I've seen quite a few comments on this blog suggesting that most of the Ancient North Eurasian (ANE) admixture found in Northern Europe today might come from Scandinavian hunter-gatherers like Motala12 and Ajvide58. It's probably obvious to most that this is not realistic, because the Scandinavian forager genomes sequenced to date show very high ratios of Western European Hunter-Gatherer (WHG) ancestry (>80%), so basically the math doesn't add up.
Nevertheless, I thought it might be useful to drive the point home using this Principal Component Analysis (PCA) based on my new West Eurasia K8 test. The datasheet is available here. You can view a spreadsheet of the results with extra samples here.
Please note that neither Motala12 nor Gokhem2, a late Neolithic farmer from south Sweden belonging to the Funnelbeaker culture, can pass for present-day Swedes. Moreover, mixing Gokhem2 with Motala12, in any proportions, will not produce a result even vaguely similar to present-day Swedes (ie. the outcome will fall somewhere along the dotted line).
I'd say one of the most obvious ways to get the right result would be to blend the Scandinavian forager and farmer with at least one other sample from somewhere below (ie. geographically speaking, east or southeast) of the Swedish cluster.
It might be possible to come up with a more precise plot location, and thus perhaps geographic origin, for this putative third source of Swedish ancestry by running some complex tests with the PCA datasheet. If anyone wants to have a go at that, and you actually manage to come up with a coherent outcome, then feel free to post your findings in the comments below.
I've decided not to bother, because as far as I can see, the options are infinite. What we really need are more genomes from the Swedish late Neolithic/early Bronze Age (LN/EBA), preferably belonging to one of the local spin-offs of the Corded Ware culture, which is thought to have originated in Eastern Europe, to provide more datapoints and help narrow down the options.
On a related note, I'm catching up on some reading this holiday season, and currently going through this book chapter which discusses the upheavals during the LN/EBA in south Scandinavia as seen through its archeology.
Rune Iversen, Beyond the Neolithic transition - the "de-Neolithisation" of south Scandinavia
See also...
Bell Beaker, Corded Ware, EHG and Yamnaya genomes in the fateful triangle
So let's play the guessing game :)
ReplyDeleteNo one wonders where was that 3rd population (here called ANE) at the beginning of the Neolithic?
We know they had to be east of the Middle Eastern, north of the South Asian, south of the Eurasian Steppe and west from China/Tibet. Not many options, are they?
Basically 2: Central Asia and Iran. And I think that it's quite obvious that Central Asia is the best candidate (for reasons I won't elaborate here for the sake of brevity).
This population, along the Neolithic, migrated (probably in non-disruptive waves) south and west, having the highest impact in Pakistan and the Caucasus, but influencing all West Asia from the north of India to Anatolia.
We know that the people from the Eurasian steppe (basically European HGs) had a good amount of admixture from this population, probably by vicinity (after all there were no natural borders between Central Asia and Kazakhstan). Probably they even got more when this Central Asian population moved to the Caucasus.
So we know that the people from the steppe (let's call them PIE) at some point expanded into the Eastern and Northern parts of Europe (all R1a high places), bringing this admixture with them (I'd rather not call this admixture ANE, but maybe Ancient Central Asian - ACA).
In the South East of Europe, this component entered probably from 2 sources: From Anatolia (mostly) and from the PIE.
The question is, how did this component entered Western Europe? Was it brought also by PIE (as in the North East) or did some amount of this CA population cross directly to Western Europe from Anatolia and SE Europe? (by the way, I don't think that the original population from CA carried much -if any- Near Eastern admixture. They did probably have HG one).
Regarding the ancient Swedish HG genomes, they do show ANE admixture, but no West Asian admixture. I think that both of these components are related to the original ACA one, but probably none of them is exactly what we are looking for, so it's hard to say if in Sweden this ACA component was present or not (MA-1 does show West Asian admixture).
END of the Guessing Game. Though I didn't make a guess as to how this component entered Western Europe (the big question), because my crystal ball didn't tell me that. I'll wait for more data before asking again :)
ANE was present in Scandinavia during the Mesolithic. Motala12 has 16% of it in this test, despite missing many markers.
ReplyDeleteSo I have no idea how you missed that, and localized ANE specifically to Central Asia/Iran for the early Neolithic, when it was obviously in Eastern Europe at least as early as in Scandinavia.
If it's assumed Swedes are simply mix of Gok2, Motala12 and a third population on that plot, the third population could plot somewhere between a point slightly to the left of Tatar/Chuvash group to one slightly to the left of the big Near Eastern group, depending on how much Motala and Gok2 Swedes have. So even with such a simple model possibilities are many.
ReplyDeleteBy the way, here's the full spreadsheet.
ReplyDeletehttps://docs.google.com/spreadsheets/d/1kd9Q9vFrL1Cra9ayqMYVFKXrUdnThmQJVMtjczLhoTs/edit?usp=sharing
Gokheim2 should not be used as a primary ancestral component of Swedes, since he was already mixed with WHG, not a pure neolithic farmer. One can see he is shifted towards Loschbour in the axis that goes to Stuttgart.
ReplyDeleteI would rather draw the axis from Stuttgart to Motala12, and then you can see the third component to create modern Swedes would be a source from somwhere East (Volga-Ural).
Chad, if you see the new full-spreadhseet of ANE K8 Davidski posted the only modern Europeans who score more WHG than Gok2 are Balts.
ReplyDeleteScandinavians, Poles, Finns, and Belorussians, have slightly less on average than what she scored.
She was probably somewhere around 65% EEF, the rest was Loschbour-WHG.
So, what makes modern northern Europeans different from Basque and Neolithic north Euros, is ANE and Near eastern.
You still need a population in Neolithic Europe who had more WHG than Gok2 and Balts, to explain admixture with Yamna and modern ANE levels.
If there were populations in-between Russia and Ireland who had ANE, that could help explain modern WHG levals, because Yamna(with low WHG) wouldn't be the only ANE source.
Stuttgart+Yamna+centralnorth HGs might be able to explain CWC and mostly explain modern central-north Euros.
It would make sense now that NE Euros and Scandinavians, have a decent amount of Mesolithic survival blood, passed what Gok2 had.
It's obvious now that Y DNA I1 and I2a2 are native WHG lineages of central-north Europe, because Gok2 and modern WHG levels prove there's a decent amount of native HG blood in north Europe.
ReplyDeleteBoth of those mega modern lineages have already been found in Neolithic Hungary.
David, I didn't miss that Swedish HGs had ANE. I wrote a paragraph about it.
ReplyDeleteThe do have ANE, but they lack West Asian admixture. My opinion is that ANE is not a very accurate representation of this Ancient Central Asian component. Just like the West Asian admixture is also not very accurate. Both are related, and both related to ACA, but I think that none is a perfect match.
So I'm not sure if what I decided to call ACA was present in Sweden or not. Maybe yes, maybe not. But in any case it wouldn't change much. As you noted in the post, more of this Eastern (and rather Southern) component is needed to explain modern Swedes. And for that we need more of this ACA component in them that arrived at a later date with PIE (R1a people).
I also didn't say this component (ANE or ACA as I prefer to call it) was exclusively in Central Asia. I explained that Eastern European people (from the steppe), did have this component at that time because they were "neighbours" with the Central Asian population and had some admixture of each other. But obviously it had to be quite higher in the population south from the steppe people, otherwise it wouldn't be possible to explain why it is today higher in the Caucasus and Pakistan and not in the steppe.
ReplyDelete@Krefter
ReplyDelete"If there were populations in-between Russia and Ireland who had ANE, that could help explain modern WHG levals, because Yamna(with low WHG) wouldn't be the only ANE source."
There's no evidence of the Hamburgian in Ireland (but there is in Scotland.)
FYI, ydna, while somewhat informative, cannot be directly associated with autosomal "components".
I'm glad you've found my new blog post.
http://www.linearpopulationmodel.blogspot.com/2014/12/the-archaeological-record-in-northern.html
Gokhem2 can't cover above northern French. She's only 4% above Brits and about modern Scandinavian levels. Her people had to mix with Motala to cover it. Yamnaya isn't enough WHG to keep it close. A better line might be Gok2 through the French and the French being 40% of the way to Yamnaya. I'll crunch numbers to verify that after work.
ReplyDeleteAlberto,
ReplyDeleteI'd say that ANE spread to the Americas, Eastern Europe, Scandinavia and Central Asia during the late Paleolithic and/or Mesolithic via population movements from the ancient (pre-East Asian) Siberian steppe.
The fact that ANE peaks in the Amazon Basin and probably the Hindu Kush nowadays doesn't mean it wasn't the main component on the Siberian steppe just before the Neolithic.
Truth,
A model involving Stuttgart might be better, but Gokhem2 is likely to be more representative of Scandinavian Funnelbeaker farmers from the very late Neolithic, which is just before the mixing took place between them, the Pitted Ware hunter-gatherers (which were very similar to Motala12), and the Battle-Axe people of Corded Ware origin.
Check out the PDF I linked to. It explains the situation in Scandinavia at around 3,000-2000 BC.
I wish I could say that my holiday reading list was so virtuous, but must confess that the trashy stuff is just too alluring.
ReplyDeleteYes, this is just my own speculation. Nothing else, of course.
ReplyDeleteWe will see if the older samples from the Samara region from about 9 kya. are mostly European HGs with some ~15% West Asian/ANE (as I suppose), or if they indeed are mostly West_Asian/ANE with only a small part of HG ancestry.
This is supposing, that the steppe belt had a common culture/ancestry. It could be that this ANE population did live to the east of what we would call the Eastern HGs (say around the Baikal lake), but somehow I find it difficult that from that location they ended up spreading most of their genes in Pakistan and the Caucasus while having less impact on the HGs near them.
Hmm.. making the French 40% Yamnaya, creates a 43NE/32WHG/ 25ANE Yamnaya.
ReplyDeleteBy removing the East Eurasian part of Kazakhs, and redistributing according to other numbers, they would be about 20% WHG, Uzbeks around 13-14%, and Uyghurs around 12%
The fact that this test very likely provides conservative scores of WHG for Europeans should make it useful in measuring WHG admixture in Asia, because it's unlikely that Neolithic ancestry from the Near East is being misinterpreted as WHG.
ReplyDeleteI think that makes the East Asian Turkic ancestors for Kazakhs around 8%NE, 0%WHG, 12%ANE, and 80% EA. Making Kazakhs about 28-30% Yamnaya.
ReplyDelete"Hmm.. making the French 40% Yamnaya, creates a 43NE/32WHG/ 25ANE Yamnaya. "
ReplyDeleteYamna wasn't even close to 50% ANE. The Karelian HGs were probably 50% ANE or slightly more.
