search this blog

Thursday, November 24, 2016

Multiple Holocene Eurasian migrations into north central Africa


The AJHG has a new open access paper on the population history of Chad. It lays out a reasonable hypothesis for the sources and timing of Eurasian gene flow into north central Africa. Just wondering, though, if the authors considered the possibility that R1b-V88 may have expanded into Africa from Iberia? After all, the oldest instance of a probable R1b-V88 to date is from an Iberian Early Neolithic sample (ID I0410, Els Trocs, Spain, Haak et al. 2015). But I don't have a strong opinion on the issue at the moment. From the paper (emphasis is mine):

Understanding human genetic diversity in Africa is important for interpreting the evolution of all humans, yet vast regions in Africa, such as Chad, remain genetically poorly investigated. Here, we use genotype data from 480 samples from Chad, the Near East, and southern Europe, as well as whole-genome sequencing from 19 of them, to show that many populations today derive their genomes from ancient African-Eurasian admixtures. We found evidence of early Eurasian backflow to Africa in people speaking the unclassified isolate Laal language in southern Chad and estimate from linkage-disequilibrium decay that this occurred 4,750–7,200 years ago. It brought to Africa a Y chromosome lineage (R1b-V88) whose closest relatives are widespread in present-day Eurasia; we estimate from sequence data that the Chad R1b-V88 Y chromosomes coalesced 5,700–7,300 years ago. This migration could thus have originated among Near Eastern farmers during the African Humid Period. We also found that the previously documented Eurasian backflow into Africa, which occurred ∼3,000 years ago and was thought to be mostly limited to East Africa, had a more westward impact affecting populations in northern Chad, such as the Toubou, who have 20%–30% Eurasian ancestry today. We observed a decline in heterozygosity in admixed Africans and found that the Eurasian admixture can bias inferences on their coalescent history and confound genetic signals from adaptation and archaic introgression.

...

It is important to note that in this work we inevitably invoke Occam’s razor to support the simplest model consistent with our data; the history of the populations studied here, including the time and sources of the Eurasian admixture in Africa, could be more complex. aDNA from Chad and neighboring regions remains a challenge given the poor DNA preservation in hot climates, but future successful efforts in aDNA research could provide additional insights and reveal additional complexities not considered by the modern-DNA-based models favored here.

Haber at al., Chad Genetic Diversity Reveals an African History Marked by Multiple Holocene Eurasian Migrations, AJHG, Published Online: November 23, 2016, DOI: http://dx.doi.org/10.1016/j.ajhg.2016.10.012 show

See also...

R1b-V88: out of the Balkans and into Africa?

56 comments:

Nirjhar007 said...

ah shit I thought it was an aDNA paper :P .

Samuel Andrews said...

The Toubo from Northern Chad have significant Eurasian ancestry but the Sara and Lala from Southern Chad have little or no Eurasian ancestry. Toubou also have 2% frequency of the lactose tolerance LCT(rs4988235) allele.


A test of the form of f4(British, chimp; X, Yoruba)/f4(British, chimp; early Neolithic farmer, Yoruba) gives the Toubou 26-30% Eurasian admixture, 0.3-2% Sara, 1.2-4.5% Lala.

The best proxies for Toubou's Eurasian ancestor are Sardinians, EF Europeans, and North Africans. The same is true for the Ethopians they tested the same way.

capra internetensis said...

Nice to see something from Central Africa. They really didn't do much beyond the Eurasian admixture stuff but I'm sure we'll see these genomes in future papers.

Looking at the f3 stats there is some subtle difference; the Toubou favour WHG (and Spanish Neolithic, Motala, Samara_HG) as ancient admixing populations a little relative to Amhara, whereas it goes slightly the other way for Yamnaya, Corded Ware, and MA1. Likewise with modern reference populations the Toubou have a lot of North and Western European populations in the most negative f3s where the Amhara have some Middle Eastern populations. I don't know if this necessarily has anything to do with the actual admixing population but the Eurasian element of the Toubou looks more western.

Aram said...

This V88 farmers were probably attracted by this.

_------
Lake Chad is the remnant of a former inland sea, paleolake Mega-Chad. At its largest, sometime before 5000 BC, Lake Mega-Chad was the largest of four Saharan paleolakes, and is estimated to have covered an area of 1,000,000 km2 (390,000 sq mi), larger than the Caspian Sea is today, and may have extended as far northeast as within 100 km (62 mi) of Faya-Largeau.[7] [8] At its largest extent the river Mayo Kébbi represented the outlet of the paleolake Mega-Chad, connecting it to the Niger River and the Atlantic.[9]

Aram said...

It would be nice to have aDNA from Sao Civilisation in Chad.

https://en.m.wikipedia.org/wiki/Sao_civilisation

Gioiello said...

I'll read the paper, but of course I was the first to demonstrate, against all (also the authors of this paper), that R-V88 came from Europe, and exactly from the Italian Refugium. I'll publish all the documentation and also letters with Fulvio Cruciani.

Shaikorth said...

