search this blog

Friday, June 28, 2019

On the origin of the Gravettians (Bennett et al. 2019 preprint)


Over at bioRxiv at this LINK. No major surprises, as far as I can see. From the preprint, emphasis is mine:

The Gravettian technocomplex was present in Europe from more than 30,000 years ago until the Last Glacial Maximum, but the source of this industry and the people who manufactured it remain unsettled. We use genome-wide analysis of a ~36,000-year-old Eastern European individual (BuranKaya3A) from Buran-Kaya III in Crimea, the earliest documented occurrence of the Gravettian, to investigate relationships between population structures of Upper Palaeolithic Europe and the origin and spread of the culture. We show BuranKaya3A to be genetically close to both contemporary occupants of the Eastern European plain and the producers of the classical Gravettian of Central Europe 6,000 years later. These results support an Eastern European origin of an Early Gravettian industry practiced by members of a distinct population, who contributed ancestry to individuals from much later Gravettian sites to the west.

...

The mitochondrial haplogroup of BuranKaya3A was determined to belong to an early branch of the N lineage, N1.

...

In addition, the N1 of BuranKaya3A carries three of the eight mutations occurring prior to N1b, a rare haplogroup most highly concentrated in the Near East, yet appearing broadly from western Eurasia to Africa. The descendants of the N1b node include N1b2, currently found only in Somalia [22], and N1b1b, found in nearly 10% of Ashkenazi Jewish haplogroups [23]. These three mutations allow us to place BuranKaya3A on a lineage apart from that which has been proposed to later enter Europe from Anatolia during the Neolithic (N1a1a) [24]. Among ancient samples, the mitochondrial sequence of an 11,000-year-old Epipalaeolithic Natufian from the Levant (“Natufian9”) [25] is also a later derivative of this N1b branch.

...

From the reads mapping to the Y chromosome, six out of six Single Nucleotide Polymorphisms (SNPs) that overlap with diagnostic sites for Y-haplogroup BT all carry the derived allele, allowing a minimum assignment to BT, which has origins in Africa, with additional derived alleles suggesting an eventual placement of CT or C, found in Asia and the Epipalaeolithic Near East [25]. Additional ancestral alleles make an assignment of C1a2 or C1b, which appear in UP Europe [1], unlikely (see Table S3 for a summary and comparative placement of Palaeolithic Y-haplogroups, and Supplementary Data 1 for a complete list of Y diagnostic SNPs).


Bennett et al., The origin of the Gravettians: genomic evidence from a 36,000-year-old Eastern European, bioRxiv, posted June 28, 2019, doi: https://doi.org/10.1101/685404

96 comments:

Aram said...

It seems that Y dna I found in Gravettian context entered Europe via Eastern Europe.

Dragos said...

Although this guy was Hg -C

Aram said...

Imho Basal Eurasians were already present in North Near East since 40.000 ybp. So they were blocking any important migration via Anatolia.
Or alternatively low sea levels were making harder mass migrations via Bosphorus.
The cases of IJ* reported by Grugni in Iran should be investigated more closely.

Aram said...

I was meaning higher sea levels. Offcourse.

Knowledgeable Geneticist said...

So proto-Gravettians and Gravettians had BT, C and I, three very distantly related haplos. It makes sense if Basals carried BT, but the Crimean sample had no Basal admixture, so then what Y haplo was associated with Basals? Since they contributed so much to Dzudzuana, surely they left a trace, at least mtdna. Perhaps Basals themselves were genetically diverse.

Dragos said...

BT relates to just poor coverage

Slumbery said...

@Knowledgeable Geneticist

As Dragos already said this guy was not BT*, his YDNA is just not resolved further. The article implies that he was probably C, but the data is insufficient to declare that with confidence.

I'd say basals were diverse with E and IJ. (The J in CHG and ancient Iran are definitely not from ANE and for now I would cross out the Villabruna related side too.)

Kristiina said...

@ Knowledgeable Geneticist "so then what Y haplo was associated with Basals?"

Maybe yDNA E and G.

Basal-rich ancients in North Africa carry mtDNAs such as M1 and U6. In Natufians there is N1b. All of these (M, U and N) have been found in non-Basal populations.

A team of geneticists should now look for a 35 kya old genome from the Arabian Peninsula or northeast Africa to shed light on this enigma of Basal.

zardos said...

Basal will be E I guess and it will turn out that a Basal related group colonised both Africa and Western Eurasia with the Subsaharans before that being much more archaic with yet unknown old substrate.
There is only North East Africa or Arabia as a possible place of origin.
Probably even a connected population continuum of Basal strains from Egypt-Yemen-Palestine?

zardos said...

The Basal related group moving South might have just split of earlier and been diverged even before that, but Basal related nevertheless. That's the only scenario which seems to make sense right now?

epoch said...

First, Neanderthal presence in the Crimea right up to the Campanian Ignimbrite eruption. Wow.

Second, a basal N1 on its way to N1b. That seems similar to the basal U6 in Muierii from the Carpathian Basin.

Third thing I noticed is the fact that they use |Z|=2 as limit of statistical relevance rather than 3. Why is this done? Doesn't that mean the conclusions are far more uncertain than those from Fu et al?

epoch said...

They also are interested to show a link with the Levantine Ahmerian culture (46,000-42,000 BP) as opposed to a local development. But have a look at figure 8 showing the outcome of f3(BuranKaya3A, x; Mbuti) with low resolution samples included. We'll find that Muierii2 has the highest degree of shared allels. There actually was a culture also considered connected to the Ahmerian in the neighbourhood: The Bohunician. And it precedes the Ahmerian.

