search this blog

Saturday, March 8, 2014

Ancient North Eurasian (ANE) admixture across Asia


Update 4/9/2014: Ancient North Eurasian (ANE) admixture across Europe & Asia

...

Studies of ancient genomes usually feature unsupervised analyses with the ADMIXTURE software. These are very informative, but only if interpreted in the right context and with caution, because they attempt to fit the ancient samples, often thousands of years old, into ancestral clusters mostly derived from present-day populations. That's like putting the cart before the horse.

So I thought I'd try a different approach, in the hope of achieving more straightforward results, and run ADMIXTURE in supervised mode, with the 24,000 year-old MA-1 or Mal'ta boy genome from South Siberia as one of the reference samples. After a lot of tweaking of the dataset, the experiment seems to have worked, because the cluster created from the ancient genome is basically identical to the MA-1-derived Ancient North Eurasian (ANE) component recently described in the Lazaridis et al. preprint.

Note also that the ANE in my analysis peaks among the Karitiana Indians at around 43%. This is very much in line with a TreeMix graph in Raghavan et al., which shows a Karitiana individual with 41.6% (plus or minus 3.4%) admixture from a clade ancestral to MA-1 (see image here).

Nevertheless, there are clearly some issues with this test. For instance, many South Asians show unexpectedly high levels of Sub-Saharan admixture (in particular, the Austroasiatic samples from India score around 6-7%, which has never been reported before). I'd say this is because they carry genetic variation indigenous to South Asia that doesn't fit well into any of the four ancestral components. The Eastern non-African (ENA) cluster, based on Han Chinese samples, captures most of this diversity, but some of it appears to be siphoned off into the other three clusters. I think the only way to really solve this problem is to include pre-Neolithic genomes from South Asia in the analysis.

By the way, I used 53K SNPs at read depth 2x or more, but varying the quality of SNPs from read depth 1x to 3x doesn't change the results very much.


Key: red = Ancient North Eurasian (ANE); green = Middle Eastern (ME); aqua = Eastern non-African (ENA); purple = Sub-Saharan African (SSA). ANE K=4 ADMIXTURE Test spreadsheet

Citations...

Raghavan et al., Upper Palaeolithic Siberian genome reveals dual ancestry of Native Americans, Nature, (2013), Published online 20 November 2013, doi:10.1038/nature12736

Iosif Lazaridis, Nick Patterson, Alissa Mittnik, et al., Ancient human genomes suggest three ancestral populations for present-day Europeans, bioRxiv, Posted December 23, 2013, doi: 10.1101/001552

See also...

First genome of an Upper Paleolithic human

Ancient human genomes suggest (more than) three ancestral populations for present-day Europeans

118 comments:

Maju said...

Am I wrong or the highest figures are in Central-South Asia, specifically among the Gujarati and Pasthun?

Davidski said...

Karitiana Indians are the most ANE-like group, with around 43% of this component making up their genomes.

Then it's the South Central Asians, like the Pashtuns, Kalash and Gujaratis. But there's a problem with them, because it seems their indigenous South Asian ancestry is inflating their ANE and SSA levels. This is probably because the analysis lacks a pre-Neolithic South Asian reference genome.

Similarly, if I had Eastern and Northern Europeans in this run, they'd be the most ANE-like, but that would be because of their high level of WHG ancestry, which is very similar to ANE.

Nevertheless, South Central Asians do carry a lot of ANE, and this correlates well with their high frequencies of Y-haplogroups R1a-Z93 and R2 (not sure which subclade).

About Time said...

Tres cool. What about Onge or Papuans as another ref point? That could help sort out India.

Seinundzeit said...

This is awesome David, thanks!

Although, it is somewhat sad that the the more accurate test was time consuming. In this one, Admixture deflates my ENA. For your other experiment, I had a perfect 16%, which, for various reasons, I know is very accurate. But here, I'm only around 12%. I had no Sub-Saharan African admixture in the previous test, but here, I'm 2%. And my ANE also went down, from 36%-37%, to 32%. Still, even if this is less accurate, I'm sure the pattern here is very robust. Also, I can finally compare myself with other South-Central Asians. Many Pashtuns, and other neighboring populations, have higher ANE than me for this test, so I'm sure some might score around 40% on your more accurate experiment! There seems to be a serious peak in South-Central Asia. Just as a side note, but is the more accurate experiment just not feasible, because of the sample sizes? This is pretty awesome, but the PCA one was so nice, the values were almost identical to the paper, and here, I think Admixture is somewhat confused for some populations, especially for more peninsular South Asians.

barakobama said...


"Nevertheless, South Central Asians do carry a lot of ANE, and this correlates well with their high frequencies of Y-haplogroups R1a-Z93 and R2 (not sure which subclade)."

I really doubt Indo Iranians brought much ANE to south Asia and parts of west Asia. If Pashuten were ~30% ANE and most of that is of Indo Iranian origin, where is the diverse hair and eye colors that early Indo Iranians had? The people who spread Indo Iranian languages I think were very similar to modern north-east Europeans who have less ANE than Indo Iranian speakers in south Asia and west Asia. My initial guess was that the high diversity of Y DNA P and mtDNA U2 in south Asia is more connected to ANE ancestry in south Asia(or south Asian ancestry in MA-1) than Indo Iranians.

Middle eastern Y DNA R1b has to be connected with ANE. Early European farmers like Stuttgart had no ANE and no Y DNA R unlike modern near easterns. Y DNA R is very absent from areas of the near east(like Arabia) with small amounts of Y DNA R. Of course Y DNA R and ANE in the near east does not correlate perfectly though like in the Caucasus. Did you do this test on the Chad in central Africa who have a very high amount of R1b V88? If they score something like 10% ANE that is great evidence R1b was brought to the middle east by an ANE people.

Maju said...

"What about Onge or Papuans as another ref point? That could help sort out India".

Papuans would be totally useless (no relation to India whatsoever), while the Onge are at best a poor proxy for the pre-Neolithic component (they are from SE Asia, not India proper). The best fix I can imagine is using a "zombie" ASI (Ancient South Indian) component, which is pretty well studied already (it has some vague relation to Onge but it's not the same thing). This "zombie" could be used also in other contexts to estimate the Indian-specific element (ANI is probably Neolithic from West Asia, plus maybe some lesser IE later influence).

Anyhow all three Kalash, Pashtun and Gujarati are Indoeuropean peoples, so there's still a doubt on whether their extra affinity (if confirmed) is intrinsic or rather induced by IE inflows. It seems likely that extra ANE-affinity in Europe is related to Indoeuropean flows, so why not in South Asia too? This could be gauged to some extent at least by comparing with Dravidian peoples.

Davidski said...

I've already tried the Papuans and Melanesians. They don't work any better than the Han, and in fact I'd say they're even worse.

I don't have the Onge, but I can try and get them. I'm also going go try and reduce the SSA noise among the South Asians by switching the SSA reference sample from Yoruba to Mbuti.

Davidski said...

Ahh, to quote myself in regards to another figure from Raghavan et al...

"It's also interesting to note the relatively high position on the list of the Kalash from South Central Asia and Lezgins from the North Caucasus."

http://eurogenes.blogspot.com.au/2013/11/first-genome-of-upper-paleolithic-human.html

Gui S said...

Considering that in your K15 test, Austroasiatic_Ho come up as having no Western Eurasian nor Siberian affinities at all, while they are about 70% South Asian in that test. One can infer that the K15 South Asian component is equivalent to :
-29% ANE
-25% ME
-36% ENA
-10% SSA
from K4.

Using that as benchmark we get :
Burusho
-26.7% ANE
-41.0% ME
-20.0% ASI
-12.3% Others (East and West Eurasian I presume)

Sindhi
-23.3% ANE
-45.1% ME
-30.0% ASI
-1.6% Others

Pathan
-25.5% ANE
-49.0% ME
-22.0% ASI
-3.6% Others

Kalash
-27.0% ANE
-50.5% ME
-22.0% ASI
-0.5% Others

Iranian
-18.3% ANE
-73.9% ME
-5.5% ASI
-2.3% Others

Tadjik
-26.2% ANE
-54.5% ME
-6.0% ASI
-13.2% Others

Davidski said...

Thanks, that looks very reasonable actually. Maybe the ASI is slightly too high, but I haven't read up on that lately so I don't really know?

Interestingly, the Burusho are still way up there with their ANE even after your correction. Here are a couple of famous Burusho dudes who've been making the rounds on the web for several years now. I doubt they're representative, but whatever.

http://i129.photobucket.com/albums/p217/dpwes/Burusho2.jpg~original

http://i129.photobucket.com/albums/p217/dpwes/Burusho.jpg~original

The thing is, the Burusho aren't Indo-Europeans, although surrounded by them. Lezgins are another ANE-rich group who aren't Indo-Europeans, but they actually live very close to the supposed Indo-European homeland in southern Russia.

spagetiMeatball said...

Yeah, but who are "indo-europeans" anymore? they are all mixed. I think the caucasian languages were connected in some way to the ancient north eurasian steppe-tundra languages, as well as uralic and turkic.

Maybe the Dene-caucasian-yenisean language family is not as crazy as people thought?

In any case, it is only afro-asiatic which has no connection at all to Indo-european, neither in phonetics nor vocabulary and all the afro-asiatic speaking folks have little ANE so we are sure that means something. They are mostly of southern eurasian ancestry.

Srkz said...

Davidski, why don't you want to include Selkup and Ket samples from Central Siberia http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE22494 in your experiments. My opinion is, they looks like ~35-40% ANE + 60-65% East Asian. I'v tryed to reconstruct their non-Siberian side, using Eurogenes K36:
7,42 Amerindian
0,30 Central_Euro
1,20 East_Central_Euro
24,53 Eastern_Euro
27,21 Fennoscandian
0,82 North_Atlantic
0,24 North_Caucasian
1,58 North_Sea
0,22 Oceanian
0,08 Pygmy
1,50 South_Asian
7,11 South_Central_Asian
27,79 Volga-Ural

Srkz said...

Selkup in K15
1,95 North_Sea
0,79 Atlantic
1,77 Baltic
28,16 Eastern_Euro
0,00 West_Med
0,69 West_Asian
0,01 East_Med
0,00 Red_Sea
1,89 South_Asian
1,27 Southeast_Asian
57,52 Siberian
5,55 Amerindian
0,35 Oceanian
0,01 Northeast_African
0,02 Sub-Saharan

Seinundzeit said...

I'm beginning to think that ANE ancestry predates EEF/ME ancestry in South Asia. I mean, if South Central Asians+Northwestern South Asians are almost at Native American levels of ANE, and even peninsular South Asians of tribal backgrounds have some substantial ANE admixture (almost matching their ME ancestry), perhaps ANE-like people existed throughout West Asia, and were the first people to mix with ASI hunter gatherers in South Asia? In fact, maybe both groups were hunter gatherers, and these South Asian ANE populations formed an eastern extension of such populations across the Iranian plateau? But eventually, EEF/ME farmers expanded throughout West Asia, and absorbed these people? I think they mention this possibility in the paper, in order to explain why Northern Caucasians have more ANE than Europeans. And if South-Central Asians have more ANE than Northern Caucasians, perhaps we could apply the same logic to South Asia?

Then again, maybe the very idea behind "ASI" involves a conceptual confusion of some sort? Perhaps, South Asia is a place where any distinction between West Eurasian or East Eurasian (or ANE and ASI) doesn’t really mean anything? For example, instead of South Indian tribal populations being West Eurasian-East Eurasian mixes, perhaps they have some sort of basal phylogenetic relationship to both West and East Eurasians? Many physical anthropologists have suggested this in the past, and it might explain some aspects of the uniparental data. For example, the position of South Asia in the differentiation of R lineages, the presence of parahaplogroup F (I'm really behind on this material, it's been quite some time since I've read serious, high quality work on haplogroups, so if I'm making terminological errors, or saying things based on outdated information, please do bear with me), y-DNA haplogroup H, y-DNA haplogroup L, deeply diverged South Asian mtDNA U and mtDNA R lineages, and the huge diversity of mtDNA M in the Indian subcontinent. Not to mention y-DNA Q in Afghanistan and Pakistan, much of which is shared with West Asians, rather than with East Asians. Also, there was a paper that made an argument for the greater genetic diversity of South Indian tribal populations versus Sub-Saharan African populations. This may, possibly, have been a fluke. Still though, it is food for thought.

Maju said...

"Perhaps, South Asia is a place where any distinction between West Eurasian or East Eurasian (or ANE and ASI) doesn’t really mean anything? For example, instead of South Indian tribal populations being West Eurasian-East Eurasian mixes, perhaps they have some sort of basal phylogenetic relationship to both West and East Eurasians?"

You are on the right track: the main South Asian component (i.e. ASI) is not any mix but something specific. Of course if SAs to position in an either/or (.XOR.) West-East Eurasian axis, as happens for example in global PCAs, they will show up as intermediate, but that's a false perception. If you remove Africans (and SAs are properly represented in the sample), they will form their own third pole invariably, even if they surely have greater affinity towards West Eurasians.

Notice that in prehistorical terms, the first UP is almost simultaneous (c. 50-45 Ka BP) in India, West Asia, Central Asia, Europe and NE Africa. This first UP almost certainly represents a single human expansion from somewhere, somewhere in South Asia or not too far away from it.

Y-DNA-wise, South Asia and West Eurasia (incl. Central Asia and the overflow into parts of Africa) share haplogroup P in its various forms (Q, R1, R2 and P*) as well as T (even IJ and G have most likely South Asian ancestors, K and F respectively). In the matrilineal aspect it's even more clear with mtDNA R being clearly most diverse in South Asia but overflowing to both West and East.

The connections are complex because there seems to be also a "Western" root connection to SE Asia (Y-DNA MNOPS, mtDNA N in general) but this one goes necessarily through India anyhow, and then there is the Northern or Siberian corridor flows which were inaugurated by peoples similar to Ma1 ("ANE") and so strongly affect Native Americans (but so little East Asians, excepted Siberian pops.)

But complexities apart, by around 40 Ka ago there were already three distinct main subcontinental populations in Eurasia: one in the West, another in the East and another one in South Asia, which are still clearly distinguishable today. The issue with ANE is: is it closer to West Eurasians or to South Asians or to neither of them (excepting backflow elements).

