RSS
Kalash girl
A new paper in The American Journal of Human Genetics, The Kalash Genetic Isolate: Ancient Divergence, Drift, and Selection, illuminates and obscures the history of this enigmatic people. Some framing is necessary here for why the Kalash are important. The Kalash are a “pagan” people who live in the uplands of Pakistan. By pagan, I mean to say that they preserve the primal religious traditions of a strand of the Indo-Iranian peoples, untouched by Islam, or, later developments which led to “higher religions” which arose directly out of Indo-European religion, Zoroastrianism and Hinduism (Sanatana Dharma). It would perhaps be defensible to depict the Kalash as the pagans of yore, a fierce people unbowed by philosophical monotheism or the quietism which explodes out of aspects of the gita.
Which brings us to another peculiarity of the Kalash: they are white. By white I do not mean white Europeans. They are not. The genetics is not in dispute, the Kalash are distantly related to the other peoples of South Asia. Some South Asians remind white Europeans (and also white West Asians) of themselves when they look at them face to face. But this tendency is heightened in isolated mountain peoples, such as the people of the Chitral valley. Among the Kalash it is more the norm the exception, ergo, legends of descent from the armies of Alexander the Great. In a previous age this paradox of an exotic and pagan barbarian people whose external appearance was white was utilized in fiction. Rudyard Kipling’s novel The Man Who Would Be King is set among the people of “Kafiristan,” what is today called Nuristan. Eight years after the publication of Kipling’s book the people of this region were forcibly converted to Islam by the king of Afghanistan. Even today if a Westerner wants to “pass” as an Afghan it is mostly plausibly as a Nuristani, because some among these people look to be Western in their outward appearance. The Kalash people were under British rule, and so were shielded from conversion to the religion of peace. Today the Kalash are surrounded by territories infested with Pakistani Taliban. Though protected by the state of Pakistan and vigilant against interlopers, it still seems unlikely that they’ll pass through the next generation unconverted.
The Kalash Genetic Isolate is open access, so I invite you to read it. I saw part of the above figure at ASHG 2014. The important aspect of this paper is that it confirms that the Kalash have a great deal of “shared drift” with MA-1, the canonical individual which represents the ancient North Eurasian people who contribute ~10-20% of the ancestry of Northern Europeans and 30-40% of that of Native Americans (and nearly as much as some Caucasian peoples). Unfortunately the tables don’t show f3 statistics of each population, so we aren’t totally clear which population is which in the ternary graphs. But we can make some guesses. The outlier South Asian group is almost certainly the Sino-Tibetan Sherpa group. The South Asian groups include the Gujarati sample from the 1000 Genomes, as well as HGDP populations such as the Sindhis. The West Asians are Iranians, Palestinians, Turks. etc. if this is correct it seems to depict South Asians as sharing a great deal of drift with MA-1. There is also a second plot which shows that Kalash share a great deal of drift with La Brana, the Mesolithic hunter-gatherer from Spain. In contrast to the result from MA-1 it does not seem that other South Asian groups share this drift. To me this is tentative support for the contention in last year’s Science paper that there was some gene flow from Europe to the Kalash over the past few thousand years.
But I need to end here on a down note. Though a lot of the results in this paper are fine, the interpretation strikes me as totally out of kilter with their own citations! They say:
LD decay showed that the Kalash were the first population to split from the other Central and South Asian cluster around 11,800 (95% CI = 10,600−12,600) years ago. This estimate remained constant even after the addition of an African (YRI) population or when the Kalash were compared with different subsets of non-African populations. The pairwise times of divergence with other Pakistani populations ranged from 8,800 years ago with the Burusho to 12,200 years ago with the Hazara.
Most of the populations and clusters that they are speaking of here did not exist when the divergence has been adduced. The Hazara for example are a compound population which emerged in the last 1,000 years due to the admixture of Mongols upon a Persianate substrate. The Uygurs are similar. The “Central” and “South Asian” population genetic clusters are refications of admixed groups which have emerged in the past ~4,000 years. That is, thousands of years after they purportedly diverged from the Kalash. The problem here is that the authors keep forcing their interpretations into a tree, when population genetic history for humans in the Holocene has not been a tree at all. As outlined in Towards a new history and geography of human genes informed by ancient DNA it is plausible that every major group of humans today (major = numerous) is the product of fusions of branches of the human race which were sharply diverged during the Pleistocene. The genomes of individuals and peoples then represent a complex and reticulated graph of interlaced histories. Reducing them to branching trees obscures rather than illuminates.
