search this blog

Friday, September 15, 2017

Modern-day Greeks & Italians vs Mycenaeans

What are the historical and linguistic implications of these qpAdm mixture models, apart, of course, from the most obvious? Please share your thoughts in the comments below. By the way, I tried a wide variety of ancients only models for the Greeks and Italians and these were statistically the most sound. If you're wondering who the Roman outlier is, see here.

Minoan_Lasithi 0.780±0.044
Srubnaya 0.220±0.044
P-value 0.909333794
chisq 7.595
Full output


Iran_ChL 0.090±0.071
Mycenaean 0.478±0.103
Slav_Bohemia 0.432±0.077
P-value 0.461783732
chisq 12.820
Full output

Anatolia_BA 0.239±0.057
Iceman_MN 0.332±0.054
Unetice 0.429±0.030
P-value 0.764439946
chisq 9.112
Full output

England_Roman_outlier 0.118±0.115
Mycenaean 0.521±0.147
Unetice 0.361±0.059
P-value 0.741956816
chisq 9.402
Full output

Bell_Beaker_Germany 0.222±0.077
England_Roman_outlier 0.210±0.134
Mycenaean 0.567±0.163
P-value 0.504442682
chisq 12.285
Full output

England_Roman_outlier 0.216±0.121
Mycenaean 0.503±0.135
Unetice 0.281±0.056
P-value 0.808464904
chisq 8.516
Full output

See also...

Ancient Greeks and Romans may have imported a whole new genetic cline into Europe (or not)

Steppe admixture in Mycenaeans, lots of Caucasus admixture already in Minoans (Lazaridis et al. 2017)


Richard Holtman said...

As always, great find and work Davidski. I much appreciated your efforts.

Ygor C.S. said...

Interesting results, thanks. More and more I wonder: considering the influx of CHG ancestry especially in Southern Europe and the linguistic landscape even during the Iron Age, and seeing by the results how close Tuscans are to ancient Mycenaeans, what if Lemnian was not actually an offshoot from Etruscan-speaking polities, but rather a remnant of pre-Greek Southeastern Europe that also expanded through Italy? I've read some not entirely fanciful hypotheses for Tyrsenian-related borrowings into Hittite, Luwian and another ancient languages of Asia Minor. It seems at least possible that Etruscan, Lemnian and Rhaetic were the few remnants of a CHG+EEF dominant Southeastern Europe before Indo-Europeanization.

mickeydodds1 said...

Just out of interest how do modern Italians compare to ancient Romans, that is 'real' Italianate-stock Romans of the Republican and Imperial eras, as to be found in remains from the Roman catacombs or the ruins of Pompeii?

Thank you.

Arch Hades said...

Should have tried some models on Sicilians that included the Canaanite genomes.

Gioiello said...

@ Ygor C.S.
"It seems at least possible that Etruscan, Lemnian and Rhaetic were the few remnants of a CHG+EEF dominant Southeastern Europe before Indo-Europeanization".

I exposed this hypothesis to the same great linguist Schrjiver, who thought that EEF spoke Hattic, but he didn't reply.

aniasi said...

I have a question. This isn't about archaeology, but how does Unetice fit in genetically with the IE expansions. Is it a late western extension of the Corded Ware people, or something different?

EastPole said...

“I have a question. This isn't about archaeology, but how does Unetice fit in genetically with the IE expansions. Is it a late western extension of the Corded Ware people, or something different?”

Early Unetice is probably a mixture of Corded Ware and some Danubian cultures. Early Unetice can be confused with Corded Ware:

Later also Bell Beakers got involved as BB in Poland cooperated with CWC and UC:

Samuel Andrews said...

Great post. Using D-stats provided by David lots of people on this blog including myself have gotten the exact same results for Italy and Greece.

Basically, it looks like Italy is a three-prong post-Neolithic mixture of native Neolithic Italians (with some Paleo Italian admixture), newcomers from northern Europe, and newcomers the Near East (Anatolia and the Levant).

Greeks are probably more or less the same mixture but ancient Mycenean genomes suggest much of the northern European-like admixture arrived recently with Slavs.

Samuel Andrews said...

Each ancestral component in Italy, except for the Levantie-stuff, was super rich in EEF ancestry. In the end, Italy is mostly one thing: EEF, but lots of it didn't arrive with the EEF Neolithic wave into Italy.

andrew said...

So, the the British Roman outlier seems like he's basically Egyptian with sub-Saharan African admixture, or maybe something similar from the North African holdings of Rome like Carthage, right?

Samuel Andrews said...

Btw, ancient Egyptians had sub-Sharan admixture but less than modern Egyptians.

Skordo said...

Do you know what part of Greece the modern Greek is from?

Mr Snow said...

Gioiello you said Pinhasi has 50 samples from ancient Italy does that include Roman Italy? When is that going to be released?

Anthro Survey said...


Bergamo makes near perfect sense. Very intuitive. An endogenous pool of EEFs, admixture with Central European Bronze Age IE-zers, and some additional "Etruscanate" influx. Then again, maybe Anatolia_BA is a composite of smth more Armenia_BA-shifted and later Roman-era minor Syrian influx in this case.

The Tuscan model is more abstract, though, because the endogenous EEF contribution and any Etruscanate ancestry is certainly bundled in "Mycenaean".

Sicilians are tricky and require North African samples to work better. That SSA must be accounted for.

Also, did you try to model Italian_South? If so, I'm guessing it had a pretty decent "Roman_outlier" signal. said...

Greeks in this experiment are non Anatolian Greeks from Thessaloniki. I would like to know how Greeks from Thessaly and Central Greece would look on that.

Anthro Survey said...


Want to hear your opinion about something.

Before I do, let me ask you this to make sure we're on the same page: do you agree that Tuscans from Arno valley, Ligurians, Lombards, Piedmontese and Rhone valley Occitans have more overlapping looks with Germans, Poles, Czechs, Hungarians, and other C. European groups than North Iberians & SW Occitans do?

If you do agree, to what do you attribute this discrepancy to?

See, in formal models, admixture studies and nMonte, they have comparable Unetice/Beaker/Urnfield-like ancestry. This can't really explain it. Btw, I normally abstain from making assumptions about how phenotype and ancestral components correlate(and Seinundzeit warns us about this regularly), but in this case, we're dealing with temporally proximal ancestry so it's much safer, imo.

In my opinion, the answer has to do with how such Bronze Age Central European groups arrived to SW Europe: trickle flow over 2200 years VS 1-2 sudden, massive migrations.

(Contrary to Nordicist claims, the Germanic contribution to these areas wasn't so huge. Besides, Catalonia was under Visigothic rule, so that part is nullified anyway.)

Skordo said...

Yes or from peloponese or dodecanese

Ryan said...

@Anthro - maybe its more recent central european heritage from La Tene, or just selection? My grandfather had rickets growing up in Czechoslovakia in the 20s/30s. There's still a lot of selection going on for paler skin.

@David - what other populations work best for Tuscans? It would be interesting to see if their West Asia / East Med ancestry could be nailed down.

Gioiello said...

@ Mr Snow

"Gioiello you said Pinhasi has 50 samples from ancient Italy does that include Roman Italy? When is that going to be released?"

A list of the about 1000 samples of aDNA in Pinhasi's lab was published and I think it shouldn't be difficult to find that. Among them 50 samples from Italy from palaeolitic to more recent times. Of course I don't know if and when they will be published. I have no special link with him.

Mr Snow said...

The one named 'Various Italian EN, MN, LN'?
Sounds like you got my hopes up for nothing. Really extraordinary that they've sampled every single shitty irrelevant culture in northern Europe from the past 10000 years yet we've got nothing from Rome.

Anthro Survey said...

Re:Tuscans-- Guessing that a more Armenia-shifted version of Anatolia_Chl and/or Anatolia_BA would work reasonably well for them.

I say more Armenia-shifted because expansion from the KA horizon starting in Chalcolithic times continued to work westward(and south-ward) well into bronze-age times. Hence, I expect than any "Antenorians"(mythical arrivals from Bythinia who landed in the Po delta) or "Etruscans" to have been even more Armenia-shifted than Anatolia_Chl was.
Maybe Aram can comment on this a bit more?

ǵenh said...

Using Greek from Thessaloniki (Greece Macedonia) as sample doesn't make so much sense.

Gioiello said...

@ Anthro Survey
I think that you all know that I studied and am studying the uniparental markers and not the autosome. If I was and am right we'll see if the Harvardian/Stanfordian/Levantinist/Kurganist model will result winning or not. I think that many people stretched the calculators of the autosome for their agenda, and in the past I ironized about the Tolemean theory made with epicicles instead of the linear Copernican one. That said, I may say that we have here (Northern Western Tuscany and Eastern Liguria where we are doing many tests) the oldest haplotypes of many haplogroups (R-M269* to R-L23-Z2110* (mine) to R-L23-CTS9219* and even an R-L277* (more diffused in Armenia and the Caucasus, but our haoplotype is the most strange found so far) for not speaking of all the series of R-L51-PF7589* and above all some subclade sof R-U152 but also R-P312 etc etc).
But so far I haven't for all these subclades the proof that they are born here, but, every sample we test, we find the oldest haplotypes known so far. For that I am waiting that the aDNA from Tyrrhenian Italy is tested and published. Only that may say if I am right or wrong.
Of course the possibility that many of these haplotypes may have come in Middle Ages with German invasions is open, but also about that we have to find the proofs. And also about other previous migrations we have to find the proofs in aDNA.

Anthro Survey said...

Could be attributed to recent La-Tene expansion, but then how do we explain why Brits/Irish are largely a "western version" of Poles in their looks(i.e. very unlike Iberians contrary to false impressions 19th century pseudo-scientists got)? They don't have a lot of recent La Tene ancestry, per history/archaeology. What did happen there, though, was a massive wave of Beaker-like folk who overwhelmed the local EEFs.
I was't talking specifically about pigmentation, but more so about other traits like large, round eyes with prominent irises, thicker hair texture, slower-aging skin and the like. In any case, I've also proposed selection in the past as a possible culprit: Po valley stands in sharp contrast to the near-desertic atmosphere of Iberia, after all.

Steven said...

Where would the Iranian Chl come from in the modern Greek?

Anthro Survey said...


Just to clarify my hypothesis visually(with a reasonably good scale) about where the Antenorians would fit along the ANF-Armenia_BA cline:

ANF/ENF-----Myc/Minoan-----Anatolia_Chl-----hypothetical "Etruscans"----Armenia_BA.


Don't always take this at face value, as I'd written about Tuscans a few comments up. It's just some extra West-Asian signal that didn't get accounted for with the chosen populations. Certainly didn't arrive in Iran_Chl form, but wrapped in a package similar to Armenia_BA, Anatolia_BA or Anatolia_Chl.

Olympus Mons said...

@Mr Snow
Some at anthrogenica seem to think we are the same person! Lol.

Thanks for mentioning "my" Shulaveri Shomu there.

Steven said...

Are modern Greeks really 50% Slavic? I would guess 10-20% would be more accurate.

Olympus Mons said...

@Mr Snow,
Any comments on their known Mtdna?

Skordo said...

I think you're right

Davidski said...


Don't have any Roman samples from Italy yet.

@Arch Hades

I tried Levant_BA, and the models were worse than with Roman_outlier, probably because Roman_outlier has minor Sub-Saharan ancestry that is needed to model Sicilians.

@Skordo, italicroots & ǵenh

I don't know where the Greeks in my model are from. But they are a little more northern shifted than some of the my other Greek sets (which have fewer markers and so can't be used here), and I removed some outliers that clustered with Turks and even Armenians.

@Anthro Survey

The Italian_South set that I have currently doesn't really have enough markers. I might try and combine a couple of sets in the future so that I have an Italian_South set for these sorts of runs.


I ran a bunch of stuff, and what you see here are models with the best fits. Tuscans were difficult to model, from memory, and I'm not sure how else to model them.


The Iran_ChL in the Greek model is either excess Bronze Age Anatolian that just happens to improve the fit because the other references aren't quite perfect, and/or very recent gene flow from Anatolia.

Ryan said...

@Anthro - my guess is the same as yours but it'd be good to see hard data.

M said...

@ Anthro Survey

"The Tuscan model is more abstract, though, because the endogenous EEF contribution and any Etruscanate ancestry is certainly bundled in "Mycenaean".

Tuscans here are the 8 Tuscan HGDP from southern Tuscany, they are the most southern-shifted. In Tuscany there is even an internal cline.

Skordo said...

It depends on what part of Greece though

Davidski said...

@Skordo & Steven

The Mycenaeans we have may not be representative of the Greeks who lived just before the Slavic invasions, and/or Slav_Bohemia may not be representative of the Slavs who invaded Greece.

Then again, they may well be. Impossible to say for now.

Anthro Survey said...


Yeap. Everyone keeps making some big deal about South Italians and Cretans, but that shared ancestry stems from a demographic expansion of Bronze Age Western Anatolians into both places. Those folks would have been essentially Anatolia_Chl+Armenia_BA(and maybe Levant_BA). When mixed with Cretan and Italian EEFs, Minoan-like people were produced. Makes perfect geographic and mathematical sense.

An expansion of Minoan-like people into Italy is dubious, though, because it necessitates a replacement of local EEFs which is nonsensical to say the least and not supported at all by uniparental marker data.


Interesting! How far south? Val di Chiana or Siena-Grosetto border? I'd always suspected there to be an internal cline in Tuscany.

In terms of post-Neolithic ancestry, I've always suspected it's like this:
Viterbo/Umbrians/SouthernMarche>Southern Tuscans>NorthMarche/Romagnoli>North Tuscans(like Gioiello)>Emilians>Piedmontese/WesternLombards>Aosta. Not sure where Veneto falls because they have Germanic influence.

Anthro Survey said...


