search this blog

Saturday, June 18, 2016

Genetics of an early Neolithic pastoralist from western Iran (Gallego Llorente et al. preprint)

Just in at bioRxiv:

Abstract: The agricultural transition profoundly changed human societies. We sequenced and analysed the first genome (1.39x) of an early Neolithic woman from Ganj Dareh, in the Zagros Mountains of Iran, a site with early evidence for an economy based on goat herding,ca. 10,000 BP. We show that Western Iran was inhabited by a population genetically most similar to hunter-gatherers from the Caucasus, but distinct from the Neolithic Anatolian people who later brought food production into Europe. The inhabitants of Ganj Dareh made little direct genetic contribution to modern European populations, suggesting they were somewhat isolated from other populations in the region. Runs of homozygosity are of a similar length to those from Neolithic Anatolians, and shorter than those of Caucasus and Western Hunter-Gatherers, suggesting that the inhabitants of Ganj Dareh did not undergo the large population bottleneck suffered by their northern neighbours. While some degree of cultural diffusion between Anatolia, Western Iran and other neighbouring regions is possible, the genetic dissimilarity of early Anatolian farmers and the inhabitants of Ganj Dareh supports a model in which Neolithic societies in these areas were distinct.

Gallego Llorente et al., The genetics of an early Neolithic pastoralist from the Zagros, Iran, bioRxiv prerprint, posted June 18, 2016, doi:

The same individual, tagged as GD13a by Gallego Llorente et al., is also featured in the new Lazaridis et al. prerprint as I1290 (see here). Not sure if there's much point in two different papers on the same sample, but at least the sequences are different.

In any case, here's a very interesting part from the paper dealing with the population history of South Asia:

It is possible that farmers related to GD13a contributed to the eastern diffusion of agriculture from the Near East that reached Turkmenistan (34) by the 6th millennium BP, and continued further east to the Indus Valley (35). However, detecting such a contribution is complicated by a later influx from Steppe populations with Caucasus Hunter-Gatherer ancestry during the Bronze Age. We tested whether the Western Eurasian component found in Indian populations can be better attributed to either of these two sources, GD13a and Kotias (a Caucasus Hunter-Gatherer), using D-statistics to detect gene flow into an ancestral Indian component (represented by the Onge). For all tests where a difference could be detected, Kotias acted a better proxy than GD13a (Fig. S9 and Table S6). This result implies that the majority of the West Eurasian component seen in India derives from the Bronze age migrations; this interpretation is supported by dating of last contact based on patterns of Linkage Disequilibrium (36).

Dating admixture events with Linkage Disequilibrium patterns is somewhat controversial, but what they're saying is more or less in agreement with what I've been bleating about on this blog for the last few years (for instance, see here). So I'm very happy to finally see others noticing the same thing.

However, I have to say that the Principal Component Analysis (PCA) in this paper is off the wall. Many of the ancient samples don't appear to cluster where they should. For instance, most of the European Hunter-Gatherers (HGs) are way too close to present-day samples, and in fact in some cases they're overlapping with them, which is just wrong. Note also the unusual elongated cluster formed by the Neolithic Anatolians, stretching all the way from where they should all cluster (south of the Sardinians), to where they really shouldn't (right next to the North Caucasians).

My bet is that the results are affected by missing markers, with the ancient samples with high rates of missing markers being pulled into the middle of the plot (towards 0.00 in both dimensions).

Below is a similar PCA that I ran using my dataset minus GD13a, which I don't yet have access to. Note the relatively tight clusters formed by all of the ancient populations. The European Hunter-Gatherers are distinct from present-day Europeans, while none of the Neolithic Anatolians (Anatolia_N) fall near the North Caucasians in dimension 2. This is of course in line with a wide range of formal stats.

See also...

Early Neolithic genomes from the eastern Fertile Crescent (Broushaki et al. 2016)


Nirjhar007 said...

I am gonna say this very gently. The papers sequencing not very trusting. Kotias Type ancestry in India was there from Neolithic at least .

After some aDNA from India and SC Asia , that will become crystal clear.

Hopefully in some weeks..

Slumbery said...


Aren't the CHG like ancestry in South Asia much higher than any EHG or ANE portion? Since it is not true for the steppe populations, South Asia cannot got the majority of the CHG like component from the Bronze Age steppe. Or at least we need to assume that BMAC (or any South Central Asian population mixed in the way) was characteristically CHG and not Iranian. However the plausible migration route for that is also Iran, so we can as well assume that a population closer to CHG that to Iran_Neolitic reached South Asia directly (we do not exactly know where the borders of the two populations were to begin with).

Kurd said...

I found their methodology problematic as far as using one dstat of the form D(Yoruba, CHG or Iran-N, Onge, S Asian) to conclude that CHG is a better proxy that Iran_N for ANI. Whereas, it is true that CHG appears to share more drift with S Asians than Iran_N, this IMO is not necessarily by virtue of S Asians being more CHG than Iran_N derived, but rather it is because both S Asians and CHG are more EHG admixed to the apparent exclusion of Iran_N, remembering that Lazaridis modeled CHG as about 70% Iran_N.

To recap ANI is the W Eurasian portion of S Asians' admixture in a 2 way ANI-ASI model where ASI is presumed to be distantly related to Onge.The jist of the discussion in Supplementary Information 9 (pg 121) of Lazaridis et al is that they used qpAdm to infer mixture coefficients for various W Eurasian test samples. Using a set of right pop outgroups defined as S7, they tested for the p-value for a rank =2 for S Asian = W Eurasian + Onge + Han. Incidentally, the p-value I believe is nothing more than the tail probability in qpAdm. The results are shown in table S9.1. So basically they are trying to reject possible sources for ANI, without making a claim about its identity. What we see in this table is that in most cases a single W Eurasian sample can be rejected as the ANI surrogate based on its low p-value. There are some exceptions that I noticed. For example I noticed that for Balochi Armenia MLBA had a reasonable p-value of 34%.

So basically what this shows is that ANI for most S Asians consists of multiple ancient pops. If we look at tables S9.2 and S9.3, we see that for most S Asians, ANI is best defined as Iran_N combined with steppe (EHG or Steppe EMBA), which sort of contradicts the conclusions here.

Table S9.4 shows the various mixture coefficients for S Asian= Iran_N + Steppe + Onge + Han. The higher the p-value the better the scenario.

So to summarize, no single W Eurasian pop seems to be a good surrogate for ANI, but rather ANI likely consists of Iran_N + Steppe, and not likely CHG as suggested here.

Rami said...

LOL , saying the majority of West Eurasian component in most South Asians derives from steppe Indo Iranians is ridiculous given what we know from Zagros Farmers. Looking at the skull types from IVC Harappa, Rakhi Garhi, these people were Caucasoids and the IVC is derived from proto IVC culture which shares strong affinities to the Neolithic Iranian cultures.

Rami said...

Excellent explanation Kurd!

Davidski said...

The steppe populations that made their way to South Asia had high ratios of both CHG and EHG. On the other hand, early Neolithic farmers that moved to South Asia were very CHG-like.

Overall, this ends up looking like Kotias-related ancestry in D stats. So their conclusions make sense.

Personally I'm skeptical that there were any largescale population movements from Chalcolithic Iran to South Asia. I think the early Zagros farmers and Bronze Age steppe pastoralists make up the bulk of ANI.

postneo said...


Lets assume kotias is the birth place of CHG.

Still kotias(georgia) to IVC mehargarh is a much smaller hop along a lush corridor compared to georgia to ukraine and then ukraine to kazakhistan to afghanistan crossing through some pretty arid areas.

Davidski said...

You'd be saying the same about the CHG, EHG and Z93 in the Altai if not for the Andronovo samples that we have.

Just deal with it already.

postneo said...

The reason yamnaya and ehg work as components for south asia in addition to Iran_N is because of shared ANE btw south asian and the steppe ancestors not because of bronze age migrations. Perhaps there were some bronze age migrations, but I bet they were small, they alone cannot account for the large ANE signature even in southern india.

We will see,

Kurd Dgk said...

@ postneo

The problem is that a single W Eurasian pop modeling of ANI is rejected by qpAdm (see my previous post). Steppe is required in combination with Kotias or Iran_N, not to mention steppe uni parental markersin S Asia. Once you add steppe as a required ingredient, steppe carried CHG is added by default. Lazaridis obtained the best p-values in qpAdm for most S Asians by using a combination of Steppe and Iran_N, and not Steppe and Kotias.

Rami said...

David u must be smoking crack if you think CHG is a better fit for South Asians than Iran_N, unless of of course the IVC was started by Yamnaya people lol, which ironically would mean Nirjar is right loool.

Kurd Dgk said...

With regards to the amount of ANE in S Asians, we are not exactly sure how much ANE Neolithic Iranians contributed to S Asia, or whether other as of yet undiscovered Neolithic Iranian lines with more ANE than the Iran_N sample we have were the ones that migrated to S Asia

Davidski said...

Hey Rami, put this in your crack pipe and smoke it.

The demographic impact of steppe related populations on South Asia was substantial, as the Mala, a south Indian population with minimal ANI along the ‘Indian Cline’ of such ancestry is inferred to have ~18% steppe-related ancestry, while the Kalash of Pakistan are inferred to have ~50%, similar to present-day northern Europeans 7.

postneo said...

"The demographic impact of steppe related populations on South Asia was substantial"

David the word used is "steppe related". In my mind, Karitiana of brazil would also be have this "steppe related" ANE admixture. Maybe I am mistaken, please correct me if so and explain how its different.

Shaikorth said...

They projected all their ancients onto their PCA's.

Jaydeep said...

Dear Nirjhar & Rami,

I think I have a little idea as to why Kotias is a better fit than GD13A for ANI admixture in South Asia.

This in itself does not mean than CHG is closer to South Asians than GD13A. The reverse is the case as you can see from the PCA.

I think the problem arises from trying to find a synthetic population such as ANI totally distinct from ASI admixture. In reality such a population may never have existed in South Asia.

Its known to most that modern Iranian people have a small amount of ASI ancestry. When did it reach Iran ? I think the Iranian Neolithic people like the Ganj Dareh already had ASI and modern Iranians have not received any ASI influx in a later period.

It could be the reason why GD13A is not a better source for ANI than Kotias. Because it has affinity with both ANI & ASI. Iranian Neolithic (GD13A & others) has ASI which Kotias most likely lacks which may also be a reason why Iran Neolithic is also so basal.

