Sunday, September 16, 2018

Celtic vs Germanic Europe


I have a feeling that ancient DNA from post-Bronze Age Northwestern Europe will be coming thick and fast from now on. To get the most out of such data I've designed a new Principal Component Analysis (PCA) that does a better job of separating the Celtic- and Germanic-speaking populations of Europe than my previous efforts of this sort (see here and here). Below are two different versions of the same PCA. The relevant datasheet is available here.

And here's a Discrimination Analysis (LDA) plot based on the 25 principal components. It further differentiates many of the populations along the east > west cline of genetic diversity.


The difference between the Germanic Anglo-Saxons and the Celtic and Roman Britons of what is now eastern England is obvious. The Anglo-Saxons could pass for Scandinavians, while the Celts and Romans both cluster between the Irish and French. This makes good sense, and is exactly what I was looking for. It's also interesting to see the presumably Celtic-speaking Hallstatt samples from Bylany, Czechia, clustering with the Belgians.

Update 14/12/2019: Pictured below is a new version of my Celtic vs Germanic genetic map. It's based on the same Principal Component Analysis (PCA) as the original, but more focused on Northwestern Europe and produced with a different program.


To see the interactive online version, navigate to Vahaduo Custom PCA and copy paste the text from here into the empty space under the PCA DATA tab. Then press the PLOT PCA button under the PCA PLOT tab. For more guidance, refer to the screen caps here and here.

To include a wider range of populations in the key, just edit the data accordingly. For instance, to break up the ancient grouping into more specific populations, delete the Ancient: prefix in all of the relevant rows. This is what you should see:


Conversely, you can leave the ancient sample set intact and instead reorder the present-day linguistic groupings into, say, geographic groupings. To achieve this just delete all of the linguistic prefixes, such as Celtic:, Germanic:, and so on. You should end up with a datasheet like this and plot like this.

Of course, you can design your own plot by using any combination of the ancient and present-day individuals and populations that I've already run in this PCA. Their coordinates are listed here. Indeed, if you're in the possession of your own Celtic vs Germanic PCA coordinates, you can add yourself to the plot. And if you're not, see here.

It's also possible to re-process PCA data via the SOURCE tab. But I don't recommend doing this with the Celtic vs Germanic data, which are derived from a fine scale analysis and don't pack much variation. On the other hand, Global25 data are ideal for such re-processing. I made the plots below from subsets of Global25 coordinates available in a zip file here. To see how, refer to the screen caps here and here.




See also...

Modeling your ancestry has never been easier

Getting the most out of the Global25

Modeling genetic ancestry with Davidski: step by step

46 comments:

  1. Did you use PCA and LDA for these two different plots in Past3?

    Well, I am definitely interested in this PCA.

    ReplyDelete
  2. This comment has been removed by the author.

    ReplyDelete
  3. Very Intersesting.
    Compared to the ancient Celts, the French are shifted south, to Italy
    At least those of the the database, that makes sense, these being supposed to be from the Lyonnaise region. Lyon being the only real Roman city in Celtic Gaul, in Gallia comata for the Romans

    ReplyDelete
  4. In the second PCA the Iron Age and Roman era Brit look like they cluster close to modern day English. From where in Europe is the Hallstatt sample from?

    ReplyDelete
  5. On your PCA Hungarians look like a mix between Poles and Italians:

    https://i.postimg.cc/DwZDQLKb/screenshot_432.png

    We also know from Cassidy et al. 2015 that Bronze Age Hungarians like BR2 were most similar to modern Poles:

    https://i.postimg.cc/4d80KQ8d/screenshot_434.png

    Bronze Age Hungarian similarity to Poles comes most likely from Lusatian culture migrating south.
    Then Romans conquered part of Hungary and settled it mixing with locals. This looks like the most probable scenario for the formation of Hungarian population. Magyars and Germanics didn’t contribute much.

    ReplyDelete
  6. Does it look like Dutch_South is closer to English_IA than Either Hallstatt or Belgian ? One would suspect Belgian to be closer to Dutch_South ?