NO doubt, but your out-of-Italy R1b model does not contradict the Kurgan theory, if I've heard you correctly. The presence of those basal clades in modern day Apuanic territory should merely be attributed to local EEF-WHGs, prior to the arrival of Bronze Age Central Europeans(carrying ~50% Kurgan ancestry). The latter introduced the more downstream clades modal in Western Europeans today. :-) Think of it as a reflux w/respect to R1b.

Anyhow, you do agree with Urnfield/Unetice expansions per archaeology, right? If so, answer my question with regards to how you think they arrived in Italy: trickle flow or a couple of massive pulses.

Also, since M brought up the cline in Tuscany, I was wondering if you've ever done autosomal testing. If you have, obtained your Global10 coords and don't mind sharing them, I'd like to try nMonte. Email me if you're down!

M said...

@ Anthro Survey

I probably did not explain it well. Tuscan HGDP is already the southern border of the Tuscan regional cluster, because it is a sample that comes from southern Tuscany. In fact, hardly a 100% Tuscan goes south of Tuscan HGDP on gedmatch. Conversely, many Tuscans plot more north of Tuscan HGDP, with northern Tuscans often ending intermediate between Tuscan HGDP and Bergamo HGDP. Obviously it's not so automatic, because we are all subject to genetic recombination, we are never exactly the result of 50% of each parent and blah blah. There are central Tuscans who are closer to the Tuscan HGDP, others who are more similar to northern Tuscans. Clearly all this if you have 4/4 grandparent from the same area and you're 100% native.

If Gioiello is from Pisa, he is a central Tuscan.

Ryan said...

@David - Have you tried modeling Tuscans as another Italian population + population X, just to see what distinguishes them from other Italians?

Anthro Survey said...


Yep, this is basically how I understood it the first time! Thank you for further clarification.

As for Central Tuscany, I've always thought it to be a strip running across Livorno, south Pisa province(like Volterra), Chianti, and Arezzo. The heavily settled Pisa proper, together with other communities in the Arno basin like Lucca, Prato, Pistoia and Florence I've always considered to be north Tuscany. Is this technically incorrect?

At any rate, that area did not have a strong Etruscan presence in the past--it was more Ligurian than Etruscan---and was not so densely populated. In the early Middle Ages, the situation reversed and Arno basin became quite the hub(didn't it?). Romagna, on the other hand, experienced both Etruscan and other East Med influences in early/mid antiquity and its relative population density didn't change much.

Because the population density of Tuscany is concentrated around the Arno basin today, it makes sense that the average modern Tuscan will cluster closer towards Bergamo than an average Romagnole will(based on what I've seen thus far).Additionally, j2a appears to attain higher frequencies in Romagna than in north/central Tuscany. This is somewhat subjective, but Romagnoli tend to exhibit more affinity in looks to modern-day Anatolia/Caucasus/north Syria region than the average Tuscan does. More "dinaric" or "east Med" types or whatever they call it.

ǵenh said...

@Anthro Survey - this PCA based on Eurogenes K15 may interest you. Red dots represent the Italian academic samples, the blue dots the average of some Italian regions based on 15 individuals per region, except Trentino, Liguria, Marche and Lazio where the individuals are between 5 and 10.

Gioiello said...

@ All
As many of you asked me about my autosome, I did pretty all the tests, and always I put them at the disposal of the researchers, were they or not citizen scientists.
Not only I did lastly a Full Genome with FGC, which was sent to Emory University for medical researches, but I used only the Y, and initially the mt, through YFull: from that my about 90 private SNPs of my R1b1a2-L23-Z2110-FGC24408, but also a complete mt beyond the FMS of FTDNA with eteroplasmies, before that FTDNA forbade to extract it, but my test was with FGC and I had already had an FMS. Justin communicated to me that a complete interpretation of my medical traits would cost 3000 dollars, perhaps from Harvard, but I am not sure. Of course I didn't spent that money, and I didn't either prothease. So far I don't want to know more about my health beyond what I know.
But I did also deCODEme. We were at the DNAforums time, and some calculator found in me more than 20 % of Ashkenazic ancestry and many thought that I'd have run in the Jews arms, but I said that very likely it was true the other way aound: that Ashkenazic Jews had Tuscan ancestry... and my persecutions strenghtened. They had already become with Rootsweb on 2007. My 23andMe (I appreciated much, in fact through it I found the parents of an American adopted, Ian Logan helped, but he lost the bet with me about her origin: she resulted full blooded Italian after more than a century of migration into the US). 23andMe at the beginning gave me 100% Italian ancestry (very likely because Tuscans were their model), now I am at 67%. I follow it even less now with Poznik at the autosome. Of course I take in no consideration la la land, Gedmatch (of Ted Kandell?): I checked it once and it gave different outputs in differnet moments. I take seriously 23andMe because, when it gave a breaking in one chromosome with my daughter (none with my son), I phased the raw data and saw that my daughter has had two back mutations in a gene of the olfaction: in fact she could make perfumes, as the olfaction was very likely more developed in our hunter-gatherers ancestors. Also Dienekes had my data and published his results. In other words I have a long history beyond me for not believing to the autosome beyond recent times of relatedness.
About Tuscany also the looks of original people are very different: I may say where a Tuscan come from from his looks.
From my analyses I am sure that my mt K1a1b1e was born here, and many mt are very old here, from Palaeolitic. The highest percentage of R-U152 demonstrates that this hg. was here before Etruscans, Ligurians and all the rest, because its percentages are the same in all these territories and also the haplogroups. I have written about my researches thousands of letters, that unfortunately are out now, because Rootsweb deleted the great part, DNAforums, Worldfamilies, eng.molgen are out etc etc, but I have all in my mind and my theory of an Italian Refugium so far stands. We'll see next, but of course I am curious to see aDNA from Tyrrhenian Italy from palaeolitic to to-day.

Anthro Survey said...

If you have your Global10 coordinates, just send them my way.

Ah, this is awesome! Thank you! It's just like I imagined it. The only samples missing are from Romagna(incl. relevant parts of northern Marche) and Emilia. From what I've seen, though, Romagnoles can be confidently placed where Tuscan HGDP samples are. Not sure about Emilians.

-Veneto probably has more variability in this day and age. If we could subtract their significant Germanic ancestry, they'd plot around Tuscans, I reckon.
-Calabria, Campania, east Sicily (and probably Salento and coastal Basilicata) are predictably at the tail end.
-Difference between Abruzzo and south Marche(old Picenum) was a bit of a shocker. I guess Ascoli is not just an arbitrary boundary after all.

Gioiello said...

@ Anthro Survey
If you have your Global10 coordinates, just send them my way".

Of course I dont know it, and here I cannot post an attachment, but, if you send me an address, I may send you my 23andMe data. Somewhere I should get also my Full Genome.

Ric Hern said...

Who were the Epirotes and how did their Giant Cattle for the time end up in Italy/Tuscany ? Romagnola, Marchigiana and Chianina Cattle.....Did some people from Epirus migrate to Italy as well ?

mickeydodds1 said...

So, what role did Unetice play in the ethnogenesis of the 'classic' Italian people and of the ultimate origin of Latin and the Italianiate Languages?

How 'typically central European' was Unetice?, what was the Steppe and WHG component?and how did this all square with the emergence of Rome etc?

Ariel said...

There is a tangible difference in look between Tuscans and northern italians, even when it comes to the ones from Emilia Romagna or Liguria that are on average more mediterranean. I'm from Tuscany, I studied in Turin and elsewhere in the north, going back and forth I was always surprised on how people look and feel different. Let's be clear though, I'm talking about groups, not individuals. Think about a population with 50% blue eyed people, it's going to feel quite different from another population with 25% blue eyed people. And I know that mine isn't a popular opinion here, but that's how I see it anyway.

Simon_W said...

Very strange these results.

22.2% Bell_Beaker_Germany in Sicilian_East and none in Sicilian_West? Archaeologically there was exactly the opposite: a West Sicilian Bell Beaker province and none in the east.

And then a Mycenaean-like majority in Tuscany! But there were no Greeks in Tuscany. And properly Mycenaean-like people can only be ethnic Greeks. I mean, that strong Srubnaya-like admixture in Mycenaeans must have entailed a Greek language. Some ancient authors have claimed that the Etruscans were Pelasgians from Greece, but as they were not ethnically Greek it's unlikely that they were like Mycenaeans. Maybe a mix of Greece_N and Anatolia_ChL or Anatolia_BA would capture them better.

Mycenaean-like ancestry in Sicily makes more sense, as the Greeks occupied large parts of the coastal area and there's also Mycenaean pottery in Sicily. Yet a Greek component of 50% and more looks dubious. And what about the indigenous pre-Greek population, the Sicels, the Sicani and the Elymians? These were not Greeks and they sure as hell were not like pure Unetice and German Bell Beaker either. Were they like Roman outlier? That's unlikely too, as the archaeological connections of BA Sicily are not primarily to Egypt, North Africa and the southern Levant, but to the Aegean and the Cypriot/East Med area.

Moreover I'm surprised by the lack of Hungary_BA admixture in these Italian samples. Both Terramare and Protovillanova were rather from around BA Hungary than from Czechia, Silesia or central Germany. The only culture that has been regarded as Unetice-related by some authors was the EBA Polada culture. Well, the cranial remains from Novilara-Molaroni also resemble the Silesian and Moravian Unetice crania, but this seems to be a very local influence possible related to the Liburnian presence there.

OK I think it's possible that the influence exerted by Terramare and Protovillanova was more cultural than genetic and that the main IE admixture was from the Polada culture.

Simon_W said...

Maybe the Bell_Beaker_Germany in eastern Sicily arrived with the Sicels. Some ancient authors (like Philistus of Syracuse) have claimed that the Sicels were Ligurians. And the serological PCA by Alberto Piazza (I know, very old stuff) shows a connection between Liguria/Piedmont and eastern Sicily in the second component (the peak being somewhere in southern Umbria).

Davidski said...


This analysis doesn't work like you assume. For instance, East Sicilians probably get Beakers instead of Unetice because of their more basal ancestry than West Sicilians, probably due to minor admixture that isn't accounted for in these runs.

In other words, you can''t read the output as literally as you have.

By the way, I actually forgot to try Hungary_BA as a reference, so there's also that.

Simon_W said...

@ Ariel

I know ou spoke quite unspecifically about looks, not about pigmentation, but in fact, large parts of Tuscany have more blond-blue eyed people than most parts of Emilia-Romagna:

Around Forli and Mirandola they're even less common than anywhere in Tuscany, Umbria and Marche.

Simon_W said...

@ Davidski

Alright, good to know.

Davidski said...

@Ryan and Simon

You should be able to do something very similar to what I did using the G10/nMonte (minus PC6) method, and tweak the models to test various ideas.

Your ancient ancestry #1

Simon_W said...

From an archaeological point of view I would see several potential sources of Anatolia_BA-related CHG enriched geneflow to Italy:

- The Gaudo culture of Chalcolithic Campania. Culturally there are relations to the Aegean or coastal Anatolia, or maybe Cyprus. A cranial series from Chalcolithic Paestum clusters with Bronze Age Cyprus. And given the geographical proximity of Cyprus to Anatolia and the Levant I would expect a Minoan-like population early on or maybe even more CHG and Natufian-shifted.

- MBA influence from around Albania-Montenegro-Bosnia in Apulia (must have been like Montenegro_BA)

- Aegean influence in southern Italy during the MBA and especially during LBA. Likely to have brought Mycenaean-like and Minoan-like admixture.

- Aegean and east Med influence in Sicily, especially during the MBA, may have brought Mycenaean-like, Minoan-like and Levantine-related admixture.

- Northern Italy had some Mycenaean presence on the Po delta, and the LBA to early IA site of Frattesina yielded rich Aegean contacts.

- Finally the colonisation of the Padanian plain by the Romans may have brought some admixture from central and southern Italy to the north.

So while there certainly was migration during the Roman age, even from Egypt to Britain etc, I doubt that it was the main cause for the position of Italy on the Europe-to-West Asia cline. Because the Roman roads could be used by everyone, and there is no reason why especially Levantines and Anatolians should have been "pulled" en masse to Roman Age Italy.

Simon_W said...

I forgot of course the Greeks of Magna Graecia in the above list.

Simon_W said...

I also think it's possible that North African-related admixture in southern Italy is partly ancient. Haven't there been proposals about North African migrants being involved in the early Neolithic of the West Med? At any rate, Portugal_BA for instance seems to have slight North African admixture, judging from my nMonte runs with my Global 10 data. When I use Portugal_BA it makes the Mozabite admixture disappear.

Rob said...

What's it about PC6 that elevates (W)HG ancestry in outputs ?
And hadn't that been taken into account with the "weighting" we'd been using ?

Simon_W said...

I suppose PC6 shouldn't be a problem if there are no Euro HG samples used?

Davidski said...

PC6 seems to inflate non-basal ancestry in a lot of samples. But it's all about experimentation, so if things work better with PC6 then by all means.

I don't recommend any sort of weighting.

Gioiello said...

Good to know that:
⦁ id:YF10958ДагестанреспубликаRUS [RU-DA]new
⦁ id:YF10600ARM
⦁ id:YF02895TUR [TR-61]

Anonymous: And say him that his subclade is probable Turkic.
Gioiello "It is a theory of mine, possible, but to be verified".
Anonymous: And his BAM can help us to know about this.
Gioiello "Of course".

Of course that R-M269-PF7562*, as some other R1b subclades migrated from Samara to central Asia, has come back with Turks is a possibility to be investigated, even though a Caucasian origin directly from Samara is also open. Anyway also in these cases these haplotypes aren't the ancestors of the Western European ones.

Alberto said...


"PC6 seems to inflate non-basal ancestry in a lot of samples. But it's all about experimentation, so if things work better with PC6 then by all means.

I run the latest Global_10 sheet without PC6 to compare the results to an older run with the 10 dimensions. Both without applying any weighting.