In other words, without trying to find a source population for a synthetic group like ANI, what if one tried to find ancestry proportion of GD13A in South Asians vs ancestry proportion of Kotias in South Asians ? I am sure GD13A will come on tops.

Davidski said...


In my mind, Karitiana of brazil would also be have this "steppe related" ANE admixture.

Only in your mind.


They projected all their ancients onto their PCA's.

OK, but so did I.

Shaikorth said...

Alright, I didn't notice because it looked quite similar to this one which you said didn't have projection bias with how Hungary-BA and the Neolithics clustered.

They used 145k transversion snp's of moderns to make the PCA, how many can be gotten out of those ancients?

Davidski said...

I can plot samples accurately as long as they have at least 20,000 SNPs out of the 250,000 SNPs that I use to run the PCA.

So I reckon almost all of the new Lazaridis et al. samples will make it into my PCA.

Gozde Gayde said...

If we measure the Steppe influence on South Asia which happened during Bronze age, we should not use CHG as key component but EHG. CHG probably difused into South Asia during Neolethic or may be Paleothic. If and only if CHG can become key component for european populations for measuring the influence of EBA Steppe ( or so called PIE). Thus if we want to learn whether Kotias or Iran_N type CHG took place in EBA Steppe difusion then we should look at D(Yoruba, CHG or Iran-N, Yamnaya, Han) stats.

Davidski said...

No one's using CHG as a key component.

CHG just looks like the best match for the West Eurasian admixture in South Asia because of the early Neolithic Zagros CHG-like admixture and Bronze Age steppe CHG proper admixture in South Asia.

The EHG that got to South Asia from the steppe, along with CHG, also helps to tilt the main stats towards CHG, because CHG is less basal than the early Neolithic Zagros stuff.

Nothing really complicated.

Jijnasu said...

Now that we have data about neolithic iran, anyone ready to make predictions about south central asia? The western south asian neolithic and the IVC populations will likely resemble populations from iran with some south asian ancestry Hope we get to see data from these regions soon

Matt said...

Hah, the discussion here is dominated by autosomal dna, even though there's not very much.... while the discussion on the comments thread about Laziridis 2016, even though there's much more autosomal dna, is dominated by y-dna and archaeology. Just shows the power of y-dna to obsess any commentators I guess.

Anyway, one positive on this at first blush is the inclusion of a present day, unsupervised world ADMIXTURE analysis with their ancient samples (an unfortunate oversight in Laziridis 2016, unless there are problems with SNP overlap?).

A comment by Gill and Sein on the Laziridis 2016 and the ADMIXTURE analysis here does make me wonder, with the findings now of initial separation between varying early Neolithic populations (it seems to me at least, Anatolia Neolithic, Natufian, Iranian Neolithic, CHG), whether we will find that the purple component here, as modal in Bengali actually represents a real fusion of pre-Neolithic fusion of West Eurasian (mainly ME, but also EHG / ANE) ancestry and East Eurasian ancestry (probably mostly EE though). Rather than the model of the first West Eurasian ancestry in South Asia being from the Iranian Neolithic. Particularly if we believe the models with ANE ancestry in East Asia at 10%.

It'd be great if Davidski could do an unsupervised ADMIXTURE analysis on the Laziridis 2016 data when it's all out. And then we can use nMonte on ADMIXTURE as a cross check against qpAdm, D-stats, PCA models.

Jaydeep said...

Dear Matt,

I think there is another possibility. Please read my comment above. What do you think about that scenario ?

postneo said...

"I think the problem arises from trying to find a synthetic population such as ANI totally distinct from ASI admixture. In reality such a population may never have existed in South Asia."

So you are either questioning the role of constructs such as ANI and ASI or saying that ANI ASI admixed earlier in Iran vs India.

In an interview Reich dated admixture of ANI and ASI at 1,900–4,200 years BP BC, he is also coauthor of

what's your take on this?

Jaydeep said...

Here is my take. Zagros Neolithic is not the source of South Asian Neolithic. I think it's the other way around. During the LGM, South Asia probably was a vast refugium. After LGM, maybe groups started migrating westward - CHG & Zagros Neolithic being the examples of such early movements. While CHG may have moved before the advent of Neolithic, Zagros Neolithic perhaps moved later on at the start of Neolithic. Therefore it is probable that Zagros Neolithic was a western extension of South Asian Neolithic and therefore should have a little bit of ASI affinity just like modern Iran ins who have it despite being admixed with the Near Eastern groups.

Reich et al have clearly said that the supposed admixture between ANI & ASI may have already happened before 4200 BP but they did not have any way to detect it. The way ANI has been modelled is itself quite suspect - It is in effect that part of South Asian ancestry which is shared with West Eurasia. So obviously you will not find ASI much at all outside South Asia. Because if it was it would have been labelled as ANI. I hope I am making sense. By definition of Reich et al, ASI is that part of South Asian ancestry which is absent or negligible out of South Asia. So how will you find ASI in Europe or elsewhere.

Jaydeep said...

Adding to the above, The ANI has been constructed on the assumption that all ancestry that South Asians share with the rest of Western Eurasia is a result of West Eurasian groups entering into South Asia i.e ANI is not native to South Asia. But what if ANI was native to South Asia and it moved out of South Asia to the rest of Western Eurasia ? The dynamics will change completely.

Jijnasu said...

@jaydeep The very fact that ANI and ASI are unevenly distributed across Indian populations shows that they do indeed represent distinct populations

Jaydeepsinh Rathod said...


I don't think so. Let me state it again - ANI is that part of South Asian/Indian ancestry which is shared by Indians with Western Eurasia which is to the north & West of India. ASI is that part of Indian ancestry which is not shared with Western Eurasia.

Now please tell me this - do you expect that in a large place like India, North Indians & South Indians would have been exactly alike even in pre-history though they may have descended from a common ancestor ? Obviously there would have been some differences between North Indians & South Indians i.e. for example some haplogroups will be more concentrated in the South while some will be more concentrated in the North. Is that not correct ?

Now, let us say there is a migration from India northwards into Central Asia. Which population is more likely to make this migration ? North Indians or South Indians ? Obviously North Indians. So who will share more ancestry with the Central Asians ? North Indians or South Indians ? Obviously North Indians. I hope this has been clear to you so far.

Now, as I said though North Indians & South Indians share a lot of haplogroups, some haplogroups will be more concentrated into North India while some will be more concentrated into South India. However, if there is migration from North India into Central Asia which haplogroups will be shared by Indians with Central Asians ? The ones concentrated more into North India. This is what is essentially the ANI. Will North Indians share more ancestry with Central Asians or will South Indians ? North Indians. Hence they will have a greater share of ANI ancestry. Which populations in India are less likely to admix with Central Asians or Europeans ? The ones deep within India i.e. South Indians. So obviously South Indians will have a greater share of population that is isolated and that did not admix with Central Asians. That isolated population is ASI. Obviosuly it will be more concentrated in the South. That is the reason from a ANI cline from North to South & a ASI cline from South to North.

I hope I have not been confusing.

Coldmountains said...

I would rather search for WHG in South Asia and Central to estimate how much steppe admixture modern Indo-Iranians have. WHG was surely not present there before Aryans. Maybe EHG really just arrived with Aryans in South Asia and Central Asia and the people which brought R2 and Q were absorbed by CHG-like people so that no EHG-like ancestry remained there like R1b in Neolithic EEFs but in my opinion at least a bit EHG-like ancestry should exist in BMAC and other Pre-Aryan cultures of this region

Nick Patterson (Broad) said...


> Its known to most that modern Iranian people have a small amount of ASI ancestry.

Do you have a reference for this?

MfA said...

They also sequenced modern Assyrians, Lors, Mazandarans, Bandari and Persians(Shirazi). It's shame still no Kurdish samples in the Human Origins set, even though most of the ancient Iranians are from Kurdistan.

la seƱora bibiloni said...

According to Lazaridis' paper, Hotu Cave sample (Mesolithic Iran) already had a bit of EHG (10% - Figs. 4a & 4b), and why wouldn't more EHG have migrated south between Mesolithic and Yamnaya? So I think you are right, the fact that could define the Yamnaya influence (or lack of it) will be more WHG than EHG

Davidski said...

I don't think Iran_Hotu has any EHG per se. To me it looks like the algorithm is just trying to compensate for the fact that Iran_N is more basal, so to fit Iran_Hotu as Iran_N it needs to "steal" a bit of EHG to get there.

Apart from that, Yamnaya (Steppe_EMBA) doesn't have much WHG. Steppe_MLBA does have a fair whack, but this would have been reduced, at least in theory, by the time the early Indo-Iranians pushed through Central Asia on their way to South Asia.

Rami said...

Davidski apparently you need ESL Lessons, the steppe was not just habited by Androvono people, others were there much before at a point. Do you really actually think Mala have direct approx 20% Indo Iranian steppe ancestry ? Then you must have been deprived of oxygen at birth. Indo Iranian steppe ancestry would be high for groups like Tajiks, but in deeper South Asia, not really.You already have EHG in Iran. Before ANI was thought to be one population , now its 2 apparently . Those groups like Baloch and Brahui even have elevated Yamnaya like ancestry but its not coming from Yamnaya or even Indo Iranians , its coming from an EHG related component already present in the region before the Bronze age. Coldmountains hit the nail, to find proper Indo Iranian ancestry you would have to look for WHG.

Davidski said...

The steppe was not just habited by Androvono people, others were there much before at a point.

And they came from Europe too. Like Afanasievo.

So when the Indo-Iranians expanded from the western steppe, they would have absorbed the remnants of these earlier migrants from the west, and become less like Potapovka/Sintashta, and more like Afanasievo as they pushed closer to South Asia.

Nirjhar007 said...

There are methodological issues in the paper . We need further retestings for the finer resolution . I do hope Guys like Kurd,Felix etc are also very avid to do so, as it is quite needed...

So, in simple sentence, no need to start narrate their conclusions yet.

Davidski said...

There are methodological issues in the paper.

And how did you come to that conclusion?

Karl_K said...


"Do you really actually think Mala have direct approx 20% Indo Iranian steppe ancestry ? Then you must have been deprived of oxygen at birth."

Why would the LD analysis look like the admixture of ANI genetics into these populations fit the timeline for this being true if it wasn't?

If they only had 5% Indo-Iranian ancestry, and 15% older steppe like ancestry, then wouldn't it look like the admixture was at a much earlier date?

postneo said...


