Monday, October 1, 2018

Greeks in a Longobard cemetery

I designed a new Principal Component Analysis (PCA) to help me test the fine scale genetic affinities of post-Bronze Age ancient samples from Southern Europe and surrounds. Below is a version of this PCA with a selection of the most Southern European-related ancients from this year's Amorim et al. and Veeramah et al. papers (for background reading, see the posts and comments here and here). The relevant datasheet is available here.
A number of people in the comments at this blog and elsewhere were especially curious about the potential genetic origins of the three most Near Eastern-shifted individuals from the Amorim et al. dataset: CL25, CL30 and CL38. Judging from my new PCA, it seems likely to me that this trio came to North Italy from the pre-Slavic invasions Aegean region. In other words, I'd say they're probably Roman era Greeks or their descendants, who, unlike most present-day Greeks, don't harbor any Slavic ancestry. That's because they cluster very strongly with present-day Greeks from Crete, and also more or less sit on a cline running from present-day mainland Greeks to Cypriots.

See also...

Mediterranean PCA update

Celtic vs Germanic Europe


  1. I reckon tied to the Exarchate of Ravenna

  2. That makes sense. It also noteworthy these possible ancient Greeks have much more CHG/IranNeo than the Minoans & Mycenean samples had.

  3. Interesting, CL121 looks more southern here than in Matt's PCA. More Sicilian-like, like they concluded in the paper.

  4. Makes me wonder how people like CL121 ended up in Collegno. Maybe it's related to Justinian's war against the Goths:

    I used to believe that the Sicilian-like and Cypriot-like admixture that can be detected using various methods in modern North Italians is merely the result of North Italian locals mixing with West Asian slaves and immigrants, creating a pseudo-Sicilian/Cypriot signal. But as there were concrete people of this type in ancient northern Italy it may be really ancient Greek and Sicilian admixture.

    For what it's worth, using the likely Burgundian Gallo-Roman CL47 and Sicilian-like CL121, I'm now able to model my own very mixed ancestry with high precision and accuracy:


    Germany_Medieval, 55.3
    Beaker_Northern_Italy, 13.6
    Italy_Medieval_Collegno:CL121, 12.5
    Italy_Medieval_Collegno:CL47, 8.8
    Hallstatt_Bylany, 4.5
    Baltic_IA:DA171, 4
    Samaritan, 1.3

  5. @Simon_W

    The Exarchate of Ravenna was the Byzanthian rules part of Italy that slowly was being conquered or assimilated by the Lombards. Greek migrants came in at least in Sicily during the attacks of the Avars and Slavs in Greece. There is some evidence that they also migrated to other parts of Italy, perhaps even being the source of teh Greek being spoken in southern Italy up until today.*.html

  6. Here's the Global25 overall similarity for these Cretan-like Medieval Italians. Forgot to check this when I was writing the blog post.

    Greek_Crete Italian_Jew Italian_South
    2.219949 2.823237 3.068355
    Sephardic_Jew Cypriot Sicilian_East
    3.081453 3.118157 3.142979
    Ashkenazi_Jew Greek_Central_Anatolia
    3.331129 3.397756

    This basically matches my new PCA, although the useful thing about the new PCA is that it fairly cleanly splits Jews from Southeastern Europeans.

  7. "the useful thing about the new PCA is that it fairly cleanly splits Jews from Southeastern Europeans."

    Yeah, I noticed that. Also a nice visualization against the whole ~50% Levantine + ~50% Italian modelling for Ashkenazis; an idea that just won't die. Proponents never answer where all the other Jewish groups who have never set foot in Italy get all this "Italian" ancestry from...

    Still it remains really interesting what binds these East Med clusters together. Clearly they all share a large chunk of very related ancestry. I've argued before that the base layer is quite old and perhaps even related to Anatolia_Chl, who at ~6000 years old is remarkably close to the modern East Med cluster.