My guess is that on ANE K8 Yamna will score 25-30% ANE, 30%> WHG, and 35-40% Near eastern.
A Yamna/Gok2 person would turn out around 40% WHG, 44.5% Near eastern, and 15% ANE. That's a very broad estimate but still pretty close to French.
IMO, you need HG contribution, beyond Gok2, to explain WHG levels in north Europe. Some of those HGs, especially in the east probably carried ANE, which complicates estimating Yamna contribution.
I should change that Turkic part to mostly Turkic and other non-Yamnaya.
ReplyDeleteKrefter,
ReplyDeleteI said that Yamnaya is about 25% ANE. Please read it again. I already said that Gok2 is not enough from Northern France and higher on levels of WHG.
A line from Gok2 to 42NE/32WHG/25ANE should mean that everyone above that line needs more Loschbour/Motala/EHG, and below needs more NE. A line from CO1 to that marker will do the same.
ReplyDeleteUpon getting Yamnaya samples we can add crosshatches along the line between those two, according to the likely input from Yamnaya. We can then know for sure what we are looking at for input outside of Yamnaya, as long as those numbers outside of Yamnaya jive with something near Gok2 or CO1.
ReplyDeleteTry and work out the best estimate for Yamnaya, or maybe a couple of different estimates, and I'll plot them on the PCA to see what they look like.
ReplyDeleteI think that is the best one, at the moment. As I suspected, Lithuanians are 67% Yamnaya, only if they have 0%NE prior to Yamnaya in this run, so that is impossible. I think that everyone in Northern Europe is sitting between 40-55% max. I will work on this some more.
ReplyDeleteAll excess after 67% Yamnaya is 100% WHG in Lithuanians, by this marker. That is what I suspected when they announced that Lithuanians could be fit so high.
ReplyDeleteChad, check this out.
ReplyDeleteThere are of course multiple ways you can mix these ancestor pops up to get close to modern scores, but what I'm trying to show you is a Yamna+Gok2+HG can come close to explaining modern north Euros.
Yamna: ANE 30%, WHG 30%, Near eastern 40%
Gok2-ish: ANE 0%, WHG 50%, Near eastern 50%.
SHG: ANE 20%, WHG 80%, Near eastern 0%
CHG: 25% ANE, 75% WHG
BHG: 30% ANE, 70% WHG
50% Gok2+35% Yamna+15% SHG= 17% ANE, 47.5% WHG, 35.5% Near eastern.
Norwegian avg: 15.5% ANE, 47% WHG, 35.8% Near eastern.
38% Gok2+38% Yamna+24% CHG= 17.4% ANE, 48.4% WHG, 34.2% Near eastern
Belorussian avg: 16.9% ANE, 47.6% WHG, 33.3% Near eastern
35% Gok2+28% Yamna+35% BHG= 19.2% ANE, 51.1% WHG, 28.7% Near eastern.
Lithuanian avg: 18.1% ANE, 52% WHG, 28.3% Near eastern.
"I said that Yamnaya is about 25% ANE. Please read it again. I already said that Gok2 is not enough from Northern France and higher on levels of WHG."
ReplyDeleteMy bad.
If Lithuanians are lets say 65% Yamna, and Yamna was 30-35% WHG, their other ancestors had to of been on avg. 60% WHG and 4% Near eastern.
ReplyDeleteThat doesn't make sense though because what was the other 36%, it couldn't have been ANE.
Yamna was too near eastern and to ANE for Lithuanians to be over 60%.
Who was in NE Europe when Yamna arrived?
Krefter,
ReplyDeleteI know the possibilities. I will work out excess and Yamnaya input for a few pops tonight.
If CWC, Bell beaker, and Unetice were a simply mix of Gok2-C01-types and Yamna, there's no way they will cluster with NE Euros on a PCA of west Eurasia.
ReplyDeleteOnly bronze age Hungarians, and some modern central Euros can probably fit as Yamna+Gok2, all other North Euros need extra WHG.
The IEs who came to France were not pure Yamna, and so modeling French as Yamna+Gok2 doesn't work.
If North Euros have a significant amount native post-early Neolithic HG ancestry, Yamna can't be the only source of ANE, which makes Yamna ancestry drop significantly. That is unless somehow all their HG ancestry came from central-west-southeast Europe.
We don't have any Neolithic DNA from Lithuania yet, so there's not much point focusing on Lithuanians.
ReplyDeleteHowever, we do have a Central European from the Copper Age, CO1, and also Gokhem2, who probably got her genetic structure from Western or Central Europe, because she lacks ANE. So maybe it's better to focus on Western and Central Europe?
I'll run Oetzi later today, so that'll give us another Central European datapoint. I expect he'll be very similar to CO1, but we'll see.
I'll do Swedes, since we have Gok2.
ReplyDeleteKrefter,
I am talking about what is pure Yamnaya, for the French. Everything else is covering pre-Yamnaya and whatever they picked up.
French probably form a line between proxy-Yamna and Gok2, because they're a mix of Spanish-like and North Sea+Baltic-like people.
ReplyDeleteIEs didn't arrive west of Germany as pure Yamna.
Swedes are too tough. We would need Gotland samples. Swedes look way less NE than Gok2, even at 40% Yamnaya. We're talking about 30% NE, pre-Yamnaya.
ReplyDeleteEven if I drop Yamnaya to 40% NE, they are barely over 30%, pre-Yamnaya.
Let me try CO1 to BR2.
David,
ReplyDeleteWhat are CO1's numbers?
David,
ReplyDeleteThat score for the Swedes would basically make them 66%Gok2, 33% Motala, outside of Yamnaya.
"Swedes are too tough. We would need Gotland samples. Swedes look way less NE than Gok2, even at 40% Yamnaya. We're talking about 30% NE, pre-Yamnaya."
ReplyDeleteBut how could all that 70% non-NE be WHG? Gok4 may have been more WHG than Gok2, but still any extra HG-admixture from north Europe will have ANE in it.
I never said it was just WHG. Refer to the comment above yours. I said 2/3 Gok2, 1/3 Motala.
ReplyDeleteOetzi clusters right next to NE1, not CO1. The East Eurasian noise looks to be the same thing that the Bedouin have, and probably either a signal of some sort of archaic admixture, or perhaps just Basal Eurasian being misinterpreted.
ReplyDeleteIDs CO1 Oetzi_Iceman
ANE 1E-005 1E-005
South_Eurasian 1E-005 1E-005
Near_Eastern 0.557066 0.682094
East_Eurasian 1E-005 0.016529
WHG 0.442874 0.28674
Oceanian 1E-005 0.014598
Pygmy 1E-005 1E-005
Sub-Saharan 1E-005 1E-005
The North Italians are very close to where Oetzi was found.
In that case, a 33% Yamnaya mixture into them, makes modern North Italians.
ReplyDelete69/31/0 with just over 33% Yamnaya gets you to their 57/34/9. Oetzi may be on the higher end.
It looks like Europeans were much further apart before steppe migrations. Right at about that mark where farming stalled for 1000 years.
ReplyDeleteAs far as NE goes anyways. Northern Europeans were above Gok2 and breaking towards Motala.
ReplyDeleteBasques look like we can fit 25% yamnaya into them. Gok2 might be a great pre-yamnaya type for northern Iberia.
ReplyDelete"It's obvious now that Y DNA I1 and I2a2 are native WHG lineages of central-north Europe, because Gok2 and modern WHG levels prove there's a decent amount of native HG blood in north Europe."
ReplyDelete"Who was in NE Europe when Yamna arrived?"
I wonder if there are two kinds of ANE due to SHG and EHG both separately sharing ancestry with a northern interior (or altitude-ier) archaic.
I emailed Laz, and he said they haven't submitted their paper to anyone yet.
ReplyDeleteLaz,
"Not yet, we're working on it."
It doesn't look like we'll get it by Christmas.
Reich said they'd post it at bioRxiv anyway, so they don't need to submit it to any journal to do that.
ReplyDeleteLast time it took almost a year from when the paper was posted at bioRxiv to when it appeared in Nature.
@Krefter
ReplyDelete"You still need a population in Neolithic Europe who had more WHG than Gok2 and Balts, to explain admixture with Yamna and modern ANE levels. "
Gokhem4 clusters with Basques rather than with Sardinians.
https://sites.google.com/site/fennobga/CCEuroPCAINDGok4.png
Is there any preparation or review involved with posting their paper at bioRxiv? Can they post it any day they want to?
ReplyDeleteThe Gokhem4 sequence is really poor, but on the usual PCA of West Eurasia it actually clusters directly just below Gokhem2, rather than with Basques. What this means is that like Gokhem2 it also lacks ANE, but has more Near Eastern admixture.
ReplyDeletehttps://drive.google.com/file/d/0B9o3EYTdM8lQZFhiQUI5Y3pYc00/view?usp=sharing
Yeah, they can post their manuscript at bioRxiv whenever. There's no review process like at science journals. They can make changes to it and post new versions. That's what they did last time.
ReplyDeleteDavid,
ReplyDeleteWhen the Yamnaya paper comes out would you like to collaborate on something to submit?
David - I don't see much in the way of Uralic groups in your K8 test. Any idea how they'd place in it?
ReplyDeleteI realize there are some pretty big bottlenecks that would skew things, but I'd think they'd be a pretty good candidate for adding some of the ANE to Swedes. I'd think it'd be easier to find a clear signal there than than looking for a proxy for Proto Indoeuropean, as I doubt even PIE would have been particularly pure in the ANE/EHG/NE sense.
I would think that the Uralic groups would form a cline from West to East, and that the slope of that cline would be the same as the line one could draw between NE Europeans and Volga-Ural samples.
Davidski I know you think Yamna's Near eastern ancestry came through the Balkans not the Caucasus, but then how do you explain Yamna's near eastern mtDNA being so different from Neolithic central Europe?
ReplyDeleteI know modern Europeans, even Sardinians, aren't in the Neolithic central European mtDNA cluster even though they have so much ancestry from similar people, but still mtDNA doesn't support a close relation between Yamna and Neolithic central Europeans.
BTW, unless you haven't already done something similar I'm almost completely finished breaking down all pre-Iron age European mtDNA. Once I'm done around Christmas, I'll have an unbelievable resource for ancient mtDNA. What I'll have compared to Jean Manco's site will be like 2014 vs the stone age.
I've noticed ancient European mtDNA follows the same patterns as autosomal DNA.
It could have been predicated last year Bell beaker was mostly Yamna and very similar to CWC-Unetice(it was by some), but people insisted on making them Iberian and western-like. The old way of comparing H vs U frequencies and what not is over.