An important issue here may be the origin of Gravettian culture. Let's not forget that Ma1 was part of that wide culture which spanned from the Bay of Biscay to Lake Baikal and not anymore to the original Altai "Aurignacian" (i.e. Aurignacian-like but not strictly so). Today I was reading Hoffecker 2013, who argues for this important culture to have originated in Eastern Europe (from the same Ahmarian roots that produced the Proto-Aurignacian but not the Aurignacian proper), however Svoboda 2012 argued for its origin in West Asia (also in Ahmarian) and re-expansion from Central Europe. Sadly the early UP in South Asia is not too well understood yet (we know that mode 4 blades were present in Madya Pradesh since c. 45 Ka but we know little of how this population relates to the ones to its West and NW), so we can't but be somewhat cautious even regarding the Gravettian for this area, which could be yet another element of ANE-like flows in whichever direction.

Seinundzeit said...

Hi Maju,

Your whole comment is exceedingly interesting. This is all very fascinating information, and I'm still trying to think this through. Frankly, your ideas here are very important, and I'd like to explore this in more detail, especially the uniparental data, and the archeological angle. I’m truly grateful that you offered your perspective on this. It's nice to see that the thoughts I had germinating in my head weren't too far out.

On a somewhat different note, I think I may have to backtrack on my first suggestion. There is a big problem with my suggestion concerning ANE hunter gatherers preceding EEF/ME people in South Asia. Both MA-1, and his distant WHG relatives, probably had really deep/dark brown skin. People like the Kalash, Burusho, and most Pashtun tribes, don’t really fit that bill. The pigmentation of South Central Asians consistently ranges from “brunette white” to “light brown”, with appreciable percentages of people with “pinkish white”/northern European-like skin pigmentation. So, if ANE is very ancient in South Asia, that might be a problem, since the ANE ancestors of South Central Asians should have been deeply pigmented people, probably of the same hue as ASI hunter gatherers. Although this could fit with Northwestern South Asians, since they are considerably more pigmented in comparison to Pashtuns and company. But even then, Punjabis are much lighter pigmented than other people from deeper inside India.

(For what it's worth, this man basically represents the average Pakistani Pashtun: http://2.bp.blogspot.com/-MVb6Lz14LEA/UhWvOoIHHQI/AAAAAAAAHrI/RNvV-UOaiNM/s1600/Shaista+Khan+Kakar.jpg. His skin pigmentation is very representative. Not too light, but still more "ruddy" than a Punjabi or Sindhi person. And his facial features are also very representative. People in this region tend to have very highly defined/"sharply cut" facial features, accompaniedzir by narrow faces and noses. The general look is very "skeletal", or "bony", if that makes sense. If you look at the two Burusho gentleman posted by David, the faces are of similar shape, but still "fuller". I probably sound like I'm talking nonsense; vague perceptions are hard to put in words.)

If this means anything, I guess that narrows things down. Either the ideas me and Maju share could be developed into the most parsimonious account/explanation offered yet, or ANE is simply the genetic legacy of Indo-Iranian migrations into South Asia. If the latter, the Indo-Iranians must have been predominantly ANE (around 90%-95%), and perhaps they were thoroughly absorbed into the original EEF/ME+ASI mixes that populated South Asia, which is why ANE peaks in South Central Asia? There is some tangential support for this in personal genomics. I have quite a few Polish, Ukrainian, and Russian “relatives” at 23andMe, which greatly confused me when I first got my results. They actually outnumber my Iranian and North Indian “relatives”. And these connections can’t be construed as recent, not by a long shot. My family has been living in the mountainous borderlands of Afghanistan and northwestern Pakistan long before historical memory, so these “relatives” have to represent something ancient and deep in time.

To be completely honest though, I have no clue whatsoever. This is all exciting stuff, but since I don’t have any technical expertise, I’m not sure if what I’m saying is absolutely wrong, or genuinely perceptive. But still, it’s fun to speculate. And I think we’ve now reached a specific conceptual point. We simply can be certain of anything. I guess prehistory was more complex and rich than we could ever have imagined.

About Time said...

@spag, Not true, Afro Asiatic is part of Nostratic meta-family with Indo European and Uralic.

barakobama said...

The ANE percentages of near easterns in this admixture(and Laz 2013) is very constant with "PCA of five ancient genomes". So the percentages may be pretty accurate and can help explain the closeness between west Asian and north European(mainly WHG+ ANE) in globe13.

http://eurogenes.blogspot.com/2014/02/pca-of-five-ancient-genomes.html

You can see near eastern populations gradually moving towards MA-1, and the order is nearly identical to your admixture results with Lezgin being the closest to MA-1. I can't find any south-central Asian populations like Burosho in the PCA's, maybe they would be even closer.

If basal Eurasian ancestry is the reason why Sardinians, Iberians, and north-west Africans are so far to the left in those PCA's I think basal Eurasian ancestry varied in ancient near eastern populations and was especially high in early European farmers. Arabian populations like Bedouin have a very small amount of ANE ancestry they are almost entirely of near eastern(probably a real thing) and sub Saharan(probably a real thing) decent, but they are in the same right-left position as central Europeans and pretty far away from Sardinians(best modern reference to early European farmers). There must be much less basal Eurasian ancestry(or whatever is moving west Eurasians to the far left) in modern near easterns than in early European farmers. That's why south-east Europeans(who have a large amount of post Neolithic near eastern ancestry) are farther down than Sardinians but not as far to the left.

Also I think ANE pushes populations farther to the right than WHG. Many near eastern populations with a high amount of ANE(~20-30%) are farther to the right than north European populations who score over 35% WHG and 15% ANE(in EEF-WHG-ANE admixture). Maybe ANE ancestry in native Americans and Siberians is the reason MA-1 seems to be part east Asian.

I am very excited by how far to the left Basque, other Iberians, and French are in the PCA's. Because admixtures also make it pretty clear that besides Sardinians those European populations have the most Stuttgart like ancestry. North-west Europeans(in my opinion Insular Celts, west Germans, and Norse) jolt to the right(of south-west Euro's) and pretty far up, mainly because they have more ANE but also WHG ancestry but their near eastern ancestry is almost entirely of Stuttgart-like origin. I still think that Indo Europeans during the metal ages are the source of the WHG and ANE modern west Europeans have and Neolithic ones did not.

There is a big north European cluster(because of similar WHG, ANE, and EEF percentages) in the PCA's and its hard to see all the populations, so I use ctrl+f. Gradually east European populations move farther towards the right I think mainly because they have the most WHG and ANE ancestry. Chuvash seem to have a lot of ANE(possibly around 30%) because they are as far to the right as Lezgin.





barakobama said...

"@spag, Not true, Afro Asiatic is part of Nostratic meta-family with Indo European and Uralic."

I wish this was true because maybe we could trace proto-Nostratic to very early west Eurasians and say WHG spoke a Nostratic or related language. But seriously what evidence is there of this? Besides Uralic languages probably originated in an east Asian people in Siberia right?

Maju said...

@Seinuzeit:

"EEF/ME people"

Just in case, notice that EEF are (or were) a mix of West Asian (plus probably some African and/or residual OoA basal Arabian) blood on one side and WHGs. So if ME means "Middle East" it should not be quite the same as EEF, which is already a European-specific mix, which, as such, surely had no impact in Asia whatsoever (or at most in some minor late backflows).

Also notice that the reference individual for WHG in Lazaridis' study, Lochsbour, appears more divergent from modern Europeans than other WHG samples. I estimate that the other WHG samples should appear intermediate between him and EEFs, so probably EEFs, and therefore modern Europeans in general, are actually more WHG-like (in the broader sense, all samples considered) than the Lazaridis' results may suggest on first sight.

"Both MA-1, and his distant WHG relatives, probably had really deep/dark brown skin".

I do not really believe that. I can agree that they may have been darker than modern Europeans but not at all so much. Notice that only one of several (known!) pigmentation alleles (SLC24A5), amounting for just 15% of the influence in the Cape Verde study, was switched to the supposed darker phenotype, if I recall correctly. More than half (by effect) of the genetic determinants of skin pigmentation are still undetermined and about 20% looks not genetic (non-heritable). So even if all four known genes would be switched to the darker phenotype, there is still some 60% uncertainty.

"... the ANE ancestors of South Central Asians should have been deeply pigmented people"...

Really? What do we know of the Mal'ta boy's skin color phenotype? Ma1 should not be much more strongly related to Paleolithic Europeans than to Paleolithic Indians (or at least some of them).

Certainly Siberians and Eastern Europeans should have suffered much stronger selection for a lighter skin phenotype than West Asians or SW Europeans just because of latitude: almost all Palolithic non-Eastern Europe lays below the 50 degrees North, while Altai and parts of Paleolithic Eastern Europe are north of it.

But in any case I think that it's not a "black and white" matter at all. Instead different shades of brown and beige, as well as the capability to tan and "untan" as needed, are involved. For example East Asians lack all the Western pigmentation alleles, but they are still pretty much white (with the expected latitudinal gradation) and almost never anywhere near black. In most cases their ancestors never lived north of the 40 degrees (latitude of Beijing) but they have obviously adapted to lower UV levels anyhow.

Even Native Americans, who must have lived at the Arctic, are quite white (but not pale white), latte color, so to say in, most cases. I can imagine that kind of shades (Mediterranean style) for WHGs but almost certainly not the tropical dark brown color of many Indians.

...

@AboutTime:

Nostratic is a highly speculative and likely wrong linguistic construct. Most versions certainly do not include Afroasiatic, which is a very clearly African family, which spread mostly in the Mesolithic (Capsian, Harifian...)

barakobama said...

Maju, I have been trying to tell everyone(especially Peter Frost) that the skin color of La Brana-1 and Loschbour is unknown, but from what is known about genes and skin color they probably had dark skin. It is good to finally find someone who agrees with me and someone other people pay attention too.

I have also been saying that there are probably many unknown reasons why Europeans(and some near eastern populations) have light skin and we can't assume its all been figured out.

La Brana-1, Loschbour, MA-1, Stuttgart, and Otzi do tell us though that the mutations associated with light skin in Europe did originate in the near east and came to Europe with farming. Even though some north Europeans may have majority WHG-ANE ancestry they are also dominated by these mutations. I have no idea why this is the case, I guess those mutations were selected for some reason.

The main three in genes SLC24A5, SLC45A2, and TYR are all just as popular in near easterns as in Europeans except the one in SLC45A2 which is about 100% in most Europeans and 50% in near easterns, plus Stuttgart and I think Otzi didn't have it. So that had to of been selected and maybe it lightens skin more than the others.

barakobama said...

"Really? What do we know of the Mal'ta boy's skin color phenotype? Ma1 should not be much more strongly related to Paleolithic Europeans than to Paleolithic Indians (or at least some of them). "

Have you seen how close La Brana-1 and MA-1 are in PCA's?

http://eurogenes.blogspot.com/2014/02/pca-of-five-ancient-genomes.html

Have you seen how close WHG(Loschbour?) and ANE(MA-1?) are in this PCA?

http://img9.imageshack.us/img9/7450/oz6i.png

Plus MA-1 was more related to La Brana-1 than to any modern populations. I think MA-1 had a lot of common ancestry with Upper Palaeolithic Europeans and their Mesolithic descendants.


barakobama said...

"Just in case, notice that EEF are (or were) a mix of West Asian (plus probably some African and/or residual OoA basal Arabian) blood on one side and WHGs. So if ME means "Middle East" it should not be quite the same as EEF, which is already a European-specific mix, which, as such, surely had no impact in Asia whatsoever (or at most in some minor late backflows).

Also notice that the reference individual for WHG in Lazaridis' study, Lochsbour, appears more divergent from modern Europeans than other WHG samples. I estimate that the other WHG samples should appear intermediate between him and EEFs, so probably EEFs, and therefore modern Europeans in general, are actually more WHG-like (in the broader sense, all samples considered) than the Lazaridis' results may suggest on first sight."

I totally agree, even with EEF possibly being part sub Saharan. Loschbour was pure WHG, unlike La Brana-1 and the Gotland hunter gatherers who obviously had a tiny bit of farmer ancestry(based on admixture results and some typical farmer mtDNA from Gotland hunter gatherers).

If there is a such thing as middle eastern(including north Africa) which is seems there is, there is a lot of WHG ancestry in EEF. Modern Sardinians are a pretty good representative of early European farmers. In Davidski's thread "PCA of five ancient genomes" Sardinians are much farther up towards MA-1 and La Brana-1 than middle easterns who have small amounts of ANE ancestry like Arabian populations and north Africans.

Look at this quote from Laz 2013.

"To determine whether a mixture of just two ancestral populations can explain the negative f3-
statistics we observe or whether more populations are required, we analyzed f4-statistics5, 22. We
began by analyzing f4(X, Stuttgart; Loschbour, Chimpanzee), which measures whether Loschbour shares more alleles with West Eurasian population X or with Stuttgart (Extended Data
Fig. 4). This statistic is positive for nearly all Europeans showing that Stuttgart has less WHG ancestry than present-day Europeans. However, it is negative for all Near Easterners, suggesting that the ancestors of Stuttgart were not unmixed migrants from the Near East1, 2, 10 (Extended Data Table 1), consistent with the clustering of Stuttgart with Europeans in the PCA of Fig. 1B.We replicated this signal in subsets of SNPs that are uniformly ascertained (Extended Data Table 2). In SI10, we estimate that the proportion of Near Eastern ancestry in Stuttgart is definitely less
than 100% and possibly as little as 61%."

They go one to say she may have had 44.4% Basal Eurasian ancestry(I really doubt its that high). I think western Europeans(and probably also non Balkan east Europeans) near eastern ancestry is almost entirely of EEF origin so they probably have more WHG than what their results say in the EEF-WHG-ANE admixture. There is a lot of modern near eastern like ancestry though in the Balkans and Italy, maybe Y DNA R1b L11 also brought some to western Europe.

barakobama said...

"Even Native Americans, who must have lived at the Arctic, are quite white (but not pale white), latte color, so to say in, most cases. I can imagine that kind of shades (Mediterranean style) for WHGs but almost certainly not the tropical dark brown color of many Indians.
"

Are you talking about "Native Americans" or the east Asian looking people in Canada who are technically also native but have ancestry from later migrations.

Example of a "Native Americans".

http://www.google.com/url?sa=i&rct=j&q=&esrc=s&source=images&cd=&cad=rja&uact=8&docid=HPiRDHHch7d6nM&tbnid=lEcYqvb7slHJRM:&ved=0CAUQjRw&url=http%3A%2F%2Fwww.city-data.com%2Fforum%2Fhistory%2F321462-ever-notice-some-full-blooded-native.html&ei=PBQdU4rvAvP7yAG-kYCICg&bvm=bv.62578216,d.aWc&psig=AFQjCNGRKbT6qaE2F0RqAxYoWJNrv67fyg&ust=1394501046602717

Here is an example of a native from Canada who has mainly recent Siberian ancestry.

https://www.google.com/url?sa=i&rct=j&q=&esrc=s&source=images&cd=&cad=rja&uact=8&docid=mjFuo98_0Jv6NM&tbnid=J5X9G1gSHDmfpM:&ved=0CAUQjRw&url=https%3A%2F%2Faudioboo.fm%2Fboos%2F118159-stamme-der-prarielandschaft&ei=1hQdU5GZGsSMyAGixYDACw&bvm=bv.62578216,d.aWc&psig=AFQjCNFvqLTOts1ZWciYrnBaTwWcA1H67w&ust=1394501150011624

Davidski said...

