There is an interesting new article that confirms by statistical means the long held theory that sickle cell disease exists in a dynamic equilibrium with the deadly disease of malaria, at least in general.
Federic B. Piel et al., Global distribution of the sickle cell gene and geographical confirmation of the malaria hypothesis. Nature communications 2010. Open access.
What I find interesting is not so much the confirmation of a correlation in Africa and Europe but the lack of confirmation in Asia and Oceania, where malaria exists endemically too and, unlike the case of America, the discontinuity cannot be explained by loss of the sickle cell allele in the arctic latitudes (Beringia). In fact the peculiarities of the geographical distribution of sickle cell disease versus endemic malaria are very intriguing:
 |
| Fig. 1 (click to enlarge) |
I understand that, in the case of Africa, the correlation is very strong, though not totally lineal. For instance the HbS (sickle cell) allele is lacking in wide areas of malaria endemism: The Horn, Southern Africa (including Bantu Mozambique but not Austronesian Madagascar), NW Africa...
So somehow these populations have managed to get rid of the sickle cell anemia in spite of being exposed to malaria in hyperendemic form. This alone is pretty interesting in itself, specially as it involves Bantu populations that are supposed to have expanded recently.
In Europe we have two situations: Greece and Albania where sickle cell exists in moderate frequencies and the rest, where it is unheard of. While Greece is one of the malaria epidemic areas, there are others where the parasite has only been eradicated in the 20th century: Italy and Iberia specially. However no sickle cell exists there. This case is very suggestive, statistics apart, with the spread of Y-DNA E1b1b1a2 (V13), haplogroup that is most concentrated in the SW Balkans.
But most intriguing of all may be Asia. The presence of the allele in West Asia makes some good sense because of its prehistorical (and maybe also historical) interactions with Africa, so the HbS allele had all opportunities to arrive, really.
But what about further East. We know of no particular "recent" genetic flow into South Asia that could explain the relatively high frequencies found in India, mostly in areas of lesser West Eurasian influence. The connections with Africa are even less likely (some small populations of recent African origin do exist but they cannot possibly have influenced the majority in such dramatic way in such short time).
It could be something arrived in the migration out of Africa of course, yet, on the other hand, further East there is no trace of the HbS allele. But malaria is found again in hyperendemic form in all SE Asia and Near Melanesia. Why? No idea. While the sampling is low density, it does seem like most relevant areas have been tested (map a), including Melanesia, Australian Aborigines, Indonesia and even Andaman.
So why these irregularities? Which are the founder effects implied? One can imagine that the northernmost historical malaria areas were free from the parasite in the Ice Age, but surely this was not the case of the Asian tropical belt. Did some sort of lifestyle, like choosing to dwell where marine breeze kept mosquitoes at bay (a usual common sense practice in tropical coasts as far as I know) help Eastern Eurasian founders prevent this disease, making the allele unnecessary? Was instead the HbS allele "stored" (by founder effect) in some South Asian demographic reservoir (and only there), gradually expanding to some nearby malaria affected areas but never participating of the colonization of the Far East by mere chance?
What about the West, where the HbS allele is also mostly absent? Here at least one could argue that malaria was probably absent in the Pleistocene, so the HbS allele was logically selected against (even heterozygous types have some lesser disadvantages if malaria is not present to compensate for them).
It is a single SNP (rs334) which encodes all this matter, so it is not possible to trace it phylogenetically in any way.
Intriguing in any case.
A couple of points.
ReplyDelete1. Recent research (within the last year or two) has established pretty definitively that Malaria is a post-Columbian disease in the Americas. So, sickle cell had no adaptive benefit for 11,000+ years in the Americas, nor was it beneficial for thousands of years in Beringia and Northeast Asia for thousands of years before then. Five hundred years is a relatively short time to "get lucky" by independently evolving the sickle cell mutation. Sickle cell malaria resistance is probably motivator for the Latin American slave trade, although the nature of the cause and effect relationship between sickle cell in African populations and malaria resistance wasn't known until much later.
2. Ethiopia, to which we trace mtDNA haplogroup L3* which is the mother of all Eurasian mtDNA lineages is not a hotbed of malaria infection, and was probably less so at the time of the Out of Africa expansion when sea levels were probably relatively low as a result of large ice caps in an ice age.
3. Sickle cell genes are most common in areas that pre-Bantu expansion were predominantly Pygmy, and for much of the pre-Bantu expansion wet Sahara period, West Africans may have lived further North. Thus, sickle cell genes are probably either introgressed from assimliated pygmy populations (I'd guess that non-L3 Africans have the sickle cell gene more than L2 and that Ls other than L2 and L3 associated with pygmies have the highest frequency) or it was a recent mutation dating from Bantu expansion into the tropics. The first wouldn't have produced much back migration to Ethopia in the Out of Africa era, the second wouldn't have happened long ago enough to impact Asia either.