Out of 64 mtDNA samples from the Pontic steppe, only one has H1(all were tested for H1-SNP), none have H3(all were tested for H3-SNP), and none had J1c(all were tested for J1c-SNP).
ReplyDeleteThat right there makes them very different from Neolithic and modern Euros.
@Davidski
ReplyDeleteWhich one is the one than that is almost as northern as Orcadians in this plot from Skoglund? And which is the one that clusters with Basques?
http://3.bp.blogspot.com/-hoBLZ0dZXto/U1luxCudqwI/AAAAAAAAJk4/J3X428YdPzo/s1600/PCA.png
@Krefter
ReplyDelete"I've noticed ancient European mtDNA follows the same patterns as autosomal DNA."
Could you elaborate on this? I find that highly interesting.
Nice going, Skoglund.
ReplyDeletehttp://linearpopulationmodel.blogspot.com/2014/12/modern-swedes-as-even-split-between.html
"Could you elaborate on this? I find that highly interesting."
ReplyDeleteThe patterns in ancient Euro mtDNA, is nothing new. Once I finish breaking down ancient Euro mtDNA, I'll be able to look at it and compare the samples to each other, and find more detailed trends.
If I knew nothing of autosomal DNA I would break up ancient European mtDNA in this way. The labels are for convince.
First pre-farming Europeans.
EHG: Eastern European Hunter gatherers.
WHG: Western European Hunter gatherers. Probably also includes Baltic hunter gatherers.
BHG: Baltic(inclu. Sweden) Hunter gatherers: Mixture of WHG and EHG. Closest affinity is to EHG.
Next the Near easterns or farmers: Defined by high diversity in mtDNA N: H, HV(H-), J, T, U3, U8a, U8b, K, N*'s(X, W, I, N1a).
ENF: Early Near eastern farmer: The few samples we have so far from early Neolithic Near easterns are very different from Europe's first farmers. They also don't have any Euro HG mtDNA; U5, U4, U2e.
Next are later Europeans who were a mix of something Near eastern and HG.
EEF: Early Neolithic central Europeans. Mixture of something near eastern and WHG, mostly near eastern though.
PIE: Copper age Pontic Steppe: A pretty even mix of EHG and something near eastern in terms of percentages of haplogroups.
Next are people who were a mix of something PIE-like and something else.
CWC-BB-Unetice: Copper-bronze age central Europeans: They are a mixture of WHG(SHG?)-EEF and PIE, but mostly PIE.
EIE: Eastern Indo Europeans: Mixture of PIE and something east Asian. Mostly PIE.
Pierre,
ReplyDeleteMixing Motala and Gok2 won't work. There isn't enough Near Eastern and ANE. If you get enough ANE, then there is almost no Near Eastern. If you get enough Near Eastern, you're only going to be around 7-8% ANE. That doesn't work, sorry. You need the third population with higher Near Eastern and ANE.
That method of mixture won't even work for Estonians. You'd still be about 8% short on the ANE, at a 50/50 mix. Gok2 and Motala doesn't work for anyone.
ReplyDeleteChad,
ReplyDelete"Mixing Motala and Gok2 won't work. There isn't enough Near Eastern and ANE."
Then there is either something wrong with Skoglunds PCA, or with Davidski's ancestry estimator.
Given Skoglund's frankly amazing genetic drift result for Gokhem2, I'll bet on Skoglund.
In which case, it's easy to construct modern Swedes (on the PCA) as I said.
It doesn't work. Even using lazardis' EEF WHG ANE. I'm sorry.
ReplyDeleteSkoglund projects his PCA plots. David's don't, so they're more accurate.
ReplyDeleteRyan,
ReplyDeleteMy analysis includes Estonians, Erzya, Finns and Moksha. These are all Uralic groups. The reason I only have three Finns is because I started with three to make sure their recent founder effect/genetic drift didn't skew the results, and forgot to add more later. Adding more Finns won't push up the ANE in Swedes.
Krefter,
I never said all of the Near Eastern admixture in Yamnaya came from the Balkans. Some did come from the Caucasus, or around the Caspian.
epoch1970,
Skoglund's PCA suffers from projection bias, so it's difficult to make sense of his plot. I've told him about it, but his view is that PCA are just a rough guide, and the meat and potatoes of the study are in the formal stats.
Here's a PCA that doesn't suffer from projection bias, or at least not as much. It's from the Kostenki14 paper.
http://imageshack.com/a/img538/2013/ehUOcC.png
Pierre,
As usual, you don't have a clue what you're talking about.
Chad,
Sure, we can collaborate on a paper.
All pca plot's are projections, so to say that Davidski's plot are not projected indicates that you don't understand this. Lior Pachter has done some talks on pca projection. I did a blog post a while back on his talk.
ReplyDelete@Krefter
ReplyDeletepretty cool
Pierre,
ReplyDeleteEnough with the semantics. No one cares.
The point I'm making is that the placement of the ancient samples on Skoglund's PCA are incorrect, because he projected them onto the PC eigenvectors computed with modern individuals.
For more information see here:
http://eurogenes.blogspot.com.au/2014/05/pca-projection-bias-in-ancient-dna.html
David,
ReplyDeleteHave you tried the hypothetical Yamnaya on a plot yet?
I can run that now. What are the best guess ancestry proportions for the hypothetical Yamnaya?
ReplyDeleteDavid,
ReplyDelete" Adding more Finns won't push up the ANE in Swedes."
Sorry, I should have been clearer, I meant more eastern and/or more isolated Uralic groups. I was thinking that that would be a way to tease the ANE out, but I think the ENA component would confound that now that I think about it.
I'd be curious as to your thoughts as to this contradiction though. From your October 20th post:
"Razib also tweeted a few times from the talk, and as far as I can tell, his main point was that the Yamnaya samples showed affinity to the Ancient North Eurasian (ANE) proxy Mal'ta boy, but were also partly of Near Eastern origin, and indeed could be modeled as a 50/50 mixture between present-day Armenians and ancient Karelian hunter-gatherers."
Contrast this from December 15:
"It also suggests that, overall, the populations that moved west across northern Europe after the Neolithic, and shifted northern European genetic structure to the east, did not carry high ratios of Near Eastern ancestry. Instead, they harbored high ratios of ANE and WHG."
Clearly the two cannot both be correct. Personally I'd suggest the former as more likely.
One thing I'd keep in mind too is that the Uralic/Germanic linguistic divisions in Fennoscandia moved around a fair bit recently, with a lot of Forest Finns migrating to Sweden and Norway before being assimilated, and a lot of Swedes migrating to Finland.
I think the average will be around 43%NE, 32%WHG, 25% ANE.
ReplyDeleteMaybe make a few ranging from 38-45%NE, 28-35%WHG, 20-28%ANE.
Chad,
ReplyDeleteThe red dot is the Yamnaya average.
https://drive.google.com/file/d/0B9o3EYTdM8lQS1NIckhTZUs4TGM/view?usp=sharing
Ryan,
Even if the Yamnaya nomads who moved deep into Europe were basically indistinguishable from the Yamnaya samples from the Samara Valley (which is close to the Urals and Caucasus) that can be modeled as around 50/50 EHG/Armenian, it still wouldn't contradict what I said.
That's because Armenians are possibly as much as 15% ANE, which means that a 50/50 EHG/Armenian mix wouldn't have a much higher level of Near Eastern ancestry than most modern Northern Europeans.
However, the people who moved west across the north of Europe weren't Yamnaya nomads, but rather various Corded Ware groups like Battle-Axe and Single Grave. I'm still of the opinion that their levels of Near Eastern ancestry were very low, and possibly on a par with those seen around the Baltic today. The only thing that will change my mind is an ancient genome that says otherwise.
By the way, I can't remember if I have any Uralic groups from east of the Urals. But the full K8 spreadsheet includes Kets, Selkups and many other Asians:
https://docs.google.com/spreadsheets/d/1kd9Q9vFrL1Cra9ayqMYVFKXrUdnThmQJVMtjczLhoTs/edit?usp=sharing
And in regards to Uralic admixture in Northern Europe potentially pushing up the proportions of ANE there, to quote myself from a few months ago...
"The various Uralic, Turkic and Mongolian groups expanding into Europe, usually after the Bronze Age, no doubt carried significant ENA, so these groups can't be the source of the fairly high levels of ANE across Europe today, because most Europeans lack ENA. Below is a graph based on two f4 tests, comparing ANE and ENA ancestry among Europeans, this time with the Han Chinese as ENA proxies. Note that most of the samples fall within a cline that runs from the Stuttgart sample to Estonians. The only outliers in the direction of the Han are groups from current or former Uralic and Turkic speaking areas of Europe."
https://drive.google.com/file/d/0B9o3EYTdM8lQN1RBR2N2QTlkOTg/edit?usp=sharing
http://eurogenes.blogspot.com.au/2014/01/another-look-at-lazaridis-et-al-ancient.html
Also, an important point: the reason the cline slopes isn't because of East Eurasian, or ENA, admixture, from Uralics in Northern Europe, but due to a lack of Basal Eurasian ancestry there.
Actually, note that the full spreadsheet also includes some Karitiana Indians, who come out 41-43% ANE and the rest ENA. So the ANE is definitely teased out from the ENA.
ReplyDeleteThat actually looks pretty close to the freehand. Draw a line from that to North Central Euros, then to LNE/EBA, and MNE, the distances look pretty close.
ReplyDeleteI think that Corded will be modeled as 73% Yamnaya solely on the fact that it is 18% ANE. It is obviously mixed with Gok2/Motala/EHG types in NC Euro at the time. It certainly should be less NE than Yamnaya. The freehand drawing showed that too. I think that Corded will be 36/46/18. Bell Beaker will be 41/44/15.
See how those Corded and Beaker numbers look. We can adjust from there.
ReplyDeleteRed = Yamnaya
ReplyDeleteOrange = Corded Ware
Yellow = Bell Beaker (from east Germany, presumably)
https://drive.google.com/file/d/0B9o3EYTdM8lQVzl5S1ZOZ01BMTA/view?usp=sharing
I think that looks reasonable. The Corded Ware dot is actually very close to me.
Are you able to plot that on a PCS that is flipped the right way? That might make it a little easier to judge the next move.
ReplyDeleteDo those all look reasonably close to the freehand, to you?
ReplyDeleteCan we then draw a line to the point where Yamnaya to Corded crosses the Gok2/Motala/AJV line? Should we say modeled as a mix of 60% Yamnaya, 40% something between Gok2/SHG that might carry 7% ANE, on the way to NC Europe? Beaker as 55% Yamnaya, mixed with something that is Gok2 like, but with a hair more WHG and 3-4% ANE, maybe from Corded being in Germany 200 years earlier. ???.. Does that sound plausible?
