The recent Jones et al. palaeogenomics paper focusing on Caucasus hunter-gatherers (CHG) has this to say about the Indo-European and Indo-Aryan expansions:
CHG left their imprint on modern populations from the Caucasus and also central and south Asia possibly marking the arrival of Indo-Aryan languages.
...
It has been proposed that modern Indians are a mixture of two ancestral components, an Ancestral North Indian component related to modern West Eurasians and an Ancestral South Indian component related more distantly to the Onge [25]; here Kotias proves the majority best surrogate for the former [28,29] (Supplementary Table 10). It is estimated that this admixture in the ancestors of Indian populations occurred relatively recently, 1,900–4,200 years BP, and is possibly linked with migrations introducing Indo-European languages and Vedic religion to the region (28).
...
Finally, we found that CHG ancestry was also carried east to become a major contributor to the Ancestral North Indian component found in the Indian subcontinent. Exactly when the eastwards movement occurred is unknown, but it likely included migration around the same time as their contribution to the western European gene pool and may be linked with the spread of Indo-European languages. However, earlier movements associated with other developments such as that of cereal farming and herding are also plausible.
To their credit, in that last quote the authors leave open the possibility that CHG arrived in South Asia in multiple waves and with a variety of groups, including Neolithic farmers. Nevertheless, I'd say their comments are still confusing and perhaps also incredibly naive, because essentially they appear to be hoping that in CHG they've identified the Proto-Indo-European and Proto-Indo-Aryan genetic component.
Indeed, a lot of people actually believe that the overwhelming part of the West Eurasian admixture in South Asia should be attributed to the Indo-Aryans. But that's just stupid.
After all, many Dravidian groups that in all likelihood never spoke Indo-Aryan languages carry significant ratios of West Eurasian ancestry. Some of this influence can be explained by admixture with Indo-Aryans, but uniparental markers suggest that much of it was brought from West Asia by the Proto-Dravidians (see here).
Below is the aforementioned Supplementary Table 10. Note that two of the Indian populations that are best modeled with D-stats as mixtures of Kotias (one of the two CHG genomes) and Onge are Dravidian speakers (Mala and Vishwabrahmin or Viskwakarma, a Malayali community). Another three are Indo-Aryans (GujaratiC, GujaratiD and Lodhi), but with high levels of Ancestral South Indian (ASI) admixture, which suggests their ancestors might have been language shifters.
On the other hand, the three populations that are best modeled as Afanasievo (a pastoralist group from the Early Bronze Age steppe) and Onge are all Indo-Aryans (GujaratiA, GujaratiB and Tiwari).
But like I say, South Asia is a complex melting pot of Indo-Aryans, Dravidians, and several other linguistic groups, so a more comprehensive analysis than a comparison of a few D-stats is needed to unravel the origins of its people in a meaningful way.
By the way, Jones et al. also argue that CHG is basically an offshoot of the so called Basal Eurasian clade, which was first described in Lazaridis et al. 2014. I'm highly skeptical of this claim, and I might check it out after I get my hands on the CHG genomes.
Citation...
Jones, E. R. et al. Upper palaeolithic genomes reveal deep roots of modern eurasians. Nat. Commun. 6:8912 doi: 10.1038/ncomms9912 (2015).
See also...
The "fourth strand" of European ancestry came from the Caucasus
What's important about groups like Tiwari, Gujarati A and Gujarati B is not that they are Indo-Aryan speakers but rather that they are members of upper castes...
ReplyDeleteDravidians were Hunter-Gatherer groups of the South, their origin is not sure.
ReplyDeleteIE's in India and SC Asia have a deeper presence, at least from Chalcolithic,
that Moorjani et al paper they cited actually reflects the shift/Migration of ANI dominant IVC/SSC people towards ASI dominant South and East, nothing else.
"many Dravidian groups that in all likelihood never spoke Indo-Aryan languages carry significant ratios of West Eurasian ancestry. "
ReplyDeleteOne of the paradoxes of linguistics and genetics in South Asia is that Dravidian genetics have some very ancient components, but the modern Dravidian language family is very young and realistically, younger than the Indo-European languages in South Asia from a linguistic perspective.
I suspect that what happened was that Dravidian languages were once predominant in Southern India and maybe even a Southeastern corner of Pakistan, then was almost completely overrun by Indo-Aryans linguistically and religiously except one little remaining Dravidian community near the middle of the East Coast of the Deccan Peninsula. This is why all Indian populations have some ANI (especially among Brahmins) and why the Hindu religion is found in Dravidian areas.
But, somehow, that Dravidian community managed a counterattack ad reclaimed most of the formerly Dravidian territory. However, all of the Dravidian languages spoken after the "counterattack" were derived from the sole surviving dialect or closely related set of local dialects of that region, and was there was a second language shift back to Dravidian languages, the Hindu religion that the Indo-Aryans had brought survived the "counterattack." The "counterattack" probably had a very modest demic effect, or involved populations that were genetically similar to start with so that the counterattacking Dravidian superstrate had more or less the same genetic composition as the "reconquered" people.
The European analogy in the historic era would be the relatively short lived Moorish conquest of much of Spain that was gradually pushed back and eventually expelled (after leaving some genetic and linguistic marks) a few centuries later. Romance languages were spoken in Iberia before and after the Moorish era (and even during to a lesser extent), but there were probably some dialects of Southern Spanish where Moorish occupation was longest lived that went extinct in the meantime. Places that were subsequently totally Spanish speaking still had some genetic trace of the Moorish invasion even after the invaders language and culture had almost completely vanished.
The conjecture I make above also explains why Southern Indian populations have older inferred admixture dates with ANI than Northern Indian populations. The first wave covers all of India. A secondary wave covers only Northern India and doesn't penetrate into the reconquered Dravidian territory. And, since inferred admixture dates heavily bias towards the last wave, the North paradoxically has the more recent inferred admixture date even though it is obvious that Indo-Aryan migration came from the Steppe to the North to the South of India.
ReplyDeleteThe problem with autosomal DNA is that similar component ratios or similar positions on PCA plots don't necessarily imply close kinship. Croats have identical EEF/WHG/Yamna ratios to Frenchmen and Englishmen, but they're not exactly genetical brothers.
ReplyDeleteCHG is certainly best proxy for teal, but the IE question is far from solved.
Croatians have very similar prehistoric admixture ratios to English and French because they basically descend from the same prehistoric populations, like all North-Central Euros, so I don't understand your point?
ReplyDelete@Davidski
ReplyDeleteOn a PCA plot, it would imply that Croats, Frenchmen, Englishmen, are more closely related to each other than to other European populations, which is just not true.
The problem here is that 1)There must be more specific strains of these components that we're not taking into account (not EEF/HG/ANE are all the same) 2) We don't know who introduced these components. We can model IE as EHG + CHG, but we don't know how this ratio was achieved and if it was a direct mixing of those two populations. All is speculation at this point. That's why IE question has major issues to resolve.
Different PCA focus on different components, sometimes really old components and other times much younger ones. It depends on the populations that are being tested.
ReplyDeleteCroatians cluster with French and English in PCA that focus on prehistoric components because, duh, they have very similar ratios of these components.
The fact that you don't understand this isn't an argument, so you have no argument.
The copper to Bronze Age transition was a watershed in Eurasian history. Communities were becoming mobile more advanced more competitive. This would have facilitated a lot of language spreads – Not only Indo-European – so it is a little bit difficult to trace any linear relationship between genetics and language.
ReplyDeleteBecause quite simply a lot of people moving as were a lot of languages m and it's all a bit of a kaleidoscope. However I think evidence from Greece and India will certainly help confirm or deny the current leading hypotheses
There's solid evidence now that R1a-Z93 moved from the steppe to India along with the Indo-Iranians around 1800 BC.
ReplyDeleteSee that's why lots of Indian scholars were so desperate about making R1a native to India; they knew they couldn't lose this battle, but what they didn't realize was that they lost it back in 1800 BC.
@Davidski
ReplyDeleteThe argument is that there is a lot of arbitrariness with these components, because the researchers design them, along with PCA plots.
On a plot, just because a point B is north-east of A, and you add a bit of a north-eastern population C to A, it does not mean population B was created from A + C. You just MODELLED them that way.
So just because adding CHG to EHG pushes towards Yamna on a plot, it doesn't mean Yamna is CHG + EHG.
The usual PCA of West Eurasia, which shows modern North-Central-East Euros as Neolithic/Yamnaya mixtures and Yamnaya as an EHG/CHG mixture, simply reflects the output from formal statistics and uniparental marker data, so I don't know why you're even bothering with PCA as an argument?
ReplyDeleteThere's solid evidence now that R1a-Z93 moved from the steppe to India along with the Indo-Iranians around 1800 BC.
ReplyDeleteSee that's why lots of Indian scholars were so desperate about making R1a native to India; they knew they couldn't lose this battle, but what they didn't realize was that they lost it back in 1800 BC.
Very Funny.
What is important now is to just take some samples from pre-2000 BC period of India and SC Asia, and to prove what was there and what was not!
Many IE communities moving south also adopted Dravidian tongue, that is also something to consider.
IMHO its not possible that Z-93 came to India around 1500 BC, it was there from before. It will be cleared soon.
I think evidence from Greece and India will certainly help confirm or deny the current leading hypotheses
ReplyDeleteAlways.
I'm pretty sure now that those skeletons from Rakhigarhi will turn out Y-HG J.
ReplyDeleteThey'll also show a lot of CHG and ASI ancestry, if genome-wide data is tested.
@Nomen
ReplyDelete"So just because adding CHG to EHG pushes towards Yamna on a plot, it doesn't mean Yamna is CHG + EHG."
These plots are not just dots on a screen. They represent actual sets of SNP differences between populations. If one later population has a mix of the SNPs from two seperate earlier populations, then it must have arisen through admixture between populations closely related to those earlier populations.
The only other possibility is that the earlier populations were themselves mixtures, and a seperate set of admixture events occurred at almost the same time but in a different order. The differences in the final populations would still be affected by seperate drift effects.
make above also explains why Southern Indian populations have older inferred admixture dates with ANI than Northern Indian populations.
