search this blog

Friday, April 8, 2016

New preprint on South Asian population history

Broad MIT/Harvard and friends have released a new preprint at bioRxiv on the genetics of South Asians. It's titled The promise of disease gene discovery in South Asia and mostly focuses on the recent demographic histories of ethnic groups from across India. See here.

In fact, it lacks any analyses with ancient genomes, which is a shame. But the supp info includes an interesting Admixture graph which shows the Paniya of South India modeled as 83% Ancestral South Indian (ASI) and 17% Ancestral North Indian (ANI). See Supplementary Figure 6. here.

ANI is shown as a sister clade of present-day Georgians, which probably means that, as far as the Paniya are concerned, it's essentially an offshoot of an Caucasus Hunter-Gatherer (CHG) population.


Nirjhar007 said...

There will be many aDNA coming from the subcontinent. I am fully confident it will change many conceptions which are in fashion now.

Seinundzeit said...

Their phylogenetic tree (with all the populations together) is pretty interesting, as it shows the immense diversity that exists in South Asia. The Pakistani populations (the ones sampled by the HGDP), as well as upper caste northwestern Indians, cluster with Europeans + West Asians. Basically, they behave like West Eurasians on that graph. On the other end, tribal South Indian populations (Kurumba, Malayan, etc) are pretty much their own thing, very distant/distinct.

But the most interesting point to note is the great separation between tribal South Indian populations and the Onge. There is no particularly close relationship to be observed.

As I've already speculated, I think most of that 80% for the Paniya isn't ENA (or West Eurasian). In this case, we really need aDNA. For whatever it's worth, both the d-stats nMonte + PC nMonte analysis support this proposition, as does the nature of the South Asian component observed in ADMIXTURE.

FrankN said...

This study presents a wonderful opportunity to identify groups that, because of endogamy or other factors, may have preserved a unique genetic structure that reaches back far into prehistory. The Paniya are one example, which actually may provide a fairly good proxy of the population that inhabited the presumed South Indian LGM refugium.

Sein - after doing some research on South(-East) Asia during the LGM, it has become clear to me why the Onge aren't a good ASI proxy. The low LGM sea level would have connected the Andamans to Sundaland, with most of the Andaman Sea falling dry, but the remaining inland sea just adjacent to the Andamans and Nicobars most likely providing sufficient precipitation to keep them well inhabitable. A nice LGM refugium, but much better connected to Sumatra than to the Indian subcontinent.

The study's Online Table 1 ( lists all included populations, sorted by their IBD score, with Onge coming in first, and Kalash 9th. In-between, we have the cluster of Malaikuarve (Rank 2), Narikuruvar (3) and Hakki_Pikki (5), all from Tamil Nadu/ Karnataka, who share IBD with each other (Supp. Table 3), but hardly anybody else. Looks like a cluster wothwhile of being explored further as pontential harbour of UP (LGM) ancestry - either connected to, or as seperate refugium to the Paniya one.

Considering that the LGM most likely resulted in a stronger SW Monsoon, and consequently a likely LGM refugium around the Indus mouth and east of it, Gondi people (IBD rank 12) also seem an apt candidate for identifying UP genetic traces. Kusunda (IBD rank 13) surely deserve a closer look, also for their linguistic isolation.

I am also quite confident that a systematic analysis of that table, in correspondence with the clustering as per Supp.Tab. 3, could bring up some groups that may provide proxies of the Bangali LGM refugium, and maybe a few other places that weren't completely turned into a desert during the LGM. However, I lack sufficient knowledge and understanding of the Indian tribal/ caste structure to forward proposals. That job requires a native - what about you, Nijhar?
I am quite confident that any reasonable proposal by you will get analytic follow-up - if not by Dave, then by Sein, Matt atl.

Chad Rohlfsen said...

Paniyas are clearly closer to the Onge than UI or CHG. Their tree shows ASI and Onge off the same ENA branch, again. Paniyas are definitely mostly ENA. It's not very close.

Qagan said...

Very Interesting! Would this mean Paniyas are around 80%+ East Eurasian or a bit more genetically?

Davidski said...

Ah, just realized that the 83/17 ASI/ANI ratio for the Paniya is actually from the Moorjani paper from a few years ago. So it's nothing new, although I don't remember seeing the graph in that paper.

Qagan said...

Oh I see. It is an old estimation. How much East Eurasian genetically do you think Austroasiatics like Kharia, Bonda, Santhal be in terms of percentage? Also would they be the most East Eurasian group out of all South Asians?

Davidski said...

It varies from a few per cent to as much as 25%, depending on the geography.

Rami said...

This is old news like from 6-7 years and will change when they get ancient genomes from the region.

How do Paniya score on K8 David?? What are their results.

Davidski said...

Here's one of the Paniya. South Indian genetic variation doesn't fit well into the K8 variation, so the ANE estimate is probably wrong. The Sub-Saharan score is obviously wrong. Can't do much more without a few pre-Neolithic genomes from India.

