We report genome-wide data for 33 Ashkenazi Jews (AJ), dated to the 14th century, following a salvage excavation at the medieval Jewish cemetery of Erfurt, Germany. The Erfurt individuals are genetically similar to modern AJ and have substantial Southern European ancestry, but they show more variability in Eastern European-related ancestry than modern AJ. A third of the Erfurt individuals carried the same nearly-AJ-specific mitochondrial haplogroup and eight carried pathogenic variants known to affect AJ today. These observations, together with high levels of runs of homozygosity, suggest that the Erfurt community had already experienced the major reduction in size that affected modern AJ. However, the Erfurt bottleneck was more severe, implying substructure in medieval AJ. Together, our results suggest that the AJ founder event and the acquisition of the main sources of ancestry pre-dated the 14th century and highlight late medieval genetic heterogeneity no longer present in modern AJ.
I’ve been asked how this modifies the narrative in my Substack piece. I’d say only on the margins.
The Erfurt community dates to the 1300’s. Interestingly it shows variation in Eastern European ancestry. The authors suggest that genetically there are actually two clusters here, though sociocultural they’re identical. The best guess is that the Eastern European enriched population migrated to Erfurt from the east (there is some difference in the isotope analysis for the teeth). Modern Ashkenazi Jews show less variability and are positioned between the two Erfurt communities in PCA and admixture space. The group without Eastern European ancestry seems to resemble Sephardi and Italian Jews. This isn’t surprising, since they confirm Ashkenazi Jews are some proportional mix of a Middle Eastern population, Italians, and Eastern Europeans, while Sephardi and Italian Jews clearly just lack the last.
The Erfurt community experienced a strong bottleneck, stronger than the one in modern Ashkenazi Jews. This implies that there are other groups out there unsampled, and modern Ashkenazim descend from that. This isn’t surprising, one feedback I got is that there are so many medieval Jews for the inferred population size during this period (going by the texts). I think one issue might be a lot of the medieval Jewish communities simply went extinct. Many of them were undergoing similar dynamics, but not all contributed to future Ashkenazi ancestry.
The homogeneity (relative) of modern Ashkenazim is probably due to late-medieval metapopulation dynamics.
Can you elaborate on the Eastern European admixture? Did it come from a prexisting Jewish community that mixed into the proto-Ashkenazi or did it come directly from gentiles who married into the proto-Ashkenazi?
“We used qpAdm to gain insight into the ancestral sources of EAJ (Methods Section 4). We modeled EAJ as a mixture of the following modern sources: Southern European (South-Italians or North-Italians), Middle Eastern (Druze, Egyptians, Bedouins, Palestinians, Lebanese, Jordanians, or Syrians), and Eastern European (Russians). To avoid bias due to ancient DNA damage, we only used transversions. Most of the models with a South-Italian source were plausible (P-value >0.05; Table S7), which would be consistent with historical models pointing to the Italian peninsula as the source for the AJ population. The mean admixture proportions (over all of our plausible models; Table S7) were 68% South-EU, 17% ME, and 15% East-EU (Figure 2A). However, the direct contribution from the Middle East is difficult to estimate given historical ME admixture in Italy [49]”[pp. 6-7].
“We also tried to model EAJ as a mixture of ancient sources [50]. The sources we used were Imperial or late antique Romans [49], Canaanites [51], and early medieval Germans [52]. These models gave poor fit (P<0.01 for both Roman sources), suggesting a missing ancestry component. Alternatively, the poor fit might reflect technical artifacts due to inhomogeneous data types: for example, the Canaanite and EAJ datasets were produced by in-solution enrichment, while the Imperial/ late antique Roman and early medieval German datasets were produced by shotgun sequencing" [p. 7].
I.e. the fake tentative of the Harvardians to demonstrate the Levantine intake in Imperial Rome as I firstly said.
The Near Eastern ancestry of these medieval Jews is only 10%-20%. Thats very low. Are modern Ashkenazi Jews actually 40-65% Near Eastern or is it closer to the medieval proportion?
Can you elaborate on the Eastern European admixture? Did it come from a prexisting Jewish community that mixed into the proto-Ashkenazi or did it come directly from gentiles who married into the proto-Ashkenazi?
latter probably. joshua lipson says in the jewish genetic genealogy community they see clear slavic mtdna in litvaks
We had a demonstration about that already when we knew that the mutation CCR5delta32 was about 50% in Jews from the Baltic and that mutation was very rare in the Roman Empire, practically non existent in Near East and higher out of the Roman Empire, thus introgressed from there in the Jewish pool.
The Near Eastern ancestry of these medieval Jews is only 10%-20%. Thats very low. Are modern Ashkenazi Jews actually 40-65% Near Eastern or is it closer to the medieval proportion?
depends on whether you pick southern or northern italians. there is a pref in their models for south italians, so that means the fraction is higher. my personal exp is south italians and sivilians are just hard to distinguish from ashkenazi jews.south italians also have 1) more roman period gene flow from the east 2) more gene flow from the east even during the bronze/iron age (haplogroup j2 flow right?)
“1) more roman period gene flow from the east 2) more gene flow from the east even during the bronze/iron age (haplogroup j2 flow right?)”