    ReplyDelete
  7. Now I wonder if the Flemish are closer to the Belgae of the Roman era ?

    ReplyDelete
  8. You guys can check where all of the samples and clusters plot exactly by plugging the datasheet into PAST.

    I wouldn't rely just on the labels to do that, because their positions are a bit skewed most of the time.

    ReplyDelete
  9. By the way, some of these English samples are from the west, near the Welsh border, where the Celtic cut of ancestry is still very high.

    And I think the level of Celtic ancestry in the Low Countries is somewhat higher than most people realize.

    ReplyDelete
  10. @Steven

    Halstatt should be western Austria no?

    ReplyDelete
  11. @All

    I updated the post with an LDA plot.

    ReplyDelete
  12. EastPole

    The making of the modern Hungarian population is in a very large part later than Roman times. There can be Roman heritage in Pannonia, but what you can see as Italian shift is probably post Ottoman migration from Croatia in a very large extent.
    Also there is a Germanic impact, but it comes from very admixed southern Germans, not Scandinavian-line ones. Indeed, the impact of Iron Age Germans is probably low.

    You should know a bit more about the history of the region.

    ReplyDelete
  13. I guess Germany has a lot of Celtic roots too. Makes sense... there are Celtic place names as far east as Poland aren't there?

    ReplyDelete
  14. I would like to see more clearly the overlap of Anglo-Saxons, modern English and Iron Age Britons. I am interested in seeing which modern English samples plot closer to which ancestral populations. These PCA's are a bit too busy with overlapping blue areas.

    Am I to presume the East/West genetic distinction is related to differing amounts of WHG admixture or is it due to more recent genetic drift?

    ReplyDelete
  15. @Tom Rowsell

    These PCA are mostly about relatively recent genetic drift. Well, a lot of this drift dates back to the Bronze Age (for instance, the Rathlin Irish_EBA samples cluster very strongly with the Irish in this PCA), but I call it recent because it defines the present-day genetic structure in this part of Europe.

    You can have a look how the clusters overlap exactly by plugging the datasheet into the PAST program and changing the colors of each population. PAST is freely available here...

    https://folk.uio.no/ohammer/past/index.html

    I did have a look at the English cluster, and it's an interesting one because it stretches all the way from near the Scandinavians to south of the Irish cluster. But this isn't surprising considering that this English sample set includes individuals from the former Danelaw regions as well as the "Celtic" Cornwall.

    ReplyDelete
  16. @Slomberry,

    "There can be Roman heritage in Pannonia, but what you can see as Italian shift is probably post Ottoman migration from Croatia in a very large extent."

    Not disputing your knowledge of Hungarian history. History offers many possible answers to ancestral origins. It usually isn't conclusive. Ancient DNA is needed in these cases.

    Most, of the non-Lombard samples from the Lombard burial in Hungary are closest to modern Italians. Many of the pre-Slavic, pre-Hungarian people of Hungary may have been Italian-like.

    ReplyDelete
  17. @Samuel Andrews

    One burial site from exactly the region I mentioned. Szólád is in the Balaton (Lacus Pelso) region and that was one main settlement area of the Romans and a region where "Latin" population indeed probably survived into early Medieval times. But that not the case for most of the Carpathian Basin.
    I do not say that there is not some Italian heritage, and there were some immigration from Italy in the 13-14th century too. But the impact is not big enough. This is still a PCA that is specifically made for a purpose (stated by Davidsky) and not very good at discerning sources of southern ancestry. What EastPole saw as Roman time local inheritance is mostly continuous gene flow and population movement from the West Balkan region during Medieval era and even later. Of course that also contain some Roman time Italian, but that is indirect in the Carpathian Basin.
    Also the Lusithanian heritage he suspects is probably simply the impact early Medieval Slavic migration (mostly, but not exclusively) and later Slovak migration from the North.
    He simply places the bulk of the population formation in too deep time in region where much happened since then.

