search this blog

Thursday, September 21, 2017

Ancient genomes from Neolithic North Africa (Fregel et al. 2017 preprint)

Over at bioRxiv at this LINK. The paper includes three ancient North African Y-haplogroup results: two instances of E-M35 from the Early Neolithic (5300-4800 BCE) and a singleton T-M184 from the Late Neolithic (3780-3650 BCE). Emphasis is mine:

Abstract: One of the greatest transitions in the human story was the change from hunter-gatherer to farmer. How farming traditions expanded from their birthplace in the Fertile Crescent has always been a matter of contention. Two models were proposed, one involving the movement of people and the other based on the transmission of ideas. Over the last decade, paleogenomics has been instrumental in settling long-disputed archaeological questions, including those surrounding the Neolithic revolution. Compared to the extensive genetic work done on Europe and the Near East, the Neolithic transition in North Africa, including the Maghreb, remains largely uncharacterized. Archaeological evidence suggests this process may have happened through an in situ development from Epipaleolithic communities, or by demic diffusion from the Eastern Mediterranean shores or Iberia. In fact, Neolithic pottery in North Africa strongly resembles that of European cultures like Cardial and Andalusian Early Neolithic, the southern-most early farmer culture from Iberia. Here, we present the first analysis of individuals' genome sequences from early and late Neolithic sites in Morocco, as well as Andalusian Early Neolithic individuals. We show that Early Neolithic Moroccans are distinct from any other reported ancient individuals and possess an endemic element retained in present-day Maghrebi populations, indicating long-term genetic continuity in the region. Among ancient populations, early Neolithic Moroccans share affinities with Levantine Natufian hunter-gatherers (~9,000 BCE) and Pre-Pottery Neolithic farmers (~6,500 BCE). Late Neolithic (~3,000 BCE) Moroccan remains, in comparison, share an Iberian component of a prominent European-wide demic expansion, supporting theories of trans-Gibraltar gene flow. Finally, the Andalusian Early Neolithic samples share the same genetic composition as the Cardial Mediterranean Neolithic culture that reached Iberia ~5,500 BCE. The cultural and genetic similarities of the Iberian Neolithic cultures with that of North African Neolithic sites further reinforce the model of an Iberian intrusion into the Maghreb.

Fregel et al., Neolithization of North Africa involved the migration of people from both the Levant and Europe, bioRxiv, Posted September 21, 2017, doi:


George Okromchedlishvili said...

So basically heavily Basal folks mixed with LN farmers? Seems legit

Ric Hern said...

Wonder if they used boats made of reeds ?

Ric Hern said...

Interesting no R1b (V88)....

Steven said...

From the article: "FST values indicate that the IAM samples are as differentiated from all other populations as Yoruba are from non-Africans (Supplementary Note 9), with the sole exception of KEB and, to a lesser extent, modern North African populations." This is confusing to me. How can they be as differentiated from other populations as Yoruba, but cluster with groups from the Near East?

Shaikorth said...

On a PCA (and ADMIXTURE) the early Neolithic samples (IAM) have a tendency to look more like modern Mozabites and other North Africans than late Neolithics (KEB) or Natufians.

EN is all mtDNA U6, LN has more European Neolithic mtDNA profile (X2b,T2b,K1a). On the Y side, one EN sample is E-M35, one LN T-M70.

They note that:
"Although ADMIXTURE analysis pointed to some relationship between IAM and Levantine aDNA samples, especially the Natufians, this is not supported by Fst-distances."

Fst-distances may be elevated by low quality and drift, especially for IAM. They note that its closest modern relative is Saharawi by Fst (Fig S9.1)

Davidski said...


Fst is crazy like that sometimes because of various factors, like recent demographic events. Formal stats (D-stats, F4-stats etc.) are more robust in comparison.

Anthro Survey said...

Regarding PCA and clustering: this is a 2D PCA and it doesn't capture all of the important variance. Should one attempt to model Mozabites on nMonte using Global10 data, the best fits, which use Natufians(not Levant_Neo), are still going to be abysmal.

I hope Piranha reads this blog entry and this post. I diagrammed out his "para-Eurasian" hypothesis which is quite relevant here and seems to be supported by this paper(from what I skimmed through thus far).

Chad Rohlfsen said...

More ancient Africans from Skoglund et al (2017)

Yes, don't pay much mind to F_st. Other stats are more important and find issues in F_st.

Chad Rohlfsen said...

Blood-types of ancients too..

AB+ here

Lathan The Great said...

They were closest to modern North Africans, and then Horn Africans.

They are more East African admixed than modern North Africans, and have no Basal Human admix. admixtures

The E-m35 is E-m81. This supports a heavily East African and Basal proto-AA.

Samuel Andrews said...

The Early Neolithic Morocco (IAM) specific component is super interesting. Natufian scores in it along with an Anatolia Neolithic component. Could that mean Natufian is a mixture of something IAM-like and Anatolia Neolithic-like.

Modern North Africans score mostly in the IAM component and cluster with IAM. But did IAM have Sub Saharan admixture like modern NW Africans?

Samuel Andrews said...

I expected mHG M1 to be derived from a Natufian-type population. Not surprised to see it in EN Morocco.

Samuel Andrews said...

Lots of new EEF-derived K1a samples in EN Spain and LN Morocco. All belong to subclades that have already been documented in Neolithic Spain and Britain and North Bell Beaker. Also, all of them are common in modern Western European mitogenomes. Not many European mitogenomes from outside of Western Europe have been published so I don't know if they also exist in Eastern Europe.

K1a4a1: EN Iberia, LN Morocco.
Has already been found in Neolithic Croatia (Cardiel), Spain, Britain, and North Bell Beaker. Today K1a4a1 is European-specific. It has a Northwest-specfic cluster; K1a4a1a2.

K1a1b1: LN Morocco.
Found all over Neolithic Western Europe and in North Bell Beaker. European-specific. Btw, Ashkenazi Jewish-specific popped up in Early Neolithic Hungary which confirms a European origin.

K1a2a: EN Iberia.
Found all obver Neolithic Western Europe. European-species.

Chad Rohlfsen said...

AIM are pretty much between Natufians and Anatolians on a West Eurasian PCA.

Chad Rohlfsen said...

Are you high? Nothing of the sorts.

Anthro Survey said...


I'm betting that Natufians can be modeled as a mixture of smth IAM-like and Levant_Neo-like. This "para-Eurasian" ancestral stream---more basal to BE---can explain their pseudo-SSA scores on various calculators and a slight shift towards SSA on PCAs even though f4s outright reject any "proper" SSA admixture when modern SSAs are used as Test pops. See the red branch in my diagram.

Indeed, it is unclear from their ADMIXTURE whether IAM had "proper" SSA admixture or this hypothetical "para-Eurasian" stuff instead. It could be a mixture of both if we consider the green sahara period. Frankly, I'm surprised they didn't do any f4s with modern SSAs in this paper!


So, if this para-Eurasian thing is real, where does this put the Ancient Egyptians? Can this explain some of their (depicted) phenotypic affinities to "proper" SSAs?

Samuel Andrews said...

Diverse array of mHG U6 in Early Neolithic Morocco: U6a1b (6 private mutations), U6a3 (4 private mutations), U6a7b. Confirms an old presence of U6 in Morocco. Four of five IAMs belonged to U6, one belonged to pre-M1b1.

mHG U6 looks like it originated in NW Africa similar to U5 in Europe. So maybe the Early Neolithic Moroccans have a lot of decent from pre-Neolithic NW Africans?

Davidski said...

Don't forget that mtDNA U6 has been found in Upper Paleolithic remains from the Balkans.

Samuel Andrews said...

@Anthro Survey,

What is Para-Eurasian? Anyways IMO Basal Eurasian was an important factor in Mesolithic West Eurasia. It looks like Iran Neolithic, IAM, Natufian have loads of Basal stuff while CHG and EEF have more Euro-HG type stuff. The "Southern" pull in the West Eurasian PCA is caused by Basal Eurasian.

We talk a lot about EEF, CHG, WHG, bla bla but maybe modern West Eurasians are better defined as a two-way mixture between something like Paleo Euros and Basal Eurasians.

Samuel Andrews said...


That U6 is so old that it's almost irrelevant. Yeah, it could be used as evidence mHG U6's phylogenetic evolution occurred mostly in Eurasia but it's nothing conclusive.

Davidski said...

Pretty sure it debunks any notion that U6 is from North Africa.

Anthro Survey said...

@Samuel Andrews

Again, it's a hypothetical branch proposed by Piranha. See my diagram and the relative drift lengths therein. Red branch.

