A mitochondrial stratigraphy for island southeast Asia

Abstract

wan at the periphery—with only sporadic mainland influence—and with the languages most likely emerging during the Holocene within “Austronesia”18 itself. This would imply an early offshoot to Taiwan and subsequent leveling of language diversity across the Malayo-Polynesian area as a result of the formation of extensive socioeconomic networks.5,6,18 The debates reviewed above can be tested with genetic data. The Out of Taiwan model would predict that at least some ISEA lineages should trace back to ancestral populations in Taiwan and, ultimately, South China, with the age of those ISEA-specific lineage clusters being no more than ∼4,000 years ago. Earlier dispersals, perhaps stimulated by climate change and sea-level rise, as suggested in several alternative models,5,6,18 would, in contrast, predict dispersals centered on ISEA and dating to the late Pleistocene or early Holocene. If several distinct dispersal processes shaped the genetic variation in the region, their signatures should each be evident in the genetic record. A number of genetic studies have already attempted to address the question of a putative Austronesian expansion, but most mtDNA studies to date have focused on Pacific islanders, with little work on the potential source populations. Because of this and because most studies only examined the fast-evolving first hypervariable segment of the control region (HVS-I), firm conclusions have been elusive. Indeed, the existing mtDNA data have been used both to support21–26 and to contradict27–29 the Out of Taiwan model. Better sampling in Southeast Asia has been achieved in a number of Y-chromosome studies,30–34 and various partitions of the data into Neolithic and pre-Neolithic have been suggested. However, different studies with similar data sets have failed to agree on how the data should be partitioned. This is a consequence of poor phylogenetic resolution—insufficient markers to allow reconstruction of the main branches of the genealogical history. This, in turn, results in poor phylogeographic resolution, since branches of the tree that may be distributed differently in space cannot be distinguished. The use of more markers will be necessary before the Y-chromosome contribution to the debate can be clarified. In the meantime, we have addressed the poor quality of previous mtDNA data, both by sampling almost 1,000 individuals from locations throughout ISEA and by analyzing the samples at a higher resolution than done previously, by including coding-region as well as control-region variants gleaned from complete sequence data. This dramatically improved data set substantiates recent suggestions that major rethinking is needed with regard to the prehistory of the region.