search this blog

Saturday, February 6, 2016

ANE admixture in Caucasus Hunter-Gatherer Kotias

The idea of significant Ancient North Eurasian (ANE) admixture in Caucasus Hunter-Gatherers (CHG) was rejected in the paper that introduced us to the CHG metapopulation (see page 3 here).

But to me it seems like an obvious conclusion when looking at the positions of the CHG genomes on a typical West Eurasian Principal Component Analysis (PCA) plot. For instance, see here. Note their massive eastern shift relative to all of the other present-day and ancient Near Eastern samples, in particular the Anatolian Neolithic farmers from Mathieson et al. 2015.

We'll probably find out what's causing this shift in the next major paper on Eurasian paleogenomics. But I'd say that the analyses below, based on D-stats and f3-stats, comparing CHG genome Kotias to the Anatolian farmers, provide persuasive evidence that CHG is indeed admixed with ANE or something very closely related.

Perhaps surprisingly, the Itelmen people of eastern Siberia produce the most significant Z-score in D-stats of the form D(Chimp,X)(Anatolia_Neolithic,Kotias), where X is every population from North and East Eurasia and the Americas in my dataset likely to harbor ANE. Mal'ta boy or MA1, the main ANE proxy, is way down the list with a non-significant (<3) result. However, I'm pretty sure that a higher coverage sequence with over 500K SNPs would come through with the goods.

Note that here Kotias shares more alleles than predicted by the line of best fit with most Amerindians, most Siberians and Eastern Hunter-Gatherers (EHG), but not East Asians or Western Hunter-Gatherers (WHG). I can't think of anything other than ANE that might be causing this, because whatever it is, it has to be shared by these Amerindians, Siberians and EHG, to the exclusion of East Asians and WHG.

See also...

CHG admixture in early western Anatolian farmers


Seinundzeit said...


Very interesting.

Ever since these genomes were analyzed, I've entertained a vague/unsubstantiated notion about the West Eurasian portion of their ancestry. Mainly, just like how the Anatolian Neolithic population was basically an even mix of WHG and "Basal Eurasian", perhaps CHG was a mix of ANE, "Proto-West Eurasian" (something like Kostenki14), and "Basal Eurasian"? But that's assuming "Basal Eurasian" is real, which is still an open question.

Is there any way to test this? Perhaps TreeMix?

Davidski said...

TreeMix occasionally picks up edges from MA1 and EHG to Kotias, but I haven't been able to consistently reproduce them. It doesn't show any other edges into Kotias.

Another thing that I didn't mention is that Kotias is probably a little more basal than the Anatolian farmers, and yet they more or less show the same relationship to East Asians, which might mean that Kotias has some East Asian admixture too.

Seinundzeit said...

I think that the MA1/EHG link that TreeMix occasionally shows is definitely real, looking at the d-stats you've done above. And some East Asian admixture would probably make sense.

But I do wonder, is "Basal Eurasian" still a parsimonious construct? Perhaps the concept needs to be reexamined, in light of Kostenki14 and Ust-Ishim (and also in light of the fact that CHG is just like the Anatolian Neolithic samples in terms of East Asian-relatedness, despite having very different relationships with West Eurasian foragers like WHG, EHG, and MA1 in comparison to Anatolian Neolithic)? Fascinating stuff.

Matt said...

Could you run a full list of:
D(Chimp,Kotias,(Native American/Siberian),ENA) and D(Chimp,Anatolia_Neolithic,(Native American/Siberian),ENA) for comparison?
Chimp Kotias Itelman Dai
Chimp Kotias Chukchi Dai
Chimp Kotias Surui Dai
Chimp Kotias Surui Dai
Chimp Kotias Itelman Atayal
Chimp Kotias Chukchi Atayal
Chimp Kotias Surui Atayal
Chimp Kotias Itelman Papuan
Chimp Kotias Chukchi Papuan
Chimp Kotias Surui Papuan
Chimp Kotias Itelman Australian
Chimp Kotias Chukchi Australian
Chimp Kotias Surui Australian
Chimp Anatolia_Neolithic Itelman Dai
Chimp Anatolia_Neolithic Chukchi Dai
Chimp Anatolia_Neolithic Surui Dai
Chimp Anatolia_Neolithic Surui Dai
Chimp Anatolia_Neolithic Itelman Atayal
Chimp Anatolia_Neolithic Chukchi Atayal
Chimp Anatolia_Neolithic Surui Atayal
Chimp Anatolia_Neolithic Itelman Papuan
Chimp Anatolia_Neolithic Chukchi Papuan
Chimp Anatolia_Neolithic Surui Papuan
Chimp Anatolia_Neolithic Itelman Australian
Chimp Anatolia_Neolithic Chukchi Australian
Chimp Anatolia_Neolithic Surui Australian

Kurd Dgk said...

@ David

You are correct, my analysis also pointed to ANE shift of both Satsurbila and Kotias compared with ANE shift of Anatolians. Even IBS shows this.

The following sorted table with most MA1 shift on top, shows Satsurbila at no.25, and Kotias at no.37, to be much higher up the list than Anatolians at no.74. This table also shows E Asians further down the list than Sats and Kotias.

