(…) the spread of agriculture in Europe was a result of the demic diffusion of early Anatolian farmers, it was discovered that the spread of agriculture to South Asia was mediated by a genetically completely different farmer population in the Zagros mountains in contemporary Iran (IF). The ANI-ASI cline itself was interpreted as a mixture of three components genetically related to Iranian agriculturalists, Onge and Early and Middle Bronze Age Steppe populations (Steppe_EMBA).
The first ever autosomal aDNA from South Asia comes from Northern Pakistan (Swat Valley, early Iron Age). This study presented altogether 362 aDNA samples from the broad South and Central Asia and contributes substantially to our understanding of the evolutionary past of South and Central Asia. The study redefines the three genetic strata that form the basis of the Indian Cline. The Indus Periphery (IP) component is composed of (varying proportions of): first, IF, second, Ancient Ancestral South Asians (AASI), which represents an ancient branch of human genetic variation in Asia arising from a population split contemporaneous with the splits of East Asian, Onge and Australian Aboriginal ancestors and third, West_Siberian Hunter gatherers (WS_HG).
The authors argue that IP could have formed the genetic base of the Indus Valley Civilization (IVC). Upon the collapse of the IVC IP contributes to the formation of both ASI and ANI. ASI is formed as IP admixes further with AASI. ANI in turn forms when IP admixes with the incoming Middle and Late Bronze Age Steppe (Steppe_MLBA) component, (rather than the Steppe_EMBA groups suggested earlier)
Dating of the arrival of the Austro-Asiatic speakers in South Asia-based on Y chromosome haplogroup O2a1-M95 expansion estimates yielded dates between 3000 and 2000 BCE . However, admixture LD decay-based approach on genome-wide data suggests the admixture between South Asian and incoming Austro-Asiatic speakers occurred slightly later between 1800 and 0 BCE (Tätte et al. submitted). It is interesting that while the mtDNA variants of the Mundas are completely South Asian, the Y chromosome variation is dominated at >60% by haplogroup O2a which is phylogeographically nested in East Asian-specific paternal lineages.
In India, the speakers of Tibeto-Burman (TB) languages live in the Seven Sisters States in Northeast India and in the very north of the country. Genetically they show a clear East Asian origin and around 20% of subsequent admixture with South Asians within the last 1000 years.The genetic flavour of East Asia in TB is different from that in Munda speakers as the best surrogates for the East Asian admixing component are contemporary Han Chinese.
I found the simplistic migration maps especially interesting to illustrate ancient population movements. The emergence of EHG is supposed to involve a WHG:ANE cline, though, and this isn’t clear from the map. Also, there is new information on what may be at the origin of WHG and Anatolian hunter-gatherers.
This paper used a complete mtDNA genome study of 113 unrelated individuals from the Melakudiya tribal population, a Dravidian speaking tribe from the Kodagu district of Karnataka, Southern India.
Some interesting excerpts (emphasis mine):
Autosomal genetic evidence indicates that most of the ethnolinguistic groups in India have descended from a mixture of two divergent ancestral populations: Ancestral North Indians (ANI) related to People of West Eurasia, the Caucasus, Central Asia and the Middle East, and Ancestral South Indians (ASI) distantly related to indigenous Andaman Islanders (Reich et al. 2009). It is presumed that proto-Dravidian language, most likely originated in Elam province of South Western Iran, and later spread eastwards with the movement of people to the Indus Valley and later the subcontinent India (McAlpin et al. 1975; Cavalli-Sforza et al. 1988; Renfrew 1996; Derenko et al. 2013). West Eurasian haplogroups are found across India and harbor many deep-branching lineages of Indian mtDNA pool, and most of the mtDNA lineages of Western Eurasian ancestry must have a recent entry date less than 10 Kya (Kivisild et al. 1999a). The frequency of these lineages is specifically found among the higher caste groups of India (Bamshad et al. 1998, 2001; Basu et al. 2003) and many caste groups are direct descendants of Indo-Aryan immigrants (Cordaux et al. 2004). These waves of various invasions and subsequent migrations resulted in major demographic expansions in the region, which added new languages and cultures to the already colonized populations of India. Although previous genetic studies of the maternal gene pools of Indians had revealed a genetic connection between Iranian populations and the Arabian Peninsula, likely the result of both ancient and recent gene flow (Metspalu et al. 2004; Terreros et al. 2011).
