I have been debating whether or not to make this a long post in the huge Western Jews’ mega thread I’ve started more than 3 years ago, which has gone a little bit dormant in the last several months (mostly since it’s been pending ancient samples we’ll hopefully get one day from upcoming studies, as has been rumored and whispered), or to start a new thread; and I feel that this kind of bold claim deserves its own thread, so people that haven’t taken part in that long mega thread will not hesitate to join in the discussion.
I’m writing this at this time since there’s seemed to have been a slight revival of the good ol’ “half n’ half” model for Ashkenazim. The half n’ half is a nickname I’ve given a while back to the widespread notion, both in academics and both in the Anthrofora that Ashkenazi Jews are basically a 50:50 percent mixture of Middle Easterners and North Italians (with some minor North European and North African). This notion has been raised first in Atzmon et al. 2010, repeated by Zoossmann et al. 2010, then the famous Costa et al. 2013, and last, but not least, by Xue et al. 2017 (which the latter also giving a very historical problematic dating for said admixture – Christian North Italy circa mid 8th century CE).
Recently, renowned blogger Razib Khan also raised Xue et al.’s premise in one of his latest feeds (no criticism here Razib, if you’re reading this, I love your blog and your work; if anything, your revival of that papers’ premise was one of the things which encouraged me to start this thread!), thus this made me realize, that while there hadn’t been a new Jewish DNA paper in several years, there had been several ancient DNA papers about Italy. In fact, there had been 4-5 different papers with tons of ancient DNA from Italy, stretching from Neolithic all the way to Medieval (and later).
And while we did discuss several of them in a glimpse in the Western Jews’ mega thread, I think it’s long overdue to finally try to look at all these new ancient samples from Italy and try see if the “half n’ half” model, which might have seemed plausible a while back, still holds.
Why North Italy seems logical
Before I’ll try to refute the half n’ half model, I’ll try to explain why it seems, on the face of it, logical and why many people are ready to accept it.
Well, as it has been shown for the past 20 years or so, the paternal ancestry of Ashkenazim seem to lay in the Middle East, and some would argue the Levant (though jury is still out on how much specifically from the Levant, and how much from elsewhere).
However, Ashkenazim seem to plot on most PCAs, as well as show close genetic distance, to South Italians and Sicilians.
As there are several Jewish historical sources foretelling the migration of few Rabbinic families from North Italy to the Rhineland around the 9th century CE, coupled with the historical narrative of the expulsion of Jews from Judea by the Romans, with many of them brought back as slaves, following the Jewish Roman Wars in the 1st-2nd centuries CE, and finally there is considerable amount of documentations (both Jewish and Roman) about the popularity of Judaism in the Greco-Roman world just prior to those wars - it seemed only logical to assume Ashkenazi Jews’ European ancestry mostly derive from Italian converts. Namely, Italian women which converted to Judaism.
Now, because the uniparental evidence suggested complete and utter predominance of Middle Eastern ancestry as per the male lineages, it was logically assumed single Middle Eastern Jewish men arrived to Italy in Roman times, whether as freemen or slaves, and converted Italian women.
So, for that, the most suitable Italian population to also cause Ashkenazi Jews to be so similar to South Italians and Sicilians, would be North Italians. And again, that’s on the route to the Rhineland, and as I’ve mentioned, there are historical documentations which discuss movement of several single Jewish notables from North Italian towns to the Rhineland.
This is, in a nutshell, why that theory seemed so plausible, and with the lack of significant ancient DNA to actually check ancient North Italian autosomal profile or uniparental commonality, it’s easy to see why many would accept it as it is.
Historical problems with North Italy being the main source for Ashkenazim’s South European ancestry
While the “half n’ half” premise, on the face of it, does seem plausible, there are several problems with it, before we even tackle the genetic stuff.
First, there’s the problem that historically, the vast majority of Jews in Italy lived in Rome and in S. Italy, not North. That means most of the genetic impact from converts will come from Rome and South Italy.
Now one of the things which have been suggested is that somehow there was an historical discontinuity, and then miraculously pristine Levantine Jews appeared in North Italy in the mid 8th century and found a secret stash of pagan Italian women there (aka Xue et al.); or that for some reason, even the conversion was indeed during the more historical plausible time frame of 1sth century BCE – 2nd century CE in Italy, then only descendants of North Italian Jews survived despite the fact this was the place suffered from most conflicts during the Fall of Rome and Migration period of the 5th-6th centuries CE.