ANE 18.07
South_Eurasian 25.81
ENF 18.02
East_Eurasian 17.64
WHG 0.01
Oceanian 12.53
Pygmy 0
Sub-Saharan 7.93

Nirjhar007 said...

Ideal time David for working on that Kum6 sample. Don't waste it.

ryukendo kendow said...
This comment has been removed by the author.
ryukendo kendow said...
This comment has been removed by the author.
Rob said...


Sounds interesting, and is agree given the natural "turnover" of populations I've oft mentioned in Europe in addition to obvious social competition. But take ur time; there's a lot of data to chew through

I have some papers I can recommend for CWC too, with regard to differential land use, site distribution, and "competing identities" viz-a-viz its contemporaries in north Europe: late TRB groups, Globular Amphora, etc; for both Poland and Germany. Just email me .

ryukendo kendow said...
This comment has been removed by the author.
Davidski said...


There's no output data of that type. I just get the coordinates, and then I unscale them with nMonte.

ryukendo kendow said...
This comment has been removed by the author.
ryukendo kendow said...
This comment has been removed by the author.
Davidski said...

Didn't keep all that. Can't re-run the tests at the moment.

FrankN said...

Ryu: I am not aware of any German database of isotope analyses. The Munich University is preparing a GIS-based DB on Strontium analyses around the Alps, but they don't seem to really get forward, and Strontium is more about mobility than nutrition. Typically, nutrition analyses are done on specific findspots and evaluated in a PhD, which means 200+ pages in German - possibly of little help to you.

Here some of the more comprehensive analyses that are in English, or at least with English abstract and figure legends:

Michelsberg culture:

Nordic FB: (check out the pollen diagram summary near the beginning, with the marked re-forestation setting in around 3000 BC) (similar to part II of the above, in English)

Late Neolithic:
Overview, w. Pollen Diagram for Brandenburg:


SW Germany:

Note that trends are quite diverging:
- In SW Germany, depopulation already sets in during the MN (early/mid 4 mBC), with de-forestation / land use indicators only rising substantially again with the Urnfield culture. During the early 3mBC, even the fertile Upper Rhine Plain south of Strassbourg appears to have been completely depopulated.
- While Schleswig-Holstein has a remarked and lengthy drop in land-use indicators after 3000 BC, such drop is also visible in Brandenburg, but then land use increases quickly. Note that we are here not talking about CW, but the quite distinct Elbe-Havel culture, FB substrate overformed by GAC.
- The Southern Netherlands show no signs of a 3000 BC decrease - to the opposite, the period 3400-2700 BC marks a peak in land clearance. Here as well we are talking about a CW-free enclave, the Hilversum culture, which more or less seamlessly blends over from FB into BB.

I would have loved to also provide analyses for ELbe-Saale, the Rhineland and the Wetterau (Frankfurt area), but couldn't find any. Elbe-Saale has hardly pollen diagrams, but for the Rhineland and Wetterau, they should exist. In fact, both regions are extremely well studied for the EN/MN (LBK/MC), and again BA/IA/migration period, but for some reason nobody has so far found it worthwhile to address the LN in-between.

Based on the observations above, I nevertheless feel that CW may be overstated, dominating widely depopulated landscapes. In-between, densely populated areas have remained (Southern Netherlands, Berlin area) that show only sparse CW influence, if at all. So far, the populations there haven't been sampled fo aDNA. But I would assume them to have played an important role in preserving MN aDNA, which sees a revival in BB and Unetice.

FrankN said...

These 2014 abstracts (German/English) may be of general interest - they cover not only Germany, but also Poland, Hungary, the Balkans, the Mediterranean, and the desertifying Sahara:

After 3000 BC, 2200 BC seems to be the next cultural turning point, with massive breaks almost everywhere (except for Britain). Then comes 1500-1200 BC (Nordic BA, Urnfield expansion, "Sea people", Dorians?). Rather than Hallstatt, which looks like a kind of Urnfield continuation, from ca. 600 BC on, LaTene, Skyths, Greek colonisation and the rise of Rome mark the next break, apparent also in German pollen diagrams.

Rob said...

Frank N

"After 3000 BC, 2200 BC seems to be the next cultural turning point, with massive breaks almost everywhere"

I would concur: the end of CWC, recession of Yamnaya and transition to catacomb, and advance of the Carpathian-Unetice axis. I.e. The start of the true Bronze Age

For LN Germany, we should mention the Muller articles you previously referenced. Some more compete sampling of the MittelElbe-Saale region would be interesting

"Der √úbergang vom Neoli-
thikum zur Kupferzeit in
Mitteldeutschland aus


Unknown said...

The paniya are i think about 50-60 percent onge because Asi is regarded as 70 percent onge and thity percent west asian and ani is 100 percent west asian or central asian so it makes sense.