mtDNA haplogroup HV14 has prominence in North/Western Europe, West Eurasia, Iran, and South Caucasus to Central Asia (Malyarchuk et al. 2008; Schonberg et al. 2011; Derenko et al. 2013; De Fanti et al. 2015). Although Palanichamy identified haplogroup HV14a1 in three Indian samples (Palanichamy et al. 2015), it is restricted to limited unknown distribution. In the present study, by the addition of considerable sequences from the Melakudiya population, a unique novel subclade designated as HV14a1b was found with a high frequency (43%) allowed us to reveal the earliest diverging sequences in the HV14 tree prior to the emergence of HV14a1b in Melakudiya. (…) The coalescence age for haplogroup HV14 in this study is dated ~ 16.1 ± 4.2 kya and the founder age of haplogroup HV14 in Melakudiya tribe, which is represented by a novel clade HV14a1b is ~ 8.5 ± 5.6 kya
The coalescence age of haplogroup U7a3a1a2 dates to ~ 13.3 ± 4.0 kya. (…)
Although, haplogroup U7 has its origin from the Near East and is widespread from Europe to India, the phylogeny of Melakudiya tribe with subclade U7a3a1a2 clusters with populations of India (caste and tribe) and neighboring populations (Irwin et al. 2010; Ranaweera et al. 2014; Sahakyan et al. 2017), hint about the in-situ origin of the subclade in India from Indo-Aryan immigrants.
I am not a native English speaker, but this paper looks like it needs a revision by one.
Also – without comparison with ancient DNA – it is not enough to show coalescence age to prove an origin of haplogroup expansion in the Neolithic instead of later bottlenecks. However, since we are talking about mtDNA, it is likely that their analysis is mostly right.
Finally, one thing is to prove that the origin of the Indus Valley Civilization lies (in part) in peoples from the Iranian plateau, and to show with ASI ancestry that they are probably the origin of Proto-Dravidian expansion, and another completely different thing is to prove an Elamo-Dravidian connection.
Since that group is not really accepted in linguistics, it is like talking about proving – through that Iran Neolithic ancestry – a Sumero-Dravidian, or a Hurro-Dravidian connection…
The Sahara was wetter and greener during multiple interglacial periods of the Quaternary, when some have suggested it featured very large (mega) lakes, ranging in surface area from 30,000 to 350,000 km2. In this paper, we review the physical and biological evidence for these large lakes, especially during the African Humid Period (AHP) 11–5 ka. Megalake systems from around the world provide a checklist of diagnostic features, such as multiple well-defined shoreline benches, wave-rounded beach gravels where coarse material is present, landscape smoothing by lacustrine sediment, large-scale deltaic deposits, and in places, tufas encrusting shorelines. Our survey reveals no clear evidence of these features in the Sahara, except in the Chad basin. Hydrologic modeling of the proposed megalakes requires mean annual rainfall ≥1.2 m/yr and a northward displacement of tropical rainfall belts by ≥1000 km. Such a profound displacement is not supported by other paleo-climate proxies and comprehensive climate models, challenging the existence of megalakes in the Sahara. Rather than megalakes, isolated wetlands and small lakes are more consistent with the Sahelo-Sudanian paleoenvironment that prevailed in the Sahara during the AHP. A pale-green and discontinuously wet Sahara is the likelier context for human migrations out of Africa during the late Quaternary.
The whole review is an interesting read, but here are some relevant excerpts:
Various researchers have suggested that megalakes coevally covered portions of the Sahara during the AHP and previous periods, such as paleolakes Chad, Darfur, Fezzan, Ahnet-Mouydir, and Chotts (Fig. 2, Table 2). These proposed paleolakes range in size by an order of magnitude in surface area from the Caspian Sea–scale paleo-Lake Chad at 350,000 km2 to Lake Chotts at 30,000 km2. At their maximum, megalakes would have covered ~ 10% of the central and western Sahara, similar to the coverage by megalakes Victoria, Malawi, and Tanganyika in the equatorial tropics of the African Rift today. This observation alone should raise questions of the existence of megalakes in the Sahara, and especially if they developed coevally. Megalakes, because of their significant depth and area, generate large waves that become powerful modifiers of the land surface and leave conspicuous and extensive traces in the geologic record.