There is no special structure to ani and asi just a diffuse cline as would be expected.

If ani is apriori handpicked to have west eurasian then any peninsular dna that made it out from South Asia would be artificially masked.

On the other hand there's no great heuristic to divide subcontinental Dna, what's a good dividing line in a continuum? Without ancient Dna from different regions of India The current ani asi is ok if used with some caution.

postneo said...

If one had Dna from imangaon vindhya Neolithic mehargarh and megalithic those would be more genuine ancestral groups

bharatiya said...

ASI and ANI don't form a diffuse geographical cline. There's a substantial difference between tribal populations and caste populations from the same region. likewise there exists smaller variations between caste groups all this indicates that ASI and ANI are real populations that underwent relatively recent admixture

Jaydeepsinh Rathod said...

Dear Dr. Patterson,

I am assuming you are the geneticist who is a co-author of Lazaridis et al 2016 paper. Let me congratulate you and your entire team for publishing the most comprehensive and brilliant aDNA paper published so far. I hope that in the near future you are able to access aDNA from India as well.

> Its known to most that modern Iranian people have a small amount of ASI ancestry.

Do you have a reference for this?

You may refer to -


Please refer to the admixture graph figure 2B. There are two components that dominate South Asians populations - K5 (light green) & K6 (dark green). K5 corresponds to ANI & K6 corresponds to ASI. If you looked at the Middle East & North Africa section of admixture graph, you will see that the dark green (K6 or ASI)component is present among the Iranians in a small but significant proportion unlike the other Middle Eastern groups.


Please refer to figure 1C. Here too you can see that the dark green component - a signal of ASI among South Asians, is also present in the Iranians & Central Asians in a small but significant proportion.


I hope this was helpful. And all the very best for your future researches.

bharatiya said...

@rami I did suppose earlier that the ratio of steppe:neolithic ancestry in south asians as a whole would be lower but it appears otherwise. I see no reason why Mala should not be 20% Indo-Aryan in ancestry. Bilingualism was widespread in ancient peninsular India, the earliest dravidian literature shows a significant amount of influence from more northern cultures. It appears for whatever reasons that the steppe pastoralists managed to outbreed older farmers

postneo said...

"There's a substantial difference between tribal populations and caste populations from the same region"
"It appears for whatever reasons that the steppe pastoralists managed to outbreed older farmers"

These regional discontinuities are due to endogamy. Also caste populations have greater IVC ancestry. Whereas tribal populations would be more HG. This is a simpler explanation than bringing in steppe and it also explains much better why IVC derived populations outbred tribal populations simply because they were a massive population to start with. An IVC population would also score higher in western eurasian relatedness simply due to their location.

postneo said...

"steppe pastoralists managed to outbreed older farmers"

In europe atleast hand there is some evidence of resurgance of HGs and some unresolved steppe like people vs first farmers

there is no dna from south asia that differentiates between "older farmers", "newer farmers".
These are just conjectures. All we have is modern DNA from INdia and some ancient DNA from Iran where we have several kinds of differentiated farmers. They seem to have lower ANE than modern south asians. This difference exists even today. For all we know this was how it was in the bronze age too.

People are free to claim anything.

Karl_K said...


"There is no special structure to ani and asi just a diffuse cline as would be expected."

As expected from what?

bharatiya said...

Agreed we do not have ancient DNA from south asia or south-central asia yet, but we do have data from most of western eurasia and modern data from India from which we can make some inferences. Archaeological data available (and if the the conference poster for the IVC DNA are to be believed) the populations of the IVC and the mehrgarh neolithic were closely related to those of the iranian plateau. We now know that the zagros neolithic alone is inadequate to model the ANI component of Indian populations, and that the early and middle bronze age DNA from the steppe seems to be a good candidate for another source of ANI component of Indian ancestry. The Y chromosomal data also seems to point towards this hypothesis.

You could of course be bold and claim that unexpected finds from South central Asia and South Asia will completely change things and maybe it might. But that doesn't change the fact that all current data seems to point to the steppe for a portion of Indian ancestry

Matt said...

Coldmountains: I would rather search for WHG in South Asia and Central to estimate how much steppe admixture modern Indo-Iranians have. WHG was surely not present there before Aryans.

I think this would be a good sense check (or else Anatolia_Neolithic fraction but may not work for all populations?). *Or* you'd need to use a method in any case which allows ANE, preferably AfontovaGora3, as a contributing population.

The ADMIXTURE analysis in this paper does not distinguish between EHG and WHG (which is a common tendency for ADMIXTURE), but based on the blue fraction (EHG+WHG), that's suggestive of a 0.33 contribution of Sintashta to Pathan or a 0.33 contribution of a Yamnaya related group without Anatolian influence to Kalash (not that I find the 0.5 steppe estimates particularly infeasible or anything).

(For Europeans, using the green Iranian Neolithic / CHG component instead and modelling it all as coming from Yamnaya seems to give estimates in line with what we know. Obviously you can't use the blue, because of contribution from both WHG and EHG).

Any kind of direct ANE contribution could be critical to model as, just as using EHG as a contributor rather than Yamnaya drops the contribution required, so would ANE contribution drop the EHG and steppe contribution required. Direct ANE contribution to ASI groups like Mala would also resolve any apparent counter-intuitiveness about these groups having steppe fraction (as much as reality can be counter-intuitive at times, so no reason to dismiss this out of hand). If this contribution is further back in time than the Bronze Age also may fit better with having a common component defined by common drift .

If ANE contribution to East Asia in the models from Lazaridis's paper is correct (again to be tested with AG3) this raises the plausibility of contribution to pre-neolithic South Asia.


@ bharatiya, IRC, what we have so far about the Indus Valley Civilization / Harappa / Rakhigarhi is -

"The first genetic exchange must have occurred from Neolithic period, between the Indus Valley and the Iranian Plateau. During the 1st millennium BC, the secondary genetic exchange based on seaborne trade could be identified".

Do we have any more definitive statements on how closely related these groups were to present day Iranians (as I assume they had no Neolithic Iranians at the time of analysis)? I may not remember if we do.

From that statement alone, it could even be a mostly ASI population with some increasing links to Iranians (via steppe or Iranian neolithic populations), but not as much as in present day people (and some increase in the 1st millennium BC).

Matt said...

From the paper's ADMIXTURE:

K13: (roughly splits South Asian populations into GD, Euro HG, Onge).

Alberto said...

We've gone through the models before, and I think we all have an idea of the proportions of the components. But where did each one come from is where we can't get an answer without aDNA.

One interesting case first highlighted by RK is the GujaratD vs. GujaratA, which seems to clearly represent an impact of an Andronovo-like population:

"Munda" 55.1
"Kotias" 19.25
"Anatolia_Neolithic" 14.7
"AfontovaGora3" 6
"Andronovo" 4.75
"Esan_Nigeria" 0.2
"Atayal" 0
"Loschbour" 0
"Karelia_HG" 0
"MA1" 0

"Munda" 37.95
"Andronovo" 23.1
"Kotias" 18.8
"Anatolia_Neolithic" 15.05
"MA1" 5.1
"Atayal" 0
"Esan_Nigeria" 0
"Loschbour" 0
"Karelia_HG" 0
"AfontovaGora3" 0

We see the Kotias, Anatolia Neolithic and ANE remain stable, while Andronovo goes up at the expense of Munda (one also can wonder why is this, since if GujaratiA was a GujaratiD + Andronovo we would expect all other components to decrease proportionately, but things are never so simple).

But this pattern does not hold with other populations, where components vary freely:

"Munda" 35.1
"Kotias" 25.7
"Anatolia_Neolithic" 18.4
"Andronovo" 11.05
"MA1" 9.2
"Esan_Nigeria" 0.55
"Atayal" 0
"Loschbour" 0
"Karelia_HG" 0
"AfontovaGora3" 0

"Munda" 26.85
"Kotias" 24.7
"Anatolia_Neolithic" 19.2
"Andronovo" 16.6
"MA1" 12.65
"AfontovaGora3" 0
"Atayal" 0
"Esan_Nigeria" 0
"Karelia_HG" 0

Nick Patterson (Broad) said...

@Jaydeepsinh Rathod

Thank you for the references; if you want to send your email
address to me
I'd appreciate it.


postneo said...

As expected from distance from ivc and the fact peninsular is bounded by ocean

Nothing anomalous about asi cline wrt west Eurasian populations as well . I would expect such clines to always exist.

It would be abnormal to think that asi was uniformly spread all over the subcontinent when such clines and regional variation are the norm in ancient Dna as well.

You simply cannot resolve

You favor a Ukrainian demographic with no effect of ivc population or the assumption that there was no difference btw ivc and asi.

There's only one sample from hotu
Is there a variation or cline btw that vs the western Iranian samples? That may be the beginnings of a clue.

postneo said...

I meant @bharatiya

Matt said...

Re: ASI in modern Iranians, in the K13 and K14 I have linked above here, there is also clear presence of the Onge component in the Iranian sample at K13 (not at a high level but around 1/10 to 1/6 of what it is in the highest Indian population), and then the ASI component which absorbs it at K14. So this is still present even in the presence of the ancient samples in this ADMIXTURE run.

bharatiya said...

I never said ivc didn't contribute to south asian ancestry. I just said steppe ancestry seems to have played a more recent and perhaps more important role.
Also I never said IVC represents ASI, rather I said the people of the IVC were probably the result of admixture between farmers from the iranian plateau (a component of ANI) and local hunter gatherers (ASI) (My guess is that they had a larger portion of iranian neolithic ancestry which would make them ANI dominant)
I'm not suggesting that they leap frogged from ukraine, we do have evidence of similar populations furthers to the east from Poltovka, Afanasievo and later from Sintashta and Andronovo.
Only some very unexpected finds from south-central asia and south asia can dislodge this hypothesis and for now the possibility seems unlikely

postneo said...

"we do have evidence of similar populations furthers to the east from Poltovka, Afanasievo and later from Sintashta and Andronovo"

Sure but they are contemporary with IVC. lets wait and see.

inferring a cline Hotu vs western Iran is tricky since Hotu is far older than other samples and its just one lone sample.

But I see that despite its age Hotu(poorly calibrated to be sure) is the most Iran_chalc shifted sample among the Iran_neolithic. We might infer a trend that zagros farmers had progressively increasing demographic impact from eastern Iran after the population further east took to farming.

postneo said...

hotu may show early signs of whatever shifted western Iran closer to Yamnaya in the chacolithic. Anyway nothing conclusive some faint signals though..

