PDA

View Full Version : Holocene Ancestors of West Eurasians



Fire Haired14
14-08-16, 00:16
At the start of the Holocene(11,500 years ago) or shortly afterwards there were at least four distinct races in West Eurasia. The origin of these races and their relationship to each other is not completely known. Older genomes are needed to learn this. Nonetheless all Modern and Metal West Eurasians are a mixture of two or more of these races and many also have ancestry from Africa or Northern Asia.

Western/Southern Europe: Known as "WHG"
Russia: Known as "EHG"
Caucasus and Iran: Known as "CHG" and "Iran_Neolithic"
Anatolia and Levant: Known as "EEF" and "Natufian".

For the two West Asian races there was at least some diversity within them. CHG and Iran_Neolithic weren't exactly the same, EEF and Natufian weren't exactly the same. A West Eurasian population having ancestry from one instead of the other, for example CHG instead of Iran_Neolithic, is significant but it is often hard to decipher.

Using D-stats from Eurogenes you can produce accurate ancestry percentages for Modern and Metal Age West Eurasians from each of these four ancestors. Below is a spreadsheet with results.

Holocene Ancestors Of West Eurasia (https://docs.google.com/spreadsheets/d/1IjnDNbzT4kJrSp0TbQm_12uVXD7YUk54FOpO_QyOHto/edit#gid=0)

Iran_Neolithic vs CHG and Anatolia_Neolithic vs Levant_Neolithic

Caucasus and Europe favor CHG over Iran_Neolithic.
Mesopotamia/Iran and SouthWest Asia favor Iran_Neolithic over CHG.
Everyone favors Anatolia_Neolithic except for SouthWest Asia and North Africa who favor Levant_Neolithic.

Concluding remarks.
SouthWest Asia is an about 40% Iran_Neolithic+60% Anatolia/Levant_Neolithic mix.
Egypt has significantly less Iran_Neolithic than SouthWest Asia and Morocco has even less.
Mesoptamia/Iran is an about 50% Iran_Neolithic+50% Anatolia_Neolithic mix, with maybe some minor EHG.
Cypriot is the same as Mesopotamia/Iran but has slightly more Anatolia_Neolithic.
The Caucasus's amounts of Anatolia_Neolithic and CHG/Iran_Neo are about equal, and then the rest is EHG which varies among ethnic groups.

CHG/Iran_Neo is over 20% in all of the Middle East but is at only 10-15% in almost all Europeans.
CHG/Iran_Neo is slightly higher in Southern Europe, especially in Sicily at 26%.
WHG has a presence in all of Europe but is absent in all of the Middle East. Peaks in Iberia, Sardinia, and Northern Europe at 10-15%.
EHG has a strong presence in all of Europe but has a weak presence in the Middle East outside of the Caucasus.
Anatolia_Neolithic is the most important ancestor for all Europeans, except Saami. It reaches over 50% in all of Southern Europe.

Saami have Mesolithic Scandinavian ancestry
Saami score the lowest Anatolia_Neolithic for Europe at 22% and highest EHG at 34.85% and second highest WHG at 13.85%, despite scoring 24% Siberian. The Saami's West Eurasian ancestors had much more EHG and WHG than any modern Europeans. The fit for Saami was bad but through other analysis I've been able to confirm they have extra Holocene European ancestry other Europeans don't have and that they have significantly less Anatolia_Neolithic ancestry than other Europeans. Evidence suggests they're normal Northern Europeans besides about 15% SHG(Mesolithic Scandinavian) ancestry and 25% Siberian ancestry. That 15% SHG is raising their score in both EHG and WHG.

Fire Haired14
14-08-16, 04:29
Notice Greece and Turkey are both mostly Anatolia_Neolithic but Turkey's other 40-45% is mostly CHG and Iran_Neo while Greece's other 40-45% is hardly Iran_Neo at all and more EHG than CHG.

Steppe/EHG admixture is mostly what sets Greece apart from Turkey. Pre-IE Greece was probably similar to modern Turkey and Cyprus. It's confusing why Steppe admixture made such a bigger impact on Greece than in Turkey, when both speak IE languages(Turkish tounge arrived in Middle ages).