Lakes, megalakes, and wetlands
Active ground-water discharge systems abound in the Sahara today, although they were much more widespread in the AHP. They range from isolated springs and wet ground in many oases scattered across the Sahara (e.g., Haynes et al., 1989) to wetlands and small lakes (Kröpelin et al., 2008). Ground water feeding these systems is dominated by fossil AHP-age and older water (e.g., Edmunds and Wright 1979; Sonntag et al., 1980), although recently recharged water (<50 yr) has been locally identified in Saharan ground water (e.g., Sultan et al., 2000; Maduapuchi et al., 2006).
In our view, Lake Chad is the only former megalake in the Sahara firmly documented by sedimentologic and geomorphic evidence. Mega-Lake Chad is thought to have covered ~ 345,000 km2, stretching for nearly 8° (10–18°N) of latitude (Ghienne et al., 2002) (Fig. 2). The presence of paleo- Lake Chad was at one point challenged, but several—and in our view very robust—lines of evidence have been presented to support its development during the AHP. These include: (1) clear paleo-shorelines at various elevations, visible on the ground (Abafoni et al., 2014) and in radar and satellite images (Schuster et al., 2005; Drake and Bristow, 2006; Bouchette et al., 2010); (2) sand spits and shoreline berms (Thiemeyer, 2000; Abafoni et al., 2014); and (3) evaporites and aquatic fauna such as fresh-water mollusks and diatoms in basin deposits (e.g., Servant, 1973; Servant and Servant, 1983). Age determinations for all but the Holocene history of mega- Lake Chad are sparse, but there is evidence for Mio-Pliocene lake (s) (Lebatard et al., 2010) and major expansion of paleo- Lake Chad during the AHP (LeBlanc et al., 2006; Schuster et al., 2005; Abafoni et al., 2014; summarized in Armitage et al., 2015) up to the basin overflow level at ~ 329m asl.
Insights from hydrologic mass balance of megalakes
Using these conservative conditions (i.e., erring in the direction that will support megalake formation), our hydrologic models for the two biggest central Saharan megalakes (Darfur and Fezzan) require minimum annual average rainfall amounts of ~ 1.1 m/yr to balance moisture losses from their respective basins (Supplementary Table S1). Lake Chad required a similar amount (~1 m/yr; Supplementary Table S1) during the AHP according to our calculations, but this is plausible, because even today the southern third of the Chad basin receives ≥1.2 m/yr (Fig. 2) and experiences a climate similar to Lake Victoria. A modest 5° shift in the rainfall belt would bring this moist zone northward to cover a much larger portion of the Chad basin, which spans N13° ±7°. Estimated rainfall rates for Darfur and Fezzan are slightly less than the average of ~ 1.3 m/yr for the Lake Victoria basin, because of the lower aw values, that is, smaller areas of Saharan megalakes compared with their respective drainage basins (Fig. 15).
Estimates of paleo-rainfall during the AHP
Here major contradictions develop between the model outcomes and paleo-vegetation evidence, because our Sahelo-Sudanian hydrologic model predicts wetter conditions and therefore more tropical vegetation assemblages than found around Lake Victoria today. In fact, none of the very wet rainfall scenarios required by all our model runs can be reconciled with the relatively dry conditions implied by the fossil plant and animal evidence. In short, megalakes cannot be produced in Sahelo-Sudanian conditions past or present; to form, they require a tropical or subtropical setting, and major displacements of the African monsoon or extra-desert moisture sources.