Jaydeepsinh Rathod said...

Dr. Patterson,

"Thank you for the references; if you want to send your email
address to me
I'd appreciate it. "

I have already sent you an email. Thank you for your kind gesture.

Thank you for the information. I think that higher Ks, purple component of ASI also appears in GD13A. At k=20, it is replaced by the very prominent presence of the Kalash component. What do you make of this ASI component already present 10,000 YBP in the Ganj Dareh sample ?

Ryan said...

Re: the crack pipe discussion - is it such a stretch to consider potentially an earlier wave of steppe ancestry into South Asia? If we believe Y haplogroup R and Q are steppe markers, then R2 requires some explaining, even if it came via Iran.

Alternatively, the population that initially colonized the steppe may have originated in South Asia, in which case it could be very hard to distinguish between old affinities that predate the steppe and newer steppe population movements prior to the Indo-Aryans arriving on the scene.

Karl_K said...


"Alternatively, the population that initially colonized the steppe may have originated in South Asia, in which case it could be very hard to distinguish between old affinities that predate the steppe and newer steppe population movements prior to the Indo-Aryans arriving on the scene."

Besides simple admixture, this also has linkage disequilibrium data to back it up.

It definitely requires some newer analysis, but at least half of the steppe-like ancestry in the south of India seems to have admixed in within the last 4,000 years.

That doesn't look like old R2 steppe admixture from 10,000 years ago.

postneo said...

"It definitely requires some newer analysis, but at least half of the steppe-like ancestry in the south of India seems to have admixed in within the last 4,000 years."

linkage disqeuilibrium only tells you time of admixing not source. Hopefully it can be re-examned with genuine data not just modern samples. The source of the "steppe like" source cold be IVC.

ASI-ANE cline in S Asia seems to be old. The hint comes from Iran Clac which is ANE shifted vs old zagros. This is because agriculture reached critical mass in S asia a bit later than levant/zagros leading to a demographic boom affecting western Iran by the Chalcolithic.

Davidski said...


Alternatively, the population that initially colonized the steppe may have originated in South Asia.

Which steppe component might be from South Asia?

EHG, CHG, WHG or EEF? Any others?

They all look like they date to the Paleolithic in West Eurasia, no?

Ryan said...

@David - ANE as a whole.

Definitely date to the Paleolothic. The Paleolithic was a long time though. People arrived at the steppe from somewhere. I'm talking 40kya+.

If you doubt the possibility, I'd put it to you to explain why how the other branches of K2b ended up in SE Asia.

Davidski said...

So you're hanging your hopes on the ancestors of MA1 and/or AfontovaGora3 coming to Siberia from South Asia, and obviously on the often hypothesized division between Paleolithic north Indians and south Indians. Because otherwise how do you explain the extremely low affinity of Ancestral South Indian-rich Panyia to MA1?

Quite apart from that, you've got the problem of the rapid and massive expansion of R1a-Z93 in South Asia, which is likely to post date the oldest Z93 samples we have from the steppe.

postneo said...

Armenia EBA is closer to Iran_chalc vs earlier Armenian_chl and later Armenia_MLBA whic are closer to anatolian_chalc

Davidski said...

Not by much.

More importantly, Iran_ChL is a new population in Iran, because it's significantly shifted northwest of Iran_N and Iran_Hotu.

So positing any serious population movements from Iran to the north during the Chalcolithic doesn't make much sense, and certainly not onto the steppes, since the uniparental markers simply don't match.

There's no obvious overlap, so the autosomal similarity is a coincidence.

Ryan said...

David - You're mistaking my position for one that it's not. I'm talking a 45kya expansion as a confounding factor. R1a expanded to India from the steppe, and brough with it a lot of ANE. I'm not denying that. I'm not some OOI zealot. Frankly, at 45kya+, South Asian DNA would probably only be barely recognizable as South Asian.

I'm questioning whether or not it is responsible for *all* ANE-related ancestry in South Asia, or if it's just the largest source. I'd ask you to actually read what I write and not throw up strawmen.

I'm not going to pretend I think there's any sort of strong evidence for or against anything I'm discussing either. I'm just raising the possibility. If you have an alternate explanation for the distribution of R2 and K2b I'm all ears.

Interesting point on Paniya. Strangely close to Ust-Ishm too.Any other South Indian populations to compare to? I couldn't see any on the chart but it gets a bit bunched up. The Paniya are mostly F-M89*, are they not? It would be interesting to see if a group like the Kadar that has a lot of R2-M124 showed different results.

All ears if you have an explanation here though. How did R2 get to South Asia? Where was it originally? When did it get there? Where was Siberia colonized from? Why are other branches of K1b centred on SE Asia?

Davidski said...


None of this is relevant to detecting Bronze/Iron Age steppe ancestry in South Asia.

As you can see from the plot I posted, Eurasian ancestry dating to ~45K YBP and associated with Y-hg K2 is not a confounding factor in recognizing the very recent gene flow from West Eurasia to South Asia.

Take a look at the plot more closely. The Kalash and Pathans almost cluster with Northern Europeans along the line of best fit. This is in accordance with their inferred levels of steppe ancestry.

If their ANE (Mal'ta cluster) affinity was in large part more ancient, this would not happen, since it would also register as affinity to Ust_Ishim.

Ryan said...

I'm not saying it's a confounding issue for Kalash or Pathans or any modern day population in the northern half of the subcontinent.

I do think it's something to consider for non-IE populations in the southern half of the subcontinent, and for populations that pre-date the IE expansion (such as these samples from Iran).

R2 seems to be most prevalent in non-IE Indians, and in particular among tribals and scheduled castes, so it seems unlikely that it arrived with R1a and IE languages. Either R2 arrived in India from a source to the north, or it arose there. Which is it, in your view?

Also, from your Jtest K14 post, which I think may be relevant here too:

That’s quite a problem, because due to their relative inbreeding, they produce strong ancestral clusters in many analyses, like in ADMIXTURE runs. However, these clusters are made up of allele frequencies from a wide range of sources and, paradoxically, it’s the relatively more outbred populations which contributed to the Ashkenazi gene pool at its formative stages that often end up showing Ashkenazi admixture in such tests, despite not having any.

Davidski said...

I don't know where R2 is from and how it got to India. But my guess is that it's from Neolithic Iran.

If so, then any ANE-related ancestry associated with R2 carried by Indians should be taken care of by using Iran_N as a reference population, as was done in the new Lazaridis et al. paper.

Of course, when the data is released I'll also try and model South Asians as Iran_CHL + Steppe_EMBA and Steppe_MLBA, just to see what happens to their levels of steppe admixture. But at this stage, it doesn't look like R2 was associated with any of those CHG-rich groups like Iran_ChL and Armenia_ChL.

Ryan said...

How would it have arrived in Neolithic Iran though? From the steppe directly I assume? I could buy that.

Davidski said...

No idea. But probably same way that P1 got to Neolithic Iran.

Ryan said...

I thought we were operating under the assumption that that P1 was in fact R2?

Davidski said...

It might well be, so if it's R2 then it definitely got to Iran the same way as R2.

postneo said...

"So positing any serious population movements from Iran to the north during the Chalcolithic"

I am not positing any far flung migrations. There's no shortage of "folks" who do, I am not one of those.

Armenia is a pretty tiny its 1/7th the size of Gujarat with 1/20th the population, so any variation is interesting. One of the Armenia_EBA happens to be R1b as well but I think thats not very significant.

What I am saying is that Iran_chalc picks up signals of an ANE population explosion driven by BMAC and IVC. It is not central to it but closer to it than Armenia. Hotu shows some early signs of the existence of such a population since it is further east.

Davidski said...

What I am saying is that Iran_chalc picks up signals of an ANE population explosion driven by BMAC and IVC.

Then why is Iran_ChL shifted northwest compared to the earlier Iran_N from the same part of the country in western Iran?

Last time I looked BMAC and IVC sites were located east of western Iran, no?

Rami said...

Your stupidity never fails to amaze me. Before you were clamoring how Baloch were recent migrants, well that turned out to be BS in light of the Lazardis paper, and even that graph which you posted showed it but you later took it down, because you were too butthurt to admit you were wrong . Obviously groups like Pashtuns and Kalash will have higher Indo Iranian steppe ancestry, as the Gandhara Grave culture is centered around where they live, coupled with the arrival of other East Iranic groups much later, but even then its inflated superficially by EHG like admixture already present. For most South Asian groups, the story is different. PJL Chamars score 0% Euro ancestry in many cases, if they had actual legit Indo Iranian ancestry it would show up but it does not. Also it makes no sense Yamnaya ancestry would be a better fit than Indo Iranians, as those Yamnaya people did not even step foot in South Asia. If you seriously think groups like Mala, Pania, or even Brahui/Makrani have 20-30% steppe admix from Indo Iranians, I truly pity you. Even South Indians are scoring ANE at levels higher than those Iranian Sintashta.

ryukendo kendow said...
This comment has been removed by the author.
ryukendo kendow said...
This comment has been removed by the author.
ryukendo kendow said...
This comment has been removed by the author.
ryukendo kendow said...
This comment has been removed by the author.
ryukendo kendow said...
This comment has been removed by the author.
Davidski said...

There really is a huge problem with modeling Yamnaya as part Iran_ChL, because the uniparental data is out of whack.

Iran_ChL has the two mtDNA diagnostic markers for that part of the world south of the Caspian, U3 and U7, and these are totally missing from the Bronze Age steppe samples to date. They shouldn't be missing if there was a substantial migration from Iran to the steppe.

This has to be explained before this model can be thought of as anything but a decent statistical fit.

ryukendo kendow said...
This comment has been removed by the author.
ryukendo kendow said...
This comment has been removed by the author.
Davidski said...

Anatolia_ChL is interesting. Pity it's a low coverage sample with few markers.

I wonder if it's similar to some of the Cucuteni-Tripolye groups, and perhaps a decent fit for a part of the southern admixture in Yamnaya?

Ryan said...

RK - The Nartufian samples are probably a pretty good bet for proto-Semitic.

Rob said...


Thanks for your insights. A couple of Q's :

* "Anatolia_Neolithic still requires a currently unsampled population slightly further west than WHG on the WHG-EHG cline to get represented accurately"

You definitely meant "further **west**"? How can something near / in Anatolia be further west than the archetypal WHG ( eg Loschbour or Ko1) ?

