Role of recent admixture in last 1500 years

[Angela]

What an edifying conversation.

 

[Sigfrido]

More like closer to pure speculation, as on top of that one would also have to be assuming that Mozabites had the same levels of SSA admixture 2200+ years ago as they do today. By the same token we can also go around speculating how much SSA admixture Italians should really have by assuming the same levels of SSA admixture in modern Middle Easterners and North Africans as those of past times.

In the same paper, tested Italians had 0% Mozabite and Yoruba. We are talking about the ADMIXTURE analysis here.

The same results are also confirmed by the ADMIXTURE analysis found in the Supplementary Info from Laura R. Botigué et al. and Lopez et al. with Spaniards being the only ones with significan african admixture, while Italians have zero.

 

[Drac II]

In the same paper, tested Italians had 0% Mozabite and Yoruba. We are talking about the ADMIXTURE analysis here.

The same results are also confirmed by the ADMIXTURE analysis found in the Supplementary Info from Laura R. Botigué et al. and Lopez et al. with Spaniards being the only ones with significan african admixture, while Italians have zero.

Lazaridis et al. did find some African DNA in Italy too. They in fact imply it is found throughout Europe but at smaller amounts. Regarding the larger amount found in Spain they in fact considered the larger sample size as a possible explanation.

We already saw the ADMIXTURE analysis of Botigue et al., "Joey" (yes, it is obvious that it is you again, back to your old games of trying to stir up trouble), and it is not difficult to see that African (both North and sub-Saharan) DNA was also found in the Italian samples (even the French had some), in fact at the higher Ks all the sub-Saharan DNA in Spain (except Canary Islands and Basque region) was "gone" while Italy (except Tuscany) still showed a bit. Plus then there's the quite larger Middle Eastern one. You also conveniently forget Bauchet et al., Li et al., López Herráez et al., Atzmon et al., Brisighelli et al., Shriner et al., etc. Things are not as clear-cut as you would like them to be.

 

[LeBrok]

What an edifying conversation.

And the total devotion and interest they pay to every percentile of SSA, and only SSA.

 

[Tomenable]

And the total devotion and interest they pay to every percentile of SSA, and only SSA.

I'm more bothered by this "Iran+Armenia" and "Turkey" admixtures. It looks like Slavic migrations were spreading a lot of "Iran+Armenia" admixture, which kind of makes no sense unless we don't know something about genetics of Proto-Slavs:

http://www.cell.com/current-biology/pdfExtended/S0960-9822(15)00949-5

Also authors of this study used Lithuanians as a proxy for Proto-Slavs ("a Slavic source represented here by a cluster of Lithuanians"). It confirms data from the previous paper that Balto-Slavic populations are similar genetically, but shouldn't Ukrainians be used instead?

And if you look at Figure 2. then there is also some kind of "SCEurope" admixture in North-Eastern Europe in times when the German "Ostsiedlung" was supposedly talking place. But "South-Central Europe" is actually associated with Italians in this study - so how on Earth did German-speaking immigrants from the HRE manage to spread Italian-like admixture instead of "NWEurope" admixture ???

 

[Angela]

I'm more bothered by this "Iran+Armenia" and "Turkey" admixtures. It looks like Slavic migrations were spreading a lot of "Iran+Armenia" admixture, which kind of makes no sense unless we don't know something about genetics of Proto-Slavs:

http://www.cell.com/current-biology/pdfExtended/S0960-9822(15)00949-5

Also authors of this study used Lithuanians as a proxy for Proto-Slavs ("a Slavic source represented here by a cluster of Lithuanians"). It confirms data from the previous paper that Balto-Slavic populations are similar genetically, but shouldn't Ukrainians be used instead?

And if you look at Figure 2. then there is also some kind of "SCEurope" admixture in North-Eastern Europe in times when the German "Ostsiedlung" was supposedly talking place. But "South-Central Europe" is actually associated with Italians in this study - so how on Earth did German-speaking immigrants from the HRE manage to spread Italian-like admixture instead of "NWEurope" admixture ???

My general feeling about this paper is that the authors may be picking up signals of ancient admixture, or, long term continuing admixture, and attributing all of it to the most recent and not necessarily strongest migrations.

In terms of signals of admixture from "Iran" or "Turkey" in Balto-Slavs I doubt that what they're picking up is any migration of Persians or eastern Anatolians directly into northeastern and eastern Europe. However, if Balto-Slavs are even partly descended from Yamnaya like people (described as half modern "Armenian" like), then I don't see why it should be so surprising that there is some trail back to populations south and southeast of the Caucasus. A trail to south central Europe might actually point to late Neolithic mostly EEF people, and an admixture that took place earlier than their algorithm would indicate. Likewise, in terms of Italians I have no doubt they're seeing a connection to "Cypriot" like populations. The question for me, is, are their dates right, and are their findings correct as to direction of gene flow.