ReplyDeleteWe can imagine the ANI type element started to Enter S Asia at least from Mesolithic period, for the IE question, It appears they came from S of Caspian to India from the beginnings of Chalcolithic (4500-4000 BC), AFAIS there weren't any migration from Steppe to India around that period David and You refer, because simply there is no data to agree with it, only to contradict, and here we are talking about a huge genetic and cultural impact, this is why some have taken shelter with the ''elite dominance'' model, but the problem is that only at best it creates hybrid types of languages and fades away with time, like happened in case of Mitanni.
There are tons of other reasons that can be found from Say Archaeology, Anthropology ,Ancient Texts etc etc.
I'm pretty sure now that those skeletons from Rakhigarhi will turn out Y-HG J
ReplyDeleteor in all likelihood R1a-Z94 or R1a-M417, in any case the fate of the Steppe Theory will be decided, that's the beauty the genetics have provided.
Npmen has a very solid point in recognizing that populations with similar percentages of a small number of admixture components are not always similar in percentages because they are closely related, or equivalently are not always closely related simply because they are close together on a PCA chart.
ReplyDeleteFor example, my children (who are half-Korean and half-Northern European) show up very close to Uyghurs in admixture percentages or a PCA chart, since Uyghurs are about 50% East Asian with an Altaic tendency and about 50% West Eurasian with a leaning that is more Northern than Southern European. But, my children aren't remotely related to Uyghurs on either the East Asian side or the West Eurasian side at all at any time since the mid-Holocene.
This just shows what happens when an admixture that could never have happened in a world with clinal genetic variation or distances that were realistically traversable before the 19th century isn't well explained by a model that implicitly provides the most informative data in cases where there is clinal genetic variation between geographically adjacent locations.
Since most cases we look at are a natural fit to the model that admixture percentages or PCAs implicitly map genomes onto, we get into the habit of thinking of samples with similar admixture percentages or PCA mappings as being closely related without consciously thinking about the circumstances where that assumption doesn't hold true.
@andrew
ReplyDeleteYour point is valid. However, no one is only using 1 or 2 PCAs to determine admixture between populations, as Nomen was suggesting.
He doesn't have a point, because he's ignoring the fact that PCA are just for illustration purposes, and if they're well designed they will reflect the reality shown in the formal statistics. Admixture output is also largely for illustration.
ReplyDeleteAnd there's no way that your daughter would cluster with Uyghurs on an intra-Eurasian plot with enough of the right samples.
This comment has been removed by the author.
ReplyDeleteJust to expand on the thought slightly, if, as Nomen suggests, you increase the number of admixture components, from the small number (3-4) necessary to explain most of the variation in most European populations, to a large number, perhaps 6-8 from the same data set, the false friends in terms of admixture percentages will tend to resolve themselves.
ReplyDeleteAt this higher resolution, for example, my children start looking like a 50% Korean-Japanese blend and a 50% Finnish-Northern European blend which is not at all what you would see if you looked more closely at the precise ancestral East Asian and West Eurasian components within a Uyghur individual, although the inherently reductionist nature of a PCA plot makes it harder to observe this kind of statistical artifact using that tool.
Nomen is suggesting that the same is probably true in the English-Croatian case -- if you increase the number of ancestral populations you are modeling from the three usually used these days for Europeans, the Croatian components probably resolve into different ancestral populations than the English ones do, but the similar subcomponents in each population get lumped together into the same ancestral population when you reduce the resolution to three ancestral populations.
Of course, the similarity could be due to a long distance migration of people from Croatia to the British Isles or visa versa. Indeed, the Old European Culture Blog http://oldeuropeanculture.blogspot.com/ spends a great deal of time on posts developing suggestive evidence that hints at just that kind of relationship between Balkan peoples (Serbians more often than Croatians) and Celts (Irish more often than English).
I'm agnostic about which is more likely to be true. The false friend hypothesis is certainly more parsimonious and favored by Occam's Razor in the absence of the kind of suggestive evidence found at the Old European Culture Blog. But, sometimes history didn't chose the most obvious route and other long distance migrations (e.g. Madagascar, Tarim Basin, Gypsies) are not unknown to history, although they aren't very common either.
@ryuk,
ReplyDeleteThere's no reason to think Teal didn't exist. Just because it's an ADMIXTURE component doesn't mean there wasn't an ancient population that is the source of the Teal signal we see. EEF is the Mediterranean component that ADMIXTURE picks up just as CHG is the Teal component.
In ADMIXTURE CHG literally is Teal. They score 100% in the what is called the teal component. No Modern pops score much above 50% in Teal. The ancient samples, including CHG, dominate the components.
Formal stats prove that CHG is the best proxy for Yamnaya's non-EHG ancestors(Teal) and for South Asian's West Eurasian ancestors(ASI).
There's also no good reaosn to think Yamnaya's Near Eastern ancestors were MA1+CHG or that CHG lacked ANE. MA1 is closer to N. Euros than to CHG just as MA1 is closer to North Europeans than to Caucasus people. CHG is slightly closer to WHG than to MA1 just as Caucasus people are. I've looked through all the evidence the authors put forward to say CHG had been isolated from ANE admixture, and none of it is good.
We already have ADMIXTURE results for the 13,300 CHG. He scores 23% ANE and 64% ENF in ANE K7. It's obvious he had ANE.
@Krefter
ReplyDelete"We already have ADMIXTURE results for the 13,300 CHG. He scores 23% ANE and 64% ENF in ANE K7. It's obvious he had ANE."
Where? I didn't see them yet. But anyway, running these samples in K7 or K8 of course will show them as a mix of ENF and ANE. That's expected. But it doesn't mean they had ANE. It just means that the best way to model them with those preconceived components is as a mix of ENF and ANE.
What I'd still like to investigate is if for example the Kalash people, who seem to lack any Anatolian-like ENF (and therefor they can't have much if any Sintashta admixture) are a pure CHG + ASI mix, or if they are CHG + ANE + ASI. I think the latter might be true, though we need to test it and see.
As you said the other day, R2 in India does suggest that ANE and R1 were around S-C Asia, so maybe the base of the West Eurasian part of S-C Asians is not just CHG, but a mix of CHG and ANE (that would look like Afanasievo, but with less WHG and more ANE, which might fit the bill of what we see in Kalash). I can't wait for David to process the genomes and start running some tests.
This comment has been removed by the author.
ReplyDelete@Ryu
ReplyDeleteYes, there is some amount of ENF/EEF in S-C Asia for sure. But it seems that the Kalash don't have it (in this admixture run they seem to get a bit of the European_HG component, but none of the EEF, in accordance with other tests).
So let's see when David creates a calculator with CHG and ANE how the Kalash (and surrounding populations) score. In the paper the only stats that can give us a hint about the real structure of S-C Asian populations are the ones showing GujaratiA/B as closer to Afanasievo than to CHG, which suggests that there might be ANE ancestry in ANI.
The current calculators can't explain CHG unless it's also assumed that they have something like 3-4% West African and 10% WHG on top of Near Eastern and ANE. As others have said there are a number of formal tests that can be performed when the sample is in public hands, and those will very likely tell more.
ReplyDeleteAndrew, formal tests are the way to approach the ancestry component issue. Razib Khan did a number of TreeMix runs and Uygurs and Hazara were resolved as basal East Eurasians with a considerable admixture edge from North Europe. Obviously this is catching something old unless South-Central Asians were Sintashta-like even in the 1000's which I think is quite implausible. The runs looked like this: http://www.unzcloud.com/wp-content/uploads/2015/08/KalashOut.71.jpg
A modern Korean + NW-European mix wouldn't show up like those Uygurs/Hazaras on TreeMix; formal tests can show the differentiation rather simply while PCA and ADMIXTURE at low K's don't, because they are better used as visualizations and require a very specific sampling to show something formal tests would show easier.
To return to the English-Croatian ancestry differentiation, D-stats and f4-stats as well as simple IBS can be used to check the estimates of Haak et al, if for instance something with a more divergent estimate appears to be closer to English than Croatians are, then there maybe something more complex going on. In any case, even if those current numbers are contradicted a bit it won't be too big of a deal. They're expected to improve as more ancient DNA is analyzed.
This comment has been removed by the author.
ReplyDeleteDavid, See that's why lots of Indian scholars were so desperate about making R1a native to India; they knew they couldn't lose this battle, but what they didn't realize was that they lost it back in 1800 BC.
ReplyDeleteEvery once in a while you throw out these poetic little nuggets. Haha
I think what is most interesting about CHG is just how distinct they are from EN. I never would have guessed a separation that took place ~25kya.
@RK
ReplyDeleteYes, the Kalash behaviour at K20 is interesting and maybe another hint about the Q population not being just pure CHG from the Caucasus. But it could also be an artifact of Admixture, so let's wait for some good tests to find out some better answers to these questions.
@ ryukendo
ReplyDeleteThere was the expansion of the Kura-Araxes culture, possibly associated with the Hurrians. The Hurrian-Urartian language family shouldn't be forgotten, although it got extinct long ago. It occupied a central position between Syria and the Caucasus. And if at a later stage there were IE intrusions from north of the Caucasus which went into Anatolia, these probably admixed with the CHG-related locals and carried this admixture further westwards.
Okunevo_EBA (Allentoft et al.) was almost in eastern central Asia and had plenty of ANE, independent of Afanasievo and Andronovo admixture, i.e. indigenous ANE, and a relatively low teal component. Mesolithic central Asians might have been similar, with strong ANE and no CHG-like admixture. South central Asia is again another matter, since a connection with the Caucasus seems possible.
ReplyDeleteBut the South Caspian probably was never an ANE hotspot and the ANE in the Caucasus seems to be mostly from the North, virtually all, if CHG wasn't ANE related.