    ReplyDelete
  18. Regarding the English similarity to the Scandinavians, there is no surprise when you consider where the Angles, and Jutes were from. There is little need to invoke the Danelaw to explain northern affinity.

    ReplyDelete
  19. I'd like to be able to plot myself and others on this. I tried using Global 25 scaled and unscaled but it doesn't appear to be compatible with either. How can I do this?

    ReplyDelete
  20. @Davidski
    I have question about origin of West_Polish samples. If it is western shifted subset of Polish samples from older North Euro PCA? Or they are new ones?

    ReplyDelete
  21. If you think about it, yes ancient DNA research has exploded in the last several years but in-depth research of modern populations has not.

    I can't think of many studies that have done in depth genome-wide research of modern pops. There was that study on Britain, ireland, Han Chinese, that's about all I can think of.

    The best there is on insightful information for modern substructure are commercial companies like 23andme and GEDmatch tests made by Eurogenes (Davidski). Both focus almost only on Europe.

    So, there's a lot more to learn in upcoming years. Modern DNA will be needed as much as ancient DNA.

    ReplyDelete
  22. This comment has been removed by the author.

    ReplyDelete
  23. @Jen - the Welsh are overwhelming Bronze Age Beaker folks. The lighter hair and skin should come from them too. There's probably some later Hallstatt admixture (David I'd be curious as to your thoughts on this aspect) and a smallish (~10-15%) older Neolithic layer but the bulk is going to be Beakers.

    ReplyDelete
  24. @Jen,

    " I read that they have the "oldest" genepool in the UK per the Welcomb Trust study; what does that mean?"

    Historical records say the Welsh descend from the Bretons. Based only on historical records, they should have the oldest British ancestry.

    Who were the Bretons? That's the next question. Based on ancient DNA, we know for sure they were of mostly British Bell Beaker origin. This means most of their ancestors arrived in a large migration wave that occurred in 2300 BC.

    By, the Bronze age (1000-1500 BC), the average British person had 90% Beaker ancestry and 10% Neolithic British ancestry. Welsh shouldn't have more than 10% Neolithic British ancestry.

    "And are the lighter skin and hair alleles dating from the early bronze Bell Beaker origins, or earlier? Were they Danube migrants or Balkan migrants, who pushed and stayed all the way to Rathlin Island?"

    The samples from Rathlin Ireland were descendants of Beaker folk. They don't originate from a separate population.

    Beaker folk were significantly more fair-pigmented than Neolithic Brits. Fair pigmentation gradually rose in frequency after 2000 BC. Beaker folk introduced typical northern European complexion it into Britain.

    Stuff, like "dark Welsh" could be more difficult to figure out. It could be normal variation in all European populations not unique to Welsh.

    ReplyDelete
  25. This comment has been removed by the author.

    ReplyDelete
  26. @Jen
    If I am remembering correctly past papers, there were different waves of Anatolian farmers, with different HG admixture. In any case, from results published for Central France, where Mediterranean and Danubian waves rejoined, both populations looked extremely similar.
    It is no surprise to find Belgian (Wallons?) close to Hallstatt, and to Celts from the Isles. BB inheritance and common cultural background for hundred of years. But I would like to have Flemish labellised as such, as from what I've seen, they are often very close to South Dutch, with Wallons nearer to East French.

    ReplyDelete
  27. Samuel,
    The version usually taught in Britain is that the Bretons were Romanized Britons (hence their name) who fled Britain to escape the Anglo-Saxon invaders.

    ReplyDelete
  28. On further reading, I guess you've confused 'Breton' - the name given to natives of Brittany, France, with 'Briton' the Celtic pre-Anglo Saxon inhabitants of the island.

    ReplyDelete
  29. @Ryan

    There's probably some later Hallstatt admixture (David I'd be curious as to your thoughts on this aspect) and a smallish (~10-15%) older Neolithic layer but the bulk is going to be Beakers.

    I don't yet have an opinion about the Hallstatt stuff pending more Iron Age samples from Northwestern Europe, and also France.

    ReplyDelete
  30. Interesting plot! But in Past I get errors because of the : in the first row and than the coloring, previous attempts with Global 25 datasets in Past went well...