It would be great if someone with the necessary mathematical and software skills, and the time, could go into the programs and the methodology used by both Busby et al and Ralph and Coop et al and see which method(s) seems most robust given the very different conclusions reached for certain populations by the two teams.

 

[Tomenable]

In terms of signals of admixture from "Iran" or "Turkey" in Balto-Slavs I doubt that what they're picking up is any migration of Persians or eastern Anatolians directly into northeastern and eastern Europe. However, if Balto-Slavs are even partly descended from Yamnaya like people (described as half modern "Armenian" like), then I don't see why it should be so surprising that there is some trail back to populations south and southeast of the Caucasus. A trail to south central Europe might actually point to late Neolithic mostly EEF people, and an admixture that took place earlier than their algorithm would indicate. Likewise, in terms of Italians I have no doubt they're seeing a connection to "Cypriot" like populations.

But we aren't talking about prehistoric gene flow such as from Yamna or ENF. We are talking about admixtures during the Middle Ages.

Check Figure 2 B - e.g. "Iran+Armenia" admixture in North-Eastern Europe and in South-Eastern Europe was in period 500 - 1500 AD:

Their Figures (charts / graphs) contradict what they write in actual text, because they don't even explain this "Iran+Armenia" thing:

http://s16.postimg.org/n21kalef9/Figure_2_B.png

 

[Tomenable]

If their dates are correct "Iran+Armenia" admixture occured in both South-Eastern and North-Eastern Europe during the Slavic expansion.

And "SCEurope" admixture occured in North-Eastern Europe during the German High-Late Medieval expansion (so-called "Ostsiedlung").

So something is strange here, why were Slavs spreading "Iran+Armenia" and Germans "South-Central Europe" admixtures ???

When you take a look at their Supplementary materials, you can see that South-Central Europeans = Italians, not Germans.

============================

On the other hand, this constant stream of "NWEurope" (dark blue) admixture in North-Eastern Europe, which - as can be seen - starts already some time Before Christ, and which continues until roughly 500 AD, is consistent with expansions of Proto-Germanic-speakers from Scandinavia:

http://s28.postimg.org/b5t5czg25/Expansions.png

"NWEurope" admixture makes sense for Celts and East Germanic tribes.

But do "Iran+Armenia" and "Turkey" make sense for Slavic expansion?

And does "SCEurope" make sense for German Medieval "Ostsiedlung"?

 

[epoch]

In terms of signals of admixture from "Iran" or "Turkey" in Balto-Slavs I doubt that what they're picking up is any migration of Persians or eastern Anatolians directly into northeastern and eastern Europe.

Those could be proxy for a number of Steppe tribes that came in the slipstream of Mongolian invasions.

And if you look at Figure 2. then there is also some kind of "SCEurope" admixture in North-Eastern Europe in times when the German "Ostsiedlung" was supposedly talking place. But "South-Central Europe" is actually associated with Italians in this study - so how on Earth did German-speaking immigrants from the HRE manage to spread Italian-like admixture instead of "NWEurope" admixture ???

Perhaps the Lombardian admixture in Northern Italians is large enough to make them look like Schwaben? Just speculating here, mind you.

 

[Tomenable]

Mongolian invasions were later, after year 1200 AD.

Also if you read the text, then authors write that they found a lot of admixture which correlates with expansion of Slavic-speakers.

But charts and graphs show mostly "Iran+Armenia" instead. So - as I wrote above - their text contradicts their figures.

Were Proto-Slavic-speakers a tribe of steppe nomads? Mainstream scholars believe that Proto-Slavic-speakers were agriculturalists.

=====================================

Even Figure 4. contradicts with Figure 2., as it shows that speakers of Proto-Slavic were genetically like Lithuanians and Ukrainians:

Is "Lithuanian-like" admixture consistent with a sum of "Iran+Armenia" plus "Turkey" plus "NEEurope" admixtures ???:

It shows that genetically "Lithuanian-like" speakers of Proto-Slavic language were expanding from 469 CE to 1150 CE:

http://s24.postimg.org/5g0g9xwqd/Lithu11.png

 

[bicicleur 2]

If their dates are correct "Iran+Armenia" admixture occured in both South-Eastern and North-Eastern Europe during the Slavic expansion.

And "SCEurope" admixture occured in North-Eastern Europe during the German High-Late Medieval expansion (so-called "Ostsiedlung").