If not megalakes, what size lakes, marshes, discharging springs, and flowing rivers in the Sahara were sustainable in Sahelo-Sudanian climatic conditions? For lakes and perennial rivers to be created and sustained, net rainfall in the basin has to exceed loss to evapotranspiration, evaporation, and infiltration, yielding runoff that then supplies a local lake or river. Our hydrologic models (see Supplementary Material) and empirical observations (Gash et al., 1991; Monteith, 1991) for the Sahel suggest that this limit is in the 200–300 mm/yr range, meaning that most of the Sahara during the AHP was probably too dry to support very large lakes or perennial rivers by means of local runoff. This does not preclude creation of local wetlands supplied by ground-water recharge focused from a very large recharge area or forced to the surface by hydrologic barriers such as faults, nor megalakes like Chad supplied by moisture from the subtropics and tropics outside the Sahel. But it does raise a key question concerning the size of paleolakes, if not megalakes, in the Sahara during the AHP. Our analysis suggests that Sahelo-Sudanian climate could perhaps support a paleolake approximately ≤5000 km2 in area in the Darfur basin and ≤10,000–20,000 km2 in the Fezzan basin. These are more than an order of magnitude smaller than the megalakes envisioned for these basins, but they are still sizable, and if enclosed in a single body of water, should have been large enough to generate clear shorelines (Enzel et al., 2015, 2017). On the other hand, if surface water was dispersed across a series of shallow and extensive but partly disconnected wetlands, as also implied by previous research (e.g., Pachur and Hoelzmann, 1991), then shorelines may not have developed.
One of the underdeveloped ideas of my Indo-European demic diffusion model was that R1b-V88 had migrated through South Italy to Northern Africa, and from it using the Sahara Green Corridor to the south, from where the “upside-down” view of Bender (2007) could have occurred, i.e. Afroasiatic expanding westwards within the Green Sahara, precisely at this time, and from a homeland near the Megalake Chad region (see here).
Whether or not R1b-V88 brought the ‘original’ lineage that expanded Afroasiatic languages may be contended, but after D’Atanasio et al. (2018) it seems that only two lineages, E-M2 and R1b-V88, fit the ‘star-like’ structure suggesting an appropriate haplogroup expansion and necessary regional distribution that could explain the spread of Afroasiatic languages within a reasonable time frame.
This review shows that the hypothesized Green Sahara corridor full of megalakes that some proposed had fully connected Africa from west to east was actually a strip of Sahelo-Sudanian steppe spread to the north of its current distribution, including the Chad megalake, East Africa and Arabia, apart from other discontinuous local wetlands further to the north in Africa. This greenish belt would have probably allowed for the initial spread of early Afroasiatic proto-languages only through the southern part of the current Sahara Desert. This and the R1b-V88 haplogroup distribution in Central and North Africa (with a prevalence among Chadic speakers probably due to later bottlenecks), and the Near East, leaves still fewer possibilities for an expansion of Afroasiatic from anywhere else.
If my proposal turns out to be correct, this Afroasiatic-like language would be the one suggested by some in the vocabulary of Old European and North European local groups (viz. Kroonen for the Agricultural Substrate Hypothesis), and not Anatolian farmer ancestry or haplogroup G2, which would have been rather confined to Southern Europe, mainly south of the Loess line, where incoming Middle East farmers encountered the main difficulties spreading agriculture and herding, and where they eventually admixed with local hunter-gatherers.
NOTE. If related to attested languages before the Roman expansion, Tyrsenian would be a good candidate for a descendant of the language of Anatolian farmers, given the more recent expansion of Anatolian ancestry to the Tuscan region (even if already influenced by Iran farmer ancestry), which reinforces its direct connection to the Aegean.
The fiercest opposition to this R1b-V88 – Afroasiatic connection may come from:
Traditional Hamito-Semitic scholars, who try to look for any parent language almost invariably in or around the Near East – the typical “here it was first attested, ergo here must be the origin, too”-assumption (coupled with the cradle of civilization memes) akin to the original reasons behind Anatolian or Out-of-India hypotheses; and of course
autochthonous continuity theories based on modern subclades, of (mainly Semitic) peoples of haplogroup E or J, who will root for either one or the other as the Afroasiatic source no matter what. As we have seen with the R1a – Indo-European hypothesis (see here for its history), this is never the right way to look at prehistoric migrations, though.