ReplyDeleteI think that looks pretty close. Maybe Yamnaya should be a bit further up and left based on the freehand plot? But it's hard to say how accurate that drawing is anyway. Yamnaya definitely should be pulling towards MA-1, and it seems to be doing just that on this PCA...
ReplyDeletehttps://drive.google.com/file/d/0B9o3EYTdM8lQcm9wY1BkU3NrV3c/view?usp=sharing
Corded Ware is just above UkrLv237, while the Bell Beaker is lost at the bottom of the tight British cluster.
I'm unable to draw lines on these plots at the moment, but if anyone else wants to have a go, then by all means.
Hmm.. Maybe Yamnaya as 41/34/25. That should bring it up and left a hair.
ReplyDeleteThat leaves us a little more room for Near Eastern survival in Europe, as well. That might be a little easier pill to swallow.
ReplyDeleteWhat about using your other plot that is turned and flipped? Can you draw lines, like the one from Gok2 to your SHG? Are you able to pull numbers from that locations, as to what type of mix it is?
ReplyDeleteHere's what I get with 42/36/22 for Yamnaya.
ReplyDeletehttps://drive.google.com/file/d/0B9o3EYTdM8lQbHBjVmx5NUxzYTA/view?usp=sharing
https://drive.google.com/file/d/0B9o3EYTdM8lQZkFsZkhMTGd0ZHM/view?usp=sharing
That must be close, but we won't know how close until we see the paper.
Anyway, this outcome really makes me curious what IR1 would get. I might have to try it, despite the lack of markers.
Btw, I really can't draw lines at the moment.
ReplyDeleteOkay, no problem.
ReplyDeleteI don't think that we can adjust the 25% ANE. That was kind of the whole premise for them calling Corded as 73% Yamnaya. We would have to lower Corded's ANE too. Does adding WHG and decreasing NE bring it up and left at all?
Try that 41/34/25.
ReplyDeleteOr, at least follow the slight slant from Stuttgart to Loschbour.
ReplyDeleteEither way, Going up would look a hair better.
Here's 41/34/25.
ReplyDeletehttps://drive.google.com/file/d/0B9o3EYTdM8lQbEJVODNsd1JaZ3c/view?usp=sharing
https://drive.google.com/file/d/0B9o3EYTdM8lQb0QxS2VnU295OHM/view?usp=sharing
It looks like that did shift it up towards Corded a little. I think that looks pretty damn good. If you want to go as low as 40/35/25, we might be okay there.
ReplyDeleteI don't know if I'd want to go any higher than that on WHG. We will be losing quite a bit of Yamnaya relation in Central Asia and the Near East. I still favor the 43/32/25, but we will have to wait and see.
I'd say 35% WHG is too high.
ReplyDeleteQuestion is, can 41/34/25 and/or 43/32/25 be modeled as 50% Armenian?
That pretty much makes North Italians 66% Sardinian, 33% Yamnaya. Basques at about 75%Gok2, 25% Yamnaya.
ReplyDelete43 can, I think. That's about 54-55% EEF. We can see what 45/20/25 does.
ReplyDeleteOtherwise,
ReplyDeleteYamnaya 43/34/23
Corded 36/47/17
Beaker 41/44/15
That makes Corded about 74% Yamnaya, based on ANE, and also brings Beaker closer to Corded, as the drawing was.
I gotta go. Need to do some late Christmas shopping.
ReplyDeleteI'd better call it a night soon. Thanks for all of the work. Feel free to play around with it. This last one may give you what you're looking for. It should put Corded more into Belarus and Poland, than Ukraine. I think that is what we're looking for, right?
ReplyDeleteEnjoy the shopping. Try that last one, when you get a chance. It sounds better.
ReplyDelete"That pretty much makes North Italians 66% Sardinian, 33% Yamnaya. Basques at about 75%Gok2, 25% Yamnaya."
ReplyDeleteDon't forget about non-Yamna, non-HG, and non-EEF ancestry in Europe. It is there, especially in southeast Europe.
South Italians for example have about as much near eastern as Stuttgart, but close to 10% ANE.
Krefter, we can attack that after figuring out Northern Europe. Late West Asian admixture with ANE might be an impossible task. We'll see.
ReplyDeleteThere are some Europeans like Basque and Lithuanians who appear to have little non-Yamna, EEF, HG ancestry, and we should model them the most we can. We should not attempt to model Greeks for example in that way, because they obviously have a more complex history.
ReplyDeleteAlso, Chad back to the debate over WHG in Indo Iranians. Sure there may have been very near eastern-people east of Samara, but bronze-iron age IEs in northern Asia had identical mtDNA to Yamna, with a high amount of EHG mtDNA. There's no doubt they had noticeable WHG.
northern Asians are scoring significant WHG in ANE K8. Kets are scoring close to 15%, and around 1% Near eastern. Even Mongolians are scoring WHG, and slightly more Near eastern
I'm not surprised about the Kets because of their EHG mtDNA, and ancient EHG mtDNA in Siberia(There's a U5a dating back to the Mesolithic).
Unless the Near eastern in north Asians is from a separate source of the WHG(Mesolithic?), I think it's evidence Indo Iranians have WHG.
Shit. I've been thinking about those angles in the drawing. Yamnaya to corded and to MNE was almost a right angle. We may need 45/32/23, for Yamnaya. Maybe as high as 50/27/23. That should be half Armenian, don't you think? At least close. Remember, it can't look pretty as a large source of our ancestry.
ReplyDeleteDavidski, since the Ughur are a mix of Turks and Tocherians, can you put them through the ANE K8 test? If they score like 15% WHG, there's no doubt Tocherians and therefore probably Indo Iranains had noticeable WHG.
ReplyDeleteKrefter,
ReplyDeleteUygurs are in this spreadsheet...
https://docs.google.com/spreadsheets/d/1kd9Q9vFrL1Cra9ayqMYVFKXrUdnThmQJVMtjczLhoTs/edit?usp=sharing
Chad,
Yamnaya can't be 50% Near Eastern, because Armenians aren't 100% Near Eastern.
I'm not as knowledgable as most who comment on this blog. To visualize it for myself I flipped and rotated the graph so that the axis is oriented like Europe North/South East/West.
ReplyDeletehttps://dl.dropboxusercontent.com/u/10788581/West_Eurasia_K8c_3.png
If you draw a line between Stuttgart and Losch, Gok (and I think the Basques) are right on the line. I can see the population mix of Neolithic Europe clearly on this line as percentages of WHG and Near East. The population of Europe today is shifted towards the Volga (Eastward).
Again, I'm sure I'm stating the obvious for most of you folk here, but it was an interesting exercise for my own visualization.
Those are Sardinians not Basques, but yeah, that's pretty much it.
ReplyDeleteDavid, this component doesn't behave like that. Stuttgart is 72% NE/ 100%eef, Brits are 41%NE/ 50%eef. That's why I went with 43%. That's just over 54% EEF. Armenians have more basal than Stuttgart, 44-45% NE might be safer. We can still stick with 23% ANE. Where do you think Karelians would fall? East of the Finns?
ReplyDeleteBut to keep things as straightforward as possible I'm just focusing on this test, and not worrying about the EEF estimates, which might soon not be relevant anyway.
ReplyDeleteIf Yamnaya are roughly 50% Armenian, then we've got the 50% worked out already, because we have Armenians in the spreadsheet.
So all we have to do is try and guess EHG. The only thing we know about EHG is that it's likely to be 0% Near Eastern, because, among other things, it clusters way up north just east of SHG. Thus, all of the Near Eastern admixture in Yamnaya must come from the Armenian average.
What else do we know? Corded Ware are 73% Yamnaya and cluster in North-Central Europe. Let's assume then that the two sample sets from North-Central Europe, Poles and Swedes, are around 73% Yamnaya.
Using all of that info, would would Yamnaya and Corded Ware be in the context of this test?
If that freehand sketch is accurate, EHG is about halfway between SHG and ANE in east-west axis (considerably more distant from SHG than SHG is from WHG) but much further "north" than a straight line between SHG and ANE.
ReplyDeleteSorry if I'm not following the question, but solving all above to guess EHG all depends on the other side of who is mixing with Yamnaya to generate Poles or Swedes. Without that being known, it varies.
ReplyDeleteStill, I tried to model the outcomes where Yamnaya is 50% Armenian and 50% X and Polish are either 73% Yamnaya and 27% Motala12 or 73% Yamnaya and 27% Gokhem2.
http://i.imgur.com/nXyy2Qp.png
It seems that Poles cannot fit as a combination of 73% population which was 50% Armenian and 50% X (aka EHG) plus 27% Gokhem2, as they would already get too much Near Eastern from this known ancestry and EHG would be required to have negative Near Eastern (which is impossible).
They can fit as a combination 73% population which was 50% Armenian and 50% X plus 27% Motala12... buuuut... this requires EHG to have Near Eastern...
Using 13.5% Motala12 and 13.5% Gokhem2 and 73% Yamnaya (who are 50% Armenian) for Poles gives a fairly sensible EHG, with the main question there being whether that is a reasonable scenario at all, archaeologically.
Same thing for Swedes - http://i.imgur.com/oBXlfAU.png
..
David, this is a minor and unrelated to the drift of the other comments and the main post, but looking at the full spreadsheet for K8, the African Americans samples average around 18% of Near Eastern+ANE+WHG and average 5% ANE, 4% Near Eastern and 9% WHG, the rest being around 80% Pygmy plus Sub Saharan.
That would seem to imply that African Americans West Eurasian ancestors would have 27% ANE, 23% Near Eastern and 48% WHG.
If you take the African Americans samples with greater than 20% Near Eastern+ANE+WHG, then you get an average of 6% ANE, 8% Near Eastern and 13% WHG, which implies 22% ANE, 29% Near Eastern and 48% WHG, or a little east of Lithuania.
Both seem kind of weird as we know mainly English and a little Spanish and French were African Americans' West Eurasian ancestors.
Although this test is not intended for non-Europeans to determine their balance of West Eurasian ancestry, I was wondering whether there would be any reasons for this besides low SNP overlap or number.
What do Latin Americans samples who should fit as purely mixes of Spanish and Native American (e.g. Mexican subsets who only fit as members of Amerind and European clusters in other tests) fit like? What about 3 way mixes of Amerind-African-European?
Some Sub-Saharan Africans score 1-2% ANE, which is obviously not ANE, but something else not covered by the components in this test. For instance, have a look at at the results of Anuak Ethiopians.