MarkoZ
14-08-16, 06:13
Notice Greece and Turkey are both mostly Anatolia_Neolithic but Turkey's other 40-45% is mostly CHG and Iran_Neo while Greece's other 40-45% is hardly Iran_Neo at all and more EHG than CHG.

Steppe/EHG admixture is mostly what sets Greece apart from Turkey. Pre-IE Greece was probably similar to modern Turkey and Cyprus. It's confusing why Steppe admixture made such a bigger impact on Greece than in Turkey, when both speak IE languages(Turkish tounge arrived in Middle ages).

So you're pinning the whole Anatolian branch on the measly 5% EHG-like ancestry in Turkey? Makes sense.

Fire Haired14
14-08-16, 07:01
Notice Greece and Turkey are both mostly Anatolia_Neolithic but Turkey's other 40-45% is mostly CHG and Iran_Neo while Greece's other 40-45% is hardly Iran_Neo at all and more EHG than CHG.

Steppe/EHG admixture is mostly what sets Greece apart from Turkey. Pre-IE Greece was probably similar to modern Turkey and Cyprus. It's confusing why Steppe admixture made such a bigger impact on Greece than in Turkey, when both speak IE languages(Turkish tounge arrived in Middle ages).

SouthEast Europeans can fit as Near East+North Europe. The non-Cupriot score for Bulgarian and Greek below are similar to the Yamnaya/EEF/WHG ratios Northern Europe has. These D-stat results are consistant with West Eurasian PCAs, the Balkans always cluster between Northern Europe and the Near East.

Bulgarian
"Cypriot" 40.95
"Yamnaya_Samara" 30.7
"Anatolia_Neolithic" 21.45
"Loschbour" 5.5
"Satsurblia" 0.8
"LaBrana1" 0.6
"Iran_Neolithic" 0
"Levant_Neolithic" 0

Greek
"Cypriot" 45.05
"Yamnaya_Samara" 25.95
"Anatolia_Neolithic" 25.15
"Loschbour" 3.85
"Iran_Neolithic" 0
"Satsurblia" 0
"Levant_Neolithic" 0
"LaBrana1" 0

Fire Haired14
14-08-16, 07:12
So you're pinning the whole Anatolian branch on the measly 5% EHG-like ancestry in Turkey? Makes sense.

I should have strawed away from making claims about genetics. 5% is enough btw.

Twilight
14-08-16, 08:00
I should have strawed away from making claims about genetics. 5% is enough btw.

Sure thing, how do I get onto D-Stats on Eurogene? :)

MarkoZ
14-08-16, 09:34
I should have strawed away from making claims about genetics. 5% is enough btw.

I think your amateurish attempts are a huge disservice to this field (which is already plagued by dilettantes), so you should probably 'straw away' from genetics altogether.

Fire Haired14
14-08-16, 11:14
I think your amateurish attempts are a huge disservice to this field (which is already plagued by dilettantes), so you should probably 'straw away' from genetics altogether.

Congrats your English is good enough for you to be an @#%&. I meant to write "I should have strawed away from making claims about •lingusitics•'. We don't know as a fact that IE languages spread from Eastern Europe with loafs of EHG and therefore shouldn't be too surprised Turkish have minor EHG.

I don't claim anything here is absolute fact. I claim it gives fairly accurate results and trust me it does.

Angela
14-08-16, 16:15
The general parameters aren't bad, and indeed have been known to us for a while, but there are obvious problems with this analysis, including...

The Sardinian breakdown-they have more WHG than that by every prior analysis I've ever seen and there's no way given the population history of Sardinia that the more "eastern" percentages are so low. Also, where is the "Moroccan" we know they have?

The lack of SSA in some southern European populations

The Moroccan numbers for Spain versus what I'm assuming is southern Italy are at the least flipped

More Iran Neolithic for Sicily than for Turkey? Really?

I could go on, but there's no point. If you want to work with some data, I would suggest Kurd's.