    Greek_Central_Anatolia Cypriot
    3.273982 4.445660
    Greek_Crete Turkish_Trabzon
    4.641108 4.770787
    Greek_Trabzon Armenian
    5.062505 5.668972
    Sephardic_Jew Italian_Jew
    5.701915 5.778981
    Druze Italian_South
    5.845886 5.935526
    Lebanese_Druze Sicilian_East
    6.030226 6.056756
    Georgian_Jew Ashkenazi_Jew
    6.279432 6.280326
    Lebanese_Christian Assyrian
    6.299322 6.406763
    Italian_Abruzzo Georgian_Laz
    6.468251 6.637642
    Lebanese_Muslim Maltese
    6.715616 6.805601

    I think it's hard to untangle as all these groups as they descend from already similar populations that subsequently mixed with each other for millenia... yet the zoomed-in PCA above does a really good job of it, without any of the the usual overlap.

  8. Could the samples which show relatedness to Greeks be representative of the genetic makeup of Northern Italy prior to the Migration Period.

  9. @Steven

    Could the samples which show relatedness to Greeks be representative of the genetic makeup of Northern Italy prior to the Migration Period.

    Impossible, unless the rest of the ancient Italian samples on that plot are mixtures of these Cretan-like North Italians and Germanics, but obviously they can't be.

    These Cretan-like North Italians are migrants from the Aegean, that's why they share so much genetic drift with Cretans and Cypriots.

  10. @Davidski

    Why is it impossible if we don't have any Roman era Italian samples to compare with?

  11. @Steven

    Why is it impossible if we don't have any Roman era Italian samples to compare with?

    Because it would mean two unlikely things...

    - that there was a massive genetic shift from the Bell Beaker period to the Roman Period in North Italy

    - that there was another massive genetic shift during the Roman Period, with lots of Germanic ancestry being spread across Southern Europe even before the Lombards got to Italy.

    But none of this works, because North Italians are largely the descendants of the local Bell Beaker population, rather than a two-way mixture of eastern Mediterraneans and Germanics.

    So if you're hoping that Roman Period North Italians will by and large cluster with Cretans, like these three, then you'll be very disappointed.

  12. This is where the two steppe-admixed Bronze Age North Italian Beakers cluster in this PCA.

    So the chances that North Italians were ever a Cretan-like population are zero.

  13. The genetic makeup of these potential ancient Aegean people fits well with being the source of East Med ancestry in Italians.

    Ancient DNA from Italy, might show people like this migrated in large numbers into southern Italy sometime in the Bronze or Iron age.

  14. @ Dave
    I also think these “Cretans” are just what Bronze Age Italics would look like
    (I state this in peace !).

  15. @Nirjhar

    Bronze Age Italics will look like the Bronze Age North Italian Beakers, and very similar to modern North Italians.

  16. @Davidski
    “Modern North Italians are just local Beakers with a bit of Southeast European admixture.”

    Your recent opinion from anthroforum is interesting. My question is what did that Southeast European admixture look like. On PCA it looks like they could be Greeks but only IE Greeks i.e. those already mixed with Hyperboreans (i.e. Slavic or forest-steppe like populations) not pre-IE Crete or Cyprus like Greeks:

    Where would Mycenaeans like Crete_Armenoi sample be located on this plot?

  17. @EastPole

    Mycenaeans more or less cluster with Sicilians in this PCA, but the result is probably something of a coincidence and isn't informative for them because they're outside the range of modern variation in the West Eurasian plot.

    I guess Crete_Armenoi would cluster between Sicilians and Greeks. But I haven't bothered to check this due to a lack of data.

  18. @ Dave
    Thanks for your reply .
    But what about the Bronze Age migration into Italy from Anatolia and surrounds - the same one which arrived to Greece and Thrace c. 2500 BCE.

    Wouldn’t that more parsimoniously explain the long noted signal in Italians, without scurrying it down to Roman Age, Near eastern merchants, or Byzantine-Greek refugees ?.

