maandag 12 januari 2015

Let's face it: "Basal Eurasian" is heavily intertwined with African ancestry; might as well stop the collective denial

These non metric attributes all support the view that most of the Neolithic inhabitants of Europe tie more closely together with each other than with the living representatives of the areas in question. The principal exception to this generalization is [...] the Muhlhausen sample, which ties closer metrically to the living inhabitants of the Middle East and North Africa.

The excerpt above describes the North Africa-affiliated "Muhlhausen" LBK sample [Brace et al 2005]. This is the exact same sample Lazaridis et al 2013 would turn to almost ten years later, to sample the now infamous "Basal Eurasian" carrier—Stuttgart. But those who've convinced themselves that the Stuttgart individual is merely an odd looking individual in an otherwise perfectly European looking population, are in for a big surprise.

There have been several reports now, dealing with the genomes of early European farmers, and there is one genetic feature they all seem to have in common. This shared genetic feature was minted "Basal Eurasian" when its existence was first inferred back in late 2013 [Lazaridis et al 2013], and it has been subjected to speculation in the anthro-blogosphere ever since.

While some bloggers seem to have hit the nail on its head in their discourse on this genetic "anomaly", others seem to be hopelessly groping in the dark, coming up with all sorts of far-fetched explanations that can't be reconciled with other data involving early European farmers (e.g. skeletal data), that has been accumulating for decades.

If there is one thing we can be absolutely certain of, it's that a genetic anomaly this consistently found among early European farmers and which is this sizable (inferred to comprise approximately 44% of the genome of the 'Stuttgart' individual in whom it was first inferred), will express itself beyond just autosomal makeup; we'd expect its underlying essence to manifest itself through all sets of ancestry informative markers, including non-genetic ones.

For instance, we'd expect academic descriptions of the skeletal remains of certain early European farmers to contain references to phenotypical traits that are atypical of early Holocene Europe. This is, in fact, precisely what we're seeing.

Skeletal analysis of early European farmers who have little genetic input from local hunter gatherers, are known to exhibit a relative craniofacial similarity with tropically adapted prehistoric North Africans  [Brace et al 2005; Ricaut & Waelkens 2008]. The general rule seems to be: the more prehistoric farmers interact with hunter gatherers, the less they resemble North Africans [Brace et al 2005; von Cramon-Taubadel et al 2013; Gallagher et al 2009].

However, limb index-oriented research such as Galagher et al 2009 may be more difficult to interpret given the essentially unchanged limb proportions of many European hunter gatherers, who often still display strongly elevated limb indices, well into the Mesolithic [Holliday et al 1999]. Hence, full body measurements are required to differentiate between elevated limb indices which result from recent ancestry from warmer regions vs elevated indices which are retentions from the original OOA migrations [Holliday et al 1997a

That the morphometric commonalities turn out to especially involve features historically used to identify African ancestery, is a curious, almost comical result, given past attempts to "flip the script" and attribute this affinity squarely to both populations belonging to a "Mediterranean", and ultimately, non-African, meta-population [Brace et al 1993; Strouhal 1971]. But I'm getting ahead of myself, first things first.

Exotic influences among early European farmers are, in and of themselves, not that surprising. Early European farmers are after all, not native to Europe. Since the popular narrative places their ultimate origin somewhere in or around the fertile crescent, a degree of differentiation relative to aboriginal European hunter gatherers is to be expected, right?

This is where the popular narrative crumbles. Early farmer immigrants aren't merely different from aboriginal Europeans; they also different from other Middle Eastern farmers in a way that intimates that they're only partially related to truly aboriginal Middle Easterners.

What we consistently see is a morphological spectrum along which expanding European farmers plot according to their degree of assimilation with local European hunter gatherers. The farmers most endogamous after having set foot in Europe (e.g. Nea Nikomedea, Mulhausen) plot on one end of the spectrum and West Eurasian hunter gatherers plot on the other end [Pinhasi & von Cramon-Taubadel 2009, von Cramon-Taubadel & Pinhasi 2011].

Not very surprisingly, the exogamous farming colonists who assimilated with hunter gatherers (e.g. some of the LBK farmers) and the aboriginal Europeans who may have adopted agriculture on their own terms, cluster somewhere in the middle.