So something is strange here, why were Slavs spreading "Iran+Armenia" and Germans "South-Central Europe" admixtures ???

When you take a look at their Supplementary materials, you can see that South-Central Europeans = Italians, not Germans.

============================

On the other hand, this constant stream of "NWEurope" (dark blue) admixture in North-Eastern Europe, which - as can be seen - starts already some time Before Christ, and which continues until roughly 500 AD, is consistent with expansions of Proto-Germanic-speakers from Scandinavia:

http://s28.postimg.org/b5t5czg25/Expansions.png

"NWEurope" admixture makes sense for Celts and East Germanic tribes.

But do "Iran+Armenia" and "Turkey" make sense for Slavic expansion?

And does "SCEurope" make sense for German Medieval "Ostsiedlung"?

maybe they got the directions of admixture wrong
they switched head with tail ?

 

[Tomenable]

In the text they say that Proto-Slavs were genetically like modern Lithuanians and - to a lesser extent - like modern Ukrainians. And this is what Figure 4. (map showing directions of gene flow in period 469 - 1150 CE and 708 - 1463 CE) shows. But when you look at Figure 2., they show - in that period when "Lithuanian-like and Ukrainian-like Proto-Slavs" expanded - mostly "Iran+Armenia" admixture, with some "Turkey" and some "NEEEurope".

Can modern Lithuanians and Ukrainians be modelled as a mixture of "Iran-Armenia", "NEEEurope" and "Turkey" ???

Maybe it's just the same thing.

======================================================

If gene flow from Iran/Armenia-like source was larger than from Lithuania/Ukraine-like source, why do they speak Slavic?:

http://s23.postimg.org/45k60rs4b/Slavicisation.png

 

[Tomenable]

maybe they got the directions of admixture wrong,

they switched head with tail ?

So you say that it could be admixture from North-Eastern and South-Eastern Europe TO Iran-Armenia, Turkey and South-Central Europe?

That would make some sense (due to barbarian migrations from the north to the south, and later due to Muslim trade of White slaves).

But I don't think that they could make such a huge mistake.

===============================

Videos about Muslim trade of White slaves (both from Western Europe via the Mediterranean and Atlantic, and from Slavic countries):

https://www.youtube.com/watch?v=WtrJOdoVnrg

https://www.youtube.com/watch?v=bZKM6rMN25w

 

[Tomenable]

"Exotic" fair-haired blonde, light brown or redhead women were definitely in demand in Muslim slave markets.

And I don't think that they didn't leave any surviving offspring, so they surely left admixture signatures as well.

 

[LeBrok]

I'm more bothered by this "Iran+Armenia" and "Turkey" admixtures. It looks like Slavic migrations were spreading a lot of "Iran+Armenia" admixture, which kind of makes no sense unless we don't know something about genetics of Proto-Slavs:

http://www.cell.com/current-biology/pdfExtended/S0960-9822(15)00949-5

Also authors of this study used Lithuanians as a proxy for Proto-Slavs ("a Slavic source represented here by a cluster of Lithuanians"). It confirms data from the previous paper that Balto-Slavic populations are similar genetically, but shouldn't Ukrainians be used instead?

Right, I would have chosen Belorussians for slavic proxy.

And if you look at Figure 2. then there is also some kind of "SCEurope" admixture in North-Eastern Europe in times when the German "Ostsiedlung" was supposedly talking place. But "South-Central Europe" is actually associated with Italians in this study - so how on Earth did German-speaking immigrants from the HRE manage to spread Italian-like admixture instead of "NWEurope" admixture ???[/QUOTE]I can't comment on this one, I haven't had time to dive into the paper.

 

[LeBrok]

If their dates are correct "Iran+Armenia" admixture occured in both South-Eastern and North-Eastern Europe during the Slavic expansion.

And "SCEurope" admixture occured in North-Eastern Europe during the German High-Late Medieval expansion (so-called "Ostsiedlung").

So something is strange here, why were Slavs spreading "Iran+Armenia" and Germans "South-Central Europe" admixtures ???

When you take a look at their Supplementary materials, you can see that South-Central Europeans = Italians, not Germans.

============================

On the other hand, this constant stream of "NWEurope" (dark blue) admixture in North-Eastern Europe, which - as can be seen - starts already some time Before Christ, and which continues until roughly 500 AD, is consistent with expansions of Proto-Germanic-speakers from Scandinavia:

http://s28.postimg.org/b5t5czg25/Expansions.png

"NWEurope" admixture makes sense for Celts and East Germanic tribes.

But do "Iran+Armenia" and "Turkey" make sense for Slavic expansion?