Also, we actually have a good documentation of continuous Jewish presence in Rome, one of the largest Jewish communities in antiquity, and even a documented papal decree by Pope Gelasius (early late 5th to early 6th century) allowing Jews to introduce slaves, if they were heathen, from Gaul into Italy. Pope Gregory in the 7th century objected to the Jews holding Christian slaves because he feared that they would be converted to Judaism. He even wrote to the Bishop of Naples that the Jews dealt in Christian slaves which they bought in the Gallic territories.
Jews are also sporadically mentioned all over the Italian peninsula during these centuries, though mostly only sparsely or in taxation papers since it seems they were doing relatively well in the Papal state and other kingdoms in Italy; and of course, during the Lombard rule in the 7th century onward we know they experienced significant tolerance compared to elsewhere in the Christian world.
So, while indeed after the fall of the Western half of the Roman Empire, many Jewish communities in the Western Mediterranean and in Western Europe seem to have vanished (or converted to Christianity?), Italy is actually one of the few places we do know of continuity, regardless of the population drop Rome itself experienced.
Second thing is, if I may dwell into Xue et al.’s proposed date of mid 8th century CE – this is incredibly implausible. Razib, in his recent blog entry, did raise a solid point and that is Europe was only “patchily” Christian in the first few centuries after Rome’s official Christianization; however where would those pristine Levantine Jews arrive from? Unless for some reason Jews elsewhere in the Mediterranean didn’t accept converts, which we know historically is incorrect since there’s ton of historical documentation for such widespread converts, these Jews would have to have arrived from the Levant directly to North Italy. The only place this could have theoretically been feasible from, would be 7th century Galilee, during the Heraclius revolt (614-625 CE). There were still ~150,000 Jews living in the Galilee at the time (modern N. Israel); however we also know that the several ones who fled, mostly fled to Mesopotamia and Egypt. Crossing the Mediterranean all the way to conflict torn and economically unstable Italy wasn’t an easy nor an attractive feat; it was expensive even during the Pax Romana era, and certainly in the 7th century. And even if several dozens did it, why would they not bring their families with them? Why would only men, as the genetic evidence seem to suggest, travel there?
In any case, we know that about 10 years after Heraclius put down the revolt, the Arabs conquered the region, and many Jews who fled to nearby areas came back. This again diminishes the plausibility of the above scenario. And considering I’ve shown there was historical Jewish continuity in Italy, even if some small amount of Levantine Jews did somehow arrive to Italy, they’d be overshadowed demographically by even the smallest Jewish community in Italy.
Third, and this can’t be underestimated, the amount of noise we would hear from the Church, as evidenced by the protest on trading Christian slaves from far away Gaul, if Jews would marry Christian North Italians en mass in the mid 8th century, EVEN under tolerant Lombard rule, would make this event widely documented historically. And there is simply no mention of such a thing.
Fourth, I will again address the date of conversion. There was a surge of popularity of Judaism in the Greco-Roman world, starting from around the 3rd-2nd centuries BCE until the 2nd century CE. We know historically this was the case; and we have Jewish sources as well as Greek and later Roman sources discussing how common was the phenomenon of proselytization. However, a series of three bloody conflicts between the Jews and Romans, stretching almost 70 years, with one of them – the Kitos War – amounting to a full-blown civil war between Jews and Romans in some parts of the Empire, seriously diminished this trend. After the introduction of the Fiscus Judaicus in the 1st century CE it had become an economic burden to be a Jew; but after Hadrian’s outlawing conversion to Judaism and circumcision, it had become almost impossible to become Jewish, unless you were born that way. And of course, as the world was less organized back then and the size of the Roman Empire usually caused any trend to slowly progress across vast distances, this doesn’t mean there were no more converts after the 2nd century CE. But, Judaism by then already had a serious competition from Christianity which on the face of it offered similar spiritual experience (and worship to the same deity) but with much less burden both economical and physical (no circumcision), and soon after (by the early 4th century) it had become the Empire’s official religion. So, again, while there’s a very slim chance that full 3 centuries after the conversion of the Roman Empire right next door to the Papal state there would be mass conversion of women to Judaism, I find it very unlikely. The date, 8th century CE, is just too damn late to make any historical sense.