But what is particularly interesting about these data, for present purposes, is that the used Natufian and PPN samples don't plot as the outgroups in multivariate space we would expect them to, if they truly were ancestral to early European farmers.

Early European farmers (depicted as blue and green centroids) plot as an outgroup to all West Eurasian populations in von Cramon-Taubadel & Pinhasi 2011, necessitating partial ancestry from a non-West Eurasian source which scores similarly negative along PCO I.
By plotting in between European farmer colonists and Mesolithic Europeans, these Natufian and PPN populations create the impression that they and their Levantine ancestors are part of a larger West Eurasian hunter gatherer meta population. Using Lazaridis' terminology: craniofacially, they plot like a WHG-like population, whose intermediate status indicates that they're just another admixture recipient of the ancestral population we're looking for, as opposed to the source.

The faillure of a representative set of seemingly native Palestinian, Syrian and Anatolian samples to plot in a way that is suggestive of ancestor status, is a huge blow to the notion that native Levantine groups constitute the ancestral populations from which European farmer colonists radiated. It also means that that ancestral population is still out there.

Backtracking the footprints of European farming colonists back to their unknown source population

Basically, we're dealing with a "ghost" population (let's call them population X for now) which, before appearing on our radar in a major way in the form of the first European farmer colonists, seems to have primarily made its existence known to us by admixing with the archaeologically visible Natufian and PPN populations. They seem to pop in and out of visibility in the regional skeletal records throughout time and space as follows:

  • At first a pulse from the ancestral homeland of Population X expands into the Epi-Palaeolithic Levant (from elsewhere).
  • At times Population X interacts with the autochtonous populations in the Fertile Crescent, creating seemingly hybrid (i.e. local and non-local) Natufian populations. Think of the Hayonim Natufian sample whose range of various body measurements seems to accomodate the African Wadi Kubbaniya and Nazlet Khater samples better than older Levantine Ohalo II H2 sample [Hershkovitsch et al 1995], and seems completely different from the more cold adapted El Wad Natufians [Holliday 2013].
  • Judging by how radically out of place some Levantine samples seem to be, even moreso than the Hayonim sample, it looks like the main body of the still expanding Population X may have left behind colonies en-route to Anatolia (think of the Natufian Shuqbah sample as described by Arthur Keith). Either that or there were more pulses than the one(s) that eventually set up colonies in Anatolia and Europe.
  • strong footprints of this north-bound population then seem to dissapear from the skeletal record in the region of northern Palestine and Syria. Then, for the first time in several millenia, we encounter populations morphometrically related to the hybrid Natufians again at the doorstep of Europe. At this point, some of them have become what we archaeologically know as the Central Anatolian "Catal Hoyuk" population.

We know Catal Hoyuk descends from hybrid Natufians because this PPN population exhibits the same morphology, albeit more watered down, that seemed completely out of place in the Levant during the Natufian period. The exact same situation in the Levant is now reproduced in Anatolia, in the sense that regional skeletal remains to their west and southwest, that are seemingly mostly biologically native, either possess closer affinity with native West Eurasian hunter gatherers or are intermediate [Pinhasi et al 2003].

It is at this point on their journey along the Mediterranean Basin, that this branch of Population X morphometrically embodies what most early anthropologists would agree is a good, and arguably one of the first, sample to exhibit what they called the "Eurafrican" and "Mediterranean" morphology.

Given its complete anomaly status among native samples in Western Eurasia and its tracable evolution from earlier Natufian hybrids in the Levant, any blogger or academic who talks about "Mediterranean" as a morphometric phenomenon that arose in isolation in West Eurasia, is groping in the dark.

No amateur blogger with self-manifactured narratives about "Mediterraneans" emerging in isolation in the Mediterranean basin can obfuscate the fact that prehistoric European farmers have a very particular and relative relationship with the hybrid Natufians [Brace et al 2005], but not with the seemingly autochtonous Natufians [Pinhasi & von Cramon-Taubadel 2009, von Cramon-Taubadel & Pinhasi 2011, von Cramon-Taubadel et al 2013].