@Capra

Iranian Neolithic-related ancestry drops relative to European Neolithic when moving towards west in North Africa according to Broushaki et al. Additionally Mandenka and Gambians show very small traces of NE1 and LBK but no WC1, which is clearly a donor for Ethiopians.

Gioiello said...

The paper at last says: "It is important to note that in this work we inevitably invoke Occam’s razor to support the simplest model consistent with our data; the history of the populations studied here".
If the authors would have used the Occam's razor:
1) they wouldn't have used a complicated autosome for demonstrating the migration and the origin of the Y R-V88 many thousands of years ago
2) they wouldn't have chosen as Eurasian populations Greeks Lebanese and Yemenites and not Sardinains, Italians, Iberians, and also Englishmen where we have the oldest haplotypes of R-V88 and the unique aDNA of the R1b hg.: Italy 14000 years ago and Iberia 7100 ya
3) they would have carried some aDNA from Middle East, where no R1b has been found (and it won't be, because it there wasn't there)
4) they wouldn't be a Jew (Haber) a Lebanese (Zalloua) a National Geographic (Wells) and haven't used funds from the Lebanese American University
5) they would have given a glance to the trees of Sergey Malyshev (smal) at FTDNA and YFull, where the oldest haplotypes of R-V88 are in Western Europe
6) they would have read my writings in these last ten years where I demonstrated that all what they say is false.

Grey said...

"Just wondering, though, if the authors considered the possibility that R1b-V88 may have expanded into Africa from Iberia?"

If you look at known trade routes from later periods you see two routes to the West African gold fields: north over the Sahara to the Med. (e.g Carthage, Tripoli etc) or east then north to Egypt

http://dragonheadmusic.com/images/750px-Niger_saharan_medieval_trade_routes.png

with the Sahel maybe important as the easiest route on the eastward path to Egypt

http://image.shutterstock.com/z/stock-vector-sahara-and-sahel-map-154667081.jpg

If you look at the R1b-V88 distribution in Africa

http://cache.eupedia.com/images/content/Haplogroup_R1b_World.png

you can see at least a trace of what might be a roughly matching U-shaped distribution going from Tripoli down to the Sahel and then east and up to Egypt.

#

I've been working on the assumption this may have been connected to the West African gold fields with miners/traders starting from somewhere like Cyprus via Egypt and the Sahel (although Iberia->Tripoli->Sahara is the same just from the other direction).

However the peak R1b-V88 is quite a way away from the gold fields and closer to the Lake Chad area - closer to the copper field in Takeda/Agadez in Niger.

https://sites.google.com/site/medievalsitesofencounter/_/rsrc/1468865144606/mali/instructions-for-labeling-the-west-africa-map/4.2%20Medieval%20West%20Africa%20Map.png

https://s-media-cache-ak0.pinimg.com/originals/fa/b6/d9/fab6d9ed1a603583719ce0c1c5af8e67.jpg

If you look back at the V88 map the NW branch of the main African distributions seems to stretch back to the Agadez region.

#

So... ancient U-shape: Tripoli->Agadez->Sahel->Egypt with the northern half of the distribution erased and the remainder nudged south by the Arab expansion?

Grey said...

for what it's worth my earlier guess (mostly based on maps) was a fork

ukraine->crete then
->crete->cyprus->egypt->chad
->crete->sardinia->iberia
but
->crete->cyprus->egypt
->crete->sardinia->iberia->chad
follows the same map logic

or even
ukraine->crete->sardinia then
->sardinia->iberia
->sardinia->chad

Grey said...

Separate to the above, reading about Fulani and Tuareg pastoralists is interesting in relation to IE especially their caste system of nobles, endogamous artisan castes and slaves.

Anonymous said...

Interesting but not exactly surprising, but it looks they might have overestimated the levels of Eurasian ancestry by using the Yoruba (and say not Dinka) as a baseline for African ancestry in Chad. The Gumuz don't possess any significant Eurasian ancestry and the Toubou likely possess less, maybe around ~15-20%, than the purported amount.

"Eurasian ancestry in Ethiopians ranged from 11%–12% in the Gumuz to 53%–57% in the Amhara. African ancestry in the Near East ranged from 7%–14% (Yemen) to 0.7%–5%
(Lebanese Christians)."

Grey said...

off-topic

mammoths

http://www.pbs.org/wgbh/nova/next/nature/mammoth-discovery-could-revise-earliest-date-of-humans-in-the-americas/

AWood said...

I'm not entirely convinced the Ethiopians have anything to do with the Toubou. R1b-V88 and T are still the dominant male lineages and both have been found in a European Neolithic context. No burials during the LBK period have been found with the common European LCT mutation, but those directly after in Central Europe have. This gets more and more interesting.

John Thomas said...

Now for that vexed question - 'the haplotype E wars' - which in its own way can only match that long fought battle over r1a1a in its intensity, and vehemence of its protagonists.
Is y DNA haplotype E ultimately Eurasian or African in provenance?
Does the near dominance of E in sub Saharan Africa imply an ancient male biased elite dominance? - with all the obvious bitter political ramifications that implies?

Folker said...

