Monday, February 3, 2014

Sources Of West Eurasian Ancestry In Eastern and Southern Africans (updated twice)

A new paper by Pickrell (preprint discussed here in 2012) observes that click language speaking, historically hunter-gatherer populations of Southern Africa (the Khoi-San) have West Eurasian admixture (in the extreme case as much as 15% but more often on the order of 5% or less) dated to 900 to 1800 years ago and predating Dutch and English colonists who gave rise to modern South Africa.

He concludes that this is probably derived from the migrations of people from East Africa.  There, in there, there was substantial West Eurasian admixture ca. 2,700 to 3,300 years ago, around the time of the arrival of Ethio-Semitic languages in Ethiopia from Southern Arabia.

I would add a few observations.

1.  Comparability to East African Hunter-Gatherer Percentages.  The levels of West Eurasian admixture seen in the Khoi-san of Southern Africa is similar to that seen in the Hazda and Sandawe who are click language hunter-gatherer populations of East Africa.  Bantu languages in Mozambique, intermediate between these two groups, also have clicks in their Bantu dialects that were presumably inherited from substrate languages that preceded Bantu expansion there.  The lack of much dilution of West Eurasian ancestry relative to hunter-gatherer populations in East Africa is notable.  It indicates a migration rather than mere diffusion of West Eurasian genes over time (something also supported by the lack of similar West Eurasian ancestry in geographically intermediate parts of Africa).

2.  Push Factors.  These hunter-gatherer population migrations were probably more than mere random wanderings.  Hunter-gatherer populations in East Africa may been retreating from Ethio-Semitic peoples, from later waves of Bantu expansion, or even from Nilo-Saharan expansion in East Africa.  Their exiles may have been a pilot wave in advance of the expansions of groups that cast them out of East African homelands.

3.  West Eurasian Ancestry In East Africans Probably Arose In At Least Three Waves.  Pickrell supposes that the nearly 50% West Eurasian ancestry found in Ethiosemetic peoples in the vicinity of the former D'mt Kingdom of Eritrea and Ethiopia happened in a single event on the grounds of maximum parsimony.

Pickrell relies to a great extent in doing so on a 2012 study by Luca Pagani, et al., so criticism of this part of Pickrell's paper really amounts to criticism of Pagani rather than Pickrell who merely relied on Pagani's findings.

In fact, other evidence makes such a simple scenario unlikely.  Instead, West Eurasian ancestry in East Africa probably arrived in at least three waves: an Upper Paleolithic back migration from West Asia, an early Neolithic back migration from West Asia, and an Ethiosemetic migration from Southern Arabic.

Moreover, the ancestry seen in the Khoi-San, is probably derived largely from the first and second waves, rather than the third.  As discussed below, the source population was from the Ethiopian fringe where Ethiosemitic influence was lightest.  These populations may have experienced only two waves of West Eurasian ancestry.

Some of the evidence includes the following:

* Methodology Mutes Impact Of Prior Waves.  The method used to date the East African-West Eurasian admixture event tends to obscure or ignore waves of admixture other than the most recent one.

* Known Upper Paleolithic Back Migration.  We know that sometime in the Upper Paleolithic era that there was a back migration to North Africa and East African that brought with it mtDNA haplogroups M1 and U6.  This may have been on the order of 10%-20% of the populations at the time - U6 has a more North African distribution, while M1 has a more East African distribution (mtDNA charts via Ethiohelix are informative on this point).

* Chadic-Cushitic Links Indirectly Date Cushites And West Eurasian Contact.  Archaeological evidence suggests that the Chadic people made their way to Lake Chad, after a period of depopulation of the region during an arid climate period, around 5,200 BCE as part of the Gobero culture.  These populations show tighter mtDNA similarities to Cushitic peoples with whom they share pottery designs and words for domesticated animals, suggesting a split of Chadic people away from Cushitic people around that time in the vicinity of Sudan.  But, the Chadic people have a distinct R1b-V88 dominanted Y-DNA profile from that of Cushitic people.  Thus, proto-Chadic men may have taken Cushitic women as wives and then migrated West to Lake Chad.  mtDNA distributions in Northeastern, Eastern and Central Africa, however, tend to follow regional patterns except for one mtDNA L3f clade particular to Chadic peoples and derived from a Cushitic clade.  The TMCRA time depth of the Chadic specific Y-DNA clade and Chadic specific mtDNA clade are consistent with this archaeologically supported date.