The "isolated Mediterranean" fairytale also can't be reconciled with the fact that the hybrid Natufians group on the side of northeast Africans and European farmers [Brace et al 2005; Schmidt and Seguchi 2008], while the seemingly native Natufians group away from northeast Africans and European farmers and towards West Eurasians [von Cramon-Taubadel et al].

Descriptions of Natufian and PPN remains consistently report post bregmatic depression, among other features consistent with Sub Saharan African ancestry [Meiklejohn et al 1992; Agelarakis 1993; Hershkovitz et al 1994; Bocquentin 2003]. Strangely, despite proximity to Africa, the prospect of these being a marker of recent African ancestry is typically not seriously considered, as in [Meiklejohn et al 1992], where it is treated as an artificial deformation.

A Natufian skull pictured in Bocquentin 2003 exhibiting many features consistent with recent African ancestry, including post-bregmatic depression (see the slight depression along the length of the vault of this skull).

Thus, it seems that the hybrid Natufians (but not the autochtonous ones) have dual ties with both early farmer colonists in Europe, as well as dynastic Nubians and other Africans. This is mirrored in treemix analyses, in which "Basal Eurasian" behaves the exact same way, by forming a "missing link" between Sub-Saharan Africans and prehistoric farmers [Skoglund et al 2014, figure 2b].

If "Basal Eurasians" were Africans, why do we mainly observe Eurasian haplogroups in prehistoric farmer aDNA?

Some bloggers have used the paucity so far of "Sub-Saharan" haplogroups in early farmer ancient DNA, as proving that "Basal Eurasian", or even tropically adapted features, are, in the case of early neolithic Europeans, not external in origin. Others have pointed out that, since "Basal Eurasian" has affinity with living West Asian populations (i.e. the Bedouin sample), it should be regarded as native West Asian in origin.

The first objection is not something someone who has done their homework, would seriously stand by. At present, there is no solid understanding of the dynamics of how haplogroup profiles change throughout the ages. All of our ideas that seem to be working well in explaining haplogroup phenomena come from living populations and they, very tellingly, didn't allow us to predict correctly what ancient populations would be like, genetically.

The Y-DNA associated with early neolithic farmers seems to belong mainly to G, a likely correlate of Lazaridis' WHG component, since it's a West Eurasian Y-DNA haplogroup. The mainly mtDNA N derived haplogroups in European farmers also seem to be correlates of Lazaridis' WHG component, since they're West Eurasian markers.

In other words, we can identify European haplogroups that are analogous to WHG and ANE, but there don't seem to be any non-West Eurasian haplogroups that can be assigned to "Basal Eurasian" (a non-West Eurasian component). Based on the phylogenetic requirements of what it'd mean to be "basal" to Eurasians, detractors would have to find some sort of novel M or N lineage to prove "Basal Eurasian" is not African.

A very tenuous thing to entertain, given the narrow gap between mtDNA M and N on the one hand, and L3 on the other hand. Not to mention, the fact that such lineages have never surfaced in living descendants of Near Eastern farmers. At present, the best translation of the notion of being "basal" to Eurasians in haplogroup language, is belonging to mtDNA L3, L4, L6 and Y-DNA DE (which has been proven to branch from a node that is basal to Eurasian CF [Karafet et al 2008], even though some still perpetuate the self-serving myth that it's a sister clade to CF).

Many people don't know that genes on the mtDNA genome contribute to the generation of cellular (and therefore bodily) heat, among other phenotypes responsive to climatic natural selection [Heulin et al 2011]. This means that individuals whose ancestors originate in a different eco-zones than the ones they inhabit, are potentially subject to purifying selection acting directly on their mtDNA genome (as opposed to just a maladaptive allele) [Heulin et al 2011; Mishmar et al 2003; Huang et al 2014].