The deep divergences being inferred here strike me as likely a function of the fact that the authors do not take into account that South Asian populations are themselves a compound of two very distinct groups. One of these, the “Ancestral South Indians” (ASI) are very diverged from the groups of western Eurasia (and, the “Ancestral North Indians”, the ANI). The peoples to the south and east of the Kalash have much higher fractions of ASI, so the calculation of a divergence that is >10,000 years before the present is simply reflecting the very deep divergence of the ASI ancestry from the West Eurasian heritage of the Kalash (note, it is important to remember that the Kalash also have ASI, but just at lower levels).
Overall, this is an interesting paper. There are notable nuggets in it. For example, phenotypically the Kalash are lactose tolerant, but they lack the common Eurasian variant in totality. That implies that there is another variant in the LCT region unique to the Kalash. This also implies that the Eurasian variant has spread relatively recently into Northwest South Asia, perhaps post-dating the arrival of the Indo-Aryans! But the discussion is marred by the straightjacket of tree-thinking, imported from macroevolutionary contexts into a population genetic one, where it is less useful.
English
Arabic
Chinese (Simplified)
Dutch
French
German
Hebrew
Hindi
Italian
Japanese
Korean
Persian
Polish
Portuguese
Russian
Spanish
Swedish
Turkish
Vietnamese
Everything you’ve written on the genetics is absolutely spot on. I just have one minor fact to add, which is very insignificant in the grand conceptual picture that is being drawn here.
Mainly, the Kalash don’t have lower levels of ASI in comparison to their immediate neighbors to the east and south (but if by east and south you really mean Punjabis and Sindhis, then yes).
As you know, Haak et al. utilized a new method with f4 statistics, for the purposes of attaining unbiased estimates of admixture in modern Europeans (with regard to Neolithic and Bronze Age aDNA samples). This method is what we have with qpAdm, part of the recent Admixtools release. This software is one of the big pragmatic “leaps” to be seen in that paper. One can try to model the Pashtun samples from the HGDP data-set, with this software. The best fit so far is this:
54% Yamnaya (presumably, late Proto-Indo-Europeans) + 32.3% BedouinB + 13.6% Dai
The Dai are good enough proxies for ASI. According to the original Reich et al. paper, “ASI” shares more drift with Dai than with Onge (even though Onge and “ASI” are a clade)!
Also, this fit is basically perfect, from a purely stat-based perspective. The lower the chisq, the better. And the chisq for this fit is only 0.491. To put that in perspective, most European populations don’t get such low chisq numbers, despite the inclusion of the now abundant European aDNA data we have at our disposal. On top of that, the tail probability is very good, and the standard errors are very low.
David at Eurogenes also tried to fit the HGDP Pashtuns with qpAdm, and found this fit:
47% Yamnaya + 43.2% Starcevo_EN + 9.8% Dai
He didn’t mention the technical details, but did note that this is a very excellent fit, works great.
Basically, the HGDP Pashtuns are around 9%-14% total ENA, most of which is probably ASI, but with the inclusion of some Siberian ENA as well.
With the aDNA we have now, and with the new methods we have now, the original Reich et al. estimates of “ASI” have been proven to be extremely bloated. In fact, Zack’s ADMIXTURE run with the “Onge-component” was far closer to the truth, in terms of absolute levels of South Asian-specific ENA (to be honest, I prefer “South Asian-specific ENA” to “ASI”).*
As this relates to the Kalash, they are also around 9%-14% ENA. You can see that in terms of shared drift, as shown in the paper. On the Han axis, the Kalash behave like other South Central Asians. On the paper’s PCA, the Kalash are as East Eurasian shifted as the HGDP Pashtuns, but deviate in their own direction in PC2.