@David

Could you combine Muierii2 and Ciaclovina to improve the number of SNPs?

Shaikorth said...

The SNP counts in f3 comparisons are extremely low so take them with a grain of salt. f3 BuranKaya3A Muierii Mbuti has less than 500 SNP's.

Even higher coverage ancients have less than 10k SNP's used, but based on them Sunghir is closest.

epoch said...

@Shaikorth

Yeah, I understand. That's why I would like to see what a combined Carpathian sample would show.

epoch said...

@Aram

"Imho Basal Eurasians were already present in North Near East since 40.000 ybp."

Yet according to Lazaridis they had no neanderthal admixture and Neanderthals lived in the Zagros mountain in that time frame.

https://en.wikipedia.org/wiki/Shanidar_Cave

zardos said...

We dont know for sure when and where the hybridisation took place. Even if Neandertals were present, there is no rule for widespread admixture having to take place.

zardos said...

Even if one wave of AMH did mix, the next might just have pushed on without.

JuanRivera said...

So the new sample belongs to the KS (Kosteki-Sunghir) cluster. The absence of 'Common West Eurasian' in the new sample (and the KS cluster in general) means that there were two early-splitting west eurasian populations, one being Villabruna ('Common West Eurasian') and the other being the KS cluster and ANE's west eurasian ancestry, which later began to admix (as evidenced by the Vestonice cluster). Though, the situation may be muddled by Ust'-Ishim-like and other East Eurasian admixture.

NeilB said...

According to Chris Stringer the Bohunician has yet to be firmly assigned to either Neanderthals or AMH as both were in the region at the time. Isn't this still the case?

NeilB said...

@JaunRivera, do you know what (if any) relationship this individual has with the remains from Salkhit? I know that site is a nearly 6000km further east and a couple of 1000 years younger, but do you know it was m/f and if the former what was the yDNA?

Davidski said...

@All

The following samples have been added to the Global25 datasheets.

RUS_Devils_Gate_Cave_N:NEO235
RUS_Devils_Gate_Cave_N:NEO236
RUS_Devils_Gate_Cave_N:NEO237
RUS_Devils_Gate_Cave_N:NEO238
RUS_Devils_Gate_Cave_N:NEO240
RUS_Kolyma_Meso:Kolyma_River
RUS_Magadan_BA:M0831
RUS_Magadan_BA:M9984
RUS_Yana_MA:Yana_Young
RUS_Yana_UP:Yana1
RUS_Yana_UP:Yana2

Same links as always. Scroll down HERE.

Arza said...

Mal`ta sits on a cline between AG3 and Yana.

RUS_MA1:MA1
RUS_AfontovaGora3:AfontovaGora3 69.8%
RUS_Yana_UP:Yana1 19%
RUS_Yana_UP:Yana2 11.2%

Distance 3.29%

And it's unlikely that AG3 is an end of this cline.

Which means that "ANE" is not real ANE. Or maybe akin to ASI-AASI we should introduce AANE - Ancestral Ancient North Eurasian. AANE would be fully West Eurasian population without Yana-like admixture.

AANE went to Siberia, mixed with locals, acquired Y-DNA R and bounced back to West Eurasia bending any possible WHG-AANE cline towards East Asia.

Variable levels of this Yana-like-shift are the reason why modelling Yamnaya with EHG doesn't work well. EHG has too much of it compared to Yamnaya.


Shaikorth said...

I'm not sure EHG has too much Yana or something like that, EHG still works best as a source for Yamnaya, and probably does in qpAdm too. In G25, if Yamnaya is fitted as a mix of hunter-gatherers without EHG (preferred over other sources) it will not really prefer AG3 over WSHG either although it will like both over Yana.


"distance%=7.5336"

Yamnaya_UKR

GEO_CHG,39.8
RUS_West_Siberia_N,34.8
ROU_Iron_Gates_HG,20.6
Anatolia_Pinarbasi_HG,4.8

"distance%=7.5483"

Yamnaya_UKR

GEO_CHG,40.6
RUS_AfontovaGora3,32.4
ROU_Iron_Gates_HG,22
Anatolia_Pinarbasi_HG,5

JuanRivera said...

About the Salkhit remains, it is only known that it's an AMH and that its mtDNA is N*.

JuanRivera said...

Given its location (near where MA1 would be later present) and its age (35kya, compared to the divergence of ANE's west eurasian ancestry from the KS cluster 38 kya and the Yana individuals 31kya), I would hypothesize that Salkhit is another proto-ANE individual similar to Yana, though with maybe some more East Asian.

JuanRivera said...

About MA1 being modeled as AG3+Yana, are there models where both MA1 and AG3 are Yana+some member of the KS cluster? May Pinarbasi_HG or Villabruna be used as "CHG" input to see if the Yana paper is correct about MA1 and AG3 having "CHG" admixture?

JuanRivera said...

And does Yana have Ust'-Ishim admixture outside of its KS cluster-like ancestry?

epoch said...

@NeilB

According to this article from 2014 by Jean-Jacques Hublin Bohunician is clearly AMH on the basis of similarities of the lithic industry with contemporary Levant Emirian and the fact that it seems a break with the MP.

https://www.sciencedirect.com/science/article/pii/S0277379114003254

Shaikorth said...