* Autosomal Genetic Clustering.  Autosomal evidence points mostly to three clusters - a Northern Berber-Egyptian-Semitic cluster, a Southwestern Chadic cluster, and a Southeastern Cushitic cluster, with each almost equidistant from each other.  All of these Afro-Asiatic populations in turn tend to cluster away from more southerly linguistic populations.  Thus, Northern Afro-Asiatic speakers are part of a significantly West Eurasian cluster, while Chadic and Cushitic populations have less West Eurasian ancestry, and the Chadic population having different kinds of West Eurasian ancestry than other Afro-Asiatic populations.

* High West Eurasian Ancestry In Cushitic Ethiopians.  Many Cushitic populations have as much as about 35% West Eurasian admixture.  This is 10%-15% less than that of Ethiosemitic populations closer to South Arabia, but too small a discrepancy to be accounted for by an initial admixture of South Arabians with Ethiosemitic peoples that then spills over to neighboring Cushitic populations.  This level of impact would be on the same order of magnitude as the Turks in Turkey, or the Indo-Aryans in India.  This would be more than the demographic impact of the Uralic population that brought Hungarian to Hungary (almost nil), less than the demographic impact of the Yaoyi who brought Japanese to Japan (on the order of 50%).

A more plausible scenario is that Cushitic populations had roughly 25% to 35% West Eurasian admixture prior to the arrival of the proto-Ethiosemetic peoples, and that Ethiosemitic invaders added a gender biased additional 10%-15%.  Meanwhile, in the pre-Ethiosemitic Cushitic population, there may have been a 5%-20% or so West Eurasian admixture percentage traceable to the Upper Paleolithic, with the balance appearing with the arrival of the Neolithic revolution probably via the Nile Valley into the Blue Nile basin.

The window of time during which the Neolithic flow of West Eurasian ancestry into Cushitic populations is quite narrow.  It had to predate the emergence of the Chadic people around 5200 BCE, and had to post-date the arrival of the Neolithic revolution in Egypt sometime after 8000 BCE.

* Neolithic Source West Eurasian DNA Likely In This Case.  It is this early Neolithic infusion of West Eurasian ancestry, rather than later Ethiosemitic contributions that probably gradually spilled over into neighboring pre-Cushitic hunter-gatherer populations with click languages over a period of more than 3,000 years.  (Niger-Congo languages were probably not spoken in East Africa until Bantu expansion and Nilo-Saharan languages were likewise probably a late arrival, since there is very little evidence of genetic profiles associated with these languages being present in East Africa until very recently.  These genetic profiles are not a background present in most linguistically Afro-Asiatic populations of the region.)

This is consistent with Pagani's finding that the East African source population is closer to Levantine than Arabian populations, but inconsistent with his findings on the timing of admixture and the number of admixture waves involved. Neolithic source populations were probably from the Levant.  Ethiosemitic population sources were almost surely from South Arabia.

Phantom West Eurasian autosomal DNA profiles for Cushitic and Khoi-San populations can help us to reconstruct genetic profiles of first wave African Neolithic migrant populations.

* Y-DNA T related clues.  Y-DNA haplogroup T is common in Somolia and to a lesser extent in adjacent parts of Ethiopia.  But, the age of the Y-DNA haplogroup T subtypes in Arabia such as Oman are too young (ca. 1600 years) to have been the source for Y-DNA hg T in Ethiopia not less than 2700 years ago with the Ethiosemitic languages.  Instead, given the much older ages of Y-DNA haplogroup T clades in Turkey and Egypt, it is fair to assume that Y-DNA hg T made its way to Ethiopia from the Levant via the Sinai and Nile River.  Thus, significant West Eurasian ancestry evidence by Y-DNA hg T must have arrived in Ethiopia considerably earlier than 2700 years ago, and quite likely during the early Neolithic.  Y-DNA hg T bearers appear to have been farmers, and why would farmers settle in the Arabian desert prior to entering Africa from there?