What people doesn’t say (just because they dislike that) is that Southern Italy had a refugium in the Palaeolithic as the North had another, and that hg J2 did come to Italy already in the neolithic probably from the Caucasus and had nothing to do with the later migration to the Levant, the known 4200 YBP event, and anyway each haplogroup has to be understood per se. Anyway I think that hg J was in the WHG as hg I long before the presence in the Caucasus and even the old migration to India from up there. Of course only the aDNA could answer my hypothesis.
I’ve been thinking about whether the early group is small enough that the non-core people who were absorbed could’ve just been something like random more townspeople/townswomen with more diverse origins from across late Antique/Medieval towns and not really a single people as such. Non-local mobility into towns seems pretty high across historical period, but most of these people didn’t really contribute much stably over time.
On that as a tangent, there could be at least a couple of reasons that lack of contribution could be. One might be that city conditions were much inherently worse, for reasons of disease and such, and the population naturally couldn’t reproduce itself. But that seems to me not so compatible with the success of Jewish people in (I presume, and its a big presumption) that same environment, with the same disease risks and at best, not much of a set of improved practices to avoid or survive diseases.
Another might be that cities just tended to be where ancient cultures “put” a lot of their more Malthusian dynamics. In that model cities aren’t inherently so much worse in terms of disease and such, but that city inhabitants are very open to economic and marriage competition to the country, with constant high levels of influx, and don’t have enough jobs and housing to support the volume of country/international immigrants and city natives. The big city is a tough place, and you’d be more likely to end up single and in the gutter. In that sense, you’d tend to see cities becoming a genetic sink where natives fail to survive and reproduce more, not to do with anything to do with disease or the environment really, but just because of sheer greater competition. That seems more compatible with Jewish groups, because in that case the unique thing is that they had closed social networks and a relatively closed economic niche, so no influx of newcomers from the country providing competition for marriage and work.
It might be useful to compare with India and the Islamic world on this sort of thing; if you have longstanding urban communities/castes there, then that kind of thing might provide another example of how long term urban populations might be possible, despite diseases, if there are social norms that separate communities and prevent people coming and competing for housing, jobs, marriage.
Paper: “We also tried to model EAJ as a mixture of ancient sources [50]. The sources we used were Imperial or late antique Romans [49], Canaanites [51], and early medieval Germans [52].”
There could potentially be lots of problems with this due to diversity within the groups of Romans, Canaanites and early medieval Germans. From memory the Romans and early med German sets can have a lot of outliers (in medieval Germany these were particularly people with cranial deformities). They might get better outcomes with identifying subsets within these large ancient dna sets and then using models with them. Find groups compatible with qpWave rank=0, then use those for the modelling.
Razib: “The Erfurt community experienced a strong bottleneck, stronger than the one in modern Ashkenazi Jews. This implies that there are other groups out there unsampled, and modern Ashkenazim descend from that.”
Yeah; so the “two population model”, Erfurt and other here, is probably very abstracted from a multi population model, where if Erfurt pop did contribute, it’s probably as one of a large number of somewhat structured, though autosomally similar, subpopulations? The “larger” population could in reality be several populations with as much founder effect as Erfurt?
This structure also might have kept the proportions of the ancestor populations variable for a long time? Shai Carmi suggests that the cemetery might be linked to a backmigration from Poland (maybe further east too?)
Yeah, as Matt alluded, this dovetails quite closely with your more recent post about cities in the pre-modern era being fertility sinks. We know that historically speaking the Jewish population was heavily focused in urban areas. They also look a lot more like a Roman-era urban dwellers rather than peasants in being an admixed population. This all being true, it’s likely that there were dozens of different urban Jewish communities in late antiquity/the dark ages, with most of them becoming effectively extinct.
“I think one issue might be a lot of the medieval Jewish communities simply went extinct.”
Agreed. This is also something we see in the near complete lack of French and German admixture despite the fact that historically, that his where Ashkenazi Jews originated.
Analysis of the Middle Eastern component is also interesting. The best match by far was Lebanese, which is a match to the original Canaanite claimed ultimate homeland of the Jewish people in the Torah. The next best matches are Syrian (right next door) and Bedouin B (a clan of Bedouins who made their way to the Negev Desert in Israel within the last 300 years and has significant introgression from other populations via exogamy, so not very informative).
Palestinians, Jordanians, Egyptians, and Bedouin A (a clan of Bedouins who have been in the Negev desert for thousands of years and is tightly endogamous), all of whom are peoples in places in or near places that figure prominently in the Torah story of Jewish origins and the Hebrew Bible is also notable. Endogamy and a stay in Egypt of only a a few generations makes the lack of Jewish affinity with Egyptians less inconsistent with the legendary traditions of the Jews about themselves.
But the lack of affinity with any of the populations that were the source of the pre-20th century Israel population of Israel, i.e. the Palestinians, Jordanians and Bedouin A, is more notable (particularly since the existence of a Jewish kingdoms in what is now Israel is corroborated by contemporaneous historical accounts outside the Bible from the region – the latter part of the Hebrew Bible is more connected to historical events than the earlier part of it).