1 MA1 100.00%
2 Karelia_HG 68.67%
3 Samara_HG 68.32%
4 Karitiana 67.57%
5 Yamnaya_Samara 67.47%
6 Poltavka 67.34%
7 Yamnaya_Kalmykia 67.27%
8 Afanasievo 67.23%
9 Kostenki14_UP 67.04%
10 Sintashta_MBA 66.96%
11 Andronovo 66.91%
12 Poltavka_outlier 66.89%
13 LaBrana 66.73%
14 Scythian_IA 66.70%
15 Lithuanian 66.68%
16 Tajik_Pomiri 66.64%
17 Finnish 66.59%
18 Kalash 66.59%
19 Loschbour 66.57%
20 Burusho 66.57%
21 Pathan 66.57%
22 Bhil 66.53%
23 Srubnaya 66.53%
24 English_Cornwall 66.51%
25 Satsurbila 66.50%
26 Hungary_HG 66.48%
27 Paniyas 66.48%
28 Czech 66.47%
29 Palliyar 66.44%
30 Sindhi 66.44%
31 Punjabi_Lahore 66.44%
32 Hungarian 66.42%
33 Ho 66.36%
34 French 66.35%
35 Tajik 66.29%
36 Kumyk 66.29%
37 Kotias 66.29%
38 Nganasan 66.28%
39 Brahui 66.27%
40 Bengali 66.27%
41 Altaian 66.26%
42 Uzbek_Afghan 66.25%
43 Balochi 66.24%
44 Makrani 66.24%
45 Uzbek 66.22%
46 Kharia 66.21%
47 Pashtun_Afghan 66.21%
48 Hazara 66.20%
49 Turkish 66.18%
50 Bulgarian 66.17%
51 Onge 66.16%
52 Turkmen_Afghan 66.15%
53 Turkmen 66.14%
54 Tajik_Afghan 66.14%
55 Hungary_BA 66.09%
56 Andamanese 66.08%
57 Armenian 65.95%
58 Kusunda 65.88%
59 Hungary_CA 65.87%
60 Kurd_N 65.86%
61 Ust_Ishim 65.84%
62 Iceman_MN 65.73%
63 Mongola 65.72%
64 Australian 65.67%
65 Syrian 65.65%
66 Dai 65.64%
67 Hungary_EN 65.64%
68 Bougainville 65.62%
69 Ami 65.62%
70 Atayal 65.61%
71 Han 65.61%
72 Jordanian 65.59%
73 Papuan 65.54%
74 Anatolia_Neolithic 65.52%
75 Saudi 65.50%
76 Stuttgart 65.49%
77 BedouinB 65.44%
78 LBK_EN 65.43%
79 Yemen 65.22%
80 Ethiopian_Jew 64.10%
81 Somali 63.72%
82 Masai_Kinyawa_MKK 63.22%
83 Khomani 62.68%
84 Yoruba 62.22%
85 MbutiPygmy 62.20%
86 Ju_hoan_North 62.19%

Alberto said...

Yes, CHG are very eastern, so ANE is the main suspect for now.

@Kurd, thanks for that list and especially for this thread that I just found with many IBS lists:

Very interesting stuff there. But to stay on topic, what I noticed is that in the Kotias IBS list, Karelia_HG, Kostenki14 and MA1 appear right at the bottom. Looking through the list to see which population without much explicit CHG or high Basal Eurasian admixture scores high, I can just think of Pulliyar, so I wonder if ASI could be involved in this.

David, do stats with something very ASI (Paniya, Pulliyar) turn out significant?

Kurd Dgk said...

@ Alberto

The relevant IBS for comparing Kotias and Anatolian's position to MA1 is the one I posted above. In addition to relevancy, the above run equalizes the playing field, by comparing all samples to MA1 using MA1's reduced number of markers.

Alberto said...


Yes, indeed. And I don't doubt that Kotias is ANE shifted when compared to Anatolian Neolithic, since it's very significantly more eastern. Just looking at other possibilities too. The IBS list of Kotias is a bit surprising, in that it shows Yamnaya, Karelia_HG and MA1 below populations like Ust-Ishim, Loschbour, Han, Dai, Ami, Atayal, Papuan, Mota, Yoruba, Mbuti... With formal stats I think that Yamnaya is at the top or near, so who knows what's going on there. Might just be a coverage issue since Stuttgart is significantly higher than LBK_EN or Anatolia_Neolithic too.

Alberto said...

Yes, looking at the IBS sharing of Mbuti, Yoruba or Ust-Ishim those same populations (Yamnaya, MA1, Sintashta, Karelia_HG...) appear at the bottom, so it seems that it's related to the coverage.

Besides, some recent D-stats showed that populations with higher CHG/ASI ratio favoured Dai over Onge proportionally, so CHG can't have any ASI.

Davidski said...