And does "SCEurope" make sense for German Medieval "Ostsiedlung"?

Turkey and Iranian connection might have come with Steppe migration prior and during Slavic expansion, as Epoch mentioned, the Huns, Kazars, Scatians, Sarmatians, Turks, Bulgars, etc. It doesn't necessarily mean Slavs spread them, just that it happened at that time. This study, however interesting, has too small resolution and only points to big scale events.

 

[Angela]

But we aren't talking about prehistoric gene flow such as from Yamna or ENF. We are talking about admixtures during the Middle Ages.

Check Figure 2 B - e.g. "Iran+Armenia" admixture in North-Eastern Europe and in South-Eastern Europe was in period 500 - 1500 AD:

Their Figures (charts / graphs) contradict what they write in actual text, because they don't even explain this "Iran+Armenia" thing:

http://s16.postimg.org/n21kalef9/Figure_2_B.png

I know. That's why I said this:

My general feeling about this paper is that the authors may be picking up signals of ancient admixture, or long term continuing admixture, and attributing all of it to the most recent and not necessarily the strongest migrations.

What migration into these areas around this time would bring Caucasus like/Iranian like/Armenian like admixture? Some of the Hunnic tribes, perhaps? Scythian like tribes? I find it hard to credit the effect was this large, and that prior admixtures weren't more important.

I feel the same way about the proposed signal of Greek admixture into Sicily during the Byzantine era. I'm sure there was some, but there's no sign of the undoubtedly much larger one created as the result of the establishment of so many Greek city states in southern Italy in the first millennium BC, so large that it was known as Greater Greece (Magna Graecia)? I find that hard to credit as well.

Ed. This conflating of the last signal of admixture with prior similar gene flows is a well known problem with Roll Off. They may not have solved this problem.

I also think, as others have mentioned, there may be problems getting the direction of gene flow right, although I doubt some female slaves from Slav areas, who would, after all, have gone to the rich, would account for this. They may, however, be picking up EHG flow into areas south of the Caucasus that was part of the Indo-European migrations.

 

[Angela]

Can someone point me to a table or area of the paper and/or supplements where the precise amount of admixture is given for some of these events? I can't find it anywhere.

I'm going blind trying to use a ruler, and it would all be rough estimation at best. Plus, there seem to be differences between what I would estimate about Levant admixture, for example, from Figure 2b versus Figure 4. In figure 2b it looks as if Italy North is at about 2%?, TSI maybe 5%?, and Italy South 10% -12%, while Sicily extraordinarily, by contrast, seems to have none. On Figure 4 a circle at about 18%? value seems to cover all of southern Italy/Sicily, and for TSI it seems to be about 10%. Can somebody explain that to me? Am I just estimating incorrectly from Figure 2b?

Also, can someone tell me the area where precise percentages are given for this "Cypriot" flow into Italy in the first millennium AD? Using my manual method and looking at Figure 2, it looks like 10% for Sicily?, 5% or less for Northern Italy, 5% for TSI 23, and none for TSI 70 or southern Italy.

If I look just at my estimates from Figure 2b instead of at Figure 4, there's a see saw for Sicily and S. Italy.
Cyprus/Levant:

Sicily: 10/0
S.Italy: 0/10-12

TSI 23: 5/5
TSI 70: 0/5
N.Italy: 2/2

That sort of makes sense to me. They are "related" groups. Depending on the "host" population, the "donor" population may look slightly different?

Still, I don't know how they get from this to the values in Figure 4.

Some things in the paper give me pause. On Figure 4 there's no recent "Levantine" flow into northern Italy, or maybe 2% if we go by Figure 2b, but maybe 15% in France? How did France get all that recent Levantine flow. It apparently didn't come by way of northern Italy. Was it by way of Spain? I don't know of any first millennium AD movements from Iberia into France other than through the Moorish forces. Maybe Charles Martel didn't arrive in time?

The results for Sardinians in Figure 2 are a little strange as well. The Sardinians nestle right around the ancient EEF samples, yet more than 50% of the ancestry in some of their clusters comes from recent migrations from northern Italy, and they also have recent ancestry from the Levant and North Africa? In terms of yDna, I don't think 50% can be traced to northern Italy.

In the same graphic it also looks like they're proposing that 20% to maybe one third of the dna in some of the Spanish clusters is from northern Italians? Am I getting that right? Northern Italians are heavily U-152, but the numbers are certainly not high in Spain.