It suggests that the Hebrews in what is now Israel were a heavily endogamous ruling caste of herders presiding over indigenous proto-Palestinian farmers which was ultimately exiled from the region in one or more events, probably culminating in the diaspora ca. 70 CE. This is at odds with a Hebrew Bible description of this process as basically one of genocide and total replacement (see, e.g., the books of Numbers and Judges).
“This is at odds with a Hebrew Bible description of this process as basically one of genocide and total replacement (see, e.g., the books of Numbers and Judges).”
Well that’s not what these books say but as Prof Kennet Kitchen showed already 20 years ago, Joshua (not Numbers) speaks of raids not conquest (in the florid and exaggerated style of the time). The conquest starts at the end of the book. And Judges does not speak at all of genocide and total replacement, on the opposite, it presents the Israelites as living only in a small part of the allotted territories and even sometimes mixing with the local population, having failed to conquer the whole land.
Now I do not know about ancient DNA of ancient Israel, maybe Razib knows about it, but it seems logical that part of the Caananite population was absorbed in the Israelite population that was originally small (50,000 people according to Finkelstein in the 11th century)and smaller than the Caananite one. So whatever the real origin of the core Israelite population that left Egypt even if that did happen – at the end, the population of the Israel and Judah Kingdoms was probably with a big chunk of Caananite origin and similar to modern Lebanon. But that’s just speculation, if there is better info about it I would be happy to know.
I think a proposal of an ethnic caste system in Israel that in some sense lasted up until the diaspora (or even at all) would need some direct testing by ancient dna rather than being inferred, before I’d have too much confidence about that (being conservative about proposals that are inferred to a great deal from modern dna or dna that long post-dates the era anyway?). Hopefully Jewish burial rules don’t outright prohibit that (exhumation of the dead is a big religious deal? it seems so from Shai Carmi’s twitter thread on this paper).
“Now I do not know about ancient DNA of ancient Israel, maybe Razib knows about it, but it seems logical that part of the Caananite population was absorbed in the Israelite population that was originally small (50,000 people according to Finkelstein in the 11th century)and smaller than the Caananite one. So whatever the real origin of the core Israelite population that left Egypt even if that did happen – at the end, the population of the Israel and Judah Kingdoms was probably with a big chunk of Caananite origin and similar to modern Lebanon. But that’s just speculation, if there is better info about it I would be happy to know.”
If this was the case, then we’d need to come up with a reason that Palestinian, Jordanian, and Bedouin A populations of Israel which predate the Zionist movement aren’t very genetically similar to Lebanon, while Lebanon is the best proxy for the Middle Eastern element in Ashkenazi Jews. Moreover, the entire time period that is relevant is historically attested, so we’d need a process to bring that about that is consistent with the historical record.
“I think a proposal of an ethnic caste system in Israel that in some sense lasted up until the diaspora (or even at all) would need some direct testing by ancient dna rather than being inferred, before I’d have too much confidence about that (being conservative about proposals that are inferred to a great deal from modern dna or dna that long post-dates the era anyway?)”
We may get ancient DNA, or we may never get it (at least in my lifetime). No harm in trying to make sense of the data that we do have until then.
I’d be open to other narratives that get one there, so long as they are consistent with the secular historical record, and I would welcome other plausible explanations of the data.
For example, perhaps a lot of the exiles stopped after a short trip to Lebanon.
Another possibility would be that Lebanese Jews were Canaanite converts to the Jewish religion from the kingdom to their south who also picked up an Aaronic priestly line of Y-DNA in the process, but not much autosomal DNA ancestry. These converts engaged in more maritime trade (a legacy associated with Lebanese Phoenicians), than the Jewish kingdom Jews (who aren’t known for their maritime legacy) causing them to end up in places like Southern Italy. Perhaps these converts are the ancestors of the people who think of themselves as part of the Jewish diaspora, while the Jews of Israel proper, who remained behind in what is now Israel, all eventually converted to Islam and lost their cultural connections to that heritage.
But, somehow or other, the Middle Eastern ancestors of the Ashkenazi Jews are not the ancestors of the people living there two thousand years after the Jewish kingdoms ceased to exist even as tributaries of the Roman Empire.
Historically, it is also worth noting that we know with a fair amount of confidence that there were a non-negligible number of Jews and members of the Church of the East (some of whom might have converted from Judaism) in the 600s CE in the Middle East at the time that Islam came into being, so assuming conventional wisdom about the importance of 70 CE to the Jewish diaspora may be overstated. But this data point still doesn’t get us all that far to unraveling what happened.
@ohwilleke
What about Palestinian Christians, where do they fall in comparison? I had been under the impression that there was a lot of admixture introduced in the Muslim period through the widespread reliance on and exchange of slaves (many of whom or whose descendants would be at some point manumitted), as well as the elite overproduction of more southerly Arabs, who ruled over the locals. Do we not have any ancient DNA for the Southern Levant from the relevant time period?
@Otanes
I think Razib has said in the past that Arab Christians today are pretty close to the pre-Islamic population of the region. They don’t have the “exotic” African or Turkic ancestry associated with the slave trade, while the Muslim Arabs do.