So as one can see, there are several crucial problems with the historicity of the “half n’ half” model. I intentionally do not dwell into the historical case FOR a non-Italian source of Ashkenazim’s S. European admixture; if any of you are interested, we can debate this in the discussion that’ll hopefully follow this opening post and you are more than invited to read the Western Jews’ mega thread which deal precisely with that.
The genetic case against North Italy being the main source for Ashkenazim
Autosomal models and genetic distances
To me, one of the biggest evidence against the “half n’ half” model is that Western Jews from communities which seem to have no history of substantial ancestry from Italy, and N. Italy specifically, cluster all too tightly with other Western Jews, Ashkenazim included, in all PCAs, including few PCAs more specifically targeted to observe genetic drift (one such PCA was prepared two years ago by Davidski when the Collegno samples became available). This wasn't really discussed in many recent papers, but as far back as 2010 you had Atzmon et al. (2010) which already discussed how all Western Jews seem to have descended from the same seed Mediterranean population. I don't know, I just can't see why non-Sephardic Romaniote Jews, Djerban Tunisian Jews and Libyan Jews would have significant N. Italian ancestry (they might have some Italian ancestry mediated either via minor admixture with Sephardim, or from ancient times when they were all part of the Roman Empire – but that much of N. Italian ancestry? Can’t see why).
The second issue is that from both Antonio el al. 2019 (but really first clue was from Amorim et al. 2018, aka the “Collegno/Lombard paper”) and the latest Potsch et al. 2021 (the Etruscan paper) and Aneli et al. 2021 (the Daunians paper), we see that Italians, even North Italians (ie in Tuscany) at the time when the forefathers of Western Jews would have been most likely to accept converts (during the Imperial era), but also up until Late Antiquity (as suggested by the 8th century dating of Xue et al.), didn’t autosomally look like contemporary North Italians or even like their predecessor Iron Age Italic population.
What we see, based on these recent studies, that from South to North, Italians of that time plotted mostly like contemporary South Italians; until they’ve received significant North shift during the early medieval period which most profoundly affected North Italians and Central Italians, making them plot the way they do today, while less affecting South Italians which is why the latter still seem to plot akin to Imperial-era Italians.
Also, if at the time when Antonio et al. 2019 was released, it seemed like maybe this South shift change was mostly limited to Rome, since it was the capital of the empire and might have received significant number of migrants from the Eastern provinces while other further localities in Italy didn’t, we now know this wasn’t just limited to Rome, but seem to have reached as far north as Tuscany (as the paper itself claim, Imperial Era “Etruscans” had ~40% of such Near East/East Med admixture) and at least as far south as Puglia (as per the Daunian paper).
So now we need to model Ashkenazim with Levantine + South Italian-like component, since this is how most Italians looked like genetically during Imperial time, at least in the relevant locations where most Jews lived, and even slightly up north in Tuscany, if we’re to accept the “half n’ half” model based on North Italians.
Again, even if we go along with Xue et al. and suggest fully Levantine Jews met up with North Italians, say Tuscans, and converted them as late as the 8th century CE, we’re still left with a somewhat South Italian-like population. This is how Italians from Chiusi (the samples are dated to 860 CE - 1073 CE, so 9th - 11th centuries), Tuscany, looked like in Early Medieval Ages:
Distance to: ITA_Chiusi_EMA_Average
0.01684813 Italian_Abruzzo
0.01684848 Italian_Marche
0.01959056 Italian_Apulia
0.02005724 Italian_Umbria
0.02058945 Italian_Molise
0.02095955 Italian_Lazio
0.02336779 Italian_Basilicata
0.02484728 Italian_Tuscany
0.02515773 Italian_Campania
0.02642290 Sicilian_East
0.02692627 Greek_Izmir
0.02805450 Greek_Thessaly
0.02839739 Greek_Peloponnese
0.02983395 Sicilian_West
0.03157958 Italian_Calabria
0.03168379 Greek_Laconia
0.03338691 Italian_Piedmont
0.03527651 Greek_Crete
0.03527716 French_Corsica
0.03604280 Greek_Macedonia
0.03651000 Ashkenazi_Ukraine
0.03655507 Maltese
0.03662124 Ashkenazi_Belarussia
0.03669431 Albanian
0.03699339 Ashkenazi_Poland
As can be seen, pretty S. Italian like. And as one goes further back to Imperial Age, it’s even more South-shifted.