  19. @Nirjhar

    I'm not aware of any significant Bronze Age migration from Anatolia to Northern Italy. Nothing like this is hinted in the ancient DNA record.

    But I can clearly see that modern North Italians are, by and large, the descendants of North Italian Beakers, rather than a mixture between Germanics and a Cretan-like population.

    So your theory is hanging on the assumption that this is just a coincidence, created by two massive genetic shifts in Northern Italy: from the Beakers to a Cretan-like population, and back to a Beaker-like population, thanks to Germanic admixture. This really isn't possible.

    Another major problem with your theory is that these three Medieval samples look quite distinct and very much like unadmixed Aegean islanders, so your theory is also hanging on the assumption that individuals and small family groups like this survived in Northern Italy from the Bronze Age into the Medieval Period, just at the time when we should expect migrants from the Aegean in many parts of Italy.

  20. @EastPole

    By the way, your two-way mixture model for modern North Italians, as a mixture between Beaker_Italy_North and Balkan groups isn't realistic, because it ignores Germanic and other types of admixture.

    Have a look where the most northern shifted Collegno and Szolad samples are located, and this should tell you what you're missing. In other words, modern North Italians are not a two-way mixture, but largely the descendants of Beaker_Italy_North with admixture from multiple sources.

  21. @ Dave

    I don’t think i theorised that Italians = Lombards + Greeks, but ok thanks for reply.

  22. @Nirhjar

    I don’t think i theorised that Italians = Lombards + Greeks, but ok thanks for reply.

    But this is the kind of model that you're proposing if those Cretan-like people from Collegno are indigenous North Italians.

    Because how else do you get modern North Italians from them without a massive admixture event with Northern Europeans?

  23. I don’t know Dave but I’m sure we’ll find out :) .

  24. The anthropological type of classical Romans who dwelled in contemporary Central and North-Western Italy has as its source the territory of south-eastern Alps. The inhabitants of southern Italy at that time were the descendants of Ancient Greeks, Illyrians and Phoenicians. These differences between more Celtic (BB) centre and Neolithic South can be traced even today.

  25. Interesting how the BBC left deep genetic traces among some contemporary populations particularly in the north. I was curious though, I remember someone else here saying that Proto-Italics were perhaps best modeled from a two way perspective as a split between the Sardinian like EEF population and perhaps Urnfield culture. I wonder though what results from Villanovan I and Terremare cultures will tell us once meaningful samples are obtained. I believe these cultural contexts fit into the greater Urnfieled system that blanketed parts of N. and C. Italy. A good contrast would be to see how these results would compare to historical samples of early period Osci, Umbri, Roman, and Latini samples among others.

  26. Before the Lombards entered North italy, the lands belonged to ostrogoths.

    Maybe as some scholars state , the Danubian culture is what exists in north-italy...we do know that North-East italy was danubian culture.....north-west italy was something else

  27. On the GedMatch calculators they look very similar to Ashkenazi Jews.

  28. @Steven

    They can't be Jews because they don't show any Jewish-specific genetic drift.

    What you're seeing at GEDmatch is that they're very similar to Ashkenazi Jews in terms of broad genetic components, but this is somewhat misleading.

  29. By the way, those Chernyakov/Ostrogothic genomes from the new paper on the genetic shifts in the Pontic-Caspian steppe are darn interesting, but I can't say anything more. We just gotta wait for the paper...

  30. @Eastpole

    "Hyperboreans (i.e. Slavic or forest-steppe like populations)"

    Do you have any links or other information regarding your Hyperborean theory? Please let me know, thanks.

  31. That is very interesting those Bronze Age Beaker Northern Italians, David. Thanks for posting that. A lot like modern Northern Italians id say but marginally more Northern shifted. What are they dated to? Are they as old as that one Sicilian Beaker? What study do those samples come from? The 2017 breaker study? Anyway thanks for sharing, didn't know they existed. Could be an example of what the earliest Indo-European speakers in Italy were genetically like.