@JuanRivera

Yana in fact prefers Sunghir as a source, distance is good but will this kind of thing agree with qpGraph? It likes Tianyuan and Hoabinhian over Ust-Ishim.

With MA-1

"distance%=2.3998"

RUS_Yana_UP:Yana1

RUS_Sunghir,50
RUS_MA1,18.8
CHN_Tianyuan,16.4
LAO_Hoabinhian,14.8

RUS_Ust_Ishim,0

Without MA-1

"distance%=4.9083"

RUS_Yana_UP:Yana1

RUS_Sunghir,67.8
CHN_Tianyuan,19.8
LAO_Hoabinhian,12.4

RUS_Ust_Ishim,0

Devil's Gate surprisingly eliminates Tianyuan from the fit completely:

"distance%=4.7267"

RUS_Yana_UP:Yana1

RUS_Sunghir,69.2
LAO_Hoabinhian,25
RUS_Devils_Gate_Cave_N,5.8

CHN_Tianyuan,0
RUS_Ust_Ishim,0

Arza said...

@ Shaikorth

If you want to test whether EHG has excess of Yana ancestry when compared to Yamnaya you should not use AG3 or WSHG. They (or EHG) are picked up because they provide better ANE-WHG ratio that overcomes the error introduced by additional Yana-like-shift that is not present in Yamnaya.

Let's replace AG3/WSHG with UKR_Mesolithic.

RUS_Karelia_HG
UKR_Mesolithic 86.8%
RUS_Yana_UP:Yana2 13.2%
RUS_Yana_UP:Yana1 0%
ROU_Iron_Gates_HG 0%
GEO_CHG:KK1 0%
Anatolia_Pinarbasi_HG:ZBC_IPB001 0%

Distance 10.2628%

As you can see Karelia_HG is shifted towards Yana compared to Ukrainian samples. Now, if Yamnaya is built on top of EHG as you claim, it should show some shift too.

Yamnaya_UKR
UKR_Mesolithic 54.6%
GEO_CHG:KK1 45%
RUS_Yana_UP:Yana2 0.4%
RUS_Yana_UP:Yana1 0%
ROU_Iron_Gates_HG 0%
Anatolia_Pinarbasi_HG:ZBC_IPB001 0%

Distance 7.7754%

But, as you can see, it practically doesn't show any.

JuanRivera said...

How about MA1 and AG3 as various mixtures of Yana, Sunghir, some ancient East Asian population and Pinarbasi_HG?

JuanRivera said...

And do Magadan_BA and Ust'-Belaya model as having steppe admixture?

Vinitharya said...

I always figured Basals were Haplogroup C, as it is the earliest differentiated exo-African haplogroup. C V-20, in particular, which still exists in a very low frequency in Europe, surviving after being swamped by the successful hunter gatherer I clades, farmer G2a and E-V13, Bell Beaker R1b, and the rugged steppe Chads of R1a.

JuanRivera said...

Yamnaya fits better when it's modeled having EHG (either Samara_EHG or Karelia_EHG). Also, Khvalynsk, "Ukraine_Eneolithic" and Piedmont have to be accounted for.

Shaikorth said...

Karelia_HG is according to qpAdm just a Sidelkino+Iron_Gates mixture (p-value for the fit is something like 0.9), and Sidelkino is best fitted as WHG+WSHG followed by WHG+AG3. Yana doesn't factor in there. Ukraine_Mesolithic is closer to WHG on the cline.

Andrzejewski said...

Samara EHG as Samara HG?

JuanRivera said...

The two denominations denote the same EHG individual.

Cy Tolliver said...

Is there anyway using either qpGraph, qpAdm, or some other tool to tease out whether Basal Eurasian is really a concrete population? Ever since the first ancient Near East genomes were released with both Natufians and Iran Mesolithic/Neolithic groups found to be heavily Basal Eurasian but not closely related and not sharing much obvious recent ancestry together, I've been skeptical BE represents a real population.

The Dzudzuana paper muddled things even further because we now have a 26,000 year old sample with Basal from the Caucasus, which I think rules out that Basal must be a an Epipaleolithic intrusion from Egypt to the Levant associated with E, as Dzudzuana is obviously too old for that (was Dzudzuana's Y-DNA noted in the original pre-print? I remember his mtDNA was U6). That Dzudzuana pre-print still hasn't been published, so I take it his genome still isn't available yet, which is unfortunate.

Andrzejewski said...

@Cy Tolliver Aren’t Semitic people the result of an admixture of Natufians and Iran Mesolithic/Neolithic?

Slumbery said...

@Shaikorth said...
"Karelia_HG is according to qpAdm just a Sidelkino+Iron_Gates mixture (p-value for the fit is something like 0.9),..."

I remain unconvinced. In our earlier discussion I told you that G25 nMonte does not only can't detect that (I would not think much of it if it were just a non-detection), but actually detects the opposite: higher eastern ancestry in Karelia_HG compared to Sidelkio.
You argument basically came down to the claim that nMonte is in general unreliable and always trumped by the infallible pqAdm. However I am reluctant to accept this, because:
1. such Balkan HG impact seems implausible and would sift (a shift, not a regional structure, because you claim the same happened in the Samara region) the WHG:ANE ratio in a way that would come up in later populations that received EHG admixture. For example Yamnaya should come out significantly WHG admixed when modeled with Sidelkio (but as far as I know it does not).
2. you could not tell any reason why the simple WHG + AG3 + CHG model for EHG would fail this way. These are very divergent populations and there is no over-fitting or any other obvious Achilles heel of nMonte in that test.

