We report new data on 155 individuals from four Tamil caste populations of South India and perform comparative analyses with caste populations from the neighboring state of Andhra Pradesh. Genetic differentiation among Tamil castes is low...reflecting a largely common origin. Nonetheless, caste- and continent-specific patterns are evident. For 32 lineage-defining Y-chromosome SNPs, Tamil castes show higher affinity to Europeans than to eastern Asians, and genetic distance estimates to the Europeans are ordered by caste rank. For 32 lineage-defining mitochondrial SNPs...Tamil castes have higher affinity to eastern Asians than to Europeans. For 45 autosomal STRs, upper and middle rank castes show higher affinity to Europeans than do lower rank castes from either Tamil Nadu or Andhra Pradesh. Local between-caste variation...exceeds the estimate of variation between these geographically separated groups...Low, but statistically significant, correlations between caste rank distance and genetic distance are demonstrated for Tamil castes using Y-chromosome, mtDNA, and autosomal data.
I don't want to get too hung up on the specific conclusions made in this paper about caste X in South Indian state Y. Rather, look at these neighbor-joining networks derived from various types of genetic data:
I've added the clarifying text. The first two figures look at Y and mtDNA lineages, uniparental lines of descent through the patriline and matriline. There's some historical, methodological and analytic reasons that these regions of the genome get a lot of attention, but suffice it to say that they're only narrow slices or specific snapshots. Fixating on the discrepancies between South Asian Y and mtDNA has resulted in a great deal of scientific controversy, but I think the last figure, which the result of 45 STRs, tells most people what they want to know. The populations of South Asia are differentiated from both those of Western and Eastern Eurasia, but when the three are considered together those of Eastern Eurasia are the outgroup (i.e., in STRUCTURE, K = 2, and your sample has a bunch of brown, yellow and white folk). You somewhat know this from just looking at people. To me it's fascinating what the differences between the uniparental and autosomal data tell us in terms of "deep history." The data from this paper comes from Tamil Nadu and Andhara Pradesh, two states in the south of India. If you know your geography, you go south and eventually there's land's end. So obviously gene flow would be from the north. In Stephen Oppenheimer's The Real Eve he argues that South Asia is the second Africa in terms of the expansion of the human species; that is, was the region of Eurasia from which humans colonized Europe, East Asia, Australia and ultimately the New World. Therefore, it is no surprise that South Asia mtDNA lineages are a mix of Western and Eastern in affinities, but overall there is some bias toward the East. Since these data are from South India, the bias is relatively strong. The impact of typically West Eurasian lineages in India declines from northwest to southeast. Nevertheless, despite Tamil Nadu's southerly location many West Eurasian Y chromosomal lineages abound. At this point some of you will want to moot the Aryan invasion theory. I'd rather table that, but just offer that there's a lot of historical evidence for northwest-southeast cultural diffusion in India (artifacts, languages, etc.), and it stands to reason that many of these culture-bearers would be males. Pinning messy genetic on messier history is not my cup of tea at this point. The autosomal data seems to imply that this northwest-southeast gene flow was persistent, and the overlay upon the original substrate came to be predominant, as evidenced by anthropologists who tended to classify South Asians as marginal Caucasoids. One model of gene flow mediated by cultural patterns suggests that patrilocality was historically dominant across most of our recent history (on the order of 50,000 years). Therefore you have sharper differences in Y chromosomal variation than you do in mtDNA over short distances as gene flow was mediated through females. But other data, such as that of the "Genghis Khan" haplotype, shows that other male lineages are not so localized, and are prone to expanding across vast swaths of territory by monopolizing access to females. Both of these dynamics are those of patriarchal cultures. It could be that for most of history male lineages exchanged women between patrifocal households, so that sons stayed home while daughters cemented bonds with neighbors. But every now and then a "supermale" lineage would arise which expanded rapidly and eliminated or marginalized most other nearby male lineages. In this case gene flow would be mediated through males as the sons, grandsons, etc., of privileged patrilineages would take up residence in various localities. During the post-expansionary stage the descendants of the supermales would settle down and exchange daughters and the cycle would eventually recapitulate itself when a new supermale came on the scene. There has been about a generation or more of genetic data on South Asia from researchers. Though there have been a few cases where extremely differentiated origins for various groups have been claimed, these are usually been based on a uniparental locus (usually Y), and there have been disputes about the phylogeny of some haplogroups (M17 most famously). The autosomal picture suggests that South Asian populations, aside from a few groups such as Parsis whose recent historical origins are known, are more closely related to each other than they are to populations outside of South Asia. But, there is also a consistent finding that the "higher castes" tend to be more "European." It doesn't take much connecting the dots to resuscitate certain Victorian ideas. I think the problem here is that using Europeans and East Asians as outgroups is convenient and sharpens inferences one can make from the phylogenies, but what you are really looking at are gene flows from populations just to the northwest (Iran, Central Asia, etc.) and northeast (Burma, etc.). Bengalis who exhibit obvious Mongoloid features are not part Chinese, rather, a substantial number of tribal peoples have obviously been assimilated into a Bengali identity. Similarly, North Indians who could pass as Iranian, can for a very obvious reason, Iran abuts North India, and there has been gene flow between the two regions! To the victors go the spoils. I think the least sensational reading of these particular data is that most exogenous gene flow to the Indian subcontinent has come form the northwest as opposed to the northeast (though some cultural elements, such as rice, almost certainly came from the northeast). Additionally, there were almost certainly secondary demographic and cultural processes triggered by exogenous intrusions in India's northwest. Consider that both Nahuatl and Quechua spread as lingua franca after the Spanish conquest, as opposed to before. By a similar process the spread the of the cultural toolkit of Dravidian elites to southern India could have been set in motion events in the northern portions of the subcontinent. The Goths moved south into the Roman Empire to escape the chaos and famine resultant from the depredations of the Hunnic Confederacy, and subsequently founded two post-Roman successor states. The Roman subjects of the Franks in the 6th century were of course Gauls who had been Latinized only the past few centuries. You get the picture. Natural history can give us inductive insights when synthesized with geography. When Eurasia and North America were connected by a land bridge most of the mammalian migrations involved Eurasian genera replacing North American ones. When North America and South America were connected by the Isthmus of Panama many northern mammalian genera replaced southern ones. The principle here is that larger landmasses for whatever reason tend to produce organisms who succeed in the rough & tumble of inter-specific competition. By analogy, South India is at a disadvantage to North India in a pre-maritime era. The victories of Vijayanagara and Mysore against North Indian polities may in part be accounted for their supply-by-sea from Europeans. But prior to this Muslim polities were at an obvious advantage when it came to materiel such as horses which could not be procured in India, not only did they have the international connections, but they controlled the access routes because of basic geographical parameters. Over the past 10,000 years Western Eurasia, in particular the region between the Mediterranean and the Amu Darya have been subject to great cultural ferment. Agriculture, literacy and urban civilization all emerged in this region. There is almost certainly an advantage to being a first adopter. Northwest India was a first adopter, and so it would have been at a cultural advantage. Textual data suggests that relatively new invaders, such as Sakas or Turks, have been demographically trivial. But, integrated over thousands of years a constant and persistent flow of population south and east, pushing local elites to the genetic margins, would have a cumulative effect. Instead of focusing on one historical-demographic event, I think we need to look at the overall historic-geographic parameters which could have shaped the nature of the genetic variation we see in any given region.