Essentially, it's a proposed lineage basal to any Basal Eurasians, but upstream of "proper" African lineages.

Natufians could be a mix of this para-Eurasian component, Basal Eurasian and Euro-HG stuff, whereas Anatolian and Levant_Neo lack it. At any rate, CHGs appear to have more BE than Anatolian Neolithics.

The SOUTHERN pull is indeed caused by BE admixture and that's why CHGs/Iran and Levant_Neo are to the south of HGs. But what causes the extra "westward" pull in Natufians seen in West Eurasian and World PCAs(towards SSA in Natufians) alike? I'm betting it's a more basal lineage and the software attempts to deal with this discrepancy by pulling it SSA-ward to account for that.

Samuel Andrews said...

I have considered mHG U6 arrived in Northwest Africa in the Neolithic. It was because if mHG U6 had been in NW Africa since the Paleolithic, like U5 in Europe, one would expect there to be a highly differentiated form of Eurasian ancestry in NW Africa that takes them off the Natufian-Anatolia Neolithic scale.

NW Africans look like an import from the Neolithic Near East with SSA admixture, while Europeans clearly have ancestry from a pre-Neolithic Eurasian population.

alexfritz said...

@Ric Herm
app the EEF is from cardial farmers who supp spread via maritime routes; so they must have had some type boats, prob even more sophisticated than reeds;

Anthro Survey said...

Frankly, I'm surprised they didn't do any f4s to examine allele sharing of IAMs with modern SSAs in this paper in the form (IAM, Eurasian;SSA;Chimp)!!!

--If it turned out that IAMs do NOT share more alleles with than other Neolithic Eurasians do----as was the case with Natufians, mind you-----that could point more clearly to a para-Eurasian lineage.
--Assuming the results do turn out to be such, subsequent f4s in the form (Natufian, other Eurasian;IAM,Chimp) would be definitive in refuting/supporting para-Eurasian notion.

Where might such a "para-Eurasian" branch have resided at one time? Someplace around the Red Sea and chock full of E-m35.

Ric Hern said...

@ alexfritz

Thanks alexfritz.

bellbeakerblogger said...
This comment has been removed by the author.
Roy King said...

Beautiful paper! It redeems my friend and colleague, Peter Underhill, in his analysis of Y chromosomal data. The Early Neolithic, "paracardial", Moroccans autosomally resemble modern Moroccans and appear to be L19(xM81) in their Y chromosomes and are likely indigenous adopters of the Neolithic package--cardial pottery along with continued hunting. Also they seem to be dark-complexioned. Later, around 3000 or so BCE, an Iberian Cardial Neolithic group influenced the Moroccans bringing T-M70 and perhaps G-P303 to Morocco as well as a lighter complexion. Is this part of the dessication event of the Sahara? Do they petroglyphs of North Africa reflect these movements? In any case, this is a monumental paper that fills in the archaeogenetic gaps in the ancient DNA research!

Lathan The Great said...

Chad, can you see what the em35 is? I may be wrong haha, someone said it was e-m81. I really don't know.

The East African may be less, this para-eurasian needs sorting so we can confirm it was EA.

What are your thoughts on this?


Lathan The Great said...

You may be right. This para-Eurasian is very interesting

Samuel Andrews said...

@Roy King,
"Early Neolithic, "paracardial", Moroccans autosomally resemble modern Moroccans"

They are definitely more similar to NW Africans than anyone else but I'm sure modern NW Africans carry significant differences. I am surprised by the high level of Neolithic continuum in NW Africa though.

Lathan The Great said...

I agree. U6, M1. And T1 all have this Balkan-Levant distribution. Awale was right that the Eurasian pre-ETS migration ancestry of Horn is Anatolian-Levantine.

Anthro Survey said...


Indeed that needs to be teased out. The similar clustering of modern Mozabites and IAM on a 2D PCA may owe itself to different reasons: para-Eurasian pulling the IAMs out, while both para-Eu and "proper" SSA pulling the Mozabites out.

See, it may well be that Mozabites are KED+later waves of "proper" SSA. As you know, KED=IAM+EEF_like, from what we can gather in this paper.

Anthro Survey said...

This is essentially how I believe things went down in terms ancestry contributions across the western Eucumene. Arrow weight is a crude indication of relative proportions. The hypothetical para eurasians are in yellow. Basal Eurasians circa the "gulf oasis" are purple. Blue=Villabruna/UHG. Green=ANE-like.

Cladogram on the bottom right for reference.

Salden said...

Roy King:

>Also they seem to be dark-complexioned.

NW African populations are significantly darker than European ones. The Euro style light haired and light skinned examples you see on the net are cherrypicked from coastal areas and cities.

Samuel Andrews:

>They are definitely more similar to NW Africans than anyone else but I'm sure modern NW Africans carry significant differences. I am surprised by the high level of Neolithic continuum in NW Africa though.

The major differences come down to modern NW Africans having admixture from the Arabian Penisula and West Africa. Earlier invasions in recorded history from Phoenicia to Rome didn't have so much of an impact.

JohnP said...

From the paper:
1) an autochthonous Maghrebi component related to a back migration to Africa ~12,000 years ago from the Levant;
2) a Middle Eastern component probably associated with the Arab conquest;
3) a sub-Saharan component derived from trans-Saharan migrations;
4) a European component that has been linked to recent historic movements.

So, let me see if I understood it:
1. The Para-Basal Northwest African.
2. Enter EN Iberians and mix, close to 50/50.
3. Stays like that until the Arab Invasions and Islamic Slave Trade, they mix (which I found shocking, I thought that at least in Carthage times there were already stronger interactions with SSA, specially concerning the hiring of Mercenaries and Slaves, but it seems not).
4. Today's North Africans.

Also, it's weird that the older IAM (-5000) seems to present some considerable SSA admixture in lower K, but not quite in the newer KEB (-3000), which begs the question if the IAM-samples were actually outliers, genetic dead-ends or both.
Also the PCA figures seems to be lacking dimensions, as two genetically different populations (albeit close) which are the KEB and Mozabites are together.

Ric Hern said...

@ Chad

Very interesting especially where stated that these Hunter Gatherers where living between the Equator and the Southern tip of Africa until a 1400 years ago.

Wonder what implications this have for the Bantu expansion especially the timeframe ?

mickeydodds1 said...

Was there any significant SSA component in the ancient samples?

Ryan said...

@Chad - thanks for posting that article. Really neat.

The deepest diversifications of African lineages were complex, involving either repeated gene flow among geographically disparate groups or a lineage more deeply diverging than that of the San contributing more to some western African populations than to others.

The latter would explain a few things, like halogroup A00, that modern human skull from 300 kya in NW Africa, and the modern human admixture in Altai Neanderthals.

Anthro Survey said...


There was likely no "proper" SSA admixture during Roman times as one key ingredient required to sail across the sands of the Sahara was missing: the dromedary. Arab conquest changed all of that.

Anyway, pretty much how I understood it, too. IAM does show a sliver of smth WHG-like at K=3, though, so evidently they were not entirely para/basal Eurasian.

The question is where it came from if it's real. There are three possible scenarios:
1. A local para/basal-Eurasian population mixes with Natufian-like Levantine migrants and acquires it from them. Ibero-Maurisians do not play a role and/or are a dead end.
2. Ibero-Maurisian WHGs mix with a para/basal-Eurasian population.
3. WHG-like signal stems from BOTH Ibero-Maurisians and Natufian-like migrants coming from the West.

It was poor scientific rigour on the writers' part to assume that affinity to Natufians necesserily stems from a Levantine migration when, in fact,a cladistically para/basal population but nonetheless geographically native to East(?) Africa could have been the culprit.

Anthro Survey said...

@Mickey and Roy King

We don't know for sure, but I doubt it. The proper f4s to at least get a hint were not run.

In the ADMIXTURE run from the supplement, IAM, Mozabites, and a handful of MENA populations share a signal with Yoruba, Mandeka, etc. at K=2. In fact, Mozabites and IAM show a comparable magnitude. We should be careful, though, because ADMIXTURE may just be lumping some para-Eurasian component together with more basal "proper" SSA lineages. The mtDNA does not seem to indicate any proper SSA admixture in IAM, but it does in contemporary Maghrebi populations.

For this same reason, we shouldn't assume undisturbed continuity from Neolithic times to the present just by looking at a 2D PCA and ADMIXTURE. The "red" in Mozabites may consist of both para-Eurasian and later SSA waves, while in the IAM only the former. A KED-like population was the intermediary, evidently.

Karl_K said...

"The research similarly shed light on the origins of another unique group, the Hadza people of East Africa.