Tomenable
14-08-16, 18:12
So acccording to your analysis there was 6.5% of EHG admixture in Iberian Middle Neolithic.

Fire Haired14
14-08-16, 19:59
The results can't be taken too literally. You need to give wiggle room. Most importantly Levant_N is basically a more African version off Anatolia_N and Iran_N is basically a more Basal form of CHG. Don't treat them as separate entities, treat them as more or less the same.

There may be a few results that don't make perfect sense. For example %6.5 EHG in Ibera_MN. Tomenable, that is definitely reflective of WHG, also it looks like Hungary_EN and LBK_EN have a little bit of EHG. Hungary_HG had some so it is believable.

Angela, the high Iran_N for Sicily is legitimate. Ashkenazi Jews score high in Iran_N as well because they have a similar makeup as Sicilians(we know this from other analysis). Sicilans have a lot more West Asian(mostly Anatolia_N+Iran_N) ancestry than any other European hence their huge Iran_N score. Notice Sicialans CHG+Iran_N score is about %10 higher than other South Europeans, which isnt a massive difference but is significant, the high Iran_N can make difference appear larger than it is. Turkish CHG+Iran_N score is higher than Sicily's.

I forgot to include Moroccan for Sardinia but their score won't change much. About Sardinian WHG, the score is definitely legitimate. Everyone in Europe scores around %10 WHG or higher except Italy and Greece. Most WHG in all of Europe was given by middle Neolithic farmers. On a Middle Neolithic European scale Sardians have low WHG, almost half as much as most in Europe had back then. Sardians are %75+ Neo European and so them having %11 WHG is no suprise, the surprise if anything is that they only have %11.

Fire Haired14
14-08-16, 20:46
BTW, I do tend to put too much faith in D-stat admixture. I've been wrong before. I was wrong about West Asia, becaise I didn't think there could be as much diversity as Iran_N and Natufain revealed. I depended too much on the data that was available which prevented me from thinking outside the box. Most of what I said was correct however.

The reason I posted this thread is because we now have a good idea who the early Holocene ancestors of West Eurasains are and the proportions of ancestry from them. Nothing about this will change very much in the future.

Hauteville
14-08-16, 20:55
Angela, the high Iran_N for Sicily is legitimate. Ashkenazi Jews score high in Iran_N as well because they have a similar makeup as Sicilians(we know this from other analysis). Sicilans have a lot more West Asian(mostly Anatolia_N+Iran_N) ancestry than any other European hence their huge Iran_N score. Notice Sicialans CHG+Iran_N score is about %10 higher than other South Europeans, which isnt a massive difference but is significant, the high Iran_N can make difference appear larger than it is. Turkish CHG+Iran_N score is higher than Sicily's.
If you use Islander Greeks, Maltese or Calabrese you can see similar percentage of Sicilians. So this state is not correct. ;)


More Iran Neolithic for Sicily than for Turkey? Really?
Doesn't make so sense...

Angela
15-08-16, 02:54
Fire Haired: Most importantly Levant_N is basically a more African version off Anatolia_N and Iran_N is basically a more Basal form of CHG. Don't treat them as separate entities, treat them as more or less the same

If this analysis can't really tell the difference between them, then stop pretending that you've separated them properly and giving them separate percentages. Nor do I have any confidence that you have proved Levant Neolithic has more SSA, going by the fact that you can't find SSA where we know it exists in southern Europe, including in Sardinia. Nor do those "Moroccan" numbers inspire confidence. We know there isn't more "Moroccan" in Sicily than in Spain just going by IBD analysis and uniparental markers and on and on.


Angela, the high Iran_N for Sicily is legitimate. Ashkenazi Jews score high in Iran_N as well because they have a similar makeup as Sicilians(we know this from other analysis). Sicilans have a lot more West Asian(mostly Anatolia_N+Iran_N) ancestry than any other European hence their huge Iran_N score. Notice Sicialans CHG+Iran_N score is about %10 higher than other South Europeans, which isnt a massive difference but is significant, the high Iran_N can make difference appear larger than it is. Turkish CHG+Iran_N score is higher than Sicily's.