Chimp Kotias Itelmen Dai -0.0141 -4.524 510071
Chimp Kotias Chukchi Dai -0.0137 -4.949 510071
Chimp Kotias Surui Dai -0.0252 -5.92 510071
Chimp Kotias Surui Dai -0.0252 -5.92 510071
Chimp Kotias Itelmen Atayal -0.0142 -3.946 510071
Chimp Kotias Chukchi Atayal -0.0138 -4.356 510071
Chimp Kotias Surui Atayal -0.0253 -5.593 510071
Chimp Kotias Itelmen Papuan -0.0591 -12.244 510071
Chimp Kotias Chukchi Papuan -0.0587 -12.904 510071
Chimp Kotias Surui Papuan -0.0698 -13.25 510071
Chimp Kotias Itelmen Australian -0.0593 -11.626 510066
Chimp Kotias Chukchi Australian -0.059 -12.157 510066
Chimp Kotias Surui Australian -0.0696 -12.305 510066
Chimp Anatolia_Neolithic Itelmen Dai -0.009 -3.851 592940
Chimp Anatolia_Neolithic Chukchi Dai -0.0092 -4.536 592940
Chimp Anatolia_Neolithic Surui Dai -0.0194 -6.436 592940
Chimp Anatolia_Neolithic Surui Dai -0.0194 -6.436 592940
Chimp Anatolia_Neolithic Itelmen Atayal -0.0088 -3.378 592940
Chimp Anatolia_Neolithic Chukchi Atayal -0.0091 -3.887 592940
Chimp Anatolia_Neolithic Surui Atayal -0.0193 -5.997 592940
Chimp Anatolia_Neolithic Itelmen Papuan -0.0494 -12.651 592940
Chimp Anatolia_Neolithic Chukchi Papuan -0.0496 -13.189 592940
Chimp Anatolia_Neolithic Surui Papuan -0.0593 -14.15 592940
Chimp Anatolia_Neolithic Itelmen Australian -0.046 -11.474 592934
Chimp Anatolia_Neolithic Chukchi Australian -0.0463 -12.045 592934
Chimp Anatolia_Neolithic Surui Australian -0.0556 -12.53 592934


Chimp Han Anatolia_Neolithic Kotias 0.0068 1.92 508715
Chimp Dai Anatolia_Neolithic Kotias 0.0068 1.882 508715
Chimp Atayal Anatolia_Neolithic Kotias 0.0069 1.73 508715
Chimp Papuan Anatolia_Neolithic Kotias 0.0028 0.711 508715
Chimp Australian Anatolia_Neolithic Kotias 0.0001 0.021 508710


The stats for Paniya would probably be almost significant with the full marker set. But like all non-Austroasiatic South Asians, Paniya also have a lot of CHG ancestry.

Chimp Paniya Anatolia_Neolithic Kotias 0.0068 1.565 111771
Chimp Pulliyar Anatolia_Neolithic Kotias 0.0058 1.372 112295

Chimp Itelmen Anatolia_Neolithic Kotias 0.0087 1.934 112296
Chimp Chukchi Anatolia_Neolithic Kotias 0.0084 1.919 112296

FrankN said...

Dave: Your analysis assumes that Mbuti and AnatNeol. are unrelated. This assumption is most likely wrong! Mbuti are traditionally using the Basenji dog, which has been shown to be a cross-breed of European dogs and Israeli wolves. OTOH, the Basenji is one of the "basal", i.e. very ancient breeds. This implies early, UP or mesolithic transfer of the Basenji from the NE into the Congo Basin.

OTOH, I have several times posted here about a putative Caucasian/ Middle Eastern post-Mota immigration into Africa. This may ultimately also have reached the Mbuti (or not).

I suggest you redo your analysis by comparing f³stats with Gorilla as outgroup. If the trends hold, they deserve indepth discussion.

Davidski said...

Yeah, Gorilla works great. Kinda hard to plot though.

I think I'll stick with Mbuti and Ju_hoan_North in these sorts of outgroup f3 tests.

Their results basically match the D-stats posted above with Chimp as outgroup. Did you notice that?

Romulus said...

is it ANE in CHG or CHG in ANE?

Davidski said...

Do you really think it's possible for, say, Surui Indians from Brazil to carry CHG and thus Basal Eurasian without anyone noticing till now?

Romulus said...

yes you're right

FrankN said...

@Dave: I have realised that my above comment may have sounded a bit rude. Wasn't meant that way, sorry. I also understand your Gorilla problem.
I saw your Chimp stats just after I had posted. I still have to digest them but they seem to confirm your conclusion.

EHG being more EEF- than CHG loaded is expectable. Magdalenian and Natufian shared dogs, and Cyprus was first visited by seafaring HGs, before EEF took over.

I also don't have a general problem to imagine ANE admix in CHG. During interglacials, the Imereti culture seems to have reached the lower Don/Volga area, providing sufficient opportunity to intemix with whoever found refuge in the Southern Urals. But the easier way would have been down the Kura, through the Southern Caspian refuge, and then along Lake Aral and the Amur Daja towards the Bactrian refuge. IOW - I would have expected more post-LGM CHG admix with Central Asians than with ANE, and Central Asians rather than Siberians and Amerindians topping your list.

The CHG appeal in Chukotko-Kamchatkans is surprising at first sight. OTOH, there is archeological and genetic evidence for West Eurasian incursion into the Cis-Bajkal region during the 4th mBC. That's too early for the CWC-type, EEF-enhanced expansion (Sintashta/ Andronovo), but fits the time frame of the preceding expansion (Andronovo) that was still EHG/CHG dominated. Why Chukotko-Kamchatkan is vocabulary-wise so close to IE is still a bit of a mistery, and it is also not spoken directly next to Lake Baikal - nevertheless, further neolithic movement of West-Eurasians towards Kamchatka, leading to genetic and linguistic interaction, isn't unthinkable of.