I proposed that it was R1a-M417 the lineage marking an expansion of Indo-Uralic from the east near Lake Baikal, then obviously connected to Yukaghir and Altaic languages marked by R1a-M17, and that haplogroup R could then be the source of a hypothetic Nostratic expansion (where R2 could mark the Dravidian expansion), with upper clades being maybe responsible for Borean.
However, recent studies have shown early expansions of R1b-297 to East Europe (Mathieson et al. 2017 & 2018), and of R1b-M73 to East Eurasia probably up to Siberia, and possibly reaching the Pacific (Jeong et al. 2018). Also, the Steppe Eneolithic and Caucasus Eneolithic clusters seen in Wang et al. (2018) would be able to explain the WHG – EHG – ANE ancestry cline seen in Mesolithic and Neolithic Eurasia without a need for westward migrations.
Dravidian is now after Narasimhan et al. (2018) and Damgaard et al. (Science 2018) more and more likely to be linked to the expansion of the Indus Valley civilization and haplogroup J, in turn strongly linked to Iranian farmer ancestry, thus giving support to an Elamo-Dravidian group stemming from Iran Neolithic.
NOTE. This Dravidian-IVC and Iran connection has been supported for years by knowledgeable bloggers and commenters alike, see e.g. one of Razib Khan’s posts on the subject. This rather early support for what is obvious today is probably behind the reactionary views by some nationalist Hindus, who probably saw in this a potential reason for a strengthened Indo-Aryan/Dravidian divide adding to the religious patchwork that is modern India.
I am not in a good position to judge Nostratic, and I don’t think Glottochronology, Swadesh lists, or any statistical methods applied to a bunch of words are of any use, here or anywhere. The work of pioneers like Illich-Svitych or Starostin, on the other hand, seem to me solid attempts to obtain a faithful reconstruction, if rather outdated today.
NOTE. I am still struggling to learn more about Uralic and Indo-Uralic; not because it is more difficult than Indo-European, but because – in comparison to PIE comparative grammar – material about them is scarce, and the few available sources are sometimes contradictory. My knowledge of Afroasiatic is limited to Semitic (Arabic and Akkadian), and the field is not much more developed here than for Uralic…
If one wanted to support a Nostratic proto-language, though, and not being able to take into account genome-wide autosomal admixture, the only haplogroup right now which can connect the expansion of all its branches is R1b-M343:
R1b-L278 expanded from Asia to Europe through the Iranian Plateau, since early subclades are found in Iran and the Caucasus region, thus supporting the separation of Elamo-Dravidian and Kartvelian branches;
R1b-V88 expanding everywhere in Europe, and especially the branch expanding to the south into Africa, may be linked to the initial Afroasiatic expansion through the Pale-Green Sahara corridor (and even a hypothetic expansion with E-M2 subclades and/or from the Middle East would also leave open the influence of V88 and previous R1b subclades from the Middle East in the emergence of the language);
R1b-297 subclades expanding to the east may be linked to Eurasiatic, giving rise to both Indo-Uralic (M269) and Macro- or Micro-Altaic (M73) expansions.
This is shameless, simplistic speculation, of course, but not more than the Nostratic hypothesis, and it has the main advantage of offering ‘small and late’ language expansions relative to other proposals spanning thousands (or even tens of thousands) of years more of language separation. On the other hand, that would leave Borean out of the question, unless the initial expansion of R1b subclades happened from a community close to lake Baikal (and Mal’ta) that was also at the origin of the other supposedly related Borean branches, whether linked to haplogroup R or to any other…
NOTE. If Afroasiatic and Indo-Uralic (or Eurasiatic) are not genetically related, my previous simplistic model, R1b-Afroasiatic vs. R1a-Eurasiatic, may still be supported, with R1a-M17 potentially marking the latest meaningful westward population expansion from which EHG ancestry might have developed (see here). Without detailed works on Nostratic comparative grammar and dialectalization, and especially without a lot more Palaeolithic and Mesolithic samples, all this will remain highly speculative, like proposals of the 2000s about Y-DNA-haplogroup – language relationships.