* " My guess is that the appearance of a mixed Levantine-Iranian basal mix in pre-Yamnaya populations is probably implicated in the late spread of agropastoral genes and memes'

Agree. Its clear enough from archaeology, if one want to face facts. The confounding issue on the steppe has been the replacement / admixture from Karelian/ Samaran -type foragers to Yamnaya agro-pastoralists was that the incoming population was less different to the WHG vs ANF scenario in central Europe. Although reserved, your comments speak volumes about linguistic implications also.

I think the only place we will see a 'native' conversion from HG to pastoralism is if / when we sample the Botai-Terek culture.

ryukendo kendow said...
This comment has been removed by the author.
postneo said...

"Then why is Iran_ChL shifted northwest compared to the earlier Iran_N from the same part of the country in western Iran?"

Yes it can easily pick that up from anatolia, not very far.

Davidski said...


Looking at rare alleles in the 1.2 million dataset that I have might be useful, but I'd need to come up with a sound methodology that takes into account false positives in the ancient DNA. If anyone has any ideas I'm all ears.

Davidski said...


Simmer down a little. Have a bottle of something that you like. You know, calm the nerves, stop the shaking.

ryukendo kendow said...
This comment has been removed by the author.
Davidski said...

Yeah, there has to be due to postmortem damage, even in the UDG treated genomes. This is why the ancient samples are often attracted to each other as well in formal stats.

capra internetensis said...


Here I was all ready to talk about Austroasiatic lol.

I was reading up on the archaeology of Ethiopia recently, and it seems like for the last fifty years people have been confidently expecting that those really ancient Neolithic sites are going to turn up in the Horn of Africa any time now... but livestock herders stubbornly refuse to show until the mid-3rd millennium BC or so, around the same time they spread to Kenya. There was contact with Arabia at this time, but more with Sudan. (This doesn't rule out people were already cultivating ensete in the highlands or whatnot.) Which is pretty freaking late, and to my mind points more to the Nile Valley while still leaving Arabia as a possibility, assuming the languages spread with pastoralism.

Egyptians obviously have their own big watery way of getting Sub-Saharan Y DNA but you have Arabians sharing 3000 year old E-V43 haplogroups with Luhya and Esan on Y-Full tree, and a fair bit of it too. That really doesn't work as Afroasiatic timing-wise.

(Though I wonder how much of it is really slaves. Eunuchs don't reproduce so well, whereas West African Muslim pilgrims with so much gold that they can crash the economy might do a pretty well with the ladies.)

ryukendo kendow said...

@ Capra
Holy shit lol

Yeah I was referring to Afroasiatic haha.

That leaves Sinai down as the better possibility.

Matt said...

Ryu: My guess is that, as small groups of humans passed through the Neolithic 'cultural bottleneck', the deformation of both genes and memes was so strong, and the resulting society so difficult for H-G societies to replicate, that all instances of highly developed agropastoral lifestyle involved at least some vertical genetic and memetic transmission, and accompanying demographic change in the ancestry stream at some point.

Or the early first Neolithic adopters simply demographically expanded a lot, very quickly, and there was no particular genetic advantage in either direction.

The existence of a Basal-EHG admixed population deep into Iran also makes a very old ANE presence in Central Asia somewhat more likely, especially if the Basal ancestry peters out as one moves deeper into the North and East. My guess is that the IVC will be Iran_N+Onge and BMAC will be Iran_N+ANE/EHG, reflecting admixture from the prior HG populations in each location.

Yes, I think this is quite clear as likely. Although only Onge in IVC (likely more Onge+some degree of Han) and only ANE/EHG in BMAC, I am not sure about (as opposed to relatively much more of one in the one place, and the other in the other). Modeling South Asia will make more sense once populations with these characteristics are sampled. Populations on the ANI-ASI cline previously modeled with little ANI will make more sense, in particular.

Ryu: Now that EEF is seen to be a mix of Iran_N and Levant_N, there is probably slight variation in the proportions depending on location, with Cardial farmers more Levantine and Danubian more Iran_N, I would bet.

I think this model of Anatolia Neolithic as Iran_N + Levant_N, at the proportions they detect, is dubious on PCA and ADMIXTURE though. The outgroups may not have had the power to model Anatolia_Neolithic correctly or distinguish on qpWave.

Further Anatolia_Neolithic shares separate drift with Europe_EN "We can rule out the hypothesis that European farmers stem directly from a population related to the ancient farmers of the southern Levant, however, since they share more allele with Anatolian Neolithic farmers than with Levantine farmers as attested by the positive statistic f4(Europe_EN, Chimp; Anatolia_N, Levant_N) (Z=15).". I don't think this will all be explained by Iran_Neolithic or additional WHG in Anatolia_N.

Ryu: Lastly, the 'Southwest Asian' ADMIXTURE component, unlike the Gedrosia, West Asian, Sardinian and Atlantic-Baltic components, does not seem to have an ancestral population behind it, being, like the 'Yamnaya' component, a product of a late, drifted combination of the 'big four' plus a bit of African, just from looking at which components it eats up.

Groups from the Arabian Peninsula get lower FST from Natufian than any populations other than Ancient Levant. Probably a reason for that. Although if you're counting Natufian as one of the "big four" (not Levant_N) perhaps your statement still clicks.

Rob: You definitely meant "further **west**"? How can something near / in Anatolia be further west than the archetypal WHG ( eg Loschbour or Ko1)?

Less EHG ancestry than Switzerland_HG, which is modelled as having less than Loschbour. (Or conversely, less close to the WHG who are ancestral to EHG, I guess?). I don't think this will actually prove to be necessary though, and that this will all depend on what they find when they rerun and republish with the dataset from Fu et al.

ryukendo kendow said...
This comment has been removed by the author.
Colin Welling said...


"that all instances of highly developed agropastoral lifestyle involved at least some vertical genetic and memetic transmission, and accompanying demographic change in the ancestry stream at some point."

This certainly happened but it went both ways. HG, Neolithic, and Steppe peoples in Europe barely mixed in ANY direction. Neolithic people from Southeast Europe traveled thousands over the span of many generations and still picked up very little HG genes. The same is true for hunter gatherers who lived right next to farmers for 1000 years but still barely had any neolithic like genes. Steppe people made there way across the whole of europe to finally settle in ireland having only lost about 2/3 of there original genetic composition. All of these cultures consisted of vertical propegations. I don't see how you can use genetics to support a special situation for farmers.

"Neolithics have an order of magnitude increase in the number of worked artifacts, highly developed notions of exclusive ownership and private property, tolerance of inequality, strong tendency to conceive of , strong tendency to value obedience in children and to beat them, and complex inheritance systems

The concept isn't property. Almost always there are responsibilities and expectations placed on the husband. Usually the husband has to make sure the wife is taken care of and he has to face the public and legal consequences of her missteps or the families failures. Having that kind of liability is not compatible with property.

Again, it wasn't a world where a rich and powerful guy got his pick, otherwise we would have seen more genetic mixing across cultures e.g. a neolithic guys mixing with hunter gatherer women.

ryukendo kendow said...
This comment has been removed by the author.
ryukendo kendow said...
This comment has been removed by the author.
Karl_K said...


"It might well be, so if it's R2 then it definitely got to Iran the same way as R2."

I strongly disagree with this statement.

Colin Welling said...


"which in fact is still the case to a certain extent a few decades ago, not that women are property and not any old type of 'caveman' scenario. "

I know what you were saying. But the responsibilities husbands had towards wives were embedded in the culture and the law, so the very concept was not property.

For decades men have been forced to financially support a woman who leaves him while she retains zero responsibilities towards the guy. This is a staunch remnant of the idea that a husband is supposed to support his wife financially, a tradition so strong that it even holds after a cultural revolution whereby wives are allowed to leave the husbands at any whim and thus relinquish any responsibilities she has towards the marriage. Much earlier wives have been allowed to leave their husbands if the husband was unfit to provide for her, with no mention of the well being of the husband. Husbands have long been a legal body shield when the wife commits a crime. Another example of the husbands prescribed "duty" to his wife is given by the fact that a wife could take her husband to court for not performing in bed and he would actually be judged, in action, by a group of women. If he wasnt getting it up, the wife had the legal right to leave him. These are just some of the responsibilities that society has imposed on husbands which show that the wife was not seen as his property, and the ownership is actually turned around.

Colin Welling said...

"This is not really true, the HG lifestyle is recovered quite easily, among Native Americans in N America for example, whose pop density was destroyed by diseases long preceding the physical arrival of Europeans, and also by Amazonians. The opposite transition is much more vertical."

I agree that it is easier to go back to hunting and gathering. All im saying is that the genetic data is showing us that, in practice, people tended to stay on one path. Thats why the hunter gatherers in europe where WHG dominant and the farmers in europe were much like the original farmers of southeast europe. Over enough time, mixing did slowly take place.

Karl_K said...


"All im saying is that the genetic data is showing us that, in practice, people tended to stay on one path. Thats why the hunter gatherers in europe where WHG dominant and the farmers in europe were much like the original farmers of southeast europe. Over enough time, mixing did slowly take place."

This could just as easily be the result of genetic predisposition.

If Basal Eurasians already had a large mix of alleles that made them more likely to be content with staying put in one location, that would explain a lot.

This tendency could be very detrimental to most hunter-gatherers, who would be better off with "wanderlust genes".

I am not convinced that this is a purely cultural difference.

Roy King said...

"Anatolia_ChL is interesting. Pity it's a low coverage sample with few markers.

I wonder if it's similar to some of the Cucuteni-Tripolye groups, and perhaps a decent fit for a part of the southern admixture in Yamnaya?"
Yes, Anatolian ChL is quite interesting. It is spatially and almost temporally similar to Kumtepe4 from the previous recent aDNA study. So, yes, both Anatolian ChL and Kumtepe4 might be related to Yamnaya.

MfA said...

Files are up

Matt said...

Ryu: Disagree, I expect that, once the preceding middle Eastern HGs are compared against the neolithic successors, extremely strong signs of recent selection will emerge.

It'll be interesting if it does (and I can imagine it happening, rather than it being insane or anything), actually I just don't see anything in these findings that make that more compelling than we'd have found it last week, I guess.

Even if you do find that signal though, how do you test against the hypothesis or how much was due to a pre-selected agricultural advantage and how much was demographic inertia? Seems you can only really say that a selective signal is found, if it's found.