Global_10 (minus PC6):

Global_10 (all 10 PCs, older run, so missing latest samples):

While many results are quite similar removing PC6 (a slight shift from Iran_N to Kotias), when it comes to S-C Asian populations the results differ greatly. For example:

Brahamin (standard Global_10):
Paniya: 49.3%
Iran_Neolithic: 31.4%
Karelia_HG: 19.3%
Loschbour: 0%
Barcin_N: 0%
Esperstedt_MN: 0%
Kotias: 0%
Nganasan: 0%
Mozabite: 0%
Israel_Natufian: 0%
Dai: 0%

Brahamin (without PC6):
Paniya: 42.3%
Loschbour: 17.3%
Barcin_N: 16.5%
Kotias: 12%
Iran_N: 7.5%
Dai: 3.9%
Karelia_HG: 0.5%
Esperstedt_MN: 0%
Nganasan: 0%
Mozabite: 0%
Israel_Natufian: 0%

To me the first model looks much more sensible than the second. I didn't have time to go through all the populations, though, so maybe someone can spot others where without PC6 the results look significantly better. Otherwise I'd advise not to remove PC6 for the time being.

"I don't recommend any sort of weighting."

I've been running both weighted and unweighted models side by side (probably hundreds of them) and while many times the differences are minor, in other cases weighting does solve some very obvious problems of the unweighted version*. I've seen a few results that look worse with weighting applied, but they're the exception. In general I don't think it matters much, but I do recommend weighting.

Though maybe Seinundzeit can chime in about this, since I read him say that he's applying some scaling that seems to be a better solution than weighting by the sqrt of the eigenvalues.

* Here a weighted version of the same set up above:

Davidski said...

Removing PC6 only improves results for Europeans, probably because it's an European specific dimension to a large extent.

Removing a different PC might improve the results for South Central Asians.

Anthro Survey said...


There would have been considerably more pull factors for Syrians and Anatolians to arrive in, say, Campania than for Gauls or Illyrians and vice versa.

These folks were quite urbanized, spoke Koine and would have shared some basic life outlooks. The Campanian plain would have felt like home to them, unlike Britain(not to mention closer proximity). Moreover, I don't think slaves formed the bulk of these arrivals as some have speculated. Glass-blowers, physicians, dock workers, architects and Christian evangelists are some of the specialties that come to mind.

According to a sentence on this interesting page here, "Greeks"(whoever they were--prolly Hellenized Anatolians) and Syrians comprised half the population of Campania during this time. An exaggeration? Perhaps. It does explain the recurring signal in places like Campania and East sicily, though, that can't be explained away by smth on the Anatolia_Chl---Armenia cline. Smth like Anatolia_BA seems insufficient to explain extra Levantine pull as well. said...


Stick to academic stuff. M'kay?

Davidski said...

Nah, don't stick to academic stuff.

Most academic stuff is garbage. Sad but true. And even a novice should know that.

Simon_W said...

I've experimented a lot using all 10 dimensions and without weighting to model my own Global 10 coordinates with ancient DNA. 1/4 of my ancestry (or 25.7% judging from IBD sharing) is from the Romagna, near and partly overlapping with the spot particularly low on blue eyed blond haired people. (The exact place names being Cesena, Meldola, Montiano and Forlimpopoli.)
The remainder of my ancestry is from the German speaking world, both from the southwest and from East Prussia. nMonte preferred a solution with Minoan_Lasithi:

"Halberstadt_LBA" 31
"Hungary_BA:I1504" 27.7
"Unetice" 16.25
"Minoan_Lasithi" 14.75
"Latvian" 5.2
"Remedello_BA:RISE489" 2.9
"Mozabite" 2.2

This would mean that my Romagnol Italian ancestry (scaled to 100%) would be like:
57.4% Minoan_Lasithi
22.8% BA central Europe
11.3% Remedello
8.6% Mozabite

It's interesting to see that there is no Levant_BA needed here. On the other hand a Minoan-like ancestry of more than 50% looks kind of unrealistic. But this is the version preferred by nMonte, and moreover the 5.2% Baltic admixture (from my East Prussian ancestry) comes close to the amounts suggested by 23andme (5.9%), FTDNA (7%) and MyHeritage (7.1%).

When I delete Minoan_Lasithi I obtain (somewhat rearranged by me; the first and the second numbers in each line refer to different runs with the same set of samples):

"Halberstadt_LBA" 34.9 / 34.85
"Hungary_BA:I1504" 31.65 / 31.75
"Unetice" 15.3 / 15

"Remedello_BA:RISE489" 3.5 / 4.25
"Mozabite" 1.35 / 1.3

"Greece_N" 6.5 / 5.4
"Anatolia_ChL" 2.6
= 9.1 / 8.0

"Mycenaean" 2.7 / 3.2
"Levant_BA" 1.5 / 1.65

Note that the Latvian admixture has completely disappeared here.

So my Romagnol ancestry would be like:

31.1 - 35.4% Greece_N + Anatolia_ChL
28.4 - 29.4% BA central Europe
13.6 - 16.5% Remedello
10.5 - 12.5% Mycenaean
5.8 - 6.4% Levant_BA
5.1 - 5.3% Mozabite

Looks quite sensible I'd say. About 1/3 "Pelasgian"; Remedello and BA central Europe reasonably strong; Greek admixture not excessive; Levantine and NA admixture in the realm of the possible.

When I also delete Unetice in order to get some Baltic admixture back, I obtain:

"Hungary_BA:I1504" 34.3 / 34.75
"Halberstadt_LBA" 33.3 / 32.55
"Latvian" 11.4 / 11.8

"Remedello_BA:RISE489" 2.85 / 2.6
"Mozabite" 1.75 / 1.8

"Greece_N" 2.55 / 2.5
"Greece_N:I2937" 0 / 0.2
"Anatolia_ChL" 4.85 / 5
= 7.4 / 7.7

"Mycenaean" 6.3 / 6
"Levant_BA" 2.7 / 2.8

My Romagnol ancestry would then be like:

28.8 - 30.0% Greece_N + Anatolia_ChL
23.3 - 24.5% Mycenaean
18.3 - 18.7% BA central Europe
10.5 - 10.9% Levant_BA
10.1 - 11.1% Remedello
6.8 - 7.0% Mozabite

Now the Greek admixture looks too strong.

Simon_W said...

@ Anthro Survey

OK; and maybe the Romagna also attracted Levantines and North Africans more than other parts of Northern Italy. There is some historical evidence for this (decline in population numbers followed by people introducing oriental cults and early Christians), but its significance hard to estimate.

Matt said...

Anthro Survey:

In my opinion, the answer has to do with how such Bronze Age Central European groups arrived to SW Europe: trickle flow over 2200 years VS 1-2 sudden, massive migrations.

(British) don't have a lot of recent La Tene ancestry, per history/archaeology. What did happen there, though, was a massive wave of Beaker-like folk who overwhelmed the local EEFs.

I'm not knowledgeable about any of this looks stuff, but on the topic of timescales, based on Martiniano et al 2017 there does appear to be a fairly low level increase of CHROMOPAINTER donation between Spain and Northwest Europe, compared to Italian Tuscan, while Spain against North Italian is even:

(These CHROMOPAINTER analyses could benefit from population substructure. In theory I could do that if I knew where the Spanish samples in their analysis came from).

So there is a question about why this is the case.

Simon_W said...

Re: my above nMonte runs; I think the most realistic solution would be in the middle between model 2 and model 3. This would give me 5.7% Baltic admixture, spot on.

And then the Romagnol mixture would be:

32.1% Pelasgian (Greece_N + Anatolia_ChL)
24.45% northern IEs
12.35% Neolith. substratum (Remedello)
(together 36.8%, mostly Italic and a little Celtic probably)
16.9% Greeks
8.15% Levant
6.05% North Africa

brenna said...

@Anthro Survey

As per Sazzini and al. it's more like Umbrians/Southern Marche>Viterbo/Northern Marche>Southern Tuscans>Northern Tuscans>Romagnols>Emilians>Western Lombards>Piedmontese>Aostans

Veneto cluster roughly like Northern Lombards/Eastern Piedmontese in the South, and with Western Piedmont around its Central-Northern parts.

You can see in one of the charts how the central Po Valley retained more Neolithic ancestry compared to its western and eastern ends, hence its slightly lower Central European input.

If Bergamo, as I suspect, is anything like Brescia, it's probably one of the most Neolithic shifted samples north of the Po.
Cuneo, and then Vicenza, plot closer to Central Europe.


The difference is even starker when you compare Tuscans with Northeasterners from Triveneto. It's not only about pigmentation, they have more robust jaws and squarer heads even compared to the Piedmontese.
Manuela di Centa, Stefano Bizzotto and Fabrizio Nonis are good examples and common types there that you can hardly find elsewhere in Italy.
I feel the difference in look is bigger than what PCA plots suggest. Despite what genome testing says, the phenotype cline goes rather like this (from Central Euro to Med):

- Südtirol/Gressoney Walser
- Aostans
- Trentino/Ladins/Piedmontese Walser
- Friul
- Veneto
- Piedmont/Lombardy
- Emilians/Ligurians
- Tuscans/Romagnols

M said...

@ Brenna

Sazzini doesn't have any sample from Romagna. He has a sample from Bologna (n =208, the largest sample of the paper) and Bologna is in Emilia, even if it's the most southern area of Emilia (the border between Emilia and Romagna is just in the province of Bologna, a few miles east of Bologna).

M said...

@ Simon_W

I doubt that your 1/4 of ancestry from Romagna may reveal something of enlightenment on all Romagnoli.

Gioiello said...

YF10512 Chatalbash/Mudyurov (Pomak), b~1870, Debren, Bulgaria Bulgaria R-L389
13 23 15 10 13 13 11 13 13 13 13 29 15 8 9 x 11 x 14 18 x 14 14 15 18 11 11 x x 14 16 18 16 x x 12 11 11 8 15 16 8 10 10 8 12 11 12 x x 15 10 12 x 13 8 14 22 21 15 x 11 x x 11 12 x 33 15 9 14 11 x x 19 13 10 12 12 10 9 12 11 10 11 11 x 12 14 x 13 10 11 21 15 21 11 22 15 11 15 24 12 24 17 10 x 17 9 11 11
B71238 Chatalbash/Mudyurov (Pomak), b~1870, Debren Bulgaria R-V1636
13 23 15 10 13-13 11 13 13 13 13 29
This sample from Bulgaria (Pomak) has been tested, and, unfortunately, from FTDNA he is tested only for Y12 and at YFull YCAII hasn't been tested, but, having DYS437=14 and DYS448=18, he belongs to the Caucasian cluster with YCAII=23-23. In fact the unique other sample with DYS437=14 instead of 15 is this
161210 Crispin Torres Rivera Unknown Origin R-P25
13 23 16 11 12-14 12 14 11 14 13 30 14 9-9 11 11 26 14 18 27 13-14-14-16 11 11 18-23 15 17 18 16 34-35 11 11 11 8 15-16 8 10 10 8 11 10 12 20-21 16 10 13 12 15 8 12 24 20 14 11 11 13 12 11 12 12
but, having DYS388=14<13, belongs to the Iberian cluster with DYS437=15, thus he had a mutation. From YFull seems that these two clusters (the Western European and the Caucasian ones) separated 16700 years ago, and only the Western European one (the highest variance in Italy) is the ancestor of the subclade with YCAII modal at 19-23.

Gioiello said...

Of course the date of separation from the Caucasian and the Western European clusters at 16700 years ago has to be verified, because many other SNPs Beyond L388 and L389 are in common (as CTS1124, CTS2705, CTS12001 etc,) and the work is not facilitated from the numerous no calls of the Big Y.
Other SNPs in common: CTS3719, CTS6941, CTS7875, CTS8029, CTS8151, CTS8359, CTS8750, CTS10490. It isn't difficult to hypothesize that the separation between the two subclades happened at the end of the Younger Dryas from the Italian Refugium as other haplogroups.

Olympus Mons said...

...Some at anthrogenica are warming up to the idea that the shulaveri-Shomu was the source of CHG into steppe. I will just seat back and wait to the day when "they always knew" it was the shulaveri all along.

Even here at Eurogenes Davidski makes a post such as Two starkly different Neolithic traditions in the Lower Volga basin …. Oh really?! You always knew all along did you?

Anyway, to keep reminding what I long have written :

a. When they were made to flee - by 5000/4900bc in the Eastern shores of the black sea, making the elusive settlements known as the Zakubanye cultures. For a couple centuries the probably were the Slobodnoe, Mehosko. The arrival of the Maykop made them flee again.

b. At the same time one probably see then already in places like Nalchik, the people that brought the package of agriculture, pastoral and general way of living that did not exist in north Caucasus but they have implemented there. by this time already moving up Samara river.

c. So now we can also start to talk about the Prikaspiiskaya culture as one of them. And Prikaspiiskaya as predecessors of Khvalynsky. And this as predecessors of Yamnya.

No shit the Shulaveri as the predecessors of them all.

Olympus Mons said...

... and that Khvalynsky R1b, what was his Mtdna? - H2 (which we know was one of the HG found on the shulaveri of Arknashen. And later Yamnaya also have H15 as they did.

Davidski said...

@Olympus Mons

You're forgetting that...

1) The R1b Khvalynsk guy (sample I0433) was almost 80% EHG, which means he didn't have much ancestry from the Caucasus

2) And the mtDNA lineages from Neolithic Armenia don't match those on the Eneolithic/Bronze Age steppe.

So there goes your theory.

Samuel Andrews said...


I got what look like really good fits for Italy, Greece, and Spain using the Global10 PCA.

Bergamo: 58.3% Barcin_N, 28.8% Yamnaya, 10.3% WHG, 2.2% CHG/IranNeo, 0.4% Natufian.
Tuscany: 53.2% Barcin_N, 23.4% Ymnaya, 8.3% WHG, 6.7% CHG/IranNeo, 4.3% Natufian.
South Italy: 48.5% Barcin_N, 13.3% Yamnaya, 6.7% WHG, 17% CHG/IranNeo, 14.5% Natufian.