4.  Copitc was probably a parent language to Cushitic from which Chadic diverged.  Increasingly, the genetic, linguistic and archaeological evidence, taken as a whole, point towards the Neolithic expansion of the ancient Egyptians who spoke a Coptic language as the parent language family to the Cushitic, Omotic, Chadic and Berber language families, and quite possibly, of the Semitic language family as well, within the Afro-Asiatic branch of the language this macro-language family.

Ancient Egyptian may in turn have had origins in a Levantine Fertile Crescent language, but that proto-language, if there was one, was probably obliterated by a Semitic derivative of ancient Egyptian's back migration to the Levant from Egypt evidence by the Afro-Asiatic specific Y-DNA E clades found there today and supported by historical evidence of long periods of Egyptian political dominance in much of the Levant.  The indications of Y-DNA E migrations into West Eurasia are rare exceptions of Africa to West Asian gene flow when mostly since the Upper Paleolithic, the direction of the flow has been back to Africa.

Also, notably, the linguistic tree of Semitic doesn't show much sign of diversification or a bushy structure prior the first attested instance of a Semitic language in the form of Akkadian, just a few hundred years after the ancient Egyptian Civilization really started to reach an advanced state ca. 3000-3500 BCE.

On the other hand, a possibly Levantine proto-Afro-Asiatic that may have given rise to Coptic ca. 8000-7000 BCE as part of a first wave Neolithic expansion, may also have survived and been, in a later form, a substrate to the branch of Coptic that became proto-Semitic ca. 3000-3500 BCE.  So early Semitic languages may be as close to proto-Afro-Asiatic as we can discern with historical linguistic methods, since its departures from Coptic might reflect retention of words and structures from a proto-Afro-Asiatic language.

UPDATE 2/04/2014 based on review of supplemental materials:

5.  The single source East African population for West Eurasian ancestry in the Khoi-San had 25% West Eurasian ancestry, consistent with modern Omotic populations on the Ethiopian fringe.  The Khoi san samples very consistently demonstrate that West Eurasian ancestry is 25% of overall East African ancestry.  This is less than is seen in linguistically Semitic or Cushitic populations, is within the range of linguistically Omotic populations, more than the Maasai, but less than another Nilotic population which is near zero, and more than the Hadza, Sandawe, a couple linguistic isolates and a Bantu population.

In other words, a 25% West Eurasian ancestry percentage in the source population for the Khoi-San suggests an East African source population on the Ethiopian linguistic fringe, rather than the Semitic or Cushitic heartland of Ethiopia.  It also strongly suggests that a single source population is responsible for West Eurasian ancestry in all Khoi-San individuals.

Notably, however, the type of West Eurasian ancestry in the Khoi-San matches the phantom ancestry inferred from East African populations more closely than any modern population, strongly confirming this sourcing for that ancestry.  Compared to modern populations that matches are closest to Southern European and Middle Eastern populations.

UPDATE 2/06/14 based on further analysis:

6.  The Ethio-Semitic demographic impact on Ethiopia may have been close in time to a demographic impact on Ethiopia associated with the arrival of domesticated cattle, obscuring the fact that there were actually two separate and parallel migration waves involved.

The notion that most of the West Eurasian admixture in Ethiopia appeared ca. 2700-3300 years ago is supported by a 2012 study by Luca Pagani, et al., but I have criticized it above, arguing for a Neolithic wave before an Ethio-Semitic wave.  But, these two waves may actually have been parallel or close in time to each other.

At least, anyway, cattle herding accompanied by a major demographic impact may have arrived close in time to the Ethio-Semitic languages may have arrived ca. 2000 BCE-1500 BCE.  If admixture between newly arrived cattle herders and existing Ethiopian populations didn't take off until several centuries after these populations started to co-exist, this could help fit the Luca Pagani admixture dates to a historical source of this admixture at around the correct time in linguistically non-Semitic Ethiopians.

