Southern Arc or Southern farce?

[Anveṣaṇam]

So I've never really done anything on AG apart from reading the threads. It's my first time starting one! Here are my two substack posts regarding the Southern Arc PIE theory and their inconsistencies with respect to Southern Arc, Yamnaya and Steppe_Eneolithic.

https://vicayana.substack.com/p/sout...southern-farce

https://vicayana.substack.com/p/sout...uthern-farce20

I'd like to know what people over here think of them!

[RCO]

A very ancient, populated and complex Iranian population since the Mesolithic full of Neolithic innovations and technologies in the same CHG-IRAN continuum from Anatolia, Caucasus, Northern Mesopotamia, Iran, Central Asia related to the same Bronze Age PIE linguistic continuum of Anatolian and Iranian languages related to J haplogroup moving and invading new lands and populations in all directions to create completely new Iranian admixed populations always related to some J1 founders in Western Anatolia, in the Steppe, in Central Asia, in the Levant and Arabia with different movements and languages in the same Bronze Age expansion. The big question about the Southern Arc is the Caspian/Iranian Eastern Wing not yet investigated.

[Guiguitargz]

One obvious comment on your first post:
What happen if you try to replicate panel F of Fig.4 of the Southern Arc paper ?

Here you extracted two "dates", when we clearly see that panel F of Fig.4 do contains some outliers that will favor higher admixture times if taken alone (and on the other hand, some other outliers favor a significantly lower admixture time).

To convince me that the paper did something wrong, you need to replicate this figure, with the same samples (== carbon dated samples) or other carbon dated samples and get a different result than them. Then, the origin of the discrepency need to be explore to know who did something wrong in the analysis.

Basically, if for the same sample(s) you get drastically different results than them, I would invite you to check deeply your results, and then contact the author of the paper to investigate on "why" you are not getting the same result than them when using the same samples.

I would have few other comments, but this one is the most critical.

[Anveṣaṇam]

Thanks for the comment.

Regarding your critique, I don't need to replicate that figure. Not because it can't be done or I don't want to. The reason is simple, I never claimed I get different results than them on same target for estimating admixture dates because I apply dates on Steppe_En and Khvalynsk, not Yamnaya.

[epoch]

I have this impression that someone needs to replicate the formal stats with these CWC samples: Plinkaigalis242 and Gyvakarai1 from Mittnik 2018 and Latvia_LN1 from Jones 2017.

These are CWC samples that form a clade with Yamnaya and are therefore likely similar to the population from which CWC originated. If they don't show similar affinity to Levant PPN in exactly the same setup we can say a likely PIE group did not have any specific ancestry to anything south of the Caucasus *even in their own setup*.

https://eurogenes.blogspot.com/2018/...m-genetic.html

[Guiguitargz]

Regarding your critique, I don't need to replicate that figure. Not because it can't be done or I don't want to. The reason is simple, I never claimed I get different results than them on same target for estimating admixture dates because I apply dates on Steppe_En and Khvalynsk, not Yamnaya.

Then it is purely a matter of interpretation.
If you don't get same admixture dates for Steppe_En and Khvalynsk compared to Yamnaya ... it means that the concerned admixture happened at least twice at different moments from different but closely related populations (or that the estimator is very unstable ... about that, I have no clear opinion, it is an estimator that I didn't know in depth).

If you didn't challenge the result they got from the samples, it seems weird to challenge the date they obtain (that is the result you apparently didn't challenge).
The admixture event they "need" for their language diffusion model apparently happened (and if I follow you well, you didn't challenge this for Yamnaya).
If you find an older but similar admixture event ... it changes nothing for their diffusion model.

Reading your claim :

Lazaridis et al., 2022 suggested that a West Asian or Transcaucasian population related to Neolithic people of Armenia or Chalcolithic people of the Caucasus to Southeast Anatolia contributed around ~21 to 26% of the ancestry of the bronze age Yamnaya pastoralists. However, the researchers also did not detect any Anatolian/Levantine–related ancestry in Eneolithic populations at Khvalynsk and Progress-2, which implies that aforementioned Eneolithic piedmont steppe and forest steppe populations did not receive any input from the south after the admixture of EHG and CHG.

You seem to assume that Yamnaya EHG/CHG admixture is sourced by the same admixture event than for Khvalynsk. Can you prove that ? (this is a genuine question).
In fact, as you didn't challenge their date for Yamnaya, but you find another date for Khvalynsk, you are already settling the question ... you need at least two distinct admixture events.

Your result didn't seems to challenge their admixture model.
In fact, if the same EHG/CHG admixture event is sourcing Yamnaya and Khvalynsk ... it would means that the admixture time estimator is unstable and provide inconsistent results (and thus you cannot know which result would be the one closest from the reality, theirs or yours ?).

So your result didn't show that their diffusion model for IE languages is wrong, it shows that the EHG/CHG admixture likely happened multiple times.