4. There have been two relatively recent events that could have brough the genes to South Asia. One is the Neolithic migration to the Indus River Valley probably from Mesopotamia, which could have brought African genes with it.
The other its the hypothetical Afro-Dravidian theory -- that Africans around 2500 BCE brought Sahel crops and cultural traits including language to South Asia arriving via trade routes from Sumeria to Harappa perhaps. The DNA trace of this may be Y-DNA haplogroup T (there is no mtDNA trace from Africa, suggesting a male dominated technology transfer population).
Since there has been no meaningful out migration of people's from India in Neolithic times, either scenario would have left no demographic impact on East Asia, which if it has a malaria resistance gene system has a different one than sickle cell. The mutation didn't evolve independently there and hasn't had an opportunity to arrive from Asia.
Also, keep in mind that demography in SE Asia has generally been from North to South in the Neolithic era, so lots of the current populatioon has little ancestry in malaria zones despite living there now.
A bit on the history of malaria can be found here. Malaria probably killed King Tut and its presence in the Nile at that time makes it likely that exposed populations were present in Mesoptamia in all Neolithic era contacts with South Asia.
ReplyDeleteIt isn't clear how old malaria is in Southeast Asia. It may be no older than sea trade from Africa and South Asia to points East, which would only be a few thousand years.
Good points Andrew, thanks.
ReplyDeleteThe only objection I have is that there is absolutely no trace of African genetics in South Asia, except maybe for very very few Y-DNA E erratics. So I'm guessing that if anything it'd be Arabia, the Persian Gulf probably, the origin of this spread.
But... another problem is that the areas with significant sickle cell allele in South Asia are not those most strongly affected by the possible Neolithic flows, which are obviously Pakistan and NW India. So I'm not sure if you're right in all.
Anyhow good ideas.
The Indus Rivey Valley connection is weaker, but if you take the Afro-Dravidian hypothesis seriously and assume that the Y-DNA T there is an African trace of that connection (my intuition from the distribution is that Y-DNA T may be malaria infested Egyptian in origin), then DNA connection is more than erratic. My guess is that Y-DNA T subtypes specific to South Asia converge to one to three subhaplotypes in a 4500 year time frame.
ReplyDeleteAlso, if you have something that is strongly selecting for a gene, as malaria does for sickle cell, you don't need a very high frequency in this autosomal trait to start with to produce a dramatic increase in its frequency over a thousand generations or so.
"... but if you take the Afro-Dravidian hypothesis seriously"...
ReplyDeleteNever heard of that hypothesis before. Dravidian is said to be related to Elamite, part of "Nostratic"... but never heard it could be related to African languages.
"and assume that the Y-DNA T there is an African trace of that connection"...
If anything it'd be an Indian influence in Africa, not the other way around.
However there's also the possibility that T actually had a Persian Gulf origin. This is also a malaria zone with relatively high sickle cell and your parallelism with Y-DNA T spread may be of some meaning, but I don't think it can be argued that T is original from Africa.
I explore the Afro-Dravidian hypothesis (which is most closely associated with Bernard Sergent), and the Y-DNA haplotype T links at length in the linked blog post from last spring.
ReplyDeleteAfro-Dravidian actually has quite a bit more evidence of multiple types to support it than the Elam-Dravidian hypothesis, and Witzel has pretty much undermined the argument that the Harappans spoke a Dravidian language, on the grounds that there isn't a Dravidian substrate in early Rig Vedic Sanskrit.
Pearl Millet is not part of the Neolithic package of India but a late introduction c. 2000 BCE.
ReplyDeleteAs for Dravidian, it can perfectly be the language family of any other Neolithic or even Paleolithic zone of India: not everything is IVC: there were many other cultures through the continent.
I mean the language can should be native to the Indian subcontinent. In fact it's my default theory about the origin of Dravidian.
However South India probably had a Bronze Age "trade" connection with NE Africa via Arabia, but I doubt this has to do with any language or genetics: it has to do with cultural transmission such the pearl millet or megalithic burial customs (also known to have existed in Ethiopia but specially Arabia (Yemen, UAE...). As with those of the Caucasus, these are late megalithic expressions (when they had vanished in the West) but are critical in understanding the (religious?) migration of this phenomenon eastward into Asia (both India and the steppes-Korea). In no case I'd expect a language to flow with them but I'd expect religious-cultural elements to do, like the zodiac or the swastika.
India who two different Neolithic packages. One was the Fertile Cresent package of crops of the Harappans that fit one climatological zone that arrives ca. 6000 BCE in the IVC. India outside IVC lacked agriculture for another 3000+ years; it had not yet experienced a Neolithic revolution.
ReplyDeleteThe other was the Sahel crop package of crops that arrived in South India ca. 2500 BCE.