On a completely different note, the neighbors of the Kalash are just as depigmented, if not more. An old physical anthropological study found that 30% of Kalash had eyes which were blue, green, or heavily mixed between those shades and brown. But the Kho of Yasin, Pakistan were at 55% (!), and a subpopulation of the Shin of Gilgit were at 40%. In terms of skin color, 70% of the Kalash had “brunette white” unexposed skin color, while 30% had very “light brown” unexposed skin color. But the Dardic population in Mastuj, Pakistan was at 80% “brunette white” unexposed skin color, 15% “pinkish white” unexposed skin color, and 5% very “light brown” skin color. In terms of physical appearance, all the peoples of mountainous northern and northwestern Pakistan are like the Kalash in terms of pigmentation.
*I call it “South Asian-specific ENA”, but qpAdm gives Iranians quite a substantial helping. In fact, qpAdm has Pashtuns and Iranians at fairly comparable levels of ENA.
Also, I just want to toot my own horn, just for a second. I predicted the ANE-South Asian connection months ago.
Then again, anyone who followed David’s work could have done so. David has done tremendous stuff with the aDNA data, and much of what he does is eventually verified in some scientific paper.
@comentator Is there a test on GEDMatch or any other service that lets you compare yourself to these ancient populations or individuals like Yamnaya people, or the La Brana sample?
I have on GEDMatch about 45% ENF and 45% WHG with about 5% East African and 5% ANE. Wondering if I could get results that give me any more fun info like this.
Razib, thanks for the summary. I’m one of those Europeans who was struck by this phenomenon in the mountains of Pakistan and the adjoining countries: Afghanistan, Tajikistan, and even a bit of China. While it is true that only a minority of these mountain people appear European, it is a common enough thing that in these areas, a person of European descent notices that they are suddenly much less regarded as “the other” than they would be in Kerala or even in Tashkent.
I am especially appreciative of your combating the tree paradigm of genetics. It seems to me there is a lot politics masquerading a genetics. Over these thousands of years, the dynamics have to be of splitting groups AND combining them and having new groups formed. In this case, since you and the previous commenter have noted that the % of European-looking folks is high among other mountain people s in this region, I’d like to posit that appearance might have a strong component of natural selection based upon environment. In the Karakoram valleys, home to the Shins, Burushos, Wakhis, and Baltis, all of whom produce the occasional light haired/eyed/skinned person, they have glaciers that are largest outside of the polar regions. This isn’t the South Asia of mangoes and coconuts, this is the South Asia of trout, apricots, and morels. These people have adapted to what has to be called a Nordic-style environment and I think we should expect a few so-called “white” people to appear over the millennia even if Alexander’s troops didn’t settle there.
Phenotype wise, they look Middle Eastern to me, some of them very fair. Some are pretty dark- pigmented. Sort of like Kurds. If they average above 10% blue eyes, that’d be quite above ME values.
I am especially appreciative of your combating the tree paradigm of genetics. It seems to me there is a lot politics masquerading a genetics. Over these thousands of years, the dynamics have to be of splitting groups AND combining them and having new groups formed. In this case, since you and the previous commenter have noted that the % of European-looking folks is high among other mountain people s in this region, I'd like to posit that appearance might have a strong component of natural selection based upon environment. In the Karakoram valleys, home to the Shins, Burushos, Wakhis, and Baltis, all of whom produce the occasional light haired/eyed/skinned person, they have glaciers that are largest outside of the polar regions. This isn't the South Asia of mangoes and coconuts, this is the South Asia of trout, apricots, and morels. These people have adapted to what has to be called a Nordic-style environment and I think we should expect a few so-called "white" people to appear over the millennia even if Alexander's troops didn't settle there.Replies: @Karl Zimmerman, @Twinkie
Keep in mind that even if temperatures in the mountain valleys stretching through this area are low by tropical standards, the total solar flux is not particularly low. In the region it can vary dramatically depending upon area – probably due to dramatic differences in cloud cover (e.g., microclimates due to topography, similar to San Francisco). In most places, though, the level of solar irritation is significantly above any European levels except for Iberia and some other areas around the Mediterranean.
That said, the relatively poor/seasonal climate does mean that people undoubtedly wear more clothing than people in the lowlands. They probably avoid going out in the winter time as well. Presuming there actually has been selection for pale skin to prevent Vitamin D deficiency, this difference may be enough to ensure that the paler variants in the initially mixed Indo-European populations won out in the highlands.