I would appreciate if somebody from here looked into that pqAdm test very rigorously and tried some alternatives.

JuanRivera said...

Some day, I tried to see if Khvalynsk has West_Siberia_N admixture by running maps, excluding EHG and then West_Siberia_N. Both produced terrible fits (though really close to each other), supporting the notion of EHG and West_Siberia_N being closely related and also rendering that approach useless.

Andrzejewski said...

“Mal’ta sits on the cline between AG3 and Yana”

“Ancestral Ancient North Eurasian”

How palpable is it, that both “EHG” and “CHG” are derivatives of sibling meta populations: EHG come mostly from ANE, CHG from Yana, and both ANE and Yana are offsprings of some Kostenki-Sunghir population?

And maybe the characterization of PIE as an amalgam of EHG + CHG is wrong and misleading, and the Proto-Indo-Europeans are a distinct population descendant from AANE *independently* from both the EHG branch and the CHG branch?

Andrzejewski said...

@JuanRivera how can you explain the demic and genetic shift from Samara HG to Khvalynsk?

JuanRivera said...

Maybe introduction of crops and livestock to the Northern Caucasus from the south. An already dense forager population would have an easier time growing crops and livestock in the comparatively humid Northern Caucasus, which in turn causes population growth, driving migration to the steppe. Having livestock then supports a higher population density than a hunter-gatherer lifestyle in dry places. All of that could account why both Khvalynsk, "Ukraine_Eneolithic" have significant Piedmont admixture, whereas Piedmont only showing a slight increase in EHG with time, with Yamnaya having even more Piedmont admixture on top of Khvalynsk and "Ukraine_Eneolithic".

Shaikorth said...

@Slumbery
It doesn't have to be specifically Balkan WHG, just something WHG, Latvia_HG+WSHG and Loschbour+WSHG still give better p-values for Sidelkino than ones where WSHG is replaced by AG3. The Karelia_HG model as Sidelkino + WHG gives a great P-value and Kale said that either WSHG or AG3 instead of Sidelkino fail so there seems to be no issues.

Re: modeling Sidelkino, nMonte and qpAdm aren't at odds there. WSHG+WHG is a bit better fit than AG3+WHG, just like in qpAdm, and adding CHG in nMonte gets the same result. But relying too much on nMonte gets you lots of Sarazm_En into the eneolithic Pontic steppe.

"distance%=3.9888"

RUS_Vonyuchka_En

Karelia_HG,41.8
CHG,32.4
Sarazm_En,25.8
Barcin_N,0
Iran_N,0
WHG,0

AFAIK qpAdm doesn't get get this stuff which makes sense.

Arza said...

@ Shaikorth

You've completely missed the point.

Open this plot:

https://i.postimg.cc/fbn2crP7/yana-EHGyamnaya.png

Data is taken from G25 scaled spreadsheet and the PCA is reprocessed.
What you see is practically PC1/PC3 view, but in fact this is a 3D plot of first three dimensions, just slightly rotated.

Why we need a 3D plot? To overlay Yamnaya on top of Globular Amphora.
According to Davidski Yamnaya has IIRC up to 15% of WHG rich EEF and GAC is the best proxy.
By placing Yamnaya over GAC I nullify any potential GAC influence on the position of Yamnaya in this plot. Simply no amount of added or subtracted GAC ancestry would shift the position of Yamnaya as Yamnaya-GAC axis is perpendicular to the screen.

As you can see, in this plot a straight line going from CHG via Yamnaya to HG groups points to Ukraine_N rather than EHG, as EHG is obviously shifted in the direction of Yana.

In multidimensional space CHG-Yamnaya line of course doesn't point to Ukraine_N as it has too much WHG, but to an empty space "over" EHG.

As a bonus you can see another interesting thing. "Yamnaya component" is older than Khvalynsk as Khvalynsk is intermediate between EHG and Yamnaya.

Michalis Moriopoulos said...

Hey, Dave, heads up:

Two of the Old Bering Ekvens in the G25 are mislabelled: I7335 & I7336. Those two are actually Ust Belaya samples. I discovered this once I saw how weirdly they plotted for putative Neo-Eskimos.

Arza said...

@ Andrzejewski
how can you explain the demic and genetic shift from Samara HG to Khvalynsk?

The only correct model is Yamnaya + EHG:

RUS_Khvalynsk_En
RUS_Sidelkino_HG 33.2%
Yamnaya_RUS_Kalmykia 20.6%
Yamnaya_RUS_Samara 19.8%
RUS_Karelia_HG 18.6%
RUS_West_Siberia_N 7.8%
UKR_Mesolithic 0%
UKR_N 0%
Baltic_LVA_HG 0%
Baltic_LVA_MN 0%
Baltic_LTU_Mesolithic 0%
Baltic_LTU_Narva 0%
Baltic_EST_Narva 0%
Baltic_EST_Comb_Ceramic_low_res 0%
RUS_Progress_Eneolithic 0%

Distance 2.1172%

Why? See the plot posted above.

Shaikorth said...

@Arza
Remember that PCA dimensions are affected by things like number of samples from a specific group on the plot, and sometimes this can be a distortion. If EHG has Yana influence Yamnaya lacks and you put Yana in pRight when modeling Yamnaya as EHG+X in qpAdm the fit should fail (because a mixture would share too much drift with Yana compared to real Yamnaya), if it doesn't we can be dealing with a simple PCA artifact.

JuanRivera said...

