Genetic History of India

Genetic structure of study populations
Our analysis based on Y chromosomal UEPs apportioned the study samples into 13 haplogroups representing 10 major Y lineages (C5, H1a*, H2, R1a1*, R2, J2, L1, F*, K* and Q3) in India. Haplogroup diversity (0.772) was found to be comparable with Dravidian speaking populations (0.723), but higher than Indo-European speaking populations (0.684) in the country [54]. Given the number of views in support of Austro-Asiatic and Dravidian speaking tribes belonging to the primary pre-historic population of India [7], [11]; the high gene diversity values observed in the study populations, which are in turn comparable to Dravidian speaking populations, are suggestive of greater antiquity, large effective size or role of gene flow in these groups. The analysis of molecular variance revealed that the extent of genetic differentiation was high among study populations which could be attributed to either the lower effective population size of these groups or the Y chromosome making them vulnerable to the effect of genetic drift which further accelerates the process of differentiation between the populations. But, accentuated differentiation due to genetic drift is expected to be accompanied with lower diversity. However, in the present study, elevated levels of gene diversity, rule out a major role of genetic drift in shaping the observed pattern of genetic differentiation. Nevertheless, the explanation for the observation of high gene diversity coupled with high-level of genetic differentiation could lie in an overwhelming genetic admixture from different sources or in an early inflow of genes from a common source followed by rapid population expansion and subsequent fission into isolated, endogamous populations. Results from Harpending and Ward analysis do not appear to favour the former explanation. Thus, the most plausible explanation for the current diversity scenario in the Indian population seems to be the suggestion made by Majumder et al. [17], in which an early demographic expansion of modern humans within India during Palaeolithic period and later fission of the populations is conjectured.



In-situ versus Ex-situ Origin of Y lineages
The indigenous versus exogenous origin of Indian paternal lineages has been widely contested. Among the study populations, six sub-haplogroups namely, C5, H1a*, H2, J2, R1a1* and R2 constituted the major paternal lineages that together accounted for 85.92% of the Y chromosomes. While the indigenous origin of sub-clades C5, H1a*, H2 and R2 are accepted, the status of sub-clades J2 and R1a1* are contested as they are believed to have been introduced in India with the demic diffusion of Proto-Dravidian Neolithic agriculturists from West Asia and the influx of Indo-European pastorals from Central Asia. It is worth mentioning here that the high frequency and associated diversity of a haplogroup is correlated with the possible place of origin of a particular haplogroup [55]. In the present investigation haplogroup H, especially H1a, represented the most frequently observed Y chromosomal lineage followed by H2 sub-haplogroup. Its higher frequency among the Indian tribes particularly among the Dravidian speaking tribes of South India and its limited presence elsewhere on the Indian subcontinent had led some scholars to denote it as a tribe-specific haplogroup [36]. However, several subsequent studies have confirmed the presence and equal prevalence of haplogroup H and its associated H1a and H2 branches across linguistically and ethnically diverse populations and in different regions of India, except the North-Eastern region [23]–[24], [54]; thus ruling it out as a tribal-specific marker and supporting the uniform distribution of haplogroup H among Indian populations. An Indian homeland for haplogroup H can also not be refuted keeping in view its higher microsatellite diversity among Indian populations [24]. Haplogroup H has also been reported from Central Asian, West Asian and Gypsy populations in Europe. However, its low frequency and prevalent diversity pattern in Central Asia and West Asia could be due to recent back migration [32], [52]. On the other hand, the established Indian ancestry of gypsy populations is surely the reason for elevated levels of H haplogroup among them [56]. All the observations, therefore clearly point towards in-situ origin of haplogroup H among Indian populations.



Haplogroup C is widely distributed in Eastern and Central Asia, Oceania and Australia [44]. As expected, all the Y chromosomes under C haplogroup belonged to the C5 sub-clade. It was observed in a frequency of 8.45% which is the highest ever reported frequency for C5 in India and whose spread is circumscribed along the coastal belt of India [24], [54]. High STR diversity in India and South-East Asia in the backdrop of haplogroup C has also been observed [34], [57]. Interestingly none of the C haplogroup derivatives frequent in South-East Asia have been reported from India. Therefore, the possibility of introduction of C5 haplogroup from South-East Asia as a result of back migration to India appears doubtful and its indigenous origin appears to be more probable [13].