Are you suggesting Natufian HGs continued to exist in the Arabian peninsula and Natufian ancestry on top of Levant_N accounts for some ancestry of SW Asians?

More or less my thinking. Probably not very much though. Or I suppose it's possible they could just be descended from Levantine farmers plus African ancestry. If populations were completely Levant Neolithic, then I might've expected them to match better the ancestry in East African populations to a greater degree than Anatolia_Neolithic / Sardinian does? Or maybe not.

Davidski said...


Thanks, downloading now.

Samaritan DNA said...

David, can you make sure to include the three unadmixed Samaritans (168723, 149532, and the Levite 149533), the Roman gladiator outlier 3DRIF-26 as well as Behar's one unadmixed Samaritan OSM537062, including on your "K10" PCA spreadsheet?

Lazaridis (2016)Fig. 1b aDNA PC Plot

Samaritan and 3DRIF-26 on the West Eurasian aDNA PC Plot
Closeup of the Samaritans and 3DRIF-26 on the West Eurasian PC Plot

Will 3DRIF-26 and the unadmixed Samaritans plot exactly with the Natufians?

Davidski said...

OK, I should have a plot by tomorrow.

Open Genomes said...

@Davidski, please create a 3-D datasheet for the worldwide samples so we can create a 3-D PC plot. This is going to be very important for determining whether the Natufians are drifted toward Sub-Saharan Africans on "dimension 1". Based on their position on the 2-D PC plot near the Bedouin A, this appears to be the case, because the Bedouin A are drawn somewhat toward Mota on dimension 1, even though they are virtually "100% Southwest Asian" (or "Red Sea", or "Arabian").

nee4speed111 said...

Could you also include copts in your plot, as they should also be plotting right next to the natufians based on the PCA in the study

ryukendo kendow said...
This comment has been removed by the author.
nee4speed111 said...

ryukendo, if you look at the PCA they provide, BedouinB who have zero or trace amounts of African ancestry plot further south of the Natufians because they have no CHG or Iranian-like ancestry to speak of. I imagine the Natufians will be similar in that regard, with the lack of CHG in them plotting them further south with Modern middle Easteners despite the fact that they have no SSA.

ryukendo kendow said...
This comment has been removed by the author.
nee4speed111 said...

Well I said BedouinB have zero or trace amounts of African ancestry, there are two sampled populations of Bedouins, BedouinA and BedouinB which despite the names are quite distinct genetically. BedouinA has typical amounts of CHG as other levantine populations, while BedouinB is almost 100% South_West Asian with CHG only existing in trace amounts in them

ryukendo kendow said...
This comment has been removed by the author.
ryukendo kendow said...
This comment has been removed by the author.
nee4speed111 said...

Yeah fair enough, I'm not an expert on Bedouin genetics but I was under the assumption that BedouinB were basically Southwest_Asians with little other admixture. I got this assumption from Gedmatch calculators which put BedouinB as having little to no SSA ancestry.

Seinundzeit said...

As we wait/look forward to David's analyses, Kurd at Anthrogenica produced an IBS list for Iran_N. If we remove ancient populations/samples, and if we remove the 23andMe raw-data files he ran (to make things more reliable), this is the list:

Makrani 65.732%
Brahui 65.714%
Iranian 65.697%
Balochi 65.696%
Abkhasian 65.596%
Georgian 65.568%
Iranian_Bandari 65.561%
Assyrian 65.529%
Kalash 65.521%
Pathan 65.479%
GujaratiA 65.447%
Sindhi 65.445%
Adygei 65.427%
Chechen 65.423%
Lebanese_Christian 65.392%
Lebanese_Muslim 65.338%
Pashtun_Afghan 65.295%
Azeri 65.289%
Burusho 65.274%
Cypriot 65.271%
GujaratiB 65.270%
GujaratiC 65.217%
GujaratiD 65.193%
Lebanese 65.156%
Saudi 65.155%
Punjabi 65.119%
Greek 65.092%
Jordanian 65.085%
Bulgarian 65.064%
BedouinB 65.024%
Yemeni 64.838%
Egyptian 64.812%
Ju_hoan_North 62.416%

IBS is a pretty coarse method, very rough/simple. And despite the simplicity, the output is often hard to properly contextualize.

But in this case, things make sense.

In line with the paper, Pakistani Makrani are the closest population to Iran_N, and the other two Balochistani populations also top the list.

Looking at this, it's pretty obvious that BMAC and IVC will turn out to be Iran_N-related, if not almost identical to Iran_N. Perhaps they'll have some extraneous genetic influences on top of an Iran_N base (like EHG/ANE, or South Asian hunter gatherer ancestry, whatever that may turn out to be), although I'm sure those won't be huge at all. Just like how despite the passage of so much time, and the treading of so much geography, Middle Neolithic/Chalcolithic Europeans were still only around 20% WHG-admixed, and still around 80% Neolithic Anatolian.

vvv666 said...

Hugely interested in pigmentation related SNPs for the Iranian neolithic samples. There is a vibrant debate on whether it is historically accurate to depict Indus Valley inhabitants with light skin in a commercial fantasy movie:

Jijnasu said...

The physical appearance of the residents of the IVC is an interesting question.
Ruchika sharma is an idiot though. Look how she throws around terms like proto-australoid. She probably bases her ideas on colonial ideas of 'bull liped' 'noseless' and 'black' enemies of the Aryans based on mistranslations of the veda no longer accepted by mainstream scholars

Alberto said...


Thanks for all the insightful comments. I basically agree with you in all, especially about ANE. So just minor comments:

- About Basal Eurasian, one problem I see with the estimates placing Iran_N as quite higher than Levant_N is that it pushes Basal Eurasian quite to the "East". So it becomes quite difficult to find a population that could be the "non-Basal" part of Levant_N. It would have to be a very divergent population, very drifted to the West (much more than WHG), that we'd still need to find. Not impossible, of course, but it makes things more complicated. While for modelling Iran_N as Basal Eurasia + Something else, we do have ANE as a divergent and very eastern branch already. Plus there are several discrepancies (relatedness to Ust-Ishim, strange Anatolia_ChL levels of BEA,...) A complicated matter, but at least it won't affect our models now that we have these basal eastern and western samples.

- About the IVC, I think we should all keep in mind that this is not an Early Neolithic phenomenon. Its mature phase is dated around 2600-1900 BCE. It already saw a huge demographic explosion. It had for long been part of a trade network with Central and West Asia. We can only guess as to how they looked genetically, so I won't argue strongly one way or another. But I'd expect that all the Chalcolithic movements we see in West Asia, Europe, the steppe,... also hit South Asia. I guess that soon enough we'll know.

ryukendo kendow said...
This comment has been removed by the author.
Alberto said...

Because I think that if Basal Eurasian is significantly closer to Iran_N than to Levant_N, then Levant_N (not to mention Natufian) will off the cline between Basal Eurasian and WHG. So this UHG will have to be quite divergent. Or do you see it possible to fit this UHG within West Eurasian variation?

ryukendo kendow said...
This comment has been removed by the author.
Seinundzeit said...


This is an IBS list for Iran_Chalcolithic:

Assyrian 65.946%
Georgian 65.937%
Abkhasian 65.887%
Lebanese_Christian 65.859%
Iranian 65.844%
Chechen 65.830%
Cypriot 65.829%
Adygei 65.772%
Azeri 65.734%
Makrani 65.676%
Lebanese_Muslim 65.654%
Brahui 65.654%
Lebanese 65.644%
Balochi 65.641%
Greek 65.630%
Iranian_Bandari 65.620%
Saudi 65.566%
Kalash 65.560%
Jordanian 65.516%
BedouinB 65.499%
Bulgarian 65.488%
Pathan 65.488%
Pashtun_Afghan 65.437%
Sindhi 65.434%
GujaratiA 65.388%
Burusho 65.343%
GujaratiB 65.303%
GujaratiC 65.221%
GujaratiD 65.169%
Egyptian 65.154%
Yemeni 65.098%
Punjabi 65.092%
Ju_hoan_North 61.902%

Iran_N almost behaved like a modern South Central Asian, it is very strongly shifted towards South Asia (especially Pakistan, seems to have a somewhat weaker affinity to both Afghanistan and India, but still very strongly shifted toward Afghan/Indian populations as well).

Not the case with Iran_Chalcolithic, the South Asian affinity is more diluted, and Caucasus/West Asian populations top the list.

So I think IVC won't be very similar to any Chalcolithic West Asian populations, but will definitely be very similar/rather closely related to Iran_N.

Interesting side note, the individual who ran these is also doing ADMIXTURE runs. He posted this:

"At K12, a separate Iran N cluster has not formed yet. The Iran N and Iran LN are part of the Indian cluster. The only Natufian sample that passed the genotype rate is clustering with BedouinB. More later this evening..."

The fact that Iran_N seems very "Indian" in ADMIXTURE says a lot I suppose.

I can't wait to see what David's analyses tell us about Iran_N proportions in Central/South Asia.

Open Genomes said...

@Nick Patterson (Broad):
In the paper you coauthored, Haber (2013), Genome-Wide Diversity in the Levant Reveals Recent Structuring by Culture we have two components from a K=10 ADMIXTURE analysis, also (roughly) plotted on a world map:

Figure 4. Comparisons of the Levantine and Middle Eastern modal components.
A) ADMIXTURE analysis based on 10 constructed ancestral components, with only the Levantine and Middle Eastern components highlighted. B) Frequency of the Middle Eastern component in world populations. C) Frequency of the Levantine component in world populations. Intensity of the colors reflects the frequency of a component in the plotted populations.

The so-called "Levantine" component may be a mix of "Anatolian Neolithic" and "CHG / Iranian Mesolithic-Neolithic". However, the "Middle Eastern" component closely corresponds to the "Southwest Asian" component found by Davidski in his CHG K10. This has been called "Red Sea" and even "Arabian" at K=36, and peaks in the Bedouin B.

The study says:
" Our estimates show that the Levantine and the Arabian Peninsula/East African components diverged ∼23,700-15,500 y.a., while the Levantine and European components diverged ∼15,900-9,100 y.a."

Clearly, the "divergence" of the "European" component in the study and the "Levantine" component is the result of admixture between WHG and the "Anatolian Neolithic" (along with later "CHG" admixture in Europe, from various sources.

However, the key is the "divergence" of the "Levantine" and "Arabian Peninsula/East African" (aka "Middle Eastern" component) reflecting the beginning of admixture between this "Southwest Asian" component and the other Near Eastern ancestral components "Anatolian Neolithic" and "CHG / Iranian".