Couldn't age be affecting that? Yana is 31kya.

JuanRivera said...

Once the Yana paper samples are available in non-sponsor nMonte, I'll be running models on Yana, MA1, AG3, Magadan_BA, Ust'-Belaya II, and modern Chukotko-Kamchatkan and Nivkh speakers.

Slumbery said...

@Shaikorth said...

It doesn't have to be specifically Balkan WHG,...

It does not change anything from my side. Such a significant extra WHG still should come up in the models fro later EHG admixed population and still no apparent reason why the nMonte test would fail this way.

As for the further part of you comment, you again give an example where nMonte fails, but that is irrelevant here. We know nMonte can go wrong (especially if you intentionally set up a test just to show that it can go wrong...), but that does not mean these are fundamentally wrong in the terms of ANE:WHG ratio (or if they are, then why?):

"sample": "RUS_Sidelkino_HG:Average",
"fit": 5.379,
"RUS_AfontovaGora3": 60,
"WHG": 35,
"GEO_CHG": 5,

"sample": "RUS_Samara_HG:Average",
"fit": 6.3222,
"RUS_AfontovaGora3": 62.5,
"WHG": 30,
"GEO_CHG": 7.5,

"sample": "RUS_Karelia_HG:Average",
"fit": 5.8392,
"RUS_AfontovaGora3": 67.5,
"WHG": 30.83,
"GEO_CHG": 1.67,

Arza said...

@ JuanRivera

Couldn't age be affecting that? Yana is 31kya.

Samples of European HGs are more or less contemporary, but some show bigger shift than the others. The effect is real and fairly recent.

IMHO ANE wave was either heterogeneous when it comes to the amount of Yana ancestry or there were multiple migrations with different levels of this Yana-shift.

Andrzejewski said...

By “Ukraine_Eneolithic” I take it you mean Sredny Stog 2 instead of SS1?

Arza said...

@ Shaikorth

That's not an artefact, that's reality.

I reprocessed the PCA to shift all important data to first few dimensions, but you can recreate almost the same view just by rotating 3D PC2/4/5 G25 plot.

This is some old plot, full G25:
https://i.postimg.cc/zBfgTvjg/YUo.png

If you'll look carefully you'll see exactly the same structure. Kotias -> Yamnaya points to WHG, EHG too much shifted towards East Asia, Khvalynsk (here Samara Eneolithic) intermediate between EHG and Yamnaya.

"Steppe component" is old. Really old. Just deal with it.

Andrzejewski said...

What’s the chance that “Steppe component” isn’t a mix of EHG/ANE and CHG/Yana, but is basically a unique, sister population to both? As in: AANE —> 1. EHG via ANE; 2. CHG via Yana; and 3. “PIE”?

Shaikorth said...

@Slumbery
The qpAdm result simply means Karelia_HG shares more drift with WHG and CHG (CHG dampens WHG signal) and in the paper's D-stats WHG did share a slight bit more drift with non-Sidelkino EHG's than with Sidelkino.

The difference between Karelia_HG and Sidelkino caused by that is so trivial that it using one or the other isn't likely to have any effect on qpGraph phylogenies etc.

@Arza
Yamnaya fits perfectly as a mix of Dereivka and Progress_Eneolithic (p-value >0.9)

JuanRivera said...

With some Khvalynsk, which makes the fits slightly better. Though the levels of Khvalynsk and Progress change with each Yamnaya group, from Yamnaya_Samara (having the highest Khvalynsk) to Yamnaya_Caucasus (having the highest Progress and lowest Khvalynsk).

Shaikorth said...

@JuanRivera
That fit was for one of those David did for Yamnaya Samara in qpAdm. It's a bit different in nMonte. Progress_Eneolithic-like groups being a vector of CHG to Khvalynsk is possible but I don't recall if it was verified with qpAdm.

"distance%=2.5823"

RUS_Khvalynsk_En:I0122

RUS_Sidelkino_HG,60
RUS_Progress_En,40


It looks good here but might be anachronistic, there are a couple of combinations that give suspiciously good fits for ancients, for example:

"distance%=1.7479"

RUS_Sidelkino_HG:Sidelkino

UKR_Mesolithic,71.4
RUS_West_Siberia_N,28.6

Davidski said...

@Michalis

Two of the Old Bering Ekvens in the G25 are mislabelled: I7335 & I7336. Those two are actually Ust Belaya samples. I discovered this once I saw how weirdly they plotted for putative Neo-Eskimos.

Thanks, I've made the changes.

MaxT said...

@Davidsk

Will you be updating Basal-rich K7 spreadsheet with new samples?

Dragos said...

I might have misunderstood something; but why are we thinking Yana & Sunghir are somehow relevant for EHG , let alone Yamnaya ?

Gabriel said...

@Cy Tolliver

The Dzudzuana paper muddled things even further because we now have a 26,000 year old sample with Basal from the Caucasus, which I think rules out that Basal must be a an Epipaleolithic intrusion from Egypt to the Levant associated with E, as Dzudzuana is obviously too old for that

I wonder if the proto-Afro-Asiatic homeland question is affected by this.

priscus said...

Haven't they basically gone beyond the basic "basal<->non-basal" model? As I understand it, the "basal" in Dzudzuana or Anatolian HGs would essentially correspond to the original idea of it, while when it comes to north Africa and with it Natufians, they are talking about "deep" ancestry, even more basal basically.
The latter would probably be better linked to Y-DNA E, especially since it seems to have some sort of link with sub-saharan africa, while the original concept as the name itself suggests didn't seem to.