ReplyDelete@RK
ReplyDeleteThat graph was actually based on David's older tests. I don't know if anyone has plotted Loschbour and Ust-Ishim, but I found some D-stats in form of D(X, Loschbour; U-I, San)
Dai -0.003 -0.33
Evenki -0.009 -0.93
Even -0.011 -1.14
Yakut -0.016 -1.66
Sherpa -0.016 -1.66
S. Indian -0.016 -1.63
Buryat -0.017 -1.65
Komi (Objachevo) -0.019 -2.02
Han -0.020 -2.25
Veps -0.021 -2.11
Kalmyk -0.021 -2.20
Andean -0.021 -2.12
Nenets -0.022 -2.25
Russian (Mezen) -0.022 -2.40
French -0.023 -2.41
Khanty -0.024 -2.53
Karelian -0.025 -2.65
Altayan -0.029 -3.09
S. Indian -0.033 -3.35
Komi (Izhma) -0.042 -4.35
Belarusian -0.043 -4.57
Sardinian -0.043 -4.42
Mansi -0.046 -4.90
Russian (Andreapol) -0.054 -6.23
I'm not 100% sure, but these are the same modern samples the study (that recent one about West Siberian population history) used in its high-coverage Treemix analysis, so they could be 40x.
Would you be able to run an analysis of South Indian castes (including Brahmins, some of whom may have been IE shifters to DR), or at least a wide variety of North Indian ones? Proportions and a potential admixture generational signature could show whether or not CHG arrived relatively early.
ReplyDeleteChad said:
ReplyDelete„looking at some f3 stats, I find it interesting that the Baden sample shows a more significant fstat with the Caucasus samples than Beaker and Hungary BA“
Good point. That was consistently overlooked by ADMIXTURE runs, as Baden/CO1 usually showed no trace of a teal-like component. Presumably because it had just CHG affinity with no ANE! But these f3 stats seem to confirm the claims made by archaeologists and craniometrical analysts alike, that Baden had some additional Anatolian input, relating it with eastern Anatolia.
@Krefter
ReplyDelete"showed no trace of a teal-like component. Presumably because it had just CHG affinity with no ANE!"
"Teal" is not the same as CHG.
LOL, I'm not Krefter. :D But I tend to second that.
ReplyDeleteMaybe even the Late Neolithic farmer from Kumtepe had CHG affinity. That would reconcile his purported "new" Caucasus affinity with the apparent lack of ANE affinity. But since he's dated to 4700 BC this would mean that the CHG affinity spread long before the Kura Araxes culture.
ReplyDelete@Karl_K
ReplyDeleteActually what the paper implies quite clearly is that CHG = Teal.
What I *suspect* is that the population that entered the steppe was not just CHG, but rather CHG + ANE, and that "teal" (as in "the component that peaks in the Kalash") might be that: CHG + ANE (and not just CHG). But this is just a speculation at this point and I'd like to see some tests to check if it's correct or not. Until then, officially, CHG = teal.
@Karl, of course they didn't mention the word "teal" in the paper, because that's a stupid word. But look at the admixture and their comments: Yamnaya is 50% EHG and 50% CHG. It's not me who's insinuating. I'm actually not totally convinced about those conclusions and would like to see other tests, but until then this is what the paper says and it might be perfectly correct (or not, we'll see).
ReplyDeleteThis comment has been removed by the author.
ReplyDeleteTed Kandell from OGF found the Kotias sample to be J2a-Y12378. "The sample KK1 from Kotias Klde, Georgia, 9955-9589 ybp (cal BP+60) is
ReplyDeleteY12630+ Y12628+ and therefore J-Y12378. However, he's also SK1314/Y12595-, Y12624-, CTS3089-, and therefore a "partial" J-Y12378.
http://www.open-genomes.org/…/Jones%20(2015)%20…/KK1/images/
This is a preliminary result from just 1 out of 5 sequences for KK1, but it seems to be consistent." CEPH also has an Adyghe who is likely from this cluster and ChrisR has several Georgians, a Chechen and an Adyghe from this group.
@Matt
ReplyDeleteRe: the WHG-EEF-ANE model, now I remembered some Dstats that I asked David to run some time ago and turned out very weird. It led me to think that there was very little ENF in S-C Asia and even less in India, in spite of the high figures showed by K8.
So, regardless of whether Kalash will be pure CHG + ASI or if they will be CHG + ANE + ASI, yes there will be a drop in ANE across the board, but probably bigger in ENF. And this could have major implications for understanding the spread of farming, for example. Populations from India are going to have very little ENF, which could mean that Neolithic there was a local development and not the product of demic diffusion from the Near East. The Anatolian Hypothesis will also take a hit from this, if that still matters to anyone.
It will be interesting to see the impact in Europe too. For example, will East Baltics have much ENF left after we account the CHG component? Will Greeks still have similar levels of ANE as Northern Europeans or will it drop even below Spanish levels?
@ryu
ReplyDeleteBasic question : enf is late and a partial descendant of chg. For symmetry
Shouldn't we also ask what percent of enf is chg?
@RK
ReplyDeleteYes, I agree. I think that there will still be ANE in S-C Asia. Though it's not going to be anywhere near the current K8 values. This changes are really going to give us some new perspective about the populations genetic structure and their relationships. Really interesting.
I think the HG in CHG is a crown west Eurasian. Not being closer to WHG or ANE, but the ancestor to both.
ReplyDeleteAn interesting detail from a previous paper about these samples:
ReplyDelete"A series of living surfaces with combustion features, faunal remains, stone and bone tools, and ornaments provide new information about human occupations in this region (a) prior to the Last Glacial Maximum (LGM) at 25.5–24.4 ka cal. BP and (b) after the LGM at 17.9–16.2 ka cal. BP. The latter provides new evidence in the southern Caucasus for human occupation immediately after the LGM. The results of the campaigns in Satsurblia and Dzudzuana suggest that at present the most plausible scenario is one of a hiatus in the occupation of this region during the LGM (between 24.4–17.9 ka cal. BP)."
So it seems that these CHG left the area during the LGM (during some 6.000 years) to find refuge somewhere else. That probably means the South Caspian shores, which was the best place around during the LGM, though it could be further south east too. In any case it provides evidence that they were not isolated in the Caucasus, but rather they entered (or re-entered) after the LGM.
@Alberto
ReplyDelete"So it seems that these CHG left the area during the LGM (during some 6.000 years) to find refuge somewhere else."
Why do you assume that the 'same people' (CHG) were there before and after this 6,000 years? There is no DNA from the earlier time point, and 6,000 years is a very long time.
I'm not assuming anything. Maybe indeed they were different people altogether. The point is that they were not an isolated population in the Caucasus since before the LGM. The samples we have apparently entered the Caucasus only after the LGM from somewhere else (South Caspian looks like most likely, but we don't really know). Whether they were the same previous inhabitants re-entering the area or new ones we don't know either, but it's not the important point.
ReplyDeleteSounds good Alberto.
ReplyDeleteIt looks like the history of J2 will be just as interesting as R1.
@Alberto
ReplyDeleteAbandonment of a few cave sites does not mean the whole region was abandoned. The Colchian lowlands were after all one of the nicer LGM refugia together with the Hyrcanian lowlands - and ecologically similar, so it isnt clear to me why humans would abandon Georgia and not the South Caspian. Though maybe they did, who knows.
I am curious how well CHG agrees with the old Caucasus calculator component. There may have been have literal Caucasus and Gedrosia populations.
It looks like Karelia forager was J1a (http://j2-m172.info/links/scientific-papers/), so he is in the same line with Kotias. Moreover, mtDNA H2b was found in Oleni Ostrov. It is rare in Europe today but found in the Near East. Karelia forager lived c. 5000 BC which means that by that time there was a trail of CHG extending from Caucasus to Karelia. There is still J2 in Eastern Europe: Southern Russia 3%, Central Russia 2.4%, Northern Russia 1.6%; J*: Mordvinians 17%, Chuvashes 16%, Tuymazinsky Tatars 10%, Kazan Tatars 15%, Burzyansky Bashkirs 3.4%. It looks like Slavs are heavily R1a but Tatars, Mordvines and Chuvashes have incorporated a lot of J2. Part of the J2 above belongs to haplotypes other than J2a, and, indeed, J2b may not have developed in Caucasus. The only ancient J2b seems to be J2b2a in Late Bronze Age Armenia. In any case, J2b has a weird distribution: it is frequent in Eastern Mediterranean and in Central Asia and India. In the west, Kosovar Albanians have a 17% frequency of J2b-M241. Also Kola Saami have J2b.
ReplyDeleteIranians have a high frequency of J2a but they mostly lack J2b. Takhar Tadjiks and Uzbeks, as well as Burusho, Kalas and Sindhi have high frequencies of J2 but it is mostly J2a, and the share of J2b is smaller. Interestingly, one of the highest frequencies of J2 in the world is in Uygurs: 17/50 J2. Afanasievo may therefore yield yDNA J.
Now we know that Indians have CHG, and it is interesting to check their J frequencies. J2a frequencies are quite similar in Dravidian and IE groups but the Dravidian J2a frequency is a bit higher (2.35-13.6% and 2 - 9.3%, respectively). J2b frequencies are clearly higher in Dravidians (3.4-19% and 2-8.3%). J2(xM47, M68) is clearly more frequent in Tamil Nadu foragers than J2a1 and J2a3, and it looks like Pulayar J2 (xM47, M68) frequency is even 16%. I think that J2b/J2a is somehow related to Dravidian languages, and the really ancient forager languages in India have died out.
It would be very interesting to get more ancient J2b yDNA. Maykop should yield some yDNA J. I do not think that Yamnaya men went to Caucasus to get a wife. Since the Mesolithic, there were probably foragers with yDNA J2a and probably J2b in Russia well before Yamnaya, but for one reason or another, it was R1b men and not J2 men who developed Yamnaya pastoralism on the steppe.
@Capra
ReplyDeleteYes, maybe, I don't really know. They do state "the most plausible scenario is one of a hiatus in the occupation of this region during the LGM", but they could be wrong. I recently read a paper about climate simulations during the LGM and it did point to the South Caspian as a refugium (in the study for warm-loving summer-green trees, but I guess that warm and humid enough for summer trees is what humans looked for too):
http://www.researchgate.net/publication/50276032_A_comparison_of_climate_simulations_for_the_Last_Glacial_Maximum_with_three_different_versions_of_the_ECHAM_model_and_implications_for_summer-green_tree_refugia
"The clear message especially from the T106EH5 simulation is that warm-loving summer-green trees could have survived mainly in Spain but also in Greece in agreement with findings of pollen or charcoal during the LGM. Southern Italy is also suggested by the models as a possible refugium for warm-loving summer-green trees, but no reliable sites with palaeo-data were available to validate it. The importance of the southern coasts of the Caspian and Black Seas for the survival during the LGM already shown by LA2007 is confirmed also when using the more advanced model T106EH5."