"They have a distinct appearance, language and genetics, and some people speculated that, like the Khoe-San, they might represent a very early diverging group from other African populations," said Reich. "Our study shows that instead, they're somehow in the middle of everything."

The Hadza, according to genomic comparisons, are today more closely related to non-Africans than to other Africans. The researchers hypothesize that the Hadza are direct descendants of the group that migrated out of Africa, and possibly spread within Africa as well, after about 50,000 years ago.

Another discovery lay in wait in East Africa.

Scientists had predicted the existence of an ancient population based on the observation that present-day people in southern Africa share ancestry with people in the Near East. The 3,000-year-old remains of a young girl in Tanzania provided the missing evidence.

Reich and colleagues suspect that the girl belonged to a herding population that contributed significant ancestry to present-day people from Ethiopia and Somalia down to South Africa. The ancient population was about one-third Eurasian, and the researchers were able to further pinpoint that ancestry to the Levant region.

"With this sample in hand, we can now say more about who these people were," said Skoglund.

The finding put one mystery to rest while raising another: Present-day people in the Horn of Africa have additional Near Eastern ancestry that can't be explained by the group to which the young girl belonged."

Karl_K said...

"The new study also found that West Africans can trace their lineage back to a human ancestor that may have split off from other African populations even earlier than the Khoe-San."

Creative said...

@Anthro Survey
A note on the dromedary as a resilient beast of burden . Assyrian's, Persians and Greeks alike used Arab mercenaries/guides in there invasion of Egypt, only meaning that Arabs were already familiar/active in the region.
The one-humped camel or dromedary (camelus dromedarius) is already sporadically attested in the Early Dynastic Period, but it was not regularly used until much later. Foreign conquerors (Assyrians, Persians, Alexander the Great) brought the camel on a greater scale to Egypt. Certainly in the Ptolemaic Period, and perhaps already under the Persians (525-343 BC), the camel (also the two-humped camel, camelus bactrianus) was used as main transport animal for the desert.

Annie Mouse said...

IAM and KEB look so different (culture, haplogroups). It is only the autosomes that suggest a connection.

Form the Supplementary Note there is a link between IAM and Cyprus in shared burials with a grind stone on the face. I wonder if they were trying to weigh down the ghosts?

I can see a Cypriot neolithilizing along the North African coast but I am not sure what a Mahgrebi would be doing in Cyprus. Substratum maybe?

Annie Mouse said...

IAM and KEB look so different (culture, haplogroups). It is only the autosomes that suggest a connection.

From the Supplementary Note there is a link between IAM and Cyprus in shared burials with a grind stone on the face. I wonder if they were trying to weigh down the ghosts?

I can see a Cypriot neolithilizing along the North African coast but I am not sure what a Mahgrebi would be doing in Cyprus. Substratum maybe?

Anthro Survey said...


It was certainly known before the Islamic era, but only gained enough popularity afterwards to be used en masse for trans-Saharan trade among the Maghrebians.

epoch2013 said...


"WHG-like signal stems from BOTH Ibero-Maurisians and Natufian-like migrants coming from the West."

Couldn't that signal be the same as the Middle-Eastern signal in the Villabruna cluster and El Miron (!) as found in Fu 2016? You can find that in Natufians and Anatolian neolithics.

Rob said...

Nice work . Still chewing through
Can wait for even older stuff from North Africa
I suspect Iberomaurisians to be very similar to Natufians, with a distinct break from
Perhaps more near eastern admixture began with late Mesolithic Capsian , neolithic, etc

Matt said...

@Steven, Fst is a really important stat and pretty easy to understand, here's a primer on what Fst is by Razib Khan:

Basically Fst is showing the proportion of genetic variation which is partitioned between groups. Intuitively it is easy to understand, in that if ~1% of the genetic variation is partitioned between groups in one case, and ~10% in another, then it is reasonable to suppose that the genetic distance between groups in the second case is larger than in the first case.

On a continental scale Fst between populations is often on the order of ~0.10. That is the value for example when you pool the variation amongst Northern Europeans and Chinese, and assess how much of it can be apportioned in a manner which differentiates populations (so it’s about ~10% of the variation).

The more serial bottlenecks a population goes through, the higher this value will tend to become, all thing being equal. Basically, as within group variation goes down, between group will tend to explain more of the variance.

So in the Americas, you have groups who show extremely high Fst - (" Seri and Lacandon show the highest level of population differentiation as measured with FST (0.136 ), higher than the FST between Europeans and Chinese populations in HapMap3 (0.11) ") - as do populations in Highland New Guinea -

The merging of populations after the go through bottlenecks also tends to reduce this figure, as e.g. comparing ancient with much more admixed modern populations.

When they say that IAM are as "IAM samples are as differentiated from all other populations as Yoruba are from non-Africans", they about the same proportion of variation is "between population" when "Yoruba+Non-Africa" is pooled as when "IAM+otherPopulations" is pooled.

This doesn't mean that they didn't go through an Out of Africa bottleneck with other Out of Africa populations (post Mota divergence, pre Eurasian divergence).

Likewise, talk of IAM clustering with Near East simply means that this is where they fit when forced to, by being projected onto a modern PCA, or via a clustering algorithm.

The formal f and d stats are an attempt to get around the "problems" for Fst of increasing genetic differentiation between groups with this serial bottleneck effect.

However, they do introduce very unintuitive results:

- Europeans modern and post Bronze Age ancients tending to be closer to Lithuanians than each other, unlike with Fst (e.g. Central Bell Beakers closest to Lithuanians by f3 sharing, rather than the NW Europeans they actually resemble in clustering and PCA and which fst finds closest to them)
- ancient Egyptians closest to Sardinians, where the more intuitive result from Fst is being closest to... Egyptians,
- Levantine populations being closest to Europeans... where Fst finds they're closest to other Levantines

The latter are particularly why Fst estimators are having a resurgence and becoming more prominent in these papers that are relating to the Mycenaeans and ancient Egyptians and so on.

This is probably because any deep rooted small intrusion tend to heavily disrupt f3 sharing, and low within group genetic diversity tends to increase f3 sharing.

Matt said...

@Steven, anyway, back to Fst, a question is, does IAM show particularly high Fst from West Eurasian groups only because of some kind of ultra bottleneck?

If that were the case, it would have high Fst from West Eurasian groups, but also *really* high Fst from African outgroups.

There's a full Fst table in the supplements (page 70) and what it shows is:

Mbuti-IAM: 0.268, Mbuti-WHG: 0.265, Mbuti-Natufian: 0.246, Mbuti-Han: 0.242, Mbuti-Europe_EN: 0.231, Mbuti-Iranian: 0.206, Mbuti-Moroccan: 0.162

Yoruba-IAM: 0.191, Yoruba-WHG: 0.201, Yoruba-Natufian: 0.183, Yoruba-Han: 0.176, Yoruba-Europe_EN: 0.162, Yoruba-Iranian: 0.138, Yoruba-Moroccan: 0.088

IAM-WHG: 0.224, IAM-Han: 0.223, IAM-Natufian: 0.210, IAM-Europe_EN: 0.175, IAM-Iranian: 0.164, IAM-Moroccan: 0.117

Natufian-WHG: 0.146, Natufian-Han: 0.167, Natufian-Europe_EN: 0.079, Natufian-Iranian: 0.085, Natufian-Moroccan: 0.071

So we can see that, relative to the Natufians, it does appear that IAM is *far* more distinct from all Eurasians, particularly West Eurasians, while not being much more differentiated from African groups (as much as they are among the most differentiated populations in this panel from Africans)....
Formal stats would be a good complement for this though.

This also shows that while Natufian are still clearly closer to WHG than Han, this is much reduced in IAM. Looking at all the distances, I suspect IAM still have some West Eurasian admixture, but they're closer to a "basal Eurasian" than Natufians are.

Matt said...

Using those Fst distances from the paper's supplement, here is PCoA of just the "West Eurasian" plot samples (albeit this means not using the North and other Africans, even though they're pretty important in this context):

1: more or less a ancient Near East+North African against ancient Europe dimension
2: more or less IAM against Levant_N
3: classic WHG->EHG and Natufian->Iran_N dimension

1+3 are approximately a "classic" West Eurasian plot.

Only thing that really sticks out as slightly odd about these scores is that TOR is pretty typically Iberia_MN as far as the paper is concerned, while these Fst estimates place it relatively closer to IAM than Iberia_MN is, and with some differentiation from all other populations...

Shaikorth said...