    ReplyDelete
  31. Fair enough. I imagine even with samples it will be pretty hard to tell too given how closely related the groups are.

    ReplyDelete
  32. Nice plots, but it would be helpful if your data points and outlines consisted of different colors and symbols. It is somewhat difficult to see which outlines correspond to which ethnic groups.

    ReplyDelete
  33. Based on archaeology I predict that there is a Celtic-like to Germanic-like population shift about midway through the Nordic Bronze Age (ca. 1200 BCE +/-) that will be revealed in ancient DNA.

    ReplyDelete
  34. @andrew

    Would you care to elaborate on that? What archaeological evidence, for instance?

    ReplyDelete
  35. https://www.biorxiv.org/content/early/2018/09/20/422295

    ReplyDelete
  36. Very interesting. The cline from Dutch_North over Dutch_Central and Dutch_South to Belgian looks beautiful! Dutch_North being close to Germany_Medieval and Belgian close to Hallstatt... a perfect continental European Germano-Celtic cline, you might say.

    Germany_Medieval looks firmly nestled in a Germanic triangle there, between Dutch_North and Scandinavians.

    The Anglo_Saxons appear distinct, with a shift towards Iceland; the pre-Danish Jutlanders were apparently more Northwestern than modern Jutlanders - which makes of course sense, because the Danes came in from the east.

    Also obvious how the modern and ancient Insular Celts are set apart from the continental Celts. This is well reflected in the craniometric affinities, too BTW. The insular Celts differed from their continental brethren by having a lower cranial index and a higher facial index, among other things.

    And it's also noteworthy that there's also a clear cranial difference between the Anglo-Saxons and the Saxons from northern Germany, precisely in line with these genomic observations: The Saxons from northern Germany were close to the other migration era and early Medieval Germanic tribes from Germany, while the Anglo-Saxons were set apart and closer to some western Scandinavian samples.

    ReplyDelete
  37. As for the French: Not sure if that French sample really comes from the Lyonnais (@ Helgenes50); I always assumed it was a random collection of French people. France is diverse, no doubt, but on the whole the French average is surely more southern than the Belgians, and it doesn't make much sense to ascribe this to the Roman influence.

    23andme does find some Italian autosomal admixture in France, but nowhere more than 1-5%:
    https://www.eupedia.com/europe/autosomal_maps_dodecad.shtml#Italian

    Nor does it make sense to think that, say, the Occitan region used to be Belgian-like in Celtic times, and got shifted towards Iberia/Basques because of Roman admixture.

    So the explanation is probably rather that there was variation among the Gaulish Celts themselves, because the various Indo-European population waves got progressively diluted with farmer ancestry towards the south/Southwest.

    I would also expect pockets of stronger farmer ancestry in refuge areas even more to the North, in woody mountainous areas. And mind you, BR2 from LBA Hungary was also relatively low on steppe ancestry, being kind of French-like in that respect.

    ReplyDelete
  38. @epoch

    1100 BCE is widely viewed as the beginning of late Nordic Bronze Age in Southern Scandinavia based upon archaeological differences in goods and objects that shift at this time period. Most suggestively cremation begins to appear in Southern Scandinavia and many metal objects related to horses are found starting at this point in time.

    In my view, the early Nordic Bronze Age was Bell Beaker in deep origins, and the late Nordic Bronze Age transition reflects the arrival of Eastern Indo-Europeans with Corded Ware roots for whom both cremation and a fixation on horses are litmus tests that appear all over Europe and Asia when these IE populations first arrive (the one outlier is that the Etruscans also cremated). In contrast, the Bell Beaker folk were not horse obsessed and did not cremate and were probably, at a minimum linguistically different from the Eastern Indo-European people, either being a different branch of IE languages or a non-IE language entirely (e.g. Vasconic) who probably also had different religious ideas than the Corded Ware people as well.