The fact that you give similar scores to both Ashkenazim and Sicilians just means you got both of them wrong. This loose talk about "West Asian" ancestry also doesn't inspire confidence in your conclusions. West Asian in terms of ancient genomes has a specific meaning and it doesn't include Anatolia Neolithic. The latter is separate from Iran Neolithic, remember? Anatolia Neolithic is connected to the first migrations and corresponds more with what was called "Mediterranean" or "Southern" in certain analyses. Iran Neolithic is separate. Look at the scores you gave Spain for Anatolia Neolithic vs Iranian Neolithic. "West Asian" is a component that in more up to date terms includes Iran Neolithic, probably some ASE, some CHG. If you go check other previous analyses you will see that the "West Asian" score is basically the same for Sicily/southern Italy and mainland Greece. In some analyses the "West Asian" is actually modal not only in Iran, or the Caucasus but in Turkey.

Most of the Iran Neolithic which came to Italy would most probably have come through Anatolia either directly or by way of mainland Greece, the Greek islands like Crete, etc. There is no way it is higher in Sicily than in Turkey.

If your analysis of Sardinia, southern Italy vs. mainland Greece etc. conflict with all the other analyses, then there's something wrong with yours or with the dstats created by someone else. How many times have we had this kind of conversation? One would think it would teach you some humility.

Go back to the drawing board, use Kurd's data, or, better yet, wait for the ancient dna from southern and southeastern Europe and for the Reich Lab paper on Caucasus and Central Asia ancient dna. I honestly don't see the point in this constant attempt to pin down these relationships in the absence of ancient dna data that we know is coming.

davef
15-08-16, 03:52
Hi FireHaired!

Seems to make sense in most ways, but...
Why do Moroccans get the highest amt of levant neo whereas Cyprus scores less than 1 percent of this component? Is it due to it being partially sub Saharan not unlike actual Moroccans?

The Anatolian Neo score for Ashkenazi is relatively low, Sicily gets 54 vs 45 for jews and England gets 46 (jews always get higher Anatolian neolithic than every Northern European and about as much as south italians, according to previous analyses) . However, we do see ashkenazi scoring 9 percent levant neo vs 0 percent in Sicily, could it be the levant neo found in ashkenazi has a lot of Anatolian neo inside of it since levant neo is very close to Anatolian?

Fire Haired14
15-08-16, 06:23
If this analysis can't really tell the difference between them, then stop pretending that you've separated them properly and giving them separate percentages.

I never claimed this, I claimed the opposite. I also made it clear that the members in the two sets of West Asian ancestors are similar but differnt. The extra Basalness/Africaness of Levant_N and Iran_N is significant. The 60% Levant_N/Anatolia_N and 40% Iran_N/CHG score in Levant is differnt from the 40% Iran_N/CHG+40% Levant_N/Anatolia_N+10% EHG score in the Caucasus, because the Levant one is more Basal.

So it's important to include all 6 ancestors to express Basalism/Africanism(whatever it is, it's pulling them away from East Asians). Iran_N and Natufian/Levant_N(Levant_N is Anatolia_N+Natufian in these stats) are gold mines. Modelling Middle Easterners as Anatolia_N+CHG never worked outside the Caucasus. They weren't basal enough. With Iran_N and Natufian/Levant_N everything is making sense. So I need them in this analysis or else I'll be getting very bad fits for the Middle East.


Nor do I have any confidence that you have proved Levant Neolithic has more SSA, going by the fact that you can't find SSA where we know it exists in southern Europe, including in Sardinia. Nor do those "Moroccan" numbers inspire confidence. We know there isn't more "Moroccan" in Sicily than in Spain just going by IBD analysis and uniparental markers and on and on.

When exotic admixture is tiny it's difficult for D-stats to detect it. It's not a big deal. The West Eurasian percentages for Sardinia, Spain, and Sicily are accurate.



The fact that you give similar scores to both Ashkenazim and Sicilians just means you got both of them wrong.