@Anthro Survey

That kind of simple and direct D-stat comparison often produces inconclusive results when Africans are concerned. For example Loschbour could seem as Mbuti-related as Yoruba or French as Ju_Hoan-related as Somalis:

Chimp Mbuti Loschbour Yoruba -0.0035 Z -1.200
Chimp Ju_hoan_North Somali French 0.0014 Z 0.914

Maybe they could test haplotypes, or build an ADMIXTUREGRAPH model which might settle it with ghost branches.

Looking at relative fst's with that drift issue in mind:

IAM to WHG/ IAM to Yoruba = 1.17
IAM to Han/IAM to Yoruba = 1.17

Natufian to WHG/Natufian to Yoruba = 0.8
Natufian to Han/Natufian to Yoruba = 0.92

Saharawi to WHG/Saharawi to Yoruba = 1.04
Saharawi to Han/Saharawi to Yoruba = 1.12

BedouinB to WHG/BedouinB to Yoruba = 0.76
BedouinB to Han/BedouinB to Yoruba = 0.82

Modern Moroccans



Some unrelated:


Han to WHG/Han to Yoruba

Ric Hern said...

@ Karl K

When comparing the earlier paper
With this one it is interesting to note that the 3000 year old remains could point to an admixture which was responsible for the creation of the Khoi-Khoi only 2000 years later (1000 AD) if I'm not mistaken. This admixture also do not show any Bantu admixture in the Early Khoi-Khoi.

So it looks like people similar to the 3000 year remains hung out between Tanzania, Zimbabwe and Northern Mozambique for at least 2000 years.

And it looks as if Bantu migration to the South from around +-Uganda only happened around 600 AD when they started crossing the Equator....

This surely does not look like a fast migration since they only reached the Northern Parts of Southern Africa (Zimbabwe) around the 12/1300s....

ak2014b said...

"More ancient Africans from Skoglund et al (2017)"

The drought is over. We're finally getting ancient genomes from nearly unsampled places.

I'm starting with this "Reconstructing Prehistoric African Population Structure" (Skoglund) and will read the Neolithic North Africa paper afterward. However, are both papers to be discussed on this blog page or will the Skoglund paper get its own separate discussion page, David?

Figure 1 is a map placing the study's 16 ancient African genomes. It looks like the samples are from east and south Africa. Is a separate paper on western and central African ancient genomes expected as well, or not on the horizon for the immediate future?

Ric Hern said...

Could we now conclude that R1b (V88) did not spread into Africa via Iberia or is it still too early to say ?

Gioiello said...

Ya Kem , unfortunately my PC has been out these last two days and only now I have it. I followed this paper on”Eurogenes blog” from my wife’s PC, but only sporadicly, but who is this stupid person? “Kisse Ellis • The first Europeans were Blacks who entered via Iberia- Gibraltar and crossed the Mediterranean - island hopping. Its rather obvious if you look at the map. But centuries of propaganda obscure the issue”. From what I read on “Eurogenes” it seems that there has been a migration from Middle East (Natufians) and also from Southern Europe, and certainly Old Egyptians were more similar to Milddle Easterners and Europeans than SSA, only present recently, as in the Maghreb, who seemed in this aDNA more similar to Europeans than now. Someone spoke of an mt K1a1b1, which is ancestor of my K1a1b1e, and my haplogroup is was born certainly in Tuscany, without any doubt. Of course it is possible that E-L19 was present in Maghreb, we’ll see from its terminal SNP at which level to put it.

Slumbery said...

Skoglund et al. says that virtually all of the ancient Eurasian ancestry in present day East and South Africans is of Levantine origin (under which they seem to mean Natufian) and then use Sardinian as reference population for Eurasian ancestry in their admixture graph in Figure 1?
This seems to be a bad choice.
Not a big problem, since their ancestry proportions in the discussion of the results do not seem to be based on that analysis, so that graph is pretty much is just decoration, but still.

Ryan said...

@ak2014b - I'd think the climate of Western and Central Africa will make those genomes more challenging to recover.

capra internetensis said...

@Ric Hern

I'm not following you. The c 800 AD South African in this study is from the Western Cape region, which was in the hands of Khoikhoi pastoralists until after Europeans showed up. The most recent of the KwaZulu Natal coast forager samples from the previous paper dated to around 100 AD, where the Early Iron Age (very probably Bantu) arrived around 400 AD. The latest Malawian sample is from around 500 BC, and the Early Iron Age arrived there around 200 AD IIRC.

So how are you getting this crossing into Tanzania in 600 AD and arriving in Zimbabwe in 1200 AD stuff? Who was farming sorghum and millet and smelting iron for the previous thousand years?

Samuel Andrews said...


Btw, mHG U6 in Paleolithic Europe and IAM means IAM definitely have some "West Eurasian" aka non-basal ancestry.

Ric Hern said...

@ Frank_K

And who did that Near Eastern people mix with ? Hadza like people ?

AWood said...

@Roy King

I agree. I suspect if we look at the certain Y lineages identified as neolithic in age in Sardinia (branches of ie: G2, T1, R1b-V88), they should be a close match to those who moved to Africa at that time.

Rob said...

@ SLumberry

It also mentions a later Iran-Neol. related component in Omotic speakers, does it not

Ric Hern said...

@ capra internetensis

Unless my reasoning failed me completely that is what I understood reading the Schlebush Paper....

As I understand it the Khoi-San were the first inhabitants of Southern Africa and then came the Khoi-Khoi. All Khoi-Khoi carry admixture from a Near Eastern source which is apparently spread rather even...

This Near Eastern admixture apparently happened plus/minus a thousand years ago with no visible Bantu admixture in the picture. Where did the Khoi-Khoi live during this time ?

The Khoi-Khoi was spread over a wide area from the Cape into the Free State and all the way into Namibia during historic times....etc.

The only place I can see for this admixture to have happened is in Zimbabwe from where the spread out....etc.

Matt said...

@ Shaikorth,

1) to be honest, I still really don't think this stuff with Fsts divided by other Fsts to get ratios is a good idea. If we want to adjust Fsts for drift levels, I really do think we should just be doing subtraction or addition. Sorry. Though this said, what do you think the differences you see show?

2) That said, the way I think of it:
(WHG to Mbuti) - (WHG to IAM) = 0.265 - 0.224 = 0.041
(Han to Mbuti) - (Han to IAM) = 0.242 - 0.223 = 0.019

Excess to WHG 0.022. Therefore IAM is probably at least slightly closer to WHG, net of WHG's level of drift.

This is a lot more even than Natufian:
(WHG to Mbuti) - (WHG to Natufian) = 0.265 - 0.146 = 0.119
(Han to Mbuti) - (Han to Natufian) = 0.242 - 0.167 = 0.075

Excess to Natufian 0.044

If we repeat with Yoruba in place of Mbuti, we get
(WHG to Yoruba) - (WHG to IAM) = 0.201 - 0.224 = -0.023
(Han to Yoruba) - (Han to IAM) = 0.0176 - 0.223 = -0.047

Excess to WHG 0.024

Pretty stable difference.

Now continuing with Moroccan
(WHG to Mbuti) - (WHG to Moroccan) = 0.265 - 0.091 = 0.174
(Han to Mbuti) - (Han to Moroccan) = 0.242 - 0.099 = 0.075

Excess to Moroccan 0.100

So I'd predict from this that f4(Chimp,X,WHG,Han) would be most negative for Moroccan (recent), then Natufian, then almost not negative at all for IAM. Ust Ishim in place of Han should mirror this pattern. We'll have to wait and see though. To me it looks like "Most Basal Eurasian yet... but not totally, and not necessarily actually the Basal Eurasian ancestors for Natufians, Iran_N, etc".

Matt said...

@Sam, yeah, I would think there would be some "West Eurasian" based on my above reasoning, and agree with you that I would expect U6 to align with that. Though I actually don't know what the divergence time of U6a would be, and whether it could be early "Crown Eurasian" possible, rather than West specific. The conundrum is that I guess I would have thought that the early North Africans like IAM would've been more "West Eurasian" than they were based on my simplistic reading of haplogroup U = European Paleolithic related.

@Slumbery / Rob, yep, there is mention of Ganj_Dareh_Iran_N related ancestry in Somali and Afar in the supplementary table S3 at least, in quite significant proportions. I actually think this may be Bedouin related and be related to some form of Persian Gulf ancestry, likely with some subtle to quite large distinction from Iran_N proper from the early sites at Ganj Dareh.

But this is all interesting stuff - the pendulum for me has swung from looking at Dienekes early calculators showing "SW Asian" (e.g. Arabian) ancestry in East Africa, to being confused when this came out as mainly Sardinian related in later formal analyses using formal stats ("Eh, how does this fit with Globe13, etc.?"), to seeing a bit more making sense now. Looks like multiple waves.

