Mitochondrial DNA Sequence Variation and Human Population Structure in the Indian Subcontinent

Abstract

An analysis of mitochondrial DNA (mtDNA) sequence variation among individuals of Indian origin was performed to compare mtDNA diversity within the Indian population to that of other human populations, and to examine Indian population substructure related to geographical location, spoken language and the caste system. A 195 bp non-coding segment of mitochondrial DNA was amplified and sequenced from 84 individuals representing a broad geographical distribution and three of the major castes. Sixty-one different haplotypes, and 54 polymorphic sites, corresponding to 28% of the sites analyzed, were found. The results obtained from the analysis of distributions of pairwise sequence differences, as well as phylogenetic analysis, are in agreement with rapid population expansion or steady growth as opposed to the maintenance of a constant population size. At 4 out of the 54 variable sites detected there was a significant difference between northern and southern populations. However, phylogenetic analysis revealed no significant population substructure with respect to either geographical distribution, language or the caste system. Some minor effects were observed in a significant under-representation of Dravidian language speakers, corresponding to individuals of south Indian origin, within one cluster, and a similar under-representation of members of the Vaisya caste in another cluster. The phylogenetic tree obtained from integration of the Indian sequences with sequences from other human populations produced a starlike cluster encompassing both Indian and non-Indian individuals. Clustering of Indians within the tree was not completely random as there were significantly fewer Indians than expected within one cluster. The results suggest that the time elapsed since the origin of the caste system has not been sufficiently long to produce significant differentiation between the major castes. Alternatively, the mtDNA fragment chosen for analysis may contain too few major polymorphisms to balance out background noise, and is therefore not adequately sensitive to detect local population structure.