They're both correct. Read the post I made before this one explaining why the high Iran_N in Sicily isn't as strange as it seems at first. If you combine their Iran_N/CHG score it isn't radically differnt from Southern Europe. The trends in this test are the exact we see for West Eurasia in all other tests.


This loose talk about "West Asian" ancestry also doesn't inspire confidence in your conclusions. West Asian in terms of ancient genomes has a specific meaning and it doesn't include Anatolia Neolithic. The latter is separate from Iran Neolithic, remember?

I don't know what else to call it. I was using the term West Asian to refer to what much of West Asia has been since 6,000 or more years ago; a CHG/Iran_N and Anatolia_N/Levant_N mix. The West Asian you're referring to arrived in Sicily with as much or more Anatolia_N/Levant_N.


If you go check other previous analyses you will see that the "West Asian" score is basically the same for Sicily/southern Italy and mainland Greece. In some analyses the "West Asian" is actually modal not only in Iran, or the Caucasus but in Turkey.

The Greece I used I'm pretty sure is the most non-West Asian of Greeks. In ADMIXTURE Sicily does score more East_MN/SW Asian stuff which definitely has some CHG/Iran_N in it because it was based on modern DNA. Plus the "West Asian" you're referring to is more CHG than Iran_N. Georgians don't have much more combined Iran_N/CHG than Assyrians, but they have loads more CHG affinity, and so West Asian components are much higher in them. This is why Turkish score so high in West Asian components as well.


If your analysis of Sardinia, southern Italy vs. mainland Greece etc. conflict with all the other analyses, then there's something wrong with yours or with the dstats created by someone else. How many times have we had this kind of conversation? One would think it would teach you some humility.

They don't conflict with other analyses if you look at the results carefully.


Go back to the drawing board, use Kurd's data, or, better yet, wait for the ancient dna from southern and southeastern Europe and for the Reich Lab paper on Caucasus and Central Asia ancient dna. I honestly don't see the point in this constant attempt to pin down these relationships in the absence of ancient dna data that we know is coming.

The primary reason I made this thread is because the ancient DNA we have for the first time can get good fits for all of West Eurasia. There's no more waiting. We have the Mesolithic/Early Neolithic DNA needed to explain everyone in West Eurasia. One way or another the new Bronze age DNA is going to show how moderns became the Mesolithic/Early Neolithic mix they are, it won't change the mix they are. I believe in the possibility new ancient DNA will change the results in this thread. I should take the results with less literalism than I do here, but in all likelihood the basic trends won't change much.

Fire Haired14
15-08-16, 14:32
Most of the Iran Neolithic which came to Italy would most probably have come through Anatolia either directly or by way of mainland Greece, the Greek islands like Crete, etc. There is no way it is higher in Sicily than in Turkey.

Keep in mind some of Sicily's CHG/Iran_N(probably 5-10%) is from the Steppe and falls mostly under North European components in ADMIXTURE. Besides that if you combine CHG and Iran_N, Turkey gets 33.75% and Sicily gets 26.5%. 23% of Sicily's is CHG and 57.5% of Turkey's is which will make the difference in West Asian component scores between Turkey and Sicily even greater. This makes differnt scores between Turkey and Sicily in ADMIXTURE understandable.

In ADMIXTURE Sicily scores slightly higher in East Med/SW Asian components than Turkey. If Turkey-like people gave Sicily their West Asian ancestry this shouldn't be the case. We don't know where it came from, it could be a mix of Levant and Turkey. People in Turkey and Levant could have been differnt from any today.

Hauteville
15-08-16, 18:16
In ADMIXTURE Sicily scores slightly higher in East Med/SW Asian components than Turkey.
Incorrect, see the spreadsheet. Turks (but Central and East Turks would score more) scores 28.8 East Med and 5.62 Red Sea (but they have South Asian and various far Asian admix); West Sicilians scores 26.74 East Med and 5.37 Red Sea, while Eastern coast of Sicily scores 26.94 of East Med and 6.73 of Red Sea.

https://docs.google.com/spreadsheets/d/19c_bZjUV_RouKyGyLHmMDw57WwAVabXFJOaso_gcuRE/edit#gid=1872836177