Saturday, October 14, 2017

3,000ybp Pastoralist proves an old point

Well, a while ago now I pointed out that, simply using modern DNA, various Southeast Africans clearly looked partly descended from peoples closely related to modern Somalis and other Horn-Africans of Erythraeic and Ethiosemitic speaking origins (see here and here) and it seems ancient DNA is now backing this up:

[1] "distance%=0.3802 / distance=0.003802"


Tanzania_Luxmanda_3000BP 50.7
Dinka 49.3

That's an nMonte model above utilizing a new 3,000 year old pastoralist sample from Tanzania belonging to the Savanna Pastoral Neolithic culture strongly tied to South-Erythraeic speaking people who began leaving the Horn of Africa for Southeast Africa some 3,000-4,000 years ago. You can see the fit is decent, showing significant Savanna Pastoral Neolithic related ancestry in this average for the Maasai Kinyawa samples in the Global-10 PCA.

We owe this new sample to a study headed up by Pontus Skoglund and it comes with a lot more details I'll be poring over to some extent quite soon but for now; I'm just putting this paper out there and sharing that we finally have some aDNA backing for what modern DNA, archaeology and linguistics have been positing for quite a while now regarding South-Erythraeic speakers and their influencing of Southern and Southeastern Africa.

Do read the study in the meantime, though.


Reconstructing Prehistoric African Population Structure, Skoglund et al.

Kostenki-14's Craniofacial Morphology

Saw this post over at Eurogenes and felt I had to read the study being shared. The study Eurogenes' author shared makes two main claims:

  • That Kostenki-14, a 36,200-38,700 year old European, does not look like Papuans and Melanesians (Australo-Melanesians) as previously claimed by authors such as M.M Gerasimov and G.F. Gebets.

  • That Kostenki-14 supposedly fits with what they call the "Caucasian complex".

The first statement would make some sense and, as far as I recall, craniometric data on other Upper-Paleolithic Europeans didn't tend to imply a strong similarity to Papuans or Melanesians or the like.  So, it would be rather odd if Kostenki-14 looked a lot like those populations. He'd be something of a strange outlier, I believe. Gerasimov and company seem to have just exaggerated the affinities based on certain traits Kostenki-14 has like marked alveolar prognathy which this new paper, for the record, finds he does actually have going.

M.M. Gerasimov (left) and A.N. Rogachev during work at Kostenki 14 site (Markina Gora), 1954.

The second statement, however, is what's suspect. Firstly, if we're talking about a so-called "Caucasian complex" where Kostenki-14 seems overall more similar to various pre-historic Europeans than to Southeast Asians, Ryukuans, Papuans and Melanesians then it seems clear that he would fit within such a complex and prove to, overall, share more craniofacial similarities with fellow pre-historic (and probably even modern) Europeans but it's otherwise clear that he does not actually fit into the more modern definition of what constitutes being "Caucasoid" in cranioform:

The position of the Kostenki 14 man in the CV I–II space is illustrated by a graph (see Fig. 1a): this individual, by the sum of craniometric indicators, is unambiguously characterized by the European complex of characters and shows no noticeable deviation toward tropical groups. Note that we are not speaking about its full similarity to any individual ancient European series. On the contrary, the results of our analysis show a sufficiently noticeable anthropological distinctness, which CV IV demonstrates (see Table 1). It separates the Kostenki 14 individual from all the series included in the analysis (see Fig. 1b). The size of differences is very great, amounting to 43% of the total variability range according to CV IV. This vector practically fully depends on one character, namely, the nose height, which is extremely small in the Kostenki 14 individual. Interestingly, the face height in this case is of little significance.

As someone who was once quite interested in craniometrics (and still somewhat am); I'd seen Kostenki-14's skull years ago and how low his nose height in particular seemed even at face-value was never lost on me:

Simply compare that to this archetypal example of a modern male "Caucasoid" skull of European origin from Bone Clones Inc., Osteological Reproductions:

Just so some are following, having just this much of an outlier nose height would would disqualify him from being an actual "Caucasoid" in the more traditional and modern craniofacial sense of the term, and this without even going into other ways in which he looks distinct such as his marked alveolar prognathism and likely rather paleolithic robusticity.

Kostenki-14 reconstruction by M.M. Gersasimov

Now, finally, there's one other statement that really stuck-out to me in the Eurogenes post and that's that Mikhail Mikhaylovich Gerasimov's reconstruction of Kostenki-14 is wrong. This is not true, as far as I know. The Russian paper doesn't even touch upon the reconstruction's validity (they actually display it without any sort of refutation against it in their study); all they seem to refute about Gerasimov's previous work on Kostenki-14 is the morphological affinity Gerasimov and company noted toward groups such as Australo-Melanesians.

The types of reconstructive techniques Gerasimov employed have been reportedly rather accurate. And reconstruction itself, when done right, can definitely be more of a science than an art.

Kostenki 14's location on map
But also, reconstruction is a different matter entirely from the sorts of craniometric measurements conducted in this new paper or that were once put together by Gerasimov and others. Their measurements or findings in that respect being off and overstating an affinity toward Australo-Melanesians does not actually play into how accurately they reconstructed Kostenki-14's looks. Though I would be interested to see other anthropologists take a crack at reconstructing Kostenki 14 in the future.

But all in all, he wasn't "Caucasoid" in the traditional sense and yes; that reconstruction, as far as we know, was roughly what he looked like in terms of facial features.


1. The Upper Paleolithic Man from Markina Gora: Morphology vs. Genetics?, Khartanovich and Zubova 2017

2. Facial reconstruction – anatomical art or artistic anatomy?, Wilkson 2010

3. Facial Reconstruction, Jenny Omstead 2011

Thursday, October 5, 2017

Lowdown on the recent Minoan and Mycenaean samples

I've had some ask me questions on how things seem regarding the new Minoan and Mycenaean samples and since I've slowly been getting back in the game lately, I figured I'd wade into things simply and swiftly.