That leaves us with the Amerindians:
"Do you really think it's possible for, say, Surui Indians from Brazil to carry CHG and thus Basal Eurasian without anyone noticing till now?"
Well, what has been noted by Skoglund/Reich 2015 is the Surui and Karitiana carrying Oceanic DNA, with significant Z-scores against Onge, Papuans, New_Guinea and Australian_WGA. That, in turn, takes us back to yDNA T, with 8% presence among Sulawesi Sea Nomads, to transfer of Coconuts from that region to the coast of Ecuador around 500 BC, plus the ample evidence for Polynesian-Amerindian contact.

Also, from that paper's Supp.Mat, Sect. 5 (CHROMOPAINTER rresults):
"The Chane shows a highly significant rate of copying from Turkish_Jew and other European populations such as Norwegians, suggesting that this single Chane individual might have cryptic European ancestry. Similarly, the single Zoro individual show a top Z-score for African Wambo. (..) The Pima show an affinity to African Tshwa (Z = 3.71), possibly suggesting cryptic African ancestry." [That's why Ju'Hoan isn't a particular good idea for such tests]

Other nice Z-scores:
Bolivian - Kinh 2.76
Aymara- Tshwa 2.24
Mayan - Turkmen 2.73
Apalai - Gui 2.76
Kaqchikel -Ukranian 2.65
Quechua - Egyptian 2.14

Rob said...


can you elaborate on the Bactrian refuge ?
As far as I know, no actual settlenents , Campsites or other human finds from the LGM have come to light. Thus, at present, LGM Refugia in Bactria, or indeed any part of Central Asia outside the Altai, remain hypothetical

Krefter said...


I just have a few more D-stats for today. I think they can confirm EHG and CHG ancestry in Italians.

Remedello_BA Italian_Tuscan Karelia_HG Loschbour
Remedello_BA Italian_Tuscan Kotais Bulgarian
Chimp Karelia_HG Samara_HG Italian_Tuscan
Chimp Karelia_HG Samara_HG Remedello_BA
Chimp Satsurblia Kotais Italian_Tuscan
Chimp Satsurblia Kotais Remedello_BA

FrankN said...

Also from that paper (as extract):

Table S3.2. Significant statistics of the form f4(A, B; Mixe, Surui).

A B f4(A, B; Mixe, Surui) Z
Chimpanzee Onge 0.00100490 4.01
Chimpanzee Papuan 0.00083341 3.29
Chimpanzee New_Guinea 0.00083884 3.17
Chimpanzee Australian_WGA 0.00085497 3.23
Mbuti Onge 0.00088618 4.06
Mbuti Papuan 0.00071469 3.30
Mbuti New_Guinea 0.00072013 3.11
Mbuti Australian_WGA 0.00073710 3.19
Biaka Onge 0.00091768 4.36
Biaka Papuan 0.00074620 3.56
Biaka New_Guinea 0.00075163 3.33
Biaka Australian_WGA 0.00076856 3.41
Yoruba Onge 0.00085923 4.21
Yoruba Papuan 0.00068774 3.37
Yoruba New_Guinea 0.00069317 3.13
Yoruba Australian_WGA 0.00071020 3.28
Dinka Onge 0.00104014 4.90
Dinka Papuan 0.00086865 4.14
Dinka New_Guinea 0.00087409 3.88
Dinka Australian_WGA 0.00089104 4.01
Ju_hoan_North Onge 0.00102585 4.79
Ju_hoan_North Papuan 0.00085436 3.94
Ju_hoan_North New_Guinea 0.00085979 3.69
Ju_hoan_North Australian_WGA 0.00087661 3.81
Yakut Onge 0.00078838 3.95
Yakut Papuan 0.00061689 3.21
Yakut New_Guinea 0.00062233 2.92
Yakut Australian_WGA 0.00063896 2.91
Han Onge 0.00079784 4.10
Han Papuan 0.00062635 3.34
Han New_Guinea 0.00063179 2.95
Han Australian_WGA 0.00064859 2.99
Yukagir Onge 0.00090198 4.64
Yukagir Papuan 0.00073050 3.87
Yukagir New_Guinea 0.00073593 3.50
Yukagir Australian_WGA 0.00075238 3.51

Differentials to the Chimp values, for the Onge stats only:
Mbuti -0,0001187 0,05
Biaka -0,0000872 0,35
Yoruba-0,0001457 0,20
Dinka 0,0000352 0,89
JuHoan 0,0000210 0,78

Not too much of a variation, but some.

Gökhan said...

@Romulus "is it ANE in CHG or CHG in ANE? "

Thats the question should be asked. I think CHG in so called "ANE"

Davidski said...

That's the question should be asked. I think CHG in so called "ANE".

The question was asked and answered. The answer is an emphatic no.

Do you really think it's possible for, say, Surui Indians from Brazil to carry CHG and thus Basal Eurasian without anyone noticing till now?

Gökhan said...

"Do you really think it's possible for, say, Surui Indians from Brazil to carry CHG and thus Basal Eurasian without anyone noticing till now?"

As how nobody noticing the "ANE" in CHG till now my answer on your question; why not it is possible?

Karl_K said...