Anyway, these are small details. The great news is they found these genomes that are going to improve our models very significantly, so let's see what we can learn from them once more tests start coming.
CORRECTION!: Karelia forager was J2a.
ReplyDelete@Alberto,
ReplyDeleteYes, the Leroy and Arpe (2007) paper is an well-done study of arboreal refugia and the Pontic-Caspian region does stand out as a likely refugium. I now think that J2a may have expanded from there southward to Iran/Iraq.
@ Ryu
ReplyDeleteRe: the differences in Afanasievo vs Malta and EHG vs Malta stats, intriguing, yet I find it hard to think through all the dimensions, and will wait to see what Mal'ta and EHG look like graphed against CHG. It may all well fall into line once the high CHG affinity of South Asians is present. I'm pretty unsure at the moment any additional MA1 ancestry than can be explained outside a pathway through extant Siberians or EHG is going to be necessary for anyone, or through Yamnaya and Yamnaya descendents (Yamnaya plus EEF, etc.) for South Asia.
I think once CHG is got by David, it might be good to ask if David can run a good large batch of D(Chimp,Ancient,Ju Hoan North,Pop) stats for a batch of the ancients, and I'm thinking these outgroups will make also help make clear what using BedouinB may have obscured.
Re: outgroup D statistics with K-14 vs others, I don't have those, but I did graph up the K-14 vs WHG for f3 based on Allentoft back in the day and ran it again today.
http://i.imgur.com/voFAB0D.png
http://i.imgur.com/XjcyQ7b.png
Also, MA1 vs K14
http://i.imgur.com/pG4z9cC.png
http://i.imgur.com/riEUecg.png
More or less basically K14 acts as MA1 would when plotted against WHG, and as MA1 would against WHG. K14 is defintely relating to the MA1-WHG clade.
But at the same time, bear in mind that in these WHG / MA1 vs K14 plots, the main dimension is 99% of the variance! So the actual real terms additional shared drift is very low with K14 beyond what is predicted by purely using MA1 or WHG as a measure. Yet this is enough to tease out a MA1 / WHG related dimension in the absence of either.
Plot all three together and...
http://i.imgur.com/ZUVKiLk.png
http://i.imgur.com/fmhO6em.png
The independent K14 dimension (dimension 3) actually seems an ENA dimension. There is nothing really CHG or EEF about it.
This (that residual relatedness to K14 after WHG and MA1 are accounted for is actually a measure of ENA) actually sort of underscores to me that, although K14 is no closer to ENA than EEF is (D stats comparing their ENA shift evaluate to 0), it doesn't have a great deal to do with EEF except what is shared through the WHG pathway (or with CHG, I think).
I would probably think of K14 at the moment as like a pre-WHG / ANE divergence (or a side clade to that divergence), that didn't get something that WHG and ANE both got that makes them closer to East Eurasians (whether this is a WHG+ANE->East Eurasians pathway or something else).
This seems simpler to me than the idea that both EEF, CHG and K14 all happened to get a similar Basal Eurasian fraction into them (despite not being very related), and at the same time, this didn't make K14 further away from Ust Ishim than WHG is, for some reason. But we'll see.
@ Nomen Cognomen's comment (btw, a username made from imperial roman family names?), kinship is... hard to define.
ReplyDeleteIf you're interested in kinship through the male or female line then you can look at y-dna and mtdna, but that is only kinship through the male or female line, not autosomal relatedness, which is another sort of kinship or sorts.
Autosomal relatedness of a autosomally European person with R1b y dna and H mtdna could be low, for instance, if the other person is an African African with high African ancestry, and who has R1b from his father and H through his mother. With massive expansions in the Bronze Age (particularly of y chromosomal haplogroups), this can also be the case for whole populations, relatively speaking.
Still, there's certainly something to the idea that populations with similar levels of Yamnaya, EEF and WHG as estimated by Haak are not necessarily the closest.
I mean, the mixes proposed by qpAdm, of ancestral components must be close to reality, but if we look at the shared drifts and qpAdm proportions, they do seem to imply relationships like English closer to Czech / Croat than Orcadian (which has quite different levels of WHG ancestry)... which doesn't seem to quite actually hold based on modern PCA positions and ADMIXTURE.
So either there are younger divergences and drift which is very important (which seems plausible, but a hard case for me, since there's not that much time for this to happen in) or there is some challenge in qpAdm and shared drifts being able to detect exactly the right admixtures with ancients.
Allowing CHG, EHG, WHG and Anatolia_EN to vary freely in models may fix some of this though.
Particularly, I think it's possible that Bronze Age Hungary, Bell Beaker, and Northwest Europe is relatively rich in CHG and WHG, relative to EHG, compared to what a Yamnaya based model would predict, while Corded Ware may be both a little richer in EHG and Anatolia_EN and less rich in CHG than a Yamnaya based model would predict. It seems possible to me that a slightly more CHG version of Yamnaya entered Hungary in the Bronze Age and precipitated the existence of the Bronze Age Hungarian population and from there maybe populations like Bell Beaker, through interactions with others?
@Matt
ReplyDeleteThat Kostenki-related dimension 3 looks like Ust-Ishim/baBal vs ANE, whatever that means. Eskimo stands out towards ANE more than Native Americans which doesn't correlate with their ENA.
Davidski, when is your new calculator coming out?
ReplyDeletecorrect me if I am wrong, but isn't the questions between the times of IJ separation from each other at 21000years and the J ydna found at 13000 and 10000 years?
ReplyDeleteThere will be no answers outside these date ranges.
@Shaikorth - interesting point about the placing of Eskimo there. Relates to a specific relatedness between North Asian groups and WHG-ANE groups that is not shared by Kostenki or Southern Asian groups (and apparently Ust Ishim and to a lesser extent Neolithic Farmers)? Bear in mind this is a very small dimension of genetic drift, but yes. I can't see it as ANE vs ENA/Ust Ishim because it loads so heavily on WHG as well as MA1, according to the biplot.
ReplyDeleteBtw, I think baBal tends to gives nonsense results in these f3s since the number of SNPs is so low (8,111 vs 300,000 for other groups), so I wouldn't worry about its position at all, I just couldn't be bothered to filter it out.
@ Capra, the Caucasus component from Dienekes old calculators seems to me likely to be a composite of EEF / Anatolia_EN ("Mediterranean" in Dienekes calculator terms although that likely include a little WHG) and CHG, as the local populations there seem to have become quite more heavily related to EEF (and also Southwest Asia) after CHG left for the steppe and South Asia.
Gedrosia seems likely to me to be the more simply CHG like one (and it seems to me to show up in Europe where it seems to me there is likely to be an excess of CHG ancestry to what would be predicted via EHG). Although on the other hand, Gedrosia could have picked up some elements of other admixture (EHG, ASI), so probably is not CHG exactly either.
Dienekes West Asian seems likely to be more what I would expect to be like CHG, but I doubt it will be 100% on the money, as neither were the EEF components, exactly.
Although I guess this all shows that it is hard to call when an ADMIXTURE component formed or if it is real. After the European results for adna with HG came in, it was easy to dismiss West Asian, probably not helped by Dienekes' linking of the component to his Womb of Nations idea of recent divergence of components from the a Caucasus birthplace of farming with heavy drift (which Europe seemed to prove false), but he wasn't to know and it seemed like a viable idea at the time. Perhaps a lesson for care in interpreting the historical story of adna in East Asia, as adna becomes available there (at the moment the main components seem likely to have been Sherpa related, Jomon related, Taiwanese Aborigine related and North Asian related)...
@Gaspar
ReplyDeleteIJ separation is probably twice as old as that, more like 45 000 years. Where do you get 21 000?
You guys are need to take a step back and look at the evidence. A handful a formal stats in the paper suggest CHG had no ANE. But can you show me in D-stats and F3 Drift stats that modern Caucasus(xwith European admixture) are closer to MA1 than to WHG or that MA1 is closer to them than to EEF. In D-stats and F3-stats MA1 isn't much closer to Georgians than to Sardinians.
ReplyDeleteThis is getting way out of hand. Once somone gets the CHG genomes it'll all be clear.
@Krefter
ReplyDeleteI don't think anyone is saying that CHG had any ANE. It's clear from the paper he didn't.
What I personally would like to test once the genomes are available is if the Kalash (and surrounding populations) are just CHG + South Asian or if they have additional ANE. That might be as easy as doing: D(Mbuti, MA1)(CHG, Kalash). I just want to check it. I have certain impression that there is additional ANE in S-C Asia, but it might all be just CHG as well.
@Alberto,
ReplyDelete"I don't think anyone is saying that CHG had any ANE. It's clear from the paper he didn't."
This is what I have a problem. You guys are making something out of nothing. When you have time can you show me how it is clear he had no ANE?
Has anyone looked at the new component scores for populations like Pashtun or Gujarat ?
ReplyDelete@Krefer
ReplyDeleteSo what are you saying, that CHG had ANE or that they didn't have?
The paper shows D stats where CHG is no closer to MA1 than to WHG. So I think that proves they didn't have ANE? Or am I missing something?
@capra
ReplyDeletefrom Karafet 2014
@Krefter
ReplyDeleteD(Yoruba, Kotias; Hungary_HG, MA1) -0.0203 -2.831
@Rob
That's going to be very interesting, but I don't know of anyone who has done that yet. I hope that David will have the genomes ready and maybe anew calculator soon to check how the new component impacts those and all other populations.
@ Matt
ReplyDelete"Particularly, I think it's possible that Bronze Age Hungary, Bell Beaker, and Northwest Europe is relatively rich in CHG and WHG, relative to EHG, compared to what a Yamnaya based model would predict, while Corded Ware may be both a little richer in EHG and Anatolia_EN and less rich in CHG than a Yamnaya based model would predict. It seems possible to me that a slightly more CHG version of Yamnaya entered Hungary in the Bronze Age and precipitated the existence of the Bronze Age Hungarian population and from there maybe populations like Bell Beaker, through interactions with others?"
