The idea that Armenians could have serious influx from Levant is not new. It was discussed many times the last 5 - 7 years that I follow all this debates. But the last paper from Skourtationi was a game changer. It demonstrated that the reason why Armenian are shifted toward East Med is not an massive influx from Levant but because Eastern Turkey in ancient times was inhabited by quite southern populations. The same finding inflated the level of Steppe in Armenia bringing it to more realistic numbers.
Here is the opinion of the user DMXX on this subject. He is the admin of Iranian DNA project.
PS. I must add that the presence of Semitic Y dna in Armenia is undeniable ( J1 -Z1884, some J2b1-s, some T-s ) Neverthless their number, their fragmented nature do not speak about any one time mass migration.
In sum, Armenians can be reliably modeled as ~85% E. Anatolian-S. Caucasian ChL-EBA and ~11% EMBA steppe.
So, I'd tentatively assert that Armenians are ~11% EMBA/~16% MLBA steppe, which is only a couple %age points less than the Kurmanji Kurdish (non-Yazidi) and Azeri averages, and is a touch higher than the average seen in Iranian Lurs.
The above also corroborates Arame's modelling - Significant differences in the source of Anatolia/Levant_N-rich admix is seen between Armenians and Assyrians. The majority of that admix in Armenians is typified by the SE Turkish EBA datapoint, whereas it's an even split in Assyrians. Levant_EBA in this context is a catch-all indicator for Semitic-related admix in Assyrians.
The overall difference in component scores here is 30% - This is substantially larger than what was inferred via ADMIXTURE many years ago (as highlighted by Alkaevli's post, the % difference was ~13%).
Ergo, with informal modelling that contains aDNA and not modern pop-modal allele frequency data, Armenians and Assyrians should no longer be considered "variants" of one another. It's analogous to claiming that Brits and Spaniards are "the same".
Either way, in conclusion:
1) Armenians and Assyrians should not be considered synonymous with one another,
2) Armenians do not have "low" (defined arbitrarily by me as <5%) EMBA steppe admix relative to neighbouring pops (they're within range for W. Asia),
3) There's reason to suspect that some of our Assyrian samples in G25 carry cryptic admixture from relatively steppe-rich neighbouring pops, skewing the averages
This is one of several examples of where, as a community, the collective perception of a given topic is unduly coloured by formative experiences with the preliminary and/or rudimentary data that were available to us in the 2005-2015 period.
I've been a spectator (or active interlocutor) to these discussions for most of that period, so forgive me for this unsourced narration regarding why there is resistance to the notion that Armenians don't have a significant (>10%) proportion of P-C steppe-mediated admixture.
The earliest uniparental studies (I have Weale et al. 2001 in mind) on the Armenians showed what we'd expect in a fairly opaque sense (they broadly clustered with other W. Asians). Later papers demonstrated that Armenians, on the Y-line, were typified by the "quintessentially W. Asian" combination of E1b1b1, G(1+2), J2a, J1(xe in peri-2010 ISOGG nomenclature) and R1b-L23, which was also observed in all populations from central Anatolia through to Iran and south towards Iraq. Later data, such as from Peter Hrechdakian's FTDNA Armenian DNA Project, broadly supported that observation.
By the early 10's, auDNA results from multiple Armenian and Assyrian individuals from the consumer testing circuit began popping up on forums. A similarity between them was readily observed. Further, the uniparental profiles were also broadly similar (per above). Folks began to assume a near-complete convergence in origins based on these factors, which were reinforced by anthropological considerations (e.g. irregular intermarriage between these two minority groups, partially facilitated by their Christian backgrounds).
The inference that the two populations were almost synonymous with one another from a genetic perspective was seemingly cemented by 2012-14, when ADMIXTURE calculators reined supreme, and the two populations certainly did appear closer with respect to the scores they'd generated. Here, we observed what Alkaevli's table perfectly demonstrates - Armenians generally appeared midway on a pan-component cline between Georgians and Assyrians, with most component scores tending towards the Assyrian end of that axis. Their "N. European" related ADMIXTURE proportions were also significantly below what was observed in groups like non-Trabzon Turks or Iranians across practically all calculators.
Although this general line of argument (an iteration of "Assyrians/Armenians are language-shifted Armenians/Assyrians" or "Assyrians and Armenians are 90+% Urarto-Hurrian-derived and that's it") seemed convincing at the time, one piece of evidence (which was curiously ignored by most, with the notable exception of former colleague and esteemed contributor 'Humanist') impeded me from accepting these hypotheses. Specifically, the surprisingly lacklustre IBD segment sharing between Assyrians and Armenians, which falls quite short of what we can observe from other known related groups in the region (f.ex. numerous Iranian ethnicities and populations that are now described as Kurdish).
In the sciences, especially when a multi-disciplinarian perspective is employed, it is crucial to validate hypotheses, control for bias and critically appraise discordant data if one of your evidence streams doesn't align with the rest. This appraisal wasn't generally taken up by the community at large, where a seeming majority seemed to implicitly accept earlier conventions without re-appraising old data in the face of new, or back-testing old ideas with new material.
In retrospect, there were always two problems with the general popular argument up to circa 2014:
1) Despite the high degree of uniparental profile overlap between W. Asian populations, there are significant differences in frequency. Additionally, there are significant differences in the types of subclades seen beyond the stereotypical W. Asian combination (f.ex. Armenians were, from memory, an unusual local hotbed for Y-DNA I2; Iranians and Kurds are broadly the most Y-DNA R1a-M17 containing para-ethnic groups; Y-DNA Q frequencies in the Republic of Azerbaijan and W. Iran, etc.). Discounting these differences in favour of a fixation on the commonality undercuts the very point of establishing contrasts between groups (I personally suspect that some observers were/are so preoccupied by the ADMIXTURE outputs of old that they instinctively looked for the common rather than the differences when it came to the Armenians)
2) ADMIXTURE's design (modal component formation based on allele frequency) makes ancestral signals "bleed into each other" in regions with a longstanding history of demographic settlement or mixing. Ergo, Armenians being 3-6% "N. European" on average in Dodecad K12b doesn't indicate much about the actual degree of admixture from a N. European-like population (in this case, MLBA steppe groups, or EMBA to a lesser degree).
Post a Comment