This would cause entire mtDNAs of a particular ancestry to dwindle in frequency, with the Y chromosomes of males with ecologically maladaptive mtDNAs suffering the same fate. This potentially explains not only the seemingly skewed pattern we see in early farming Europe (i.e. WHG haplogroup profiles, but partially "Basal Eurasian" genomes), but also the fact that the first ancient E-V13 Y-haplogroup and the most ancient mtDNA L lineages so far have been found in Spain [Lacan et al 2011; Fernandez et al 2005], which has a climate similar to that of coastal North Africa)

Strouhal [Strouhal 1971] cites an anecdote detailing the seemingly absurdly high, lopsided mortality among ethnic Sudanese soldiers in an Egyptian army in 1824. The supposed death toll was 17.000 out of 20.000 soldiers, while the Egyptians remained healthy under similar conditions. Mind you, this supposedly took place in southern Egypt (Aswan), not in the Levant or Europe. But then again, the desert temperatures fall so drastically at nightime in the Sahara that it might as well be.

If one speculates that backfiring efficient heat dissipation (due to their highly tropically adapted bodyplans) was the cause behind much of the ethnic differences in cold resistance, it could be argued that those with maladaptive mtDNA mutations contributing to reduced protection against cold, would be especially vulnerable. However, Strouhal cites pneumonia as one of the causes of death, suggesting there is more to this catastrophy than mere cold.

In any case, I would expect to see a drastically different mtDNA landscape in this population after such an ill-prepared military campaign, possibly with already present "Eurasian" mtDNAs (already having gone through similar selection in the Upper Palaeolithic during OOA), predominating in the surviving Sudanese population.

To illustrate my point, I'm putting up my own minor investigation into the autosomal ancestry of the neolithic Avellaner cave specimen from Spain, and the results seem to support my case. While only 1/13 African haplogroup has been found in this sample (n=7), their probabilities of being Sub-Saharan in the 3 population analysis seem to be much higher than one would predict from the presence of the E-V13 carrier.

Ave01 — 11% probabilty of being SSA (K1a - G2a)
Ave02 — 10.5% probabilty of being SSA (K1a - G2a)
Ave03 — 1.2% probabilty of being SSA (H3 - G2a)
Ave04 — 82.4% probabilty of being SSA (T2b - x)
Ave05 — 1.5% probabilty of being SSA (T2b - G2a)
Ave06 — 23.3% probabilty of being SSA (K1a - G2a)
Ave07 — 10.8% probabilty of being SSA (U5 - E1b1b1a1b [V13])

Microsattelite data taken from [Lacan et al 2011] and processed using popaffiliator

It should be pointed out that, since these specimen originate from Spain, they may have come in contact with the (pre)predynastic Egyptian influenced southern neolithic pulse, which entered mainland Europe via Spain, from Africa. (For those unaware of this migration, it's not a controversial one; there is plenty of data pointing in this direction [Anderung et al 2005; ; Pardinas et al 2014]). 

If so, this would make these results less extrapolatable to other European farmers that entered mainland Europe via the Aegean as some of those SSA probabilities could potentially mean SSA in a literal sense, as opposed to "Basal Eurasian". 

Additionally, it should be noted that I've specifically used the three population analysis (as opposed to the 5 population analysis) because the populations used to model the North African and Near Eastern categories in the latter analysis are not devoid of outside genetic influences. The program is not concerned with correctly inputing ancestry in such a scenario; it merely assigns probabilities.

10 opmerkingen:

  1. This article is interesting: but, since I'm merely a browser, rather than a student of this stuff, it would be difficult to offer any kind of coherent judgment. (Although I've picked up a lot of random info.) Why don't you post it on some of the more popular forums--- like Dienekes, Forum Biodiversity, Rhazib Khan, et. al. ?

  2. Behind the scenes I've had email conversations with personalities in the anthro blogosphere. Generally speaking, they're not receptive to the idea of Basal Eurasian having at least a partly African origin (ironically, some "Afrocentrics" reject the idea as well). Admitting such would just complicate their narrative and invalidate a lot of what they've been writing about.

    For instance, they'd have to admit that there is now a candidate autosomal correlate for Y chromosome E and the Palaeolithic language(s) Afro-Asiatic descends from. Such an autosomal ancestry has so far been elusive, which has led to remarks that Y chromosome E and D can't be African because they're both found in Eurasia without an apparent African autosomal signal. So yes, accepting an African origin for basal Eurasian is a can of worms to certain commentators; it presents a conflict to their pre-existing beliefs.

    It's also the case that I know a lot more about this subject than I'm letting on here, so the issue really isn't uncertain in my mind. I was about to write a paper on this with all my best arguments and material in it, but I ended up shelving the project for various reasons. I might still publish it in the future.