I agree it would be interesting to see
But I think a few of us long suspected that ....;)
@Rob, Matt
ReplyDeleteSo CHG could be associated to R1b maybe, apart from J2? We have the R1b BA_Armenia that looks in admixture pretty much like 60% CHG/40% EF. And Gedrosia has been associated with R1b too (and obviously Bell Beaker). Would be interesting if R1b turned out to have excess of CHG over EHG.
@Alberto,
ReplyDeleteYou have to look at how modern populations score in D-stats. I'm pretty sure all West Eurasians, including Caucasins, are closer to WHG than to MA1. Plus, MA1 isn't much closer to Sardinians than to Georgians in D-stats. So, that destorys the arguments made by the paper that CHG had no ANE.
In the paper:
ReplyDeleteD(Yoruba,MA1;Kotias,Stuttgart) = D:-0.0153, Z:-2.380
D(Yoruba,MA1;Kotias,NE1) = D:-0.0083, Z:-1.362
D(Yoruba,MA1;Satsurbalia,Stuttgart)= D:-0.0096, Z:-1.423
D(Yoruba,MA1;Satsurbalia,NE1)=D:-0.0006,Z=-0.084
So there is some statistic with the CHG a little closer to MA1, but it falls short of significance. There could be something there in either direction with MA1 having some CHG related ancestry, or vice versa (not too surprising given geography). Not really that CHG defined as ANE+anything though, on this evidence (and this is evidence of 13300 BP antiquity).
At the same time
D(Yoruba,Onge,Kotias,NE1)=D:-0.0024, Z-0.0558
D(Yoruba,Onge,Kotias,Stuttgart)=D:-0.0030, Z-0.0678
D(Yoruba,Onge,Satsurbalia,NE1)=D:0.0009,Z=0.0202
D(Yoruba,Onge,Satsurbalia,Stuttgart)=-0.0004,Z=-0.085
So that's pretty equivocal that these CHG and European Early Farmers seem about as shifted towards ENA (EEF have minor WHG compared to Anatolia_EN, but that shouldn't change matters too much).
Re: South Asia and steppe ancestry, quick toy model in a spreadsheet, assuming different pre-Sintashta fractions:
http://i.imgur.com/CpPfwb2.png
http://i.imgur.com/Qt50h4O.png (these seem pretty extreme to me, but not necessarily impossible I guess)
(assuming ancestry proportions from Mathieson et al's interpretation of Sintashta and Andronovo - http://s1.postimg.org/8s7j5xipr/EDF2.png - which might differ if modelled with CHG as well.)
@ Krefter, if your argument is what I think it is, you're saying: CHG could have MA1 related ancestry (ANE) and be as close to WHG as MA1, as modern West Eurasians were modelled as having ANE and are closer to MA1.
Certainly true, but the difference is we had samples that could mix with ANE (WHG, EEF) to fit modern West Eurasians as having MA1 related ancestry while still being closer to WHG. And admixture statistics supported the idea this was the case (although it now seems Yamnaya / EHG are better than MA1 was).
For CHG, this is not true. We have nothing that would mix with MA1 to make CHG and then be the appropriate closeness to WHG (exactly the same as it is to MA1) and all other populations. There's no particular reason to model CHG as MA1 admixed, unless we want to make up an extra ghost population to do so, and there seems like no benefit to imagining this ghost existing, in terms of modeling other populations and population history.
Before going on with the theorizing, I think we need more ancient DNA from south of the Caucasus (namely, Iran). How much do we have from there? Maybe it's all J and Teal, or maybe it's not...
ReplyDeleteWe need to determine what clades of J2a are Indo-European. J2a is particularly popular in Greece, and maybe was related to the Myceneans. J2b and R1b-L23*/Z2103 are closely related in the Balkans, and since J2b is found in Saami I find it hard for it to be near eastern, since J2b2 is also exclusively found in the Balkans and Northern India, with it being rare in the middle east.
ReplyDelete@Krefter
ReplyDeleteOk, I think I also get your idea. But it's also based on the idea that modern Caucasus have ANE, while they might not have. We don't have any population that we know for sure that is a mix of MA1 and ENF to check if it would behave the same as CHG. We only populations that actually descend from CHG and not from MA1. So it's hard to tell.
Even in the hypothetical case that CHG were a 60/40 mix of ENF and MA1, the important thing now is that they created their own drift and we can track their ancestry easily. Much like Native Americans can be a 60/40 mix of East Asians and ANE, but then they are drifted enough to be able to make a Native American component. In this last case, it's not useful because we're not tracking back-migrations of Native Americans in Eurasia. But if we were tracking them, getting a real Native American genome instead of having to use East Asian + ANE would be a boost. So that's what we got now, a genome of CHG which will help us track the admixture and movements of this population. no matter if it's a mix of ENF and ANE or if it's completely unrelated.
As a side note, yes, it would be interesting to explore the chances of it originally being a mix of ENF and ANE, but just for the sake of knowing how plausible it is. For practical purposes we're better of using the CHG component than trying to split it again into something else.
Nomen
ReplyDeleteWe need more j2 studies in Europe, but from what I recall there are some interesting patterns. J2a in southern Greece (incl Crete) and central Italy, J2b in northern Greece, the Balkans, and the Black Sea littoral (incl. up to 10% in Ukrainians).
J2a is common Pakistan, but j2b is rare. India has both. Caucasus is dominated by the J2a-M92 and 67 varieties IIRC.
Turkey had some regional variation (see the Cinnioglu study); i can't recall exactly
Most of what we know about frequencies of deep Y DNA subclades is from commercial testing. Very few non-Europeans do commercial testing. So, we don't know much about Y DNA J.
ReplyDeleteCHG looks like a composite of ANE and ENF but drifted due to isolation
ReplyDeleteMy PCA of Satsurblia.
ReplyDeletehttps://drive.google.com/file/d/0B9o3EYTdM8lQM3h1cTNmcGFfYVE/view?usp=sharing
@Davidski,
ReplyDeleteA nice game changer! Satsurbia is the magnet that draws much of the Near East away from Early Neolithic farmers. Again, revenge of the Mesolithic!
This comment has been removed by the author.
ReplyDeleteIf any has time can you run these Formal stats....
ReplyDeletehttps://docs.google.com/spreadsheets/d/1TPByWTZL2eOhq21QvpeV-PFe7Ev04FAcxeD-72fyMnA/edit#gid=354458588
All evidence CHG lacks ANE is.....
>East Asians are equally related to CHG and EEF. But MA1 is slightly more related to CHG than to EEF.
>CHG is closer to WHG than to MA1.
There could be MA1-type ancestry in CHG that is masked by as much WHG admixture in ENF(Old Near Eastern, shared by EEF and CHG).
There might not be real MA1 ancestry in all of West Asia. I'm thinking North Caucasus has a lot of Steppe ancestry, but besides that there isn't MA1-related blood. Instead the only people with legitimate MA1 blood are Siberians, Amerindians, EHG(therefore Europeans), and maybe South Asians.
How can we figure out if WHg is masking MA1 ancestry in West Asia? Is there anyway? Because it could be that West Asians don't have WHG or MA1, but a close relative of both. And that EEF has WHG and EHG had ANE.
Krefter,
ReplyDeleteI'll run some stats when I have Kotias because it has a lot more markers.
rk,
https://drive.google.com/file/d/0B9o3EYTdM8lQTE9QU2tFd3ZJV3c/view?usp=sharing
https://drive.google.com/file/d/0B9o3EYTdM8lQMHVQRkVUUndnSUk/view?usp=sharing
https://drive.google.com/file/d/0B9o3EYTdM8lQa0RfVVVwM3U5RkE/view?usp=sharing
https://drive.google.com/file/d/0B9o3EYTdM8lQZTkxLVJTNGtCYXM/view?usp=sharing
@Davidski,
ReplyDeleteWere you able to create a Y BAM file for Satsurblia? It will be useful to drive down its current J status (if possible) to J1 or J2.
I won't be able to do any better than YFull. I know Kotias is already up on their tree, and Satsurblia might be too.
ReplyDeletehttp://www.yfull.com/tree/
Roy,
ReplyDeleteIs the Kotias Klde sample from Mesolithic Georgia 100% confirmed to belong to Y-DNA haplogroup J2a? I ask this because I have seen some people call it/him a ''partial J2a''.
@Gaspar
ReplyDeleteThe only Karafet 2014 paper I know of is "Improved phylogenetic resolution and rapid diversification of Y-chromosome haplogroup K-M526 in Southeast Asia", and I don't recall that saying anything about IJ. Could you direct me to the specific place?
Y-Full puts the TMRCA of IJ at 39 900 to 45 200 years ago, and Karmin et al put it at around 45 000 years ago. I am surprised to hear that any study came up with such a young date.
K8 results. Lots of missing markers though.
ReplyDeleteK8 Satsurblia
ANE 0.2485
South_Eurasian 0.042
ENF 0.6126
East_Eurasian 0
WHG 0.0692
Oceanian 0
Pygmy 0.0043
Sub-Saharan 0.0234
Satsurblia scores like modern South Caucasus with extra WHG and Sub Saharan noise.
ReplyDeleteLooking at Eurogenes K15 and ANE K8 results for the Caucasus I realized....
North Caucasus has European admixture and South Caucasus has more CHG. This is obvious by WHG scores in K8 and North Sea/Atlantic/Baltic/(espec.)East Euro in K15.
West Asian or Teal is highest in the South Caucasus where ANE is lower. I expect CHG to have MA1-related ancestry but it wasn't very much.
We're begging to learn the origins of ANE in West Asia. It didn't come from a European(WHG-rich) population. The super high ANE in North Caucasus is from European or North Asian admixture though. ANE in most of West Asia probably comes from CHG_types.
Capra
ReplyDeleteMaybe Gaspar is referring to the karafet paper from 2008 (New binary polymorphisms reshape and increase resolution of the human Y chromosomal haplogroup tree) ??
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2336805/
Maybe I'm wrong, but I think that main sourse of ANE in East Europe and North Caucas came from Seroglazovka culture.