Haplogroup J is predominantly found among the populations of West Asia, North Africa, Europe, Central Asia, Pakistan, and India [44] and widely linked with the spread of agriculture from the Fertile Crescent that extends from Israel to Western Iraq. In the Indian subcontinent two sub-clades of J - J2a and J2b have been reported. Consistent with the previous studies, a higher proportion of J2 in West India as compared to North and South regions of India has been recorded in the present study [24], [54]. Two models pertaining to the homeland for Indian J2 sub-clades have been contested, West Asia and Central Asia. Cordaux et al. [35] had proposed the Central Asian homeland for Indian J2 sub-clade mainly because of higher frequency of J2 in Central Asia. It is worth mentioning here that J1 sub-clade which appears in appreciable frequency in Central Asia is largely absent from the Indian as well as most of the West Asian populations [44]. Haplogroup R is represented by two sub-clades R1a1* and R2 among the study populations. After haplogroup H1a, haplogroup R1a1* represented the most frequently occurring haplogroup. Haplogroup R is widely distributed in Central Asia, Eastern Europe, West Asia and the Indian subcontinent [58]–[59]. The higher frequency of haplogroup R1a, up to 63% in Central Asia and its relatively lower occurrence in other regions has been linked with Central Asian origin of R1a clade [36]. However, later studies [24], [54] showing higher prevalence of R1a sub-clade along with high microsatellite diversity among the tribal populations of India lend support to probable South Asian origin of R1a sub-clade as suggested by Kivisild et al. [34]. This is further substantiated by almost the complete absence of other derivatives of haplogroup R1 among the Indian populations, which is expected in case of the inflow from Central Asia [23]–[24], [41].

In the present investigation sub-clade R2 occurred with a frequency of 9.51%, which is similar to its frequency reported from other Indian populations [24], [34], [36], [54]. The frequency of R2 decreases as one goes further west from India and its frequency is almost negligible in Europe. Moreover, its frequent occurrence among Dravidian speaking groups as compared to Indo-European or Austro-Asiatic speaking groups of India can be attributed to its indigenous Indian origin.

The MDS plot (Figure 4) also reflects the closeness of South Asian populations with West Asian and Central Asian populations possibly due to overlapping of haplogroups. In comparison to the world populations the overall mean haplogroup diversity among the study populations was relatively higher than in the European or East Asian populations [32], [48] whereas it was found to be lower than that of Central Asian and West Asian populations [49], [51]. As mentioned earlier the high frequency and diversity of a haplogroup is indicator of the possible place of origin of a particular haplogroup [55]. Consequently, the observed haplogroups in the present investigation could be apportioned into three Y lineages based on their possible origin. These lineages include Central Asian, West Asian and indigenous Indian Y lineages.

Khurana P, Aggarwal A, Mitra S, Italia YM, Saraswathy KN, Chandrasekar A, et al. (2014) Y Chromosome Haplogroup Distribution in Indo-European Speaking Tribes of Gujarat, Western India. PLoS ONE 9(3): e90414.

Fig. 1
Schematic phylogeny of South Asian autochthonous mtDNA haplogroups, based on ML age estimates. Node ages for haplogroup U2 were estimated in an independent analysis. Colours correspond to the putative origin of each branch

Although haplogroup M in Asia has been shown to depart from a strict molecular clock [62], we found no evidence for a clock violation when performing a LRT (p > 0.05). Curiously, however, we found violations to the molecular clock for South Asian R lineages (p < 0.00001). Since ML analysis is partly based on the tree structure, it averages the branch lengths and provides similar estimates to a previous relaxed clock [63]. The values indicated throughout the text are therefore ML estimates (corrected for purifying selection). This is not observed in the global mtDNA tree [30, 64] and seems peculiar to the haplogroups in South Asia, due to demographic effects, as we argue below.

There are two major founder clades detected in South Asia (haplogroup N is very rare and its age does not correspond to a founder age). As previously, the age of haplogroup M, at 50.1 [44.8; 55.5] ka, and R, at 64.5 ka [55.9; 73.2] are younger than the Mount Toba eruption (~74 ka), suggesting a later arrival [2]. Haplogroup R and several of its subclades (R7, R30, R31) appear older than M, but this may be illusory—see below. The older clades in R predominate in the west and south of the Subcontinent, supporting a southern coastal route of primary colonization [1–3].