I often tend to think PCAs (principal component analyses based on autosomal SNPs) are more straight-forwardly telling and useful for getting a point across and the one above really helps summarize how these new samples generally look.

The Minoan samples

In the PCA, the Minoans nearly join what I'd dub the EEF/ANF/LNF/LHG continuum. A continuum formed by populations seemingly rich in VHG-related ancestry and Basal Eurasian ancestry with Natufians (LHGs/Levantine Hunter-Gatherers) so far proving the most Basal Eurasian and Early-European-Farmers (EEFs), descended from a mixture between Anatolian Neolithic Farmers (ANF) and European Hunter-Gatherers of the so-called Villabruna-cluster, proving the most VHG-related.

This continuum's source is ultimately West-Asia with its most VHG-related end (various EEFs) acquiring more VHG-related ancestry in Europe after their ancestors migrated to the region from Anatolia, bringing agriculture along with them to the formerly Hunter-Gatherer dominated region.

The Minoan samples from Lasithi and Odigritia practically do fall within this continuum and would seemingly sit right in-between Anatolian-Neolithic-Farmers and Levantine-Neolithic-Farmers (LNF) if not for a clear eastern-pull being present within them which implies ancestry outside of this continuum giving them an elevated Ancient North Eurasian-related affinity and the study does address that in its abstract alone:

"Here we show that Minoans and Mycenaeans were genetically similar, having at least three-quarters of their ancestry from the first Neolithic farmers of western Anatolia and the Aegean and most of the remainder from ancient populations related to those of the Caucasus and Iran."

Basically, they've found that the Minoans have ancestry from what I'd call the aforementioned continuum's opposite continuum: a continuum of sorts between Caucasian Hunter-Gatherers, Iranian Neolithic Farmers (INF) and a Hunter-Gatherer sample from the Hotu cave in Iran. These particular pre-historics seem to be largely composed of Ancient North Eurasian-related ancestry and Basal Eurasian ancestry as well as perhaps some Villabruna-related ancestry as in the case of CHGs.

And the paper shares some formal-stat based models in its supplementary information to back what its abstract says up:

The models above basically show that the Minoans can be modeled as part something related to Neolithic Anatolians and part something related to Neolithic Iranians and Caucasian Hunter-Gatherers with those models, as you can see with the second set, showing the lowest standard errors/fitting the best. The paper basically summarizes the nature of the Minoan samples in the researchers' opinions with this bit in the supplementary information pdf:

"The Minoans themselves could be modelled as a mixture of Neolithic Anatolians and Caucasus hunter-gatherers, but they could not be successfully modelled as mixtures of later populations."

The emboldened bit is somewhat interesting as I doubt, for obvious reasons, that these Minoans are really a direct intermixture between Neolithic Anatolians and Caucasian Hunter-Gatherers or Neolithic Anatolians and Iranian Neolithic Farmers. It's obviously much more likely that they got such ancestries and affinities by way of proxy from later populations. I.e. from neighboring populations in Anatolia carrying something like Chalcolithic Iranian-related ancestry:

[1] "distance%=0.5658 / distance=0.005658"

Greece_N:Klei10   68.8
Anatolia_BA:I2683 31.1

In the above nMonte model using Global-10 PCA positions, a population akin to Bronze Age Anatolians that intermixed with the earlier ANF/GNF type inhabitants of Crete would have been responsible for the more eastern-pull carrying CHG/INF/IHG related ancestry. And I'd say it's more sensible to assume this sort of way is how the admixture made its way to the area (by way of proxy) but I'm of course not at all detracting from the authors' findings which is that these Minoans look to carry ancestry related to the CHG/INF/IHG continuum. [Important Note]

That being said, I'd also say the lowdown is that these Minoans are visibly distinct from modern Cretans and neighboring mainland Greeks. They're notably more ANF/GNF-related and seem to lack the steppe-related influences in later inhabitants which makes good sense given that they are generally accepted by scholars to not have been Greek/Indo-European speakers and, as a result, seem to trace the majority of their ancestry back to pre-historic West Asia.

However, given that you can model even modern mainland Greeks as being close to a 60% Minoan population - :

[1] "distance%=0.0682 / distance=0.000682"

Minoan_Lasithi   48.25
Ukrainian_West   20.05
Yamnaya_Kalmykia 16.25
Minoan_Odigitria 13.90
Polish            1.55

[1] "distance%=0.0499 / distance=0.000499"

Minoan_Lasithi   59.40
Srubnaya         30.75
Ukrainian_East    4.25
Minoan_Odigitria  3.65
Srubnaya_outlier  1.95

- we're not remotely talking population replacement here. We're just looking at a shift over-time toward Europe caused by various migrations (like that of early Greek speakers and medieval Slavic speakers) bringing in elements such as ancestry from the Pontic-Caspian Steppe.

The Mycenaean Samples

As both the PCA and the study impart, there's not a huge difference at all between the Minoan and Mycenaean samples but there is a difference.

The Mycenaeans come quite close to entering yet another continuum of sorts. This continuum, however, is much more modern. It's been dubbed by some, including myself, as the Eastern Mediterranean continuum and is essentially a cluster inhabited by Sicilians, the Maltese, Western Jews, various Greek islander populations such as modern Cretans, and, finally, Cypriots. 