Now that we have more ancient DNA from several studies, we can also look back at the Aegean-like/Near Eastern shifted samples from 6th century CE Collegno, N. Italy, featured in Amorim et al. 2018. It seems very likely these samples were simply late survival of Imperial-like genetic profile; the fact that they were found in Collegno and were shown to be natives rather than new migrants makes sense in the context of both Antonio et al. 2019 an Potsch et al. 2021 (Etruscan study), and also stretches just how far north that South Italian-like genetic profile used to reach (as far north as modern Turin !).
Now if you try to model Ashkenazim or other Western Jews with S. Italian-like proxy, because of how close S. Italians (and Aegean Greeks) are to Western Jews, you’d get very little Middle Eastern component, which we know can’t be the case because of how substantial the Middle Eastern component is on the uniparental paternal side, and also how in general paternal subclades in Ashkenazim or Western Jews are very different from Italians.
The third issue is an issue I’m having with how Xue et al. actually constructed their reference groups. Xue et al. seem to have created a reference group representing the possible South European admixture in Ashkenazi Jews from the following populations: Bergamo Italians, Tuscans, Sicilians, Abruzzese and Greeks:
xue_et_al_references.png
While perhaps back in 2017, the difference between N. Italians and S. Italians and Greeks wasn’t as common knowledge as it is today, but you all will agree with me that today, in 2021, we know very well these are distinct populations from one another (specifically Sicilians and Greeks vs. Bergamo Italians). Another problem is this also the exact historical geographic route Jews would make when moving from the East Med to N. Italy onward to the Rhineland.
Now any of you with a good knowledge of IBD sharing etc., I raise this as a question: what would we see, if say Jews with already East Med rather than pristine Levantine genetic profile migrated up north to Italy from say, Rome, and met very few pagan North Italians female slaves on their way up north to the Rhineland around the mid-8th century - would that still yield the result that Jews admixed with N. Italians in the mid-8th century rather than the East Med ancestry they'd might have already included few centuries earlier? I’m leaving this as an open question to people who are more knowledgeable in the field of IBD sharing, which I am admittingly not an expert in, because IMO this might actually explain what has took place and why they’d got this unusual dating for the admixture event.
Lack of substantial uniparental lineages
I left the uniparental issue, or lack of uniparentals really, last, but it’s one of the biggest evidence against North Italy (or Italy in general) being the biggest European admixture source in Western Jews, Ashkenazim specifically.
But before I dwell into this, I’d like to begin by extend my sincere gratitude to members Principe and leorcooper19 which greatly assisted with their uniparental “mastery” at putting this together.
We’ve looked into the Vagnari villa study (featuring 30 Roman era samples from Puglia), Amorim et al. 2018, Antonio et al. 2019, Aneli et al. 2021 and Potsch et al. 2021. In total, when discarding post 8th century CE samples and samples before the Iron Age, we’ve amounted about 229 ancient samples, from 9th century BCE to 9th century CE. We’ve looked at all the maternal subclades in the papers, and cross referenced them with the known statistics of modern Ashkenazi maternal subclades.
This is what we’ve come up with:
Ancient Italian mtDNA % among Ashkenazi Jews H1aj1a 1.4% T2b25 0.98% T1a1 0.33% T2a1b 0.31% HV1a'b'c 0.28% K1a4a 0.20% J1c-C16261T 0.20% V15 0.17% X2e2a 0.14% H5a1 0.14% H47 0.08% N1ba2 0.08% H5'36 0.06% U8b1b1 0.06% T2g 0.03% T2b 0.03% U5a2b 0.03%
So, overall, 17 lineages (out of about 80-90 total maternal lineages among Ashkenazi Jews for the more conservative estimates; some estimate total of 124 maternal lineages though), amounting to an occurrence of about 4.52% in Ashkenazim.