The phylogeography of haplogroup M is complex. While some older lineages (e.g. M2, M6, M32’56, M36, M39) originated in the western or southern regions of the Subcontinent (similarly to R), others trace to central India (M4’67, M35, M52) or the east (M13b, M31, M42b, M61, M49, M50 and M60). We need to tease out these more detailed patterns to explain the discrepancy in the age estimates.

If we perform regional estimates simply by considering all samples of each region, no discernible patterns are apparent, with M age estimates in the south and east showing similar ages (Table 2). However, when we take into account the inferred source for each clade and repartition the data on that basis, the re-estimated age for M in the west becomes 55.3 [45.1; 65.9] ka—higher than across the rest of the Subcontinent (Table 2). This suggests an early expansion in the west, similar to R, and a common origin and spread of both M and R along the southern coastal route, as also suggested recently from analyses of ancient DNA (aDNA) [65]. Although M has previously been dated to an earlier age in East Asia [30, 66], the lower age of M in the east of the Subcontinent versus the west argues against an eastern origin of M as recently proposed [35].

Table 2
Age estimates (in ka) of haplogroup M in different regions of South Asia: (1) using the raw modern geographic distribution and (2) considering the most probable origin of each major haplogroup and including only basal lineages of each region

ML ρ whole mtDNA ρ synonymous clock
West 47.7 [41.3–54.2] 37.4 [31.6–43.2] 39.0 [28.8–49.2]
South 47.2 [41.5–53.1] 42.4 [36.7–48.3] 40.0 [31.4–48.6]
East 47.7 [42.5–53.0] 42.4 [38.4–46.6] 43.9 [37.1–50.8]
Central 43.6 [38.1–49.1] 40.8 [35.4–46.3] 41.4 [33.0–49.7]
(2) West 55.3 [45.1–65.9] 44.5 [32.5–57.0] 50.6 [29.7–71.4]
South 48.9 [42.1–55.8] 47.5 [39.2–56.0] 41.1 [29.6–52.6]
East 45.2 [38.8–51.8] 40.8 [34.6–47.0] 40.1 [31.3–48.9]
Central 39.5 [31.9–47.2] 33.0 [26.8–39.3] 34.80 [23.2–46.5]


This result suggests that an ancient western ancestry may have been disguised by further re-expansions of haplogroup M in South Asia. Several branches of M (M38, M65, M45, M5b, M5c, M34, M57, M33a) display signals of dispersals from the east and the centre dating to ~45–35 ka, and M4’67 (which is only separated by a single mutation from the root of M), with a possible origin in central India, displays an extraordinary multi-branching structure dating to 38.0 [30.1; 46.0] ka, suggesting a major expansion at that time. If we consider that a root type of M could have survived for ~10,000 years after it arose (as is evident from modern clades within that age range), it is plausible that re-expansion created a secondary founder effect within M that decreased the overall age estimates. Such a scenario would impact even more on ρ than ML estimates, which is indeed what we see (Table 1). An expansion 45–35 ka would also fit well with the palaeoenvironmental and archaeological evidence [2, 67, 68], and is further supported by an increment in N e associated with M across South Asia from ~40 ka (Additional file 1: Figure S1).

The next major discernible signal in indigenous lineages begins ~12 ka, at the Pleistocene/Holocene transition. Various star-like clades dating 12–9 ka suggest a rapid expansion across the Subcontinent, namely M6a1a (11.4 ka), M18a (9.2 ka), M30d (12.1 ka), R8b1 (11.6 ka) and U2b2 (9.2 ka), all from a southern source; and R30c + 373 (12.4 ka), from the west. An increment in N e is also observed at this time in the BSP for haplogroup M in the west and south (Additional file 1: Figure S1).

We also see a further increment in the last few millennia. BSPs for M in the west and centre show an increment in the last 2.5 ka (Additional file 1: Figure S1), associated with the emergence of several subclades in the west (M2a3a + 4314, M2a1b, M2c + 1888 + 146, M30a2, M5a3b, M6a1 + 5585 + 146 + 1508) and centre (M2a1a1b, M3b, M3a1a, M63, M5a2a2 + 234, M5a3a and M61a + 5294).