How could this be?

First off, the Cushitic-Chadic split may have happened in Sudan, before Cushitic languages had advanced up the Blue Nile into Ethiopia proper.

Second, as summarized in a previous post here, where it was noted that herding arrived to East Africa later than one might naively expect (emphasis in bold added):
In east Africa pastoralism became established late in the third millenium BC, and over the next 1000 years domestic livestock slowly moved towards Tanzania. These cattle were probably humpless, for the earliest evidence for humped cattle comes from ancient Egyptian paintings dated to around 1500 BC. These authors believe that cattle with cervico-thoracic humps (i.e. with humps on their necks) were first introduced into the Horn of Africa. . . .

The quite late arrival of pastoralism in East Africa, which is also some to relict hunter-gatherer populations which are absent in North Africa and in West/Central Africa to the North of the Congo jungle, suggests that herding (and probably farming as well, at least in less arable land) may have come much sooner to North Africa and West Africa than to East Africa (with Southern Africa not experiencing agriculture until the Bantu expansion arrived there after it had swept the rest of the continent). Indeed, the arrival of cattle herding in East Africa could conceivably have predated Bantu expansion by only a thousand years or so.

If one makes the plausible assumptions that the ancestors of Cushitic populations predate Nilo-Saharan populations in East Africa who in turn pre-date Bantu populations in East Africa, and that all of these language families were spread by cultures that employed herding or farming, the late arrival of cattle herding to East Africa (ca. 3000 BCE to 2000 BCE), the fairly well dated arrival of Bantus in East Africa (ca. 1000 BCE-500 BCE), and the known paleoclimatic wet sahara/dry sahara/chad basin size changes quite tightly constrain and compress the timing of Cushitic and Nilo-Saharan expansions in East Africa. It also probably puts the time that the source cultures for these language families arrives in East Africa sometime after the ethnogenesis of the culture that gave rise to the Niger-Congo languages that are dominant in West Africa (which necessarily predated the expansion of its Bantu branch of that language family ca. 1500 BCE-1000 BCE).

Tishkoff argues based on genetic evidence associated with the digestion of cow's milk by adults and archaeological evidence that:
Cushitic-speaking Afro-Asiatic populations . . . are thought to have migrated into Kenya and Tanzania from Ethiopia ~5,000 years ago and practice a mixture of agriculture and pastoralism . . . [while] the Nilotic-speaking Nilo-Saharan populations. . . are thought to have migrated into Kenya and Tanzania from southern Sudan within the past ~3,000 years and are strict pastoralists.
Thus, according to Tishkoff's account, Nilo-Saharan linguistic populations arrived in East Africa proper only a few centuries before the Bantu expansion reached the area, and the Cushitic population was expanding south from Ethiopia only shortly after Egypt and Sumeria developed written languages at about the same time as the high point of the Indus River Valley civilization, although their arrival in or ethnogenesis in Ethiopia clearly must have happened earlier. 
Tishkoff doesn't reach a conclusion about whether a milk digestion gene specific to Africa arose first in Cushitic or Nilotic populations, but another quote from the same article strong suggests that the Cushitic people, rather than the Nilotic population was its source (and hence that the Cushitic people had cattle farming): "The absence of C-14010 in the southern Sudanese Nilo-Saharan–speaking populations suggests that this allele either originated in or was introduced to the Kenyan Nilo-Saharan populations after their migration from southern Sudan."