This is essentially a continuum for populations that serve as a sort of bridge between mainland Southern Europe and the Levant with Cypriots being the most Levantine shifted population:

[1] "distance%=0.4026 / distance=0.004026"

Lebanese_Christian 71.2
Greek              28.8

And Sicilians being the most mainland Southern Europe shifted population:

[1] "distance%=0.5027 / distance=0.005027"

Italian_Tuscan     65.7
Lebanese_Christian 34.4
These Mycenaeans look to lie somewhere in-between the two and have, as the PCA implies, something of a lower eastern-pull than usual, bringing them just a smidgen closer to the EEF/ANF/LNF/LHG continuum. The study itself imparts the following:

"We have successfully modelled Mycenaeans as a mix of (i) Neolithic populations of Anatolia, Neolithic Iran or Caucasus hunter-gatherers, and eastern European huntergatherers or Upper Paleolithic Siberians, (ii) Neolithic Anatolians and Chalcolithic-to-Bronze Age people from Armenia, or (iii) Minoans and Bronze Age people from the Eurasian steppe (or from mainland Europe after the arrival of steppe ancestry there)..."

This is indeed quite easily noticeable via means such as nMonte when a non-academic like myself takes a look:

[1] "distance%=0.2597 / distance=0.002597"

Minoan_Lasithi   85.0
Sintashta         8.4
Yamnaya_Kalmykia  6.6

[1] "distance%=0.2985 / distance=0.002985"

Minoan_Lasithi      84.1
Srubnaya            15.9

The general picture just seems like they're basically close to a majority Minoan-like population that just had some steppe admixture (~10-15%) tacked onto it. This makes sense given that the Mycenaeans were Greek speakers, they were in fact the first known culture to have written Greek down, and the Indo-European language they were speaking would have most likely been brought to Greece by people substantively descended from Bronze Age Steppe pastoralists very similar in ancestry to the pre-historic peoples of the Yamnaya, Srubnaya and Sintashta cultures.

Although, I wouldn't simply interpret this as "The Mycenaeans are descended from the Minoans and are just them + some steppe ancestry." I'd say it's more sound to assume there were Minoan-like peoples in the areas of mainland Greece that these Mycenaean samples are from and it was these people who acquired early Greek alongside steppe admixture intermixed with a complex cultural framework that was already present via being influenced by the likes of the Minoans to herald the first true pre-modern Greek civilization.

Now, there's more information to be had like the uniparental (Y-DNA and mtDNA) results of some of these new samples or some phenotypic data acquired via genomic analysis but this post was ultimately just concerned with summarizing the autosomal affinities shown by these Minoan and Mycenaean samples so I'll leave it at that for now. I also advise reading the full study + its supplementary materials if you wish to know more, of course.


1. Genetic origins of the Minoans and Mycenaeans, Lazaridis et al. 2017


1. You can understand how nMonte works by going here. And my thanks go to the author of the Eurogenes genome blog for the PCAs used for nMonte and the one shared visually in this post.

2. I've noticed that some people are making a big deal about the lack of either R1a or R1b subclades found among the Mycenaeans given that they, unlike the Minoans, would have been Indo-European speakers, and I'd just like to point out that the Y-DNA sample-size here is literally n=1. 1 sample that's J2a1, that's it. I'm sure once we have more samples some R1b and the like will pop-up given the obvious steppe ancestry these Mycenaeans carry.

3. This is a somewhat decent discussion on the study to check out, though it slightly goes off the rails at times later into the thread.

4. For those wondering why I haven't made a post on the recent African papers; this post was mostly done several weeks ago, I just made some minor edits today and posted it. I'm a little busy lately but will try to make posts on those African studies soon.

Monday, August 14, 2017

Some interesting nMonte models for Oromos

So I recently got around to utilizing nMonte with David's Global 10 PCA and have been able to run models for several global populations and I might share some of these in due time but, in the mean time, I've been working on running models for several Horn & generally East African populations but don't want to crowd this post up with models for all of them so I'll be focusing on Oromos with this post just to sort of introduce people to how I might conduct future nMonte related posts.

You can understand how nMonte works by going here.

At any rate, let us begin:

At first, the global 10 PCA didn't have Ari samples but did have the ancient Mota sample present whilst also lacking Yemenite Jews and Copts so I tried to create a model for the Oromos utilizing Somalis, Negev Bedouin-Bs and Mota where I essentially got this:

[1] "distance%=0.298 / distance=0.00298"

Somali   78.25
Mota     14.20
BedouinB  7.55

Quite a strong fit there and with arguably sensible Mota/Ari-related ancestry most likely owed to Omotic speakers alongside nearly 10% later West-Eurasian ancestry shown through via the model's need to utilize the Negev Bedouins but with quite a lot of Somali-like ancestry being present. This model would basically impart that a 78% Somali, 14% Mota and 8% Bedouin-B population would be very close to being identical to this Oromo average.

Now, in the following case, I utilized Somalis, Ari-Blacksmiths and Yemenite Jews since Aris and Yemenite Jews are more historically sensible. Remember always that a model should make sense (i.e. historical sense) and fit with other analyses and not just statistically fit well:

[1] "distance%=0.3255 / distance=0.003255"

Somali                   74.8
Ethiopian_Ari_blacksmith 19.4
Yemenite_Jew              5.7

A noticeably poorer, but not all that much poorer at all, fit with a climb in most likely Omotic-speaker related admixture which makes sense since Mota is quite an ancient individual who seems to have little genuine Eurasian ancestry unlike the probably more Ari-like population to have admixed with the ancestors of various Oromo speaking subgroups.

The later West-Eurasian ancestry has gone down a little with this model as well which made me wonder if it might be interesting to include both Yemenite Jews as well as Negev Bedouins in the next model but the overall round up for the above model would be that this Oromo average would be extremely close to a 75% Somali, 19% Ari-Blacksmith and 6% Yemenite Jewish population.