Thessaly: 52.3% Barcin_N, 29.3% Yamnaya, 6.5% WHG, 8.2% CHG/IranNeo, 3.8% Natufian.
Peloponnese: 46.6% Barcin_N, 29.3% Yamnaya, 5% WHG, 11% CHG/IranNeo, 8.2% Natufian.
Macedonia: 48.6% Barcin_N, 30.75% Yamnaya, 3% WHG, 12.2% CHG/IranNeo, 5.4% Natufian.

Cantabria: 48.2% Barcin_N, 21.5% Yamnaya, 22% WHG, 1.1% CHG/IranNeo, 3.6% Natufian, 3.5% Morocco.
Murcia: 44.3% Barcin_N, 21.4% Yamnaya, 18% WHG, 4.4% CHG/IRanNeo, 3.6% Natufian, 8% Morocco.
Portugese: 44% Barcin_N, 20% Yamnaya, 19% WHG, 3% CHG/IranNeo, 3% Natufian, 11.8% Morocco.

To get them I modelled the test as a mixture of already mixed proposed ancestors. First I modelled the proposed ancestors as Yamnaya, WHG, Barcin_N, Natufian, and IranNeo/CHG. Second, I modelled the test with those mixed populations. With their results I could get Yamnaya, WHG, Barcin, etc. scores for the test population.

Davidski said...

By the way, OM, just in case the science and maths confuse you: considering this Khvalynsk guy was ~80% EHG, it's extremely unlikely that both his R1b and H2 were from the Caucasus.

Gioiello said...

The SNP V1274/7613234 (G>A), assigned to Mangino (Italy) from FTDNA, belongs also to Puerto Rican HG00640 but not to YF10512, thus separating the Western European R-L389 from the Italian Refugium from the Caucasian haplotype:

HG00640 R-L389* A
YF10512 R-L389* G

Anthro Survey said...


It's more difficult to gauge EEF ancestry from PCAs(unlike post-Neo stuff), but the first figure you showed is somewhat conclusive. Yeah, decreased C.Euro affinity and decreased West_Asian affinity together suggest greater persistence of endogenous Neolithics in Lombardy. This is why I am always so cautious with nMonte and formal models which utilize stuff like Minoans and Anatolia_chl only to remain with <20% local EEF. I always treat these as mathematical (allelic) sums of endogenousEEF+Western Anatolian migrants and never at face value.

I'm willing to bet that this EEF is slightly higher in Alta Padana: Como, Tessin, Varese, Lecco, Bergamo and Brescia/Garda. The climate moderated by lakes in this area is relatively mild compared to the lower plain, which would have been more suitable for Bronze Age invaders instead. Moreover, those areas have plenty of valleys and hills to "hide" from invaders. e.g. Val Camonica. It's also true that you'll encounter more Iberian-looking folks(albeit less pigmented) among these Lombards than in other parts and not so many East-Med or North Euro typologies. E.g: Claudio & Patrizio Sala, Gian Maria Volonte, Demetrio Albertini, Roberto Donadoni, Caravaggio. You really won't find Zucchero or Flavio Delbono lookalikes there.

So, with regards to your phenotype cline: this mostly holds for East-Med/Balkanic/WestAsian typologies, but less so for "West Med" types which don't follow any particular cline. Just a small correction: Po valley--North Tuscany--Romagna/North Marche---the rest in this.

Anthro Survey said...


If I understand correctly and the centimorgan thresholds are relatively low here, results aren't so conclusive. In such a case and given the pops used, these results could simply indicate that Iberians share more "bronze-age invader" ancestry w/Dutch and Brits, while Italy received theirs from other groups.

Besides, instead of using contemporary samples, it'd be better to carry the same analysis out(N. Italian vs Iberian[various]) with the full panoply of Bronze Age groups: from beakers to urnfields to unetice. With that in mind, several runs like this could be made, ratcheting the cM thresholds up with each run. Theoretically, if N. Italian ancestry involves more dramatic infusions, Urnfield, etc.(?) should "stay off the line" and lean in their direction at higher thresholds. Then again, if such an infusion took place before ~2200BC or earlier but Iberia's trickle began a millenium later, results can mask it since more time=more recombination action. I doubt such a turn of events, though.

Send Davidski a request to be plotted on the Global10, in that case. That's how you'll receive your coords.

Olympus Mons said...

a. Mtdna from SSC that we know thus far is not caucasus or not caucasus...its I1, h2, and h15. So that is what i said.

b. Ssc were exogenous to caucasus. Read my last paragraph. So we dont know where they came from. This is not a playstation game with levels. There was lots of different people from lots of diferent places. Currently we know close to nothing.

B. As I keep on stating they share the same lifestyle and traits as the iron gates people. There was something similar in the shulaveri to Ovcarovo gorata and the iron gates, then fikirtepe at black sea shores, then hagoshrim...

So, grow out of that childish view view of levels in a playstation game. Its ok to have ones ancestry from the most historically irrelevant and uneventful region of the planet. It does not day anything about you as a person.

In case you get confused with reality...if the admix you so much promote had anything to it....the region where it was prevalent would be a noteworthy region in human hjstory. In fact one can argue the opposite,can one not?

Vincent said...

Yes Greeks are 43% Slavic and only Bergamo is related to Iceman. Great work! LOL

Amateur backyard genomics with bias don't impress me. Peer-reviewed science does.

Greeks and Italians are on a cline --near Minoans and Mycenaeans-- between Anatolia_N and Yamnaya/CHG/Iran_N. Whatever Levant or African admixture they have is negligible, even Sicilians who have the "most" of all.

Gioiello said...

@ Anthro Survey
Send Davidski a request to be plotted on the Global10, in that case. That's how you'll receive your coords".


Gioiello said...

@ Anthro Survey


I thank you, Davidski, but of course this is Arab to me. Hope that it is useful for you all.

Gioiello said...

From these data it is sure that I am Italian from at least (documented) 1000 years, and probably long before. What that does mean ... I don't know
Italian_Bergamo,0.01804,0.02916,0.00044,-0.00862,-0.00102,0.02386,0.01056,-0.00198,-0.0026,0.0013 Italian_CentralSicilian,0.01395,0.02855,-0.001,0.01025,0.00225,0.0164,-0.0032,-0.0024,0.0042,0.0018 Italian_EastSicilian,0.0143333,0.0286,-0.0009667,0.00935,0.0018167,0.01715,-0.0036167,-0.0004333,0.0041333,0.0016167 Italian_South,0.01462,0.02896,-0.00156,0.01066,0.0034,0.0146,-0.00376,-0.00014,0.0034,0.00222 Italian_Tuscan,0.0173,0.02892,-0.00036,-0.00238,-0.00048,0.01888,0.00508,-0.00048,0.00202,0.00154 Italian_WestSicilian,0.01466,0.02894,-0.00027,0.00715,0.00096,0.01632,-8e-04,-0.00061,0.00181,0.00232

but I am sure that R-V88 and R-L389 etc were in the Italian Refugium in the Late Palaeolitic and expanded with the Younger Dryas.

ǵenh said...

No way north Italians are 50% blue eyed, especially in Turin where north Italians are today a tiny minority. I grew up in north Italy myself. Considering blue, gray and green eyes, at most Piedmontese are 40%, with Emilians and Tuscans who are 31%, Ligurians are 29%. An accurate study of the color of the eyes and hair is that of Livi/ Biasutti, which was realized a few years after the unification of Italy and before the great demographic changes that occurred in Italy because of the migrations from the south to the north and central Italy. Unfortunately, it was based only on males, and if sexual dimorphism exists, regional averages can change. However, the maps of Livi/Biasutti show a clear pattern.

Northeastern Italians can be different also from Emilians and Romagnoli, let alone from people of Turin or Milan where north Italians are today a minority, also because north east Italy is the area that has had less migrations from the rest of the country, compared to other areas of north Italy.

@Anthro Survey - I've already samples from Romagna and Emilia but are still too few. The few Romagnols I have seen are similar to people from Marche. The Emilians are intermediate between Italian Bergamo (North Italian HGDP) and Tuscan HGDP, someone is very close to North Italian HGDP, someone leans towards Tuscan HGDP but I need more samples to make accurate regional averages. However, as you can see yourself, North Italian HGDP (Italian Bergamo) can not fully represent the variability that exists in northern Italy.

>Difference between Abruzzo and south Marche(old Picenum) was a bit of >a shocker. I guess Ascoli is not just an arbitrary boundary after all.

Those are regional averages. I have seen some Abruzzese entering the cluster of the center of Italy but the majority bring the regional average closest to southern Italy. Anyway Italians are on a cline, there is nothing strange.

Matt said...

@ Anthro Survey, I don't know if that would be better (what would be the advantage compared to just repeating Coop et al with finer scale groups?), anyway, unfortunately can't do much of that at the moment, however, what I could do is compare the mean CM length provided by Davidski with the chunk counts of coancestry from Martiniano 2017:

This should provide some info as to whether any populations have an excess of chunkcount compared to length. In theory length indicating more recent coancestry, and/or from a more dramatic population expansion.

Black lines intercept zero, red are bivariate fit.

Notes: 1) Bergamo in Davidski's data is a proxy for NorthItalian here, btw, 2) removed Russians as their CM length is very high with all populations in Davidski's data and I wasn't sure about that.

ǵenh said...


your results based on david's models.

Anatolia_BA 34.2
Unetice 34.1
Iceman_MN 31.7

Mycenaean 50.5
Unetice 38.9
England_Roman_outlier 10.7

Mycenaean 67.2
Slav_Czech 31.1
Iran_ChL 1.7

Results for Greeks and Italians.


Anatolia_BA 49.2
Unetice 34.3
Iceman_MN 16.4

Mycenaean 67.15
Unetice 30.10
England_Roman_outlier 2.75

Mycenaean 62.5
Slav_Czech 29.1
Iran_ChL 8.3



Unetice 43.2
Iceman_MN 34.3
Anatolia_BA 22.4

Mycenaean 52.2
Unetice 45.9
England_Roman_outlier 1.9

Mycenaean 57.4
Slav_Czech 42.6
Iran_ChL 0.0



Anatolia_BA 42.4
Unetice 35.5
Iceman_MN 22.1

Mycenaean 52.70
Unetice 37.85
England_Roman_outlier 9.45

Mycenaean 62.4
Slav_Czech 31.4
Iran_ChL 6.2



Anatolia_BA 66.5
Iceman_MN 18.1
Unetice 15.4

Mycenaean 46.0
England_Roman_outlier 30.1
Unetice 23.9

Mycenaean 78.8
Iran_ChL 12.2
Slav_Czech 9.0



Anatolia_BA 62.1
Iceman_MN 22.1
Unetice 15.8

Mycenaean 54.8
England_Roman_outlier 23.3
Unetice 21.9

Mycenaean 81.10
Slav_Czech 10.15
Iran_ChL 8.75



Anatolia_BA 63.5
Iceman_MN 21.2
Unetice 15.2

Mycenaean 55.6
England_Roman_outlier 23.4
Unetice 21.1

Mycenaean 80.90
Iran_ChL 9.65
Slav_Czech 9.45



Anatolia_BA 57.1
Iceman_MN 21.9
Unetice 20.9

Mycenaean 44.9
Unetice 29.2
England_Roman_outlier 25.9

Mycenaean 75.00
Slav_Czech 15.75
Iran_ChL 9.25

Samuel Andrews said...


I don't see any bias in David's analysis. He has simply posted good fits for Greeks and Italians using only ancient DNA. Yes, with models that assume most Greek's Steppe ancestry is from Slavs will give them super significant Slavic ancestry. The fit works. Maybe it isn't realistic but it works.

And Italians do have significant East Meditreaen ancestry from people similar to Anatolia_BA and Jordan_BA. They aren't a simple EEF, Steppe mixture. Look at the results I posted for Italians earlier in this blog....

Bergamo: 58.3% Barcin_N, 28.8% Yamnaya, 10.3% WHG, 2.2% CHG/IranNeo, 0.4% Natufian.
Tuscany: 53.2% Barcin_N, 23.4% Ymnaya, 8.3% WHG, 6.7% CHG/IranNeo, 4.3% Natufian.
South Italy: 48.5% Barcin_N, 13.3% Yamnaya, 6.7% WHG, 17% CHG/IranNeo, 14.5% Natufian.

Italy has extra layers of CHG/IranNeo and Natufian that can't be explained by Barcin_N or Yamnaya. Neolithic ancestry from Italy is certainly high in most of Italy but probably doesn't reach 50% anywhere except Bergamo. That is unless CHG and Natufian stuff was already in Italy during the Neolithic. said...

"Most academic stuff is garbage. Sad but true. And even a novice should know that."

The original article of Lazaridis et al 2017 has the Greeks from mainland (Thessaloniki, Central Greece and Thessaly) scoring additional 13-18% north east European admixture compared to Myceneans, which de facto invalidates your "results". According to them, Cretans have no additional NE admix which means that they are the most Mycenean-like of the bunch. They did not analyze other Greeks like Pontians, Anatolians, Cappadocians, Peloponnesians,... who make a sizeable part of the Greek population as well. said...

Also the paper of Sarno et al 2017 have Islander Greeks and Calabrians/Sicilians scoring 7% North East Euro admixture compared to 15% of Mainland Greeks and Albanians. Cypriot Greeks have 0% of it. Which means that there is a 7% of post Mycenean North East European admixture in the mainland if we trust Lazaridis et al 2017.

Stamatoyannopoulos et al estimates the slavic input in the Pelopponese of 0-14% averaging out 7%.

Rob said...

7% seems underestimate, even for Peloponnesus. Doesn't really match clues from Y markers and archaeology said...

@Samuel Andrews
"And Italians do have significant East Meditreaen ancestry from people similar to Anatolia_BA and Jordan_BA. They aren't a simple EEF, Steppe mixture. Look at the results I posted for Italians earlier in this blog...."

No they don't. They are just Steppe enriched Minoans -aka CHG admixed Anatolian farmers- with 2-3% of Moorish admix in the far South.