OK, in this run I replaced the Yoruba with the Mbuti as SSA references to try and reduce the SSA noise among the South Asians. But as far as I can see, the results look basically the same.

https://docs.google.com/spreadsheet/ccc?key=0Ato3EYTdM8lQdGFadDBTZktaTlE2S2J6ZTg5eTBCOVE&usp=sharing

Here's another run, this time with a Ket and Selkup added. I don't think their ANE results are accurate, because Kets and Selkups have Eastern European (and thus WHG) ancestry, which probably inflates their ANE. But it's interesting to note that adding these two individuals increases the ANE among the Karitiana by 1%. In other words, the ANE cluster becomes slightly more Amerindian/Central Siberian.

https://docs.google.com/spreadsheet/ccc?key=0Ato3EYTdM8lQdGpiOXQzamJnckprc09YT3VOd1Iyd1E&usp=sharing

Finally, here's a run with the number of markers increased from just over 50K (at read depth x2 or more) to well over 100K (at read depth x1). Does it look better than the original?

https://docs.google.com/spreadsheet/ccc?key=0Ato3EYTdM8lQdFVyTWRiOVpPWUoxYW9nRTRuZU16Z0E&usp=sharing

By the way, I was just told that the ancient samples from Lazaridis et al. will be available in VCF and eigenstrat binary formats, so in other words it'll be easy to add them to my dataset. Can't wait for that.

Seinundzeit said...

Thanks David!

I think the 100K run is the most accurate. But the changes are extremely small, at least in my case.

K4_ADMIXTURE_Test
ME=53.85%
ANE=31.94%
ENA=11.68%
SSA=2.51%

K4_ADMIXTURE_Test4_x1_depth_read
ME=53.18%
ANE=32.30%
ENA=12.22%
SSA=2.28%

And that is pretty exciting, this should prove to be very interesting.

barakobama said...

"By the way, I was just told that the ancient samples from Lazaridis et al. will be available in VCF and eigenstrat binary formats, so in other words it'll be easy to add them to my dataset. Can't wait for that."

FINALLY!!!!!!!!!!!!!!!!! Also is it possible to put ancient samples into the ANE K=4 ADMIXTURE? If Loschbour and the Motalas score 100% ANE that's more prove MA-1 had a close relationship to Mesolithic Europeans.


Another question: Are you able to find out if a population is more related to Loschbour or Stuttgart? I think WHG ancestry is higher in Europe than EEF-WHG-ANE results because of a high amount of WHG in EEF.

Davidski said...

I won't be able to have a close look at the relationship between EEF, WHG and ANE until I get the Stuttgart and Loschbour genomes, which should be very soon.

But I've just tried a K=5 run of Eurasia with La Brana-1 added as reference alongside MA-1, and the program just kept going and going. I don't think it could solve the problem because it couldn't split ANE and WHG. I'll try again later, maybe overnight, and see what I have in the morning.

Srkz said...

Here's another run, this time with a Ket and Selkup added. I don't think their ANE results are accurate, because Kets and Selkups have Eastern European (and thus WHG) ancestry, which probably inflates their ANE.
Thank you, David! Many Siberians have high Eastern_Euro level, and very low Baltic, North_Sea and Atlantic. At the other hand, Europeans with high Eastern_Euro always carry Baltic, North_Sea and Atlantic. Compare Selkup and Chuvash:
Selkup Chuvash
North_Sea 1,95 13,39
Atlantic 0,79 5,79
Baltic 1,77 13,51
Total 4,51 32,69
Eastern_Euro 28,16 36,15
Maybe Eastern_Euro mostly represents ANE-type ancestry, while other 3 mostly represents WHG-type ancestry?

Srkz said...

Hmmm. North_Sea sometimes also represents ANE-type ancestry:

Selkup Chuvash Burusho Kalash
North_Sea 1,95 13,39 2,29 5,71
Atlantic 0,79 5,79 0,70 1,38
Baltic 1,77 13,51 1,31 2,72
Total 4,51 32,69 4,30 9,81
Eastern_Euro 28,16 36,15 14,77 14,53

Amerinidian 5,55 2,43 2,04 1,49

PS Sorry for bad English

Davidski said...

Yes, all of the Northern and Eastern European-specific ADMIXTURE clusters are part ANE and WHG.

That's why I'm sceptical that there's absolutely no WHG in West and South Asia, and especially in the North Caucasus and the Hindu Kush, where the North/East European ADMIXTURE clusters usually show up and decent levels.

Davidski said...

I just made an update. I'll leave things for now until I can get the Onge samples and the new ancient genomes.

Maju said...

@Davidski:

... "this time with a Ket and Selkup added. I don't think their ANE results are accurate, because Kets and Selkups have Eastern European (and thus WHG) ancestry, which probably inflates their ANE".

Besides the point that I could not recognize significant European admixture among Selkups in previous tests (they quickly form their own Siberian cluster) and that Kets' neighbors the Khanty are archetypal of the Central Asian (or West Siberian) distinctive autosomal component, there is another important problem here that needs us to resort directly to the Raghavan study (the Mal'ta aDNA one), because it is there where we can best discern this issue best.

In the PCA analysis, Siberian populations form a V shape with the vortex towards East Asia, the left branch tending to Mal'ta (not Europe but directly to Ma1) and the right branch tending to Native Americans. The main article does not include any population-specific legend but in the freely available supp. materials the same figure is reproduced with population names (fig. SI 10-A). And it is there where we can see that it is precisely Kets, Selkups and Khanty, the populations that most strongly tend towards Ma1 in all Siberia, even more than Altaians.

There is no deviation whatsoever towards Europe, if anything slightly towards Native Americans in fact. In fact, Aleuts are between them and Ma1.

So I would say that Kets, Selkups and Khanty are among the modern populations most closely retaining the Ma1 ancestry type ("ANE"), although they clearly show intense East Asian admixture, being roughly half-way between Ma1 and East Asians.

This highlights the likelihood of the Central Asian or West Siberian component of Hui Li 2009 being an homogenized mix of Ma1 plus more recent East Asian ancestry (proto-Uralic in essence most probably).

Whatever the case the strong Ma1 affinity seems perfectly justified with no need to resort to tenuous hypothetical European admixture.

Davidski said...

I was very careful to pick a Ket and Selkup with the least European affinity I could find, because some of the ones in my dataset show very strong European admixture. Nevertheless, you can see that the Ket and Selkup here have 8% and 13% ME admixture, respectively. Where did they get that, if not from some European group, like the Andronovo nomads or Russians?

Anyway, what's your opinion about the presence of ME, but total lack of ANE, among some Cambodians and Malays? Where would they get the ME if not from India? In that case, where did their ANE go?

Maju said...

PS- Also in fig. SI 21 (shared drift with Ma1), the sorted scores after NAs are:

1. Naukan (East Siberian)
2. Ket (West Siberian)
3. Lithuanians
4. LSFIN
5. Russians
6. Estonians
7. Mordovians
8. Chutki (East Siberian)
9. FIN
10. Khanty (West Siberian)
11. Maris
12. Selkup (West Siberian/NE European)
13. Koryak (East Siberian)
14. Shors (Altai)
15. Orcadians
16. Ukrainians
17. Tundra Nentai
18. CEU
19. Hungarians
20. Kalash

Maju said...

"you can see that the Ket and Selkup here have 8% and 13% ME admixture, respectively. Where did they get that, if not from some European group, like the Andronovo nomads or Russians?"

Affinity ("admixture" would need to be demonstrated somehow and in this case it's clear that the ME component acts as proxy for others in essence). But, well, in any case, the ME affinity is much much lower than among the Kalash (8% vs 57%), so even if it represents Eastern European admixture, it is quite low.

Somehow it seems to me that you're trying to push the idea that ANE=Eastern European. And that's clearly a wrong idea in the general terms, even if there is a kernel of truth in what regards to Europe, where Eastern European (or IE) genetic influence seems clearly associated with greater ANE. Instead, regarding Siberia and Central Asia the situation seems quite differen and probably lower ANE means greater East Asian or other admixture. East European admixture is more likely to result in no significant effect in the Ma1 affinity element, considering that Siberians and East Europeans have similar apportions of it.

Davidski said...

I suppose that the Kalash got most of their ME directly from the Middle East, not from the southern Urals, where presumably their ANE-rich Indo-Iranian ancestors came from. But the ME carried by the Selkups probably came to Western/Central Siberia via Europe, right?

Anyway, I'm not denying that the Kets and Selkups have unusually high affinity to MA-1. What I'm saying is that this affinity is inflated a little, or a lot, by varying degrees of Eastern European admixture, which they surely have, and possibly also non-ANE Siberian ancestry that is ancestral to the East Asian type of ancestry carried by Amerindians.

Why do you think I didn't put any Lithuanians in this analysis? Their extremely high WHG would make them more ANE-like than the Karitiana, which wouldn't be right. It could also skew the whole analysis, by changing the ANE cluster into a partly WHG cluster.

DOD879 said...

@David "Anyway, what's your opinion about the presence of ME, but total lack of ANE, among some Cambodians and Malays? Where would they get the ME if not from India? In that case, where did their ANE go?"

I don't want to make too much out of admixture data (drift, recombination, selection...) and would rather trust ancient genomes. But a hypothetical situation for the ME in SE but no ANE could go as far back as human migration itself:
Migration from Africa-> ME-> India -> SE Asia -> India -> Siberia -> India

The recent South Indian adventure into Cambodia probably did add some admixture, but may not have been big, or they were washed out over the centuries.

About Time said...

@David, is there any way to apply this analysis (same clusters) to Europeans to see how they perform?

I'm guessing they'd come out ANE+ME, but would be good to see. Could also be a test of the strange African % that pops up in Finland/Lithuania in the LD stuff.

Davidski said...

I can't include Europeans in this test, because they'd change the character of the ANE cluster to something resembling WHG. But I did add Turks to the latest run, and the results look fine. Here's the link...

https://docs.google.com/spreadsheet/ccc?key=0Ato3EYTdM8lQdG1KU0RzN2dVMHFsYVM3NHpsYXNVTHc#gid=0

In any case, the K=5 of West Eurasia with La Brana-1 and MA-1 as separate clusters seems to have worked out. The results appear very reasonable anyway. I'll post that spreadsheet later.

Grey said...

"Yeah, but who are "indo-europeans" anymore?"

If the model is a sequence of "Out Of x" events where later expansions mostly (but not completely) over-write previous expansions and you assume a sequence:

1) Out of Tropics
2) Out of Africa
3) Out of India

where the OOI population expands over most of Eurasia (with some OOA surviving either as pockets or as admixture) and then imagine step 4 has two origins i.e.

4a) Out of Middle-East
4b) Out of South China

and you imagine those two expanding from their respective origin until they hit the Himalayas (over-simplifying here) then what would remain of the OOI population distribution?

1) India itself
2) Northern Eurasia

with the OOME and OOSC in between.

So maybe some kind of boomerang took place?

OOI -> ASI & ANE
ANE -> IE
IE -> ANI

(This sequence might also cover the presence of the ME component mentioned in some of the earlier posts i.e. the ME component might actually be the remnant of the OOA component from the expansion before OOI.)

barakobama said...

Davidski, based on alleles in SNP rs12913832 it seems the Indo European kurgen people in your home turf during the Eneolithic and bronze age were mainly EEF or just simply near eastern.

http://dienekes.blogspot.com/2014/03/dark-pigmentation-of-eneolithic-and.html

Out of 94 samples only 16% had the "blue eye" mutation in SNP rs12913832(unlike modern Ukraines who have 65% and (north)Europeans who have 71%), 43.2% had the light skin mutation in gene TYR(more than modern Europeans and near easterns) out of 44 samples, and only 43% had light skin mutation in gene SLC24A5(modern Europeans have about 100% except for Sardinians who have a little over 50% and is more than modern near easterns who have about 50%).

According to SNPedia if you have C,C or C,G alleles in SNP rs16891982 and your European you are 7x more likely to have black hair. 67% of these ancient east Europeans had it. Stuttgart had C,C in SNP rs16891982 and modern Sardinians have the lowest amount of light skin allele G in Europe.

The dark eyes but likely light skin remind me of this.

http://dienekes.blogspot.com/2013/06/ancient-steppe-populations-hints-of.html

And of course of Stuttgart, Otzi, and modern Sardinians. So far ancient DNA shows early Indo Europeans were probably mainly EEF. Why do Indo Europeans seem to be the source of the rise of WHG and ANE and where did the mainly light eyed bronze age Siberian Kurgen people come from? Another good question is why are modern Russians, Ukrainians, Finno Urgics, and Turks in that area mainly WHG+ANE not EEF? I am just basing their EEFness on eye color though modern Finno Urgics and Turks in that area are mainly dark eyed(~60%) despite being mainly WHG+ANE.

I can't wait to get DNA results from those ancient Pontiac steppe people, I think there is a good chance they were very similar to Stuttgart or to modern south-east Europeans. Their Y DNA results will also probably be interesting maybe they had mainly haplogroups more typical to the Balkans and that probably are of near eastern origin.

Tobus said...

@Maju: Notice that only one of several (known!) pigmentation alleles (SLC24A5), amounting for just 15% of the influence in the Cape Verde study, was switched to the supposed darker phenotype, if I recall correctly.

You recall incorrectly - Loschbour and La Brana both have double copies of the ancestral (dark) versions of SLC24A5, SLC45A2 and TYR... showing the same configuration of known pigmentation alleles as Sri Lankans, Melanesians and Australia Aboriginals. Mal'ta is the same but with no coverage at SLC24A5.

Maju said...

You're right Tobus. Now let's see what we know of those alleles:

· SLC24A5 (15% weight in Cape Verde pigmentation study), today nearly fixated in Europe, correlated with skin color in a study from India.

· SLC45A2 (<10% weight in Cape Verde pigmentation effect), today still variable in Europe, correlated with skin color in India and Europe

· TYR (<10% weight in Cape Verde pigmentation), in Europe NOT correlated with skin color but only with freckles, correlated with skin color in India.

Together they weight just ~30% in the Cape Verde study, the rest are other genes (of which only one is known and has little weight) and a 20% of non-genetic, non-inheritable variance.

