The inclusion of our 35 novel sequences (Table ) into the phylogeny of haplogroup R allows the recognition of eight new subclades within six haplogroup R branches unique to the Indian subcontinent (Fig. , [see Additional file 1
]). We refine here the internal topology of haplogroups R5, R6 and R8, and describe two novel sub-clades of hg R7, to be discussed below in detail. Subclade R5a is defined by a deletion at nucleotide positions (np) 522–523 and one control region mutation at np16266. R6a is defined by two control region substitutions (at sites 16129 and 16266). In haplogroup R7, two new subclades R7a and R7b can be identified (for details see further down). A new subclade of R8, called R8a, is defined by a single coding region substitution at np 5510. Haplogroup R30 splits into two subclades R30a and R30b, the former supported by ten coding region substitutions and the latter by 24 coding and control region mutations. Similarly, in haplogroup R31 a new subclade R31a can be distinguished by 17 control and coding region mutations. Coalescent estimates suggest an ancient branching pattern in hgs R30 and R31, dating back almost to the earliest diversification of the superhaplogroup R itself. This most probably occurred soon after the out of Africa dispersals into the Indian subcontinent [see Additional file 1
Geographical, Linguistic and Haplogroup Affiliations of Completely Sequenced mtDNAs.
Figure 1 The most parsimonious tree of haplogroup R7 complete mtDNA sequences observed in the Indian subcontinent. This tree was redrawn manually from the output of median joining/reduced network obtained using NETWORK program (version 4.1) http://www.fluxus-engineering.com (more ...)
Comparison of patterns of haplogroup distribution in relation to linguistic groups reveals that the frequency of the R7 clade is several times greater among AA (Munda) speakers than among Dravidian and Indo-European speaking populations (Table , [see Additional file 2
]). Geographically, the distribution of R7 in India is centered on the AA "heartland" (Bihar, Jharkhand, and Chhattisgarh) [see Additional file 3
]. Similar to R7, haplogroup R6 is significantly more frequent among the AA speakers than among other linguistic groups (Table , [see Additional file 2
]). PC analysis based on frequency data of the hg R subclades confirms that the majority of Munda speaking populations cluster separately from others mainly because of higher hg R7 frequency (Fig. ). However, only 50.6% of the variation can be explained by the first two principal components. Interestingly, hg R6 is placed within the main cluster, which is comprised of populations from all language groups. Based on these preliminary results we focused on R7 as a potential AA-associated marker.
Frequency of Autochthonous R Subgroups Among Different Language Groups of India.
Figure 2 Principal component (PC) analysis of R5-8, R30 and R31 lineages in Indian populations. Munda group and a few Indo-European/Dravidian populations collected from Bihar, Jharkhand and Chhattisgarh states, predominantly cluster with haplogroup R7. Haplogroup (more ...)
In general, the elevated frequency of hg R7 among the AA speakers of India can be explained by two alternative scenarios. Firstly, one may consider a possible origin of R7 among AA (Munda) speakers, possibly already outside India. Under this scenario the presence of R7 in some Dravidian and Indo-European speaking communities would be explained by its later introgression from the Munda communities, or by language shift of some Munda speaking groups into Dravidian/Indo-European languages. Secondly, an origin of R7 may lie among non-AA populations of India, with the presently observable higher frequency of R7 among AA resulting from founder effect(s) due to random genetic drift. To test these two scenarios, we carried out a detailed analysis of R7 mtDNAs in populations speaking different subgroups of AA languages, as well as among IE and Dravidian-speaking populations of Indian subcontinent.
Complete mtDNA sequence-based topology of hg R7 divulges two deep-rooted subclades (Fig. ). R7a is defined by four and R7b by six coding region mutations and, in addition, by two control-region substitutions (146 and 16311). We calculated the time to the most recent common ancestor (MRCA) for all R7 major sub-clades (Fig. and , Table ), applying different calibration methods [30
]. All the AA individuals coalesce to the founder R7a1 that dates back to between approximately 3 Ka and 7 Ka ago, depending on the mutation rate used. The coalescent times of R7 variation among Dravidians and Indo-Europeans are older. In other words, the only R7 lineage found by us in AA speakers of India – R7a1 – is nested within the R7 lineages found among Dravidian and Indo-European speakers of India (Table ).
Figure 3 The reduced-median network of 152 mtDNAs belonging to haplogroup R7. Each sample represented on the diagram has been sequenced for the HVS-I region and genotyped for the coding region mutations that are indicated. Circle sizes are proportional to the (more ...)
Coalescent times of hg R7 subclades estimated from HVS-I data.
Geographically, the distribution of R7a frequency is concentrated towards Bihar, Jharkhand and Chhattisgarh States, while R7b has its frequency peak in Andhra-Pradesh (Fig. and ). The frequency of R7a is higher among AA (Munda) speakers, while R7b is most common among Dravidian speakers from Andhra-Pradesh, although the overall frequency of R7b is much lower than that of R7a (Fig. and ). A Mantel test showed a significant correlation between genes and geography for the Indian R sub-clades, but no such correlation for the relationship between genes and languages (Table ). The spatial autocorrelation analysis favoured a clinal pattern for the distribution of hg R7 [see Additional file 4
]. At the local (i.e. district) level, R7 is present in Bihar, Jharkhand, Chhattisgarh, Madhya-Pradesh and the northern districts of Andhra-Pradesh (Adilabad, Warangal and Khammam), whereas elsewhere in India it is virtually absent, including among other AA groups inhabiting Orissa and Maharashtra states [see Additional file 5
Figure 4 The frequency distribution of R7a and R7b clades in Indian subcontinent. The upper panel (a, b) shows the spatial distribution (%) of these clades in Indian populations. Isofrequency maps were generated by using Surfer7 of Golden Software (Golden Software (more ...)