Edit note – I can’t figure out why my href link results in all caps. Must be something internal to the commenting system. Apologies.
I've been told more than once that Tajiks in the Pamirs are much more likely to have brown or even red hair than Tajiks in the lowlands. Has there ever been a genetic comparison of the Tajiks of the Pamirs and, say, the Fergana Valley?
Does anyone know of any relatively recent papers or books on Kalash religion? Everything I’ve seen so far is at least 100 years old.
“the Kalash are lactose tolerant, but they lack the common Eurasian variant in totality. That implies that there is another variant in the LCT region unique to the Kalash.”
Reminds me of the second half of this post: https://westhunt.wordpress.com/2014/05/19/phenotypes-vs-genetic-statistics/
http://www.people.fas.harvard.edu/~witzel/KalashaReligion.pdf
A list of texts on kalasha language + religion in English is maintained here
http://kalashapeople.blogspot.com/p/books.html
A discussion of Kalsha language is provided by Elena Bashir in Ch. 22 of the following book;
The Indo-Aryan Languages
By George Cardona, Dhanesh Jain
It has some discussion of religion in pages 905-990
Nature in the Kalasha Perception of Life by Birgitte Sperber, a chapter in a book entitled "Asian perceptions of nature: a critical approach" by: Ole Bruun and Arne Kalland.
Natural Resources and Cosmology in Changing Kalasha Society, A Book By Mytte FentzReplies: @Yudi
Razib – how do the Burusho fit in here? They’re very high in R1a1a but non Indoeuropean. The reason I ask is I’d think there’d be other non-Indoeuropean influences underlying the region to various degrees.
two hypotheses come to mind
1) R1a1a was not limited to IE people
2) gene flow from IE groups into these groups
3) some of both
i lean more to 1 than 2 in my ratio, but i'm at low confidence.Replies: @CupOfCanada
In this regard the Burusho don't differ in any way from Indo-Iranians.
Razib – how do the Burusho fit in here? They’re very high in R1a1a but non Indoeuropean.
two hypotheses come to mind
1) R1a1a was not limited to IE people
2) gene flow from IE groups into these groups
3) some of both
i lean more to 1 than 2 in my ratio, but i’m at low confidence.
My, my, the heirs of an Alexandrian conquest, the heirs of a Genghisid conquet, this has it all.
This seems to be a recurring theme of yours. How much confidence do you place on this plausibility?
What happens if you separate the constituent genetic inputs of today’s Hazara (to the extent that the ancestral inputs can be ascertained) and then re-assess the genetic distance to Kalash? What do you see then?
"This seems to be a recurring theme of yours. How much confidence do you place on this plausibility?" in response to 'it is plausible that every major group of humans today (major = numerous) is the product of fusions of branches of the human race which were sharply diverged during the Pleistocene'.
I'm surprised you have difficulty accepting Razib's comment. I would have thought by now most would have accepted the overwhelming evidence that it is so.Replies: @Razib Khan, @Twinkie
I am especially appreciative of your combating the tree paradigm of genetics. It seems to me there is a lot politics masquerading a genetics. Over these thousands of years, the dynamics have to be of splitting groups AND combining them and having new groups formed. In this case, since you and the previous commenter have noted that the % of European-looking folks is high among other mountain people s in this region, I'd like to posit that appearance might have a strong component of natural selection based upon environment. In the Karakoram valleys, home to the Shins, Burushos, Wakhis, and Baltis, all of whom produce the occasional light haired/eyed/skinned person, they have glaciers that are largest outside of the polar regions. This isn't the South Asia of mangoes and coconuts, this is the South Asia of trout, apricots, and morels. These people have adapted to what has to be called a Nordic-style environment and I think we should expect a few so-called "white" people to appear over the millennia even if Alexander's troops didn't settle there.Replies: @Karl Zimmerman, @Twinkie
Despite Afghanistan’s isolation from modern trade and migration (save the refugees who fled to Pakistan), Westerners who have been involved in the Afghan civil war have long noted the striking and colorful ethno-racial cauldron of the area: Pashtuns, Tajiks, Uzbeks, and Hazaras, and their intermixtures, for that matter, that produce a whole gamut of ethnic appearances. The French, in particular, were quite taken with the Tajiks.