We can assume that this in fact is the "Natufian" ancestral population that we see in your recent study.

Did the admixture go *into* Africa 23,000-14,700 years ago (the start of the LGM to the Bolling Interstadial, i.e. the Kebaran Period) or did it arrive in the Near East *from* Africa?

The Y-DNA can provide us an answer. The Natufians were all E-Z830.
Here are the tMRCAs of E-Z830 and the distribution of the various subclades according to
E-Z830 YFull tree

The tMRCA of E-Z830 is: formed 23,900 ybp, TMRCA 19,200 ybp
This is just about right for the LGM, c. 23,000-18,000 ybp.
E-Z830 splits at 19,300 ybp into two subclades, E-V1515, and E-PF1962. E-PF1962, which has a tMRCA of 18,900 ybp, has a completely Near Eastern / Eurasian distribuion, with no hardly any representatives in Africa (except a few very derived examples along the North African coast).

On the other hand, E-V1515. which has a tMRCA of 12,500 ybp, is too late to have a Natufian expansion, although 12,500 ybp would be just right for the PPNA.
E-V1515 splits into E-V1700, which in YFull has an Arabian distribution (the Portuguese is in a recent Saudi subclade) and a tMRCA of 10,900 ybp, again in the PPNA.
We know however that E-V1700 is common in Eritrea and northeast Ethiopia.

E-V1486 however, with a tMRCA of 5,600 ybp (the Chalcolithic / Early Bronze Age) is *very* widespread in East Africa, and is found as far south as Namibia. It has a distribution that corresponds to East African (Luhuya / LWK) and Khoisan-speaking Nama cattle herding:
Map of E-M293 disitribution

Open Genomes said...

@Nick Patterson (Broad) (continued):
The question is, is E-V1700 under E-V1515 a later arrival from East Africa to the Levant, or is a "back-migration" from the Levant?

A map from Trombetta (2015), Phylogeographic Refinement and Large Scale Genotyping of Human Y Chromosome Haplogroup E Provide New Insights into the Dispersal of Early Pastoralists in the African Continent can give us the answer:

Fig. 3 Maps of the observed frequencies for haplogroup E-V1515 and its major subhaplogroups.
(A) Haplogroup E-V1515,
(B) haplogroup E-V1700,
(C) haplogroup E-V1486,
(D) haplogroup E-V2881,
(E) haplogroup E-M293,
(F) simplified phylogeny of E-V1515, showing the estimated age of the nodes (ky).
Place of sampling of chromosomes carrying rare paragroups E-V1515*, E-V1785*, and E-V1486* is shown as blue asterisks in panels (A), (B), and (C), respectively.

It would seem E-V1515, the the sister clade of very likely Natufian E-M123* (already discovered in aDNA from Bronze Age Armenia) is African in origin and *not* a back-migration from the Near East.

The source of E-V1515 would appear to be somewhere along the Red Sea coast of Eritrea and Eastern Sudan.

Interestingly, this is the region of the Beja (known as the Wedjet in Ancient Egyptian), who speak a language that is either a very divergent form of Cushitic, or an independent basal branch of Proto-Afro-Asiatic.
Wiki - Beja language classification (with sources)

Perhaps this "Natufian" component not only represent proto-Afro-Asiatic, but also it's clearly of *East African* origin and not a back-migration from the Levant?
Is this area of the Sudanese Red Sea coast the proto-Afro-Asiatic "Urheimat"?

We also know the Natufians had craniofacial features that show they may be a mix of East Africans and Eurasians:
Natufians - Craniofacial research - Wiki (with a link to the original study)

Therefore, E-M123 would represent the initial settlement of East African proto-Afro-Asiatic speakers in the Levant:
YFull E-M123 tree
The tMRCA of E-M123 is 18,900 ybp, and it appears that E-M123 did undergo a rapid expansion after the Bolling Interstadial at 14,700 ybp.
In particular, E-Z849 with a tMRCA of 14,100 ybp has a basal sample from Iran.

Were the "Basal Eurasians", who could have mixed with local Levantine Hunter-Gatherers during the Geometeric Kebaran Culture to form the Natufians in fact the "Basal Afro-Asiatics"?

The question comes about about the symmetric relationship of "Basal Eurasian" to Mota (Y-DNA E1b1a2-M329 vs. Natufian E1b1b1b2a-M123), who lacks any Eurasian admixture. Mota plots most closely with the Gummuz (ie. an Ethiopian name for the Nilo-Saharan-speaking South Sudanese farmers) and the click / Khoisan speaking Hadza.

The YFull tMRCA of the common ancestor of Mota in E1b1a2-M329 and the Natufians in E-V830 (likely E1b1b1b2a-M123) is 42,600 ybp, as old as Oase I from Romania who was part Neanderthal.

YFull E1b1-P2 tree, tMRCA 42,000 ybp

We wouldn't expect a hypothetical East African "Basal Proto-Afro-Asiatic" population to be much closer to a Nilo-Saharan or Khosian Mota than any other nearby population, except for the fact that they would lack Eurasian Neanderthal admixture.

Open Genomes said...

@Nick Patterson (Broad) (continued):

We know that there was an influx of Levantines into East Africa around 1,000 BCE. T-M184 was found among the Levantine PPNB farmers and significantly, in the LBK. (The LBK sample does in fact have about 10% "Southwest Asian" while the Starcevo-Koros samples from Hungary do not.)

T-M184 has a African distribution that extends to the Lemba of South Africa.
Haplogroup T-M184 wiki, with map of its worldwide distribution

T-M184 in Africa is associated with Cushitic-speaking intensive cultivators in East Africa, particularly the southernmost Afro-Asiatic speakers, the Iraqw (Siraqwa), of the Great Lakes region of East Africa.

The Siraqwa People were associated with the "Stone Bowl" cultural complex but they went extinct in the Middle Ages and were absorbed by various surrounding Nilotic and Bantu-speaking peoples.

The Iraqw are 11% T-M184. Cushitic T-M184 may be the source of T-M184 among the Lemba of South Africa.
Iraqw people - wiki.

T-M184 has a tMRCA of 26,900 ybp (before the LGM) with a basal sample from Armenia, and a primary rapid expansion with a trifurcation in the main T-L206 subclade at 16,000 ybp:
T-M184 YFull tree

Sample I1707/AG83_5 from Ain Ghazal in the Levantine Neolithic PPNB, is T-M184 (P322- FGC3945.2- equivalent to T-L162). It was T-L162 that was found in the LBK sample I0797/KAR16A. it seems that while T-L162 was also a carrier of this "Levantine/East African" component.
Interestingly, it's the T-Y15711 subclade of T-L162 which is found in Sudan, with a tMRCA of 4,400 ybp. but the entire branch just above it, T-Y15127, has a tMRCA of 8,700 ybp, in the PPNB.

Both other sublcades of T-L206, have tMRCAs old enough for the PPNB. T-Y11151 has a tMRCA of 15,400 ybp, and T-L131 in particular sems to have a rapid expansion around 11,100 ybp.

The locus of greatest diversity of T-L206 is in Southern Iraq and the Persian Gulf region, not the Levant. Perhaps it was T-M184 which brought a kind of "CHG / Iranian Neolithic" component to the Levantine Neolithic?

Open Genomes said...

@Nick Patterson (Broad) (continued):

Another source of much later back-migration to Africa from the Levant is R1b1c-V88. While R-V88 split from R1b1-M415* just after the LGM at 17,200 ybp, the expansion of R-V88 is 10,200 ybp, too late for the Natufians, but during the PPNA-PPNB transition.

R-V88 YFull tree tMRCA 10,200 ybp

R-V88 was found in sample I0410 from Els Trocs the Cardial Ware Culture in Neolithic Iberia. R-M18 is a basal clade of R-V88 which is almost exclusively found in Sardinia, with one exception from Lebanon. The route that R-V88 took to Spain in the Early Neolithic is debatable.

R1b-V88 is of course associated with the Afro-Asiatic Chadic languages in Africa:
Cruciani (2010) Human Y chromosome haplogroup R-V88: a paternal genetic record of early mid Holocene trans-Saharan connections and the spread of Chadic languages

able 1. Frequencies (%) of Y chromosome R1b subhaplogroups in Africa and Eurasia

The "Chadic" Cameroonian / West African clade of R-V88 is R-Y7771 (the parent clade of R-V69). The tMRCA of R-Y7771 is 5,200 ybp, but the tMRCA of its parent, RY8447, is 7,000 ybp, and just above that the tMRCA of R-Y777 is 7,600 ybp. R-V35, under R-Y7777, is found in Sardinia (along with R-M18). This appears to be about right for R-V88 joining the Early Neolithic Cardial Ware culture.

Does sample I0404 from Els Trocs have any additional "Natufian" or Levantine Neolithic admixture? Or even some additional African admixture, of any sort?

R-V88 is not only found among 4% of Yoruba (Hausa is a Chadic language, but Yoruba is not), but it's also found in Central Africa among Bantu language speaking groups. R-V88 in fact is found as far south as Namibia. There may be more than one stream of Eurasian admixture in Sub-Saharan Africa which could confuse the ADMIXTURE analysis.

There is yet other piece of evidence for an East African origin of "Basal Afro-Asiatic" ancestry:
Fernandez (2014) Ancient DNA Analysis of 8000 B.C. Near Eastern Farmers Supports an Early Neolithic Pioneer Maritime Colonization of Mainland Europe through Cyprus and the Aegean Islands found an mtDNA L3 in a PPNB sample from Tell Halula along the Middle Euphrates in North Syria 6800-6000 BCE (8800-8000 ybp).

Interestingly, E-V22, under E-V68 which is the other half of E-M35 from E-Z830, has a tMRCA of 8,600 ybp. E-V22 is quite common in the Levant and among Assyrians, and is found as far east as Eastern Iran:
E-V22 YFull tree, tMRCA 8,600 ybp

However, it may be that mtDNA L3 in the Levant predates the arrival of E-V22. mtDNA L3 is generally East African, and L3h in particular is associated with Afro-Asiatic Chadic speakers, in the same way as R-V88:

Cerny (2009) Migration of Chadic speaking pastoralists within Africa based on population structure of Chad Basin and phylogeography of mitochondrial L3f haplogroup

L3 was not found in the PPNB of the Southern Levant, but it obviously did not just skip the Southern Levant on the way to the Middle Euphrates in North Syria.