[1] "distance%=0.2732 / distance=0.002732"

Somali                   72.10
Ethiopian_Ari_blacksmith 20.65
BedouinB                  6.40
Mota                      0.85
Yemenite_Jew              0.00

This seems to be the very best fit yet and what's interesting is that, as I've noticed before and suspected, Negev Bedouins are favored by these models over Yemenite Jews which is very interesting as this also holds for Habeshas as well, as will be noticed via later posts I make.

I do wonder why given that it's obvious that at least some notable portion of the later West-Eurasian admixture in the Highlands has to be owed to the Proto-Ethiopian Semitic speaking community which would have migrated over from Southwestern-Arabia/Yemen. But we also see a slight upping in the most likely Omotic-speaker tied ancestry thanks to Mota's introduction, I assume. This tallies off into:

72% Somali, 22% Ari/Mota and 6% Negev Bedouin-B being extremely close to this Oromo average, and I keep emphasizing that this is an Oromo average because that is actually one problem with these particular models. I'm utilizing the average clustering position of the modern populations present in the datasheet (since that's all that is available) and while that isn't such a big deal for more homogeneous sets like the Somalis or Tigrinyas; it can obviously not give us the whole picture for more heterogeneous clusters like the Oromo samples.

Different subsets of the Oromo cluster will likely end up with different results under these models (i.e. the more Borana and Borana-like samples will assumably not show the Negev Bedouin-B/Yemenite Jewish affinities) and these results are ultimately what you get for an averaged out population containing all those differing samples.

But, in the end, the results look generally sensible in that they can spot that Oromos generally have substantive Ari-like ancestry whilst many, like Habeshas and Agaws, carry later West-Eurasian influences that Somalis have far less of or generally don't carry at all while the remainder of their ancestry looks quite Somali-like as ADMIXTURE had implied with other closely related populations in the past.

Being more accurate about the substrates in Ethiopian Semitic

Several times in the past; I've stated that the Ethiopian Semitic languages have an Agaw/Central-Erythraeic substratum and while this is quite true for the likes of Amharic, Argobba and Tigrinya... It's not entirely true for Harari, the Gurage languages & Tigre. 

Tigres, unlike other Ethiopian Semitic speakers, were not historically a predominantly sedentary highland farmer population but were instead lowland semi-nomadic pastoralists similar to many Bejas, Sahos, Afars & Somalis. Alongside this, their language doesn't have a Central Erythraeic substratum but a Beja/North-Erythraeic substratum which implies that some of their ancestors and predecessors were most likely Beja speaking semi-nomads. [3]

In fact, a people in Eritrea and Sudan called the "Beni-Amer" are often understood to be a mixture between Bejas and Tigres and will speak either both languages or one of the two, though they're often considered one of the tribes of the Beja.

Old photograph of a Tigre man with a slain lioness

And, as I've pointed out in the past, a good amount of the peoples on what is now the Eritrean coastline were seemingly referred to by the Greco-Romans essentially as "Barbaroi" (non-Greek speakers) of some sort or other whilst their seemingly pastoralist life-style was sometimes alluded to. These peoples at one point, unlike the Barbaroi in North-Central Somalia who seemed largely independent, were seemingly under the control of the Aksumite Monarch and even "policed" the shores of the Eritrean coastline on his behalf:

"One Metrodorus, a philosopher, is said to have penetrated to further India in order to view places and see the world. Inspired by his example, one Meropius, a philosopher of Tyre, wished to visit India with a similar object, taking with him two small boys who were related to him and whom he was educating in humane studies. The younger of these was called Aedesius, the other Frumentius. When, having seen and taken note of what his soul fed upon, the philosopher had begun to return, the ship, on which he traveled put in for water or some other necessary at a certain port. It is the custom of the Barbaroi of these parts that, if ever the neighbouring tribes should report that their treaty with the Romans is broken, all Romans found among them should be massacred. The philosopher's ship was boarded; all with himself were put to the sword. The boys were found studying under a tree and preparing their lessons, and, preserved by the mercy of the barbaroi, were taken to the king. He made one of them, Aedesius, his cupbearer. Frumentius, whom he had perceived to be sagacious and prudent, he made his treasurer and secretary. Therefore they were held in great honour and affection by the king."[1]

Whether or not they're linguistically the "descendants" of these peoples is rather irrelevant, in life-style and the areas they inhabit; modern Tigres, Bejas, Sahos and Afars on the Eritrean coast are essentially the successors of these tribes and most likely, in some great part or entirely, descended from them.

Some Gurage folk beside some of their traditional dwellings

The other group are speakers within the Harari and Gurage groupings of "Southern Ethiopian Semitic" whose languages actually have a Sidamic substratum. By Sidamic I mean "Highland East Erythraeic" which is a subbranch of East Erythraeic alongside Lowland East Erythraeic, the latter being the node Oromo, Somali, Afar and Saho belong to.

This makes sense given that the speakers of the various Gurage languages have seemingly lived in the Central-South Ethiopian Highlands for quite a long time, to a point where their languages have acquired rather pronounced influences from the local Erythraeic languages:

"Sidamo is the substratum language of the Gurage speaking region. Sidamo influenced the Gurage cluster in the phonology, morphology, syntax, and mainly in the vocabulary."" [4]

Even their dwellings/huts which are pictured above are more similar to those of close-by Oromos and Sidamic speakers, in my humble opinion, than to those of Amharas or Tigrinyas. They also look to lack Hidmo type dwellings (some Hidmos) which were historically found among Tigrinya speakers alongside round-huts which seems somewhat representative of how Aksumite farmer peasantry (what would've been the majority of the populace) might have lived in:

"Some clay models of houses survive which illustrate the architectural style of the smaller
Aksumite dwellings. A round hut, with a conical roof thatched in layers, and a rectangular doorway, is one type from Hawelti (de Contenson 1963ii: pl. XXXVII, b-c). A second type from Hawelti is rectangular, the doors and windows also rectangular, with a roof supported by beams whose `monkey-head' ends can be seen below the eaves. The roof has a small parapet and there is a waterspout to drain it (de Contenson 1963ii: pl. XXXVIII-XXXIX)." [1]

Granted, Amharas historically tended to lack this type of dwelling as well, from what I've seen; and, for at least the last 300-500 years or possibly more, look to have mostly inhabited some form of round-hut, whether wattle & daub or mud-mortared stone in construction. [note] Unlike Argobbas (their close linguistic relatives in a supposed "Amhara-Argobba" linguistic node) who do have some form of pseudo-Hidmo type dwellings alongside some huts to this very day.