  3. I'm curious to know their justification for rejecting your ideas, not their context of motivation. Anyway, thanks for the response---- I'm just trying to learn a bit more.

  4. It’s not so much that they have specific evidence against what I’m saying. They don’t. They look at the same data and interpret everything differently (see my previous post). So, for instance, where I might say a “Basal Eurasian”-like signal is found in African populations along the African side of the Red Sea, they will look at it differently and see it as evidence of backflow from Eurasia.

    They think Basal Eurasian is somehow a genetic component that has remained isolated in the Levant, Arabia or elsewhere, and that we’ve so far not yet detected the place of origin of this population. Their expectation is that it will show up in Eurasia with enough aDNA sampling.

    Another argument I’ve heard is that this component peaks in Bedouins and not Africans. But this is to be expected in my hypothesis because Basal Eurasian was introduced to Europe via Bedouin-like populations, not directly by any African population. Moreover, while the Bedouin samples preserve their early holocene ancestry fairly well, modern Egyptians don't, so we have to settle for Cushitic speaking Africans, who are relatively flawed proxies for speakers of the ancient Egyptian language (see Pagani et al 2015, where "Ethiopian-specific" and "Egypt-specific" haplotypes are quite distinct, although there is overlap).

    As I said in my previous post, another thing they’ll bring up is that populations with Basal Eurasian don’t have any Sub-Saharan-specific ancestry (this is actually false).

    That’s basically what I keep hearing.

    Hope that answers your question and good luck figuring things out.

  5. Oh, I've also hear the argument that "skeletal remains don't matter" (no matter how much they resemble Nubians), because, according to these commentators, "genetics is everything". I've heard this puzzling response even when I cite something like this:

    "As for the nasal index, the Nahal Homer skulls manifest a low value (48.1) compared to preceding (e.g. Natufian = 52.5; Taforalt = 52.1) and contemporaneous populations (North Africa average = 50.6: Levant average = 55.1), an indication of their long and relatively narrow nasal aperture. Only in two North African Neolithic groups is such a low index found (La Meskiana and Mougharet el Aliya). The Nea Nikomedeia from Greece and Nubia 117 from southern Egypt manifest very similar nasal dimensions (59.4 and 59.7, respectively); however, while the high index in the Nubian group may be explained by Negroid gene flow into the population, the reason for such a high value in a European group is unknown."


    Mind you, this small Greek Nea Nikomedea sample (I think it consisted of four crania at the time this paper was written) is actually the predecessor of the Stuttgart population in which Lazaridis et al first discovered "Basal Eurasian". From the looks of it, they are even more pure carriers of "Basal Eurasian". Note also that the nasal index of the Levantine average was 55%, which is comparable to the Badarians and many Nubian samples. It's certainly much, much higher than the original, Upper Palaeolithic Levantine average.

  6. Very interesting! Llorente et al.'s recent paper purporting to show that there is vast Eurasian introgression into Africa complicates the story you're telling, but I think it complicates everyone's story more than is clear from most discussion online, and I don't think it actually pushes directly against your theory at all. The actual fact is that in 2500BCE there was a relic population in the Ethiopian Highlands, at least one individual of which was not related to the Basal Eurasians, but the Ethiopian Highlands are very much what Johanna Nichols, coming from the historical linguistics angle, has termed an accretion zone. In other words, it does not at all establish when the Eurasian introgression into Africa began, and it does not touch the question of where Basal Eurasian comes from, other than making Ethiopia a poor place to look for it. But despite the highly divergent Afro-Asiatic languages in Ethiopia, this is not where I would be looking for the origin of Afro-Asiatic. As an accretion zone, Ethiopia picks up and retains highly divergent languages, much like the (better-studied) Caucasus is like a poorly-erased palimpsest reflecting all of the language, culture, and gene spreads in its vicinity. Ethiopia was ill-connected to the origination of agriculture, and inventions like the plow and extensive use of grain seem not to have arrived there until after 2000BCE, apparently from Egypt via Nubia. I would identify that time frame as a likely one for the introduction of the pre-blended Basal Eurasian + Near Eastern genetic admixture. If this is correct, it would imply such an admixture had already become current in Egypt, some indefinite period earlier than 2000BCE. This still leaves open the possibility that so-called Basal Eurasian is secretly something like early Egyptian or Nubian--or from somewhere else in Africa--before the absolutely vast quantities of admixture and displacement took place in those cultures over their history. I would be very interested to see how various Ethiopian skeletons plot in your non-metrical analyses, given the new data!