    The late Bronze Age transition is a better fit to the apparent shallow time depth of the Germanic languages than the early Nordic Bronze Age, the timing coincides with a generalized collapse of Bell Beaker aligned cultures to Corded Ware aligned cultures in Europe around the time of Bronze Age collapse, and if Urnfield is proto-Celtic or proto-Italo-Celtic, or pre-Celtic, the transition that gives rise to Germanic in the Nordic area has a parallel counterpart in Urnfeld/Halstatt in the South that gives rise to the Celtic languages of Europe which also have a shallow time depth (too shallow to date back to Bell Beaker). Davidski's latest post shows genetic support for the Southern part of that transition.

    ReplyDelete
  39. @Simon-W interesting thoughts! But may be I'am going to disturb this image:

    "Very interesting. The cline from Dutch_North over Dutch_Central and Dutch_South to Belgian looks beautiful! Dutch_North being close to Germany_Medieval and Belgian close to Hallstatt... a perfect continental European Germano-Celtic cline, you might say.

    Germany_Medieval looks firmly nestled in a Germanic triangle there, between Dutch_North and Scandinavians. "

    " And it's also noteworthy that there's also a clear cranial difference between the Anglo-Saxons and the Saxons from northern Germany, precisely in line with these genomic observations: The Saxons from northern Germany were close to the other migration era and early Medieval Germanic tribes from Germany, while the Anglo-Saxons were set apart and closer to some western Scandinavian samples."


    "And it's also noteworthy that there's also a clear cranial difference between the Anglo-Saxons and the Saxons from northern Germany, precisely in line with these genomic observations: The Saxons from northern Germany were close to the other migration era and early Medieval Germanic tribes from Germany, while the Anglo-Saxons were set apart and closer to some western Scandinavian samples"


    Look at my and my parents results, we are from 'outmost' North Dutch heritage.

    My result:

    https://www.mupload.nl/img/o1pdgr7t9qced.png

    Mother from Drenthe ('Saxon')
    https://www.mupload.nl/img/p6rv21.png

    Father from Groningen ('Friso-Saxon')
    https://www.mupload.nl/db/k83zqifoxhgw

    We come close to the early Anglo-Saxons and Scandinavians. All three in the range left of the North Dutch on the way to the Scandinavians.

    So how can I combine this with your thoughts that the Saxons of Northern Germany in this respect are different from the Anglo-Saxons.

    As fare as I know there were two major streams/influxes to the North Dutch area. First Chauci/Saxons especially in the fifth century. And second more Scandic kind of tribes in the beginning of the sixth century (especially in Westergo/Friesland).

    I guess that much of the North Dutch in the PCA are in general sense 'above the Rhine', that's not always the outmost part and may be more of 'general' Dutch ancestry.

    And for what it's worth, my personal 'cranial abbreviation' is a bizygomatic width of 160 mm.....;)

    I'm curios about your vision about this all.

    ReplyDelete
  40. @Simon W:
    "23andme does find some Italian autosomal admixture in France, but nowhere more than 1-5%:"

    Doesnt 23andMe claim their admixture predictions are "recent" (less than 500 years), wich makes it possibly based on "chunks" (pieces of DNA where several thousand alleles in a row are at least half identical) instead of a allele frequence based concept like "admixture" or PCA.

    They dont state the 500 years there anywhwere but say recent. From my memory, several yerars back it was claimed "chunks" beeing identical can be false positive if they are older than 500y years. But I recall to have seen (here on Davids block) that chunks distribution apears to make sense (mimicking the language groups for example) even if they are the size they would be after 5000 years. And only then, sink into noise.

    Of course, I saw maps of the "recent" 23andMe anchestry on Eupedia and these maps dont look recent to me at all. At least most of them dont.

    the Scandinavian one could be recent as it states that there is almost no Scandinavian anchestry in Britain. wich we know is impossible, except, if the map truely shows only younger than 500 years migration.