Aside from this, the Gurage folk also look to display numerous cultural traits shared with the local Erythraeic speakers of Central-South Ethiopia such as an apparent historic presence of Waaq worship (Sky-God seemingly worshiped by many pre-Islamic Somalis, Afars & Sidamic speakers as well even some Oromos today) along with a strong presence of Muslims to this day given that at least one of the Medieval Muslim polities of Ethiopia's hinterland was likely "Gurage" in origin and a neighbor to the likes of the Sidamic speaking Hadiya polity.

1800s depiction of Harar

The final group in question are intriguingly Adares/Hararis whose language seems to form a node with the East-Gurage languages called "Harari-East Gurage"; they are a people who've, at least for the last ~300-500 years, inhabited the town of Harar in Eastern Ethiopia (East Hararghe) which, around the mid-1800s, had a population somewhat below ~15,000 people:

"During his visit in the 19th century (1854), Richard F. Burton describes the town as having roughly 8,000 inhabitants with 2,500 of these being Hararis, 2,500 being Somalis and 3,000 being Bedouins who come and go. Though in this case, "Bedouin" is simply a term for nomads in the region whether they were Somalis, Afars or Oromos and doesn't necessarily connote an Arab origin. At this point in time, predating the conquest of Harar by Menelik II, there was no mention of Oromo, Amhara, Tigray-Tigrinya or Gurage residents who are now common in the town whilst Hararis and Somalis who are now minorities looked to be the bulk of the population at the time." [note] [2]

You'll probably see several at times outlandish claims about how old the town is but, as far as I've noticed through actual evidence for the time being, the earliest mentions of the town that I've seen only really date to the early 1500s during which case it's historic sections weren't seemingly as well-built as they are now or during the mid-1800s (it's surrounding wall wasn't built until later into the 1500s, for example). Richard F. Burton, who visited during the mid-1800s, is perhaps more diminishing of its early past than I'd be though:

"Harar was a mere mass of Bedouin villages during the reign of Mohammed Gragne, the “left-handed”" [2]

It probably existed before the 1500s but as a much smaller and more irrelevant settlement. Granted, Harar was never truly that big historically. It's importance was as a town of Islamic learning, basket weaving, textiles and trade but it wasn't exactly some sort of pre-modern "Metropolis".

But that's not what's interesting here... What's intriguing is that the language of the local Adares/Hararis has a Sidamic substrate the same way the various Gurage languages do. Enrico Cerulli was one of the first to discover this many years ago and it seems to be common wisdom at this point. Although, what's extra-intriguing is that he suggested the prior inhabitants of the area where Adares/Hararis live now were Sidamic speakers and that the Ethiopian Semitic language was grafted onto theirs as they were assimilated.

I find it a bit implausible that Sidamic speakers lived that deep into Eastern Ethiopia and so close to Northwestern Somalia. Though I suppose it's not impossible prior to the spread of the Oromo people. Nevertheless, it's more likely, to me, that the linguistic predecessors of Adares/Hararis perhaps lived where the Gurage peoples do and acquired their substrate there then moved to East Hararghe to find people mostly like Somalis but I suppose one can't be absolutely sure.

I've encountered the autosomal DNA results of at least two Hararis who were tested via 23andme and one was, oddly, a distant relative of mine and, aside from being 1/8 Pakistani (paternal great grandfather), his Horn African side seemed somewhat like a crossing between Somalis and Tigrinyas. The other fellow, as is visible above, seemed essentially identical to your average Tigrinya. 

Not exactly what I'd expect from people who might be predominantly of Central-South Ethiopian origins (one might expect higher levels of Ari Blacksmith-like ancestry? [note]). Or perhaps it implies, as some prior indications might have, that Sidamic speakers are very similar to Northern Ethiopian Highlanders from a genetic perspective. 

At any rate, all of these little details I'm getting into now are for another time, the point of this post was outlining that not all of the Ethiopian Semitic languages chiefly have Agaw/Central-Erythraeic substrates the way Tigrinya & Amharic do. Harari and the Gurage languages have a Sidamic substratum whilst Tigre has a North-Erythraeic/Beja substratum, for instance.


4. Gurage Studies: Collected Articles, by Wolf Leslau


1. If anyone has access to Sidamo, Alaba, Kembatta or any other such Sidamic speaker genome; sharing its results would be quite appreciated, same goes for Harari or even Tigre results (granted, I've seen a number of what look to be Tigre results already).

2. I edited the text from Munro-Hay's book to say "Barbaroi" instead of "Barbarian" as the English word "Barbarian" is a poor translation of what Greco-Roman terms such as "Barbaros" (βάρβαρος), "Barbaroi" and so on actually meant: [-] [-]

3. This is an old post I largely prepared almost 7 months ago but never got around to finishing up the last few details for and posting so apologies if anything at all seems vaguely outdated though it mostly looks tip-top to my eyes now.

Friday, January 20, 2017

Human Genetic Diversity ≠ Discrete or Pure Races

One thing I've noticed is that a lot of much more "lay" people who glance over at population genetics tend to mistake Human genetic diversity, which is real, as validation of discrete races based on racialist models from the 18th-to-20th centuries or even current models ("Black", "White", "Brown" etc.).