On a personal note, I'm not sure that ancient DNA is the most adapted tool to study language diffusion (particularly when we speak about populations for which we don't know what language they spoke).
Yet, I don't think the southern arc paper is settling the question ... many other diffusion models can still work (even "exotic" stuff).
But I also don't think that your posts are challenging the model they propose for IE language diffusion.

[Ariel90]

It seems crazy to me that we just assume that this "levantine" component is relevant. It is, if it actually exist, a trace admixture. It seems implausible that such small ancestry brought an entire language family in the steppe. Also this is "ad hoc" argument only used in this case, because trace admixtures are usually, if not always, irrelevant.

[Ariel90]

So your result didn't show that their diffusion model for IE languages is wrong, it shows that the EHG/CHG admixture likely happened multiple times.

No at all. It implies that weak results should be contextualized or ignored, and that we should follow the principle that governs all modern science (Occam's razor). Which one is the simplest explanation? That two distinct events, in completely different time frames, gave birth to similar populations even if the source populations didn't exist in the later time frame? Or simply that one population generated the other? Eneolithic steppe + Ucraine Neolithic is a perfect fit for Yamna. The alternative model requires to put in the steppe imaginary populations that didn't exist anymore in their unmixed form in that time frame, and that weren't even near the steppe as contemporary Caucasus populations had obviously nothing to do with Yamna.

[Anveṣaṇam]

Then it is purely a matter of interpretation.
If you don't get same admixture dates for Steppe_En and Khvalynsk compared to Yamnaya ... it means that the concerned admixture happened at least twice at different moments from different but closely related populations (or that the estimator is very unstable ... about that, I have no clear opinion, it is an estimator that I didn't know in depth).

If you didn't challenge the result they got from the samples, it seems weird to challenge the date they obtain (that is the result you apparently didn't challenge).
The admixture event they "need" for their language diffusion model apparently happened (and if I follow you well, you didn't challenge this for Yamnaya).
If you find an older but similar admixture event ... it changes nothing for their diffusion model.

It's not a matter of interpretation or of semantics. And wasn't it common knowledge that CHG-related ancestry arrives in two waves according to Lazaridis et al. 2022 itself? 1st a pure CHG wave forms Progress-2 and Khvalynsk and then a 2nd CHG wave arrives but this time with some ANF/Levant_PPN-related ancestry and supposedly with "PIE" or "Core PIE" languages. I've also mentioned the reason why Steppe_EN & Khvalynsk don't get the same results as Yamnaya. That's because of confounding, read a bit more about DATES in the paper Chintalapati et al. 2022. DATES usually gives the admixture date for the most recent admixture event, and that's why Yamnaya's admixture dates are younger than that of Steppe_EN and Khvalynsk.

You seem to assume that Yamnaya EHG/CHG admixture is sourced by the same admixture event than for Khvalynsk. Can you prove that ? (this is a genuine question).
In fact, as you didn't challenge their date for Yamnaya, but you find another date for Khvalynsk, you are already settling the question ... you need at least two distinct admixture events.

Your result didn't seems to challenge their admixture model.
In fact, if the same EHG/CHG admixture event is sourcing Yamnaya and Khvalynsk ... it would means that the admixture time estimator is unstable and provide inconsistent results (and thus you cannot know which result would be the one closest from the reality, theirs or yours ?).

So your result didn't show that their diffusion model for IE languages is wrong, it shows that the EHG/CHG admixture likely happened multiple times.

Please, read the Lazaridis et al. 2022 paper again. I'm not assuming any such thing. Yamnaya cannot be modelled without Steppe_EN so it's not possible for Yamnaya to have a vastly distinct northern source than Steppe_EN. In all of Lazaridis et al. 2022's plausible qpAdm models (p ≥ 0.05), Steppe_EN is a source for Yamnaya. So, they both (Steppe_EN/Khvalynsk and Yamnaya) do get formed from the same CHG waves.

And it was their own result that CHG admixes twice, not mine. I just provide more nuance for these two admixture events.

On a personal note, I'm not sure that ancient DNA is the most adapted tool to study language diffusion (particularly when we speak about populations for which we don't know what language they spoke).
Yet, I don't think the southern arc paper is settling the question ... many other diffusion models can still work (even "exotic" stuff).
But I also don't think that your posts are challenging the model they propose for IE language diffusion.

Sorry to say but I think that's because you haven't read the paper properly.

[Anveṣaṇam]

I have this impression that someone needs to replicate the formal stats with these CWC samples: Plinkaigalis242 and Gyvakarai1 from Mittnik 2018 and Latvia_LN1 from Jones 2017.

These are CWC samples that form a clade with Yamnaya and are therefore likely similar to the population from which CWC originated. If they don't show similar affinity to Levant PPN in exactly the same setup we can say a likely PIE group did not have any specific ancestry to anything south of the Caucasus *even in their own setup*.

https://eurogenes.blogspot.com/2018/...m-genetic.html

I will try if time permits, @epoch.