Steppe ancestry in Chalcolithic Transcaucasia (aka Armenia_ChL explained)

In 2016 Lazaridis et al. published a paper featuring five ancient samples from the famous Areni-1 cave complex, in what is now Armenia, dated to the Chalcolithic (see here). This is how they described the ancestry of these ancients, which they labeled Armenia_ChL, in the supplementary PDF to their paper (page 94):

We do not have a pre-Chalcolithic sample from Armenia. We first model it [Armenia_ChL] as a 2-way mixture of any of WHG, EHG, CHG, Iran_N, Levant_N, Anatolia_N (Table S7.18), but we find no pair of these populations that could be ancestral to Armenia_ChL. We next model it as a 3-way mixture (Table S7.19), and determine that Armenia_ChL can be modeled as 18.3±1.5 EHG, 29.2±2.4% Iran_N, and 52.5±2.2% Anatolia_N. In the absence of a pre-Chalcolithic sample, we cannot be certain whether the Neolithic population of Armenia (which borders Anatolia from the east) was similar to that of Northwestern Anatolia and experienced gene flow from the east and north, or the reverse.

Since then, a lot of opinions have been posted in the comments at this blog and elsewhere about the possible origin and significance of Armenia_ChL. It seems to me that many people see Armenia_ChL as more or less an example of the indigenous Neolithic and Chalcolithic peoples of the South Caucasus. But some have argued that Armenia_ChL was in large part of Central Asian origin and concocted various mixture models to try and back up this rather strange claim.

To me, it was always obvious that Armenia_ChL harbored very recent admixture from the Pontic-Caspian steppe, because I couldn't reconcile its relatively high level of Eastern European Hunter-Gatherer (EHG) ancestry with a deep origin south of the Greater Caucasus range.

Moreover, in any decent Principal Component Analysis (PCA), like the one below, Armenia_ChL appears to form two subtle sub-clusters, with three of its individuals "pulling" more strongly towards Eastern Europe. This suggests that the EHG admixture in Armenia_ChL was present at variable levels and thus likely to be recent, because it didn't yet have time to diffuse evenly throughout the population.

Also, two out of the three Armenia_ChL individuals who are "pulling" north belong to steppe-specific mitochondrial (mtDNA) haplogroups. Armenia_ChL I1634 belongs to mtDNA haplogroup H2a1, which is seen in ancient samples from the Pontic-Caspian steppe associated with the Khvalynsk, Sredny Stog and Catacomb cultures, while Armenia_ChL I1409 belongs to mtDNA haplogroup U4a, which is found in numerous ancient samples, especially foragers, from the Pontic-Caspian steppe and other parts of Eastern Europe (see here). Coincidence? Surely not.

The idea that Armenia_ChL represents a long-standing indigenous Transcaucasian population also took a major hit recently with the release of the Wang et al. manuscript on the genetic prehistory of the Greater Caucasus. The preprint included samples from the Eneolithic Caucasus dated to earlier than Armenia_ChL (4594-4404 calBCE vs 4330-3985 calBCE) which looked typically Caucasian and lacked any discernible signals of ancestry from the steppe. Below is a PCA from Wang et al. featuring both the Eneolithic Caucasus and Armenia_ChL samples.

Unfortunately, modeling the recent ancestry of Armenia_ChL is still difficult, because the genotype data from Wang et al. haven't yet been released, so currently there is still no pre-Armenia_ChL sample available from the Caucasus for me to work with.

The earliest post-Armenia_ChL sample is Armenia_EBA I1658, dated to around a thousand years too late (3347-3092 calBCE). However, this individual is associated with the Kura-Araxes culture, which is generally seen as a direct successor to the native Neolithic cultures of Transcaucasia, and appears to be practically indistinguishable from the Eneolithic Caucasus trio in the Wang et al. PCA. Thus, pending the release of a pre-Armenia_ChL sample I might be able to use Armenia_EBA I1658 as an effective proxy for such a population.

Below are a couple of successful two-way qpAdm mixture models for Armenia_ChL and Armenia_ChL I1634, featuring Armenia_EBA I1658 and Sredny_Stog I6561 (the output for Armenia_ChL I1409 looked wobbly, probably due to a lack of markers). The reason I decided on Sredny_Stog from the North Pontic steppe as the surrogate for the steppe ancestry is because of the position of Armenia_ChL in the Wang et al. PCA relative to Eneolithic Caucasus, which suggests gene flow into the former from a more westerly steppe source than, say, Khvalynsk from the Samara region. Using these reference samples, the inferred ratio of steppe admixture in Armenia_ChL is around 15%, which I think makes sense, more or less, considering its position in both of the PCA above.

Armenia_ChL
Armenia_EBA_I1658 0.862±0.050
Sredny_Stog_I6561 0.138±0.050
chisq 17.038
tail prob 0.148174
Full output

Armenia_ChL_I1634
Armenia_EBA_I1658 0.836±0.065
Sredny_Stog_I6561 0.164±0.065
chisq 13.813
tail prob 0.312808
Full output

The presence of a significant, unambiguous signal of steppe ancestry in a group from a rich archeological site in Chalcolithic Transcaucasia might be very important in the context of the Proto-Indo-European (PIE) homeland debate. That's because it suggests that there was a movement of peoples potentially speaking dialects of PIE from the Eneolithic Pontic-Caspian steppe, the main candidate for the PIE homeland based on historical linguistics data, into cultural hubs south of the Caucasus, which may have acted as early dispersal points for Indo-European languages into other parts of the Near East, such as Anatolia. Admittedly, though, I'm still a fan of the Balkan route for the introduction of Hittite and other Anatolian languages into Anatolia, despite recent claims in scientific literature that this scenario wasn't corroborated by ancient DNA (see here).

Citation...

Lazaridis et al., Genomic insights into the origin of farming in the ancient Near East, Nature volume 536, pages 419–424 (25 August 2016), DOI: https://doi.org/10.1038/nature19310

See also...

Yamnaya: home-grown