They observe something like the above PCA (principal component analysis) based on 166,000 autosomal SNPs and they assume that because populations cluster apart; this means the discrete  races they believe in exist... That you can, for instance, divide most of the Homo Sapiens Sapiens (HSS) population like this:

This is, quite simply, not true. Simply because a massive portion of the HSS population is not entirely or overwhelmingly descended from one "discrete race" that then formed into another discrete race that then formed into another and another until their ethnic group came into existence.

Instead, the actual truth about several Human populations today is much more in-line with what you see above. In order to properly explain this I'll need to throwback to something I went into in an older post of mine where I pointed out the two main mechanisms by which Anatomically Modern Humans (HSS) diverge through:

Genetic Drift

  • Genetic Drift: Population A & Population B are more or less "identical" and each have 1,000 people in them but they are geographically separated for 10,000 years... Population A doesn't experience any odd population losses and just continues to prosper from the original 1,000 roster but Population B loses 800 people due to an Earthquake very early on and the descendants of Population B find themselves being descended, 10,000 years into the future, from only those 200 folks who survived the Earthquake; losing a lot of their population's prior genetic diversity and diverging from Population A as a result of this alongside later mutations and selection, and so on. (more on genetic drift here and here)
  • Admixture: Population B's 1,000 remained intact in terms of numbers but they, even within a short time-frame, discovered a far away Population C with ancestry distinct from Population A & B and then intermixed with them to form a new mixed population that is now divergent from Population A due to being a mixture between Population B & C.

The thing about the old racialist model where we were all usually divided into four mega-races known as Mongoloids, Negroids, Caucasoids and Australoids- :

Natural History gallery of the Horniman Museum, London

-is that, even though the anthropologists who originally devised this worldview weren't familiar with the Human genome, it is very based around Genetic Drift and divides Humans up almost like they are "proto-subspecies" of sorts ("biological races" are the level right below subspecies, usually). 

For example, there was a Caucasoid race and then, via some intermediaries, "Aethiopids" and "Nordids" developed and Somalis and Danes developed from these two sub-races respectively. There might, at times, be some acknowledgement of admixture playing a part (Aethiopids are sometimes acknowledged to be slightly "Negroid" influenced, for instance. Or Maasais are noted to be an "Aethiopid+Negroid" mixture of some sort) but the model is still very dependent on the idea that Human phenotypic and thus genomic diversity was mainly shaped by natural selection and this is partly where it falls short.

By this logic, you would have to explain the differences between West Eurasians, as they are in that regional PCA above, as being mostly caused due to the formation of sub-races. I.e. There was a single ancestral West Eurasian population (Caucasoid race) and Northern Europeans formed as a sub-race of this population because their ancestors were separated from the ancestors of "Arabs" for tens of thousands of years and then natural selection & mutations, and therefore genetic drift, took place and that's mainly why we see genetic diversity here.

However, that's simply not the case. 

Instead, none of these West Eurasian populations are separate/discrete races from one another. They in fact share very recent ancestry from, for now, what look to be three or so core pre-historic populations. Villabruna-related West Eurasian Hunter-Gatherers (VHGs a.k.a "WHGs"), Ancient North Eurasian-related peoples (ANEs) and the theoretical "Basal Eurasians". 

They are, in large part, the product of admixture and not simply genetic drift where there was a Population X and then Populations Y and Z descend from it, were separate for a long time, then developed different mutations that altered the genotypes and phenotypes of a few of them, and then went through natural selection and genetic drift to select for those traits and become distinct entities. 

What makes them cluster apart is actually, to some great extent, the genetic drift between the highly divergent pre-historic populations they descend from. For instance, the ancestors of ANEs and VHGs, even though these two groups are closer to each other than they are to Eastern Non-Africans (ENAs), diverged apparently over 35,000 years ago [note]; and Basal Eurasians, whatever they might turn out to be in the end, are more divergent from ANEs and VHGS than they are from ENAs. 

It is ultimately the distinctions in how much ancestry these modern West Eurasians trace back to these pre-historic populations that differentiates them and makes them cluster apart. For example, Finns will pull more toward VHGs than Saudis will as they carry much more ancestry related to such pre-historics than Saudis do whilst Saudis will pull much more toward Basal Eurasian-rich pre-historic West Asians (i.e. Natufians and Neolithic Levantines) as they carry much more of ancestry related to such pre-historics than Finns do. [note]

West Eurasians are thus differentiated very much by admixture rather than merely just genetic drift. [note] So, in their case, various old racialist models fall flat. There was no single ancestral population that sub-races developed from and then yet more sub-races developed from and then the modern groups are just descendants of those separate sub-races... Instead, West Eurasians are interrelated, and most likely also look similar, because they are the result of very recent inter-mixture between these distinct pre-historic groups. [note]

And the thing is, this is similar for several ethnic and regional groups across the world. As a matter of fact, several (or often the majority) of the "native" ethnic groups in all of the following Old-World regions share upwards of 10-20% of their ancestry from arguably within the last 5,000-30,000 or so years via the likes of Basal Eurasian, VHG and ANE related ancestries:

Regions of interest colored in red
Essentially none of those populations can be considered truly "discrete races" from one another despite their differing phenotypes (i.e. Tamils compared to Somalis or ethnic Russians compared to Mauritanians) and distinct plotting points in global PCAs, because they did not develop the way species and subspecies usually develop which is mainly via substantive genetic drift over a long period of time. 

These particular populations mainly differ, again, because of the ancestries they don't share and the levels of ancestries they do share (i.e. East African cluster-related ancestry in Somalis not being present in Tamils and "ASI" ancestry not being present in Somalis and then differing levels of pre-historic West Asian-related ancestry in both groups). They do not differ because they're all downstream developments from an Ancestral-HSS population that mainly differentiated via mutations and subsequent drift.