Oddly enough Lochsbour had blue eyes (this is brutally more certain from genetic parameters, being the only pigmentation trait for which most of the genetic causes are known).

As I said before East Asians are white and do not have any of those Western alleles (or so little that they are irrelevant). East Asian depigmentation pathways are still unknown but they obviously work very well (even if they are not as exaggerated as some Western variants and seem quite lower in pheomelanin, the red pigment that gives a ruddy shade to so many Western skins).

Not only East Asian pathways are unknown but more than half of the Western genetic pathways to depigmentation are also unknown so far.

My big question is: considering that vitamin D, almost only generated in the skin under UV effect, is so crucial for so many body functions but, critically, for early brain development, do you really believe that people could be living generation after generation in Altai or Ukraine or even in Southern France without experiencing some marked depigmentation? I do not. Also why do you think that these alleles would only be selected in relatively sunny areas such as West Asia or Greece (needed to explain EEFs carrying them and not WHGs)?

I would rather think that various genetic pathways to depigmentation were at play although for some reason, maybe just numbers?, the EEF one became dominant eventually.

About Time said...

@Grey, I wonder if OOI (Out of India) was more out of Central Asia. India seems like a migratory cul de sac bc of difficulty passing over mountains to get in/out.

But lots of migrations started out near Khorasan/Central Asia. In better climate, steppe could be reasonably ok for game hunting, with rivers to support camps or settlemts. In arid spells everyone would empty out bc it would all turn to desert (like Aral Sea right now).

About Time said...

@Maju, we so far know EEF from Stuttgart, so not ultra sunny. Maybe EEF underwent evolutionary selection closer to there than to Greece (but without access to seafood maybe - or allergies?) and became a specialized type of "cloudy weather agriculturalist."

The new paper showing dark pigmentation in Kurgans 3000-2000 bc mentions seafood as dietary factor preventing need for depigmentation in northern non-farmers.

I'm trying to remember who it was that later on lived by the sea and never ate much seafood (strangely). Maybe Mycenaeans, but I can't find a quick reference offhand.

Seinundzeit said...

Hi David,

Is it possible for you to make a GEDMatch calculator with these results?

Maju said...

@About Time:

India (South Asia) is no "cul-de-sac" but actually the only practicable corridor between East and West Asia before the colonization of Altai, what only happened at the beginning of the UP (c. 47 Ka ago).

Not only it is a "corridor" but it was also the first sizable affluent land that Homo sapiens found when migrating out of Africa. First, between c. 125-100 Ka., they settled in Arabia and the Persian Gulf (then an emerged marshland) but their numbers must have been limited because of the small area implicated and its relative aridity, even in a pluvial period. But since c. 100 Ka ago. we see clear evidence of H. sapiens in South and SE Asia and they almost certainly did not go via Altai, not yet (all evidence points to only other species living there so early).

So after the OoA there was an Asian expansion with initial centrality in South Asia and with a secondary center in SE Asia. A tropical species colonizing tropical and subtropical areas quite rapidly. From there people went to Australasia (quite early it seems, although the exact dates are still debated), to NE Asia and to West Eurasia. These last move corresponds to the initial Upper Paleolithic, since c. 50-45 Ka ago and is also partly at the origin of Native Americans and Siberians via Altai.

... "we so far know EEF from Stuttgart, so not ultra sunny. Maybe EEF underwent evolutionary selection closer to there than to Greece"...

Why would selection act so fast on recent arrivals and not on long time residents like Lochsbour. I rather think that they carried the alleles from farther SE, unless there are still hidden surprises.

"I'm trying to remember who it was that later on lived by the sea and never ate much seafood (strangely)."

Early Neolithic Britons: they switched to milk and never again ate much fish until the Danelaw: http://forwhattheywereweare.blogspot.com/2014/02/neolithic-peoples-from-britain-and.html

Grey said...

@About Time

"I wonder if OOI (Out of India) was more out of Central Asia."

I dunno. I think OOI makes the most sense for the 3rd step in the sequence (for the reasons Maju gives) but I am sure the steps after that are more complex than I described. That was just to illustrate the idea of how the same base population might be split in two by later expansions.

Tobus said...

@Maju:
Together they weight just ~30% in the Cape Verde study

Note that the SLC45A2 and TYR mutations identified in multiple other papers (rs16891982 and rs1042602) aren't the same ones as in the Cape Verde paper (rs35395 and rs10831496), so those 2 sites aren't included in this estimate. The Belezal paper comfirms the role of SLC24A5 and suggest 3 new sites (plus general ancestry) as factors, but as far as I know there's been no confirmation of the role of these new alleles.

It's also worth noting that the "x%" in Belezal is a little misleading compared to other papers as they are reporting the percentage of the total variance, whereas papers like Lamarson 2005 calculate the percentage of the average difference. Both agree that the absolute effect of a single SLC24A5 allele is roughly the same, they are comparing it against a different baseline.

As I said before East Asians are white and do not have any of those Western alleles

The genetics of depigmentation in East Asian are largely independent to those of Europe, the only directly associated allele is rs1800414 in OCA2 (which incidently, doesn't have a North/South cline as we'd expect from UV, but an East/West one, centered in southeastern China), so a comparison between the genetic configuration of East Asians and Europeans is not much use in terms of skin colour.

do you really believe that people could be living generation after generation in Altai or Ukraine or even in Southern France without experiencing some marked depigmentation?

It's what the data we have to date indicates, so yes, I really believe it. Consider that you don't just magically turn white if you move to a high-UV area, there has to firstly be a functional mutation that lightens skin tone, and secondly it has to make you reproduce to a significantly higher degree than everybody else. It's likely any vitamin D benefit only became an issue in this regard when diet changed. There are many pigmentation mutations like albinism and red hair/freckles that have *not* been selected for due to negative side-effects (skin cancer being the obvious one), so it seems that a random mutation that significantly increases vitamin D intake but still maintains enough UV protection to be functional is a rare mutation indeed. I see no reason why modern humans couldn't have remained dark-skinned for a long time before such a mutation arose, and before such a mutation made such a significant reproductive difference that it became fixed (ie within the last 10,000 years).

As I've pointed out in a number of other discussions, a large part of the rejection of this data is a preconcieved notion that Europeans have always been "white people". If this same DNA was retrived from a NY crime scene this morning, the police would be looking for a black man, no questions asked.

barakobama said...

"As I've pointed out in a number of other discussions, a large part of the rejection of this data is a preconcieved notion that Europeans have always been "white people". If this same DNA was retrived from a NY crime scene this morning, the police would be looking for a black man, no questions asked.'

Well actually if they would say this person is non west Eurasian. Plus skeletal features and genetic DNA testing are important. I have not heard many people say Europeans have always been pale skinned, before any of this new ancient DNA everyone was saying blue eyes are 6,000 years old, light hair is 6,000 years old, light skin is 6,000 years old.

The reason it is hard for me to believe Mesolithic Europeans did not have light skin, is because there is strong correlation between light hair-eyes-skin and Mesolithic ancestry today. Light hair-eyes are connected there are states to prove it so for that reason it makes sense to me light hair existed in Mesolithic Europe and therefore light skin probably did too.

I don't know how to explain an EEF origin of light hair, and why Finnish who are some of the most WHG+ANE descended Europeans have lighter skin than Sardinians who are the closest modern relatives to early European farmers.

The pigmentation results of Eneolithic-bronze age people from Ukraine, Bulgaria, and Russia are beyond unexpected. They had around the same eye color percentages as northern-near easterns but much lower percentages of rs1042602 A/A than modern Europeans and near easterns, and around the same amount of rs16891982 G/G as modern near easterns. They also had a high amount of mtDNA U5 so I really doubt they were a typical modern near eastern population, but they possibly were mainly EEF.

When did all those light skin mutations become dominate in Europe and why are they dominate in so many different ethnic groups if just 5,000 years ago they were not dominate in some? The whole theory of EEF-hunter(WHG+ANE) mutts being the origin of Euro-paleness does make some sense when considering ancient DNA but still there are inconstancies.

Davidski said...

If these groups carried a lot of Y-hg R, which I expect they did, due to male mediated admixture from the northeast, then they were significantly ANE.

Their mtDNA structure is very similar to that of the Unetice samples from Germany, so not exactly the best match for EEF we've seen to date. If you want to know what EEF mtDNA was like, then look at the stats for LBK here.

http://i129.photobucket.com/albums/p217/dpwes/Ancient_mtDNA_PCA.png~original

http://i129.photobucket.com/albums/p217/dpwes/Ancient_mtDNA_table.png~original

Maju said...

@Tobus:

How can you explain that the Cape Verde study, which is so systematic in locating active sites, finds markers that are different than in other studies? For me it is a key unavoidable reference because it's the only one directly studying a Euro-African mixed population and one in which the Euro component is not North European but West Iberian (Galician and Portuguese).

While Iberians are not totally lacking the extra ANE influence attributable to IE migrations, they are in the lowest ranks and most of their ancestry should be explained in terms of pure WHG-EEF mix. Yet Portuguese skin color (phenotype) is similar to that of Polish (darker than Irish but lighter than Italians) [→ see here].

In that study (Candille 2012) it was obvious that skin color variation in Europeans is poorly explained by Genetic variance, at least anything they could identify:

The fact that we did not detect reproducible associations with skin or hair color suggests that, unlike eye color, skin and hair pigmentation variation in Europe are not determined by major loci.

"The genetics of depigmentation in East Asian are largely independent to those of Europe"...

What I'm trying to explain is that the genetics of depigmentation in Europe or in general West Eurasia are not well understood either. Also, by pointing to East Asia (which I agree should be a different evolutionary process), I meant to indicate that there is a lot of depigmentation genetics which we know nothing or nearly nothing about and that similarly unknown (but surely distinct) genetic pathways may well have been present in West Eurasia and specifically in Europe and Central Asia/Siberia, where the pressure towards depigmentation must have been overwhelming.

A fish rich diet alone cannot explain anything because there will be always groups with poor access to such vit. D rich foods and/or periods in the history of every population in which such resources were not easily available. Also judging only on such speculative argument, the darkest populations in Europe should be the Danes, who have always got a fish-rich diet.

"It's what the data we have to date indicates, so yes, I really believe it".

It is what certain genetic data, interpreted in certain way, may suggest. As I have tried to explain the certainty is very low in fact.

...

Maju said...

...

"Consider that you don't just magically turn white if you move to a high-UV area"...

Low UV area, right? You don't turn white or more precisely whiter, because it is a matter of gradation and not just black and white, not yet. But those who have lighter shades already will have much healthier offspring (in many aspects but crucially in brain development) and soon the population will have forcibly evolved to lighter shades even with no novel mutation implicated (just selecting withing the available diversity). This probably happened in the matter of few generations, centuries, so in the time of dozen millennia (UP chronology) the evolution must have been even much sharper.

"It's likely any vitamin D benefit only became an issue in this regard when diet changed".

I can't but seriously doubt the fish diet as miraculous evolutionary blocker. It did not work that way in East Asia and should not in Europe nor Central Asia either. Even if it may have been of some help in the adaptive process, it is not something you can rely on so clearly as happens with direct skin photosynthesis.

"... so it seems that a random mutation that significantly increases vitamin D intake but still maintains enough UV protection to be functional is a rare mutation indeed".

Speculative. In fact there seem to be many such mutations, judging on the difficulty to identify them on the genetic landscape. Also it was a matter of life and death and we know that in such cases evolution gets its job done (otherwise death and extinction).

"... a preconcieved notion that Europeans have always been "white people"".

It's not a matter of "preconception" but of survival. If it would not be such a crucial issue, whiter shades of skin color would never have evolved (what for?) It is however the most clear case in all human biology in which extremely strong evolutionary pressures are central. It is almost unavoidable that peoples adapt skin-color-wise to their latitude, there's no way around it.

"If this same DNA was retrived from a NY crime scene this morning, the police would be looking for a black man, no questions asked".

Nope: they'd use AIMs (ancestry identification markers) and that would point clearly to Europe. Irrelevant anyhow.

Seinundzeit said...

Although computationally intensive stuff, a fineSTRUCTURE/ChromoPainter run with MA-1 could be very informative. As Falush notes at his blog:

"The fineSTRUCTURE model we published earlier this year is very similar to the STRUCTURE model, except that instead of assuming that a population is characterized by a set of gene frequencies, individuals in each population have a characteristic rate of sharing genetic material with individuals in their own population and with individuals in the other populations. This turns out to be a more general model because sharing of genetic material can be considered in terms of sharing tracts of DNA rather than individual markers. This allows it to take into account genetic linkage, a pattern of heredity that is a consequence of the fact that DNA is a linear molecule organized into chromosomes. It also turns out to be a much more computationally tractable model for datasets with large numbers of genetic markers because there is no need to estimate the genetic composition of the populations at each marker."

I think this sort of model could work better when dealing with MA-1 and contemporary populations. Also, it would be very interesting to see what "region" he clusters with.

Seinundzeit said...

Ah, I just found this ChromoPainter/fineStructure analysis of La Brana 1, by Anders Pålsen:

http://fennoscandia.blogspot.com/2014/03/la-brana-1-closest-to-basque-sardinians.html

Using the unlinked model, which is supposed to reveal broad genetic affinities, La Brana 1 clusters with Finns and Saami people. Anders also notes an affinity to Lithuanians, Basque, and Scandinavians. In the linked model, which is supposed to be more powerful for genealogical connections, La Brana 1 clusters with West Europeans. Makes sense. I think he analyzed La Brana 1 with only European populations. Since ANE ancestry probably peaks outside Europe (likely in South Central Asia+Northwestern South Asia, followed by the Caucasus), perhaps a similar analysis of MA-1 with Asian populations could prove to be very interesting?

Davidski said...

When I'm doing these ADMIXTURE tests, I know that the Karitiana Indians should be around 40% ANE, Lezgins around 28%, etc., so I can make corrections to the analysis when modern genetic drift is skewing things, and then be fairly sure that the rest of the results are in ballpark range at least. But the problem with doing fineSTRUCTURE runs on the ancient samples is that there's nothing specific to cross check the results with.

Seinundzeit said...

A very good point. With no real reference point to cross check the results, I suppose interpreting the fineSTRUCTURE output could be problematic.

Tobus said...

@Maju:
It is what certain genetic data, interpreted in certain way, may suggest. As I have tried to explain the certainty is very low in fact.