@ All, btw generally on Skoglund 2017, I'd also for extra context worth a revisit to at Busby's 2016 paper on (mainly) TVD distances in Africa and really deep sampling of modern day Africa -

E.g. for one -,1500/0/default.jpg - the Malawi population which this paper finds has no admixture from earlier populations is very tight in TVD and finestruture PCA with West Africa.

There are some differences in scattering and structure in Busby's paper though. The clinal space in the finestructure PCA between East Africa and South Africa looks a hell of a lot fuller than the equivalent PCA in Figure 1 of Skoglund 2017 (where it is empty except for the ancients).

Of note to me in particular is that there is quite a lot of substructure within the West African population. All the Bantu ancestry populations from West-Central to East Africa seem to form a pretty tight cluster, while there's a good amount of deep structure within West Africa itself from the far West to West-Central... This is interesting, but not unexpected - expansion into the rest of Africa by West Africans was by the eastmost West Africans, who were best placed to exploit the advantage over non-food producers. Structure within West Africa probably retains more ancient depth.

capra internetensis said...

@Ric Hern

Oh, you are using their linkage disequilibrium dates. Those are usually on the low side, sometimes way low. Schlebusch et al actually say "Based on these results, we suggest a migration from East Africa into southern Africa, resulting in admixture with local hunter-gatherers ≥1.5 kya" and "this observation is consistent with archaeological evidence for the arrival of migrant Iron Age farmers of West African descent to the eastern parts of southern Africa at ~1.7 kya", so they don't take their own dates too seriously. Evidence of sheep-herding shows up a few centuries before the Iron Age farmers, which may be the signature of the ancestral Khoe pastoralists.

As for where the admixture took place, was necessarily in just one place, but previous studies found that Nama were autosomally closer to southern Khoisan like ǂKhomani than those from further north. That could mean that the mixing was in the south of South Africa. But I'm not sure how far I trust those results.

Shaikorth said...

Those Fst's were basically to compare relatedness to Africans.

If I do something like that with addition and subtraction a rough pattern with similar leanings stays but the relations don't look as good.

Doing with subtraction
Ethiopian_Jew to Yoruba minus Ethiopian_Jew to Natufian = 62-86 = -24
IAM replacing EJ = 191-210 = -19
Saharawi: 98-80 = 18
Algerian: 107-84 = 23
BedouinB: 144-81 = 63
Sardinian: 156-89 = 67
TOR: 181-110 = 71
Armenian: 145-86 = 59

Now the excess "Yoruba relatedness" Ethiopian Jews show compared to IAM looks like that Saharawi shows compared to Algerian (5), but IAM doesn't appear as SSA-shifted as EJ's based on the results so far, while Saharawi and Algerians are both part of the North African metapopulation and less differentiated.

Now the ratios (X to Yoruba/X to Natufian) - EJ and IAM are clearly differentiated from each other using this, while Saharawi and Algerian look more similar to each other.
EJ 0.72
IAM 0.91
Saharawi 1.23
Algerian 1.27
BedouinB 1.78

Ric Hern said...

@ capra internetensis

If the mixing happened in the South of South Africa it would mean one of two things. One would be that those "Cushetic" Pastoralists inhabited most of South Africa before the Bantu migration...or that some Sea Voyage brought Near Eastern people to the South where they could mix...

Anyway if you read the whole paper it looks like there is going to be a lot of dating to be sorted out in the near future compared to the current dating of human remains....

Chad Rohlfsen said...

I'm anxious for an East African older than Mota. I'm very certain, more than ever, that all Africans have Eurasian admixture. Looks pretty inevitable.

Slumbery said...

@Rob&Matt: Yes, I saw the Iran_N part, however Sardinian is not exactly a stellar proxy for that either.
Nevertheless, I have thought about this and understand their choice better now. That ADMIXTURE test just uses Sardinian as a representation of West Eurasian, as it uses Japanese as representation of East Asian. Just to get the general picture. It should work as far as all pre-levantine-migration Sub-Saharan African groups are significantly closer to each other than to any Eurasians.

Chad: But when? Based on this study it had to happen a very long time ago. The genome-wide impact of migrations like the one carrying R1b-V88 is not visible here, also the 12 000 y BP Natufian migration in North Africa (mentioned in Fregel et al.) does not seem to reach even Ethiopia.

Chad Rohlfsen said...

All it will take is an older reference. The migration is likely Iberomarusian with some later Levant ancestry, followed by Cardial.

Samuel Andrews said...


So you think IAM has lots of Iberomarusian stuff? Also, I'm thinking maybe an Anatolia_Neolithic-like pop in the Levant farmed before a Natufian one. Then it migrated to Anatolia and the ones who remained mixed with Natufian hunter gatherers, hence the Anatolia Neolithic/Natufian mix we see in the Levant's first farmers. What do you think of that?

Seinundzeit said...

I think that these early Neolithic Moroccans, along with those foragers from the Balkans, will prove to be of immense interest, when it comes to attempts at explicating the genetic composition of populations on the Natufian/Levant_N/Anatolia_N cline.

I mean, it seems that IAM are mostly BEA, and the Balkan foragers in Mathieson et al. (2017) are the closest thing we've seen yet to the UHG construct.

Kristiina said...

@Chad and Sam

The estimated age for the whole of haplogroup U6 is around 35 kya and the age of Muierii2 in Peştera Muierilor Romania is 33 kya BP. The oldest modern U6 haplotypes are in fact in the west, e.g.
In Maghreb and Europe
U6a7 29000 (IAM x 1),
In Maghreb:
U6a6 21900,
U6a1 18600,
U6a1b 17100 (IAM x 2)
In Maghreb, Europe and West Africa:
U6a3 18800 (IAM x 1)

Also some European haplotypes are old: U6a1 18600, U6a1a 13100 and U6a1a2 16200. Compared to these, East African haplotypes are younger, e.g. U6a2a1 Ethiopia 12700. Therefore, at the moment, the phylogeny and ancient evidence suggest an Iberian route for U6 in Africa. The age of U6 in North Africa coincides with the appearance of backed bladelet industries during Last Glacial Maximum (~23-18 KYA) and the Iberomaurusian culture c. 20 000 kya.

Matt said...

@Shaikorth, the patterns of excess relatedness in your example there, with TOR and Sardinian closest to Natufian relative to using Yoruba as an outgroup to measure drift, and IAM least close, those seem correct, and I would strongly expect them to be confirmed with formal f3 outgroup statistics.

However, the place of IAM on that line is not because it is more SSA shifted; it's because it's an outgroup, apparently to a very significant degrees, which shares little ancestry with Natufians, unlike all the other populations (and most extremely for Ethiopian Jews who have a strong dose of both relatively Natufian related and Yoruba related ancestries).

Let's add Han onto that line using the equivalent figures: 0.176 - 0.167 = 0.009

So Han are "SSA shifted" relative to Saharawi? They're just not very shifted towards Natufians, and unlike other populations don't share recent ancestry.

This stuff will all be checked out with f3 in time, however in my experience of testing comparisons f3 outgroup stats strongly correlate with (Fst Outgroup-to-X)-(Fst Outgroup-to-Y). (Plus subtraction is constant independent of the two African outgroups used, in this example, which is not the case if you used Mbuti in place of Yoruba in the multiplication).

@ Sein: "I mean, it seems that IAM are mostly BEA, and the Balkan foragers in Mathieson et al. (2017) are the closest thing we've seen yet to the UHG construct."

Certainly looks like they have high BEA *of a sort*; though this said I'm not sure the BEA ancestry in this population will actually be any closer to the BEA that is theorised in the ancient Near East than e.g. Onge will be to Kostenki14, and in fact may even be more deeply diverged in time depth. (Not that I necessarily think you're saying that it was).

I think it strengthens the case that a population Basal to All Eurasian groups is likely to have existed somewhere within Northeast Africa, for sure. Should be good to see if Davidski / Chad can add this population to simple qpGraph.

Davidski said...

Not sure if anyone's noticed but those ancient African genomes from the new Skoglund paper are now available at the Reich lab website.

I'll try and do something with them soon. I'm a bit overextended with stuff at the moment.

Shaikorth said...