The South Italians from Calabria in your analysis have both more WHG and Natufian admix than old Greeks. In any plot they would also plot in the same place, which invalidates your amateur "results".

You and the rest of the bunch -ancient Romans were just like Danes- are going to feel the pain and finally shut up the fuck up when ancient results from Italy will finally come out.

Rob said...

Yeah, I'm not seeing any North African or Levantine in mainland Italians

"Barcin_Neolithic:I1099" 64.3
"Kotias:KK1" 17.2
"Karelia_HG" 16.95
"Villabruna:I9030" 1.55
"Levant_Neolithic:I1704" 0


"Barcin_Neolithic:I1099" 55.4
"Kotias:KK1" 14.25
"Levant_Neolithic:I1704" 10.3
"Karelia_HG" 10.1
"Iran_Neolithic:I1290" 6.95

Samuel Andrews said...

"You and the rest of the bunch -ancient Romans were just like Danes- are going to feel the pain and finally shut up the fuck up when ancient results from Italy will finally come out."

Lol, I don't think ancient Romans were Danes. That's hilarious. I expect them to be similar to Tuscans but I'm open for other possibilities.

Ain't nothing wrong with having "Near Eastern" ancestry. And it ain't impossible to be European but have Near Eastern ancestry. The pre-historic past is a mystery. Almost anything is possible. There's not much we can figure out with modern DNA and archaeology.

In 2006 people would call you crazy if you said just about all Europeans are 30% Neolithic Anatolian. And then there's some common people who think the Egyptians were the first humans and all of humanity has a common ancestor in the last 10,000 years.

WHat I'm saying is we have no basis to say to any theory on pre-historic origins "Well, that's crazy." ALmost anything is possible for the pre-historic past which modern data has told us almost nothing about.

You seem to think significant and relatively recent Near Eastern ancestry in Italy is crazy based on what you know. But you and everyone else really knows nothing about the pre-historic past until you either travel back in time or collect ancient DNA.

We've already collected ancient DNA and it's pretty obvious Italians do have "recent" Near Eastern ancestry.

Samuel Andrews said...


In PCA and sometimes with formal stats (if you use bad outgroups) Italians can be modelled as Barcin_N+Yamnaya. That's what happened with your model [Anatolia Neolithic, plus 35% (EHG+CHG)]. But when one includes the right outgroups (like a CHG outgroup or a EHG outgroup) this model doesn't work.

Also, ADMIXTURE from Dienkes, Davidski, and others has been showing for years that Italians score in Southwest Asian and Caucasus components which Neolithic Europeans do not score in.

It's time people finally admit there's "recent" Near Eastern ancestry not just in Italy and Greece but also in the Balkans, Spain, France, Western Germany, and even into England. No ancient DNA studies have investigated this issue yet, it isn't an established fact yet, but it nonetheless is a very real phenomenon. The Steppe+EEF+WHG model doesn't work for everyone.

Mr Snow said...

@Samuel Andrews
"I expect them to be similar to Tuscans"
We have a few Etruscans samples, they were much more central European than modern Tuscans:
So we can deduce that the Roman part of Tuscan's heritage was rather South Italian/Sicilian-like.
The reason for that is that there was a large scale invasion in southeast Europe in the bronze age of Mycenaean/Sicilian-like Indoeuropeans directly from the source of Indoeuropean languages, Shulaveri Shomu, instead of derivative Indoeuropean cultures like Yamnaya and Corded ware that were mixed with EHG. Those new Indoeuropean invaders did love taking Steppe wives though, probably the blue eyes that got them charmed.

Samuel Andrews said...

@Mr. Snow,
"Those new Indoeuropean invaders did love taking Steppe wives though, probably the blue eyes that got them charmed."

Seriously, can we please stop with the racisms. Gosh dang it, man!

"Mycenaean/Sicilian-like Indoeuropeans"

qpADM results for Sicily using Mycenaen as a proposed ancestor.

England_Roman_outlier 0.216±0.121
Mycenaean 0.503±0.135
Unetice 0.281±0.056

Also, South Italy results compared to Mycenean results I got using David's Global10 PCA.

South Italy: 48.5% Barcin_N, 13.3% Yamnaya, 6.7% WHG, 17% CHG/IranNeo, 14.5% Natufian.
Mycenean: 68% Barcin_N, 13.4% Yamnaya, 0% WHG, 16% CHG/IranNeo, 2% Natufian.

Roughly similar but Mycenean lacks both WHG and Natufian (yes, Levantie ancestry) which Sicilians/South Italy do have.

Rob said...

@ Sam

It has EHG and CHG there, and my model doesn't imply Italy is simply Yamnaya & EEF.
I have no doubt that some recent Near Eastern admixture exists, and there's nothing 'wrong' with it, because it's always been happening.
But most of it is Bronze Age.

"Hungary_CA:I1497" 60.5
"Armenia_EBA" 19.95
"Yamnaya_Samara" 18

"Hungary_CA:I1497" 65.4
"Yamnaya_Samara" 23.2
"Armenia_EBA" 9.1
"Loschbour:Loschbour" 2

South Italy is even more complex, some recent Near Eastern (still pretty minor), some Bronze Age Anatolia, etc.
But it's still guesswork without aDNA.

Chad Rohlfsen said...

I think there is extra Levantine and North African as well. I'm getting it here. I have a lot of samples most don't. This includes a whole lot of Berbers.

Gioiello said...

David, I thank you, and I saw the interpretation of genh, but the problem is Always the same of ten years ago: Italians are Always believed come from elsewhere, but I am sure (as I was the only one to foresee Villabruna), that R-V88 and R-L389 (and also R-L51) came out from Italy, so we should see how much the other places have of Italian. I am waiting the aDNA. And I have a Full Genome, done at the best with FGC, and if I had the programs and the computers I could run those data, also with the Shi Huang method, but I have done my studies on the uniparental markers and now only the aDNA may demonstrate if I am right or wrong. You have my data, you may use them. If for some test I have to pay that little fee, consider me among your customers. I spent at least 5000 dollars in my tests (and in those I paid for others) thus no problem in paying these few dollars. The last I paid were 37 dollars at Ysq for DYS381, DYS520 and FGC22956 for one (very likely) very old R-U152-L142, the cluster with the DYS385b fractionated, that Argiedude and me discovered, and there is the possibility that he is of German descent found close matches in Poland for DYS385b=14.2.

Ashton D. Barella-Lee writes on a FB thread of mine: On serious Tests, I have no Ashkenazi, but then on others like, I have a small amount. Regardless, we already know full well that I have no Ashkenazic Ancestry. I am a mixture of Tuscan, Laziale, Abruzzese, and Campano, so which sides are being mistaken for the Jewish, I don't know, nor does it really matter, since I already thought that it was pretty well established that Central Euro Jews had at least some South Central European Admixture. As for my mt, which is from my Tuscan side, we see on genbank that I am one of the three sole samples of Basal J2a1* - 14133G found from testing. The other two are a modern Irishman and archaeological dna from Hungary (pre Hungarian). Of the three, my Haplotype is by far the oldest, which so far fits the model of an origin in or around Tuscany or Italy, and a branching out northwards. ( I wote a lot about his mt on Anthrogenica before my banishment).

Samuel Andrews said...


Can you run this qpADM for Tuscany and Sicily using the same outgroups you used for the above outgroups...





Davidski said...

Looks like they all fail.

Ric Hern said...

Italy sure looks like a Prickly Pear. Many Bronze Age and Iron Age invasions as Archaeological and Historical evidence shows. Terramare, Remedello Polada,etc. plus Gauls,Epirotes,Carthaginians,plus the (Roman era importations from all over the Empire) and Germanic Tribes.

It surely looks than more than a days work to pinpoint who came from who....looks like a Pizza to me. Heheheeeh ;)

Harry Gindi said...

I saw a talk a while ago where it was mentioned that strong signals of Steppe ancestry (in the form of R1b haplotype individuals) in Ireland arises in the paleogenetic record (in the samples on Rathlin Island) as early as 2400 BC, and it was hypothesized that perhaps this migration came as a sort of horseshoe shape, from the steppe, down through the near east onto the Aegean coast before 2800BC, then by means of a maritime migration across the mediterranean into either southern France or along the southern coast of Iberia, then up the Atlantic coast, then across the English channel and the Irish sea to make it in time for a 2400BC presence of R1b in Ireland.

If this were the case, perhaps the pre-Doric Invasion inhabitants of Thrace and Greece were already speaking Indo-European languages (pre-proto-Italo-Celtic?), and they were, in Greece, later either displaced or absorbed by a later Indo-European migration from the steppe, coinciding specifically with the expansion of chariotry into the Aegean and the Doric/Mycenean infiltration at some point between 2000 BC and 1800 BC, sort of right on schedule.

I haven't been keeping up with the latest and greatest paleogenetics, but has this idea been refuted? It seems like it might explain the late-departure features of ancient Greek, while also explaining the early departure of Italo-Celtic, the west-east cline of R1b -> R1a, and the fact that Corded Ware/Pre-Germanic didn't extend much further west of Denmark (since they would have been blocked in their advance by distinct horseback-riding steppe peoples who left earlier but took the circuitous

Aegean->Mediterranean->Atlantic maritime route.

Anthro Survey said...


Alright, I ran you using David's PC6 voiding method with a constraint of only 4 populations you see below(Unetice and Levant are averages, mind you). For those interested, the fits for the Italian populations were pretty good----all under 0.35. Gio a bit worse(~0.5), but it's normal since he's an individual, not an average.

On the whole, you are somewhere between the HGDP South Tuscans and Bergamasque. Note that in the case of Bergamasque and yourself, the Anatolia_Chl and Neolithic farmer ratio is almost the same, but the Kurgan component is higher in Bergamasque. In South Tuscany(heartland of Etruria in antiquity), on the other hand, there is a markedly higher Anatolia_Chl:Neolithic ratio but a comparable Kurgan signal to yours.

Unlike genh, I stuck to using Anatolia_Chl, not BA, because it is a North-Western Anatolian sample and most proximal to the hypothesized Etrurian place of origin(central-west "Lydia"). The BA sample hails from a different eco-geographical part of Anatolia.

In reality, I believe the bringers of CHG to EEF-like peoples of Crete, Greece, and Italy were on a cline between Anatolia_Chl and Anatolia_BA, but closer to Chl. Cypriot-like, essentially. Definitely not Minoan-like.

Note also how in Southern Italy, there is no escape(yet again) from Levantine affinity. Like Chad said, there's definitely smth to that.

On the whole, I think this model is a great relative comparison of things, but I'm afraid it undercuts Yamnaya-related ancestry by about 3-5 percentage points relative to the formal models we've seen if we take Unetice=65% Yamna-like.

Iceman_MN:Iceman 46.4
Anatolia_ChL:I1584 27.4
Unetice 26.3

Anatolia_ChL:I1584 41.4
Iceman_MN:Iceman 29.5
Unetice 29.1

Unetice 39.6
Iceman_MN:Iceman 38.7
Anatolia_ChL:I1584 21.6

Anatolia_ChL:I1584 46.9
Iceman_MN:Iceman 26.8
Unetice 13.3
Levant_BA 13.1

Italian_Tuscan Italian_Bergamo Italian_South
0.006962586 0.009262268 0.018573013

Gioiello said...

@ Harry Gindi

If that were true, we should get in the Isles the Eastern European R1b haplotypes, i.e. above all some R-L23 subclades, but in the Isles Reich&Co found the downstream subclades of R-L11. The idea is possible, but haplogroups and times don't fit. That they thought that the path is right, but the origin wasn't in Yamanay but in Italy and older than it is thought.
Of course they presuppose that R-L11 formed during the voyage from Yamnaya to the Isles through the Mediterranean Sea. Of course the sea may not disprove this hypothesis as the aDNA in the Balkans and Hungary demonstrated that this fact didn't happen... so they continue to dream, but I am waiting that aDNA of the Tyrrhenian Italy from Palaeolitic to BA is published.

Ric Hern said...

Harry as far as I know the R1b in Ireland originated in Germany near Quedlinburg and Kromsdorf. Autosomaly they do not have influence from Iberia but rather from Central European Farmers who can not directly be linked to Iberia.

So my personal view is that the R1b of Rathlin originated in the Western Ukraine, pushed down into Czechoslovakia and Hungary by the Corded Ware Expansion or even a bit earlier from where they migrated up the Danube and down the Elbe rivers or following the Carpathians into Central Europe....

Ric Hern said...

@ Harry

It certainly looks like R1b people West of the Dnieper River tended to prefer hilly country and River Islands for defensive purposes. So I think it is reasonable to track their migration all along such type of environment. Foothills of the Alps and Carpathians....

Anthro Survey said...


Thanks again. Yeah, not terribly conclusive, it seems, but leans a bit more towards Spaniards in the length and the last two length/count slides---even for more eastern folks Poles and Germans. Could be that a recent trickle has a slight "edge" over(or is roughly equivalent to) a less recent dramatic expansion when it comes to this. Hmmm.
Why does the plot always have to be so thick? :D

Anthro Survey said...

-Oh, I knew all about the cline. It's just interesting how circa the Abruzzese Apennines environs, there is a particularly sharp gradient: in other words, the rate of change in PCA clustering per diagonal kilometer is highest here. It's also worth noting those "transitional" folks don't constitute a significant proportion of Italy population. Essentially, when this is taken into consideration, we're left w/two main clusters if we exclude Sardinians.
-If those Romagnoli and Emiliani are willing to invest a bit of money and are willing to share, have them contact Davidski for Global10 coords.
You can share any data by emailing me at Actively piecing it together from regions of N. Italy and France.
-Also, like northern or southern Marche? Marche has a clear ethno-linguistic bifurcation going back centuries. Northerners are Romagnol speakers and not "heirs" to the Picenes whose northern limit was Ancona;North "Picene" language wasn't related.