Kristiina said...

@ Dragos

I agree. Yana guys who lived 30 kya ago are P1, and that line came extinct.

On the other hand, the oldest R1a is from Ukraine c. 10 kya ago (8700 BCE?) and the oldest R1b is from Villabruna more than 10 kya ago.

These rather have a connection with Malta but that line came extinct as well. If we had a 20-10 kya old R1 sample that would be more useful.

Ric Hern said...

Interesting article, if I can find it again claims that 36% of Europe could have been inhabited during the maximum of the LGM. This study basically took into account Climatology,Plant and Animal presence of the LGM, Current Hunter Gatherer distribution etc...So it looks like there could have been populations living across Southern Russia and the Ukraine during this time. Low density compared to the Iberian Peninsula but populated non the less...So in essence there are still wide open spaces to explore....

epoch said...

An interesting little piece of information in the description of the site in the Supp Info

"During the excavation season of 2001 110 and 2009-2011, this site has yielded a Middle Palaeolithic layer (Micoquian, Kiik-Koba type, layer 111 B) located above an Early Upper Paleolithic layer (Streletskian or eastern Szeletian, layer C) and 112 below three Aurignacian layers (layers 6-5 to 6-3), followed by three Gravettian layers (layers 6-113 2, 6-1 and 5-2) and one Swiderian layer (layer 4), all densely layered across 1 to 2 meters of depth."

Swiderian is supposed to be the culture that brought Baltic HG to the Baltic. It has been associated with the Magdalenian (Sometimes also with the Solutrean based on willow leaf blades IIRC) from what I read and Baltic_HG has a substantial amount of El_Miron, just as Polish GAC has 3% GoyetQ2.

epoch said...

Also, the site was unoccupied for roughly 20,000 years after 34 ka BP. I don't know how far north that can be extrapolated. But it is interesting to see it repopulated with the Swiderian culture, which I think is associated with WHG.

Aram said...

epoch

My assumption is based on this. http://eurogenes.blogspot.com/2018/01/unadmixed-basal-eurasians-lived.html Partly confirmed by Dzudzuana.
Also notice that the sample with one of most low level of Neanderthal admixture is found in NE Iran Mesolithic.

Aram said...

CHG as a two source admixture. Pinarbasi could be a proxy for Dzudzuana.


[1] "distance%=13.5553"
GEO_CHG

Anatolia_Pinarbasi_HG,51.2
RUS_AfontovaGora3,48.8

then little bit better

[1] "distance%=12.8627"
GEO_CHG

RUS_MA1,64
Anatolia_Pinarbasi_HG,36

and then Yana

[1] "distance%=12.6292"
GEO_CHG

RUS_Yana_UP,100

I messed something or it should be like this?
And here as a north-south axis.

[1] "distance%=10.6222"
GEO_CHG

IRN_Ganj_Dareh_N,59.4
RUS_Samara_HG,22.4
Anatolia_Pinarbasi_HG,18.2

Davidski said...

@MaxT

I've added these high coverage samples to the Basal-rich K7 spreadsheet...

Kolyma_Meso:Kolyma_River
Devils_Gate_Cave_N:NEO240
Yana_UP:Yana1
Yana_UP:Yana2

The Basal-rich K7

But it looks like the Yana genomes are way too old for this test, because their ancestry proportions look like garbage. You should see much more sensible results with the Global25.

George said...

Hi Ric,

Human population dynamics in Europe over the Last Glacial Maximum
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4500234/pdf/pnas.201503784.pdf

Dragos said...

@ Epoch

''Swiderian is supposed to be the culture that brought Baltic HG to the Baltic. It has been associated with the Magdalenian ''
'
Yes that is the old and up to now dominant theory. However, the Magdalenians seem to have been largely replaced ....

Ric Hern said...

Yes. Thanks George.

Cy Tolliver said...

@priscus

My understanding was that Natufians still had proper Basal ancestry, with some additional deep ancestry related to ancient North Africans. i'm very curious to finally resolve the relationship between the Basal ancestry as it exists in Natufians, Anatolian/Dzudzuana HG, and Iran foragers - is it truly all derivative of a single coherent source, or were there multiple diverged Basal populations floating around in Southwest Asia in the late Pleistocene?

epoch said...

@Dragos

But have a look at extended data figure 6 from the Dzudzuana paper. There Latvia_HG has a firm chunk of ElMiron. Less then Loschbour, but still. Also, the 3% GoyetQ2 in Polish GAC points to some survival.

https://www.biorxiv.org/content/biorxiv/early/2018/09/20/423079/F9.large.jpg

Dragos said...

@ epoch
Sure; but a few percent survival’ is a far cry from the long established view that Magdalenians or their direct ancestors were the main recolonisors of post Ice age Europe

epoch said...

@Dragos

A suggestion of what may have happened after LGM: Expanding from Iberia the Magdaelenians followed reindeer, one of the two favourite Magdalenian prey, to the north once the climate warmed. From Italy the Epigravettians expanded north and east, but they were far less discriminate in what game they followed [1]. Magdalenians in Southern France and Iberia hunted the other favourite Magdalenian game: Horses. Mind you, at that point Iberian Magdalenian already had pretty WHG ancestry.

Then the Laacher See eruption hit and as we see tephra deposits in an important area for reindeer you can imagine this must have had a huge impact on both the reindeer herds and wiped away most of the Northern Magdalenians.

