Resolving the ancestry of Austronesian-speaking populations

Abstract

There are two very different interpretations of the prehistory of Island Southeast Asia (ISEA), with genetic evidence invoked in support of both. The "out-of-Taiwan" model proposes a major Late Holocene expansion of Neolithic Austronesian speakers from Taiwan. An alternative, proposing that Late Glacial/postglacial sea-level rises triggered largely autochthonous dispersals, accounts for some otherwise enigmatic genetic patterns, but fails to explain the Austronesian language dispersal. Combining mitochondrial DNA (mtDNA), Y-chromosome and genome-wide data, we performed the most comprehensive analysis of the region to date, obtaining highly consistent results across all three systems and allowing us to reconcile the models. We infer a primarily common ancestry for Taiwan/ISEA populations established before the Neolithic, but also detected clear signals of two minor Late Holocene migrations, probably representing Neolithic input from both Mainland Southeast Asia and South China, via Taiwan. This latter may therefore have mediated the Austronesian language dispersal, implying small-scale migration and language shift rather than large-scale expansion.

Fig. 1

Founder analysis results for ISEA, assuming Taiwan as source, for mtDNA (female lineages) and Y-chromosome variation (male lineages). a Probabilistic distribution of mtDNA founder clusters across migration times scanned at 200-year intervals from 0 to 70 ka, using two criteria for founder identification, f1 and f2; b probabilistic distribution of Y-chromosome founder clusters across migration times scanned at 200-year intervals from 0 to 70 ka, using two criteria for founder identification, f1 and f2; c proportion of founder lineages in a four-migration model for mtDNA and Y-chromosome variation using two criteria for founder identification, f1 and f2; d probabilistic distribution of each individual lineage in mtDNA and Y-chromosome variation in a four-migration model chromosome using two criteria for founder identification, f1 and f2. Individual founder clusters with more than 2 % frequency in overall ISEA (sink populations) are indicated at the left-hand side of each plot

Fig. 2

Frequency map of probable Neolithic markers (lineages argued to track one or other of the dispersals associated with Neolithic ceramics) in mtDNA and genome-wide data. a Pooled frequency of candidate “out-of-Taiwan”, “Neolithic II” mtDNA haplogroups, based on founder analysis. b Possible “out-of-Taiwan”, “Neolithic II” component in the genome-wide data when considering 10 ancestral populations in the ADMIXTURE analysis. c Pooled frequency of candidate MSEA “Neolithic I” haplogroups in ISEA. d Possible MSEA “Neolithic I” component in the genome-wide data when considering 10 ancestral populations in the ADMIXTURE analysis. The outline map was obtained from http://www.outline-world-map.com

Fig. 3

Reconstruction of ancestry in Asian populations using ADMIXTURE. Considering a five ancestral populations (K = 5) and b 10 ancestral populations (K = 10)

Fig. 4

Schematic tree of haplogroup M7. The tree is scaled using maximum likelihood and a time-dependent molecular clock for whole-mtDNA genomes

Fig. 5

Phylogeographic patterns in haplogroups M7c3c, E and B4a1a1. a ML ages of key clades in the test for an “out-of-Taiwan” pattern; ρ founder ages from Taiwan into ISEA; ρ founder ages from Taiwan and the Philippines into the rest of ISEA. b Detailed view of the most relevant time-frame for the data in a. c–e Increments in expansion of haplogroups B4a1a (c), E (d) and M7c3c (e), measured from Bayesian skyline plots as effective population size change per 100 individuals per 100 years, in Taiwan and ISEA