Good to see that other people are now considering that R1 V88 could have been from Iberia, and could have travelled to Chad following paelorivers (who existed in North Africa during the Green Sahara:
https://www.sciencedaily.com/releases/2013/09/130911184712.htm

Ravai said...

Who would form the calls by FTDNA as "Invader Metal Age" ?? Are there any specimens in Gedmatch?

Thank you

Regards

Gioiello said...

@ Folker "Good to see that other people are now considering that R1 V88 could have been from Iberia, and could have travelled to Chad following paelorivers (who existed in North Africa during the Green Sahara:
https://www.sciencedaily.com/releases/2013/09/130911184712.htm"

Very likely from Italy to Iberia 7500 years ago and from Iberia to Northern Africa. I am saying that from many years and have letters from Cruciani and Scozzari I published in many fora now deleted but also on www.worldfamilies.net etc.

bellbeakerblogger said...

A plausible scenario could be made that V88 & M269 of the Iberian-Tyrrhenian sphere made a significant impression on parts of North Africa in the Early Bronze Age and these communities only later became Berberized in the Iron Age before spreading deeper into Africa.

Personally, I think the Maritime Impresso folk have some surprises to show us. Els Torcs may not be an outlier and it's possible the Atlantic littoral, Sudan and blotches across the Mediterranean littoral had farmer-dairy communities that were more different from the Balkan-Danubian farmers than is currently believed.

Yeah Grey, Clovis first has been dying a slow death.

Gioiello said...

@ bellbeakerblogger

It is clear from so long that many Nigerians like Auwalu Mussa (I had his data once) are closely linked to Berbers and perhaps Fulanis, and that DYS464 pretty Always is 12-12-15-15 demonstrates that in some cases the link is recent, and that is also on the YFull tree, even though I think underestimated for an 1.17 factor.

AWood said...

How come R1b-V88 is nearly non-existant in NW Africa but Central and the Levant it reaches very high levels? Seems like it travelled the eastern Mediterranean rather than the west. Follow the most likely scenario, not the uphill battles folks.

Olympus Mons said...

@Aram.
Have you read in my Shulaveri2Bellbeaker the suplement - Suppl II Egypt (Stay up North!)?

That part of the history is very vague...but I do adress the SAO Kotoko and so forth in the context of the V88.



Folker said...

Simply because Y haplogroups modern distribution represent recent migrations, which erased more older ones. So modern distribution is not a good clue about ancient one. And we don't have aDNA from North Africa. Could be very surprising, with little contribution to modern populations.

Simon_W said...

@ David Rabaez

I'm not sure if I understand your oddly formulated question, which moreover seems completely off topic. But since no one else is going to reply, I'll address it quickly. FTDNA now models Europeans as a mix of hunter-gatherers (Loschbour, La Brana 1 and Motala), early European farmers (Stuttgart, Iceman, LBK) and so called "Metal Age invaders" which are based on Corded Ware and Yamnaya samples - so the latter is what we more typically call "steppe admixture". (Plus they add non-European admixture in case someone has it very evidently.) This model looks crude for our present-day standards, and their results look very different from what I've seen here in Davidski's analyses or in other ancient-DNA calculators on Gedmatch, as their inferred proportions of steppe admixture are much lower, about half of that inferred by others. And moreover to me it looks completely wrong that FTDNA suggests that I (1/4 Italian, strong southwest European component) have more Metal Age invader admixture than my fully East Prussian grandmother.

Ryan said...

My guess is ~15,000 kya - Steppe > Ukraine, Spain, Italy, Anatolia
~10,000 kya Italy > Tunisia
~8,000 kya Tunisia > Sahara

Unknown said...

On mtDNA in Chad:- http://onlinelibrary.wiley.com/doi/10.1111/ahg.12040/full?wol1URL=/doi/10.1111/ahg.12040/full&regionCode=GB-EN&identityKey=bb4992ba-170e-44b3-ad14-1cf19b52a9d2 "In Daza (Toubou sub-group), we identified only M1 and U6 (no H1, H3, V, or JT) clades, suggesting rather the pre-Neolithic formation of their maternal gene pool."

capra internetensis said...

Looking at the tree of 1000 Genomes + WGS Y haplogroups in the supplement, there are 2 Toubou and 1 Laal R1b-V88s. One of the Toubou is on a branch with the Peruvian HG01947, who is under R1b-FGC20890 (Jewish clade presumably), the other is on another branch with the Laal, which I assume is Y7771. So both branches of R1b-Y8447 appear to be in Central Africa, and their TMRCA should be for that node.

That age would fit an initial spread of Y8447 still during the second, pastoralist-dominated phase of the Green Sahara. From the inadequate data presently available the FGC20973 branch looks predominantly Semitic and the Y7111 branch mainly African, so a spread between the Near East and Northeast Africa at this time would make sense. I don't know about non-Jewish Europeans but Sardinians at least have V35, the sister branch of Y8447.

This is also the time of Cardial Ware, we know they had at least a little of it, and they landed in Lebanon and North Africa as well as Italy and Spain.