I agree that we don't know everything about pigmentations in genetics but what we *do* know tells us a lot. We may not know what causes the small differences between Europeans (~5% of the range used by Beleza), nor what causes the large range of dark skin phenotypes in sub-Saharan Africa (~40-50 of this range), but we do know, for certain, that each allele (or LD equivalent) of the SLC24A5, SLC45A2 and TYR genes (and possibly ABPA2) reduces skin pigmentation to a measurable degree. In total these alleles amount to ~30% of the global range, which is ~23 melanin units - some 75% of the average Afican/European difference used by Lamarson. Applying this to a gradient of visible phenotype (say the global readings in Jablonski 2000), this means a sample homozygous for the ancestral version of these 3 sites, assuming all the unknown genetic components are the derived light-skin versions, will have a skin tone darker than modern Native Americans and close to the middle of the range seen in modern South Asians. If any of the "unknowns" were ancestral as well then the skin tone could well have been darker than this.

You are focussing on the unknowns and saying "we can't tell", but the bottom line using the "knowns" tells us enough - these samples had skin that is considerably darker than what we see in modern Europeans populations, probably as dark or darker than what we see in modern Indians.


How can you explain that the Cape Verde study, which is so systematic in locating active sites, finds markers that are different than in other studies?

The explain it themselves, the SLC45A2/TYR in other studies are probably in LD with the SLC45A/TYR with the ones they found, and KITLG either has too small an effect to be noticed or was due to design errors in the original experiment (that latter makes sense to me since all non-Africans have it including Melanesians and Australian Aboriginals)

Low UV area, right?

Yes, my mistake.

I can't but seriously doubt the fish diet as miraculous evolutionary blocker. It did not work that way in East Asia

East Asia doesn't show a North/South cline - perhaps UV wasn't the primary or only selective force at work?

Nope: they'd use AIMs

They'd probably use 8plex, at least in the first instance. I think it's relevant because it shows that priors are at play here - if the samples were in India not Europe would you be questioning the skin colour?


Grey said...

@Tobus

"a large part of the rejection of this data is a preconcieved notion that Europeans have always been "white people"."

That may be true but personally I'm sure they looked a lot like people from India when they arrived but I'm not convinced how long they stayed that way.

.

"There are many pigmentation mutations like albinism and red hair/freckles that have *not* been selected for due to negative side-effects (skin cancer being the obvious one),"

Well that's the point. If you paint a white wall, white then you won't notice. Was an earlier flawed version of skin lightening selected for originally until an improved version came along later?

(incidentally turning the red hair multi-colored in the process)

.

"if the samples were in India not Europe would you be questioning the skin colour?"

Did the earliest written sources describe Indians as being particularly pale skinned?

Maju said...

@Tobus:

"East Asia doesn't show a North/South cline"

I would say it does (Koreans, Japanese and North Chinese are generally "whiter" than South Chinese and these generally "whiter" than at least some SE Asians), however the cline is (barring Negritos) more "caramel to cream" than "chocolate to cream". This may imply some sort of "recent" mass flow southwards (unconfirmed, but there's also replacement known to have happened in northwards direction after Neolithic).

In Europe there's not a clear N/S cline either. As I explained before Portuguese and Poles are about the same shade of skin color (although Poles are generally blonder in hair color and Portuguese tend to have greater frequencies of dark eyes, what matters here is skin color, all the rest being seemingly under no selective pressure whatsoever, unless it is some sort of cultural one). The cline seems more SE to NW in fact, what is a bit counterintuitive when we consider that the alleged light skin alleles were carried only by EEFs, whose genetic influence is rather the opposite.

"... if the samples were in India not Europe would you be questioning the skin colour?"

I would question in most cases because we simply do not know enough of skin color genetics. However India is mostly a tropical region, so the environmental pressures are very different and therefore the hypothesis would sound much more reasonable.

"You are focussing on the unknowns and saying "we can't tell", but the bottom line using the "knowns" tells us enough"...

They do not. Because what we really don't know too much on this matter yet.

In Europe there are native people with darker complexions (not black but certainly "brown" -caramel, olive-, I have a couple of cousins like that) and also with extreme hyper-white phenotypes. If known genetics were at play, it'd be possible to discern these phenotypes from the known alleles, however it seems it's not possible or only very approximatively so.

BTW, I have to insist that TYR in Europe only seems related to freckles and not general skin color. Only in India (and to a lesser extent, Cape Verde) it has been "found" to be correlated with skin color but a confounding factor is probably at play there.

Tobus said...

@Manu:
I would say it does (Koreans, Japanese and North Chinese are generally "whiter" than South Chinese and these generally "whiter" than at least some SE Asians)

Yet Siberians, Aluets and Alaskans are all darker... Jablonski (2000) fails to show a N/S cline when taking objective measurements of unexposed skin, and the distribution of the known lightening allele OCA2 is decidedly not centered on the North. Many images of "white"-looking Japanese and Chinese are due to makeup, skin whitening or extreme avoidance of sunlight so I don't have much faith in subjective assessment of skin colour in East Asians... objective measurements show the darkest Europeans are lighter than the lightest East Asians.

Because what we really don't know too much on this matter yet.

We have multiple papers by multiple labs all saying the same thing... how much more evidence do you need? If we know for sure that certain alleles reduce skin colour by X melanin units, it's common sense that a person with those alleles has to be at least X melanin units lower than the lightest person on record... if that puts them in the region of South Asians then that's the lightest we can expect them to be, regardless of what the unknown elements are.

In Europe there are native people with darker complexions (not black but certainly "brown" -caramel, olive-, I have a couple of cousins like that) and also with extreme hyper-white phenotypes.

All the studies I've seen that take controlled measurements of skin colour (multiple samples from unexposed skin, such as Candille) report only a very small variance of skin colour in Europeans. I suspect a large part of this subjectively assessed "olive" vs "pale" difference is due to environmental effects. The effects of tanning are not included in the genetic effects measured in Belezal (which are all based on unexposed skin), so consideration of tanning ability can only make the Mesolithic samples darker, not lighter.

Tobus said...

@Manu (cont):
BTW, I have to insist that TYR in Europe only seems related to freckles and not general skin color. Only in India (and to a lesser extent, Cape Verde) it has been "found" to be correlated with skin color but a confounding factor is probably at play there.

Shriver (2003) found it significant in African-Americans too, even after they controlled for individual ancestry.

Tobus said...

@Grey:
Did the earliest written sources describe Indians as being particularly pale skinned?

The earliest written sources are a good 4,000-5,000 years later... more than enough time for a strong selective sweep to take place. Not to mention that descriptions from early writers aren't always 100% accurate or consistent - if they contradict the genetic data, they are probably exaggerating or stereotyping.

About Time said...

@Grey "Did the earliest written sources describe Indians as being particularly pale skinned?"

Source: The Aryan Invasion Theory: A Reappraisal by Shrikant Talageri. (Even if you disagree with the author's conclusions, the arguments and evidence he musters are unusual and worth a look).

There is no evidence anywhere in the Rigveda to show that the Vedic Aryans called themselves fair-skinned in contrast to their "aboriginal" enemies.

Bhargava cites I.100.18. However, in this verse, the composer does not refer to the Vedic Aryans, but to the Maruts who (like their father Rudra) are often described in the Rigveda as being white or red in complexion. The line goes: "The mighty Thunderer with his fair-complexioned [svitnya] friends won the land, the sunlight, and the waters"... Bhargava's contention that the verses describe the Vasissthas and Trtsus as white-complexioned is not borne out by the verses which in fact describe the hair-style and the white robes of these warriors.
(p. 199-200)

Rudra is the Rigvedic God who is generally accepted as a "non-Aryan" God borrowed by the Aryans from the naties. ... Ironically, the Rigveda repeatedly stresses (e.g. throughout hymn X.96) that Indra is greenish-brown in colour. Vishnu is also known to be dark or dark-blue in colour. On the other hand, Rudra (II.33.8) is described as white-complexioned. (p. 202)

A.D. Pusalker chooses "Rama and Krisna..." as the "true types of Aryan heroes who were pioneers of the spread of Aryan culture and civilization all over India". Again, we find that the ancient texts, and the entire range of Indian tradition pertaining to Rama and Krisna, are unanimous in describing them as being dark in colour. (p. 203)

@Maju (we discussed fish diets in another thread). I found a reference while looking into Grey's question:

Earlier, Gimbutas also points out that 'the presence of bone or antler harpoons, points, fish-hooks and fishbones in Kurgan villages indicates that fishing was an important means of subsistence.'

And here we find a major point which not only does not show a "splendid coincidence", but in fact contrasts sharply with the linguistic reconstruction of proto-Indo-European culture. The Indo-European languages do not have a common word for "fish." Childe also points out: "fishing is never mentioned either in the Veda or the Avesta, and the repugnance felt by the Homeric Greeks for a fish diet is notorious," and also: "it is notorious that early Aryans, even in a maritime region, eschewed a fish diet."
(p101-102)

By the way, "fish" is one of the few IE type of words that shows up as a borrowing in Sumerian. http://dienekes.blogspot.com/2011/05/case-for-euphratic.html

About Time said...

Sorry to double post. Two more things come to mind.

Achaemenid art (Persian c.500 BC) clearly shows dark skinned people.

http://www.historyandcivilization.com/Picture-Gallery---Achaemenid-Empire---Sculpture---Friezes.html

But, ancient Bactrians used light stones to show their skin color. Artistic license or just a natural use of color, who knows.

http://commons.wikimedia.org/wiki/File:Kaunakes_Bactria_Louvre_AO31917.jpg

http://commons.wikimedia.org/wiki/File:Margiana_or_Bactria,_goddess_Nana_.jpg

http://commons.wikimedia.org/wiki/File:Seated_Female_Figure_LACMA_M.2000.1a-f_%281_of_3%29.jpg

http://commons.wikimedia.org/wiki/File:BMACFemaleHead3rd-Early2ndMilleniumBCE.jpg

The garment has been compared to a Sumerian kaunakes. http://upload.wikimedia.org/wikipedia/commons/thumb/0/0b/Standard_of_Ur_901bis.jpg/220px-Standard_of_Ur_901bis.jpg

georgy chich said...

@Tobus:
"Many images of "white"-looking Japanese and Chinese are due to makeup, skin whitening or extreme avoidance of sunlight"

You're not right. Look at this Japanese girl without makeup http://img-fotki.yandex.ru/get/9260/208125452.bc/0_cad9f_c33b364a_orig

or legs of this girls http://img-fotki.yandex.ru/get/3601/tihiro191.1/0_99dd_1eb5081d_orig

now compare with Greeks
http://img-fotki.yandex.ru/get/42/captainalex2007.c/0_16765_cef3937a_orig

http://img-fotki.yandex.ru/get/4007/vladinetta.11/0_39ff8_2c82d353_orig

Grey said...

@Tobus

"The earliest written sources are a good 4,000-5,000 years later... more than enough time for a strong selective sweep to take place."

My point was that no-one quibbles with the idea of ancient Indians being dark because there's no reason to think otherwise.

There are two reasons to quibble when it comes to Europeans: 1) current coloring, which I agree doesn't prove anything and 2) written records which if accurate, push back the time available for this selective sweep to have happened.

For example, the recent Kurgan paper in the other thread talks about strong selective pressure over 5000 years but it's not 5000, it's closer to 2000 and from a selection pressure - a switch to agriculture - which ought not to apply to pastoralists or at least not as much.

That doesn't prove it didn't happen but...

Also if the ANE were brown and the WHG were brown then as the ANE moved west to the Baltic they would get darker and darker until the Sardinian farmers arrived and made them blond...

It doesn't make a lot of sense to me.

.

On the other hand I can imagine

Indian

http://postgradproblems.com/wp-content/uploads/2013/10/00500f1f2e2adff7cc657558614d24e4.jpg

->

red phenotype through some kind of partial albinism

http://i.imgur.com/Ew5G7OL.jpg

and then red phenotype plus Sardinian

http://upload.wikimedia.org/wikipedia/commons/b/b8/Folk_Costume_of_Sardinia_in_Oliena_5.jpg

->

average euro


.

@About Time

If the steppe dudes came down in a sequence of raid and expand conquests over a long time period that would explain the striking mismatch between y dna and mtdna but it might also imply that they got darker on the way south so

Tobus said...

@georgy:
You're not right. Look at this Japanese girl without makeup

Subjective interpretations from images have no weight against objective measurements in controlled conditions. Lighting conditions, contrast settings, colour adjustments etc. all affect the final colours and there is no control from the genetic history of the subject nor their relative exposure to sunlight or skin-care regimes. We can all cherry-pick images to make our point, but unless it backed up by controlled objective measurements it's a subjective opinion, not a scientifically verifiable fact.

@Grey:
My point was that no-one quibbles with the idea of ancient Indians being dark because there's no reason to think otherwise.

That's exactly my point too - it's our preconceived theoretical notions that make us doubt this data, not the scientific validity of it.

For example, the recent Kurgan paper in the other thread talks about strong selective pressure over 5000 years but it's not 5000, it's closer to 2000

Their "last 5,000 years" is meant to emphasis that the sweep was still ongoing during these times, not that it started then (from the paper: "the selection pressures that initiated the selective sweep during the Late Pleistocene or early Holocene were still operative").

We have Stuttgart with 50% lightening alleles at 7,500kya and he was unlikely the very first person to have them... so we could well be talking somewhere in the vicinity of 10,000 years from go to whoa (confirmed by the paper: "The selective sweeps...implicated in the lightening of skin pigmentation, are estimated to have begun between 11,000 and 19,000 y ago").

Grey said...

@Tobus

"it's our preconceived theoretical notions that make us doubt this data, not the scientific validity of it."

It's not a theoretical notion it's a bunch of people writing down that another bunch of people in a certain area were particularly pale (relative to them).

If that is correct then it subtracts 2,500-3000 years from the selection time which multiplies how strong the selection pressure must have been to unusually high levels.

On top of that the selection pressure in question is supposed to have been caused by a shift from animal protein to grains because those grains didn't have certain vitamins.

However the group in question had a fish diet and then switched to pastoralism. They should have had **less** selection pressure than the farmers.

So the current model says a group who ought to have had the least selection pressure had exceptionally strong selection pressure.

It maketh no sense.

.

nb I don't doubt the data at all. I just think it's more likely that an improved version of depigmentation was selected over an earlier flawed version of depigmentation. I think there's missing data not that this data is wrong.

About Time said...

@Grey, well yes steppe was main "habitat" for these cultures - but they were always interacting with and more or less dependent on population centers to the south. That's where they got rich, got technology, got wives, etc.

So it's impossible to really talk about Scythians without talking about Anatolia, Assyria, Persia, Afghanistan, and Pakistan.

So eg Sakistan is in Iran. http://en.m.wikipedia.org/wiki/Sistan

Scythopolis was even further south. http://en.m.wikipedia.org/wiki/Beit_She%27an

Its like the Goths. Yes they were a northern linked culture, but they got rich and famous in southern places, then went back home when that ran its course.