But reindeer hunting cultures recovered the land and it is remarkable that Mesolithic WHG samples from both the area of the previous Ahrensberg culture (Loschbour) and the Swiderian culture (Latvia_HG) showed Magdalenian ancestry.

We saw that the Magdalenian ancestry was only slowly replaced by WHG in Iberia, with even a Azilian sample being half GoyetQ2, compared to the almost full Villabruna ancestry of Bichon, also an Azilian sample.

[1] That may follow from this paper, which shows a different mobility for Epigravettians when compared to late Gravettians: https://www.nature.com/articles/s41559-019-0900-8

Matt said...

Few basic G25 plots (no reprocessing) showing where Yana UP clusters in relation to other populations: https://imgur.com/a/xMOAjLT

Yana seems to AG3 as UP Europeans are to Loschbour (MA1 to AG3 as El Miron to Loschbour).

Cladistically relative to outgroups populations may be similar, but lacks the specific drift path that defines AG3 in G25.

Closest populations to Yana_UP in G25 distance measures are other UP Europeans (Kostenki, Sunghir, GoyetQ116, to extent Vestonice) and populations to a lesser extent contemporary populations who have a strong mixture of different Eurasian population and so don't have too much specific drift down one Eurasian path or another (Turkmen, Uzbek, Baskhir, Burusho). Reasonably similar to patterns with GoyetQ116 and Kostenki.

Cy Tolliver said...

@Matt,

Do you have any opinion on where Yana's Asian ancestry came from? Something related to Tianyuan?

WeightofAudio said...

N1b happens to be my mtdna haplogroup. Interesting. Perhaps an epiGravettian lineage that survived for a while restricted to the Tisza and southeastern Europe, and then after a bit of time, dispersed in all directions, including northeast europe (in my case)? I know the sample wasn't confirmed N1b-proper, but it's obviously a lineage that transitioned into it, or a like kind.

Dragos said...

@ epoch
The transition from magdalenians in northern latitude precedes the Lecher see; and relates to the Azilian expansion (also from southern France); which must have had a better hunting strategy for the new climate with denser forest cover.
As you say; Magdalenian ancestry remained highest in Iberia ; but elsewhere it was overridden more expediently

I’m not sure that the eligravettian expanded out of Italy . The details of Gravettian -> epigravettian remain murky, both in Italy- Balkans and Eastern Europe

Samuel Andrews said...

@David,

Any idea when the Dzudzuana genomes will be available? Also do you have a link to Chad Rofhlsten's blog. I remember he did posts on Upper Paleolithic genomes.

Samuel Andrews said...

N1/N1b mtDNA in Gravettian Europe is a surprise. N1 & N1b seemed like good candidates for basal Eurasian mtDNA from southwest Asia.

U5b2b in Italy dating 19ky makes me think there was WHG in Italy that long ago. I wish we had the genome of that sample. Villabruna, 14ky Italy, belonged to U5b2b btw. Possibly WHG, or at least western European WHG, comes from Italy.

TLT said...

@Arza

>AANE went to Siberia, mixed with locals, acquired Y-DNA R and bounced back to West Eurasia bending any possible WHG-AANE cline towards East Asia.

Wasn't Yana modeled as part early west Eurasian (Sungir-like) and part early east Eurasian (something similar to a Tianyuan-like population) in that Yana paper? I guess that most of the ANE autosomal DNA comes from the Sungir-like west Eurasian source (which originally almost certainly had the C1a2 yDNA) while the Tianyuan-like east Eurasian source provided either P* or P1* into the mixture that resulted in ANS/Yana population. IDK what the mtDNA associated with this Tianyuan-like source would be like. Perhaps some kind of an mtDNA R* as in similar to what was found in Ust'Ishim?

TLT said...

The mtDNA haplogroup N is no surprise since a Dzudzuana (a population partly descended from a Gravettian-like source) sample had it. Someone pointed out that yDNA G could have come from a basal source, so a basal yDNA G + a Gravettian-like mtDNA N1 would result in a mixed Gravettian-like + basal Eurasian Dzudzuana population later on.

Ric Hern said...

As far as I can remember P (MNOPS) is a descendant of K and K split from IJK. Ust Ishim was already K2a. Was Tianyuan Haplogroup K, otherwise I can not see how it could be ancestral to P ?

Matt said...

@Cy, I'll have a think about that. I don't think at the moment I can see any evidence that Yana is much closer to any present day or Holocene inferred East Eurasian group - e.g. AASI, East Asian, Onge. Papuans look more distant due to the 3% Denisovan infusion, other than that, not huge differences.

Some plots: https://imgur.com/a/yBMhPdX

Using Biaka and Global25 distances to try and construct some measures of differentiation adjusting using Biaka for lower levels of (captured?) differentiation in Global25. Colour scheme as Davidski's usual for Global25 datasheet.

May be worth doing big blocs of D/f4 statistics and seeing what happens.

Ric Hern said...

@ TLT

In this paper they rather think that Basal will be found Somewhere among the Early Ahmarian population which predates both the Aurignacian and Gravettian...if I understood correctly..

TLT said...

@ Ric Hern

>As far as I can remember P (MNOPS) is a descendant of K and K split from IJK. Ust Ishim was already K2a. Was Tianyuan Haplogroup K, otherwise I can not see how it could be ancestral to P ?