@AWood

I don't reall see a reason to favour an Iberia-Morocco route fro V88 either. However, the Maghreb has been drowned in the recent expansion of E-M81, and then a bunch of Arab stuff on top of that, so the frequencies 3000 years ago may have been very different than they are now. Going to the far west of the Sahara, the sample of 189 Saharawi (Western Sahara + Mauretania) from Bekada et al (2013) has 7% R1b-V88, which is a very respectable frequency considering the degree of dilution.

It's not particularly common in the Levant, as far as I know, except for that one hot spot around the Dead Sea. It would be nice to have better sampling in that area, Lebanon is the only place I know of other than Sardinia and Corsica to have R1b-M18.

Davidski said...

@David Rabaez

FTDNAs "Invader Metal Age" is not a real ancient population. It's just a cluster they found by running lots of modern and I think some ancient samples.

Grey said...

BBB

"Yeah Grey, Clovis first has been dying a slow death."

Yeah I was thinking more of them using pools to store meat and a possible relation to the LGM and the possibility of a single mammoth kill lasting a band for weeks or even months if they dug cold lockers.

Gioiello said...

@ capra internetensis
"Looking at the tree of 1000 Genomes + WGS Y haplogroups in the supplement, there are 2 Toubou and 1 Laal R1b-V88s. One of the Toubou is on a branch with the Peruvian HG01947, who is under R1b-FGC20890 (Jewish clade presumably), the other is on another branch with the Laal, which I assume is Y7771. So both branches of R1b-Y8447 appear to be in Central Africa, and their TMRCA should be for that node.

That age would fit an initial spread of Y8447 still during the second, pastoralist-dominated phase of the Green Sahara. From the inadequate data presently available the FGC20973 branch looks predominantly Semitic and the Y7111 branch mainly African, so a spread between the Near East and Northeast Africa at this time would make sense. I don't know about non-Jewish Europeans but Sardinians at least have V35, the sister branch of Y8447.

This is also the time of Cardial Ware, we know they had at least a little of it, and they landed in Lebanon and North Africa as well as Italy and Spain".

The oldest subclade of R-V88, just R-V88*, is Sexton/Saxton from the Isles, and Marchesi from Italy, of course tested wrongly from FTDNA as M269. Another R-V88* is very likely in Boattini 2013 from Macerata, Le Marche, Italy.
African and Middle Eastern R-V88 are all recent, less than 7000 ya, i.e. from the sample found in Iberia 7100 years ago and come very likely from Italy.

"Lebanon is the only place I know of other than Sardinia and Corsica to have R1b-M18".

R-V88-M18 is found amongst Druzes and not Lebaneses, but we don't know their haplotypes or SNPs. Samples (two clusters) are in Corsica and Sardinia, and are older than R-V88-V35 and all the others known so far. We are waiting that the Druze samples are tested and submitted to YFull. I think that they come from Sardinian mercenaries of the second millennium BC.
I remember you all that no R1b has been found in aDNA in Middle East, Anatolia, Iran, but Villabruna, Italy, is R1b1 with two SNPs at the P297 level and R-V88 is found in Iberia 7100 years ago.



Gioiello said...

@ AWood

"How come R1b-V88 is nearly non-existant in NW Africa but Central and the Levant it reaches very high levels? Seems like it travelled the eastern Mediterranean rather than the west. Follow the most likely scenario, not the uphill battles folks".

Inconsistent argumentation, with no knowledge of genetics, history and all the rest. No knowledge of haplotypes, haplogroups, mutation rates, trees etc etc, as those used in the past, but here we aren't on Anthrogenica and you may ban me. In these years you all have been defeated by me.
Only Davidski may save his arguments (in part), that R1a and IE languages (only the satem ones I think) expanded from Eastern Europe.

Aram said...

Olympos

I didn't read the Sao part. Can You give the link? But I think V88 is too old to be linked with Sao.

Gioiello said...

@ capra internetensis

"That age would fit an initial spread of Y8447 still during the second, pastoralist-dominated phase of the Green Sahara. From the inadequate data presently available the FGC20973 branch looks predominantly Semitic and the Y7111 branch mainly African, so a spread between the Near East and Northeast Africa at this time would make sense. I don't know about non-Jewish Europeans but Sardinians at least have V35, the sister branch of Y8447".

All these people descend from one person (R-Y8447) who is 7000 years old. From a descendant of his (R-FGC20973) descend 3600 years ago YF04009SAU and HG01947 PEL. The Jewish cluster R-FGC21047 is only 1050 years old, thus what you say is unjustified. This subclade of R-V88 may have expanded during the Bronze Ages (as the same bellbeakersblogger said above) and the Jewish cluster have introgressed in the Jewish Sephardic pool in Iberia.
Without aDNA all what you say is only a prejudice, i.e. a thought without (sound) judgement.

Ravai said...

Thanks @ Simon_W, thanks Davidski!!!

Regards

Olympus Mons said...

@Aram,
just go to http://shulaveri2bellbeaker.blogs.sapo.pt/ and Ctrl F to "Egypt (Stay up North!)". Its one of the supplements.

around middle of that suplement I start adressing Farafa oasis, Nabta and moving into sub-Saharan Africa.