ReplyDeleteMA-1 does score some Sub-Saharan admixture in many tests, so there might be an ancient Eurasian or archaic link that's causing this. But I don't think I'll be able to fix it.
Davidski, do ANE K7 and K8 not work well with west Eurasians with minor Sub Saharan and east Asian?
ReplyDeleteThis test works well for all West Eurasians, including those with various admixtures, and it'll be available shortly. So stay tuned.
ReplyDeleteStrange. I didn't notice that at all as I was only looking at the groups with more than 15% Eurasian at first. I could see that might add into the true West Eurasian ancestry via CEU to inflate the ANE balance of their West Eurasian components. Relatively high WHG maybe due to chance on small sets of markers?
ReplyDeleteI guess that happens in the Yoruba / Mandenka who are the most likely candidate for the African Americans' African ancestry as well?
Yes, it happens with the Mandenka, and probably Yoruba, but I'd need to check the latter.
ReplyDelete@Davidski
ReplyDeleteI had a look at this graph:
https://dl.dropboxusercontent.com/u/10788581/West_Eurasia_K8c_3.png
Am I right in seeing that modern day Sardinians are on the "Neolithic" population line.
And if so, isn't it possible that this shift of most Europeans toward the Baltic is mostly due to a long term fusion between North and South Paleolithic/Mesolithic/Neolthic Europeans.
Yes, there's some influence from the Volga/Steppe, but much of it could have come secondarily from the Baltic/Scandinavia, as most of North and South European populations fused during the Neolithic.
I'm not saying that there couldn't have been a minor wave from the Steppe during the Bronze age, or later, but I suggest that it is not higher than about 20%.
I know you don't want to focus on the near east, but a better method might involve lumping Stuttgart with NE1 and Sardinians, then pulling out a good chunk of the WHG and any ANE. Modern Bedouins could be some crazy mix from later migrations and drift.
ReplyDeleteI think the 43% NE works for Yamnaya. Corded and north central euros can be fit as 73% yamnaya, but aren't literally that. That would leave about 5% NE with something close to motala making up the rest. I think the true number is closer to 55-60% yamnaya, mixed with something between Gok2 and your SHG composite. That seems more likely with their route through western Russia and Belarus.
I wonder why it's people who live by the North Sea and Baltic Sea who have the most HG ancestry passed a Gok2-Yamna mix.
ReplyDeleteFor Europeans that do form a cline between Gok2 and Yamna, it's probably not because they lack excess-HG ancestry. It's probably because they're a mix of something like modern far northern Europeans(North and Baltic seas) and something more Near eastern than Gok2.
I noticed by simply glancing at admixture scores last year that French look like a mix of Irish and Spanish, and Spanish look like a mix of Irish and Sardinian. Maju and others though of course called me crazy.
But it's pretty obvious north Euro(CWC, BB, Unetice) and south Euro-like(C01, Otzie) populations mixed with each other during the copper-iron ages to create central Europeans. That sounds simplistic but is probably very true.
Well Marnie, we have three samples from the very late Neolithic/Copper Age from three locations in Central Europe that fall along the Neolithic cline with the Sardinians:
ReplyDeleteGokhem2 from Sweden
Oetzi from the Alps
CO1 from Hungary
A <20% replacement scenario doesn't work with these genomes, because they have 0% ANE, and most of the people who live at these locations and west of them now have well over 10%.
So it is what it is: a major population turnover across Western and Central Europe during and/or after the Copper Age.
Well, Davidski,
ReplyDeleteA <20% replacement scenario doesn't work with these genomes, because they have 0% ANE.
Could it perhaps be possible that your "ANE" is particularly biased toward picking up Bronze Age and later Steppe "ANE"?
Gokhem2 and Oetzi were very southern looking, even in the Neolithic. There are other ancient DNA samples that don't look like this. I know it's been discussed, but what about the Motala samples?
Even on the original Lazarides PCA, from January, if you take Oetzi like samples and average them with Motala like samples, you would get something like Southern French, Basques, Pais_Vasco and Spanish samples.
From here, you're talking about a very minor shift to express most of the population of Europe.
Have a look:
http://linearpopulationmodel.blogspot.com/2014/01/mesolithic-western-european-hunter.html?m=0
"A <20% replacement scenario doesn't work with these genomes, because they have 0% ANE."
ReplyDeleteCould it perhaps be possible that your "ANE" is particularly biased toward picking up Bronze Age and later Steppe "ANE"?
Gokhem2 and Oetzi were very southern looking, even in the Neolithic. There are other ancient DNA samples that don't look like this. I know it's been discussed, but what about the Motala samples?
Even on the original Lazarides PCA, from January, if you take Oetzi like samples and average them with Motala like samples, you would get something like Southern French, Basques, Pais_Vasco and Spanish samples.
From here, you're talking about a very minor shift to express most of the population of Europe.
Have a look:
http://linearpopulationmodel.blogspot.com/2014/01/mesolithic-western-european-hunter.html?m=0
Marnie,
ReplyDeleteThat won't work. It could be what NE Europeans were before Yamnaya, but you still need another 10% ANE. Yamnaya is probably in the 23-25% range, so you need to add 40% of that or more for the Northern half of Europe and 25-40% for the South.
You're assuming these figures actually mean anything
ReplyDeleteLooks like people are just massaging numbers around , without producing anything meaningful. Sorry.
Mike,
ReplyDeleteThey are plotting about where they are supposed to. Corded Ware was supposed to plot with North Central Europeans, and Yamnaya was SE of there a ways. Bell Beaker was barely south and west of Corded. We have accomplished that and stayed in that supposed 73% range.
If anyone thinks they can do better, or have an idea, please do share.
I've got a bit of a hypothesis here.
ReplyDeleteI think that the pastoralists from the Caucasus are going to be pretty close to modern Armenians, and the Mesolithic Yamnaya hunters are going to be about 65%WHG, 35%ANE. The pastoralist will be about 85% NE and 15%ANE.
I think that the pastoralist are either after ANE spreads into the South Caucasus (via Iran), or from the opposite direction. Steppe and Caucasus mingling started in Maykop's precursor.
It's probably more complicated than that, and likely that Sredy Stog/Dereivka had EEF/NE that they spread into the area first. Yamnayans probably had some NE, prior to the arrival from the Caucasus, of pastoralists. It could totally change the Mesolithics as 50/50, mixing with Stedny Stog, and then Caucasus folks. Lot's of possibilities!
ReplyDeleteIs anyone in contact with Laz, Reich, and others? I thought their paper would be out by now, it's the 1 year anniversary of their last paper.
ReplyDeleteJust fitting some things together. Not to be taken literally...
ReplyDelete50% Yamnaya
50% Gok2
46.5/42/11.5 French?
50% Yamnaya
25% Gok2
25% Motala
34/49.5/16.5 Looks like Corded
50% Yamnaya
37.5% Gok2
12.5% Motala
40.25/45.75/14 Kind of like Beaker
50% Beaker
50% CO1
50.5/42/7.5 Basque?
50% Beaker
25% Corded
25% Gok2
42/46.25/11.75 Northern France?
Actually, that Beaker and CO1 would be Spaniards, right?
ReplyDeleteBeaker and Gok2 makes Basques.
Spanish kind of look like SW French with some recent Near eastern admixture. I suspect there's some significant(probably not 20%) recent Near eastern ancestry in Spain, but little in SW France.
ReplyDeleteI would model with SW French instead of Spanish.
SW France had several Greek colonies. The same will go there. It's just for fitting purposes. Again, not literal.
ReplyDelete60% Yamnaya
ReplyDelete30% Gok2
10% Motala
40.8/43.4/15.8 Czech/SE German maybe?
I'm ready to bet money now that R1b-L11 was spread with IE languages and by BB or Hinxton(Iron age) type people.
ReplyDeleteWhat do you think Davidski?
The reason I think this 5000YBP R1b in Bell beaker, Bell beaker being mostly Yamna, and a signal of North Sea-type ancestry in all western Europeans.
I'm for example 50% North Sea, 20% Spanish, and 25% south-central Euro and can easily fit as a Frenchmen.
During the 20th century and bronze age the same elements admixed independently to create me and French.
@ Krefter
ReplyDeleteI have assumed in the past that looking at Central Asians stripped of East Asian ancestry could give us some clues as to how the Yamnayans looked like.
The WHG:ANE ratios of all pops there are very low. While the increasing amount of ANE and NE vs WHG as one goes deeper East to East Asia and South to South-Central Asia probably means that the figures are artificially depressed due to local admixture with ANE and NE-rich populations, I find it hard to believe that more than half of pre-mongol pre-turkic ancestry derived from locals, which would be necessary to get from a 1:1 or >1:1 WHG:ANE ratio in the Yamnaya population to what we have in Central Asia today. If corded ware is 73% Yamnaya, and Europe was almost certainly more densely populated than the Central Asian desert, there is no reason why the intrusive nomads should not become genetically even more dominant in Central Asia than they were in Europe.
@ Chad
I think one problem with all the figures thus far is that a line drawn from Armenian to Yamnaya would imply the EHG population was in the vicinity of the SHGs, when the drawn pca figure from the talk implies this isn't the case. The EHGs there are far to the east of the SHGs.
I suspect that Yamnaya is probably further east in the plots, such that EHG is east of any European population today. I'm also not sure we can assume that Central-North Euros are 100% Corded Ware. Last of all it seems the 73% yamnaya figure for Corded Ware was derived from a simple doubling of the 36% non-local 'EHG' ancestry, which tells us very little about how much ANE was derived from Yamnaya as opposed to existing beforehand from either Motala or EHG. As you said before, Reich did not think Yamnaya was a good source for the ANE in Corded Ware.
Ryu, that would probably be accomplished by making Yamnaya more near eastern. I can't raise the ANE, as Corded has to stay in North Central Europe, so changing the yamnaya ANE wouldn't be wise, with the 73% deal. We could maybe make Yamnaya 25% ANE and Corded 18.5, but no more. We'd be exiting north Central Europe after that. Adding a percent or two of Near Eastern would help. That 50% wasn't EHG, but Karelians, from what I remember. Our Yamnaya isn't far from that line. Karelians should be just east of the Finns.
ReplyDeleteWe can see where 44/31/25 goes. How about that David? See if that puts us between Armenians and Karelians.
ReplyDeleteNevermind, David. That's too far East. 46/32/22, might put us between Armenians and something just NE of Kargopol Russians. I think Karelians would be near there. What do you think?
ReplyDeleteThat should be much closer to the middle, and as Reich said, not a good source for direct ancestry, but I think something more Western Steppe, with less NE and more WHG. The ANE could even be a smidge higher away from all that Near Eastern ancestry.