    But some other maps.... apear to mimmick Hallstatt or stuff like that. But of course one cant rule out that the old migration routes are still used til the modern day. lol

    ReplyDelete
  41. @ andrew

    Nice theory, but cleaarly wrong:

    Before there was (purely indirect!) Bell Beaker influence in Northern Europe, the Corded Ware was already present there, in form of the Single Grave and Boat Axe cultures. The latter are merely historical special names for the local variants of the Corded Ware.

    And as you might know, the Norse branch of R1a, Z287, is attested in Denmark already in Corded Ware times, well before the first R1b is attested.

    That's the one thing I would criticize on you theory. The other huge problem being: The Corded people inhumated their dead, they did as a rule not practice cremation! Their typical burial rite in central Europe included the orientation of the dead body along an east-west axis, the females on their left side with the head towards the east, the males on their right side, with the head towards the west, both sexes looking towards the south. In lesser Poland and the Ukraine on the other hand, the dead Corded people were oriented along a North-south axis, but also with the females on the left side and the males on the right side, with both sexed facing towards the east. And by the way, that's rather similar to the standard burial rite in the steppe admixed eastern Bell Beakers, they also oriented their dead along a North-south axis, the difference being that they had the males on the left side and the females on the right.

    Moreover you're clearly misguided by the Name "Urnfield culture". That's a historically transmitted designation for a certain complex of cultures, but cremation alone isn't its decisive feature. Because cultures who predominantly cremated their dead already existed long before the Urnfield and Unetice cultures. In the Carpathian Basin, cremation was already widespread and common in many cultures during the EBA! And as a matter of fact, there was no Corded Ware in the Carpathian Basin, only Bell Beaker. Cremation was predominant in the following EBA cultures of the Carpathian Basin: Nagyrev, incrusted potters aka North Pannonian culture, Nyirseg-Zatin, Hatvan, Vatya (the inhumation graves where we got DNA from are mostly from late Vatya) and Vattina. Cremation and inhumation were equally common in Verbicioara and Kisapostag. These are all EBA cultures who date long before Unetice and Urnfield.

    ReplyDelete
  42. Ermm correction sorry! They date long before the LUSATIAN culture and Urnfield, of course. Unetice was EBA too, but an inhumating EBA culture, at that.

    ReplyDelete
  43. @ weure

    Thanks for the info, that's indeed interesting. (And BTW funny how you end up more northern than either of your parents - the fascinating effects of uneven inheritance!) The thing is, I don't remember which sites the Saxon cranial series from Germany were from. I think one may have been from Anderten near Hannover... In any case the tribal/linguistic area of the Saxons in early Medieval Germany extended quite far to the south, roughly to the Harz mountains and Göttingen, and it included Westfalen:
    https://justpaste.it/7dav0
    That may make the difference - so maybe those close to the North Sea coast were similar to your family and more Anglo-Saxon-shifted.

    ReplyDelete
  44. @ Fanty

    Yes, somewhere on their site (I don't recall where) I also read they aim at a time depth of about 500 years or so. And indeed, their method is based on the classification of short alignments of SNPs. They even phase the genomic data, if the customer provides a kit of a parent or a kid as well. This explains their greater accuracy and reliability compared to some other competitors. Otherwise they do some pseudo-phasing based on statistical data, which is much less reliable.

    I think they still suffer from "falses positives" in the sense that short chunks by chance can lie close together and are merged into long chunks by the algorithm. But then again I don't consider this to be much of a problem, because the short chunks are real nonetheless. And all our recent ancestry is built up from ancient ancestry anyway. Their new "ancestry timeline" feature however suffers a lot from this problem, and I don't take it very serious.

    A thing to keep in mind though, is that they don't try to elucidate the deep origins of different types of chunks. They merely look in which countries they are typical now. Hence "French&German" chunks are just chunks that happen to be typical for France and Germany, whereas "broadly Northwestern" or "broadly southern European" chunks are those that are common in several different countries. And as the West Germanic tribes for example migrated a lot, broadly Northwestern DNA is better correlated with West Germanic influence than the French & German category, for instance.

    ReplyDelete

Read the rules before posting.

Comments by people with the nick "Unknown" are no longer allowed.

See also...


New rules for comments

Banned commentators list