Ravai said...

I am not an expert in the field, I am in learning. I hope you can use this data. Thanks!

Spanish people id gedmatch:

ID I0410, Els Trocs, Iberia, 5.100 ybp, M641265

Minimum threshold size to be included in total = 150 SNPs
Mismatch-bunching Limit = 75 SNPs
Minimum segment cM to be included in total = 3.0 cM

Country, id, "Largest Segment(cM)", "Total of segments > 3 cM", "Matching Segments"

Andalucia (Spain) T811060 6.70 13.80 3.00
Catalunya (Spain) T143876 4.30 8.00 2.00
Basque (Spain) T339288 3.50 3.50 1.00
Basque (Spain) T307182 5.20 8.70 2.00
Catalunya (Spain) T506756 4.20 17.90 5.00
Catalunya (Spain) T078035 5.40 5.40 1.00
Catalunya (Spain) T346472 3.80 3.80 1.00
Catalunya (Spain) T130022 4.70 8.40 2.00
Catalunya (Spain) M480828 7.50 12.20 2.00
C. Valenciana (Spain) M381432 5.70 14.00 3.00
Catalunya (Spain) T380559 3.50 13.20 4.00

Regards














Gioiello said...

Sample ID HG 7278990/S10054 8337108/M8363 22442159/ZS4440
REFSEQ G G G
YF07201 R-V88* A A T
The proof that YF07201 is FTDNA 431539, Sexton
_b2. R1b-V88* 431539 George Sexton, d ~1689, Westfield MA Unknown Origin R-PF6290
13 24 15 10 13-15 12 12 11 13 13 29 17 9-10 11 12 26 15 19 28 12-12-15-17 11 11 21-24 15 13 16 18 34-35 12 11 10 8 15-15 8 11 10 8 11 9 12 22-22 16 12 12 12 15 8 12 22 20 15 13 11 13 11 11 12 13 32 15 9 15 11 24 26 19 13 11 12 12 10 9 12 11 10 11 11 30 12 12 25 13 11 10 21 15 18 13 23 16 12 16 24 12 24 18 11 15 18 9 11 11
and that also Marchesi belongs to the same cluster
R1b1c (V88) or likely based on STRs
266227 Giuseppe Marchesi b. 1856, Piedmont, Italy Italy R-M343
13 24 15 10 13-15 12 12 12 13 13 29 18 9-10 11 12 27 15 19 28 12-12-15-15 11 11 20-24 15 13 16 17 32-35 12 11 10 8 15-17 8 11 10 8 10 10 12 21-22 16 13 12 12 15 8 12 22 21 14 13 11 13 11 11 12 13 31 15 9 13 11 24 26 19 13 11 12 12 10 9 12 11 10 11 11 30 12 12 24 14 11 10 23 15 18 11 23 16 12 16 24 12 24 18 11 15 18 9 11 11
and also the sample from Boattini 2013.

Seinundzeit said...

For what it's worth, the proportions look a little different, if we change up the West Eurasian and Sub-Saharan African references (and use a different method).

Utilizing David's excellent global PCA data-sheet, I took advantage of my tried-and-tested reference population setup.

It's the same one I've used on West Eurasian populations as diverse as Moroccans, Egyptians, Lithuanians, Russians, Irish, Italians, Portuguese, Greeks, Armenians, Iraqi Jews, Iranians, Balochis, the Kalash, Pashtuns, the Pamiri peoples, etc. Not to mention the successful models I've obtained for various South Indian and Austroasiatic populations.

I wanted to see if this setup could prove fruitful in an East African context. The output:

Ethiopian_Anuak

39.30% Mota
30.85% Gambian
26.35% Esan_Nigeria
3.50% Israel_Natufian
Distance=0.4448

Ethiopian_Gumuz

53.35% Mota
38.85% Gambian
6.40% Israel_Natufian
1.15% Iran_Neolithic
0.25% Esan_Nigeria
Distance=0.4258

Masai_Kinyawa

39.75% Gambian
31.15% Mota
27.40% Israel_Natufian
1.20% Iran_Neolithic
Distance=0.7674

Ethiopian_Oromo

45.15% Israel_Natufian
25.95% Mota
25.60% Esan_Nigeria
3.25% Iran_Neolithic
0.05% Gambian
Distance=0.6167

Somali

46.85% Israel_Natufian
27.20% Gambian
15.70% Mota
6.10% Esan_Nigeria
4.15% Iran_Neolithic
Distance=0.6371

Ethiopian_Amhara

55.60% Israel_Natufian
24.15% Esan_Nigeria
12.85% Mota
7.05% Iran_Neolithic
0.30% Gambian
Distance=0.7153

Egyptian

29.40% Levant_Neolithic
16.40% Israel_Natufian
23.85% Iran_Chalcolithic
10.50% Barcin_Neolithic
7.45% Iran_Neolithic
6.85% Esan_Nigeria
5.55% Gambian
Distance=0.0547

BedouinB

84.6% Israel_Natufian
15.4% Iran_Neolithic
Distance=1.4704

The fact that East Africans chose Natufians as an ancestral population, despite the presence of so many similar/related options (Levant_Neolithic, Barcin_Neolithic, LBK_EN, Iberia_EN, etc), is of considerable interest.