Mantel correlation test of Autochthonous R Subgroups to assess the significance of correlations between gene and geography, or language.
The overall higher than average frequency of R7 among the AA speakers of India may superficially be seen as supporting the model that places the origin of this haplogroup among AA speakers, possibly even outside India, assuming the language phylum would have arisen elsewhere. Indirectly, such a scenario would be also supported by the Y chromosome evidence (haplogroup O2a, for details, see Introduction). However, the much higher diversity of R7a and R7b sub-clades among non Austro-Asiatic populations of India suggests that the source of haplogroup R7 is not among the maternal ancestors of all Austro-Asiatic tribal groups, but that they acquired this haplogroup via local admixture, together with the rest of the South Asian mtDNA lineages that make up their extant maternal lineage pool. Furthermore, the presence of only a single recent founder branch of R7, i.e. R7a1, among widely dispersed AA populations of India supports the founder event scenario by introgression of this lineage from the local non-AA populations before the range expansion of Munda speaking populations within India. If indeed R7 did have its origin among some so far unsampled populations of the present-day Myanmar or Cambodia, we would then expect to see different sub-divided AA populations losing by drift different sub-branches of R7a and R7b (to explain their reduced diversity), and the admixed Dravidian and Indo-European speaking populations would be expected to have obtained a subset of the R7 variation observed in AA speakers, which is not the case. While the occurrence of R7a1 among Dravidian and Indo-European-speaking populations living close to the AA populations (Fig. ) could be explained by language shift or secondary admixture with AA speakers, sub-haplogroup R7b appears to be restricted to Dravidian-speakers of the southern part of India (Fig. and ). Nevertheless, this haplogroup is also reported in two Indo-European populations (Kolcha and Rathwa) whose local tradition speaks about their ancient split from the Gond (Gondi subfamily of Dravidian language group) population of Central India and further migration to Gujarat. Thus, from the data and analyses shown here, it is most parsimonious to conjecture that R7 originated in India among non-AA, possibly in Dravidian speaking populations.
To test further the two hypotheses, a Dravidian origin for R7 with admixture and founder effects, versus an external AA origin of R7, we examined whether the spread of R7 among the different Munda sub-groups in India, as defined by the language trees [26
], is uniform. This would be expected if R7 was present among the ancestral AA speakers prior to the diversification of the language family into numerous branches. Consistent with the non-AA origin of R7, we found the distribution of R7a1 among AA populations to be profoundly skewed towards the Kherwari sub-branch of the North Munda languages which accounts for ~90% of the AA R7 samples (Fig. ). Conversely, R7 is very rare in the South Munda group. It is completely absent in Koraput Munda speakers and marginally present only in the Kharia tribe of Madhya Pradesh (in total 3 out of 431 South Munda samples) (Fig. ). This finding yet again strengthens the argument that only a subset of Indian AA groups has acquired one sublineage of R7a1 in situ
after their arrival to Indian subcontinent from local non-AA groups through admixture. Thus, we fail to find from the evidence of the extant maternal lineage pool of the Austro Asiatic speakers of India any major lineages that show signs of potential origin outside India. Overall, the enigma of the origins and demographic past of the AA speakers in India remains, for while the East Asian contribution to their paternal gene pool seems evident, the maternal side of their genetic heritage appears to be autochthonous to Indian subcontinent. This suggests that introduction and spread of AA speakers into India involved a complex and sex-differentiated demography, involving both exogenous males and local females.
Figure 5 The frequency distribution of haplogroup R7 in different branches of the Austro-Asiatic language family of India.
In brief, our high-resolution study of haplogroup R7 suggests that this haplogroup originated in India among non-AA population most probably Dravidian, and that the Munda (mainly Kherwari group) speaking populations have acquired a subset of it only relatively recently. The highest frequency of haplogroup R7 among Austro-Asiatic tribal groups can be explained, thus, by their regional admixture with other local Indian subcontinental populations followed by random genetic drift, rather than being a genetic marker of their own. The spread of R7 as well as other ancient sub-clades of haplogroup R in India follows predominantly the geographic rather than linguistic landscape of the subcontinent. The geographic correlations are further manifested in the distribution patterns of the sub-clades: R7a being more common in northern India while R7b is more frequent in the southern parts of the subcontinent. Because Dravidian speakers harbour all the twigs of R7 identified so far, the haplogroup may have arisen among the matrilineal ancestry of the present day Dravidian speakers. However, it is important to caution that autochthonous basal mtDNA lineages in South as well as Southeast and East Asia appear to be significantly more ancient than any linguistic reconstruction offers to present day language families. This would imply that linguistically significant relationships among Indian populations may be superimposed on, and masking, demographic events of much greater antiquity. Our results also remind us, once again, that phylogenetically established within-haplogroup diversity is more informative than mere frequency in establishing the direction of gene flow between populations, language groups and geographically defined regions.