ReplyDeleteIs there any way that North Eurasians (MA1) can be modeled as being slightly mixed between a proto-CHG and something related to Kostenki? If this is the case it could explain the presence of mtDNA H in modern ANE admixed Siberians and X in Native Americans, both possibly originating from a Caucasus source. R1 would have split (30kya) early into Central Asia and be associated with a mixed ANE signature while R2 would maintain the original CHG West Asian like signature in West and South Asia. MA1 being mixed could also explain why the CHGs are being expressed as part MA1 in ADMIXTURE. Anyway just some thoughts and would be equally satisfied for this to be shown unfeasible.
ReplyDeleteCHG has a lot of Basal Eurasian ancestry. MA1 (ANE) and EHG have none.
ReplyDelete@ Matt
ReplyDelete„It seems possible to me that a slightly more CHG version of Yamnaya entered Hungary in the Bronze Age and precipitated the existence of the Bronze Age Hungarian population and from there maybe populations like Bell Beaker, through interactions with others?“
Seriously? Did you notice the outgroup f3 stats in Supp. Fig. 1 showing that both Kotias and Satsurblia share more drift with Hungary_CA than with Bell Beaker and with Hungary_BA? So if anything the incoming Yamnaya reduced the CHG affinity in Hungary, rather than having particularly high affinity to it.
@ Alberto, re: Gedrosia
ReplyDeleteDon't forget, the R1a-dominated Corded Ware samples all have far more Gedrosia than modern Britons.
It's just crazy that in David's PCA Satsurblia is extremely far away from Hungary_CA, and Hungary_BA is closer. :-o
ReplyDeleteIt's unlikely that Hungary_CA has admixture from CHG. They probably just share ancestry from a third source.
ReplyDeleteIt might just be that they have more similar ratios of Basal Eurasian than Hungary_BA and CHG.
SC Asia and India should show some ANE say around 10-15 % even after the inclusion of CHG in ADMIXTURE, but if CHG eats all the ANE there, then it will become very interesting, concerning the high Concentrations of R1a,R2a and U2 Mtdna's there.
ReplyDelete@Nirjhar
ReplyDeleteYes, it's going to be interesting to see what happens with that. I think that S-C Asians have a higher affinity to MA1 than CHG, but on the other hand I think that this is not going to show up in Admixture and probably CHG will replace all the ANE in S-C Asia. Maybe because the CHG-like population ancestral to S-C Asians had higher MA1 affinity than CHG themselves, but otherwise was quite similar (just like there is a small difference between Loschbour and KO1 in affinity to certain populations).
But then I wonder, if this happens, how relevant is ANE going to be once we have a CHG component? We'll have to see how that turns out.
This comment has been removed by the author.
ReplyDeleteThis comment has been removed by the author.
ReplyDeleteDoes anyone know the hair color, eye color and skin tone of CHG??
ReplyDeleteYes, it's in the paper: dark hair & eyes, lighter skin than WHG
ReplyDelete@Rob
ReplyDeleteI just noticed Karafet 2015 paper ( March ) has a different date.
with a new branch called I3 for a Georgian sample.
http://genome.cshlp.org/content/suppl/2015/02/18/gr.186684.114.DC1/Supplemental_Text.pdf
So my numbers are now wrong with Karafet latest change.
The split of IJ is the same time that LT split from K , so whatever that is , thats the date
https://en.wikipedia.org/wiki/Paragroup
@Rami
ReplyDeleteYes, he was brown eyed and dark hair, but fair skin.
@Rami, truth, Gasper
ReplyDeleteAlso, Satsurblia from 13,300yo was a carrier for blue eyes. This confirms the few West Asians with blue eyes today didn't have to receive it from European admixture.
There's no markers that correlate with skin tone in a significant degree. Although for the most part Sardinians and West Asians are a Brown tone, so it's probably save to assume CHG, EEF, and all Paleolithic/Neolithic West Asians were to. I definitely wouldn't describe them as fair, except compared to Indians or Africans.
Simon W: Seriously? Did you notice the outgroup f3 stats in Supp. Fig. 1 showing that both Kotias and Satsurblia share more drift with Hungary_CA than with Bell Beaker and with Hungary_BA? So if anything the incoming Yamnaya reduced the CHG affinity in Hungary, rather than having particularly high affinity to it.
ReplyDeleteI noticed that after posting that comment, actually ;) Thanks for asking.
But actually I'm thinking about the balance of EHG vs Kotias compared to assuming Hungary_BA is WHG+EEF+Yamnaya, not so much whether CA is closer to Kotias than Hungary BA (although it seems like it should be, so surprising if it's not). Which will take some modelling, to see if the EHG:Kotias ratio of Hungary_BA's ancestry that isn't WHG or EEF, is higher or lower than Yamnaya.
The f3 statistics there are seem generally, interestingly, not indicative of any strong patterns that line up with our prior assumptions for whatever reason (e.g. Sweden_MN and Spain_MN at opposite ends of the distributions, while more consistency with many other populations). The level of sharing is a little lower compared to Bichon, also, I noticed (but this is also true of EEF compared to Bichon/WHG with present day populations?).
@Krefter makes sense because when I went to a Druze regions in Israel years ago , I noticed an unusually high amount of people with blue eyes ,as well as blondism esp compared with other groups in the region. Druzea are an archaic West Asian group they do not have little or no Euro admixture.
ReplyDeleteAlso agree they could not be fair like Europeans are today. Probably a shade of brown but question is were they like lighter West Asians or darker west Asians, whose skin tone is comparable with NW South Asians.
This CHG seams to correspond with the Baloch/ANI component, which peaks in Baloch and Brahui , I am assuming they will score the highest CHG , followed by the Kalash and Burusho peoples
Laz is convinced CHG is Yamnaya's Near Eastern ancestor. He's probably disappointed his team didn't discover them first.
ReplyDeletehttps://twitter.com/iosif_lazaridis
@Matt
ReplyDeleteThanks, you're probably right that Caucasus and Baloch/Gedrosia are composites rather than reflecting different CHG-like populations. Or at any rate drift specific to different CHG-like populations is probably only part of the story as far as ADMIXTURE is concerned.
@Gaspar
Thanks, you had me a bit confused there. The 2015 paper you linked to is Karmin et al not Karafet et al (though Karafet was one of the authors as well). The split date is ~45 000 years ago.
@Krefter:There's no markers that correlate with skin tone in a significant degree
ReplyDeleteSLC45A2 and SLC24A5 are both there and both are reliable predictors of skin tone. Both of the CHG samples are homozygous derived for light skin at one, and homozygous ancestral for dark skin at the other.... just like Razib Khan IIRC. This suggests they were darker than your average modern West Asian, and somewhere in the range of modern South Asians.
@Tobus,
ReplyDeleteSLC45A2 and SLC24A5 do show correlation but there's a lot more involved. On 23andme thread people posted results and there are Europeans with an ancestral allele for a single one(I think I've seen some with ancestral allele for both) who have white skin. There are West Asians and South Asians who have double derived alleles for both who have brown skin. Latinos are the people to study to see if SLC45A2 and SLC24A5 have a strong effect.
If CHG is in fact 24% ANE as your K8 run suggests Davidski, I think that's a major blow to your contention that R1a couldn't have passed through somewhere south of the steppe before the Indoeuropean expansion.
ReplyDeleteYou're exaggerating.
ReplyDeleteR1a is found exclusively in prehistoric samples that have very high ratios of EHG ancestry. What evidence do we have that CHG carried R1a or was even partly EHG?
@Krefter:
ReplyDeleteThe research into these two alleles is really solid, and has repeated multiple times under empirical conditions using different populations (here's one including Latinos: http://www.researchgate.net/publication/6118939_The_golden_gene_%28SLC24A5%29_differentiates_US_sub-populations_within_the_ethnically_admixed_Y-SNP_haplogroups).
You'll have to excuse me if I don't take anecdotes from 23andme as solid evidence, as personal impressions of skin colour are notoriously subject to both subectivity and environmental factors, and when using photos, the lens and lighting. Empirical research is done with skin reflectance measurements sampled from unexposed skin (typically under the arm). I suspect many of these claims would fall apart if tested empirically.
Be aware that SLC24A5 is widespread outside of Europe (see https://en.wikipedia.org/wiki/SLC24A5#/media/File:Ala111Thr_allele_frequency_distribution0.png), and almost certainly didn't originate there. So although it's generally associated with white Europeans, virtually all the dark people in Eurasia and North Africa also have the derived "light skin" version of it and are correspondingly lighter skinned than sub-Saharan Africans and Melanesians. This means derived SLC24A5 carriers can range from South Asian to Scandinavian in skin tone, depending on the remaining skin colour alleles. SLC45A2 for instance: West Asian populations have high (often >50%) frequencies of derived SLC45A2, South Asian pops typically have <10%. Based on our current understanding of skin colour genetics, CHG, with 0% derived SLC45A2, is probably going to be more towards the South Asian end of the scale than the West Asian.
David - No evidence yet, but if we assume R1 and ANE are linked, I don't think it's unreasonable to think that CHG could have received R1 along with its ANE heritage as well. I'm not saying that that must be the case, just that it reduces the certainty that R1 was confined to the Steppe, as ANE was apparently not.
ReplyDeleteOn a mostly unrelated note, would you mind if I sent you a somewhat technical and personal question via email?
Sure, as long as it's not too personal.
ReplyDeletePersonal for me, not you, and thanks. Will do.
ReplyDelete@ David
ReplyDelete"It's unlikely that Hungary_CA has admixture from CHG. They probably just share ancestry from a third source.
It might just be that they have more similar ratios of Basal Eurasian than Hungary_BA and CHG."
On a second thought, agreed. CO1 is too similar to MN farmers in PCA and other tests to have CHG admixture.
@ Matt
ReplyDeleteI see, yes it's not straightforward. Will be very interesting to see if the Hungarian Yamnaya had more or less CHG than other Yamnaya.
@ Capra
ReplyDeleteThe Dodecad K12b Caucasus component is quite strong in some EEF like Ötzi (22%), and also in Sardinians, so already for that reason it's dubious that it's a variant of CHG. And all the modern Caucasus pops have both Caucasus and considerable Gedrosia; without the latter modern Caucasians are impossible.