And the cline noted in the paper is perfectly replicated here. Basically, we have the Anuak at around 3% West Eurasian, the Gumuz at around 8% West Eurasian, the Maasai at around 25%-30% West Eurasian, the Oromo and Somali at around 50% West Eurasian, and the Amhara at around 60% West Eurasian. Personally, I feel these estimates are probably much closer to the mark.

One thing that really struck me though, the Egyptians are way more cosmopolitan with regard to different kinds of Near Eastern ancestry. It seems they can't be construed as contributors of West Eurasian ancestry to East African peoples.

I mean, they have substantial amounts of Neolithic Anatolian and Chalcolithic Iranian admixture, in contrast to the overwhelmingly Natufian + minor Neolithic Iranian ancestry seen in East Africa, not to mention a preference for the later Neolithic Levantine samples.

By contrast, Yemeni Bedouin samples (I have been told that the BedouinB are actually just samples collected from Bedouin people in Yemen) can clearly be construed as an almost perfect ancestral population for East Africans.

Which isn't surprising, as these East African populations have very ancient links with the Arabian peninsula/Yemen.

By contrast, I guess they've been far more isolated from the populations of Egypt/the Levant.

All in all, fascinating stuff.

Shaikorth said...

Can you try that model again, but with Mozabites as an added source.

Seinundzeit said...

Shaikorth,

As requested, I added Mozabites as a source population, yet the results are absolutely identical, no changes.

Which isn't surprising, as the Mozabites model very differently, using my standard setup:

Mozabite

44.55% Israel_Natufian
20.95% LBK_EN
21.20% Gambian
8.25% Iberia_EN
3.50% Loschbour
1.15% Iberia_MN
0.40% LBK_EN
Distance=0.5669

Unlike the Yemeni Bedouins, or the West Eurasian half of many East African populations, the Mozabite seem to have ancient links with some southern European populations.

As you know, there is a long history of interaction across the whole region of the Mediterranean sea, and the Mozabite were obviously affected by these networks of gene flow/population movements (we're talking about links/exchanges that probably date back to the Neolithic, or even before).

For what it's worth, here are Algerians and Tunisians, using my standard setup:

Algerians

19.80% Israel_Natufian
21.25% Gambian
19.60% Barcin_Neolithic
17.30% LBK_EN
14.85% Levant_Neolithic
2.45% Iran_Neolithic
2.05% Iran_Chalcolithic
1.55% Samara_Eneolithic
0.85% Iran_Hotu
0.30% Esan_Nigeria
Distance=0.0313

Tunisians

23.70% Levant_Neolithic
22.40% Barcin_Neolithic
19.30% Gambian
16.75% Israel_Natufian
8.15% LBK_EN
7.80% Iran_Neolithic
1.80% Iran_Chalcolithic
Distance=0.0964

The fits are exceedingly good/tight, not sure why. Also, they (along with Mozabites) are always consistently around 20% Sub-Saharan African.

Anyway, again, these populations seem to display ancient links with some Mediterranean groups in Europe, and (much more distantly) with populations in West Asia.

Furthermore, as shown previously, the Egyptians seem to have strong ties with the Levant, and even with Mesopotamia (and probably with North Africans proper).

So, it really does seem that East Africans are quite distinct from North Africans proper, and from people similar to these Egyptian samples from the northern Nile delta (other Egyptians, and North Africans proper, might be modeled differently, as these regions hold populations of considerable diversity, and people from these parts do show great variability with regard to the amounts of Sub-Saharan African admixture).

Again, I guess the affinities of East Africans lie with isolated Arabian peninsula populations/the Yemen region. I mean, the BedouinB are supposedly Bedouins from Yemen, and they look like perfect proxies for the West Eurasian portion of East African ancestry. At least going by the percentages they independently score, when analyzed with the setup I like to use with David's PCA data-sheet.


idurar said...

^ Interesting results. Although it seems the Israel_Natufian is absorbing some actual African ancestry in East Africans, Egypt and the Near East. Northwest Africans on the other hand don't seem to be affected, as expected, since the type of African ancestry there is different from what is found in the first group. As for Egyptians not being a good proxy for the West Eurasian ancestry of Horners, this should only apply to Egyptian Muslims. Copts are very likely to match better the West Eurasian ancestry of Horners (especially Cushitic speakers) than Arabian populations, although many Copts do have very clear recent West Asian (and sometimes European) ancestry.

Shaikorth said...