After much discussion, we’ve also decided to include N1b1b1, since few of the Collegno samples have turned up under that subclade, AND its direct parent subclade, N1b1b seem to have been found in BA Swiss samples (Furtwaengler et al. 2020) – so to cover all possible maternal subclades which might have entered into Ashkenazim via Italians, we’ve decided that it should be included, despite the fact that it arrived to North Italy most likely with the Germanic invasions, and it’s not native to Italy (the two Lombard samples from Collegno belonging to that subclade, SZ22 and SZ13, both show no trace of South European admixture and seem to plot close to North Europeans). Theoretically, it could have also entered into Ashkenazi gene pool in Central Europe, since that’s where it originated from, but alas – I think there’s a plausible enough case made to include it.
N1b1b1 is one of the major Ashkenazi maternal lineages, and really changes the picture. It’s occurrence among modern Ashkenazim (it varies a bit between the different sub-Ashkenazi community one look at) can be estimated to be around 5%.
So, with N1b1b1 included, one can double the amount of Italian-originated maternal lineages to an occurrence of ~9.5% (+-0.5%) in Ashkenazim.
However, one has to remember this list is very “liberal”, as I said above, we tried to estimate the maximum extent of maternal lineages which could have entered Ashkenazi gene pool in Italy.
In reality, several of the subclades which were listed above are probably late arrival, three in particular - X2e2a, N1b1a2, H47 – seem to have origins in Chalcolithic and Bronze Age Anatolia, and could have arrived the ancestors of modern Ashkenazim in Levant, and not necessarily in Italy, as can be demonstrated in the following tables:
X2e2a:
X2e2a.png
N1b1a2:
N1b1a2.png
H47:
H47.png
So, without those three lineages, and without N1b1b1 we’re at ~4%, and with those three and N1b1b1 we’re at ~10%.
And considering what we know of Ashkenazi history – that is – a strong founder effect and a genetic bottleneck, if indeed N1b1b1 entered in North Italy, judging by the other suspected Italian subclades, there’s no way it was originally that prevalent, but got inflated as result of said bottleneck / founder effect.
But, even if we’re to stay with 10%, which really is the maximum amount from the current known Ashkenazi maternal lineage and current known maternal lineages from ancient Italian samples (which as mentioned here, is not negligible as of 2021) – it’s simply way too low. Contrast that to the utter and complete predominance of Middle Eastern subclades on their paternal side, and a high likelihood most of them are specifically Levantine, Ashkenazi maternal subclades clearly do not have predominant Italian origin.
The paternal side of things isn’t too different. While as mentioned, and verified in several academic studies, there’s a clear Middle Eastern predominance among the paternal lineages of all Western Jews, Ashkenazi included (~80% of the paternal lineages are definitely of Middle Eastern origin), there are also few lineages which are in high confident of Italian origin:
Ancient Italian Y-DNA % among Ashkenazi Jews J-Y33795 1.10% R-L4 0.71% I-Y11261 0.65% R-Y101215 0.45% J-Y99200 0.40% R-Y104180 0.24% I-BY18705 0.21%
So as can be seen, only 3.76% of Ashkenazi paternal lines can be considered, in high probability, as has been acquired in Italy.
So where does this leave us?
If I can summarize all of this, I think that if we consider all the data I've laid down here, both historical and genetic, with the latter being gathered in the last few years from recent papers and the ever increasing sampling size, it is safe to assume that half n' half model of Jews as 50% Levantine/ME and almost 50% North Italian (as in modern North Italian-like, not Imperial North Italian-like) is almost certain invalid. The dating proposed by Xue et al. also seem completely out of place. However, I would also go further and say that if one look specifically at the uniparental evidence, it seems the biggest European source, which I’m still certain is South European, was most likely acquired from sources outside of the Italian peninsula. Now I need to clarify here – all of this doesn’t mean that Western Jews, or more specifically Ashkenazi Jews, didn’t acquire a substantial part of their South European ancestry while their ancestors lived in Italy. For instance, if substantial amount of the women converts in Italy during Imperial and late Republican era (2nd century BCE to 2nd century CE) were from Magna Graecia, it would still count as Italy geographically, but won’t be of distinct Italian origin per se. Same goes for instance for maybe few of those Gallic slaves from Late Antiquity Gaul (modern France) – as we’ve seen, Italian Jews were allowed to trade in those; and it might be that Jews got some admixture from those – however, while geographically this would have happened in Italy, this admixture shouldn’t be considered of distinct Italian origin.
In any case, I do hope for a fruitful discussion on all of this – so feel free to add your own comments and thoughts !
Reply With Quote
.png)
Originally Posted by Claudio