Btw, here is a representative set of Lombards w/Lombard surnames mostly born before the southern migration waves, wouldn't you agree?:
Luigi Riva;Angelo Anquilletti,Giacinto Facchetti; Gian M. Volonte;Giovanni Lodetti, Alessandro Volta;Paolo Pulici;Claudio&Pat. Sala;Gigi Meroni;G. Meazza;Donadoni;Baresi brothers(football);Arturo Merzario; Regazzoni(formula 1);Vittorio Brambilla(formula 1);Pierino Prati;R. Boninsegna;Nazzareno Canuti;Graziano Bini;Aristide Guarneri;Emiliano Mondonico

Essentially, French folks! ;-D Overlap with Brits and Catalans to some extent, as well. Pigmentation-wise, nothing extreme--not like in the Veneto.

Harry Gindi said...


How about a route where they skip the whole voyage around Iberia and moved overland north of the Pyrenees to the Atlantic coast? Also, given the length and distance of the route by water, couldn't the R1b in Rathlin be a result of a founder effect?

It just seems weird to me that if the Irish Indo-Europeans arrived on the way through Corded Ware that they wouldn't have brought corded-ware material culture along with them. It also seems to give a weird and unexpected linguistic phylogeny. My intuition is that linguistic drift, like genetic drift, should arise from periods of isolation. The clustering of Italic and Celtic, despite them being separated by the formidable topographical boundary called the Alps, vs the relative lack of geological formation separating Germania and Gaul would lead me to think that if the Rathlin men came from Ukraine -> Czechia -> Germany -> Low Countries -> British Isles, that Celtic and Germanic should cluster more tightly than Celtic with Italic.

That is, why don't the linguistic phylogenies show Celto-Germanic as a group and the archaeological record show a long band of Corded Ware stretching from Germania through the low countries, into the British Isles and Ireland?

Harry Gindi said...


Could you give me the paper reference on the different R1b clade distribution? Why should we get R-L23 rather than R-L11?

Gioiello said...

@ Harry Gindi

Look at the maps of "Eupedia" on hg. R1b, even though the trees of Maciamo are frequently wrong.

Ric Hern said...

@ Harry

There have been some that proposed a Italo-Germano-Celtic Language. There have been some proposing that Celtic, Italic, Tocharian and Hittite share Archaisms not found in other Indo-European Languages.

Most propose that Tocharian split very early from Proto-Indo-European. This could mean that Proto-Italo-Celtic-Germanic could have also split earlier but Proto-Germanic maybe recieved more input from the Late Proto-Indo-European Corded Ware Expansion than Celtic and Italic.

If this expansions happened within a 500 year timeframe from a Common-Indo-European root area it could be that the Proto-Indo-European spoken in Central Europe were more like Dialects rather than completely different languages.Maybe Corded Ware was in the proses of creating a dialect levelling effect....

Ric Hern said...

@ Harry

So as I see it Proto-Celtic evolved Central Parts of Germany, Proto-Italic in the Southern Parts of Germany and Proto-Germanic in the Northern Parts where Corded Ware maybe had a more significant influence on the Plains.

huijbregts said...

I have a few observations about the dimensions of Global10; my example has been the model of Norwegian.

- Dimensions 4, 6 and 7 show the most variance and structure. For Europeans 4 is nearly identical to 7.
- The most densely sampled clines are the EEF migration and the steppe migration.
- These clines are especially distinct on the plot PC4/PC6. However these clines are not aligned to PC4 and PC6, but are tilted by some 30 degrees.
- Early last year Davidski has posted 'West and Central Eurasian' K9 ('Adventure'). The two major clines are here on PC1 and PC2 and they are nicely aligned with the dimensions.
Even the nMonte estimated admixture percentages are virtual identical to Global10 (at least in the Norwegian example).
- Rotation of the dimensions 4 and 6 by 30 degrees does change the scores 4 and 6, but does not change admixture percentages.
- Dropping columns from the dataset usually does not change admixture percentages. The exceptions are dimensions 6 and 9; the effect of dropping 9 is contrary to dropping 6. A side effect of dropping a dimension is that it seems to increase the overfitting (less degrees of freedom).
- Keeping only the first 6 dimensions does not change admixture percentages

The models have a striking resilience. It is nearly impossible to get deviating admixture percentages. Yet Davidski claims that the results are improved by dropping PC6.
Indeed it takes drastic remedies to change the admixture percentages (again: in the Norwegian example).
I can think of only a few explanations:
1 somehow the model is not linear
2 some weirdo is hiding in the dimensions 11-20
3 the PCA is estimated on a superset of the presented dataset and the subset is not respresentative
4 maybe it is not wise to calculate a 10-dimensional Euclidean distance

Davidski said...

Dropping PC6 does seem to bring the levels of hunter-gatherer ancestry to more realistic levels, especially in Northeast Europeans. The differences are pretty big in the basic three-way model: Yamnaya, WHG and EEF.

I don't know for sure, but this might have something to do with PC6 being heavily affected by modern Northeast European-specific drift. If so, this can't be avoided when creating the PCA at this stage, and not until I have many more high quality ancient samples to work with.

Dropping other PCs doesn't help to improve European models, as far as I can see. And indeed, it seems that dropping PC6 is a bad idea when testing most samples from outside of Europe.

Simon_W said...

@ M

"I doubt that your 1/4 of ancestry from Romagna may reveal something of enlightenment on all Romagnoli."

Well, if you want to call its significance into question because it's just 1/4, then I have to say you're wrong, because my other ancestry is also very well known and when modelling my position with ancient samples there's not room for doubt about what influence must be from where. And the Global 10 coordinates being a PCA it's not susceptible to the merging of different components into other components like admixture calculators are.

But granted, there's always individual variation, so more individuals would be preferrable and then we could work with an average.

I expect even some minor difference between the northern, central and southern Romagna. It's probably a gradient. Unfortunately the Romagna is never sampled in autosomal studies.

Simon_W said...

I did consider the possibility before that I randomly inherited a higher share of my Romagnol grandfather's North African and Levantine admixture than the expected 25.7%. And with the scaling to 100% a smaller deviation could be bloated up. That's possible, but how strong could that effect be? These ancient exotic segments should be very small by now and because of the long endogamy of the population I guess they should be pretty evenly distributed across individuals and genomes.

Simon_W said...

BTW I didn't know that paper by Sazzini et al. It's not a bad effort. From their tree-mix models this one looks best to me:

Because there the Sardinian samples cluster together and are the first ones to branch off, as it should be.

And the other samples are more or less bifurcated into a northern and a southern cluster. I'm wildly speculating now, but I even get the impression Grosseto, Brescia and Padua form a "Proto-Villanovan cluster", while Vicenza, Cuneo and Bologna form a "Terramare cluster". And Ancona is basically southern, even more than L'Aquila, but it receives some northern admixture from the "Terramare cluster". Grosseto on the other hand is pulled southwards by southern, Ancona-like admixture. And Bologna receives some "Basal Italian" admixture, lol. I could imagine that Forli-Cesena might be close to Bologna, but also with an admixture edge from Ancona. Or it might be somewhat outlying, like Pistoia and Perugia.

Samuel Andrews said...

R1b L21 in Bronze age Rathlin Ireland isn't an isolated incident. There was loads of R1b P312 in Germany a couple hundred years earlier. That's where the R1b in Bronze age Ireland is from. It took it a route across mainland Europe from the Steppe to the British Isles. Nothing suggests a route across the Meditreaen sea.

Harry Gindi said...


I can believe the whole version you've stated except to answer the following question: If this were all the case, why don't we see corded ware material culture extending across the entire northern coast of Europe, through to the low countries, and into Ireland. The migration timeframe we need, in order for Steppe people to arrive at Rathlin in time, is the heyday of corded ware. It seems like we would expect to see archaeological findings between 3000BC and 2500BC giving evidence for a corded ware migration westwards. I am not aware of such findings.

Harry Gindi said...


Gioiello just said that it's R-L11 and not R-L21.

Also, the evidence suggesting a mediterranean maritime route is circumstantial, I agree, but there are questions about a northerly route that seem to me unanswered, especially concerning the lack of material culture evidence.

Simon_W said...

I'm now going to say something the Lega Nordicists are not going to like. But I think the lower incidence of light hair in large parts of the Emilia-Romagna isn't the legacy of pre-Celtic people, but is in part caused by the Roman influx after the Roman conquest of Gallia Transpadana. It's pretty obvious:

Pesaro, Rimini, Bologna, Modena, Parma and Piacenza were Roman and Latin colonies, respectively. They're all in the darker area. Ravenna and Reggio Emilia were not colonies and they're lighter. The darker areas immediately south of the Po didn't have colonies, but they didn't have noteworthy other cities either, the cities located there were founded much later.

Also, if you look at the topography:

It's striking how the more hilly areas have lighter hair while the plain has darker hair. Usually older population layers tend to be stronger in the mountains. Because later invaders tend to settle the plains first, they don't head straight to the mountains and hence it's unlikely that the darker hair colour in the plain is a legacy of the Celtic population layer that had absorbed the Etruscans and Umbrians.

And the distribution of agricultural land to Roman settlers, called centuriation, entailed that the Roman influence wasn't confined to the few colonies, but it was all over the place:

Ric Hern said...

@ Harry

If I remember correctly there is some influence from the Schönfelder Group evident in Ireland.....

Harry Gindi said...


I'd be interested if you can give a reference! Thanks!

Simon_W said...

It looks like the Anatolia_ChL/Anatolia_BA admixture in Italy spread both from east central Italy (Marche, especially Ancona) westwards and from southern Italy northwards. yDNA J2 is more common in the Marche than west of it:
And a cranial series from Roman Age Marche clustered rather with Sicily and Anatolia, while another series from Latium clustered with Iberians.

I still think a Cypriot admixture in the Gaudo culture could have brought some Levantine affinity. Both craniometrically and culturally this is quite possible. And Cyprus is almost as close to Syria as it is to Anatolia. No wonder that in the paper by Sarno et al. 2017 the Near Eastern component was nowhere stronger than in Cyprus, among the populations considered.

Vincent said...

@ Samuel Andrews
"I don't see any bias in David's analysis. He has simply posted good fits for Greeks and Italians using only ancient DNA."

Then open your eyes. This whole blog is a bunch unqualified amateurs ignoring the pros and playing with data to get the results they want--like 'more steppe' in some pops and 'more Levantine/African' in others. Then Davidski says "most academic stuff is garbage" because it gives the opposite results that he doesn't like, and everyone agrees with his delusion and keeps doing their own 'research'. LOL

These are the actual facts:

- Sazzini (2016) modeled all Italians as a 3-way mix of Sardinian, Caucasus-Iran and Russian, with the diff being more CHG in the south.

- Ralph & Coop (2013) showed Italy's gene pool goes back to the Bronze Age before the Roman Empire, and recent admix has been negligible.

- Lazaridis (2017) put Greeks and Italians on the same Anatolia -> steppe/CHG cline as Minoans and Mycenaeans, with some even being more 'northern'.

Everything else is BS.

huijbregts said...

@ Davidski

"Dropping PC6 does seem to bring the levels of hunter-gatherer ancestry to more realistic levels, especially in Northeast Europeans. The differences are pretty big in the basic three-way model: Yamnaya, WHG and EEF."

Suppose there is an ANE dimension at PC20.
In the Global 10 this component will be eaten by the Yamnaya dimension PC6.

Davidski said...


Ralph and Coop's results for Italy are often misrepresented, and this is what you did just now.

Ralph and Coop found that Italian genetic structure was likely to be very diverse and complex, probably due to multiple gene flow episodes into Italy.

They certainly didn't find that the Italian gene pool was created during the Bronze Age, but rather that there was no significant admixture from Northern Europe dating to the Migration period, unlike in the Balkans, where it was significant.

And try not to extend yourself about what Lazaridis did and didn't do unless you've spoken to the man. In fact, why don't you e-mail him and ask if his last paper rejected the possibility of recent North African and Levant admixture in Italians?


Suppose there is an ANE dimension at PC20. In the Global 10 this component will be eaten by the Yamnaya dimension PC6.

Perhaps there is, but even if so, this isn't very important, because it's not something that shows up via other methods, and the Global 10 wasn't designed to discover anything on its own, but rather to replicate findings from other, more robust, methods.

So if removing PC6 from the Global 10 improves the results for Europeans in the context of other, more robust, methods, then I think it should be done.

Samuel Andrews said...

"This whole blog is a bunch unqualified amateurs ignoring the pros and playing with data to get the results they want--like 'more steppe' in some pops and 'more Levantine/African' in others."

Show me how David plays with data? He focuses on Steppe stuff because that's his personal preference but doesn't manipulate data. He gets the same Steppe ancestry scores as ancient DNA studies.

It's understandable how you and others get the impression he and many others in the blogosphere have an agenda like to prove Steppe ancestry is superior.I think David puts a lot of energy into pushing the Steppe PIE stuff for purelly academic reasons because lots of "experts" refuse to accept discoveries made with ancient DNA and many in the blogosphere refuse to for ethnocentric reasons (eg, Indians who don't like the idea their language and R1a Z93 are ultimately from Europe).

And I think you just don't like the idea of Levantine ancestry in Italy for some reason. It's not just analysis like qpADM and PCA which show it, ADMIXTURE has been showing it for years. Look at Dodecade K12b. Notice Southern Italians score a decent amount in "Southwest Asian" which is a componnet that peaks in Natufians and modern Levantie/Arabia/East Africa. It's a signal of Natufian aka Levantie-specific ancestry.

Notice I didn't claim Levantine ancestry arrived with ROman slaves or that Italians were Nordicists before then. Don't project racist personas onto me. I'm just going with the evidence, it looks like there's some Levantie ancestry in Southern Italy.

huijbregts said...

@ Davidski
My intention was not to 'discover' new components.
I was wondering whether the discrepancies between qpadm and Global10 might be the result of dropping higher dimensions than PC10.

Gioiello said...