  32. These North Italian Beakers are from the Olalde et al. Beaker paper. Not sure what their dates are.

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

  33. @ epoch

    But the Exarchate of Ravenna, properly speaking, was founded after the Lombard invasion, around 584 AD. The Byzantines had previously reconquered the whole Italy from the Ostrogoths. And the Exarchate was what was left of it, so to say, after the initial Longobard conquests starting in 568 AD. However, the southern outliers from Collegno were locally raised, according to isotopic data, in contrast to many of the Longobards. So people like those outliers were present in Collegno before the Longobards arrived, and thus, before there was an Exarchate of Ravenna.

    I think this article from the Italian Wikipedia on the Gothic war offers some useful insights:'Italia

    The population of the local civilians shrank drastically because of famines and plagues, cities were destroyed in the course of the war, and civilians became refugees roaming around in their own country, a bit like in Syria today. But not all parts of Italy were severely affected, especially Sicily and Ravenna weren't. So maybe there was some repopulation going on in the aftermath of the war that - to some extent - involved Sicilians.

    Another thing also mentioned in that paragraph: The Byzantine rulers were quite ruthless with regards to taxes. To the degree that some people in Sicily, Sardinia and Corsica decided to become tax refugees and fled to the barbarians.

  34. Judging from the modern distribution of the Germanic R1b-U106 in Italy, Longobard admixture seems to be mostly a North_of_the_Po-Thing. In particular there's little reason why anything of this should be present in the Romagna, which was named that way because it belonged to the eastern Romans (Byzantines) for a prolonged period, in contrast to the "Lombardia".

  35. Re: North Italians as basically a continuation of North Italian Beakers, I do think it's a little more complex, though.

    First of all, we know that the true average of the North Italian Beakers was less steppe admixed than the pre-set Beaker_Northern_Italy average from the Global 25 pop average sheet. Because in a sample of three there was one without steppe admixture. This sample size may be too small to compute their true coords reliably, but in any case we know they were on the whole less steppe admixed than Beaker_Northern_Italy.

    Secondly, we know there were later, additional incursions from beyond the Alps into Northern Italy. When nearly all of them can be doubted, at least the Celts most certainly came from beyond the Alps. So this pushed the level of steppe admixture up. And using just the steppe admixed North Italian Beakers here would result in too low Celtic admixture.

    I for one always use the average of all three North Italian Beakers when modelling my ancestry. And it strikes me how yet I don't get any Celtic ancestry in addition to what I got from my German speaking father and grandmothers (12.7% in my latest model), and how my roughly 25% Romagnol ancestry can be modelled very well as a nearly 50-50 mix of North Italian Beaker and CL121. That's suggestive of a rather thorough replacement of the Gaulish Senones and Boii in the low-land part of the central Romagna by the Romans (at that time still mostly of Latin origin) - quite in line with the historical sources. Maybe the early Latins were still North Italian Beaker-like. At least the Roman Age cranial remains around Rome show some affinity to the remains of Iberians from Catalonia. But I'm sure Celts survived and were Romanized in many other parts of Northern Italy.

  36. (de Saint Michel) I am the Y-DNA match to CL121. We created a new branch (R1b-BY197053) on the Y-Haplotree from my Big Y700 results. We also shared the private SNP BY197459.

    I am also a Y-DNA match to a scientific sample called Sardinia 1031, on YFull he's called ERS256993 from Cagliari, Sardinia. We share the private SNP BY197776.

    My origins results on FTDNA: 27% England, Wales, and Scotland, 16% Central Europe, 13% Ireland, 28% Iberian Peninsula, 7% Sardinia, 3% Italian Peninsula and 6% West Slavic.

    Very interested in your comments about my origin.

    Jeffrey Mitchell


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