One peculiar thing here is that PCA-based modeling seems to overestimate West African in everyone. Here are various Broushaki fits for Africans:

Egyptian: Iran_N 0.3327 LBK 0.4699 Mota 0.1523 Yoruba 0.0451
Saharawi: LBK 0.6617 Mota 0.1555 Yoruba 0.0885
Oromo: NE1 0.0001 LBK 0.199 Mota 0.7911
Gambian: NE1 0.0003 LBK 0.0003 Mota 0.2401 Yoruba 0.7372
Kikuyu: Mota 0.6857 Yoruba 0.3029
Luhya: Mota 0.3456 Yoruba 0.6197

There is no high coverage Natufian genome but it would probably split into Eurasian Neolithic, with maybe some Mota. It surely wouldn't take enough Mota away to explain the difference and prevalence of Gambian in PCA models. Perhaps the PCA has relatively too few Africans so the dimensions showing intra-SSA variation (West and East Africans are very distinct) don't get enough weight.

Seinundzeit said...

Idurar,

I was really struck by how tightly the Algerians fit, seems like the model is very good for them.

Although, I think we might possibly find that these estimates of West Eurasian admixture (with regard to East African populations) are more accurate, compared to what we've seen previously.

I think the estimates have changed, since the Natufians are far more close to being the direct ancestral population for East Africans, while previous efforts have used Sardinians, or Early European farmers.

Early European farmers seem to be quite different from Natufians, and thus from the ancestral stream of populations from which East Africans are partially derived.

For example, LBK_EN:

66.90% Israel_Natufian
31.55% Loschbour
1.55% Iran_Neolithic
Distance=1.4524

LBK_EN have a pretty substantial shot of WHG-related admixture (compared to Natufians), so I'd imagine that they were never quite optimal for East Africans. Rather, they were just the best we could do, before uncovering more Near Eastern aDNA.

Although, physical anthropologists have hypothesized African admixture for Natufians (since all of them were very gracile, since most of them had strong alveolar prognathism, and since some of them seem to have had broader noses, etc). And as per aDNA, it even seems that they had very dark skin pigmentation.

Nonetheless, formal stats don't seem to show any shift towards Sub-Saharan Africa, so the vaguely African affinities that were noted can probably be chalked up to physical variability (many ancient West Eurasian populations, in the Mesolithic, looked strikingly different from contemporary populations in West Eurasia, as per physical anthropology).

Regardless, the paper in question claims that the Amhara can be construed as 57% West Eurasian (using European Farmers), and here they are modeled as being 60%, so it isn't much of a difference. And the Gumuz are more West Eurasian-admixed in the paper, at 12% versus 7% here.

Shaikorth,

That's interesting. I think what you say concerning dimensions makes sense.

Anonymous said...

Seinundzeit,

This particular study seems to be overestimating the level of Western Eurasian ancestry in East Africans and populations with significant said ancestry, the Amhara, Gumuz, and Toubou included. In particular, the study models all of the sampled African groups as Yoruba + X Western Eurasian group; however, the African ancestry in East Africans is best modeled by the majority component in the Dinka and Mota for example. The Gumuz generally exhibit very little if any non-African ancestry, so the use of the Yoruba as a proxy would explain their unusual results in this case. This is just another example of deep-rooted population sub-structure in Sub-Saharan Africa.

The Natufians are still a puzzle, while D-stats and f4-stats don't seem to be picking up on any African ancestry in the Natufians, other measurements indicate otherwise, e.g. PCA, treemix, etc. I am personally in the opinion that the Natufians possessed some ancestry from a population native to North Africa that's not perfectly modeled by Mota or any other modern SSA population.

Davidski said...

I do have one Dinka individual. A full genome no less. I've added him to the Global_10 datasheet.

https://drive.google.com/file/d/0B9o3EYTdM8lQaU5CRFFfdjN4OW8/view?usp=sharing

Seinundzeit said...

Gihanga Rwanda,

I think African populations will always constitute really complex puzzles, considering the deep-rooted population sub-structure you mention. I mean, this continent is humanity's "birthplace", one would expect great complexity, and probably many streams of unsamples "ghost populations"

Also, what I've found is that every method provides different percentages, even for well studied/"simple" populations. Personally, I think it might be impossible for all these methods to fully converge.

Honestly, I guess all we can do is hope for more aDNA samples (besides Mota, the whole African continent is terra incognita with regard to aDNA). And with every set of newly acquired samples, we should just keep trying all methods available to us. At some point, results will begin to stabilize (to an extent) across different methodologies.

I mean, I still remember how analyses of Central Asia and South Asia were always fraught with uncertainty/subject to intense debate, and then we got lucky by acquiring Bronze Age steppe samples + Neolithic Iranian samples, both sets of which had a huge effect together, in terms of bringing clarity/shedding light.

Personally, I'd exercise a sort of skepsis (in the original sense of the term) with regard to these East African results.

On first examination though, they do strike me as exceedingly reasonable, and the patterns are of considerable interest. Mainly, I'm fascinated by the genetic isolation seen with regard to the Nile delta/greater North Africa, and the strong affinity seen with some populations from the Arabian peninsula.

In addition, when it comes to the Natufians, I do incline towards the view that the Natufians lacked any ancestry related to present-day populations in Sub-Saharan Africa (but I do find your ideas on this to be of great interest). The African affinity displayed in analyses like PCA are likely a by-product of Natufian-related populations having a very strong/intense role in the ethnogenesis of some East African populations, creating a "pull" between them.