@ Samuel Andrews
" I'm just going with the evidence, it looks like there's some Levantie ancestry in Southern Italy"

R-Y16852 FGC14617 * FGC14610 * Y16852+1 SNPs formed 4600 ybp, TMRCA 3800 ybpinfo

R-Y23838 Y23838 * Y23999
id:YF10966RUS [RU-IVA]new
id:F38 aDNA in Iran from Yamnaya
id:YF04770 [Russia]
R-Y11410FGC14628 * FGC14599 * FGC14623+23 SNPsformed 3800 ybp, TMRCA 1100 ybpinfo
id:YF04142 [Grijalba, Spanish Noble of Visigoth origin]
R-FGC14600FGC14614 * FGC14611 * FGC14600+1 SNPsformed 1100 ybp, TMRCA 550 ybpinfo
R-FGC14600* [Jewish cluster of the Spyra etc.]
id:YF03999DEU [DE-HE]
id:YF03792ROU [RO-BT]
R-Y21258Y21258/YFS515862formed 550 ybp, TMRCA 375 ybpinfo
id:YF06539LTU [LT-KU]

Davidski said...


I was wondering whether the discrepancies between qpadm and Global10 might be the result of dropping higher dimensions than PC10.

I think this is unlikely. More likely it's due to recent genetic drift in Northern Europe. And there's no way around this problem except to use ancient reference samples, which is not possible currently, or manipulate the output to match results from qpAdm etc.

Simon_W said...

I guess in the Forli/Forlimpopoli area there may have been a predominance of Campanians among the Roman settlers. There's no better explanation for the special scarcity of light hair there. Modern Campania isn't that extreme, it even has a rather light haired area in the hinterland. But I'd bet this comes from the Longobards, being centered around Benevento. Calabrian or Apulian migrants on the other hand are not that likely, given the high cephalic around Forli.

Simon_W said...

@ Vincent

It's not even the case that all the pros agree with your simple view. Busby et al. 2015 argued for massive Levantine-, North African-, Cypriot- and Armenian-related admixture in Italy, dating to the Roman age and later. Check Figure S3 for that. I'm not saying they were definitely right, but the findings of academic papers have to be discussed, not uncritically bought.

Simon_W said...

@ Vincent

And the paper by Ralph and Coop that was published in 2012 is silent about the Levant and North Africa. No samples from these regions were considered.

Simon_W said...

It's obvious how certain circles have no problem in assuming "noble" Greek admixture, or from the sturdy CHG from the Caucasus, who lived in the far north of West Asia anyway, but when it comes to the Levant, the fun ends. :D

Vincent said...

When you know how to interpret Ralph & Coop correctly, as scientist Razib Khan does, it shows exactly what I said:

"Then there was Peter Ralph and Graham Coop’s 2013 paper, The Geography of Recent Genetic Ancestry across Europe, which reported lots of deep regional structure across Italy.

This is important because it suggests a local stability to the demographic character of the regions for a long time. Probably earlier than the period of the Roman Empire. Though one can imagine scenarios of demographic replacement which would produce this result, they’re generally less parsimonious than the model whereby modern Italian population structure maintains the general outline it had at the beginning of the Iron age."


The Busby study was kind of a mess, using experimental software and coming up with weird results. The massive Levant/NA admix was in a tiny South Italian sample and also the French sample, but then there was almost none in the Sicilian or Greek samples. None of that makes much sense, and it doesn't agree with other studies like Sazzini that have all South Italians/Sicilians clustering together or this one that says the Mediterranean Sea has prevented recent admixture.

Davidski said...

Make no mistake, my qpAdm models for the Greeks and Italians will look pretty good as more ancient DNA from Greece and Italy is published.

The ancestry proportions will change somewhat, and more proximate sources of admixture will be identified, but you will indeed see that Italians have recent North African and Levant admixture.

You can huff and puff all you like, but at some point you will accept it.

Vincent said...


I don't need to speak to Lazaridis to understand his studies. I guess you do, but that hasn't worked out too well since he basically shot you down when you tried to deny that CHG was similar to Iran_ChL.

Anyways, nobody's claiming there's NO recent North African and Levant admixture in Italy/Greece, only that it's much much lower than what you claim/wish, because that's what the overwhelming majority of evidence says.

You should probably be more worried about how Asian-shifted Polish and other East Euros are. That ANE's a bitch. :)

Davidski said...

Yamnaya doesn't have Iran_ChL admixture. It makes no difference if Jesus himself says that it does if the reality is different.

This is plain as daylight and eventually it will have to be shown in a paper dealing with the topic. And then I'll send you a paper copy so you can swallow it.

In regards to the matter at hand, you should definitely e-mail Lazaridis, because you're claiming that his last paper contradicts my qpAdm models. Clearly, you don't believe me when I tell you that this is not the case, so you need to hear it from the horse's mouth. said...


This coming from the dude who said for years that ancient Greeks were full of steppe and North euro admix... said...

@Simon W

Both Sarno et al 2017 and Fiorito et al 2015 found no evidence of Levant admixture in their IBD analysis. With North Africans it's a different story because they have some ancestry from Roman colonists and expelled Muslims, who were mostly native Sicilian converts.

Anthro Survey said...


I doubt those Roman colonists were Italy_South-like even if they came mainly from South Italy. Samnites soldiers from the Apennine foothills hailed from "warrior caste" backgrounds. This meant an excess of BA Central European ancestry, less Anatolia_Chl-like and a lack of any Levant-shift altogether. Italic languages are clearly in the same family w/Celtic and Germanic. It's also consensus that Italic military traditions cannot be called "Mediterranean" but descend from earlier Central European practices. Terni Culture is an example.

More likely that those colonists from Lazio, Umbria(which included hilly portions of Romagna), Sabinum, Etruria and Samnium resembled modern Umbrians and/or modern Romagnoles/Tuscan_HGDP instead.

Look at the map of centuriation and the PCA genh provided for me earlier. Again, Romagnoles seem to hover around Tuscan_HGDP. Yet, Emilia(Bologna) is considerably far from that, though. Why such an irregularity on a level, contiguous plain with comparable degrees of centuriation? Imho, that reflects long-standing genetic structure in the Romagna which was def subject to east med influences as was much of Italy's Adriatic seaboard. Marche was not necessarily an epicenter of that, though, esp on Hg grounds---as Maju would be quick to point out.
So, what if ancient, pre-Roman Romagnol commoners were like modern ones? Won't be easy to tease apart colonial ancestry. :-(

As for the dark hair maximum circa Forlí? I wouldn't ascribe such a fine resolution to this study. After all, the map is incorrect about Umbria: Tuscans---esp Lucchesi---are considerably lighter than any Umbrians group. Ask Gio.

Making firm precise conclusions about ancestry from generic, singular traits isn't safe. It's subject to confounds and with such traits, existing correlations tend to hold, but over wide stretches of land. Best to look at overlays of several such traits.

Yes, it's quite obvious that Romagna as a whole is the darkest and probably owes it in part to having more "Antenorian" ancestry. After all, plenty of evidence for that influence in pre-Roman times, right? Could also just be a natural gradient, too. Btw, from what I've seen, the peak is actually around Cesena-Faenza. Forlivese tend to be lighter. More importantly, fewer West-Asian looking people there.

Anthro Survey said...


As for older populations strata in the hills---

It's often true, but not always and Italy is a (counter)example where newcomer Urnfielder(?) Italics tended to occupy high ground in southern Italy right off the bat. Langobards followed suit. Perhaps they found climatic conditions suitable to their Central European tastes?

It does seem to be the case, though, that indigenous EEFs presence was higher in the mountains(look at G2a) whereas West Anatolian newcomers and later Greeks settled the lowlands and valleys.

In terms of pigmentation(if the map is accurate), I often find that such alleles can amplify in the mountains given smaller population sizes and drift. E.g: Tizi Ouzou Berbers, Kalasha, Pamiri, Nuristanis, Kashmiri, North Caucasus, etc

Gioiello said...

@ Anthro Survey

Sincerely I am laughing for all this your interest towards Italy!

"As for the dark hair maximum circa Forlí? I wouldn't ascribe such a fine resolution to this study. After all, the map is incorrect about Umbria: Tuscans---esp Lucchesi---are considerably lighter than any Umbrians group. Ask Gio".

You ask, and I answer.

You know that I am studying the uniparental markers, and I didn't reach any sure proof neither about my R1b1a2-L23-z2110-FGC24408: if it is old Italian or if it came from Caucasus even in Middle Ages. So far we have the most part of samples tested in Italy, but I am waiting for sure proofs.

It seems that "Lucchesi" are clearer for their huge Longobard presence, but nothing sure without aDNA or SNPs matches. Sincerely it seems to me absurd that all our R-U152 are come recently to Italy and above all with Longobards.But we'll see next results.

I have alwas thought that darker "Romagnoli" were due to Byzantium and the link with Eastern Mediterranean Sea. But also here we need proofs.

Steven said...

If R1b and I2a2 can be found in the Balkans before the Bronze Age then how do we know that they were brought by the Indoeuropeans?

Simon_W said...

@ Vincent

I'm the last one who would doubt the deep regional structure of Italy. Hence there might be places with Levantine admixture and others without it. Actually, Ralph and Coop didn't even include the few Turks and Cypriots they had into the detailed graphs in their supplement. So the paper is mainly about IBD sharing with Europeans. And what you said about Busby isn't quite right. According to Figure S3 heavy Levantine and North African admixture is also in two Tuscan clusters and even in one North Italian one. And in a South Italian one like you said. It's indeed weird that the algorithm didn't notice any Levantine admixture in Sicilians. I don't quite buy this either. By the way, I don't believe the Mediterranean Sea has prevented recent admixture, that's nonsense. Even in the Bronze Age there were good ships. What did prevent the most recent admixture was the religious barrier: Muslims and Christians as a rule didn't mix. And then the fact that many people prefer to be sedentary and stay in their homeland. While some people may have chosen to migrate they were not numerous enough to affect the large local population of the Roman Age beyond low levels. It may have been easier in the Chalcolithic and Bronze Age, when local populations were smaller.

Simon_W said...

@ Anthro Survey

The maps of light hair colour that I and other users have posted here are in fact based on an enormous piece of work that was accomplished a very long time ago: An examination of about 300'000 Italian army conscripts of 1859 – 1863 by Rudolfo Livi. So I wouldn't call this into any doubt, how tiny it may be. This was a very large sample size from all parts of Italy and, importantly, it was collected at a time when Italians were still very connected to their roots, i.e. it shows the condition of Italy before Italians started to migrate and mix, i.e. it shows the structure when it was deepest. Of course Italy today is no longer like this! People have migrated and mixed and structure has been worn down. And maybe now Forli is lighter haired than Cesena and Faenza, why not? But this doesn't say anything about the past. For inferences about the past it's better to use the old info. And it's also important to keep in mind what each map really shows. The map about the incidence of light hair doesn't show the average hair colour, that is, a place with lots of brown haired people and just a few blondes would score lower in the map than a place where black and blond hair are both common. And then someone has made an arbitrary decision what tones of hair colour are considered „light“, that's somewhat subjective. Judging from the numbers reported, I suppose they used a rather loose meaning of „light“.

Simon_W said...

And yeah, no doubt, the interpretation of these maps isn't that easy. Basically light hair is common in north-central and northeastern Europe, but less common in western Europe or the Celtic fringe of Britain. It wasn't common in Yamnaya either. So a scarcity of blond hair doesn't necessarily mean a lack of north Europoid IE admixture. Rather it means a relative paucity of north-central and northeast European admixture.

Simon_W said...

That paper by Sazzini et al. 2016 in fact gave me a strong hint that the MDLP K23b analysis of my half Sardinian – half Cesenate DNA relative is plain wrong and that the Eurogenes K15 again got it better. Because, as I've mentioned before, I found a DNA relative whose father is from Cesena and the mother from Oristano. Oracle-4 of his MDLP K23b results gave him 50% Kosovar + French Jew + Sardinian as the top result. But Sazzini et al. showed that Oristano is just very Sardinian, there's nothing un-Sardinian about it. It was rather their other sample, from northeastern Sardinia that showed some mainland admixture. Now, according to Eurogenes K15 my DNA relative is like 50% Sardinian + 50% Tuscan. Makes complete sense. So his Cesenate ancestry would be Tuscan-like. But the 4 pop approximation is also interesting: 50% Sardinian + French + Italian Jewish! This seems to suggest the involvement of a Celtic-Gaulish element and a Levantine admixed one. Second best approximation is 50% Sardinian + Cornish + Cypriot. So again a Celtic element, and Cypriots are similar to Anatolia_ChL / Anatolia_BA.

Interestingly enough my own Cesenate admixture seems to be rather poor in Anatolia_BA-like ancestry. My phased maternal half according to Eurogenes K15 is like French + North Italian. The French part comes from my non-Italian Swiss/Southwest German grandmother. Indeed in most analyses I scored relatively low on CHG dominated components, hence my Italian admixture is rather North Italian-like, and I tend to score high in Iberian components, so I would expect a relatively strong EEF legacy.

On a related note, I've just decided to give Ötzi a try and add him on the population sheet, after all he's from northern Italy too. And now the nMonte results have completely changed! Levant_BA and Mozabite have disappeared completely, but also Minoan_Lasithi, they're all gone. It's amazing, now I get rather strong EEF scores from Ötzi:

"Halberstadt_LBA" 30.35
"Hungary_BA:I1504" 22.7
"Unetice" 22.25
"Iceman_MN" 18.8
"Anatolia_ChL" 5.9

"Halberstadt_LBA" 30.4
"Hungary_BA:I1504" 22.55
"Unetice" 21.85
"Iceman_MN" 18.8
"Anatolia_ChL" 6.05
"Latvian" 0.35

without Unetice:
"Halberstadt_LBA" 38.25
"Hungary_BA:I1504" 22.05
"Iceman_MN" 17.2
"Latvian" 11.8
"Anatolia_ChL" 10.7

ǵenh said...

@Anthro Survey - I agree with you if you include also these Lombards as part of the Lombard spectrum.