Also the ADMIXTURE analysis in the paper at K=17 shows no CHG component in the Copper Age samples, and one of these must be CO1, though it's not labelled which one.
ReplyDeleteBut re: Hungarian Yamnaya, judging from previous ADMIXTURE runs, I'd expect they had low CHG (they had low "teal" after all).
ReplyDeleteWaiting for tests with the CHG genomes still to get a better understanding, but regarding the CHG-ANE question, I think that the whole ANE concept was mostly sustained by the CHG component. Once we separate them, I'm really not sure that ANE is a component at all. Needs to be investigated, but Admixture consistently ignoring it in hundreds of runs with different samples (including MA1 and AG2, and also EHGs) might not be a coincidence.
ReplyDeleteNow that we know that CHG/Teal is a real component, going back to the Haak et al. Admixture (which has high resolution, unlike Jones et al.) becomes quite interesting. The rainbow that MA1 shows probably reflects a reality. A mix of (proto-?)WHG, something S-C Asian (Kalash-like? But more ancient, with some Onge-like affinity) and some ancient East Asian components.
So for West Eurasians, once we have a WHG and CHG clusters, it's going to be quite irrelevant, since only the non-West Eurasian parts will show up (something specific South Indian and East Asian).
From ancient DNA is going to be difficult to find this out, because these are very old migrations. To detect the South (or South-Central) Asian ancestors of MA1 moving north, then mixing with WHG-like people there, and then with something East Asian in Siberia is not very feasible. But maybe finding some model that works for MA1 can be done. The assumption that because it's old it contributed to different populations might be the wrong one. It now looks more like different populations from Europe, South Asia (including CHG) and East Asia contributed to MA1.
We'll have to learn about this.
Alberto
ReplyDeleteI was going to say that I'm not surprised that mal'ta died off; but related populations survived nearby - trans-Baikalia and Altai. Archaeologists constantly point to this region as the origin of Mesolitiic microblade technology which reached western Russia by 9 ky BP.
Another - more remote possibility - is that some ANE survived far to the west - in an already admixed form with WHG (somewhere in SW Ukraine and Moldavia) during the LGM. The absence of EEF in Yamnaya makes this admittedly unlikely
@Rob
ReplyDeleteYes, these populations obviously existed. What is unclear is they're real genetic structure. Moving geographically in an area between Europe, (South-)Central Asia and East Asia might have determined their genetic makeup. What is unclear is, if this population really was the same one, isolated before, during and after the LGM, why didn't they create a strong component that easily shows in admixture, even when having the presumed "original" samples? Why do EHGs are so close to WHG if they stayed isolated from each other for 10-15 Ky? Native Americans are very likely a mix of several different things, but they did create a component due to isolation and drift that very clearly shows up in Admixture. WHG did too. CHG did too. Why ANE didn't?
If we find a model that works good for MA1 we might get answers to these questions. Maybe more than real population continuity there is population similarity due to geographical area and the mix of similar components, even in different times.
We need to investigate and learn this to be able to understand things better. The ANE-based doesn't seem to work now that we have CHG, but it does leave a few lose ends that we'll need to tie somehow (their relevance is probably minor, but still we can't ignore them completely).
This comment has been removed by the author.
ReplyDelete@RK
ReplyDeleteYes, sure, admixture is not 100% reliable, it gives many different outputs and variation based on modern populations, etc... But it also can't be 100% wrong all the time.
CHG component was already pretty well represented in Haak et al. just by having Yamnaya genomes, not even the actual CHG. And now CHG do get their own cluster where they score 100%. WHG also when included (for example again in Haak et al.) get their cluster with 100% score. EEF don't get 100% but that's because they have WHG too. But in any case, Admixture does a good job with the ancient components even when unsupervised and mixed with so many modern samples. So yes, it's not absolutely smart, but also not absolutely stupid.
ANE has been elusive in absolutely every single run I've seen unless forced by some "zombie" ANE sample in supervised mode. This is what I mean. It's very strange. We have MA1 and AG2, we have EHGs, we have many Siberians and Native Americans. Yet, ANE never appears as a component that links these populations in any significant way.
"MA-1 score very significantly in S Indian in every analysis incl. Jones et al., and the Kalash component favours MA-1 over EHG in this analysis. And also, there is R2 and many U clades in India. I would't be so quick to dismiss that."
Yes, and this has often been interpreted as MA1 ancestry in South Asia. But it's time to think it over again and admit that it's much more likely South(-Central?) Asian ancestry in MA1.
That's why finding a model that works for MA1 would be very helpful. WHG + Kalash (?) + something ENA, maybe. The problem is that MA1 is very old, so with modern populations is difficult. We're still missing many pieces, but now with CHG we can make a step forward to elucidate all this.
This comment has been removed by the author.
ReplyDelete@RK
ReplyDeleteThe point about getting a component in Admixture is not so much about being ancient or not, but about being "unique" in some way. Small, isolated populations are subject to high drift, and admixture can see this. We've learned that the LGM in the northern parts of Eurasia left small pockets of isolated populations in a few refugia, so these populations are "specific" enough to be detected by Admixture. ANE looks like a perfect candidate for this specificity, but surprisingly it doesn't show up, anywhere. That's quite a paradox for me.
Another paradox is the situation of EHGs. If they are direct descendants of MA1-types who 24.000 ya. was already quite different from WHGs, wouldn't after 15.000 years these two populations be much more different instead of much closer? The only way they can be closer is if EHG is admixed with WHG (which looks like the more likely scenario, obviously). But then, if we have WHG and EHG (as a mix of WHG + a putative ANE population descended from MA1), why doesn't admixture detect this component like it could detect the CHG component when the EHG + Yamnaya samples? Maybe someone has a good explanation for this, but I don't.
Another thing, do you know someone who has measured with good methods the amount of West Eurasian and East Eurasian in Native Americans? All I've seen is an assumption that if they are 40% MA1 it means they are 40% West Eurasian. But neither I've seen any reliable test for Native Americans to confirm this, nor have I seen it for MA1.
If MA1 is 100% West Eurasian, we should expect that Dstats like:
D(Mbuti, East Eurasian; MA1, West Eurasian)
would be insignificant, right? (West Eurasian could be WHG, CHG, LBK_EN... and East Eurasian could be Han, Dai, Kharia, Kinh,...) I would like to see if those kind stats are really insignificant or not. Have you seen any of such kind?
This comment has been removed by the author.
ReplyDelete@Alberto
ReplyDeleteLoschbour MA1 Korean Chimp -0.0054 -0.814
Loschbour MA1 Kinh Chimp -0.005 -0.739
Loschbour MA1 Atayal Chimp -0.0046 -0.665
Loschbour MA1 She Chimp -0.0045 -0.664
Loschbour MA1 Ami Chimp -0.0042 -0.625
Loschbour MA1 Lahu Chimp -0.0038 -0.568
Loschbour MA1 Dai Chimp -0.0074 -1.11
Loschbour MA1 Han Chimp -0.005 -0.752
Loschbour MA1 Japanese Chimp -0.0048 -0.715
Loschbour MA1 Miao Chimp -0.0023 -0.345
Loschbour MA1 Mongola Chimp -0.0039 -0.583
Loschbour MA1 Naxi Chimp -0.0054 -0.808
Loschbour MA1 Oroqen Chimp -0.0051 -0.768
Loschbour MA1 Thai Chimp -0.0011 -0.169
Loschbour MA1 Tujia Chimp -0.0053 -0.803
@Carpra
ReplyDeleteThanks! Those look indeed very consistent.
Though I still find it strange. I went back to the original paper's supplementary information and I don't see there anything much that indicates that MA1 is completely West Eurasian. On the contrary, he looks pretty ENA, though probably of an extreme form that is not related to modern East Asian and mostly is related to Native Americans:
- The admixture at the optimal K9 (page 55) clearly shows MA1 with a lot of ENA. Only 34% is clearly WE, while the rest is either ENA or a mix of WE and ENA (depending much if we consider Kharia as mostly WE or ENA).
- On the PCA, Figure SI 10 A (page 64) MA1 is clearly between West Eurasians and ENA, though mostly towards Native Americans and less towards Oceania or East Asians.
- On TreeMix models, he appears first (Figure SI 13, page 71, 4 migrations edges) to stem the area in between ENA and West Eurasians. Then (Figure SI 15, page 73, 4 migration edges) again from the same place between ENA and WE, but this time with some extra ENA migration edge from between Dai and Papuan. Finally (Figure SI 16, page 74, 4 migration edges), he does appear in the WE branch, but gets a migration edge from Karitiana, who is in the ENA branch between Han/Dai and Papuan.
- Finally, using MixMapper (Figure SI 18 B, page 79), MA1 appears as a mix of Sardinian and Han, and subsequently Karitiana as a mix of MA1 and Han.
So all in all, the paper strongly suggests that MA1 is a clear mix of West Eurasian and ENA, though an ancient and extreme form of ENA not related to East Asian and Oceania, but to Native Americans and probably some Siberians.
@Capra, you once mentioned that it would be easy to estimate the proportions of West Eurasian and East Eurasian in Native Americans using f4 ratios. Any idea if someone has done that or how exactly and who could do it? I doubt that Native Americans have more than 15-20% West Eurasian, but would love to see a real solid test about it.
Ah, also:
ReplyDeleteLoschbour EHG Han Chimp -0.0151 -2.884 341763
Loschbour EHG Kinh Chimp -0.0146 -2.722 341763
Loschbour EHG She Chimp -0.0144 -2.665 341763
Loschbour EHG Dai Chimp -0.0137 -2.506 341763
And EHGs have about half the ANE than MA1.
This comment has been removed by the author.
ReplyDelete@RK
ReplyDeleteThe stats that Capra Internetensis posted are very consistent, as I said. I'm far from ignoring them. But I think your answer is not very fair or that helpful.
I went to the original paper as you suggested and found several tests that show ENA admixture in MA-1. Now you tell me that I should ignore the Admixture results they got, I should ignore the Treemix results they got, I should ignore the MixMapper results they got,...