ReplyDeleteI think Matt showed what the problem with the model is. That is probably what Reich sees. One thing that I see about Corded Ware, is that it has the West Asian mtDNA, but also a shit ton of the Narva stuff. That may be why their Near East crashed and the WHG rose up, while leaving that ANE up around 73% of Yamnaya.
ReplyDelete@ Chad
ReplyDeleteThose Karelians were ancient Karelians, from which Laz postulated the EHG population. It is practice in scientific papers not to say that "Europeans are Gokhem + Loschour" but "Europeans are EEF + WHG" so that the tree can show WHG branching from Loschour and EEF branching from Gokhem. In this case Karelians were classified as EHG, and we know roughly where that scores on pcas, which is far from just slightly east of the finnish.
I think the issue here is the chain of implication between North Central Euros as 100% Corded Ware, plus Corded Ware 75% Yamnaya, plus Yamnaya as 50% Armenian. Reasoning from these alone makes for a high-WHG Yamnaya which results in contradiction between the results obtained and the general position of EHG in the plot.
One or more of these has to go to get a logically consistent picture. I think the 73% Yamnaya one is the weakest link, cos I suspect that this figure is simple extrapolation from 36% Karelian/EHG ancestry. It could just be that Corded ware is 36% EHG which implies 75% Yamnaya if Yamnaya was the sole source, but like Reich said not really, because blah.
If a motala-ancient Karelian continuum existed across northern Europe, there is no reason why populations with ANE higher than Motala but lower than EHG could not have increased EHG affinity in northern Europe, or for that matter in Corded Ware itself, independently of Yamnaya. This would explain why Yamnaya has Caucasus contribution, but this contribution reaches a mminimum in Balts today.
Yes, I agree that this is complicated. I do think that Corded is certainly mixed with something that is a good amount of ANE, outside of Yamnaya. Beaker too, for that matter. There is Narva stuff popping up in the South Baltic around that time. Also the arrival of Pitted Ware in the North, from the East Baltic. It will make a dent in the Yamnaya numbers. When we have the Yamnaya and EHG numbers, this will make more sense. I hope...
ReplyDeleteDavid,
Could you please plot 70WHG/30ANE, 60WHG/40ANE, and 50/50, with Armenians? I want to see how those dots look compared to the hypothetical Yamnaya and Armenians. (78 Near East, 4 WHG, 15 ANE, 3 South Eurasian) The plot with red dots may be easier to see, or whatever is easiest for you. Thanks!
I think an EHG between 30-40% ANE will be needed to line up with Armenians, and staying within 73% of the ANE in Yamnaya, while keeping Corded in North Central Europe.
ReplyDeleteRyuk, when did Reich say Yamna isn't a good source for ANE in CWC?
ReplyDeleteThere's no doubting anymore Yamna had noticeable WHG. Patterson wouldn't have suggested Maykop as PIE if that wasn't the case. CWC wouldn't be clustering in central-north Europe if that wasn't the case.
I think EHG's easterness on the hand drawn plot is off. It wouldn't make sense for EHG to be something over 60% ANE, if CWC was 36% and clusters in central-north Europe.
CWC was probably around 18-20% ANE. Yamna was probably 25-30% ANE. This is foundational knowledge now.
If proto-Indo Iranians were lets say 30% WHG(in ANE K8) and Kurds are 15% proto-Indo Iranians, only 4.5% WHG will be contributed to them. IEs possibly took a more complicated route into west and south Asia than into Europe, which slowly diluted their blood, especially WHG because the people they mixed with had close to none.
@ Chad
ReplyDeleteSW France had several Greek colonies. The same will go there. It's just for fitting purposes. Again, not literal.
Greek colonies were mainly in SE France .The SW French are right next to the Basques and were not colonized by the Greeks.
Helgenes,
ReplyDeleteThat's correct, sorry.
Kester,
The EHG can't be over 60% ANE, that would make Yamnaya over 30%, which can't happen if we want to keep Corded around Belarus and Poland.
David,
Try this... I will literally make it half Armenian and something with no NE.
34/40/26 for Yamnaya
34/47/19 for Corded
38/47/15 for Beaker
This will make the 73% Yamnaya very feasible.
That puts Corded at exactly 73% of the ANE of Yamnaya, and it should fit the rest as well. Corded will stay around Belarus too.
ReplyDeleteHow does that look?
@ Krefter
ReplyDeleteReich said that in an email exchange with Chad. Thats why I think the 73% figure is highly suspect.
I never said that Yamnaya did not have WHG. I just said that the figures we have now have WHG too high in relation to ANE.
You said yourself in previous posts that Andronovo and other cultures in central Asia were more or less identical on both sides of haploid markers as Yamnaya, which implies that there could not be too much mixing with locals, unlike those in IE-ized European cultures like Baden or CW.
I'm just saying I find it impossible to believe that proto-slavs and germanics in the Central European forests, which had strong farming cultures going, preserve the genetic legacy of Yamnaya steppe herdsmen better than the far more similar Indo-Iranian herdsmen on the Central Asian steppe. While there was farming in Central Asia it was very limited. If, even with farming, thick forests and all that, we take at face value that a 73% contribution from yamnaya to Corded Ware was effected, then what must the ratio be in places like kazakhstan?
If we take that into account, plus the ddriving up of WHG and some ANE in corded ware vis-a-vis Yamnaya, I think we would get a far more coherent picture.
Using the hand drawn plot from David, a graphics editing program and basic geometry, I tried to give a rough idea of where EHG would plot on the PCA being toyed around with, so that we could try and plot Yamnaya as 50% EHG and 50% Armenian.
ReplyDelete"Bell Beaker" should lie roughly between EHG and MNE at about a 45/55 split. I labeled the location of MNE as best as I could and then used the Bell Beaker location to plot a line and give a rough idea of where EHG would be given the current plot.
http://tinypic.com/view.php?pic=2e30c2o&s=8#.VJo_XF4AKA
That remainder after the 73% Yamnaya, is just more WHG like than Gok 2. That fits a lot better now. Once David plots those new numbers, it should look better.
ReplyDeleteRyuk, central Asians are not a simple *pure* east Asian and Yamna mix. I don't think it's safe to base anything about Yamna on them.
ReplyDeleteLaz suggested a massive population turnover in Europe caused by Yamna. If they were anything under 30% WHG, that's not possible. We'd be talking about a just as large comeback of local hunter gatherers, if Yamna was only 20% WHG in ANE K8.
If that is the case... Patterson has a point, but there are many possibilities here. Pre-Yamnaya Kazakhstan was likely just NE/ANE, with not much WHG. So, the door is still kind of open for a steppe dispersal of IE. Let's wait and see what Maykop was like, with haplogroups and their dispersal.
ReplyDelete@ Krefter
ReplyDeleteI'm not sure it came from Yamna, as much as from Corded Ware, which I suspect is less close to yamna than most people suspect.
Does anyone know exactly where are the Northern French on the PCA ?
ReplyDeleterk,
ReplyDeleteYou're misinterpreting what Reich said. Chad posted the quote, which said that the Yamnaya samples from the Samara Valley didn't look like the perfect source of the eastern admixture in modern Europeans, but they were probably related to this source.
I can think of a couple of groups closely related to the Samara Yamnaya that would fit the bill, including Yamnaya from further west and Corded Ware. Wouldn't you agree?
Also, the whole point of the ASHG presentation on the Corded Ware was to inform that this group wasn't of local Central European origin. You should read the abstract from the talk.
http://eurogenes.blogspot.com.au/2014/09/corded-ware-culture-linked-to-spread-of.html
It seems to me like you're trying very hard to find an alternative explanation to the Indo-European expansion for the massive shift in the genetic structure of Northern and Eastern Europeans after the late Neolithic. But it won't work, because Corded Ware came from the east at just the right time, they were "surprisingly" eastern, as I was told by someone who was at the ASHG talk, and the main clades of R1a that expanded across Northern and Eastern Europe during and after the Copper Age are closely related to the main Indo-Iranian R1a clade.
By the way, the early Slavs were highly mobile and mainly cultivated millet, which entered Europe from the steppe. Stable isotope data also show that the Hallstatt Celts cultivated millet.
So the close position of the Corded Ware and Samara Yamnaya on that freehand plot might well be correct, and in fact other Yamnaya groups might cluster even closer to Corded Ware.
Chad,
I can't run any plots at the moment, but will tomorrow.
Okay, sounds good.
ReplyDeleteOkay. Let's go with these, with the Armenian numbers, and EHG at 60 WHG/ 40% ANE.
ReplyDelete34/40/26 for Yamnaya
34/47/19 for Corded
38/47/15 for Beaker
@ David
ReplyDelete"which said that the Yamnaya samples from the Samara Valley didn't look like the perfect source of the eastern admixture in modern Europeans, but they were probably related to this source"
This is exactly what I've been saying--that a direct line drawn from Yamnaya to Corded Ware to North-Central Europeans for us to infer backwards from NC Europeans today does not produce accurate estimates for Yamnaya.
As for the millet subject, the main issue here is the relative density of the admixing population in Europe vs Central Asia, with Europe being more densely populated and ecologically more hostile to horses and chariots and a mobile lifestyle than Central Asia is. Mallory already raises issues about horses and chariots in Europe--and he is a supporter of the steppe hypothesis.
The point I am making is that, while the ANE increase most certainly derives its ultimate origins from the oldest PIE cultures like Yamnaya, the WHG increase probably cannot be traced to this, but to some secondary local phenomenon that increased WHG within Corded Ware.
Therefore to say that groups like Lithuanians and Estonians strongly resembled the earliest PIEs is probably counterfactual.
Ryu, in case you're interested, here is a PCA where the position of a sample is determined by the balance of each of the three components Near Eastern, WHG and ANE. Includes the West Eurasian panel and others with over 15% West Eurasian.
ReplyDeletehttp://i.imgur.com/UOA6Oec.jpg - on a grid
http://i.imgur.com/XuvSCT3.jpg - rotated
(http://i.imgur.com/7ykfnid.jpg - rotation plus some distortion that I feel may better represent the distances between the components, but that's just subjective based on my read of the FST table and Laziridis).
The populations with dots are coded as in David's PCA, while the squares are other samples with the West Eurasian components.
Aqua square - Karitiana (i.e. 100% ANE)
Blue squares - Northeast Eurasians /Siberians
Cyan Squares - Kazakhs
Red Squares - Central Asians (incl. Pakistan, Iran, Afghanistan)
Yellow-green squares - Southeast Asians (obviously with Indian ancestry)
Yellow squraes - Maris
Orange squares - East Africans (Somalis, Maasai etc.)