"why not it is possible?"

Because. If all non-Africans had Basal Eurasian, then it wouldn't be basal to anything.

That's how we can tell that Amerindians do not have it, while CHG does.

If Amerindians had CHG ancestry (via ANE), then they too would have Basal Eurasian ancestry (as would the Mal'ta boy).

This doesn't mean that populations ancestral to these populations did not mix with one another in their pasts, but it means that EEF and CHG got an extra dose of something that was an outgroup to other Eurasians, and so it is part of the definition of those groups.

Matt said...

Thank David. Yeah, I think you can see my thinking here. I was wondering if the D stats were boosted up by a differing relatedness to ENA for Kotias and Anatolia_Neolithic, so my idea was to take the pairs


As in theory all that should be driving the D(Chimp,Pop,Siberian,ENA) stats is the ANE fraction.

So for example

D(Chimp,Kotias,Itelman,Dai)-D(Chimp,Anatolia_Neolithic,Itelman,Dai)= -0.0051

meaning the Kotias stat is more negative than the Anatolia_Neolithic stat and presumably that reflects a greater affinity to the non-Dai related part of Itelman's ancestry (ANE).

Problem is this is kind of hard to actually interpret, as that doesn't really explain the significance or lack of, for the stat (no Z scores).

So, could you also run:

Anatolia_Neolithic Kotias Itelmen Ulchi
Anatolia_Neolithic Kotias Chukchi Ulchi
Anatolia_Neolithic Kotias Surui Ulchi
Anatolia_Neolithic Kotias Itelmen Dai
Anatolia_Neolithic Kotias Chukchi Dai
Anatolia_Neolithic Kotias Surui Dai
Anatolia_Neolithic Kotias Itelmen Atayal
Anatolia_Neolithic Kotias Chukchi Atayal
Anatolia_Neolithic Kotias Surui Atayal
Anatolia_Neolithic Kotias Itelmen Papuan
Anatolia_Neolithic Kotias Chukchi Papuan
Anatolia_Neolithic Kotias Surui Papuan
Anatolia_Neolithic Kotias Itelmen Australian
Anatolia_Neolithic Kotias Chukchi Australian
Anatolia_Neolithic Kotias Surui Australian

I was also thinking given those D-stats for the Australians and Papuans above, compared to the stats for the East Asians, how to test whether that was solely due to Denisovan influence, and it seems like these stats would show the difference:

Chimp Kotias Han Denisovan
Chimp Anatolia_Neolithic Han Denisovan
Chimp Kotias Atayal Denisovan
Chimp Anatolia_Neolithic Atayal Denisovan
Chimp Kotias Papuan Denisovan
Chimp Anatolia_Neolithic Papuan Denisovan
Chimp Kotias Australian Denisovan
Chimp Anatolia_Neolithic Australian Denisovan

By comparing the pairs like D(Chimp Ancient East Asian Denisovan) to D(Chimp Ancient Oceanian Denisovan) which should mask out the Denisovan influence.

again if poss.

Romulus said...

Maybe the CHG in ANE is some sort of Basal CHG/WHG (YHG IJ) that came along with the affinity that Mal'ta Boy shares with WHG? Along with the Venus figurines from Eastern Gravettian.

George Okromchedlishvili said...

There is no CHG in ANE. Otherwise it would be more distant from East Asians and closer to Africans than WHG. This is not the case

Davidski said...


Anatolia_Neolithic Kotias Itelmen Ulchi -0.0055 -2.447 508715
Anatolia_Neolithic Kotias Chukchi Ulchi -0.0042 -2.351 508715
Anatolia_Neolithic Kotias Surui Ulchi -0.0054 -1.555 508715

Anatolia_Neolithic Kotias Itelmen Dai -0.0065 -2.512 508715
Anatolia_Neolithic Kotias Chukchi Dai -0.0054 -2.369 508715
Anatolia_Neolithic Kotias Surui Dai -0.0066 -1.857 508715

Anatolia_Neolithic Kotias Itelmen Atayal -0.0065 -2.118 508715
Anatolia_Neolithic Kotias Chukchi Atayal -0.0053 -1.964 508715
Anatolia_Neolithic Kotias Surui Atayal -0.0065 -1.738 508715

Anatolia_Neolithic Kotias Itelmen Papuan -0.0106 -2.827 508715
Anatolia_Neolithic Kotias Chukchi Papuan -0.0096 -2.721 508715
Anatolia_Neolithic Kotias Surui Papuan -0.0107 -2.53 508715

Anatolia_Neolithic Kotias Itelmen Australian -0.0137 -3.399 508710
Anatolia_Neolithic Kotias Chukchi Australian -0.0126 -3.253 508710
Anatolia_Neolithic Kotias Surui Australian -0.0137 -2.976 508710

Chimp Kotias Han Denisovan -0.7249 -100 438370
Chimp Anatolia_Neolithic Han Denisovan -0.7152 -100 506351
Chimp Kotias Atayal Denisovan -0.7251 -100 438370
Chimp Anatolia_Neolithic Atayal Denisovan -0.7151 -100 506351
Chimp Kotias Papuan Denisovan -0.718 -100 438370
Chimp Anatolia_Neolithic Papuan Denisovan -0.7102 -100 506351
Chimp Kotias Australian Denisovan -0.7188 -100 438366
Chimp Anatolia_Neolithic Australian Denisovan -0.7121 -100 506346

Matt said...