There is virtually no overlap of the two because they require different seasons to thrive. Pearl millet was not introduced ca. 2000 BCE to the IVC. They have the wrong seasonal weather patterns for it to thrive there. The Harappan traces of it around then are in a frontier trade post warehouse on the SW coast of India. The Neolithic arrived in the two climate regions of India at two different times that had distinct groups of crops. The evidence that the two separate agricultural societies had anything to do with each other at first beyond arms length trade is almost nil.
(Arguably there was also a third separate wave of crop transfer and Neolithic revolution in NE South Asia from Southeast Asia ca. 3000 BCE associated with Austroasiatic peoples that brought the crop set of East Asia.)
As a default hypothesis, Dravidian as native to India is plausible. This is a good default. The people of India had to speak something before the arrival of people who spoke IE languages, Tibeto-Burman languages, or Austro-Asiatic languages, and each of those seems to have a pretty clear origin outside the region. Dravidian languages, per se, are not found outside South Asia. Those people could have been effectively isolated from modern humans to the east and west for tens of thousand of years erasing any linguistic connection traceable to the original Out of Africans. The lack of a Dravidian substrate in early Sanskrit also pretty much rules out Dravidian language speaking Harappans in the IVC, and hence any strong source of influence and contact from, for exampe, Sumerian or Elamite or Akkadian for Dravidian languages (although a South Asian source for Sumerian or Elamite couldn't be ruled out).
Given the apparently young age of Dravidian Babel (and while linguistic dates aren't terribly helpful, I don't think that once can fairly say that there are entirely meaningless), in an indigenous Dravidian scenario, you pretty much have to infer that South Asian Neolithic community's dialect of the region became dominant and was the source of all modern Dravidian languages to the extinction of all other dialects ca. 2500 BCE to 3000 BCE in approximately the location attributed to Proto-Dravidian in South India. The dialects of Dravidian would be far more different from each other if they were native language of India.
But, if indeed, there are coincidences between Dravidian and the Niger-Congo languages that are too great to attribute to chance at an immense time depth, then the other evidence corroborating Afro-Dravidian connections, most powerfully the crops in that time frame, but additionally other customs and cultural pratices, then the case of Dravidian being an imported language with some substrate influence starts to look pretty strong, particularly if one can point to a plausible population genetic link.
And, the distribution of Y-DNA haplogroup T in South India, even if it isn't proximately African in origin as the Afro-Dravidian hypothesis suggests, is a good fit for the expansion of the Dravidian language community that erased other local dialects in the South Indian Neolithic associated with the arrival of agriculture based on Sahel crops at this time, and is a poor fit for any other known historical cause. The historical known invasions pretty much all came from the West, not the Central Eastern Coast of India.
Alright, I am realizing how little I know about the Neolithic of India (found this article that confirms what you say more or less). Till date I had not even questioned that the NW Neolithic, maybe with the help of rice agriculture from the East and some local domestications (some vine is Indian-domesticated, can't recall which) simply spread to the South, probably by mere cultural diffusion. I did not even ponder if wheat, rye, barley, lentils and such could even grow in tropical climates (probably not) and I imagined that millet/sorghum from the North would have been adapted too. I mean: if they can grow Palestinian cereals and herd Iraqi sheep in Norway... why not in Kerala?
ReplyDeleteBut maybe not. But rice is certainly a crop fit for the tropics as are also the zebuine type of cow, the all-resistant goat and the water buffalo, which possibly arrived with rice and Austroasiatic tribes.
But in any case, I really have a problem with the Niger-Congo link: Niger-Congo was then restricted to West Africa (no Bantu expansion yet), what makes a connection to India almost impossible (would it be Nilo-Saharan or Afroasiatic... but Niger-Congo is totally out of range). Also such a recent language split would be more recent than Indoeuropean (or Austroasiatic) and absolutely recognizable.
Also, as mentioned there's no known African genetics in India, other than the few populations of recent African ancestry. A language introduction should imply at least some notable genetic flow in pre-civilization (or pre-hierarchical) phases. So if Dravidian is from outside India, I can only think it arrived from West Asia, which is the commercial hub that brought all the neolithic elements ultimately, including African millet and later Megalithic cultural expressions (Arabia Peninsula must be the link for this), and whose lineages are indeed found in India, including the South.
It may be the case still that Elamo-Dravidian is correct, if Arabian peoples then spoke, at least partly, such languages. This would be quite unsurprising for the Persian Gulf, I dare say, with Semitization surely being from a later date, when camel domestication allowed the crossing of deserts (not much earlier than 2000 BCE, I understand) and/or when Mesopotamia became Semitic-speaking, a slow process starting c. 3900-3500 BCE.
The default hypothesis indeed remains that it is a native Indian language, experiencing local expansion in the context of Neolithic possibly.