When I said a Tianyuan-like population, I am really referring to any kind of an early east Eurasian population. K2a is certainly something that comes from early east Eurasian and all of the non-P subclades of K2b are now a days mostly found among Negritos, Papuans and some related populations (yet again, east Eurasians). I think that some early upper paleolithic proto-Negrito/Papuan-like population has to be ancestral to P* in the form of K2b carriers. The current idea is that K2b came out of southeast Asia/Sundaland and perhaps emerged while already diversified into various descendant subclades like P* (instead of these subclades diversifying outside of Sundaland). The later mutations responsible for P1* however could have happened further up north later on.

Ric Hern said...

@ TLT

How much Denisovan in Yana ?

Matt said...

Re; different East Eurasian ancestry which could have contributed to Yana UP, it seems that it did not share too much drift with Tianyuan, judging by Global 25 distances as a measure: https://imgur.com/a/Ofuch34.

Judging by a scale where Ust Ishim is 1, Tianyuan is 0, then East Asians (today): 0.1, Onge: 0.27, AG3: 0.68, Villabruna: 0.73, Yana: 0.78, GoyetQ116: 0.83, Sunghir: 0.86. Assuming about 22-29% East Eurasian in Yana as per paper, then other side of Yana about 0.6 on UI->Tianyuan scale.

Kind of a simple calculation and probably better to use D-stats for this (not totally sure how these distances are behaving), but I think it bears out that any East Eurasian ancestry in Yana must have been almost a trifurcation between East Eurasian+West Eurasian (like Ust Ishim is), and not have shared very much drift with any subsequent ENA group.

(Paper states: " Demographic modelling of the high-coverage individual Yana1 using a site-frequency-spectrum-based framework indicates an early divergence of the ANS lineage at about 39 ka (95% confidence interval (CI) 32.2–45.8 ka), concomitant with substantial gene flow (approximately 29%; 95% CI 21.3–40.1%) from East Asians, an event that probably occurred very soon after the latter diverged from West Eurasians 43.1 ka (95% CI 33.4–48.6 ka) (Fig. 2a, Supplementary Information 7)."

E.g. effective almost trifurcation with West Eurasians and main trunk ancestor of Tianyuan etc, possibly before Tianyuan split as a very early branch at almost itself the split of East and West Eurasian - https://www.cell.com/current-biology/fulltext/S0960-9822(17)31195-8. Not any ancestor contributing much greater to e.g. "East Asians" compared to Onge etc.

Chronologically the split between the East Eurasian 22-29% of Yana and surviving East Eurasian lineages+Tianyuan must have happened slightly prior to the split of Yana's main 78-71% ancestry from Sunghir, K14, etc.)

TLT said...

@ Matt:

Could the split between the east Eurasian contributors to Yana and generic early east Eurasians (like Tianyuan for example) be as old as the formation of yDNA K2b?

Cy Tolliver said...

@Matt,

Super interesting stuff as always from you. Tianyuan seems less related to contemporary Asians than UP Euros do to contemporary Europeans. And regarding your 1-0 scale, I wouldn't have expected Villabruna to be more Tianyuan shifted than Yana.

The fact that the "Asian" side of Yana is apparently even more basal to Tianyuan makes me wonder if it can even be called Asian - maybe it's some sort of modeling artifact, or perhaps that component itself is some kind of early hybrid West-East population that mixed into the more obviously West Eurasian side of Yana. I think if we're able to get more ancient DNA from Central Asia/western China (maybe even South Asia too) we'll find that the split of West and Eurasians is not as clean as it's often made out to be.

Matt said...

@cy, thanks, I'd say it needs checking with formal stats as well; those would adjust for differing drift levels from African in the "row" populations while I only adjusted for differences in the "column" (not totally sure what impact that would have although I think adjusting only in the "column" would give a better idea of how much drift is shared relative to differences in drift in the "columns" without strongly pushing all drifted populations downstream together?).

But I would also say think Kale from Anthrogenica who is looking at Yana with lots of deep formal stats suggests "now that I have Yana, I can say that if Yana is proto-ANE, then East-Asians have about 5% ANE. It is not that ANE's East Eurasian derives from an East-Asian related branch." (https://anthrogenica.com/showthread.php?15693-Pleistocene-Siberian-Genomes-Ancient-North-Siberians-(Sikora-et-al)&p=579195#post579195).

The plots I gave above don't change too much with Hoabinhian and Jomon in place of Tianyuan, at least in terms of Yana UP's place in them: https://imgur.com/a/cnjUGcj

The tree modelling in Sikora 2019 for East Eurasian ancestry in Yana was OK, but I think at getting at the source of East Eurasian ancestry and when it branched, they'll do better to use the new Jomon samples and Tianyuan as well rather than just using moderns+Devils Gate as they did. Ideally Hoabinhian as well (if quality allows!). Cover the McColl et al model of East Asian as Tianyuan related+Onge/Hoabinhian related. I think the tree Sikora presents where the East Eurasian ancestry in Yana looks to split off from an East Asian related branch looks a bit questionable anyway.

@TLT, I don't think I have a strong idea about it; I'd think the age range seems like it could be possible, but I'm not clear one way or the other if it is the simplest solution given the distribution of k2b clades (in modern day and sparse attestation in ancient dna).

Ric Hern said...

Maybe the split between C and F when we look at Tianyuan MtDNA Haplogroup B which looks as if it spread with Y-DNA Haplogroup C early on towards the East ? Or maybe the K2b/K2a split if Tianyuan Y-DNA was along the NO lines. However the affinity between Goyet and Tianyuan seems to hint at the C/F split ?