Thanks. For those stats, they show the same thing (that the non-ENA portion of the Siberians and Native Americans specifically is more related to Kotias than Anatolia_Neolithic) but generally don't reach quite the Z = 3 / -3 significance level.

Also maybe potentially interesting that the most significant stat that reaches Z=-3 is

Anatolia_Neolithic Kotias Itelmen Australian -0.0137 -3.399 508710

while the least is

Anatolia_Neolithic Kotias Surui Ulchi -0.0054 -1.555 508715

The Ulchi IRC should be the closest population to Native Americans / Siberians minus ANE, so that stat should segregate out the ANE portion of the Native Americans the best, but is also the least significant. While the stat for Itelman vs Australian should be the one of this set which most separates off Northeast Asian related vs Basal Eurasian ancestry (but with the confound of ANE)...

(Denisovan ancestry in Australians and Papuans should affect the D(Chimp,Ancient,Siberian,East Asian/Oceanian) stats comparing an Oceanian stat vs the East Asian stat, but I don't think that should matter here, as the difference in the stats with Chimp would be due to Chimp-Denisovan sharing, while Denisovan wouldn't affect attraction to Kotias / Anatolia_Neolithic).

There's kind of a similar thing looking at the stats with Denisovan. Kotias has more sharing with the ENA groups along the path they don't share with Denisovan, and that's most increased compared to Anatolia_Neolithic for Atayal and least for Australian.

e.g. D(Chimp Kotias Han Denisovan) - D(Chimp Anatolia_Neolithic Han Denisovan) = -0.7251 - -0.7151 = -0.01
D(Chimp Kotias Australian Denisovan) - D(Chimp Anatolia_Neolithic Australian Denisovan) = -0.7188 - -0.7121 = -0.0067

Although again, no idea if that would translate into a statistically significant or non-sig difference.

So I guess I would take from all that, that the differences between Kotias and Anatolia_Neolithic in relation to Native Americans and Siberians is probably still mainly an effect ANE related ancestry plus also maybe slight and lower effects of less overall Basal type ancestry and some East Asian related drift pushing up the stat a little.

FrankN said...

@Rob: "can you elaborate on the Bactrian refuge ?

"Bactrian refuge" might not have been the most appropriate term. "South Central Transoxania", i.e. the Tajikistan side of the Amur Daya, the Foothills of the western Tian Shan and the Pamir, in interglacials also the Fergana Basin, would have been more precise (but also more clumsy). A general discussion of that area, pointing out areas where LGM finds might be expected, is found here:

None of the sites included in Tab 1 of that study covers the LGM, but Ogzi-Kichim (S. Tajikistan, 21-17 ky calBP) comes pretty close. Nearby Shugnoo isn't AMS dated, but palynologically dated to 25-20 kyBP. Interestingly, both sites are Mousterian (Neanderthals).

For other species, the refugium function of the area is documented here:

- Persian Walnut:

- Dice snake:

A bit more unspecific ("somewhere in Central Asia") are the following two (lynx and brown bear), but note that in both cases "Central Asia" excludes the Altai, and relates to the Pamir/ Tian Shan area:

I also tried to figure out something about pistachio (pistacia vera), for which earliest cultivation evidence stems from BA Usbekistan, but the species isn't particularly well researched. A refugium in the Zagros Mts. is documented, but it is unclear whether it relates to P.vera or the mediterranean P.atlanticus.
In any case, an area that served as likely glacial refuge for walnuts, lynx, bear, and warm-loving fish-eating snakes should definitely have been allowed human survival.

As to the Altai, this recent reconstruction of climatological and botanical conditions during the LGM might be of interest.

"The models projected LGM predominance of desertsteppe
across most of the non-glaciated area of the Russian Altai, probably associated with areas of typical steppe, tundra grasslands and some other habitat types, including forest patches in streamvalleys. (..) A different history was suggested for the precipitation-rich area in the northernmost Altai (north of Lake Teletskoye), where the LGM models suggested occurrence of patches of open forest of Larix sibirica and Pinus sibirica in forest-tundra and forest-steppe landscapes. These forests may have provided the LGM refugium for the temperate forest species that currently occur in this precipitation-rich area."

Note that the model just covers the Russian Altai, not the more southerly parts in Kazakhstan, China and Mongolia.

Rob said...


The study you cite confirms what I've read elsewhere - Central Asia was site-rich archaic homo species continuing from the MP, but not for AMH. In fact, the number of AMH sites in the UP are very low, and the region appears to only become well populated during the Mesolithic, and, of course, subsequent Neolithic

Chad Rohlfsen said...

Malta-Buret is not Gravettian related. There are many differences, including with the statues. They are very different, including one lacking faces and representations of breasts. Statues representing the feminine are not uncommon and do not represent "Gravettian" influence by default. Do feminine statues in Africa represent Gravettian influence? Of course not. Neither do they in Siberia.

Romulus said...


You're totally wrong.

Rob said...

Chad is correct. Only older studies thought Mal'ta is 'Gravettian'. Now, the Gravettian is seen to be only within Europe, and mal'ta is its own culture - Malta- Buret..