The Dravidian language family consists of about 80 varieties (Hammarström H. 2016 Glottolog 2.7) spoken by 220 million people across southern and central India and surrounding countries (Steever SB. 1998 In The Dravidian languages (ed. SB Steever), pp. 1–39: 1). Neither the geographical origin of the Dravidian language homeland nor its exact dispersal through time are known. The history of these languages is crucial for understanding prehistory in Eurasia, because despite their current restricted range, these languages played a significant role in influencing other language groups including Indo-Aryan (Indo-European) and Munda (Austroasiatic) speakers. Here, we report the results of a Bayesian phylogenetic analysis of cognate-coded lexical data, elicited first hand from native speakers, to investigate the subgrouping of the Dravidian language family, and provide dates for the major points of diversification. Our results indicate that the Dravidian language family is approximately 4500 years old, a finding that corresponds well with earlier linguistic and archaeological studies. The main branches of the Dravidian language family (North, Central, South I, South II) are recovered, although the placement of languages within these main branches diverges from previous classifications. We find considerable uncertainty with regard to the relationships between the main branches.
With every new paper using these revamped pseudoscientific linguistic methods popular in the early 2000s, including glottochronology, Swadesh lists, phylogenetic trees, mutation rates, etc. I feel a little more like Sergeant Murtaugh…
Featured image, from the article: “Map of the Dravidian languages in India, Pakistan, Afghanistan and Nepal adapted from Ethnologue . Each polygon represents a language variety (language or dialect). Colours correspond to subgroups (see text). The three large South I languages, Kannada, Tamil and Malayalam are light red, while the smaller South I languages are bright red. Languages present in the dataset used in this paper are indicated by name, with languages with long (950 + years) literatures in bold.”
The rugged topography of the Himalayan region has hindered large-scale human migrations, population admixture and assimilation. Such complexity in geographical structure might have facilitated the existence of several small isolated communities in this region. We have genotyped about 850,000 autosomal markers among 35 individuals belonging to the four major populations inhabiting the Himalaya and adjoining regions. In addition, we have genotyped 794 individuals belonging to 16 ethnic groups from the same region, for uniparental (mitochondrial and Y chromosomal DNA) markers. Our results in the light of various statistical analyses suggest a closer link of the Himalayan and adjoining populations to East Asia than their immediate geographical neighbours in South Asia. Allele frequency-based analyses likely support the existence of a specific ancestry component in the Himalayan and adjoining populations. The admixture time estimate suggests a recent westward migration of populations living to the East of the Himalaya. Furthermore, the uniparental marker analysis among the Himalayan and adjoining populations reveal the presence of East, Southeast and South Asian genetic signatures. Interestingly, we observed an antagonistic association of Y chromosomal haplogroups O3 and D clines with the longitudinal distance. Thus, we summarise that studying the Himalayan and adjoining populations is essential for a comprehensive reconstruction of the human evolutionary and ethnolinguistic history of eastern Eurasia.
The possible scenarios based on potential sample results in terms of Y-DNA and mtDNA haplogroups seem to be generally well described, and I would bet – like Khan – for some kind of an East-West Eurasian connection. This is all pure speculation, though, and after all we only have to wait one month and see.
Out of the potential models laid out by Joseph something struck me as plainly wrong. From the section about R1a and Vedic Aryans (emphasis mine):
In the ancient DNA from Rakhigarhi, scientists identify R1a, one of the hundreds of Y-DNA haplogroups (or male lineages that are passed on from fathers to sons). They also identify H2b — one of the hundreds of mt-DNA haplogroups (or female lineages that are passed on from mothers to daughters) — that has often been found in proximity to R1a.
There is no reason whatsoever to think that this would be the research finding, but if it is, it would cause a global convulsion in the fields of population genetics, history and linguistics. It would also cause great cheer among the advocates of the theory that says that the Indus Valley civilisation was Vedic Aryan.
And it goes on to postulate reasons why such a big fuss will be created about the potential finding of haplogroup R1a, and its implications for the Out-of-India Theory. A global convulsion, no less.
It seems that all new methods involving admixture analysis, PCA, and other statistical tools to study Human Ancestry are still irrelevant for most, and indeed that Archaeology and even Linguistics are at the service of the simplistic identification of ancient languages with modern haplogroup distributions.