About their conclusions and the f3 stats:
"SI 14.7 Evidence for basal East Asian gene flow into MA-1
Putative western Eurasian ancestry in Native Americans does not exclude the possibility of some gene flow into the MA-1 lineage from eastern Eurasian populations. To test this, we compared f3(Yoruba; MA-1, X) and f3(Yoruba; Sardinian, X), where Population X represented one population from a set of worldwide populations. Under a model where MA-1 is from the same lineage as the Sardinians, the ratio of these two statistics for unrelated populations is expected to be 1.0, but East Asians and Oceanians were both observed as being closer to MA-1 than to the Sardinian (Figure SI 27). Since the ratio is ~1.09 for both Oceanians and East Asians in the outgroup-ascertained data, the MA-1 lineage may have absorbed some ancestry from populations ancestral to these groups. Indeed, this is also consistent with the admixture graphs inferred using MixMapper (SI 12). However, a note of caution here is that the MA-1 data is of lower quality than the Sardinian data.
...
Figure SI 27 Contrasting shared drift between different modern populations and, Sardinian and MA-1 using an outgroup f3-statistic.
All East Asian populations (pink) are closer to MA-1 than to Sardinian, and no difference is seen with Oceanians (dark green)."
I'm not ignoring anything or jumping to conclusions, I'm just raising questions to try to understand what is this elusive ANE thing. But I'll leave it here for now. It's not very relevant anymore for West Eurasians, so maybe some time in the future we can learn more about it.
@Alberto
ReplyDelete*In principle* it would be easy, but we have to make assumptions about the history of the reference populations which may be (probably are) wrong. Anyone with admixtools and the appropriate genomes could do it. It would just be f4(Amerindian, East Eurasian; WHG, Outgroup)/f4(MA-1, East Eurasian; WHG, Outgroup) to find the proportion of ANE. Or alternatively f4(Amerindian, WHG; Onge, Outgroup)/f4(Han, WHG; Onge, Outgroup) to find the proportion of Northeast Asian. Ideally using a variety of reference populations to check how robust the estimate is. The idea is to set up f4(X, B; C, O)/f4(A, B; C, O) so that the A-related component of admixed population X forms a clade with A against C and C forms a clade with A'X(A) against B (related to the other component of X) while O is an outgroup to all of them (or near enough). That way the A-related component of X and A itself will be equidistant from C, so the affinity of X to C vs A to C will be proportional to the amount of A-related ancestry in X, and unaffected by the nature of B and C. (Notice that using Onge and Karitiana would already violate the assumptions, albeit weakly.)
@capra
ReplyDeleteThanks. I guess I get the concept, though I'd need to see the stats and maths to really understand it better.
Yes, we always have to make assumptions with these things, but even then it would be ok as a working hypothesis. What I'm surprised is that if it's this easy no one has ever cared to do it. Before MA1 we thought that Native Americans were 100% ENA. No one checked those f4 ratios? After MA1 we all think they are 40% West Eurasian. Again, did someone check that? Apparently no, that I know of.
Under the assumptions that Sardinian is 100% West Eurasian, and Han is 100% ENA (and the assumption that ANE is clearly closer to West Eurasian than to ENA), I kind of doubt that Karitiana would turn out 42% "Sardinian", but I could be wrong. In any case, I think another problem might be the assumption that outside Africa there is only WE and ENA. Native Americans don't seem to sit in between West Eurasians and East Asians, but probably have part of both plus something else that pulls them away from a cline.
But anyway, no point in speculating about this now. Will wait for future findings to see if they inform us about this ANE thing any better.
Thanks again.
@Alberto,
ReplyDelete"The stats that Capra Internetensis posted are very consistent, as I said. I'm far from ignoring them. But I think your answer is not very fair or that helpful.
I went to the original paper as you suggested and found several tests that show ENA admixture in MA-1. Now you tell me that I should ignore the Admixture results they got, I should ignore the Treemix results they got, I should ignore the MixMapper results they got,..."
RK just refuted all lines of evidence that MA1 has any ENA.
Ryukendo, you say that “all east asians are also equidistant from ma-1, ie onge and dai are the same distance to ma-1.” With this do you mean only Onge and Dai or do you mean that also Tibetans, Ainu, Japanese, Ittelmen, Koryaks, Chukchi and Eskimos are equidistant from Ma-1? In admixture runs they are all fully East Asian.
ReplyDeleteDai and Onge are both very southern and geographically close to each other. Shouldn't we use for example Ainu instead as they should be 100% East Asian but more northern version.
@Alberto
ReplyDeleteIf EHGs are closer to East Asian than MA-1 is and have less ANE, then this can hardly be evidence for the East Asian affinity of ANE/MA-1. I wouldn't say EHG have less ANE though really, because EHGs are not significantly further from Amerindians than MA-1 is. So if we model EHG as a mix of WHG and ANE, then the ANE in EHG is more Amerindian-like than the ANE as represented by MA-1, or alternatively MA-1 is a mix of ANE proper and something else (probably extinct). If ANE formed in a refugium during the LGM and then moved to the New World, which is questionable on grounds of timing, then EHG and Amerindians could be more closely related because they share that extra drift that MA-1 did not experience. But the Amerindian type of ANE could already have been in Beringia or wherever at that point, in which case ANE is a pre-LGM ancestry with a lot of independent drift in different populations.
ANE is pretty much by definition what draws Amerindians and Europeans together. Amerindians are consistently closer to West Eurasians than East Asians are, and this goes all the way back to K-14:
East Asian Native American Kostenki-14 African -0.0025 -5.3
But being closer to Amerindians does not make you consistently closer to East Asians.
Chimp Karitiana EHG MA1 -0.0002 -0.033
Loschbour EHG Han Chimp -0.0151 -2.884
Loschbour MA1 Han Chimp -0.005 -0.752
So the parsimonious solution is that, however complex the actual population history involved, ANE is a branch of West Eurasian.
Regarding the evidence from the MA-1 paper, don't forget that Sardinians and other modern West Eurasians have Basal Eurasian (whatever that is), and MA-1 does not. TreeMix and MixMapper may model gene flow from Crown Eurasians into MA-1 to compensate for it. East Asians being closer to MA-1 than to Sardinians is part of the evidence for Basal Eurasian, and basal East Asian gene flow is a possible alternative explanation, but it is not specific to MA-1/ANE.
ADMIXTURE makes MA-1 a jumble of modern components with high ANE affinity - Sakilli and Estonian, + smaller amounts of Karitiana and Inuit - and excludes the low ANE ones - Bedouin, Han, and Even, except for a little Papuan. Since MA-1 is highly distinct from the non-ANE ancestry in any of these groups, he gets a mix of them, same as Ust' Ishim does (but with more bias toward West and South Eurasian). BTW, the South Indians in the sample are not Kharia and other Austro-Asiatic-speaking tribal East Indians, but Sakilli and other Dravidian-speaking low-caste/tribal South Indians, and they all have Papuan + Han components on top of their specifically South Indian component, so the latter is not particularly East Eurasian.
On PCA an ancient genome, lacking specific modern ancestry, will tend to plot toward the middle. MA-1 is close to the centre but deviates towards Europeans and Amerindians (away from East Asians). Or alternatively, he is placed between Amerindians and Europeans + South/Central Asians. You could say that on PC1 he is as East Asian as South Asians are, but the PCA is dominated by Amerindians and this is not really an East-West Eurasian axis as such.
I am not saying that MA-1 cannot have any East Asian ancestry, just that the evidence of formal statistics is a lot less ambiguous than the output of model-building programs.
@Alberto
ReplyDeleteFor more evidence, look at Lazaridis, where Basal Eurasian is introduced for the first time. This completely expunges the need for any ENA in ANE, and cements the idea of ANE as a west eurasian lineage.
The reason why MixMapper will never work is because it has no ability to postulate ghost populations as part of its calculations, so of course it places amerind as sardinians plus east asian. ADMIXTUREGRAPH, and the human mind itself, does have such an ability, so that graph is moot.
Jyst place the results of model-based algorithms or methods dependent on the type and number of samples, e.g. pca or ADMIXTURE, on a lower footing from now on, follow where the formal stats lead.
This comment has been removed by the author.
ReplyDeleteRyukendo, so I gather that at least North East Asians are not equidistant from Ma-1, but in fact, I already knew, that in the recent Ket paper they estimated that the amount of ANE, i.e. West Eurasian ancestry, in Eskimos is 66.7%.
ReplyDeleteHowever, I admit that Ma1 looks like being distant from Southeast Asians, so I stick to my belief that ANE represents to a considerable extent the third Eurasian component that developed close to India.
With this new CHG, I see that India has become (more) West Asian quite recently and this explains the findings that “the analysis suggests that the 8th century Roopkund population comprised two groups of genetically distinct individuals. The majority showed genetic affinity with present day European and Middle Eastern populations, while the others displayed common haplogroups with the Austro-Asiatic population of the North Indian Himalayas.” (http://eurogenes.blogspot.fi/2015/08/archeogenetics-of-roopkund-skeletons.html)
@RK
ReplyDeleteThanks, but no problem, it was more about the "choose this and ignore that" than about the tone. As you said, we have to look at all the tests and contrast the information to have something really solid. And yes, I agree about ANE likely not having ENA from the area between China to Thailand and to Oceania. At least MA1, because EHG do seem to have a bit of it.
For now I'm not going to explore further this because it's not really relevant right now and besides I don't have the tools to make the tests I'd like to. But some time in the future it'll be interesting to retake the topic of ANE. Maybe if we get a Mesolithic sample from around the Altai region that gives it some more solid base for tests than MA1 is.
@Capra
ReplyDelete"But being closer to Amerindians does not make you consistently closer to East Asians.
Chimp Karitiana EHG MA1 -0.0002 -0.033
Loschbour EHG Han Chimp -0.0151 -2.884
Loschbour MA1 Han Chimp -0.005 -0.752
So the parsimonious solution is that, however complex the actual population history involved, ANE is a branch of West Eurasian. "
We should compare apples to apples meaning Northern Amerindians and East Asians, not Karitiana. I suspect that proximity to Northern Amerindians will more consistently mean proximity to East Asians.