Purple squares - North Africans (Egyptians, Moroccans, etc.)
Brown squares - African Americans (who have their position due to a minor issue Davidski's explained with low percentages of ANE showing up falsely in West Africans adding on to their true Eurasian ancestry)
The black triangles are some ancient populations, should be fairly obvious which.
Davidski, have you sent ANE K8 to whoever the GEDmatch guy is?
ReplyDeleteNo, I have not. Two different K8 tests will be available soon. I'll make an announcement about that here after the New Years.
ReplyDeleteReich's paper won't be online for a while IMO.
ReplyDeleteWell, he said in an e-mail a few weeks ago that it'll be posted at bioRxiv within a couple of months. So it should be available soon, unless something went wrong or they're expanding the study to include more samples. In any case, I'm quite certain we'll see it online in early 2015.
ReplyDeleteDavid,
ReplyDeleteMerry Christmas!
Is the second k8 using European Neolithic samples to create a Near East component, by pulling out most WHG and any ANE?
Just going by eye, the Bedouins are more than half a grid section further from the line from Loschbour through Stuttgart, than Sardinians are. I am not sure why they would be so far on the ANE side of the line if they had no ANE, plus the fact that they have SSA, which I think should pull them further South of that line, towards the side NW Africans are on. I just think that might clear up the fact that CO1 and Gok2 showed a slight ANE shift, when only Stuttgart was involved.
ReplyDeleteI'd love to see an Ancient Near East representative that didn't involve populations that likely received later African, ANE, and South Asian genes.
Your call though, man!
Another interesting alternative, would be to use Ust-Ishim, pluck out any ENA, if he has any, and throw a pinch of Loschbour into him and see where he plots. He could bring out any SSA in Near Eastern farmers if any exists. If you would discover the difference between Stuttgart and Ust-Ishim's Basal Eurasian, that would be huge!
ReplyDeleteIt would certainly be a worthy first paper from us. The models for modern pops would be more accurate than any before it, if you ask me. It could be a project to work on, while we wait for the Yamnaya, Corded, and Beaker genomes. What do you say?
ReplyDeleteI wonder if Reich and co. have decided to test the Maikop samples before releasing their paper on Yamnaya?
ReplyDeleteHere's that plot:
ReplyDelete34/40/26 for Yamnaya - red
34/47/19 for Corded - orange
38/47/15 for Beaker - yellow
https://drive.google.com/file/d/0B9o3EYTdM8lQZnVCZTZYUFE5bDg/view?usp=sharing
Beaker's too eastern IMO, but the yellow dot is probably OK for Beaker/Unetice etc., in other words a mix of various LNE/EBA samples.
The two K8 tests will be identical to the tests in this spreadsheet. The second test, like the one in sheet 2, will estimate low-bound Near Eastern ancestry in Europeans.
https://docs.google.com/spreadsheets/d/1x8pm8sVcHqceiNFJMO082kxaBF5ePr4__bAK05VQRFw/edit?usp=sharing
Also, the reason the Bedouins are shifted east is because they have some sort of Eurasian ancestry that is not described well by any of the clusters in this test, and so comes out as ~2% East Eurasian. I suspect that this is what shows up as Ust'-Ishim-related ancestry in the BedouinA and Egyptians in the f3 test that I ran recently.
https://docs.google.com/spreadsheets/d/10gaCNMpadAjueogTKEthjYxfHXiTb5ILwYNXDjXr_hQ/edit?usp=sharing
I do have a few ideas for papers. But I doubt we can do anything on Yamnaya until we actually analyze at least one Yamnaya genome.
And I don't think it'll be possible to split the Stuttgart and Ust'-Ishim Basal Eurasian types without an ancient early Neolithic genome from the Levant or Arabia.
Off now for some Christmas drinks. Merry Christmas!
Looks good, David! We could make Beaker less ANE, but it's already less ANE than Czechs, which was a requirement. That freehand had Beaker closer to Corded than this plot is, so I guess we can wait for the paper.
ReplyDeleteAs for the Ust-Ishim, I was thinking that we just put some Loschbour like stuff in it, and then use it as a component (Hypothetical NEF), without the Near East component. That should flush out what is different about Stuttgart, modern Near Easterners, and possibly what is truly SSA in Africa. Maybe, maybe not, but it's worth a shot while we sit and hypothesize about Yamnaya. At least we have these other genomes.
If we mix Ust-Ishim and Loschbour, we would have to set it as 0%SSA, ENA, ANE, and the like, in order to pull out everything else that people have, right? I'd like to play around with that, and see where it goes.
Or, we could simply use Ust-Ishim, take out Ust-Ishim stuff from Kostenki, and just call it Basal Eurasian. That could find everyone's true Basal, WHG, and ANE.
The LNE/EBA on that freehand plot isn't just Bell Beaker, it's Unetice as well.
ReplyDeleteBtw, Helgenes50, the French are marked on this plot.
https://drive.google.com/file/d/0B9o3EYTdM8lQUnNQRFJaRmtJSm8/view?usp=sharing
Okay, maybe give beaker two more NE and 2 less ANE.
ReplyDelete@ David
ReplyDeleteThanks!
Merry Christmas!
Chad,
ReplyDeleteTwo more NE and two less ANE might be too high.
https://drive.google.com/file/d/0B9o3EYTdM8lQOUE4Qk9WNk9jT2s/view?usp=sharing
Maybe 41 NE, 46 WHG and 13 ANE is a safer option? It's well within range of modern variation, which I suspect Bell Beakers were. Who knows though without seeing an actual Bell Beaker genome?
https://drive.google.com/file/d/0B9o3EYTdM8lQZ2JaT19HNHJuVjQ/view?usp=sharing
Maybe 41 NE, 46 WHG and 13 ANE is a safer option?
ReplyDeleteIf this result is correct, Bell Beaker must be very similar to a Northern French.
Between an English and a French.
Although 46% seems a bit high for a French, even for a Northern, but for the ANE and for the Near eastern , it's close.
ReplyDeleteBut if that fits to the Bell Beakers, we don't speak of the first Bell Beakers of Spain!
No way in hell the first Iberian Beakers are R1b. I bet they'll be kind of like CO1.
ReplyDelete@ Chad
ReplyDeleteThat's possible, but CO1 is close to the Neolithics
Yeah, I think around CO1 is safer than Gok2.
ReplyDeleteI'll be done tonight for tomorrow breaking down all Paleolithic-bronze age European mtDNA.
ReplyDeleteWhat do you guys think of the reference point spreadsheet I have? It's the source for all the comparisons between different ancient people's mtDNA I'll do later today.
https://docs.google.com/spreadsheets/d/1HcAhe7QvggT792VruuoZX6IsTg4LhWXV-Z_ZfTk2KGA/edit?usp=sharing
Of course I'm not done putting all the samples in there yet. I'll list my sources on the spreadsheet.
IF that reference point isn't detailed enough people can look at my personal notes.
I'll list all HV1 matchs(not all because some mean nothing, and I'll give the reasons why), and the few HV1+2 matches. I'll list all the likely founder effects(there are very few) within samples of the same sites, cultures, etc.
I'll make google maps showing every site from a certain era and whether a haplotype or haplogroup is found in it that site and at what frequency, like FTDNA does.
I'll show very detailed haplogroup frequencies, and detailed comparisons.
I might make videos showing how the frequencies of a haplotype/haplogroup is different in different regions/eras/cultures.
I think we all can agree Jean Manco's site does not give a detailed enough description of ancient mtDNA.
ReplyDeletehttp://www.ancestraljourneys.org/europeanneolithicdna.shtml
At some point someone had to make something more interactive and detailed.
If one of the ancient genomes was slightly closer to Loschbour than Gokhem on the same line, CO1 ? the Swedish would be on a line between this ancient genome and Yamnaya
ReplyDeleteI think Northern Europeans were between were mostly gok2 like. With some Motala as you head east. I think CO1 was 10% more Near Eastern than Gok2. David can clear that up.
ReplyDeleteIf it's the case, CO1 could be a best fit for the French, Gokhem has a lack of Caucasian, what is not the case for CO1 and this component is more present among the French than among the Northern European.
ReplyDeleteCaucasian mainly of LBK origin
Yes, I think just under northern French is the cutoff for Gok 2. Gok2 could cover the basques and Pais Vasco.
ReplyDeleteKrefter,
ReplyDeleteLooks good. Let us know when you're done.
David,
ReplyDeleteIf you get a chance, shoot me an e-mail. chadrohlfsen@hotmail.com
The paper that speaks of Europeans mutating 50% faster than others, is out..
ReplyDeletehttp://biorxiv.org/content/early/2014/12/24/010314
I believe it is only covering skin pigmentation changes.
ReplyDeleteThere's no debating that mutations associated with light skin, hair, and eyes have been selected for in Europe over the last 5,000 years. Wilde. 2014 was the first study to confirm this with a large sample set.
ReplyDeleteYamna, Neolithic Hungarians, and the Ajvs, can represent the populations who mixed to explain most of the modern European gene pool 6,000-4,000YBP, and they all probably had brown or olive skin.
What's surprising about this quick and widespread change, is that it occurred multiple times and in unrelated people.
Everyone's ancestors from the Irish to the Sami, went through it. So many people went through that you don't find anyone today pigmented like the Yamna or Ajv58.
The selection for the lactose persistent mutation is more surprising. People who barely admixed all suddenly were able to take in lactose after just a few hundred years before only a small minority couldn't.
What happened to the people who couldn't take in lactose, did they just not reproduce. It's very confusing how evolution could occur so quickly in humans who you would expect be conscious of such a widespread and quick change.
It's kind of creepy when looking at my lactose-associated SNPs that I have at least one derived allele in each one, because I'm not unique and one way or another am a result of natural selection.
ReplyDeleteThere's no people group in the world totally isolated and innocent. Every culture and genealogy has a history of corruption, natural selection, and foreign influence.
David,
ReplyDeleteI'm going to try and breakdown what it is that I am trying to see with Ust-Ishim.
I am not trying to separate Basal Eurasian. That is not possible. What I am thinking is that by creating a synthetic Ancient Near Easterner, we can isolate what it is about Stuttgart that is different. My feeling is that Basal Eurasian and WHG is not going to create Stuttgart. Actual SSA is going to be needed, on top of the Basal and WHG.
This can be approach by three ways.
1) Just use Ust-Ishim. Then make him a component, simply Basal Eurasian. We can then see what everyone's true Basal Eurasian percentage is, plus what else is there.
2) Pull Basal out of Kostenki, and add it to Ust-Ishim to get more coverage.