Seinundzeit said...

David,

Thanks!

I'll include him, we'll see how the results change.

Seinundzeit said...

Yeah, the Dinka sample really allows for much better fits, and cleaner models. But the proportions of West Eurasian admixture aren't much different, although now I'm seeing some Levant_Neolithic.

Again, same setup I always use, plus Dinka sample:

Somali

50.35% Dinka
41.15% Israel_Natufian
5.3% Iran_Neolithic
2.80% Esan_Nigeria
0.25% Levant_Neolithic
0.10% Biaka
0.05% Yoruba
Distance=0.1678

Ethiopian_Oromo

50.65% Dinka
37.15% Israel_Natufian
4.80% Levant_Neolithic
4.15% Iran_Neolithic
3.10% Biaka
Distance=0.2583

So, the Somali and Oromo are now closer to 45% West Eurasian, while previously they were around 50% West Eurasian. Not much of a difference, and these Dinka-based estimates probably match those found in the paper.

Unknown said...

If V88 (aka R1c aka r1b1c aka R1b1a2) was in Chad 5,700–7,300 years ago.

Then V88 was in NorthAfrica/Med (Tunisia or Egypt) in 3,700 BCE (at least) and possibly as early as 5,300 BCE).

This predates the Aegean Bronze age which did not start until 3,200 BCE.

So R1b was in the Mediterranean/North Africa in the neolithic or earlier. This is actually a very important fact. Clearly R1b did not arrive in Europe in the Bronze age, it was already in some part of North Africa by then. We now have three pieces of evidence putting R1b in the mediterranean. Villabruna in 12000BCE Italy, El Trocs in Spain and the Chadean R1b data.

I had always assumed Central African V88 was a late mutation/expansion of a relict haplogroup and as a result meaningless in terms of the R1b main picture. It is starting to look like potential evidence for the timing and path of the main R1b expansion into Europe.

Its also a reminder that we know a lot LESS than we think we do about ancient European DNA. There could be multiple populations in isolated valleys and lakes that we don't know about yet. And much of the ancient coast is under water.

Really all we know is that:

14,000BCE (Magdalenian) R1b is in Siberian Russia (Central Asia)
12,000BCE (Epigravettian) R1b1a is on the northern Mediterranean (Villabruna).
5,500BCE (Mesolithic) R1b is on the eastern Volga (Samara)
5,000BCE (Neolithic) R1b1c in Spain (El Trocs, V88)
3,700BCE(Neolithic) R1b1c is on the southern Mediterranean coast (V88 expansion).

Absence of proof is not proof of absence but its worth noting that R1b is also completely absent from the many samples from the Near East neolithic so we can assume it was not there in the Paleolithic or mesolithic. Probably.

Paleolithic Central Asian men (1/1)
Paleolithic northern European men (0/0)
Paleolithic eastern European men (0/10) A mix.
Paleolithic western European men (0/5) All I.
Paleolithic southern European men (1/2) No Iberians or Greeks tested.
Paleolithic North African men (0/0)

Conclusion: R1b is not likely to be in Eastern Europe but it is a paleolithic European haplogroup (somewhere.... most probably south). It could be in Central Asia or North Africa also. At this stage we really dont have the evidence to be sure of anything.

Unknown said...

Mesolithic Central Asia (0/0)
Mesolithic northern European men (0/6) All I2
Mesolithic eastern European men (1/11) Mostly R1a. R1b is very eastern.
Mesolithic western European men (0/4) Mostly I.
Mesolithic southern European men (0/1) No Italians or Greeks tested.
Mesolithic North African men (0/0)

Conclusion: Unlikely to be common in Eastern Europeans but could be just about anywhere else.

Gioiello said...

@ Annie Mouse

Do you remember when I called you "Annina Topo"?
Thank you. I think having brought infinite proofs in these last ten years that R1b1 (except L389- which was in Central Asia/Caucasus/India) was in the Italian Refugium.

Unknown said...

@Gioello

:) I still lurk even if I dont talk much.

Ric Hern said...

When did the domestication of cattle spread into North Africa ? Which people can be connected to this spread ? Cattle were domesticated in the Taurus Mountains.This was about a thousand years before the start of the Sahara Subpluvial...Were R1b(V88) people responsible for this spread from the Taurus ?

Ric Hern said...

Some Y-DNA from Tenerians and Kiffians could throw some light on the subject...

Ric Hern said...

At what time did domesticoated cattle appear on the Islands around Italy ? Are there any DNA evidence that shows a connection between African cattle and Italian cattle ? If it spread from there, how did they manage to transport cattle over the seas at such an early date ?

Ric Hern said...

If R1b(V88) were connected to the spread of domesticated cattle into Africa then there is clearly a problem with the spread from Italy. Why ? Because the Islands around Italy only show evidence of the introduction of cattle during the Middle Neolithic between 5000 and 4000 bC. while in Southern Egypt we see domesticated cattle at 7000 bC. with genetic connections to the Lavant and Taurus Mountains where cattle were first domesticated.