Tishkoff also recaps the archaeological evidence for cattle domestication in Africa:
Archeological evidence suggests that cattle domestication originated in southern Egypt as early as ~9,000 years ago but no later than ~7,700 years ago and in the Middle East ~7,000-8,000 years ago, consistent with the age estimate of ~8,000-9,000 years (95% c.i. ~2,200-19,200 years) for the T-13910 allele in Europeans. The more recent age estimate of the C-14010 allele in African populations, ~2,700-6,800 years (95% c.i. ~1,200-23,000 years), is consistent with archeological data indicating that pastoralism did not spread south of the Sahara and into northern Kenya until ~4,500 years ago and into southern Kenya and northern Tanzania ~3,300 years ago. . . 
The main crops of the West African Sahel and some of the Ethiopian Highland crops are indigenous African domestications that were not part of the Fertile Crescent agricultural package. Fertile Crescent crops are a poor fit for parts of Africa beyond the Mediterranean fringe and the Nile basin; they are confined to regions further north than the regions where Fertile Crescent livestock can thrive without evolving to adapt to African weather and diseases. The quoted post dealt only with cattle husbandry, and one needs to look at dating for goats and sheep herding and for indigenous African crops to reliably date the West African and East African Neolithic (I've seen some references, that I'll have to track down, that have proposed an early date of arrival for these herd animals than for cattle.) It could be that early African agriculture didn't include cattle until other parts of the agricultural package were better established.

One possibility that could reconcile the various lines of evidence, and also relax some of the tight constraints in timing discussed above, is to envision the Cushitic languages expanding to the herding of sheep and goats, and the Nilo-Saharan expansion as being boosted by the addition of cattle husbandry to the Cushitic food production package (although, as noted above, Tishkoff's sources, at least, doubt that scenario). . . .

Recent East African autosomal genetic data also makes clear that there was not any significant demic impact of West African Niger-Congo populations in East Africa prior to the Bantu expansion into that region, although the domesticated crops of West Africa and East Africa respectively a testaments to the mutual exchange of locally domesticated plants between the regions at some point in time. . . .

Another implication of this narrative, if it is accurate, is that within the Afro-Asiatic language family, the Berber and Coptic language families probably significantly predate the Cushitic and Chadic language families. (These speculations don't resolve one way or the other the relative origins of Berber and Coptic languages on one hand and the Levantine Semitic languages on the other, although Ethio-Semitic languages surely broke off as a branch of the Semitic languages after all of the language families within the Afro-Asiatic macro-language family were established, probably sometime in the Bronze Age).


Davidski said...

Italians (from Bergamo) are supposedly the best modern proxies for the ancient West Eurasian admixture in East and South Africa. They're in third place after the two inferred ancestral populations (see Fig 1C). Something doesn't add up there. In theory, it should be the Sardinians.

andrew said...

Sardinians are a good proxy for first wave Neolithic farmers in Europe. All such farmers, presumably derive from areas on the border between the Fertile Crescent Neolithic (i.e. initial West Eurasian Neolithic which really extended into some of the fringes of SE Europe), near the Northernmost extent of the Fertile Crescent.

The Neolithic revolution arrived in Egypt and the Indus River Valley about a thousand years before it arrived in Europe. And, in the case of Egypt, presumably involved migration from the Southern Levant, near the far Southwestern tip of the Fertile Crescent.

It does not surprise me that separate waves of migrants whose origins are separated by a thousand years and about 700 miles would have different genetic profiles.

Perhaps Bergamo received significant gene exchange with North Africa and Egypt during the ancient Roman era that pulls it in the direction of the West Eurasian component the made its way into East Africa, while its location may have shielded it from Central Asian genetic influences.

andrew said...

It is also worth noting that Y-DNA haplogroup E, sister clade to Y-DNA haplogroup D, emerged almost immediately after the Eurasian CF clade. But, Y-DNA E peoples in African do not as a rule have Neanderthal ancestry, while Y-DNA D people in Europe do. This implies that Y-DNA E probably arose either in Africa, or back migrated from Arabia prior to Neanderthal admixture occurred in its subpopulation of proto-Eurasians (and given the mtDNA associated with it, also before mtDNA M and N emerge).

E looks like a parallel African development to CF in Eurasia at about the same time. D maybe emerges from the DE clade in Africa, perhaps as a wave after CF leaving it to marginal territory.

Joe Lyon said...

E does have beanderthal ancestry. It s ust populTions in subsaharan africa dominated by a or b that dont. They even probably have a very tiny trace. E backmigrated into africa afterbgetting an extra dose of neanderthal genes, but moving into mostly non-neanderthal africa diluted it considerably. Likewise when F absorbed all that non-neanderthal hss dna in india and south chiba, neanderthal levels in F got diluted lower than the amount in haplogroup C.