3) Mix Ust-Ishim and Loschbour, to an extent, to create a synthetic Near Eastern Farmer. Isolating true Basal Eurasian, with true WHG, should tell us what everyone, including European farmers were.
There is something about Ust-Ishim that is very interesting. He seems to be equidistant to everyone, until we run into modern Europeans. If Stuttgart were simply just Basal Eurasian and WHG, should Stuttgart and Europeans not be closer to Ust-Ishim, instead of farther away than East Eurasians. I think that this is a sign of SSA admixture. That is the only thing that is not equidistant when comparing Ust-Ishim to East Eurasians and Ancient Europeans.
The second interesting thing is the supposed connection to Papuans. My feeling is that this isn't actual admixture from Papuans, but the longer strands of Neandertal DNA causing him to seem more admixed with archaics, like the Papuans are. I think that using all of the Neanderthal and Denisovan snp's from Felix's work, we can isolate some of those markers in Ust-Ishim. If we remove a few markers here and there, we then break up those longer stretches of markers. If we remove say 20-30 markers, or however many you wish, we can then run it again, on D stats, to see if that moved Papuans away from him.
If that does not work, we can then assume that admixture may be real. We can then isolate those Papuan markers, and remove a few at a time, until he is no closer to Papuans than anyone else. That will give us true Basal Eurasian, and a place to start from , with the creation of components and running it against ancient and modern samples.
Does that clear it up at all?
Thanks!!
@ryu
ReplyDelete"If a motala-ancient Karelian continuum existed across northern Europe ... increased EHG affinity in northern Europe, or for that matter in Corded Ware itself, independently of Yamnaya."
Say there was a continuum with 10% EHG at the west end and 40% at the east end and the eastern population moved west then you might think the average would end up at 25%.
Alternatively if the 40% population A moved into the territory of the 20% population B and and the two amalgamated
(for example if B was two tribes fighting and one B tribe sided with the A incomers against their traditional enemies)
amalgamated in the proportion 70% population A and 30% population B then and then moved further west together the average EHG would be
70% pop A x 40% EHG = 28%
plus
30% pop B x 20% EHG = 6%
or 34% EHG
So the final proportions of mixing will depend on the form of the mixing event.
I think the form will often be deduceable through the imbalances of ydna, adna and mdna.
If the original ydna survives in quantity then a migration would more likely have been an amalgamation event or elite replacement event.
If the original ydna doesn't survive in quantity then it was more likely to have been a far squishier event.
Hence why I think something different happened in the north of the Carpathians northern channel compared with the Danube->France central channel.
Chad,
ReplyDeleteI already made synthetic early Near Eastern farmers. They cluster with Yemenite Jews. Look for ENF here.
https://drive.google.com/file/d/0B9o3EYTdM8lQc3k1dGdoR01aNk0/view?usp=sharing
This is probably very close to where early Neolithic genomes from the Near East will cluster, although perhaps the oldest samples will be positioned further southwest, and basically in line with the Sardinians across the horizontal axis.
As for the experiment ideas, they're generally too complicated and well beyond what I'm capable of. Also...
1) I can't make Ust'-Ishim a component because he doesn't represent a single component in modern populations. I can only get components out of samples if they represent very robust and well defined components.
2) I can't pull Basal Eurasian out of Kostenki14 because I don't have a Basal Eurasian reference, so I wouldn't know what to look for. And to be honest I doubt very much that Kostenki14 even carries the so called Basal Eurasian that Stuttgart has.
3) Mixing Ust'-Ishim with Loschbour won't create a synthetic Near Eastern Farmer. Have you seen where Ust'-Ishim and Loschbour cluster on global plots? Mixing the two will produce some sort of Central Asian sample that probably never existed as a real population.
@ryu
ReplyDelete"I'm just saying I find it impossible to believe that proto-slavs and germanics in the Central European forests, which had strong farming cultures going, preserve the genetic legacy of Yamnaya steppe herdsmen better than the far more similar Indo-Iranian herdsmen on the Central Asian steppe. While there was farming in Central Asia it was very limited."
I think this can be explained by relative population density effects.
Say neo/copper pop. density is on a ten point scale such that
HG ~ 1/10
Farmer ~ 10/10
Pastoralist ~ 4/10
If nomadic pastoralist tribe A displaces nomadic pastoralist tribe B and they have roughly the same population density then the invading population can't incorporate large numbers of the displaced population so the final population might be 100% A males and 80-90% A females.
If nomadic pastoralist tribe A (pop. density 4/10) displaces only the top half of farmer tribe B (pop. density 10/10) then the A population may only be a minority of the total population going forward. The males may still end up becoming the majority of ydna due to elite effects but the females don't.
So maybe
step 1) western steppe nomads ~ 100% Yamnaya
step 2a) some western steppe nomads remain where they are ~ 100% Yamnaya
step 2b) some western steppe nomads mix with northern HGs ~ 70% Yamnaya
step 3) western steppe nomads displaced by eastern steppe nomads ~ 10% Yamnaya
So if this is correct the western steppe nomads *were* more Yamnaya originally but later steppe expansions changed it because a nomad on nomad displacement is less able to incorporate the original population than a nomad on farmer displacement as one is
low pop. density -> low pop. density
and the other is
low pop. density -> higher pop. density
@Krefter
ReplyDelete"The selection for the lactose persistent mutation is more surprising. People who barely admixed all suddenly were able to take in lactose after just a few hundred years before only a small minority couldn't."
I think this is another population density effect.
If you say for the sake of argument the population densities on a ten point scale were
HG ~ 1/10
Farmer ~ 10/10
Pastoralist ~ 4/10
and further assume that the farmer pop. density is a maximum value that was only reached in the optimal areas for neolithic farming and those optimal areas were in the south along the med. coast then what you'd get is a cline of farmer pop. density from south to north and south to northwest.
This would create a line of equilibrium at some point between farmers and HGs where their pop. density balanced.
This line of equilibrium is clearly shown (imo) by the range of LBK (and equivalent farmer groups).
So if the Atlantic region west of LBK was still HG (pop. density 1/10) (apart from coastal Atlantic megalith fishing settlements) and some population either arrived or developed a pastoralist alternative (pop. density 4/10) then a potentially very small group could expand dramatically into that under-exploited niche between the Atlantic coast and the edge of LBK.
(In this model they would also likely over-run the western edge of the farmer's also as they'd have a higher pop. density than the most marginal farmers along the edge of the LBK range.)
i.e. I think the LP thing represents a population explosion specifically along the Atlantic coast followed by a slow expansion further inland over time.
@Matt
ReplyDelete"the African Americans samples average around 18% of Near Eastern+ANE+WHG and average 5% ANE, 4% Near Eastern and 9% WHG, the rest being around 80% Pygmy plus Sub Saharan.
That would seem to imply that African Americans West Eurasian ancestors would have 27% ANE, 23% Near Eastern and 48% WHG."
It's a shame it is too un PC to happen but African Americans would otherwise be a perfect test case for selection in place.
In particular if neolithic farmer dna had advantages that went into and were carried forward by Euro dna then those same advantages would have gone into some African-Americans via Euros.
#
@Krefter
"I'm ready to bet money now that R1b-L11 was spread with IE languages and by BB or Hinxton(Iron age) type people."
Or both.
I think there may have a hare and tortoise event with maritime BB as the hare and also being changed through attaching themselves to the pre-existing Atlantic megalith culture.
My mental picture is small groups of mostly male maritime BB copper miners landing in places along the Atlantic coast like Ross Island in Ireland, trading for local wives from the local HGs and then experiencing a dramatic population explosion due to cattle and LP.
At the the same time the tortoise part of the same original population as the BB is taking the much slower overland route west to Hallstatt.
.
If it's correct that the form of a mixture event can often be deduced by divergent end results in ydna, mdna and adna then your mtdna thing could be invaluable.
#
@ Chad
"I think one problem with all the figures thus far is that a line drawn from Armenian to Yamnaya would imply the EHG population was in the vicinity of the SHGs, when the drawn pca figure from the talk implies this isn't the case. The EHGs there are far to the east of the SHGs."
Could that result if both EHG and SHG had the same minor but highly divergent component?
If you draw a line from the Armenians to Yamnaya, and then another one of the same length from Yamnaya to wherever at the same angle, the result does show a location similar to where EHG should be.
ReplyDeletehttps://drive.google.com/file/d/0B9o3EYTdM8lQcWxNV1BLX2hCNHc/view?usp=sharing
Keep in mind also that this plot doesn't show the average positions of the WHG and SHG clusters, just Loschbour and Motala12, both of which are extreme examples of WHG and SHG, respectively. If I also had La Brana-1, Ajvide58 and other samples, things would look quite different.
By the way, here's where pure ANE, ENF and WHG samples cluster on the K8 plot.
https://drive.google.com/file/d/0B9o3EYTdM8lQNTVNQmFCQVR2T28/view?usp=sharing
@David
ReplyDeleteDon't you think that CO1 whould be a best fit for the French than Gok2.
This is at least what shows your PCA if CO1 is placed lower than Gok2 on the same line.
Gok2 seems to be a best fit for the Northern French, closer to the English.
That makes sense, since, as Normand in the K13 and K15 oracles, I am between the Basques and the West Norwegians.
I'm not sure. It's impossible to say who was in France during the Copper Age, just before the shit hit the fan.
ReplyDeleteMaybe they were like Oetzi, or a mix between Oetzi and Gokhem2?
CO1 is technically an East-Central European from the Copper Age, that's why I'm thinking Oezti is a better fit for France, but who knows?
Rightly ! CO1 is just between Oetzi and Gokhem.
ReplyDeletethere's no doubt that we don't know the solution, not now !
But always from this PCA, Oetzi seems to be a best fit for the SE Europeans
Davidski If you draw line from Armenians to Yamnaya, and then another one of the same length from Yamnaya to wherever at the same angle, the result does show a location similar to where EHG should be.
ReplyDeleteAlthough the literal end point of the line you posted there cannot exist within the three variables used, as it is outside the triangle of ANE,WHG and Near Eastern. You could not combine the three ANE-WHG-Near Eastern in such a way as to actually sit at the end point of that line.
(all possible combinations of ANE-WHG-Near Eastern that sum to 100% would be within this triangle - http://i.imgur.com/c9tFEUj.png because all the PCA analysis is doing is creating a triangle on which to plot the scores within the three proportions, with some contribution from minor components).
The closest you could get would be the point that intersects the Loschbour-Motala vector at 66% WHG, 34% WHG - http://i.imgur.com/YauAEnb.png
That would have to contribute around 60% or over of the Yamnaya point with Armenian to work. That's not a problem if we're treating the ASHG ancestry proportion as an approximate I guess, if that's a good idea to do.