So, based on that map, the Chinese of Sichuan should have the most European-like appearance, eh?
I’ve been told more than once that Tajiks in the Pamirs are much more likely to have brown or even red hair than Tajiks in the lowlands. Has there ever been a genetic comparison of the Tajiks of the Pamirs and, say, the Fergana Valley?
It’s not helpful to simply focus on R1a, you need to look at the subclades.
In this regard the Burusho don’t differ in any way from Indo-Iranians.
Dan,
Unfortunately, besides the test at GedMatch, there really isn’t anything else.
CupOfSoup,
For what it’s worth, the HGDP Burusho are quite low on R1a1a, compared to their neighbors. Out of 20 HGDP Burusho samples, only 2 have R1a1a. By contrast, R2 attains a frequency of 40% (the R* is actually R2)!
Twinkie,
We now have samples for Pamiri language speakers, Yaghnobi speakers, and a cosmopolitan sampling of Tajiks from Dushanbe. The Pamiri speakers, who are stereotyped for looking strongly European, are much more genetically South Asian and much more genetically Northern European in comparison to both Yaghnobi speakers and Tajiks proper. This dual heightened affinity to both South Asia and Northern Europe are probably indicative of greater steppe ancestry, in comparison to Yaghnobi and Tajiks proper. The Yaghnobi are considerably less East Asian-admixed in comparison to Tajiks proper, although they have the same affinity to both Northern Europe and South Asia as Tajiks. In a way, the Dushanbe Tajiks seem to be basically Yaghnobi, but with some Turkic admixture. The Pamiri are somewhat distinct from both, although Pamiri East Asian levels are intermediate between Yaghnobi and Tajiks proper.
"the Kalash are lactose tolerant, but they lack the common Eurasian variant in totality. That implies that there is another variant in the LCT region unique to the Kalash."
Reminds me of the second half of this post: https://westhunt.wordpress.com/2014/05/19/phenotypes-vs-genetic-statistics/Replies: @Vijay
Witzel wrote about the Kalasha religion here :
http://www.people.fas.harvard.edu/~witzel/KalashaReligion.pdf
A list of texts on kalasha language + religion in English is maintained here
http://kalashapeople.blogspot.com/p/books.html
A discussion of Kalsha language is provided by Elena Bashir in Ch. 22 of the following book;
The Indo-Aryan Languages
By George Cardona, Dhanesh Jain
It has some discussion of religion in pages 905-990
Nature in the Kalasha Perception of Life by Birgitte Sperber, a chapter in a book entitled “Asian perceptions of nature: a critical approach” by: Ole Bruun and Arne Kalland.
Natural Resources and Cosmology in Changing Kalasha Society, A Book By Mytte Fentz
http://www.people.fas.harvard.edu/~witzel/KalashaReligion.pdf
A list of texts on kalasha language + religion in English is maintained here
http://kalashapeople.blogspot.com/p/books.html
A discussion of Kalsha language is provided by Elena Bashir in Ch. 22 of the following book;
The Indo-Aryan Languages
By George Cardona, Dhanesh Jain
It has some discussion of religion in pages 905-990
Nature in the Kalasha Perception of Life by Birgitte Sperber, a chapter in a book entitled "Asian perceptions of nature: a critical approach" by: Ole Bruun and Arne Kalland.
Natural Resources and Cosmology in Changing Kalasha Society, A Book By Mytte FentzReplies: @Yudi
Thanks for the links.
Unfortunately, besides the test at GedMatch, there really isn't anything else.
CupOfSoup,
For what it's worth, the HGDP Burusho are quite low on R1a1a, compared to their neighbors. Out of 20 HGDP Burusho samples, only 2 have R1a1a. By contrast, R2 attains a frequency of 40% (the R* is actually R2)!
