Abstract
The origins and affinities of the ?1 billion people living on the subcontinent of India have long been contested. This is owing, in part, to the many different waves of immigrants that have influenced the genetic structure of India. In the most recent of these waves, Indo-European-speaking people from West Eurasia entered India from the Northwest and diffused throughout the subcontinent. They purportedly admixed with or displaced indigenous Dravidic-speaking populations. Subsequently they may have established the Hindu caste system and placed themselves primarily in castes of higher rank. To explore the impact of West Eurasians on contemporary Indian caste populations, we compared mtDNA (400 bp of hypervariable region 1 and 14 restriction site polymorphisms) and Y-chromosome (20 biallelic polymorphisms and 5 short tandem repeats) variation in ?265 males from eight castes of different rank to ?750 Africans, Asians, Europeans, and other Indians. For maternally inherited mtDNA, each caste is most similar to Asians. However, 20%–30% of Indian mtDNA haplotypes belong to West Eurasian haplogroups, and the frequency of these haplotypes is proportional to caste rank, the highest frequency of West Eurasian haplotypes being found in the upper castes. In contrast, for paternally inherited Y-chromosome variation each caste is more similar to Europeans than to Asians. Moreover, the affinity to Europeans is proportionate to caste rank, the upper castes being most similar to Europeans, particularly East Europeans. These findings are consistent with greater West Eurasian male admixture with castes of higher rank. Nevertheless, the mitochondrial genome and the Y chromosome each represents only a single haploid locus and is more susceptible to large stochastic variation, bottlenecks, and selective sweeps. Thus, to increase the power of our analysis, we assayed 40 independent, biparentally inherited autosomal loci (1 LINE-1 and 39 Alu elements) in all of the caste and continental populations (?600 individuals). Analysis of these data demonstrated that the upper castes have a higher affinity to Europeans than to Asians, and the upper castes are significantly more similar to Europeans than are the lower castes. Collectively, all five datasets show a trend toward upper castes being more similar to Europeans, whereas lower castes are more similar to Asians. We conclude that Indian castes are most likely to be of proto-Asian origin with West Eurasian admixture resulting in rank-related and sex-specific differences in the genetic affinities of castes to Asians and Europeans.
Shared Indo-European languages (i.e., Hindi and most European languages) suggested to linguists of the nineteenth and twentieth centuries that contemporary Hindu Indians are descendants of primarily West Eurasians who migrated from Europe, the Near East, Anatolia, and the Caucasus 3000–8000 years ago (Poliakov 1974; Renfrew 1989a,b). These nomadic migrants may have consolidated their power by admixing with native Dravidic-speaking (e.g., Telugu) proto-Asian populations who controlled regional access to land, labor, and resources (Cavalli-Sforza et al. 1994), and subsequently established the Hindu caste hierarchy to legitimize and maintain this power (Poliakov 1974; Cavalli-Sforza et al. 1994). It is plausible that these West Eurasian immigrants also appointed themselves to predominantly castes of higher rank. However, archaeological evidence of the diffusion of material culture from Western Eurasia into India has been limited (Shaffer 1982). Therefore, information on the genetic relationships of Indians to Europeans and Asians could contribute substantially to understanding the origins of Indian populations.
Previous genetic studies of Indian castes have failed to achieve a consensus on Indian origins and affinities. Various results have supported closer affinity of Indian castes either with Europeans or with Asians, and several factors underlie this inconsistency. First, erratic or limited sampling of populations has limited inferences about the relationships between caste and continental populations (i.e., Africans, Asians, Europeans). These relationships are further confounded by the wide geographic dispersal of caste populations. Genetic affinities among caste populations are, in part, inversely correlated with the geographic distance between them (Malhotra and Vasulu 1993), and it is likely that affinities between caste and continental populations are also geographically dependent (e.g., different between North and South Indian caste populations). Second, it has been suggested that castes of different rank may have originated from or admixed with different continental groups (Majumder and Mukherjee 1993). Third, the size of caste populations varies widely, and the effects of genetic drift on some small, geographically isolated castes may have been substantial. Fourth, most of the polymorphisms assayed over the last 30 years are indirect measurements of genetic variation (e.g., ABO typing), have been sampled from only a few loci, and may not be selectively neutral. Finally, only rarely have systematic comparisons been made with continental populations using a large, uniform set of DNA polymorphisms (Majumder 1999).
To investigate the origin of contemporary castes, we compared the genetic affinities of caste populations of differing rank (i.e., upper, middle, and lower) to worldwide populations. We analyzed mtDNA (hypervariable region 1 [HVR1] sequence and 14 restriction-site polymorphisms [RSPs]), Y-chromosome (5 short-tandem repeats [STRs] and 20 biallelic polymorphisms), and autosomal (1 LINE-1 and 39 Alu inserts) variation in ?265 males from eight different Telugu-speaking caste populations from the state of Andhra Pradesh in South India (Bamshad et al. 1998). Comparisons were made to ?400 individuals from tribal and Hindi-speaking caste and populations distributed across the Indian subcontinent (Mountain et al. 1995; Kivisild et al. 1999) and to ?350 Africans, Asians, and Europeans (Jorde et al. 1995, 2000; Seielstad et al. 1999).