That fst comparison is not meant to be an general measure of SSA affinity for Eurasians or something, Han lacks the Natufian and Neolithic farmer and associated UHG of Saharawi, Bedouins etc. so they have excess Natufian relatedness compared to it. If we compare to Iran_N, another ancient Near Eastern population but one that has less to do with Northwest Africans:
Ethiopian Jew -12
IAM 14
Saharawi 30
Han 57

I'm not so sure IAM is an outgroup, it's always somewhere between Ethiopian Jews and modern North Africans when comparing Fst to Eurasian population and SSA so the high overall Fst could be the result of the sample's low quality. If we compare distances to Yoruba to distances from Han, which among the Eurasian groups here should be the most neutral to IAM and Natufians (Yoruba/Han ratios in brackets, ratio again more in line with the PCA results):

EJ: -49 (0.56)
IAM: -32 (0.86)
Saharawi: -12 (0.89)
Algerian: -7 (0.96)
KEB: 7 (1.05)
Natufian: 16 (1.1)
BedouinB: 25 (1.21)
TOR: 40 (1.28)

Matt said...

@Shaikorth, well, we'll sort these things out with f4 statistics I guess. My guess would be that IAM is closer to having the majority of its ancestry from an outgroup, either with a position between Mota and the Basal Eurasian in Natufians, or at the same level as Basal Eurasian, and is not otherwise shifted towards Africans under statistics f4(Chimp, African, IAM, Eurasian).

(This is what we're talking about isn't it? Whether there is any specific shared ancestry between IAM and groups from further south in Africa, or whether IAM is just mainly some mix of populations at varying stages in the bottlenecks between Mota and the bulk of Out-of-African ancestry?).

Steven said...

Are any of these up on Gedmatch yet?

Chad Rohlfsen said...

What I see from the West Eurasia PCA is that AIM is less Basal than Natufians, but maybe further from Iranians. AIM does look less Basal than Natufians though. It could be the hunter under the BE is less differentiated to the Western side.

I'll put these samples through the ringer after they come out.

Shaikorth said...

Guess so, although there is the question of what's the point between Mota and Eurasians where a branch no longer is "SSA".

Also, even deeper African ancestry could go undetected in a simple D-stat test unless the SSA reference is right (so there could be a room for ghosts there).

Mozabite Natufian Mota Chimp 0.000312 Z 0.664
Mozabite WHG Mota Chimp -0.000605 Z -2.160
Chimp Ju_hoan_North Somali French 0.0014 Z 0.914
Yoruba reference instead of Mota gives a significant stat for Mozabites and so on. I'd expect Mota to work better in qpgraph.

Roy King said...

I am dismayed by the commentators obsession with SSA and phenotypic skin color. The parsimonious explanation of IAM's link to West Africa is likely due to the wet Sahara phase when there was not the geographic barrier we see today, so that the term Sub-Saharan and its derivative SSA bore little meaning at the time. We have evidence of this continuity with Tishkoff's 2009 study of some autosomal markers that included the Dogon of Mali who actually have a large Moazabite autosomal component, unlike the Mende or Yoruba populations. We should be studying the Tuareg and the Dogon and the Berber populations in comparison to IAM. Later, there is clearly an immigration of Iberian farmers to Morocco which correlates with their changing autosomal position and even later, there is likely an influence historically via slavery or Islam of now Sub-Saharan ancestry in North Africa.

Roy King said...


The Dogon and Bambara of Mali have been studied autosomally and the data are publicly available:
Can you add these data to your analyses?

Davidski said...

Marker overlap with my current datasets is relatively poor.

Roy King said...

Too bad! My hunch is that the IAM Early Neolithic samples will reflect an admixture between the wavy-line pottery of the Western Sudan (mirrored in contemporary Dogon) and the Natufians. The later SSA of Nigeria and Senegal drifted away from these founding lineages of the wavy-line pottery horizon. To do SSA admixture, one should use groups like the Dogon or even samples like Mota rather than Yoruba.

Davidski said...

My impression from this paper is that IAM are not a mixture of Natufians and Sub-Saharan Africans, but rather an early Natufian-related, but highly differentiated, back to Africa population from Eurasia.

Davidski said...


Please don't use the ancestry of the commentators that you're arguing with as part of your arguments.

Gioiello said...

Ya Kem, but if you remember what I wrote also here about the tree of PF2431, it seemed that an old line was present in Northern Africa, but the main line from which also your haplotype descends was present from old times in Europe. Thus, as I said above, it is important to know the terminal SNP of this sample, because it could belong to the "African" line and not to the "European" one. I read the post of Roy King you quoted... but the recent aDNA from Africa seems to demonstrate that there were only hgs. A and B, as I, and many Others, are writing from so long. Even E, and not only R-V88, back migrated to Africa.

Seinundzeit said...


"Certainly looks like they have high BEA *of a sort*; though this said I'm not sure the BEA ancestry in this population will actually be any closer to the BEA that is theorised in the ancient Near East than e.g. Onge will be to Kostenki14, and in fact may even be more deeply diverged in time depth."

An intriguing notion; personally, I very much look forward to further sampling of North Africa/Southwestern Asia, so that we can fill in the spatio-temporal gaps.

"I think it strengthens the case that a population Basal to All Eurasian groups is likely to have existed somewhere within Northeast Africa, for sure... "

I completely agree; these samples seem to be indicative of the possibility that we'll find more representatives of populations in the broader BEA group (as opposed to the Crown Eurasian cluster) in northern and northeastern Africa.

Regardless, I'd love to see models of Natufians/Levant_N/Anatolia_N as mixtures between IAM, Balkan foragers, and CHG/Iran_N.

Of course, neither IAM nor the Balkan foragers are ideal representatives of Near Eastern BEA and UHG, but they are the closest we've seen yet; so it'll be exciting to see what David/Chad can work out, and it'll quite fun to try some things myself with nMonte.

Samuel Andrews said...


Btw, Balkan (almost all Bulgaria) and Hungary Neolithic mtDNA has few significant matches with my modern NW European mitogenomes. I'm begging to think Funell Beaker might turn out to be the biggest EEF contributor to NW Europe. And also, that Globular Amphora is the biggest contributor to Balts and Slavs. ALso, Neolithic Iberian mtDNA has already shown significant matches with modern Iberian mtDNA.

Valfol said...

Perfectly this could be due to a "back migration" into Africa of a Natufian-like population (Eurasian-African hybrid population). Being Y-DNA E as African signal and mtDNA as Eurasian signal.

Natufian as well as IAM are Y-DNA Africans and mtDNA Eurasians.

IAM with Paleolithic mtDNA M (Y-DNA C linked mtDNA) and U6 probably from a influx of a Paleolithic Balkan population into Africa (Y-DNA IJ linked mtDNA)

Bronze said...


nope, both IAM and natufians are euroasian on the paternal line, not african. Y-dna E originated in eurasia and backmigrated into africa and mixed with native SSA women. We know this as we now have two ancient dna samples with y -dna E and with no ssa admixture, and the pcas show both natufians and IAM to be completely unrelated to ssa groups.

Matt said...

@ Chad, would say projecting onto the West Eurasian space could be misleading if this population is as distant as it appears in the World PCA. Projection onto World PCA would seem possibly more informative, as there's a lot more genetic space there to land samples that are well out of range.

Looking at the paper again, there is something of a WHG and Natufian affinities in the outgroup f3 scores, which are present.

There's also the ADMIXTURE in their paper, which fits some North African groups as mostly composed of the IAM component (I'm not sure if they actually included modern day Africans in the same admixture though, which could be a confound here).

Be interesting to see how deep or shallow this is compared to what we see in Natufians. The Fst scores could be heavily exaggerated - I'd hope not as they have seen fit to place these in their preprint.

I suspect we may see a revision of this preprint attempting to answer more questions with formal stats and qpGraph well before anyone else has a chance to analyse these in those terms (everything we've commented on here has probably been gazumped by Reich lab etc members contacting the lead authors already).

(Or it could be released tomorrow!).

@Shaikorth: "what's the point between Mota and Eurasians where a branch no longer is "SSA"? "

Yeah, good question I think this is mostly semantic really. If they are a population which branches just after Mota, don't appear to have gone through quite the same bottleneck as Eurasian populations (even including Basal Eurasian?) and live in Africa, I have no problem simply calling these ancient people Africans, or for precision Ancient North Africans (to emphasise the lack of any deep sharing with ancient East Africans like Mota, or the ancient South Africans and West Africans which exist in theory). Or even "North Saharan Africans" (though technically may not be 100% precise).

In that case, neither "Eurasian" in the sense of "Shorthand for all people who descend from an Out-of-Africa bottleneck (plus Neanderthal ancestry)" or "Actually historically lived in Eurasia" would really fit.

Shaikorth said...


They have West Africans included in ADMIXTURE, figure S7.1. Only lowest K values show differences between non-outlier Mozabites and IAM.

Another look at Fst correlation:

Matt said...