Davidski said...


Plus also maybe slight and lower effects of less overall Basal type ancestry.

Based on the formal stats and TreeMix analyses I've run, I think CHG is more basal than Anatolia Neolithic, and that's what is causing the problems, because the heavy basal is canceling out most of the ANE. The lower the basal in the reference, the lower the ANE signal in CHG.

Chimp MA1 Kotias Anatolia_Neolithic -0.0104 -2.039 367109
Chimp MA1 Satsurblia Anatolia_Neolithic -0.0061 -1.016 298843
Chimp MA1 Kotias LBK_EN -0.0055 -1.047 366152
Chimp MA1 Satsurblia LBK_EN -0.0016 -0.254 297959
Chimp MA1 Kotias Esperstedt_MN 0.0057 0.777 310282
Chimp MA1 Satsurblia Esperstedt_MN 0.0083 1.002 254241

Rob said...

Would it have anything to do with the fact that the CHG is based on much older samples than Anatol _farmer ; and that the latter has other components mixed in (apart from UHG / WHG, other unknowns )?

Aram said...

As far as I know Sardinians don't have ANE. But they have CHG. So how to explain this apparent contradiction?

Krefter said...


Can you run these? I think I'm making progress. These could provide absolute prove of EHG, CHG, and SW Asian ancestry in South Europe. I think it's the best D-stat strategy for finding admixture.

Remedello_BA Italian_Tuscan Kotais Italian_Bergamo
Remedello_BA Italian_Tuscan Karelia_HG Italian_Bergamo
Remedello_BA Italian_Tuscan Cypriot Italian_Bergamo

HungaryGamba_EN Greek Macedonian Karelia_HG
HungaryGamba_EN Greek Macedonian Kotias
HungaryGamba_EN Greek Macedonian Cypriot

Spain_MN Basque_French French_South Karelia_HG
Spain_MN Basque_French French_South Kotias
Spain_MN Basque_French French_South Cypriot

Spain_MN Spanish_Cantabria Spanish_Andalucia Karelia_HG
Spain_MN Spanish_Cantabria Spanish_Andalucia Kotias
Spain_MN Spanish_Cantabria Spanish_Andalucia Cypriot
Spanish_Aragon Spanish_Galicia French_South Morocco

Nirjhar007 said...

It is interesting that some have proposed a strong connection between Proto-Indo-European and Basque, yes Basque. There is a paper -

Interested people can check it out. I know some of them are technically weak but there are some impressive ones, I think Kristiina will also agree if she takes a look.

So as we know Basque may have come in the neolithic into Europe, this really gives an interesting curve ball , the connection with Sumerian is really strong, there is Uralic also, and Hurro-Urarian do show some notable levels of connection .

Davidski said...


Remedello_BA Italian_Tuscan Kotias Italian_Bergamo -0.0139 -3.831 323047
Remedello_BA Italian_Tuscan Karelia_HG Italian_Bergamo -0.0169 -4.446 348151
Remedello_BA Italian_Tuscan Cypriot Italian_Bergamo -0.0022 -1.468 370855

Iberia_MN Basque_French French_South Karelia_HG 0.0025 0.923 531901
Iberia_MN Basque_French French_South Kotias -0.0008 -0.299 478939
Iberia_MN Basque_French French_South Cypriot -0.0073 -5.677 554482

Iberia_MN Spanish_Cantabria Spanish_Andalucia Karelia_HG 0.0072 2.273 531901
Iberia_MN Spanish_Cantabria Spanish_Andalucia Kotias 0.0102 3.559 478939
Iberia_MN Spanish_Cantabria Spanish_Andalucia Cypriot 0.0009 0.52 554482

Spanish_Aragon Spanish_Galicia French_South Algerian 0.0095 7.966 594924

Hungary_EN Greek Macedonian Karelia_HG 0.0104 5.436 412623
Hungary_EN Greek Macedonian Kotias 0.0156 9.94 438563
Hungary_EN Greek Macedonian Cypriot -0.0028 -4.466 459207

Nirjhar007 said...

Well, the link is not alive, check his works here -

Krefter said...

Awesome!! Do you think those stats are prove of EHG and CHG in all of South Europe?

Gökhan said...

"Awesome!! Do you think those stats are prove of EHG and CHG in all of South Europe?"

Spread of CHG and EHG seems relatively independent as far as there are negative correlation between CHG and EHG admixture stats in Southern European and Balkan pops.

Davidski said...

This is very interesting, although hardly surprising. Eight of the Anatolian farmers help to produce a stronger ANE signal in Kotias than the rest.

Chimp Surui Anatolia_Neolithic2 Kotias 0.0148 3.359 505722
Chimp Itelmen Anatolia_Neolithic2 Kotias 0.0142 3.486 505722
Chimp MA1 Anatolia_Neolithic2 Kotias 0.0139 2.6 365166

Chimp Iberia_Mesolithic Anatolia_Neolithic2 Kotias -0.024 -4.981 438436
Chimp Loschbour Anatolia_Neolithic2 Kotias -0.0243 -4.775 431483
Chimp Bichon Anatolia_Neolithic2 Kotias -0.0177 -3.462 346338

This is where they cluster (red X).