Basically, you cannot assume that two populations cluster apart or are genetically distinct (by Human standards) simply because they are discrete and pure entities the way the Caucasoid, Negroid, Mongoloid and Australoid model, or any sort of racialist mindset, tends to imply. [note]

"The peoples of Africa"
Nevertheless, I suppose one could argue that certain populations are genuinely "discrete" in that they have not shared certain ancestries in well over 35,000 years. For instance, this can be said about West-Central Africans when compared to East Asians but here things do get a bit dicey as well since, while you can assume they're discrete from one another, they themselves are probably not, to some great extent, "pure" or mostly pure entities. 

By that I mean... They too are probably, in some part, the result of admixture rather than mostly or entirely being linear developments from a singular ancestral population which is how the old racialist model might paint things. 

For example, the quite diverse mtDNA profiles (simply based on their non-M&N lineages) of groups like Omotic speaking Southwestern Ethiopians, Niger-Congo speaking West-Central Africans and Nilo-Saharan speaking Southern Sudanese people tend to imply that they are probably the result of admixture between distinct pre-historic populations within Africa itself. [note]

Some of these ancestral populations were possibly even as distinct from each other as the San are from modern West-Central Africans (time divergence appears greater than the time-divergence between West-Central Africans and the Han-Chinese, and genetic drift (based on Fst) is comparable to the drift between the Han-Chinese and the English).

Groups that would count as "Negroids" within Africa should also not be seen as some sort of genetic monolith. They're not... And even the old racialist model didn't truly imply as much. There's often a West-Central African cluster ("Niger-Congo" above) and an East African-cluster ("Nilo-Saharan" above) in ADMIXTURE runs, for instance. The Fst between these two clusters, as an example, is a little over 1/2 the Fst between the East Asian and European clusters above.

Also, based on Haplotypic data, the time divergence between some of these "African" ancestries (i.e. the African elements in Somalis and the African elements in Yorubas) implies they possibly haven't shared ancestry in over 30,000-40,000 years or so which is comparable to the, so far, supposed time-divergence between ENAs and the ancestors of European Hunter-Gatherers. [note]

These African populations are also often, but not always, mixtures between these distinct modern clusters (as can be seen above) which are quite likely, in my humble opinion, not "pure" clusters either. We may just find that, like West Eurasian clusters from before we had ancient DNA flooding in, they are mixtures between distinct pre-historic African groups. In fact, I'd say this is quite likely given the mtDNA and sometimes Y-DNA profiles of these ethnic and regional groups. They even show slight signs of such mixture at the lower Ks of certain ADMIXTURE runs... [note]

So whilst some of these African populations, that mostly or entirely lack West Eurasian admixture, can be seen as separate from Out-of-Africa groups like ENAs in a manner that can somewhat be drawn onto a "tree"; they too look to be, like West Eurasians, not "pure" entities that linearly developed from one ancestral race that merely developed into discrete sub-races in order to form their current genetic diversity. Admixture played an important role here too, for certain. (also see note 2 below the reference list)

"The Races of Asia"

Things also look grim for the old "Mongoloid race" in that large subsections of it (Native Americans & Siberians) are not even entirely Eastern Non-African in origins. Native Americans are mostly a mixture between Ancient North Eurasian-related people and Han-like people and roughly the same is true for many non-European Siberian populations (Kets, Nganasans et al.) whilst old-school racialism failed to notice that East Asians and "Australoids" were closer to each other than they were to West Eurasian or African populations. [note]

Some like to re-envision these old racial classifications and now, to them, the only true Mongoloids are those in East Asia (the Han, Japanese, Koreans etc.) and, for example, the mixture that formed modern West Eurasians fits with the discrete and pure "Caucasoid race" idea but this is merely someone trying too hard to make these old racial classifications stick when they know the population genetics doesn't truly fit with them as they were traditionally envisioned. 

"American peoples"

And the traditional Mongoloid classification did include Native Americans as a subgroup of it. In fact, one vaguely humorous side note about the term Mongoloid is that its namesake, the Mongols, are not even of purely Eastern Non-African or particularly Han-like descent. See here

At any rate, it seems quite possible to me that a pre-historic population in East Asia is responsible for the traits we traditionally associate with East Asians (i.e. one whom important derived EDAR alleles formed among) and being partly or mostly descended from this population is probably responsible for certain phenotypic traits we see among East Asians like the Han or the Japanese (epicanthic folds et al.)... But even then, East Asians, as a whole, are likely not "pure" or "discrete" entities either, and this is definitely the case for some of them.

The Japanese, as one example, carry Jomon Hunter-Gatherer type ancestry which comes from a population that seems closest to modern East Asians but is still distinct from pretty much all of them (It also looked physically distinct from them as well. See here.) so admixture certainly played a part in the formation of the modern Japanese. And, while I'm still skeptical about the validity of this, various East Asian populations might also be part Ancient North Eurasian-related which again points to admixture playing a role in developing their modern genomic profile rather than them being linear developments from one single ancestral Mongoloid race. [note]

"The Human Races"

Now, one of the main reasons I'd say the old racialist approach fails is because, in a sense, they attempted to divide Humans based on phenotype-hinging taxonomy. There was no knowledge of DNA when these "races" were divised and they were divised, usually, but not always, as these mostly pure and discrete races and sub-races based around, for the most part, the outwardly visible phenotypic traits of various populations (i.e. their craniofacial traits).