@Shaikorth, thanks, I sort of glossed over the HGDP admixture for the Human Origins panel one, but that is certainly relevant.

Chad Rohlfsen said...

World PCA does odd stuff too, like putting WHG right next to Anatolians, and Natufians displaced towards Africans. There's obviously stuff between Africans and HG, and potentially several things. I really wouldn't call Mota SSA either. He is, without a doubt in my mind, admixed with "Eurasian".

Matt said...

OK, but all the academic stuff (e.g. Skoglund's latest from last week) so far is that Mota simply branched off from a common ancestor with Eurasians later than other Africans did.

That is, Mota is not preferentially admixed with any Eurasian group, therefore either a) mixed with a pre-divergence Eurasian (pre-Basal Eurasian divergence, pre-everything), or b) just split from proto-Eurasian later than other Africans. b) is more parsimonious, so b).

Slumbery said...

YDNA E can be African without being tied to SSA.
I am not saying that this is a theory I prefer. I am just saying generally that something can be African without any connection to SSA, because as far as we know SSA (or the precursors of it) was never dominant in the Mediterranean part of Africa, so the lack of SSA in Natufians is not a decisive evidence on the matter.
Also YDNA E is very old. It is possibly older than anything we could recognize as SSA. It surely went trough a lot of pruning and re-expansion, so the homeland of E and the homeland of the current E lineages might be two entirely different story.

Steven said...

The back to Africa migration theory is absurd. That the entire African continent was overrun by a near eastern group replacing nearly all of the preexisting y haplogroups in Africa is implausible. The topography of the continent would make it difficult to replace the majority of preexisting lineages in SSA. An East African origin of E1b1b is far more likely.

Samuel Andrews said...


Any idea if Mittnik 2017, Mathieson 2017, and Olalde 2017 are combining forces to make a massive paper? Or if all of them are sampling new genomes and plan to publish something in 2030?

Valfol said...

Slumbery, you pointed out fairly well all inconsistencies from Bronze answer.

Anyway IAM is closest to SSA than KEB.

Chad Rohlfsen said...

Only due to lack of older samples there. Mota's day will come. All Africans are admixed. Even Mota. Mark my words.

Bronze said...


its not absurd at all, its almost confirmed that this is what happened. If E1b1b originated in east africa you must explain why natufians completely lack any SSA admixture while modern east africans and even Mota have significant SSA ancestry. its far more likely that near eastern y dna replaced most of the native african y-dna due to superior technology. And the reason E is so widespread all over africa today is due to more recent expansions such as the bantu migration.

Steven said...


Natufian has North African DNA. This component is shifted towards East Africa on PCA plots so perhaps a population migrated out of East Africa to North Africa and evolved into the North African component. This component could have been spread from North Africa into the Levant.

capra internetensis said...

Lack of relatedness between Sub-Saharan Africans and Natufians rules out gene flow from SSAs to Natufians, but not gene flow from Natufians to SSA, because fairy dust. This law is called "Bronze's Arrow".

Ric Hern said...

As far as I understand Haplogroups D and E evolved somewhere in Asia....Macro Haplogroup DE....

Anthro Survey said...


I completely agree regarding the d-stats, but I still say it wouldn't hurt to run a battery involving modern African populations and IAM as we may get a few valuable clues.

With regards to where the point on the branch between Mota and Eurasians ceases to be called SSA----It is semantics, but if such a para-Eurasian ghost population is real, and if that component is no longer be present in modern Africans to a significant degree, most folks will not call it SSA given the tendency to project modern biogeography into the past.

Anthro Survey said...

@Shaikorth, Sein & Matt

By the way, getting some Tenerian, Kiffian or other Green Saharan DNA wouldn't hurt either. My instincts are telling me that most of these groups had a strong affinity to IAM and/or the potential para-Eurasian branch. After the Sahara desertified, many such groups may have migrated to adjacent regions(like Egypt in the north or West Africa in the south) and/or went extinct.

It's believed by many that the Egyptian civilization was initiated by migrations to the Nile valley during the desertification period of the Sahara. It's possible that any such ghost ancestry was quite present in Old Kingdom Egyptians, being gradually diluted by migrations from the Levant over the centuries. It could offer a fresh explanation, for instance, why one Pharaoh was E1b1a(modal in WEST Africa TODAY), as opposed to E1b1b, since the haplogroup's hypothesized place of origin is placed on the eastern edge of the Sahara. After all, we would expect a diversity in E1b in any groups derived from such a para-Eurasian ghost from East Africa.

Bronze said...

Natufians do not have north african dna, its these ancient north africans (IAM) who have natufian related admixture (ydna E m35). And regardless neither natufians or these ancient north africans have any sub saharan admixture. So E is clearly eurasian in origin. that new skoglund paper on african population structure showed that all y-dna in ancient africa was A or B, except one relatively recent lineage in kenya from around 1500 AD which was E. This strongly points to the fact that E lineages are not ancient within Africa, so they likely came from the middle east or north africa, from a population which was completely distinct from SSA, basal eurasian makes the most sense.

Bronze said...

*all y-dna in ancient sub-saharan africa so far is A or B, with the exception of Mota which had eurasian admixture and plotted close to eurasians on pcas*

Ric Hern said...

I think it was a Ping Pong between Western Asia and the Green Saharas for thousands of years for Haplogroup E especially...and Haplogroups A and B floated around South of the Sahara during this time.

Ric Hern said...

Some Kiffian DNA would be nice since their skulls look rather interesting especially their eye sockets...maybe that was the furthest extent to the North of SSA influence during the Green Sahara period ?

Ryan said...

@Ric - "I think it was a Ping Pong between Western Asia and the Green Saharas for thousands of years for Haplogroup E especially...and Haplogroups A and B floated around South of the Sahara during this time."

Agreed. The Mousterian Pluvial from 50,000 KYA to 30,000 KYA would match the timing. What the hell happened to haplogroup DE though in all the space between Tibet and Africa?

Ric Hern said...

An idea of how the Landscapes looked back then could help. People didn't usually move far from a water source and usually did not settle more than 5 kms from a water source...

Ric Hern said...

@ Ryan

So basically it is difficult to find this water sources because most of it in the Sahara and Western Asia are today covered by sand.Maybe the reason why Denisovans moved into Tibet is to be closer to the River sources to avoid the changing river flow routes downstream ?

If most humans did the same then they would have preferred to migrate through Mountainous areas which could have also provided them with shelter and security against maybe some Palaeolithic samples from the Levant, Southern Caucasus, Zagros etc. will throw light on Haplogroup DE ?

Ric Hern said...

@ Ryan

Or maybe the Atlas Mountain Range ?

Ryan said...

@Chad - I'd just ignore him.

@Ric - Yah, I agree that the deserts from the Sahara to the Thar are probably where DE was hanging out from 50,000 KYA to 30,000 KYA. The distribution is still peculiar though. Why is it so different from CT? What separated the two? Does DE correspond with haplogroup M?

Not sure I follow re: mountains. During desertification they would be the last/only places with a favourable climate though.

Ric Hern said...

@ Ryan

Yes like in the Tarim Basin where watercourses changed drastically due to shifting sanddunes the most favourable areas would have been near Mountains....

Gioiello said...

Yannik Kemouche, but if you remember what I wrote also on your FB page about the tree of PF2431, it seemed that an old line was present in Northern Africa, but the main line from which also your haplotype descends was present from old times in Europe. Thus, as I said above, it is important to know the terminal SNP of this sample, because it could belong to the "African" line and not to the "European" one. I read what Roy King and others wrote ... but the recent aDNA from Africa seems to demonstrate that there were only hgs. A and B, as I, and many Others, are writing from so long. Even E, and not only R-V88, back migrated to Africa.
Genetiker writes: “The two Tanzania samples from 600 AD, like the 1500 AD sample from Kenia, the 2500 BC sample from Mota Cave in Ethiopia, and the modern sample from the Hadza people of Tanzania, have significant amounts of each of the Bushman and Pygmy, Negro, and Natufian-related components. Note that the 1500 AD Kenya sample belongs to the Caucasoid Y haplogroup E1b1b, as the Natufians did. The amount of the Natufian-related autosomal component in African populations correlates very well with the fequency of the E1b1b haplogroup in those populations. The Hadza, for example, have on average 18.20% of the Natufian-related component, and they have an E1b1b haplogroup frequency of 15.0%”.
That the Hadza and the wa-Sandawe were linked to the languages of the Bushmen and even the click languages were linked with the Camitic languages, thus introgressed into Africa (practically we have no survived language of the original African hgs. A00, A and B, admitting that also A and B are African and not the A00 alone), was demonstrated more than a century ago from the great Italian linguist Alfredo Trombetti.