I still wonder if these Kurgan graves were from elite tombs who were more married in with southern elites (like Bartatua's son Madys who was part Assyrian and maybe even part Jewish). So they would be darker partly just from Bronze Age mixing. Selection could be social/sexual selection too, not just natural selection.

Davidski said...

Check out the latest update on the GAGA + PCA + Mclust global cluster analysis of MA-1 and La Brana-1.

Seinundzeit said...

Thanks David!

This GAGA approach looks very promising, rather interesting.

Gill said...

Shouldn't the run with MA-1 as the Mammoth steppe reference be equivalent to your other run with MA-1, ME, ENA, and SSA? I wonder why it's showing different results.

Also somewhat curious, my ME made a jump to close the gap to the other Punjabi Jatt whose ME didn't really change. I wonder what that change represents.

Davidski said...

Adding Europeans changes the analysis, and shifts the Ancient North Eurasian cluster to the west, even when MA-1 is used as the reference. That's why I left them out from the first run.

In fact, the difference between the La Brana-1 and MA-1 inspired Mammoth Steppe clusters is almost non-existent. So I suspect what we're actually seeing there is something like a Western Mammoth Steppe component in both runs, although not quite the Western European Hunter-Gatherer (WHG) from the Lazaridis paper.

Contrasting the Asian ANE with the European shifted ANE might actually be useful at some level. For instance, if there were multiple waves of ANE-related gene flow to South Asia, then maybe the difference between these two clusters represents the contribution from the later, more westerly wave from somewhere like the Southern Urals? But I'm just speculating now.

Davidski said...

No, wait, that's wrong...not the difference between the two, but the European shifted ANE might be the later wave.

Everest said...

David, can you perhaps post some Admixtools data? It's got some nice programs like f4 ratio estimation, f3 statistic, D statistic, etc. It is also pretty easy to use.
http://genetics.med.harvard.edu/reich/Reich_Lab/Software.html

The Harvard guys love this software. Even Lazairidis et al relied more on this software than anything else.

However, what's missing is the qpGraph software, due to which we can't calculate basal Eurasian ancestry for example. I asked Nick Patterson about it, and he says the tool relies on some commercial software, so it cannot be made publicly available.

Davidski said...

My plan was to wait until I have more ancient genomes in my dataset and that software package improves a bit so the output is easier on the eyes for people from the personal genomics scene. I'll definitely put it on my system when the Loschbour and Stuttgart genomes are released.

Everest said...

I see. I remember having to make changes to the software. If you need my copy, let me know.

Also have some scripts that creates Zombies. I think the main use is probably to compare the various components.
For example, both "Baloch" and "Northern European" of Harappa have some Native American admixture, which probably points to ANE admixture. NE European turned out to be 3% Native American and 0.5% Beringian, and Gedrosia turned out to be 2% Native American. The numbers may not be accurate, but it shows certain affinity.

Seinundzeit said...

Thanks for the update David!

Very interesting, it seems the new ENA cluster is rather strongly shifted in an ANE direction.

Seinundzeit said...

This is a pretty interesting figure from the Anzick paper:

http://www.nature.com/nature/journal/v506/n7487/fig_tab/nature13025_F4.html

For the neighbor joining tree, I suppose the results are what we'd expect. You have African hunter gatherers being the most divergent, followed by Sub-Saharan African farmers, and finally, a split between West and East Eurasians. MA-1 is clustering with West Eurasians, and is closest to the "Indian" sample. For what it's worth, that Indian genome is from Southern India.

Grey said...

@About Time

"So it's impossible to really talk about Scythians without talking about Anatolia, Assyria, Persia, Afghanistan, and Pakistan."

For 90% of steppe history I think that's true but I don't think it was true right at the beginning after the first horse domestication.

I think there was a mammoth steppe population and at a certain point one segment of that population domesticated horses. I think that segment will be the ones who spread across the whole steppe at that point.

After that phase then who lived where on the steppe changed lots of times. I'm just talking about the very first phase.

Maju said...

@Seinudzeit: the problem is that such kind of graphs are approximate reconstructions, not strong evidence on themselves. It's not abnormal for example that admixture with an outgroup (for example East Asians) would drive a population (Ma1 in this case) to an apparent higher position in the tree.

In the same paper you have the ED Fig. 4, whose part related to Ma1, I just uploaded to the cloud for your convenience:

→ https://drive.google.com/file/d/0Bx7bO1-VAF2UajM5RDNWSlhYZFE/edit?usp=sharing

There are no Asian Indians in it but it is still obvious in it that the "range" of Ma1 affinity is as follows:

1. Native Americans (~0.14)
2. Russians (~0.135)
3. Most Europeans, Kalash, Burusho, Pathan, Uyghurs, Hazaras, Yakuts (~0.13)
4. Most East Asians, Sardinians, Balochi, Brahui (~0.125)
5. Melanesians, Makrani, Druze (~0.12)
6. Papuans (~0.115)
7. Palestinians, Negev Bedouins (~0.11)

I believe that the basal Ma1 affinity for West and (probably) South Eurasians is lv3, while for East Asians is lv4 (they diverged from Ma1 earlier). Variance up or down these levels implies admixture with third populations, surely "Denisovans" in the Melanesian case but Homo sapiens in the other cases, either residual "Basal Eurasian" from Arabia or African. Some of this Paleo-Arabian and/or African admixture must have been present in Neolithic flows from West Asia, which affect Sardinians the most in Europe and Balochi/Brahui the most in South Asia.

However the graph lacks enough samples from South Asia and Siberia to be totally sure about this interpretation.

About Time said...

@Grey, hmm hard to say. Lots of unknowns. Medes had 2 capitols, summer in north and winter in south. Cherokee did something similar moving north/south seasonally. So not limited to state/semi state societies.

We get a biased view of "nomads" because we see them mostly late, after massive agricultural expansions when best land areas have been used up by farmers and nomads are pushed out to fringes.

Arabia used to be a big hunting ground (rock art at Timna State Park shows big men on chariots hunting with bows and arrows). Lots of tumuli in Oman, etc and Gulf in general, barely anyone excavates it but it shows links to Af/Pak.

Arabs make a big deal of horse breeding and even have semi legend stuff that was woven into Islam but is IMO probably much older. http://en.m.wikipedia.org/wiki/Al_Khamsa

Arabs have legends about "vanished tribes" that lived in the area and are gone now. (Maybe from cold periods when people moved south en masse - like the Hyksos).

IMO in paleo/Mesolithic times, the "map" of where populations were might have looked different. People spread out since then. The fact that empires always tried to get those areas (Assyrians, Persians etc) can be practicality but also their old notions of what was "our turf."

Big (ANE?) morphological types def show up in Sumer / Ur etc. Empires look for the best of everything, and northern hunter bands made for good fighters. If they weren't hired, they might raid instead, especially in cold spells.

Seinundzeit said...

Hi Maju,

Thank you for the link! Very interesting.

You make some very important points here. Looking at these figures, I do find one specific detail to be very fascinating. South Central Asian populations are in the same "range" as Europeans, despite lacking the substantial WHG admixture found in Europeans (which is very close to ANE), and having significantly more "Basal Eurasian" admixture than Europeans. This is probably indicative of how much ANE admixture these populations possess. I think that most of these more northerly South Asians are around 30%-40% ANE. For comparison, I think Native Americans range from 14%-38% ANE, according to Raghavan et al., (I think. Although, my memory might be faulty). And, if I'm recalling things correctly, the highest ANE among Europeans would be around 19%.

I think ANE and WHG are, respectively, eastern and western "variants" of a single biogeographic group. The western variant is best represented today among living Northern and Eastern Europeans, and the eastern variant is best represented today among living South Asians and Native Americans, followed by northern West Asians. I think this probably sums everything up, despite being somewhat cardboard and simplistic.

Gill said...

Something I had mentioned to David in an e-mail is a phenomenon where South Central Asians (Afghans, Pakistanis, and Northern or high caste Indians) are showing varying types of European admixture. Some are showing entirely Northeast European (which in the Harappa project peaks in Finns at 79%) and in the various Eurogenes calculators are showing either the same or Western European admixture, including traces of Basque, Iberian, and North Atlantic.

So the question is, is the European admixture in South Asians similar to Northeastern Europe which is somehow broken down by the software into West European when the East Eurasian admix signal is too weak or is the original ancestral signal related to Western Europe to begin with and the software is mixing it with the East Eurasian admixture already present in South Asians to get inflated values of Northeastern Europe in some individuals? I've noticed European admixture amounts can be somewhat "normalized" when using only Basque and/or Sardinian as European components in supervised mode in Admixture.

Keep in mind that using Basque doesn't mean actual Basque ancestry, it just means the software looks for a component west-shifted in allele frequencies compared to Eastern Europe. This could reflect East Eurasian admixture (similar to Uralic) in Europeans after they branched off from South Asians and within the past 2000-3000 years.

I'm inclined to favor the link with Germanic Europe just because the supposed contributors of European-like admixture to South Asia that we know about were very similar to Sarmatians and other Pontic-Caspian steppe populations who, back then, were probably not very East-admixed. These groups migrated throughout north central and western Europe settling in France, southwest Norway, Netherlands, etc. Their descendants living in those areas have mixed less with outside groups in the past 2000-3000 years than Europeans in the East who have mixed with Turkic/Tatar/Uralic groups after the branching off from South Asian bound groups.

Maju said...

@Seinudzeit:

I can't agree with your conclusion:

"The western variant is best represented today among living Northern and Eastern Europeans, and the eastern variant is best represented today among living South Asians and Native Americans, followed by northern West Asians".

There seems nothing supporting those ideas, at least in most aspects.

On one side the data from Lazaridis, strongly suggests that the western variant (WHG) is well represented among most Europeans, especially if we would use a better proxy than Lochsbour, which seems a bit "extreme" and not the best representative.

On the other side there is no clear data from formal comparisons of the type f(Yor,A,X) for South and Central Asia.

As for West Asia, it does seem to me that at least part of it has some other influences (residual First Arabian and/or Mesolithic NE African), which is what drifts them apart from WHG and Ma1. This component was brought to both Europe and South Asia with the Neolithic. For that reason West Asians and peoples with notable West Asian influence may appear more distant. In Europe this extra distance is already factored for in the Lazaridis paper using EEFs (WHG+West Asian mix) but in West/South Asia it is more difficult to measure because of the lack of local pre-Neolithic samples to compare with.

Also South Asia probably harbors some important internal diversity, rooted in various ages, which is being overlooked at least in autosomal analysis. It's for example very possible that NW or in general Northern South Asians were originally closer to West Eurasians than those from the South, which were more isolated since soon after the OoA (semi-arid Deccan buffer). These kind of likely distinctions are no easily found in autosomal studies of South Asia more interested in simplifying structures like ANI/ASI or in comparing them with their western and eastern neighbors.

Independently of Ma1, etc. I'd love to see a good autosomal study of South Asia or at the very least India because it has intrinsic interest.

Maju said...

@Gill

There are a lot of problems with your supposition.

1. The Kurgan period (Corded Ware notably) was surely not as influential demographically in NW Europe as the previous Megalithic/Funnelbeaker one (see: http://forwhattheywereweare.blogspot.com/2014/02/neolithic-and-chalcolithic-demographics.html).

2. There was some back-migration from Poland to Eastern Europe with CW and even its precursors (Luboń, Globular Amphorae).

3. These Western IEs from (initially) Poland and Eastern Germany were a segregated group from the original Kurgan Eastern Europeans which must have admixed heavily with the locals.

4. It's almost certain that the Tatar+ genetic influence in Russians, Ukranians, etc. is not so strong, especially considering that Tatars themselves are mostly of European ancestry rather than East Asian, as are other Turkics West of the Caspian (only the Nogai, who are Mongol, retain major East Asian influence). Most of the Oriental-like genetics of Eastern Europeans actually come from Uralics and/or possibly related populations who left a legacy also in places like Ukraine as soon as the Neolithic (see: http://forwhattheywereweare.blogspot.com/2013/09/ukraines-neolithic-and-bronze-age.html).

So I see no reason to think that Eastern Europeans or less admixed Central Asians like Tajiks or Pashtuns are not a much better proxy for the early Indo-Iranians than your Germanics, even if they are not an exact match.

Gill said...

Turkic admixture was not a central part of my "supposition", I just mentioned it offhand because it's an entertainable possibility with historical basis in that region which is impossible in Western Europe. You can toss that out and my supposition still stands without it.

In my experiments with admixture (using Basque, Sardinian, and Caucasian components plus three Siberian components (Altaic, Uralic, Paleo) and a slew of South Asian ones) I've found that HGDP Russians were coming out as ~80% Basque and ~12-15% Uralic Siberian. For comparison, HGDP Orcadians were coming out as ~90% Basque, 0-5% Sardinian, and ~4-5% Caucasus.

North Sea and North Atlantic individuals (from Hapmap's CEU) were coming out similarly but with slightly more Sardinian than the Orcadians.

All these groups registered a little bit of South Central Asian. Curiously, not Baloch (which was a component from HGDP Balochis and Brahui) but often to the East of it like HGDP Pathan or North Indian (Metspalu Kshatriya/Brahmin). It was almost as much as the European (Basque) that the South Asians were registering which was 5-10%.

Lastly we're talking about trace amounts of European in South Asians, like 10% or less if we're talking about Western European admixture (could be 10-20% if we're talking about Northeastern Europe, i.e, Finns). No one said the contribution by these groups from the Pontic-Caspian steppe to Western European DNA had to be anything more significant than that. We know they migrated to these areas, a remnant of less than 20% isn't asking for much.

Gill said...

Tajiks vary in autosomal profiles depending on which part of Central/South Asia they are from and they have significant East Asian admixture not found in South Asians. Though they have plenty of European admixture, the almost drastic difference in the rest of the makeup doesn't point to any indication that they share similar ancestry. Their European admixture could easily have been picked up later than the South Asians whose European admixture must be very, very, old.

Gill said...

Here are some Eurogenes results for South Asians including some Iranians, Kurds, and Central Asians:

https://docs.google.com/spreadsheet/ccc?key=0AuXBmvmgdkfVdFMtRHVlZDBuQ3lMcjhxMDE4V3JoYlE&usp=drive_web#gid=7

Gill said...

Sorry, forgot to add, in my admixture run Ukrainians were coming out as 80-85% Basque, 5-10% Caucasus, 4-5% Uralic, 0-5% South Asian.

Seinundzeit said...