I'd say the reason they help to produce a stronger ANE signal in Kotias is because they're more basal and less ANE. If so, what this means is that at least some Anatolian farmers already had CHG and even ANE admixture, but the differences between them are so subtle that it's not possible to confirm this directly with D-stats.

Btw, Krefter, I don't know. Need to think about it.

Ryan said...

"Another thing that I didn't mention is that Kotias is probably a little more basal than the Anatolian farmers, and yet they more or less show the same relationship to East Asians, which might mean that Kotias has some East Asian admixture too."

Perhaps this was mediated by ANE (though not MA-1)? Or ANE itself was already somewhat intermediate?

Or the alternative is just African admixture into Anatolian farmers, which would match the archaeological record. I'd note La-Brana is closer to East Asians than modern Europeans IIRC (or at least was closer to Ust-Ishm). African admixture would explain that too.

Alberto said...

I think that genetically these models work good, and at least with what we have today they're the best hypothesis. But has someone tried to fit it into reality? If WHG and ANE descend from Kostenki and split some 28-30 Kya, and assuming this happened in the northern parts of Eurasia (since Kostenki predates the split and he was already there), how do we end up with ancient Near Easterners being a 50/50 mix of WHG and Basal Eurasian? (Similar for CHG and ANE, but for this we have even less data to speculate).

One could hypothesize a big retreat from Europe to the Near East during the LGM, but do we have any evidence for such thing? And where did the WHG uniparental markers go? Or did WHG go from West Asia to Europe during the Gravettian? (But then where did the Basal Eurasian population come from to the Near East, and when? And again the problem with little haplogroup sharing between WHG and Anatolian farmers, if these latter ones are 50%+ WHG descended.)

Karl_K said...


"Those admixture results were based on an old sequence of La Brana 1. We now have a high coverage version of this sample, and it doesn't appear to have any special affinity to Southeast Asia or, as is often mentioned in the comments here, North Africa."

Matt said...

Davidski: Based on the formal stats and TreeMix analyses I've run, I think CHG is more basal than Anatolia Neolithic, and that's what is causing the problems, because the heavy basal is canceling out most of the ANE. The lower the basal in the reference, the lower the ANE signal in CHG.

I can see why that would be true (lower the BE, lower the MA1 signal), and that that's why the comparisons between AN and Kotias here are stronger than with Kotias and the WHG admixed EEF in Jones 2015.

But at the same time, for formal stats, how's that compatible with the stats:

Chimp Han Anatolia_Neolithic Kotias 0.0068 1.92 508715
Chimp Dai Anatolia_Neolithic Kotias 0.0068 1.882 508715
Chimp Atayal Anatolia_Neolithic Kotias 0.0069 1.73 508715
Chimp Papuan Anatolia_Neolithic Kotias 0.0028 0.711 508715
Chimp Australian Anatolia_Neolithic Kotias 0.0001 0.021 508710 ?

BE is usually found by shift away from ENA relative to the outgroup (in the absence of actual gene flow from ENA), and these show more shift towards ENA for Kotias than Anatolia_Neolithic (albeit not significantly).

(The ones for Papuan and Australian I also think are less significant because of Denisovan, hence the whole cross comparison of D(Chimp,X,East Asian/Oceanian,Denisovan) stats).

Ryan said...

Karl - I'm talking about the D stats in the Ust-Ishm paper.

Matt said...

Re: Krefter stats, I guess

Remedello_BA Italian_Tuscan Kotias Italian_Bergamo -0.0139 -3.831 323047

Remedello_BA Italian_Tuscan Karelia_HG Italian_Bergamo -0.0169 -4.446 348151

Remedello_BA Italian_Tuscan Cypriot Italian_Bergamo -0.0022 -1.468 370855

means something like

"The genetic drift that differs between Remedello_BA and Italian_Tuscan is similar to the genetic drift that differs between Italian Bergamo and Kotias and is similar to the genetic drift that differs betweens Italian Bergamo and Karelia HG.

The genetic drift that differs between Remedello_BA and Italian_Tuscan is not similar to the differences between Italian Bergamo and Cypriot"

Although, I guess if that is the interpretation then

Iberia_MN Basque_French French_South Karelia_HG 0.0025 0.923 531901
Iberia_MN Basque_French French_South Kotias -0.0008 -0.299 478939
Iberia_MN Basque_French French_South Cypriot -0.0073 -5.677 554482

means something like

"The genetic drift between Iberia_MN and Basque_French is not like between French_South and Karelia_HG, nor French_South and Kotias.

Instead it is like between French_South and Cypriot".

Davidski said...


As you say, the stats aren't significant, but they might reflect shared ancestry between Kotias and East Asians to the exclusion of Anatolian farmers.

If Kotias was less basal or Anatolian farmers more basal, the stats might well be significant.

By the way, I've got solid evidence of CHG admixture in at least some Anatolian farmers. I'll post it later today.

Alberto said...


Thanks, your reading of those Dstats is exactly what I've been thinking about the last couple of days to explain some weirdness in some stats with certain populations. Just before I managed to put it together and asked Tobus about it. Not sure if I got it right, but for what you wrote there it seems that maybe yes.


I asked you a question about Dstats on a mail you posted here once, I hope you don't mind. Not sure if you use that mail, though, it was some tobus73... If you could check it I'd really appreciate. Thanks!