Somalis, Oromos, Habeshas and the like were craniofacially most similar to each other and thus formed an "Aethiopid" sub-race of the Caucasoid race as they were supposedly more craniofacially similar to West Asians, North Africans, various South Asians and Europeans than to other Africans and it went on and on from there. [note]

Plot based on craniofacial data from a 10 year old non-racialist peer-reviewed craniofacial study

Some of the main problems with doing this would be:

  • Admixture: The incredibly important role admixture played in forming the modern gene pools of several HSS populations is recurrently down-played or ignored depending on the group being studied or whose work you're reading. I.e. Danes can merely be an off-shoot of Nordids rather than, basally, a mixture between ANEs, VHGs and the, for now, theoretical Basal Eurasians. A certain "craniofacial type" (accompanied by some other phenotypic traits) that occurs on average within a population is ofttimes assumed to be the mark of a "discrete" and, ofttimes, "pure" race of its own. [note] (also check note 6 in the notes section below)
  • Hinge entirely on phenotypic traits: People often overestimate how much genomic divergence and relations correlate with physical differences. Differences in pigmentation (skin, hair and eye color), hair-type and craniofacial traits among Modern Humans are mainly controlled by seemingly only some hundreds of SNPs among ~10 million SNPs in our entire genomes. Hence why, until genomics set-in for a while, nobody noticed the Andamanese were genetically closer to East Asians than to Africans and West Eurasians. Two populations that are substantially drifted and/or haven't shared ancestry in seemingly over 20,000-40,000 years (Aris and Yorubas, for instance) can look quite similar whilst two that look very distinct can be closer than many would expect (the Han and the Andamanese). [note]

Current racialism is even less scientific and much more arbitrary as it often heavily fixates on one trait: skin pigmentation (though other outward phenotypic traits are definitely taken into account).

In this case Australian Aborigines, Papuans and Yorubas, in many people's eyes, count as part of the same "race" ("Black") because they look vaguely similar and are dark-skinned, and people with pale-skin from Europe are usually seen as "White". And then various South Asians and West Asians are often darker than Europeans (especially Northern Europeans) so they're "Brown" and it goes on and on from there.

This form of racial classification is often pretty arbitrary, to be quite frank. For one, a person's membership in a racial group can seriously depend on who's classifying them or the culture of the place they're in...

In the United States, people of part "Black" and "White" ancestry are often considered "Black" whilst this may not be the case in South Africa or certain Caribbean countries where specific terms exist for such people (see here and here). Where I grew up (the UAE) you will see Arabs, Iranians and the like being considered "White" when they certainly wouldn't be seen as such if they went to various parts of Europe or, at times, even the US. 

These currently popular racial groups are simply not well defined, especially from a scientific viewpoint. The Early Modern form of racialism, while obsolete, was much more thought out and wasn't totally wrong on everything [note]. But current racial groupings, at times, merely look like any sort of phenotypic, cultural, ethnic or genetic group suddenly counts as a "race". 

This is why, in my humble opinion, you get notions like "the Jewish race" when Jews (particularly Western and certain Mizrahi Jews) are basically just an ethno-religious group (emphasis on the ethno) and of course not some sort of discrete race when compared to West Asians or Europeans or North Africans and various others.

This is also why various "Latinos/Hispanics" in the US such as ones of Mexican origins, despite commonly (not always) being of mixed origins (Native American, European and sometimes African), are often labeled as their own race category because they form a distinct cultural and sometimes physical group even though they're obviously not actually a "separate race" from "White" Americans (sharing in ancestry from Europe and even in the ANE-related ancestry in their Native American roots), for example. 

This sort of treads on the very notion of what races were originally supposed to mean within a Human context as they were very much centered around a taxonomy inspired notion that we, as Homo Sapiens Sapiens, could be divided into into proto-subspecies based on phenotypic traits (with craniofacial measurements at the forefront) rather than being a continuum of mixtures that oddly count as their own discrete and pure races in the minds of many.

At the end of the day this form of racialism falls flat on its face for similar reasons as the older form, along with its way of dividing everyone based on phenotypic traits being even more shallow (hence why Australian Aborigines and Yorubas can both count as "Black" in the minds of some, mainly based on their similarity in pigmentation and their somewhat superficial facial similarities).

But, there's one thing to keep in mind here... I'm not telling you that Human genetic diversity does not exist and that we're all 100% the same. As the title says: "Human Genetic Diversity Discrete or Pure Races". Homo Sapiens Sapiens genetic diversity, however high or low overall, does exist as the very title of this post basically espouses but what this post means to point out is that the racial groupings and population-structures you most likely believe in, whether based on old and new conventions, do not well and truly correlate with population genetics. [note]

Nevertheless, one is free to create their own divisions based on population genetics, in my humble opinion. I.e. I tend to group "West Eurasians" together as they're closer to each other than to most outside groups, then the same is true for Eastern Non-Africans, West-Central Africans, Horn Africans, and so on.

Then within these groups you can near endlessly create other regional sub-groups that do, genetically, seem closer to each other than to the rest of their macro-grouping like Southern Europeans, Arabians and the like as subdivisions within "West Eurasians" [note] but the main thing to grasp is that these often aren't pure or even always discrete entities that you can draw onto a tree structure like this:

Hence, to a great extent, why both old and new racialism aren't in line with the actual science. Old racialism comes closer than new racialism but they both fall short in the end. Actual Human genetic diversity is a good degree more complex.



1. Link

2. Link

3. Link

4. Link

5. Please do read all of the "[notes]" including the four above as they're crucial for fully grasping this post. If you haven't read them; you have not actually read this post.

6. I also strongly recommend going through a comment section tangent below between myself and a user by the name of Onur. Some key things are elaborated upon there.

Special thanks go to the author of Vaêdhya and an administrator at Anthrogenica who goes by the pseudonym "DMXX" as well as a long-time friend in this field who goes by pseudonyms such as "Lazara Masho" (23andme) & "Lank" (Antrogenica) online for reading through this post for any inaccuracies and handing out some advice on how to improve it.