@Maju,

You make a very important point here, WHG admixture is definitely present at substantial percentages throughout Europe. In fact, if we take into account EEF being around 56% "WHG-like", all Europeans can be construed as predominantly WHG. Nevertheless, it was my impression that WHG peaks in Northeastern Europe. And in regards to South Asia, this conclusion seems to be entailed by much of the data. For example, if one calculates shared genetic drift between MA-1 and different populations, South Central Asians are ahead of North Caucasians, who are around 20%-29% ANE, based on Lazaridis et al. This is rather suggestive. Also, neighbor joining trees consistently join Europeans and South Asians together, in direct contrast to Southwest Asians. Even Southwest Asians with only 1%-3% Sub-Saharan African admixture are further removed from Europeans in comparison to South Asians, even in comparison to Indian populations with substantial ENA admixture. On top of that, we could point to the peak in R1a among South Central Asians. Using my ethnic group as an example, I believe R1a is consistently found at 50%-60%. Not to mention R2. I should also note the presence of y-dna haplogroup Q among Pashtuns. I'm not completely sure, but I think most Pashtun Q lineages are shared with West Asians, and don't seem to be recent arrivals from East Asia (again, my knowledge of this is fuzzy at best, I could be wrong about haplogroup Q among Pashtuns). And I believe David tried various separate analyses, some of them not involving ADMIXTURE, and they consistently revealed the same patterns (ANE always seemed to peak in South Central Asia). Based on all of this, I'm assuming we will eventually see these findings corroborated in a formal/academic context. But only if researchers actually work on this issue. So far, I think genetic variation in South Asia has been tackled from a rather "cookie cutter" perspective by researchers.

Then again, I'm still tempted by the notion that South Asia is somehow "basal" to both West and East Eurasia. In fact, I wouldn't be surprised if ancient DNA from South Asia calls into question the very distinction between "West" and "East" Eurasia. The good thing is that we can surely look forward to ancient South Asian DNA. In fact, we might see something exciting, this year.

Davidski said...

Gill,

Have you read the Lazaridis preprint? It shows formal admixture graphs that separate ANE from ENA. ADMIXTURE has trouble with that, and will produce skewed results with the wrong references. I explain some of that stuff here...

http://eurogenes.blogspot.com.au/2014/01/another-look-at-lazaridis-et-al-ancient.html

Also, another problem with your theory is that both ancient and modern evidence argues against any strong links between Western Europeans and prehistoric steppe groups. For instance, the Andronovo and Scytho-Siberian nomads carried R1a not R1b , while the mtDNA of Yamnaya and Catacomb groups was high in U4 and U5a, not U5b. Also, ANE generally falls and EEF goes up as one moves from Eastern to Western Europe, and it's ANE that is supposed to be the ancient steppe component.

In fact, Basques are one of the few groups in Europe that can be modeled as having no ANE ancestry. Others include nearby French, North Italians and Sardinians. So how can these groups represent the Scythians and Sarmatians, or in fact any early Indo-Europeans from the steppe?

It's likely that Western Europeans do have some ancestry from the Eastern European steppe, but it's only minor. A larger part of their genetic makeup comes from the Mediterranean and the Near East.

In fact, even the Western European Hunter-Gatherer (WHG) component peaks in Northeastern Europe today, rather than in Western Europe. This is likely where some of the "Uralic" influence comes from, because La Brana-1 often shows it too when tested with ADMIXTURE.

Gill said...

Thanks for the link, David! That was a fascinating read.

Just to be clear, I agree fully with your theory written in that blog post. In order for, at max, perhaps 20% of the autosomal content of certain regions of Northwestern Europe to correlate to a possible late Bronze/Iron age ancestor from the Pontic-Caspian steppe that they could potentially share with some South Central Asian groups, there doesn't need to be much correlation in haplogroups. We're not talking about a big autosomal footprint. And the only reason I suggest such a late connection when the earlier Proto-Indo-European ANE migration would be enough is one possible way of explaining the admixture results showing affinity to Western Europe in South Asians. It certainly isn't to Basque. I did not mean to imply that at all, I just used Basque as an extremely West-shifted component averse to East Eurasian in order to normalize our European signals so they were not artificially inflated by our ambiguous East Eurasian-like admixture, not all of which is from ANE and may in fact be indistinguishable from ANE to the software. It did give very stable results and better ethnic clustering. And the affinity is not just to "normal" ANE if the software is to be believed because otherwise it'd be registering as all Eastern European.

The two ways of interpreting it are:

1) It's probably the Uralic and other relatively late-arriving East Eurasian-like admixture in modern day Eastern/Northeastern Europeans which is partially responsible for this effect. But the Uralic doesn't get higher than 15%. SO that alone doesn't explain why the South Central Asian affinity isn't tracking it completely. The only option left is Northwestern Europe. It doesn't need to be a lot of admixture in common from there, just enough to account for 10-20% overlap (I'm being generous and doubling it).

2) It's actually the WHG in Eastern Europe which is pushing the admixture further West towards more EEF/ANE areas. Any ancient EEF signal in South Asians is being caught up in the Western European components. Some EEF is preferable to no WHG I suppose. It's possible WHG is being picked up as Uralic.

However, the admixture isn't Basque (in K36 for example). The Western European signal is mostly North Sea with some North Atlantic. The Eastern European signal is highly influenced by the Mordvin/Kargopol (Eastern Euro) and Chuvash populations (Volga-Ural). The North European signal (Fennoscandian) is lowest. If I do my own run with Orcadian, Basque, and Russian, most will settle into Orcadian with some bleedoff into the others (usually Russian, with some individuals mostly Russian). So, it's not Basque. It's just that Basque when used in a dataset with a lot of Europeans with no other component but Sardinian, acts as a pretty decent substitute for West-leaning European without fluctuating as much within ethnic groups as Eastern European but still varying enough between different ethnic clusters.

There is no Uralic Siberian in South Asia that I could find any direct affinity to and traces of Altaic Siberian are very small any further east of Afghanistan. Any Arctic admixture is usually Paleo, Beringian, or Amerindian.

Gill said...

Pet theories aside, it's happening in South Asia too. I was 33% ANE in your run but my European/Caucasian admixture in most calculators are both leaning to the West, areas of relatively lower ANE, and my ENA-derived components are raised (much more than the 3-4% ENA difference in your run between me and the other Punjabi, our full difference in Harappa stretched out to a 12% swing minimum in just the two South Asian components). In South Asians at least, the slightest bit of ENA or noise interpreted as ENA can set off a chain reaction in the components as interpreted by Admixture and come out with noticeably "shifted" results.

I recently did a little experiment in weeding out noisy regions (started off by trying to just extract the major ENA segments as you suggested). What I noticed was that Gedrosian/Baloch is a very stable component in Punjabis, not that much affected by noise (relatively speaking anyway, it did fluctuate by as much as 3% after a lot of pruning). But the South Indian, European, and Caucasian components fluctuated significantly, South Indian most of all. South Indian breaks down into 50% Caucasian, 8% Papuan, 42% East Asian. Northeast European is the component peaking in Uralic-heavy Northeastern Europe (~80% in Finnish). Caucasian here is modal in West Caucasians but it was dragged along for the ride by S-Indian whose Caucasian half most resembles it. If it weren't for S-Indian, the Caucasian component would be stable too (in fact, most of the fluctuation in Caucasian was dependent on S-Indian's movements). I think this also is related to the fact that Gedrosian is "local" or native to Punjab, while S-Indian is not, even though it's on the same subcontinent.

Also, when I checked the results from an admixture run in byseg mode, almost half (!) the range covered in our raw data was covered by NE-Asian, SE-Asian, Papuan, Siberian, Beringian, Amerindian, W-African, E-African, Pygmy, and San (what I considered "noise" in Punjabis). There's just a lot of East Eurasian/West Eurasian noisy overlap. Your run showed this actually, the ANE+ENA was at around 50%. The amount filtered out for me was 49% and for the other Punjabi whose data you ran, it was 47% (curiously, and probably somewhat coincidentally, close to our ANE+ENA).

Gill said...

^ Sorry, one correction. The SW-Asian component is the closest to the Caucasian half of S-Indian, but most of it just transfers over again into Caucasian.

Gill said...

Sorry, another thing. Sein mentioned that these results are similar to Geno 2.0. Here are some Geno 2.0 results:

https://docs.google.com/spreadsheet/ccc?key=0AuXBmvmgdkfVdFMtRHVlZDBuQ3lMcjhxMDE4V3JoYlE&usp=drive_web#gid=11

The Pashtun is Sein, the Punjabi Jatt Muslim is me. Unfortunately Geno 2.0 uses a small number of SNPs for their test, but it's not too far off. Mediterranean+Northern European+ Native American would correlate to ANE (perhaps Northeast Asian as well for Punjabis). Southeast Asian would correlate to ENA (and also perhaps ASI).

Davidski said...

Gill,

Right, not Basques.

OK, the signal shared by Northwest Europeans and South Central Asians, in particular Western Brits and Kalash, has turned up in a number of analyses. I have no idea what it means, but in the past I've speculated that it might be due to related and very specific ANE migrations from the southern Urals both to the west and east, an artefact of a similar mix of EEF/ANE ancestry among these groups, or the fact that the markers were ascertained in people of mostly British origin (in fact, the CEU bunch).

The first theory is supported by ADMIXTURE and PCA analyses. See, for example, dimensions 1&7 here...

https://drive.google.com/file/d/0B9o3EYTdM8lQeEUzeWZ2TWpZbzA/edit?usp=sharing

But the problem is that neither ADMIXTURE nor PCA results are easy to interpret in the context of ancient population movements. What we need to see are formal mixture tests with ancient genomes as references.

Seinudzeit and Maju,

After Amerindians and some Siberians, it's the Lithuanians who share most drift with MA-1 (so they're in fact the most ANE-like West Eurasians). A lot of this is indirect affinity though, via indigenous European hunter-gatherer ancestry. But it's worth mentioning, because the question remains whether it's actually possible to draw a line between European and ancient Siberian hunter-gatherers? In fact, maybe MA-1's relatives further to the west were ancestral to La Brana-1?

http://i129.photobucket.com/albums/p217/dpwes/MA-1_shared_drift.png~original

Another problem is that those phylogenetic trees will show different results when admixture edges are added, especially for more recently admixed samples. So look what happens when an admixture edge is added to the Indian genome...

http://i129.photobucket.com/albums/p217/dpwes/Mixture_tress.png~original

That's not to say Indians don't carry a lot of ANE. I personally believe North Indians and other South Central Asians carry as much or slightly more ANE than Northeast Caucasians. But it's hard to prove that, because MA-1 is such an old genome that we don't know whether it has some sort of South Asian admixture, or it just didn't have time to differentiate from Southeast Asians as much as extant West Eurasians?

So what I'm saying is that maybe the high ANE affinity among South Central Asians can't be explained by one thing, and what we're seeing is a combination of deep links between South Asia and ancient Siberia, as well as a more recent event of mixture between Bronze Age steppe people and Indians?

By the way, those two figures came from the Raghavan et al. study if anyone's interested. The second one is from the free supp info. See here...

http://eurogenes.blogspot.com.au/2013/11/first-genome-of-upper-paleolithic-human.html

Maju said...

@Gill: what I meant was in essence that Uralic genetic influence in Eastern Europeans was already very important before the IE expansions and certainly before the Indo-Iranian one. Actually IEs, by reason of mere prolonged contact, acted as carriers of Uralic genetic influence to other parts of Europe, the same they did with the extra West Siberian genetics ("ANE"). And they must have carried it also to South Asia for whatever genetic influence the steppe riders had over there (minor, I think).

As David says, the original steppe IE influence (incl. the related Uralic-like and extra ANE) dilutes rather quickly as we move to the West in Europe. Germans and Scandinavians seem to have a sizable chunk of it (judging on genetic markers like R1a and others) but Brits and French not too much. Genetically speaking, one thing is "Indoeuropean" and another one "European", especially Western European.

Maju said...

@Seinudzeit:

"it was my impression that WHG peaks in Northeastern Europe".

Around the Baltic, at least judging on the Lochsbour-based stats estimated by Lazaridis. It is however unclear how much of that corresponds to genuine Western European forager legacy or extra Eastern European forager one brought by Indoeuropeans and their precursors in migration to that area: the Pitted Ware peoples of Dniepr-Don roots. One thing to take in account is that no Eastern European specific sample was ever considered and personally I do think there are those two elements acting: genuine Western European forager remnants and somewhat distinct Eastern European forager genetics carried westwards by IEs and Pitted Ware.

We have to understand that unlike most of Europe, Eastern Europe was not directly affected by the mainstream European Neolithic flows, so it's only logical that Eastern Europeans were/are more purely of European forager ancestry (WHG-like), regardless of their Siberian inputs (ANE, Uralic).

In other words: part of that "return of the foragers" that we witness in much of Europe since the Chalcolithic corresponds to IE (and other) expansions from the East. In the case of Baltic area populations (i.e. Lithuanians and such) it'd be interesting to discern how much of it is Western (rooted in Magdalenian, etc.) and how much is Eastern (rooted in Eastern Epigravettian), because there must be some distinction between both populations. Judging on the modern presence of yDNA I2 and mtDNA U5 (the genetic markers dominant among NW foragers), it'd seem that the Western European forager persistence is not too large and that a (related but necessarily distinct) Eastern European forager element is being confused for WHG.

"neighbor joining trees consistently join Europeans and South Asians together, in direct contrast to Southwest Asians".

Do you mean in autosomal DNA, I'm not really familiar with that consistency. It is true that in some cases Arabo-Palestinian groups may be placed upper in the tree because of their African-like admixture (which is probably in part aboriginal Arabian or "Basal Eurasian" rather than strictly African but looks the same in shallow analysis for it contrasts with the rest of Eurasia) but in general Europeans are closer to West Asians than to South Asians (what makes good sense considering that European ancestry, be it forager or farmer, comes mostly from West Asia).

(continues)

Maju said...

(continued from above)

"I should also note the presence of y-dna haplogroup Q among Pashtuns".

It should be clear by now that yDNA Q is most basally diverse and hence original in West Asia (or maybe South Asia?). Not just Pashtuns but Iranians and others from that wider region have some important diversity of early derive yDNA Q. Siberian-NA Q clearly stemmed from West-South Asia via Altai. And, before the recent findings re. Ma1 and such, it was the most clear evidence (along with archaeology) of a West→East flow through the Siberian corridor in the Early Upper Paleolithic.

Y-DNA wise, MNOPS probably coalesced somewhere near SE Asia (either Indochina or the lower Ganges) some time before the UP. The migration westwards of P (R, Q) through Northern India, along with that of mtDNA R, can be easily mapped, and is surely central to the origin of West Eurasians as distinct "continental" population. That must have happened (regardless of "molecular clockmaking") just before the beginning of the UP, which is roughly dated 50-40 Ka years ago (cal.) in West, South and Central Asia, as well as in Europe and NE Africa (LSA).