I really hope some R1a subclade is found among the samples, so that stupidity can reach the lowest possible level in discussions among amateur geneticists obsessed with haplogroup R1a’s role in the expansion of Indo-European speakers. Maybe then will the rest of us be able to overcome this renewed moronic supremacist trends hidden behind supposedly objective migration models.
This page allows historical linguists to compare and scrutinize proposed prehistoric lexical borrowings from the perspective of Proto-Indo-European. The first entries are all (135 in total) extracted from my master’s thesis “Foreign elements in the Proto-Indo-European vocabulary” (Bjørn 2017). Comments are encouraged at the bottom of each entry. New entries will be added, also on request.
Take this not as the conclusion, but an invitation to join the conversation.
So, we welcome the invitation, and hope that this new project thrives.
India represents an intricate tapestry of population substructure shaped by geography, language, culture and social stratification operating in concert. To date, no study has attempted to model and evaluate how these evolutionary forces have interacted to shape the patterns of genetic diversity within India. Geography has been shown to closely correlate with genetic structure in other parts of the world. However, the strict endogamy imposed by the Indian caste system, and the large number of spoken languages add further levels of complexity. We merged all publicly available data from the Indian subcontinent into a data set of 835 individuals across 48,373 SNPs from 84 well-defined groups. Bringing together geography, sociolinguistics and genetics, we developed COGG (Correlation Optimization of Genetics and Geodemographics) in order to build a model that optimally explains the observed population genetic sub-structure. We find that shared language rather than geography or social structure has been the most powerful force in creating paths of gene flow within India. Further investigating the origins of Indian substructure, we create population genetic networks across Eurasia. We observe two major corridors towards mainland India; one through the Northwestern and another through the Northeastern frontier with the Uygur population acting as a bridge across the two routes. Importantly, network, ADMIXTURE analysis and f3 statistics support a far northern path connecting Europe to Siberia and gene flow from Siberia and Mongolia towards Central Asia and India.
Among the most interesting results (emphasis mine):
Our meta-analysis of the ADMIXTURE output shows that the IE and DR populations across castes shared very high ancestry, indicating the autochthonous origin of the caste system in India (Figure 2). f3 statistics show that most of the castes and tribes in India are admixed, with contributions from other castes and/or tribes, across languages affiliations (Supplementary Table 4 and Supplementary Note). The geographically isolated Tibeto-Burman tribes and the Dravidian speaking tribes appear to be the most isolated in India. Linear Discriminant Analysis on the normalized data set clearly supports genetic stratication by castes and languages in the Indian sub-continent
Our meta-analysis of the ADMIXTURE plot in Figure 4A quantifies the ADMIXTURE results (darker colors indicate higher pairwise shared ancestry). Indian populations show a greater proportion of shared ancestry with the so-called Indian Northwestern Frontier populations, namely the tribal populations spanning Afghanistan and Pakistan. Central Asian populations share higher degrees of ancestry with IE and DR Froward castes. Uygurs share high degrees of ancestry with Indian populations.
f3 statistics (all negative Z-scores are shown) indicate Chinese and Siberian ancestry contributing to the Tibeto-Burman tribal speakers. On the other hand, the Mongols and the Europeans have contributed significant amounts of ancestry to the Indo-European and Tibeto-Burman forward castes. F3 statistics also show that the Central Asians are an admixed population with signs of admixture from Caucasus and other parts of Europe.
Among the results for proportions of shared ancestry between Indians and Eurasians (FIG. 4), there is an obvious influence of European admixture (Caucasus, and Southern, Central, and Northern EU), potentially from the Yamna-Corded Ware expansion, in IE_ForwardCaste, which is lessened in IE_BackwardCaste and also in IE_Tribal, while DR_ForwardCaste shows again more admixture than IE_Tribal, but diminishing with lower castes and quite low in DR_Tribal.
Ancestry from Central Asia is strong with a similar pattern, which hints at the influence of Sintashta, Andronovo, and BMAC influence in the expansion of the Steppe component, even more than a later Turkic component.
Quite interesting is then the influence of admixture in these different ethnolinguistic groups, Indo-European and Dravidic, which points to an initially greater expansion of Indo-European speakers, and later resurge of Dravidian languages.
Featured image contains simplified origin and data of samples studied, from the article.