Ric Hern said...

No I don't think that Click Sounds originated with the Cushetic or Nilotic peoples.Most of the Click Languages are found in Southern Africa or areas directly bordering is pretty much Khoi-San related...

Ric Hern said...

Please Help.

Why was Haplogroup BT* only found in China among the Hui people ?

capra internetensis said...

* just means "not any subclade we tested for". What subclades did they test for in these Hui?

Ric Hern said...

I'm a bit confused. Some recent papers talk about a migration of AMH out of Africa at 50 000 years ago. How does that make sense ? When we see Ust Ishim Man dating to 45 000 years ago ? Did all the Mutations between BT and F and Migration as far as Kazakhstan happen in a timespan of 5000 years ?? Or maybe they are referring to the Ancestors of E only.... ?

Ryan said...

@Ric - That reference for BT* on wikipedia is almost certainly wrong. It's in Russian and I don't speak Russian, so... yah, but extraordinary claims require extraordinary evidence, and a sentence on wikipedia referencing a paper most of us can't read isn't much in the way of evidence. The table is readable in English but I have no idea how reputable these guys are. See for yourself.

Salden said...

Lazaridis speaks about the NW Africa paper.

capra internetensis said...


99% chance Ryan is right and the BT* is bogus. I wouldn't assume that the paper even says it's real BT*(xB, CT). I can't count how many times I've seen people do this, assuming samples from some old paper were tested for all possible sub-branches.

50 000 years is probably too low anyway, but the big expansions of Y hgs C, D, and F (and K) and of mt hg M and N are all basically around that age, so that does seem to be when the ancestors of modern humans spread across Eurasia.

Ric Hern said...

@ Ryan


Ric Hern said...

@ capra


bellbeakerblogger said...

I goofed. They are not Beaker, the remains in Ifri n'Amr o'Moussa (IAM) were dated to the Early Neolithic and Kehf el Baroud (KEB) right before the Beaker period.

The KEB is the one that's really interesting given the contacts between Western Iberia during this time.

Anthro Survey said...

Unless he knows something we don't, it seems Dr. Laz made a hasty assumption in the tweet posted by Salden.

There is no reason to be convinced at this point that the pull we see is due to SSA(at least not due to SSA components modal in MODERN Africans).The 2D PCA will produce the same result if there was admixture from a ghost(?) branch basal to Basal Eurasians but crown to all other Africans(assuming significant statistically significant drift lenghts).

The mtDNA is not suggestive at all of mainstream SSA admixture.

Ryan said...

@Anthro - that PCA means that IAM shared drift with SSA that most other Eurasians lacked. Maybe the picture is different on different dimensions but I think that's pretty unlikely since the SSA/Eurasian distinction is what drives PC1.

Who says the admixture had to be on the female side either?

Chad Rohlfsen said...

Natufians and Iranians shift the same way with no SSA. It doesn't have to mean SSA.

Anthro Survey said...


Granted, it wasn't the largest sample, but I only see T and E-m35 over the time period sampled. Statistically speaking, we'd expect there to be a hint somewhere in there. Modern Maghrebis, however, d have visible SSA introgression evidenced by their maternal Hgs(and paternal to a limited extent) as do modern Muslim Egyptians relative to the ancients.

The PCA suggests a shorter drift path to the SSAs relative to BEs, WHGs, ANE, and ENAs on the main trunk of the HSS tree. That doesn't necessitate direct admixture, though. It could simply mean IAM is the result of an admixture from a ghost branch which is more basal to the BE component found in Natufians, Levant_N or Iran_N/CHG(and more proximal to SSAs on the main trunk).

Oh, I'm sure the picture is clearer in multiple dimensions. Ever tried modeling Mozabites in nMonte using global10? Let me tell you: it's a bitch and the closest distance% anyone's gotten to was 3.0, which is a crappy fit, essentially. Nonetheless, which populations get us even that far? Natufians and, IIRC, Ethiopian Jews. There is more to this than meets the eye. I can't wait to use IAM as an input population.

Natufians, I suspect, share more of some basal component with IAM to a greater extent than Levant_Neo or Iran_N does; btw, they're thought by many to have had North African ties. Not only do they pull towards the SSA direction on 2D and 3D PCAs alike like Chad said, but they also score SSA on some calculators. Formal stats explicitly reject direct SSA admixture, though, no matter which SSA populations are used.

Anthro Survey said...

Btw, if we're talking about ghost branch of humanity refluxing back to Africa, it would have to have taken place quite a long time ago---perhaps 30KYA or earlier(based on TMRCA of the exclusively African Y hg E2). It's reasonable to suspect that DE split in Eurasia, whereby D trekked eastward while E came back to Africa.

Millenia later, this E-carrying lineage bounced back to Eurasia yet again, and, by Natufian times, was quite widespread in the Levant. Predynastic and Old Kingdom Egyptians were probably its descendants in large part, resembling IAM. Ramses being E1b1a would be consistent with him carrying ancestry from such a group with a high E1b diversity, not necessarily with just SSA ancestry. Descent from such a basal, more tropically evolved group can also explain the pseudo-negroid skeletal, cranio-facial and integumentary features depicted in Egyptian art.

Ryan said...

@Chad - Yah, I guess just basal Eurasian would show up the same wouldn't it. My bad.

xyyman said...

Anyone with any objectivity knows these researchers are lying and playing the status quo game. The way they sample, the reference populations used, the analytical methods (statistics) used is all geared towards deception. The ultimate goal is maintain the belief that Europeans are not a subset of Africans. That modern Europeans are the be it all. They know it is a lie based upon the results from aDNA studies emerging. That is why they are skewing results and data. They will lie and mislead as long as possible and forums like Davidski, Dienekes play a part with members whose role is to keep the myth going. Dienekes has come around and don’t post bs anymore

Eg look at the new paper on aDNA on prehistoric Africans. They hypothesize that the “ near Easterners backmigrated” into Africa. But used modern Sardinians as the reference population. They cannot use modern Near Easterners because these people there now are essentially Ottoman Turks. And these Turks in the Levant are NOT related to the ancient Levantines who were essentially Africans(North mixed with SSA just as they are today). The researchers know this that is why they switched to Sardinians. But The researchers screwed up because they focused their discussion on 3000bp Tanzanian to draw their hypothesis but forgot that their chart shows that the 81000BP Malawian_Hora carried “European” ancestry(just as I speculated BEFORE the paper was released). TheMalawi_Hora had European ancestry BEFORE the Hunter gatherer European La Brana (5000bp). Do you see the “pickle” these lying Europeans are in. They cannot used modern Levantines(because they are Turks and it will screw their hypothesis), The Bedoiuns may also be a gamble because we know Natufians/Bedoiuns are NOT ancestral to modern West Europeans. They had no other choice but to use Sardinians and then lying about the relation with Levantines withOUT providing data. But I am onto their games now.


The paper is fascinating at many levels. Eg up to about 2000bp South Africans carried no Eurasian(European and Asian-Japanese) ancestry. Yet 8100bp_Hora from Malawi carried mostly European(neglible Asian) ancestry. Before European ancestry entered Europe. lol! This proves there was indeed some substructure in Africa. But South Africa was the last hold out(we may have to wait on aDNA from West Africa) of an ancient African population. A00 In South Africa?

Do you understand what I just said. EUROPEAN ANCESTRY IN AFRICA “BEFORE” IT ENTERED EUROPE!!!!!!! lol!

Davidski. Do some runs with Hora_8100BP, La Brana, Onge and Natufians? Lol!

Ric Hern said...

Yes African in an Very Ancient Sense not in the Modern Sense.In a Modern Sense it can be more accurately described as From Africa and not African.

Khoi-San ancestors already split as early as 260 000 years according to the latest papers and if Hadza is somewhere in the Middle of this their ancestors split around 130 000 years ago....

Ric Hern said...

Or if Hadza is something in the Middle of 350 000 year ago Cameroon and 260 000 years ago Khoi-San then abviously their ancestors split as early as 300 000 years ago....

Valfol said...

You can find "Negroid skeletal" traits among Natufians, obiously rather diluted. This is a common trait among E1b and also whole DE.

Also Mesolithic Nubians, likely-E1b, have strong "Negrid skeletal" traits.

Ric Hern said...

@ Valfol

Obviously because Haplogroup E is very old....

Chad Rohlfsen said...

Oh, whatever. E didn't come about in an SSA population. You guys are completely idiotic sometimes.

Ric Hern said...

Didn't Neanderthal also have an longer skull ? So it is a very ancient trait among homo predating SSA is what I actually mean....