"South Asia is somehow "basal" to both West and East Eurasia".

I can't but agree in general terms. However once the "critical mass" was reached and South and East Asia consolidated their differentiation, since then, South Asia has got much more intense relations (via AfPak) with West Eurasia, what blurs the distinctions. And autosomal DNA is very sensitive to "recent" changes (because of recombination) being much more difficult to infer deep ancestry using it alone. In this sense I strongly recommend to try to have a holistic view, considering also haploid markers and known archaeology/prehistory in order to get the most complete perception.

Maju said...

@David:

"After Amerindians and some Siberians, it's the Lithuanians who share most drift with MA-1 (so they're in fact the most ANE-like West Eurasians)".

Well, the formal tests attribute more ANE to Estonians but let's not be nit-picky. For me what that implies is that a significant part of their WHG component is not directly from Western European foragers but from Eastern European ones (IEs, Pitted Ware), who would have been in more direct contact with ANE type peoples since Gravettian times.

"... look what happens when an admixture edge is added to the Indian genome..."

It can well be an artifact. It almost certainly is. I do not think that East Asian genetics heavily influenced South Asians, just where such direct influence is noticeable as in Orissa or the Himalayas.

I would insist in using in all analysis more samples for South Asia, which is a continental region in its own right and to perform more analysis with specific focus on the subcontinent. There is a particular complexity regarding South Asia, which, as Snudzeit suggests, is at least partly "basal" for all Eurasia but also, as I added above, more akin in some aspects to West Eurasia. They are not a simple mix in any case but more a mix of SA-specific components (surely more than just one but generically synthesized in the ASI component), Western ones (represented by ANI) and much less important Eastern ones (Austroasiatic and TB elements in essence).

"That's not to say Indians don't carry a lot of ANE".

Actually ANE and East Asian affinity are almost opposite. They should not relate in any obvious way. Non-Siberian East Asians have low ANE affinity.

"maybe the high ANE affinity among South Central Asians can't be explained by one thing"...

Agree. It's likely that several elements are at play here: on one side remote affinity identical to the general West Eurasian affinity, rooted in the early UP, and then probably some extra ANE affinity via IE flows (same as in Europe). Neolithic flows from West Asia should instead be neutral or rather negative re. ANE affinity (same or similar to what happens in Europe too).

Davidski said...

Well in that case the genetic difference between the Western and Eastern European hunter-gatherers wasn't all that big, because Poles are right up there when it comes to allele sharing with La Brana-1. I bet that Lithuanians would do even better if they were in that analysis, judging by their results in my Mammoth Steppe test in which La Brana-1 was the reference.

Also, I don't think that graph is showing East Asian admixture in the Indian genome, because the admixture edge starts up the tree from the Dai. That looks like the so called Ancestral South Indian (ASI) component to me.

By the way, I can't use more South Asians in the Asian ANE test, because they dominate the analysis and turn the ancient ANE cluster into a modern South Asian one. As per above, I probably need some ancient genomes from South Asia to get around this problem. The Northeast Europeans do exactly the same thing, as you can see in the Mammoth Steppe test, but at least there I got figures that were close to the overall hunter-gatherer related ancestry among Europeans reported in the Lazaridis et al. paper.

Maju said...

"I can't use more South Asians in the Asian ANE test, because they dominate the analysis and turn the ancient ANE cluster into a modern South Asian one."

Fair enough. But, anyhow, wouldn't it be possible to run some sort of supervised test, as done with "zombies", forcing the samples to align with this or that or the other? Using Ma1 as one of the reference populations and the others picked among tentatively "purest" populations (such as the Irula or Palestinians or whatever), it'd be possible to get a better notion of what is what.

But I understand that it is complicated, of course.

About Time said...

@Seinundzeit, there is also the Gedrosia which tends to be Scottish/British, and the West Asian % that maxes out in the Balkans and continental Germanics. Wondering if the Welsh have any local pattern, but no data so far.

Looks like maybe 2 waves, one older (Gedrosia) and one younger (West Asian). The younger one might fit with the 700bc German/Austrian/Orcadian admixture in Kalash (north Iranian related? Like Medes who did set up city of Mundigak in Af/Pak area).

Older one might be Hittite or epi-Hittite/Assyrian, but that's just a guess. (Hittite was court language, Assyrian was trade language at the time - and Aramaic later shows up in eastern areas with Kamdahar Edict of Ashoka).

About Time said...

Double posting to clarify my thoughts. I think in Kalash and Orkney we are seeing very peripheral traces of events that happened "at the center" (West Asia) but left traces in fringe places.

I've been throwing around a lot of obscure labels like Medes, Assyrians, Akkadians, Hitties, etc. without being clear about how this connects to Europe or tiny groups like Kalash.

Here's my picture:

1. Main middle eastern empires were around Tigris/Euphrates. They spoke versions of Semitic as lingua franca. That's Akkadian/Assyrian/Aramaic - which functioned as the "English" of its day.

2. Those empires were often conquered by rugged mountain men from fringes. Often IE like Gutians, Hittites (same thing?), Mittanians, Medes, Persians. These military elites were numerically smaller but moved around a lot more. They interacted more with fringe populations at outskirts.

3. Fringe groups were influenced, either by direct conquest/settlement or imperial skirmishes and copycat technology and trade agreements. I think Hallstatt, maybe even Bell Beaker fits this pattern.

These little pops mostly kept their own minor languages (Germanic, Kalash, etc) but might have absorbed (at times) settlers or military garrison pops from the main empires at the center. This is what I think the Gedrosia in Orkney or German/Austrian admixture in Kalash signifies

Not that Kalash or Germanics were in any way important in the Median or Neo-Assyrian empires. Just that insofar as they received any mixture, it was from those more active imperial zones of the day.

Just like many people around the world have a little British or Spanish or Dutch ancestry. Not because imperial outposts like the Filippined or Sri Lanka were central to interactions, but bc they received mixture from the active imperial populations of the day.

Gill said...

If the Orcadian and Northwestern Europe signal is coming off the Gedrosian, that would be interesting since I registered Orcadian on FTDNA's Population Finder and had 6% North Sea in K36, but my Gedrosian/Baloch has been a few points below the average for Punjabis.

I've done a lot of tinkering with regards to filtering out various types of segments and I couldn't get the Gedrosian or the Atlantic-Mediterranean signal to change character. The Gedrosian especially was a very stable and consistent signature. It could be that in Jatts from the East (Eastern Punjab and Haryana), their Gedrosian is bleeding off into Western European, and the reverse as we move westwards.

Another thing is that by combining a Western European population with Siberian Uralic, I was able to conjure up a fake Northeastern Europe component that functioned just like the original. That's why I kept thinking the Western European signal was split off somehow from Northeaster Europe. Gedrosian is an interesting theory.

About Time said...

@Gill, what you and Seinundzeit are saying about waves with increasing levels of East Asian (and IMO maybe West Asian too) admixture makes sense.

Older waves (maybe still more ANI like) got pushed out to remote places like Britain, Basques, Norwegians. Later waves end up squished into North-Central Europe and West Balkans. Later waves to Baltic (squished from older range in North Eastern Europe). Later still are Ukrainians and Russians. Last of all Tatars, Chuvash, etc.

Seinundzeit said...

Maju,

I absolutely agree with what you're saying, and I think your points here are exceedingly pertinent. The data surely supports many scenarios, all of which will have to take into account some intense complexity and conceptual richness. And I also believe that a fresh, holistic perspective is absolutely warranted. Looking at all of the evidence, the notion that South Asia (especially peninsular South Asia, with an eye towards South India) represents a unique pole of genetic variation versus West and East Eurasia (albeit much closer to West than to East Eurasia), and has a genetically constitutive relationship to both regions, seems very reasonable to me.

David,

I think your points here are absolutely correct, and completely warranted. I do think that the peak in ANE ancestry among South Central Asians is the product of many different processes and events. To be honest, I still think the fact that it's been named "Ancient North Eurasian" is really just a reflection of biased preservation. If it does peak in northern South Asia, followed by the northern Caucasus, I think it's plausible that ANE-like people existed throughout West Asia, and were absorbed by "Ancient Near Easterners" rich in "Basal Eurasian". In addition, it's probable that excess ANE admixture entered South Asia from the steppe. And in regards to the results you've obtained, the best way to verify all of this is using the Onge, once you acquire those samples. They belong to the same clade of peoples as the "ASI", so they will nicely capture actual ENA ancestry for South Asians. If the results don't change, even when using the Onge, I suppose all of this is solid, northern South Asians have moderately more ANE than Northern Caucasians. But, the results could definitely change. Also, for what it's worth, the Indian sample is from Karnataka, in Southern India.

About Time,

You make a very interesting point here concerning Gedrosia. In another experiment David tried, I ended up being 36% ANE, which is actually identical to my Gedrosia score on various calculators. Everest, who made a comment above, calculated shared drift between ADMIXTURE components and MA-1. Interestingly, Gedrosia and Northern European shared the highest amount of drift with MA-1 (I'm not sure if he included the Native American component). It's of some relevance that I shared more drift with MA-1 than Gedrosia. Surprisingly, a Punjabi Jatt friend of mine shared more drift with MA-1 than the Northern European component!

Davidski said...

About Time,

Unfortunately, the intricacies of ADMIXTURE mean that any detailed discussions about components like Gedrosia, Caucasus, West Asian, or whatever, aren't very useful.

That's because ADMIXTURE usually bases its clusters on the most isolated, genetically drifted, and/or endogamous populations in the dataset. These clusters then resonate in many other populations, usually the more genetically diverse/outbred groups that have a hard time creating their own clusters.

Obviously, such results do prove that the people concerned share ancestry. But working out the direction of gene flow, or even where it came from in the first place, because it could have come from a third party, just from the ADMIXTURE output is impossible.

Luckily, we now have a few ancient genomes and it appears that the culprits for these mysterious but seemingly relatively recent links between Europe, the North Caucasus, South Central Asia and Siberia are in large part the relatives of MA-1, who most likely expanded from somewhere around the Urals at the end of the Neolithic.

You can actually see that MA-1 clusters between the Volga-Ural and the Northeast Caucasus in my latest cluster analysis. I think that's a pretty good approximation where about 15-25% of the DNA of modern Northern and Eastern Europeans came from during the Copper Age.

http://bga101.blogspot.com.au/2014/03/updated-eurogenes-k13-population.html

Indeed, check out the division in the first tree between the WHG/ANE-rich (top half) and Mediterranean-rich (bottom half) West Eurasians. It's the same thing as what ADMIXTURE has been showing us for years, but a lot less cryptic because here we can see where the ancient genomes are sitting.

Davidski said...

There's a new preprint at BioRxiv that reports ANE ancestry among Pathans by showing that they can be fit as a mixture of Samaritans and Karitiana Indians. See figure 2 at the link below, and then refer to the supp info PDF for details:

http://biorxiv.org/content/early/2014/03/21/003517

http://biorxiv.org/highwire/filestream/920/field_highwire_adjunct_files/0/003517-1.pdf

Other Pakistani groups aren't showing the same Karitiana signal. For instance, the Balochi and Brahui are fit as mixtures of Cypriots and Paniya. But the paper just reports a single outcome for each test group, so we can't see whether these Pakistanis can be fit as partly Karitiana. I think they probably can, but it'd be a less likely fit than what's reported.

So either the Pathans have unusually high ANE for their region, or perhaps their minor East Asian admixture is helping them to get that result (because Karitiana are themselves a very ancient mixture of ANE and East Asians). But the Burusho have more East Asian ancestry than the Pathans, and they're fit as a mixture of Georgians and Naxi.

Maju said...

But, David, that study's results are terribly imprecise and the supp. material is not at all more detailed (no figures given, only names ref. pops.) According to it Lithuanians are also an
unmixed population, while we know that their ANE affinity is one of the highest.

When I was told of this study days ago, I replied to my correspondent something like: the study seems very roughly "correct" but it is so extremely imprecise that it can't be relied upon. It's just a very rough draft in its results. If at least they had some decent supp. material in the line of the Pääbo team studies, which are great data mines, but not at all.

Notice that Iranians are described in the same type of two-way admixture as are Pashtuns but, on the contrary, other Pashtun neighbors which score high in ANE affinity (Burusho, Kalash) are described very differently: Burusho as Caucasus-India mix (even if we know that they have minor East Asian admixture, much more than Pashtuns) and Kalash as unmixed.

Never mind about using such horrible reference pops. as Samaritans or Druze, which are the product of recent endogamous ultra-homogenization (and that way they appear as "unmixed", I guess). The Samaritans are just 800 individuals!

For me it is not useful to pay much attention to this study: it can only be a vague draft.

Davidski said...

It looks like a comment piece to me, in which the authors are mostly trying to spread the word about ancient DNA, and also saying that using modern DNA alone to learn about human admixture history is a futile effort. So it's not surprising that their f3 analysis isn't very detailed, because it's just there to generally back up their claim that most human groups show signals of admixture.

The Pathan and Iranian results are important because, apart from my own experiments here, they show for the first time that there are populations southeast of the Caucasus that carry considerable ANE. I'm sure more detailed results are on the way, possibly along with complete genomes from Central Asia, but this paper isn't for that.

Seinundzeit said...

It was quite surprising that the Pashtuns seem most similar to ANE-rich Caucasians, as well as Iranians, rather than neighboring Pakistanis. Also, correct me if I'm wrong, but don't the HGDP Colombians have less ANE ancestry than Karitiana Indians? If so, it's interesting that the Lezgians can be best construed as mixtures between Samaritans and Columbians. So far, they are the most ANE-rich population recognized in a formal context. Since the Pashtuns are best fitted as a mixture between Samaritans and Karitiana, I think it is reasonable to suppose that they probably have more ANE than the Lezgians. It'll be extremely exciting to see some ancient Central Asian genomes.

Davidski said...

By the way, Maju, the Burusho are fitted as Georgian and Naxi, so Caucasian and South Chinese, not South Indian.

Maju said...

"... the Burusho are fitted as Georgian and Naxi, so Caucasian and South Chinese, not South Indian".

Alright. My bad.

They are still high in ANE affinity, what, as happens with the Hazaras, is not detected in the Pickrell & Reich paper (as Native American admixture). Same with Lithuanians. The problem seems to be persistent and therefore not really useful to explain ANE affinity or inform of it in any useful way.

aniasi said...

Any chance that this could be added to the Gedmatch admixture utilities? It would be interesting to see what happens with other samples and ANE.