from.

    In my opinion, there were two types of WHG. One from Italy, one is from Balkans. The one from Italy carried mtDNA U5b and settled Western Europe. The one from Balkans carried mtDNA U5a and settled Eastern Europe and mixed with ANE from Central Asia.

  10. SAI, the best modern techniques using site frequency spectra show selection continues on pigmentation in NW europe (britain). i can agree more than 50% of the selection on pigmentation (depending on area) was done by 1000 BC. but there is a definite residual, and more than definite in some ways.

    if you question my motives again i will ban you. you are on a short leash. i don’t know if you are a sperg, but behave like a normie for once.

  11. @Samuel Isaac Andrews

    So you are proposing a link between Italian WHGs and the ones that moved into western Europe, presumably being the same ones that are partially ancestral to Magdalenians? Interesting, however I don’t think that the U5a/b thing was all that clear cut. There was some U5b in eastern Europe during the mesolithic. So it was more like Italy + Balkans being the main reservoir, with Italy or at least the Italian group moving into western Europe being 100% U5b and the Balkan group being mostly U5a but with considerable U5b as well, which would make sense since it is closer to the main LGM reservoir and so it would get input from both groups of lineages.

  12. Yes, that is what I am proposing. And it is key point Magdalenians as you mentioned that were already 50% WHG and from the U5b type of WHG. Which means WHG is older 20,000 years old. And by 20ky, had already split into Western and Eastern versions.

    The oldest U5b2b is in Southern Italy dates 19ky but its genome hasn’t been tested. I’m sure it will turn out WHG.

    I suppose you’re right it is simplistic to strictly divide U5a WHG and U5b WHG.

  13. The preliminary data of the (very low-coverage and so tentantive) samples in the Eurogenes Global25 that I saw also showed some somewhat more “southeastern”, in intra-European terms, individuals appearing in the eastern/Alsatian Iron Age Hallstatt-to-La Tene set compared to the Bronze Age Alsatian set. This could be related to more central European kind of ancestry arriving with Hallstatt (whether also Celtic with it or not), but at any rate I have the feeling that, as the authors admit, it would seems hard to tell finer differences apart in this kind of situation.

    There’s less variance on average, though we also see some more “southwestern” individuals in the southern French Iron Age whether due to some migration from Iberia or further local increase in pre-Beaker-related ancestry that don’t seem to appear in their Bronze Age set, and there’s still also that seemingly inconsistent (real or not) easternization that should be clarified with more, and higher quality, IA samples. Especially since the LBA-IA Urnfield to La Tene period in central Europe is important for the most mainstream kind of theory of the spread of Celtic so further clarification is welcome. I don’t think this particular period is illuminated much with this paper but it’ll be interesting if that BA-IA change turns out to be the case with future papers.

    Matt, at first I thought that maybe it’s just an issue with the low coverage but then I also notice that on their PCA there are two sets of red pentagons, where pentagon = Steppe and red = Yamnaya. One is the more outlying usual Steppe_EBA set but the other plots with Corded Ware. Could they have also included Steppe_MLBA individuals as “Yamnaya” in their analysis or was that just for the sake of convenience and simplicity on the PCA, which I assume goes for the Chalcolithic Anatolian individuals that have the same color as the Neolithic ones? This might also explain why some other individuals get no Villabruna but instead are modelled as Yamnaya + Barcin, which you wouldn’t expect based on both PCA position and geographic location, and also the general apparent inflation of Yamnaya-related ancestry. The modelling of the G2a Beaker PEI2 that plots with Neolithic populations seems harder to explain like that though and maybe is just due to coverage.

  14. The Iberian Magdalenian survival paper also showed a tad Magdalenian in Global Amphora Culture, and modeling in the Dzudzuana paper also in Kunda.

  15. The selection for pigmentation is interesting.

    I think I tend to agree with Razib that Scythians weren’t in the range of Central/Northern European pigmentation. We should question their light hair, eyes and “Northern” features too in light of evidence of recent selection and the fact that they were ~20% East Eurasian. (

  16. Off Topic.

    The genetics of Canaan,

    One recent article in Cell

    “The Genomic History of the Bronze Age Southern Levant”

    and from two years ago in:

    American Journal of Human Genetics.

    “Continuity and Admixture in the Last Five Millennia of Levantine History from Ancient Canaanite and Present-Day Lebanese Genome Sequences”

    main author Marc Haber.

    Mr. Razib Khan,

    Some 8-9 years ago when I started to follow your posts, you had quite a few of them related to Jewish genetics.

    Perhaps these 2 recent articles may shed additional light on the topics from long long ago.

    With great Respect,

  17. @Matt: That study proves what I proposed all along, that there was a large scale takeover of a HG clan leading to the spread of I2a and increased WHG ancestry in the Middle Neolithic. If nothing else pops up in Northern Europe (they just notice a stable PWC contribution in the Northern fringe areas), it seems that most if not all of this I2a/WHG expansion can be attributed to Northern Italy-Southern French HGs taking over Cardial cultural elements and wives, forming their own Neolithic ethnicity and expanding from there.

    From the paper:
    “However, the situation is different in today’s France, where we observe not only the highest HG ancestry proportion overall when compared to other regions in Europe but it is also found in the oldest individuals from the southern sites of PEN and LBR. This observation is also supported by uniparental markers. Y chromosome lineages in western early farmers in the southern region are exclusively derived from HG (I2a; table S5 and text S5). In contrast, mitochondrial DNA results show a more universal Neolithic diversity profile, as previously reported [i.e., (4, 5)], with only two haplotypes (U5 and U8) that are potentially of HG origin (table S4 and text S4).”

    As for the French Bronze Age to Iron Age transition, the shift seems to be big enough for the wave like expansion of Iron Age Celts, but more fine scaled analyses are needed, because most of the movements to the West happened on the base of regional BB ancestry with not too much ending up in the far West of what started in Central Europe in LBA to EIA.

  18. @Obs, no they don’t phrase what in they see in the interactions that way, and in fact quite differently.

  19. They speak of peaceful Lala-Land interaction, but they were different people which didnt intermix, but became people apart. For thousands of years after the HG clan Neolithics emerged. And these Southern French expanded North, thats what they write too.
    So its what I’m saying. The only difference is that they don’t explain the different expansions in a meaning full way, they just document it and say its all peaceful interaction, what it wasnt. The European Middle Neolithic was one of the most brutal eras in human (pre-) history. Male clans didnt disappear through peaceful interaction.


Comments are closed.