Open Genomes said...

@Nick Patterson (Broad) (continued):

Archaeologically, are the "Basal Eurasians" who are actually the "Basal Afro-Asiatics" in fact the Ramonian/Mushabian culture, which appears to enter the Southern Levant from East Africa around 16,000 BCE (18,000 ybp), at the end of the LGM, which also happens to be the tMRCA of Y-DNA E-M123 found among the Natufians of Rakefet Cave?

Radiocarbon Context Database 16,000-12,500 calBCE showing the southern Levantne Ramonian/Mushabian Culture and the Northern Levantine Geometric Kebaran Culture

As the Altai Neanderthal study demonstrated, the "admixing Neanderthal" was closest to the Mezimksaya Neanderthal from the Caucasus. It's likely that Neanderthal admixture took place around Mt. Carmel in the Levant. Neanderthals were present in the Upper Paleolithic Levant and there's no reason to think that any Levantine Upper Paleolithic population would have entirely avoided admixing with them, or with the descendants of those who admixed with them, over the course of about 30,000 years.

What all this means is that:
1. "Basal Eurasian" is very likely to be "Basal Afro-Asiatic", centered among the Beja on the Red Sea Coast of Sudan, the probable Afro-Asiatic Urheimat.
2. These "Basal Afro-Asiatics" would not be closer to Mota than Near Easterners, becuase their tMRCA with Mota would be around 42,500 ybp, about the time of the tMRCA of Y-DNA P (Y K2a2) with NO* (Y K2a1-M2308, Ust'-Ishim and Oase 1)
3. There will be confusion with "LBK-like" ancestry among *modern* African populations, becasue of several subsequent post-8.2 Kiloyear Event population exchanges.
4. The Natufians themselves may have "Geometric Kebaran" Upper Paleolithic Levantine Hunter-Gatherer admixture. This doesn't show up on the Y in the three samples from Rakefet Cave on Mount Carmel, but it may show up in the seven samples from Wadi Hammeh 27 that are being sequenced by Latrobe University.
5. The definitive proof of an African origin of the "Basal Afro-Asiatics" is the fact that this group has no Neanderthal admixture. This would make sense only if they originated in the Afro-Asiatic Urheimat of the Red Sea Coast of Sudan (or an adjacent region).

Karl_K said...

@Open Genomes

"5. The definitive proof of an African origin of the "Basal Afro-Asiatics" is the fact that this group has no Neanderthal admixture. This would make sense only if they originated in the Afro-Asiatic Urheimat of the Red Sea Coast of Sudan (or an adjacent region)."

You may be correct, but the lack of Neanderthal admixture is not in any way proof of location.

As long as the Out-Of-Africa population had split before the first admixture with Neanderthals occurred, then the Basal Eurasians could have been genetically isolated, but within Eurasia.

It may complicate your haplogroup story, but is just as reasonable at this point.

The "Basal Afro-Asiatics" could have been a back migration from Eurasia of a Basal Eurasian population that lacked Neanderthal admixture.

Alberto said...


Thanks for sharing that! Yes, it's interesting that Iran_N has stronger affinity to South Asians, though I think quite expected. We have to see better what Iran_N really is (does it have any ASI? Is it Basal Eurasian + EHG?), but its connection to South Asia is undeniable.

The subject of the IVC is complicated to argue, because we're pretty much in the dark regarding ancient DNA. I just thought it was good to remember that it's not Early Neolithic, because that's what it seems to me that people are figuring out. It's a Bronze Age culture. 4000 years later than the Early neolithic in South Asia. And with strong ties to Central and West Asia by then.

The case of Balochi and Brahui is interesting. From people arguing that both were recent migrants to the region (since both speak languages related to far away areas), to know people arguing that they are relic populations from the Neolithic and Bronze Age in South Asia. Maybe they are, but maybe their inflated Iran_N affinity is because they really migrated from NW Iran "recently"? (I mean Balochi in this case, obviously).

Regarding Chalcolithic movements, well Iran_Chalcolithic doesn't do bad in the IBS with S-C Asia either. Though I was thinking more of Central Asia, where they could be ANE shifted confronted to Iran_ChL. We'll see how Armenia_ChL does for S-C Asia.

Mc Michael said...

A proper interpretation of the data, is that Neolithic Iranians (or possibly only those restricted to Ganj Derah), only AS EARLY AS 10kybp, did not contribute European gene pool. This does not rule out the likely possibility that later Iranian Neoliths, and Iranian populations onward, did not. Just based on the significantly more eastern position of Modern European samples (relative to older samples), in the figure, it's pretty clear, in fact, that the total Iranian contribution to the European gene pool is greater. More-so, any conclusion based on the geno-typing of a single sample, should not deserve much confidence. Just another vile attempt from bankrupt steppe theorists, to "Westernize" the origin of the modern European gene pool.

Davidski said...

Wishful thinking on your part.

There are a number of ancient Iranian samples available now, and it's clear that ancient populations from Iran did not contribute to the modern European gene pool.

The reason modern Europeans are shifted east relative to some ancient Europeans, is because of the expansion of Bronze Age steppe pastoralists deep into Europe from the Eastern European steppes.

And obviously, if modern Europeans have a lot of steppe ancestry, but lack ancestry from Iran, then the ancient steppe people didn't have any ancestry from Iran.

jay parisi said...

Very interesting. I agree the plot seemed odd to me, as I thought GD13a as well as other Iranian Neoliths would be farther away from modern South Asians. Would you guess that GD13a would be more distinct from modern South Asians on your PCA, then the PCA generated in the study? Thanks.

Davidski said...

GD13a is the Iran_Neolithic individual sitting furthest to the right on the plot here.

The Iran_Neolithic cluster is actually pretty close to South Central Asians (SC_Asia) on my plot as well.

Neolithic Iranians will always cluster close to modern-day individuals from the Indus Valley region, because they're very closely related to them.

Mc Michael said...

Actually on that plot you posted, (, Neolithic Iranians along with GD13a look about as far from modern S. Asians, as it is to other Near Easterners and Caucaus groups (and remember True South Asian is defined by an indigenous ASI component, of which we have no example, and which purer examples among modern populations of S. Asia are less frequent). And just look at the East shifted position of Anatolian Chaloliths, compared to the Anatolian Neoliths (whom themselves were largely Iranic, as admixture analysis reveals). It's pretty obvious that their intermediate position is the result of admixture with Iranian Neolithics who were moving Westward (and vice versa with respect to Iranian Chaloliths). Without samples from Eastern Turkey, it's easy to polarize Iran. But as far Iranian genetic contribution to Europe, I think it's glaring. The Steppe and the Caucauses had a significant influence from the Iranian plateau, not only as supported by PCAs, but HG data as well.

Davidski said...

Neolithic Anatolians don't have any admixture from Neolithic Iran. What you're seeing there is an abstract model of reality that just happens to fit the data.

There weren't any migrations from Neolithic Iran to the west or north, so Europeans and West Asians don't have any Neolithic Iranian ancestry. This is a fact.

What they do have is Caucasus Hunter-Gatherer ancestry, which entered Anatolia and the steppe at different times, and shifted the genetic structure of Anatolia to the east. That's why Anatolia Chalcolithic is positioned east of Anatolia Neolithic.

All of this will become clearer in future papers with more data and better models.

Mc Michael said...

"Neolithic Anatolians don't have any admixture from Neolithic Iran. What you're seeing there is an abstract model of reality that just happens to fit the data."

It's the simplest possible scenario, and it goes well with data. However, I myself don't agree with estimates as high as 40%, as was suggested by Lazaridis, but the estimate of ~15% Iran_N + WHG + Levant_N fits snugly. But keep in mind, that what's been labeled as "Neolithic Anatolian", is based on very limited sampling at the far west of modern Turkey (in other words the predominately WHG/Levantine end).

"There weren't any migrations from Neolithic Iran to the west or north, so Europeans and West Asians don't have any Neolithic Iranian ancestry. This is a fact."

That would lead to a bit of mystery why a population would completely confine their movements to the east. Second, there is a range of early haplogroups distributed through Iran (J2, R1b/R1a, IJ linker, perhaps even originated in Iran), and the presence of their derivative types, in modern populations, to the west, are in-line with a model of westward diffusion - through the Neolithic period. Third, PCAs of ancient populations are very much in agreement with a westward influence; again, the eastward position of Anatolian_Chl and the position of modern Europeans (compared to WHG), is best explained by Iran_N influence - not with a fabricated an Anatolian Neolithic ancestral group. And later Eurogenesis at the Iranicized Steppe (as well as EHG), pulled European populations further Eastward.

"What they do have is Caucasus Hunter-Gatherer ancestry, which entered Anatolia and the steppe at different times, and shifted the genetic structure of Anatolia to the east. That's why Anatolia Chalcolithic is positioned east of Anatolia Neolithic."

Ok, but that seems like an awful lot footwork to avoid what can basically be explained with Iran_N movement Westward and North. CHG, itself is highly is highly N_Iranian + EHG. It cannot be associated with a unique ancestral source (not one with an significance, anyways), as Levantine, EHG, WHG, and Iran_N, or ASI. So to say that Anatolian_CHL and the Steppe is CHG, is just saying that they have some secondary Iran_N admixture. This may in fact be true, that the admixture is secondary, but it doesn't change the fact.

Jeremy Battle said...

Indeed, judging from that PCA (in the new study BELOW), the Tepecik samples from South Central Turkey, do appear to be more leaning toward the Caucauses and Iran, which is still in line with the idea that there was reciprocal gene flow between Anatolian and Iranian Neolithics. And keep in mind, there is still plenty of ancient DNA geographically East, which which has not yet been sampled. Also note the intermediate position of Bronze Age/Iron age samples, with respect to WHG and Caucaus/Iran (EHG/Iran_Neolithic), most of which are in close proximity to the position of modern-day North and Western Europeans. This is suggestive of the possibility that Bronze Age/Steppe populations emerged as the dominant type in Europe.

Davidski said...

That PCA is garbage because it suffers from projection bias. It's a common problem in ancient DNA studies.

And Central Anatolian farmers didn't have Neolithic Iranian ancestry, they had Caucasus Hunter-Gatherer-like ancestry, which was probably native to eastern Anatolia.

Not sure how else to communicate this, but at some point you people will have to grasp the data and accept that Neolithic Iranians only migrated to South Asia, not to Anatolia or the European steppe.