Likewise, I find it odd that there's a little chunk of Armenia/Iran in TSI 70, then smaller in Sicily and Northern Italy, but TSI 23 and southern Italy barely get a sliver? Is that conne

I'm also having trouble understanding the TSI numbers in general. Goodness knows, I 'm aware that there's phenotypical variety in Tuscany if nothing else, but SIX clusters in that province alone? Then there are significant differences in admixture and admixture events between the two groups they closely examine, TSI 23 and TSI 70. Just for one example that I didn't yet raise, TSI 70 seems to have a huge chunk of SWE (France or Spain), almost 60%?, but TSI 23 has none? Also, as I mentioned above, Cyprus gene flow is found in TSI 70 (at 5%?), but none is found in TSI 23 (0)? The Caucasus goes into TSI 70, but barely a sliver into TSI 23?

Other than in the far northwest, there are no geographical barriers to gene flow in Toscana. Plus, it was my understanding that the HapMap samples all came from one town around Florence. (I know a few Li et al samples from other places were also included, but there were only 8 of them.) How could there be such a difference?

I'm also confused by the fact that while in the text the authors seem to be proposing the invasions by the Moors as the vector for SSA, the figures show that the North African into some of the areas with the higher SSA is quite low. On Figure 2b, it shows North African only going into Sardinia, yet the West African goes not only into Sardinia but into Italy, Spain and France. How is that possible? The SSA is there, at about 5% if I'm estimating correctly from Figure 4? Wouldn't the SSA have been transmitted by means of people who were largely North African? Don't the proportions seem off? Perhaps the non SSA portion of the Moors is so EEF like that it doesn't register in their algorithm, but doesn't that call into question the ability of their method to "see" not only the date of the admixture but the admixture itself?

I really had hopes that this paper might clarify some of the outstanding questions about the population history of Italy, but I have to say I'm more confused now than I was before. Just in terms of this "Levantine" flow that they're proposing into Italy, I could see it as a result of the Moorish invasions, but they're proposing that the "Levantine" admixture came during the Roman era if I'm reading it correctly, so probably tied somehow to Roman slavery I would think. Fine if that's the case, but weren't the Gauls and the Germans and the British manumitted and allowed to "admix"? Why only the eastern Mediterraneans? Also, as far as the "Cypriot" ancestry is concerned, they're dating it to after the Roman period, at least for the northern Italians, so it's even harder to understand. I do see that they've changed the wording since Hellenthal et al, as well as drastically lowering the percentages. Now they're saying that: "in the northern Italians (itali 13: 677 (362-989CE) involving Cypriot and French like sources". Of course, it's too late for the Gallic invasions, but it would just about cover any Byzantine input during the Gothic Wars. So, maybe 2% or so "Cypriot" into northern Italy by way of them might make some sense? (5% into TSI and 10% into Sicily?)


One last anomaly of sorts. I don't see any recent gene flow from the Balkans into Italy at all, yet that's a gene flow that we know happened, both from history (the migration of the Arbereshe) and from the IBD analysis of Ralph and Coop.

By the way, that Ralph and Coop analysis based on IBD segments didn't see any signficiant gene flow into Italy even from Turkey, much less from the Levant after at least 400 BC. I guess 2%, 5% and 10% "Cypriot" might not qualify as "significant", but those "Levant" numbers in Figure 4 certainly would...

 

[LeBrok]

To be honest, I got lost analyzing their graphs. Things were more confusing than helpful. But firstly, I would like to know how on earth they can tell the time of admixture from just looking at the genes?

 

[Angela]

To be honest, I got lost analyzing their graphs. Things were more confusing than helpful. But firstly, I would like to know how on earth they can tell the time of admixture from just looking at the genes?

This is what the paper says:
" We can infer the date of admixture by modeling the decay of LD between ancestral chunks, which decreases more rapidly the longer ago admixture occurred." I didn't find anything more specific.
http://www.cell.com/cms/attachment/2036746823/2051577514/mmc1.pdf

From page 23 of the supplement, this is what they see as the "challenges" (among others) for their program:
3.even in theory our approach finds it challenging to distinguish distinct continuous “pulses” of admixture and continuous migration over some time frame, because of the difficulty of separating exponential mixtures [S? If the time frame were narrow, we expect to infer a single admixture time within the range of migration dates. Where we infer two admixture dates, in particular with the same source groups, the exponential decay signal could also be consistent with more continuous migration, and so we conservatively refer to this as admixture at multiple dates".

Oh, I forgot to mention this doozy of a comment, after all these graphs and percentages etc:
"When we compare the relative differences between pre- and post-admixture groups, we observe no appreciable difference between them, suggesting that admixture has not had a significant impact on genetic variation in West Eurasia (Figure 4G)."

HUH? Then what was this all about, and how valid were their "admixture" numbers?