Twinkie,
We now have samples for Pamiri language speakers, Yaghnobi speakers, and a cosmopolitan sampling of Tajiks from Dushanbe. The Pamiri speakers, who are stereotyped for looking strongly European, are much more genetically South Asian and much more genetically Northern European in comparison to both Yaghnobi speakers and Tajiks proper. This dual heightened affinity to both South Asia and Northern Europe are probably indicative of greater steppe ancestry, in comparison to Yaghnobi and Tajiks proper. The Yaghnobi are considerably less East Asian-admixed in comparison to Tajiks proper, although they have the same affinity to both Northern Europe and South Asia as Tajiks. In a way, the Dushanbe Tajiks seem to be basically Yaghnobi, but with some Turkic admixture. The Pamiri are somewhat distinct from both, although Pamiri East Asian levels are intermediate between Yaghnobi and Tajiks proper.Replies: @Twinkie
Thank you. That is VERY fascinating. Among the Pamiris, Yaghnobis, and the Tajiks from Dushanbe, what kind of genetic distances are we talking about? Can you provide some sense of scale regarding the respective affinities for South Asia and Northern Europe, for example?
Could this help explain the dreaded middle eastern autosomal result? http://dna-explained.com/2012/07/24/the-dreaded-middle-east-autosomal-result/
Or this? http://dnaconsultants.com/_blog/DNA_Consultants_Blog/post/Anomalous_Mitochondrial_DNA_Lineages_in_the_Cherokee/
@ Twinkie
“This seems to be a recurring theme of yours. How much confidence do you place on this plausibility?” in response to ‘it is plausible that every major group of humans today (major = numerous) is the product of fusions of branches of the human race which were sharply diverged during the Pleistocene’.
I’m surprised you have difficulty accepting Razib’s comment. I would have thought by now most would have accepted the overwhelming evidence that it is so.
"This seems to be a recurring theme of yours. How much confidence do you place on this plausibility?" in response to 'it is plausible that every major group of humans today (major = numerous) is the product of fusions of branches of the human race which were sharply diverged during the Pleistocene'.
I'm surprised you have difficulty accepting Razib's comment. I would have thought by now most would have accepted the overwhelming evidence that it is so.Replies: @Razib Khan, @Twinkie
yep. it’s pretty obvious. you just need to know the population genomics and run the data yourself to convince. check to both for me 😉
"This seems to be a recurring theme of yours. How much confidence do you place on this plausibility?" in response to 'it is plausible that every major group of humans today (major = numerous) is the product of fusions of branches of the human race which were sharply diverged during the Pleistocene'.
I'm surprised you have difficulty accepting Razib's comment. I would have thought by now most would have accepted the overwhelming evidence that it is so.Replies: @Razib Khan, @Twinkie
How is asking for his confidence level of what he terms a “plausible” explanation “have difficulty accepting”?
I happen to suspect that this idea is very much on the mark. I just wanted to know Mr. Khan’s level of confidence and, if he’s amenable, to know what factors increase or decrease that confidence. And I happened to have expressed similar ideas elsewhere on Unz Review (without the benefit of Mr. Khan’s level of expertise on the matter since I am not a geneticist).
What’s with all this motive suspicion and faulty mind-reading? Not every question is some sort of a snarky challenge or internet penis size contest.
confidence high. definitely not certain. perhaps not very high.
Thanks for the reply. What prospective discoveries in the field would increase that confidence or decrease it?
better frameworks to detect/distinguish clinal vs. pulse admixture events.
two hypotheses come to mind
1) R1a1a was not limited to IE people
2) gene flow from IE groups into these groups
3) some of both
i lean more to 1 than 2 in my ratio, but i'm at low confidence.Replies: @CupOfCanada
Do you think it’s plausible that the “specialness” of the Kalash comes from a non-IE substrate?
Davidski, Commentator – that question applies to both of you too.
They're like the Basques of the Hindu Kush. That's why they've been confusing the bejesus out of many population geneticists.Replies: @CupOfCanada
There’s nothing really special about the Kalash. They’re just an extreme isolate since the Bronze Age with a very small effective population size.
They’re like the Basques of the Hindu Kush. That’s why they’ve been confusing the bejesus out of many population geneticists.
They're like the Basques of the Hindu Kush. That's why they've been confusing the bejesus out of many population geneticists.Replies: @CupOfCanada
What about the lactase persistence Razib mentions?
Most Europeans became lactase persistent since the Bronze Age, so why not the Kalash with their own allele?