Giovanni Trapattoni, Gigi Radice, Gabriele Oriali, Luciano Re Cecconi, Rino Marchesi, Pier Luigi Pizzaballa, Angelo Domenghini, Pierino Prati, Giuseppe Dossena, Aldo Boffi, Angelo Longoni, Giuseppe Savoldi, Giuseppe Signori, Pierluigi Casiraghi, Roberto Baronio, Daniele BoneraGiacomo Agostini, Gianni Clerici, Felice Gimondi, Arianna Fontana, Lucia Peretti, Emiliano Brembilla, Luca Tencati, Deborah Compagnoni, Ugo Tognazzi, Massimo Boldi, Mario Monti (half Emilian), Agostina Belli (Agostina Maria Magnoni), Johnny Dorelli (Giorgio Domenico Guidi), Gianni Brera, Vittorio Feltri, Aldo Busi, Mina (Mina Anna Mazzini), Paolo Nespoli.

ǵenh said...


Anatolia_BA 59.6
Unetice 40.4
Iceman_MN 0.0


Mycenaean 44.9
Unetice 39.5
England_Roman_outlier 15.6


Mycenaean 47.2
Slav_Czech 33.2
Iran_ChL 19.6


Anatolia_BA 41.6
Unetice 39.8
Iceman_MN 18.6


Mycenaean 53.15
Unetice 39.80
England_Roman_outlier 7.05


Mycenaean 56.90
Slav_Czech 35.45
Iran_ChL 7.65


Anatolia_BA 49.8
Unetice 35.0
Iceman_MN 15.2


Mycenaean 54.5
Unetice 35.1
England_Roman_outlier 10.4


Mycenaean 60.40
Slav_Czech 29.75
Iran_ChL 9.85

Simon_W said...

In my above nMonte run some of the Iceman_MN is undoubtedly from my German speaking ancestors. It can't be all from my Italian grandfather because then there would be no place left for northern admixture in his genome, which is impossible, lol. I took care of this problem by adding French_South to the population sheet, so that the broadly central European influence is now encircled by Halberstadt_LBA, Unetice, Hungary_BA and French_South. The latter population is especially useful to take the Gaulish gradient from BA central European towards more southern, EEF admixed places into account. The result:

"Unetice" 28.3
"French_South" 26.4
"Hungary_BA:I1504" 21.4
"Halberstadt_LBA" 9.2
"Iceman_MN" 7.4
"Anatolia_ChL" 7.3

Hence my Italian ancestry from Cesena, scaled to 100%, would be roughly:
42.8% BA central Europe
28.8% Iceman_MN (EEF substratum)
28.4% Anatolia_ChL

And actually the Anatolia_ChL isn't even low here. For comparison, Davidski obtained in his nMonte model for Tuscans just 20.9% Anatolia_BA. So it's still all as expected, Cesena still slightly more southern than the Tuscans.

Simon_W said...

Judging from Sazzini et al. 2016, Bologna is also close to Tuscany in the PCA, it's not overly different from the Romagna:

Fiorito et al. 2015 on the other hand showed that Ferrara (in the very north of the Emilia) is rather northern in the PCA and close to Liguria, but rather distinct from Tuscany (mostly Siena):

Simon_W said...

So the bigger difference seems to be rather southern Emilia-Romagna (along the ancient Roman Via Aemilia) vs. northern fringe of the Emilia.

Now, in terms of yDNA Bologna is also very close to Tuscany, as suggested by this neighbour joining tree from Susi Pelotti et al. 2008:

So you could wonder if that special relationship between Bologna and Tuscany is a result of both places having been part of ancient Etruria, in spite of the difference in blond hair frequency (which could be ascribed to different pre-Etruscan substrata).

But I would say both places just happen to be similar mixtures of northern BA European, Anatolian and EEF strains, that's what makes them similar in the PCA, and the similarity in the yDNA is most of all caused by the high incidence of R1b-U152 which isn't genuinely Etruscan, I suppose.

Simon_W said...

I think the basic autosomal structure of Italy can be best seen in the first TreeMix run from Sazzini et al., without admixture edges:

There we see, among other things, a southern cluster and a northern cluster. And Bologna is sort of „off-northern“; it's still rather northern than not, but the most divergent one of the northern samples. It's off-northern in the same way as Ancona is „off-southern“.

And apart from these northern and southern clusters we see the Tuscan samples Grosseto and Pistoia belonging to neither cluster. Grosseto is very basal to everything and Pistoia is even at the basis of the Sardinian branch. In a way it makes sense because Tuscany is central Italy and hence it can be expected to be neither northern nor southern.

Now this TreeMix run with some admixture edges gives us further valuable hints as to how the above structure arose:

There you see Bologna entirely in the northern branch, but it receives strong admixture from a population that's rather basal to all. This is what has pulled it southwards and made it off-northern. I don't think it's far fetched to believe these were the ancient Romans and Latins, as these were from west-central Italy like the Tuscans.

You further see Ancona receiving some northern admixture; this is what made it off-southern.

And you also see Grosseto as being a mix of a clearly northern basis (which explains the high incidence of R1b-U152) that was pulled southwards by some rather east-central Italian admixture.

Simon_W said...

There are several possible reasons why the Romagna at least up to Forlimpopoli is a little more southern than Bologna:

- The land up to Cesena was added to the Roman colony of Rimini early on, in 268 BC. The land up to Forlimpopoli was added shortly later, in 236 BC. So this part of the Romagna was Roman by then, just like Picenum and the Ager Gallicus in the Marche, they were a geographical unit. The rest of the Emilia-Romagna was conquered and settled quite a bit later, in the 170s BC.

- The land up to Cesena had belonged to the Senones, and according to the sources they were treated worse than all other Gaulish tribes, having been so threatening to Rome in the past. The sources speak of expulsion and genocide. Only in marginal areas "archaeological evidence points to the continued presence of Senones; the necropoleis of Montefeltro, Montefortino, Cagli, Serra San Quirico, and S. Paolina di Filottrano show Gallic remains." (source: )

- The lowland in the southern Romagna up to Cesena is narrow, jammed between the Adriatic and the Apennines, and the centuriation was pervasive there. Again in the words of the above source: "The earliest centuriation, located between the Marecchia and Savio Rivers, allows room for 6,000 colonists’ plots, which would not leave space for original inhabitants to remain in the centuriated area." But further north and northwest the plain gets wider, leaving more space for the Gaulish locals.

- Isolation by distance is another reason: Cesena is geographically closer to the Marche than Bologna is.

Simon_W said...

@ Anthro Survey

"It's also consensus that Italic military traditions cannot be called "Mediterranean" but descend from earlier Central European practices."

Well, I'm not overly convinced of this point. I'm far from being an expert on this matter, but I always thought the traditional Italic warfaring style was centered around throwing javelins at a distance, quite different to central European practices. Though I suppose when going back to the Final Bronze Age then we may see more typically northern practices.

Simon_W said...

@ Anthro Survey

"I doubt those Roman colonists were Italy_South-like even if they came mainly from South Italy. Samnites soldiers from the Apennine foothills hailed from "warrior caste" backgrounds. This meant an excess of BA Central European ancestry, less Anatolia_Chl-like and a lack of any Levant-shift altogether."

My old historical atlas indeed shows an expansion of Sabellic tribes in the 5th and 4th centuries BC from somewhere around L'Aquila that also led the Samnites to their seats in the south. But judging from how southern L'Aquila is in the paper by Sazzini et al. ( ) I doubt these Samnites were anything like Tuscan-like even. I mean, L'Aquila is deeply in the Apenninic mountains, it doesn't look likely that it was overrun with Middle Eastern migrants in the Roman age.

But then again this also means that the Roman colonists in the north cannot have been predominantly Samnites or any other Oscans.

Simon_W said...

On a second thought, the TreeMix run from Sazzini et al. with the largest number of migration edges is even more instructive! This one:

I've drawn these admixture migrations into a map of Italy, for a better overview:

It's striking that most migrations go from the north southwards.
More precisely speaking, we see migrations from the north to Ancona/Marche, to Perugia/Umbria and to Grosseto/Tuscany, and the latter goes with lesser intensity also the south. I would say it's very likely that these patterns were caused by the expansion of the Terramare network down the Adriatic coast into the Marche and of the Protovillanovan culture to central and southern Italy.

There is only one large countermovement that goes from central Italy to Bologna/Emilia-Romagna. According to the TreeMix graph it stems from an area inbetween Tuscany and L'Aquila, but closer to Tuscany. I've interpreted it in my map as Roman, which makes sense.

Interesting is also the weaker admixture edge from Sicily to Grosseto. Is this real, maybe caused by an exodus of Sicilians during the Muslim conquest, or is it a consequence of direct Levantine and North African admixture? It's possible that this sort of movement also went into the Romagna.

Simon_W said...

Very instructive is also the TreeMix run with international populations, again with the full number of migration edges:

- It shows that North Italians are Bulgarian-like, I'm interpreting this as Balkanic with northern admixture.

- Central Italians are basically Iberian-like (slightly closer to Sardinia) with West Asian admixture that has a considerable Levantine pull.

- South Italians are basically Cypriot-like, and hence essentially Anatolia_BA, but with some northern admixture.

Recalling that ancient Etruscan genomes seemed Bulgarian- and Iberian-like in a PCA this makes perfect sense. So it strongly looks like the Anatolia_BA and Levantine admixture in Italy was not associated with the Etruscans. I'm starting to think that the Etruscans might really have arrived in form of the Protovillanovans; Lemnian might be derived from a west-east movement of early Etruscans (after all they also sailed to Campania). And the Italics might be derived from the Terramare culture and/or the Laterza culture.

Simon_W said...

In my above nMonte runs nMonte preferred to use Minoan_Lasithi to represent my east Med admixture, and this made the Levantine admixture disappear. But even more than Minoan_Lasithi it preferred to use the Iceman, which also made the North African admixture disappear. Both preferences by nMonte were of course due to better fits. But the question remains if these better fitting models are also more realistic... After all, Ötzi is older than the Remedello samples and from deep within the Alps, which are a refugium. I would expect that Remedello is a better proxy for the last mainland Italian EEF remnants. And so maybe what really happened is some Levantine and North African gene flow to the Romagna, maybe from Sicily like the above TreeMix run suggested it for Grosseto.

Simon_W said...

Correction: The Bulgarian-like position of the North Italians in the international TreeMix graph is probably a consequence of a similar EEF + Anatolia_BA + IE Steppe composition. And given the 23.9% Anatolia_BA David received for Bergamo, the Etruscans may still have been associated with Anatolia_BA admixture; I don't believe this could be all from the Romans.

And these are 4mix results of my 50% Romagnol + 50% Sardinian DNA relative:

It's evident that the best approximations are reached with the addition of some "Italian_Jewish". It works better than anything from Sicily. That's a hint for some ancient Levantine admixture in the area. There also seems to be some ancient South Italian admixture involved, but how much exactly is hard to tell as long as there's a complete absence of relevant ancient Italian DNA. The only thing that seems certain from Sazzini et al. being that the Emilia-Romagna is a mix of regular North Italian ancestry (Etruscan-Celtic) + considerable Roman admixture.

Simon_W said...

It very much depends on what the pre-Roman Gaulish inhabitants of the Romagna were like. If they already were North Italian-like (as a consequence of strong Anatolia_BA admixture maybe from the Etruscans) then the Romagna would be ca. 60% pre-Roman. However, if they were French-like, then the Romagna would be about 40% pre-Roman.

Roberto Mocci said...

What I've always found weird is that Sardinians have much less bronze age Levantine admixture than other South Italians, despite Sardinians having direct contacts with Levantines since the bronze age and despite the many Phoenician cities of iron age Sardinia.

Some of the archaeologists leading the excavations in Sardinia had already said that the majority of the population of the Phoenician cities in Sardinia was most likely native to the island, now science seems to support that thesis;

But I wonder why Sardinians adopted Phoenician culture en mass like that during the iron age without mixing much with the Phoenicians.

Simon_W said...

Yes, good question!

BTW, I think I found a perfect nMonte model for myself which seems to imply that the particularly dark area from Forli to Cesena is mostly a mix of Ötzi-like farmers (Terramare-Palafitte no doubt) with a strong Aegean wave of migrants (Tyrsenians?) with little Italo-Celtic admixture:

"England_Anglo-Saxon" 50.25
"Halberstadt_LBA" 0.9
= 51.15% = my Germanic ancestry

"Iceman_MN" 13.7
"Remedello_BA:RISE486" 1.25
= 14.95% = my Terramare-Palafitte substratum of the Po valley

"Anatolia_ChL" 6.2
"Minoan_Odigitria" 3.1
= 9.3% = my Tyrsenian (?) ancestry from the Aegean

"England_Roman_outlier" 0.6
= MENA admixture

"French_East" 17.3
= my Gaulish ancestry

"Latvian" 5.1
"Lithuanian" 1.6
= 6.7% = my Prussian Baltic ancestry

It's quite possible that with the advent of the Terramare-Palafitte culture from the circum-Alpine area older Remedello-like people were quickly outnumbered. Or maybe they were always more Ötzi-like in the eastern Emilia-Romagna than in Remedello di Sotto which is more to the northwest. And in this model Anatolia_Chl is stronger than the Minoan-like stuff - fits with a Tyrsenian origin in the northeastern Aegean. And together the Ötzi-like component plus the Tyrsenian influence make up almost all my Italian ancestry (which would be 25.7%). This explains why the Forli-Forlimpopoli area is so dark. They were just a pocket of Etruscan remainders where the Celts were rare. Actually from Cesena itself no Celtic finds have been excavated. And the Anatolia_Chl influence on the whole isn't low in this model, definitely stronger than in Tuscans. The MENA admixture is nearly identical to what I got on 23andme. The East European/Baltic admixture too. In sum: This must be it!

Roberto Mocci said...

There is no archaeological evidence for a migration from Western Anatolia to Italy in the late bronze age, just saying, I'm very skeptical about that theory.