  7. Thanks for your well-thought out reaction, Michael.

    I’m open to more complicated explanations, but I personally don’t see a game-changing quality to this specific aspect of Llorente et al 2015. Based on Hodgson et al 2014 and other data, this was to be expected. In Hodgson et al, the Ethio-Somali component (which is a “Basal Eurasian” variant), is equidistant between the Niger-Congo component and Eurasian genetic material. This is peculiar given the very early East-West African split times, but not when you look at it from the perspective of later contributions from East Africa and the Maghreb in the early Holocene. These contributions manifest in West/Central Africans by their ubiquitous and considerable amounts of L0a, L3f and A3b2-M13 haplogroups and relatively rare U6a5a and U6a5b. Both sets of haplogroups carry an “EEF”-like signal in the respective African source populations, so there is really no need go overboard with scenarios involving G2a carrying European farmers. At least not before the well-documented presence of Kiffian, Tenerean, Hassi el Abiod, Asselar and other Saharan populations have been considered.

    The linking of “additional” traces of Neanderthal in Sub-Saharan Africans reported by Llorente et al, to European farmers also deserves a healthy dose of skepticism, in my view. It seems to me to mostly serve as an extra layer of defense in the event people make short shrift of their hypothesis of European farmers deep in Central Africa. The way I see it, Prufer et al 2013 have already made similar observations, but they were more determined to get to the bottom of this. Instead of just using crude statistics and running with the results as is, Prufer et al went a step further and screened potential Neanderthal fragments to filter out false positives. Some of their observations:

    “Applying this analysis to our data, we find a significant correlation that is strongest when we use West Eurasians as the non-African reference population, suggesting a history of West Eurasian related geneflow into the ancestors of Yoruba. Figure S13.4 shows the decay that we observe when we treat YRI as an admixture of West Africans and Europeans. The decay of LD is highly statistically significant (Z = 5.3 for the significance of detection of admixture LD), and has a genetic distance scale of its decay corresponding to 332 ± 63 generations or 9618 ± 1825 years assuming 29 years per generation

    According to Prufer et al, the percentage of Neanderthal in Yorubans is a meagre 1% of the amount of Neanderthal found in Eurasians. Not only that, but the admixture date in Yorubans rivals the formation of Sardinians and EEF. Thus, EEF may not even have existed when this EEF-like ancestry reached the ancestors of the Yoruba. Not very reassuring for bloggers who insist on linking EEF-like ancestry in West Africans to Bronze Age European farmers.

    Having said that, you’re right that the picture is more complicated. It’s clear from looking at Llorente et al’s D statistics (Prufer et al produce a similar list, BTW) that there is more to this signal in West Africans. Europeans with comparatively more WHG contributions (e.g. Belarussians, Lithuanians) also show up high in these lists. This could be, in some way, related to the fact that West/Central Africans (including Pygmies) consistently tend to have a low level of relatedness – of as yet undetermined direction – to WHG genomes (e.g. Oase I, La brana). To me, evidence of West African contact with prehistoric Maghrebi populations who carried late-WHG haplogroups, such as mtDNA V, T, H and various CRS haplotypes, fits this aspect of the data well.

    Since the East African and Maghrebi haplogroups in West/Central Africa (L0a + L3f + A3b2-M13 and U6a5a + U6a5b, respectively) are all incipient/early Holocene in age, they call to mind Prufer et al’s Yoruban LD admixture event of >9ky. These haplogroups could represent either a mixed Maghrebi + East African population (e.g. possibly a population like Erg Ine Sakane) or two distinct events.

  8. Deze reactie is verwijderd door de auteur.

    1. I deleted your spammy double post. But I'm going to leave this reply because it shows that some of my detractors are grunts with one-word sentences who couldn't think up a decent counter argument to save their lives.