Abstract
Austronesian languages are spread across half the globe, from Easter Island to Madagascar. Evidence from linguistics and archaeology indicates that the 'Austronesian expansion,' which began 4,000-5,000 years ago, likely had roots in Taiwan, but the ancestry of present-day Austronesian-speaking populations remains controversial. Here, we analyse genome-wide data from 56 populations using new methods for tracing ancestral gene flow, focusing primarily on Island Southeast Asia. We show that all sampled Austronesian groups harbour ancestry that is more closely related to aboriginal Taiwanese than to any present-day mainland population. Surprisingly, western Island Southeast Asian populations have also inherited ancestry from a source nested within the variation of present-day populations speaking Austro-Asiatic languages, which have historically been nearly exclusive to the mainland. Thus, either there was once a substantial Austro-Asiatic presence in Island Southeast Asia, or Austronesian speakers migrated to and through the mainland, admixing there before continuing to western Indonesia.
Introduction
The "Austronesian expansion" is one of the most significant migrations in human history, spreading languages and culture from Madagascar to Easter Island. While linguistic and archaeological evidence (like red-slipped pottery) strongly points to Taiwan as the starting point roughly 4,000–5,000 years ago, genetic evidence has been historically complex. Some studies suggested the expansion was purely cultural diffusion without large-scale migration, while others noted a sharp genetic divide between western and eastern Indonesia. This study addresses these controversies by analyzing genome-wide data from 56 populations to distinguish between multiple waves of migration and determine the origins of mainland genetic signatures in Island Southeast Asia (ISEA).
Methods and Data
The researchers utilized a sophisticated modeling approach to move beyond simple clustering methods:
- MixMapper 2.0: A tool developed for this study to build phylogenetic "scaffold" trees of unadmixed populations and map admixed populations onto them as two-way or three-way mixtures.
- Statistical Tools: $f$-statistics ($f_2$ and $f_4$) were used to measure allele sharing, and ALDER software estimated the timing of admixture events via linkage disequilibrium analysis.
- Datasets: The study integrated 18,412 SNPs from the HUGO Pan-Asian SNP Consortium and the CEPH-Human Genome Diversity Panel (HGDP), validated with a subset from the Affymetrix Human Origins array (9,032 SNPs).
Key Insights: A Complex Migration History
The Universal Taiwan Connection
Every sampled Austronesian group—from the Philippines to Fiji—harbors ancestry (30–90%) most closely related to aboriginal Taiwanese (Ami and Atayal). This finding effectively settles the debate in favor of a "Taiwan-centered" migration model over models of pure cultural diffusion.
The Austro-Asiatic "Surprise"
A major discovery is that western ISEA populations (e.g., Sumatra, Java, Borneo) carry a substantial (10–60%) genetic component from a source nested within mainland Austro-Asiatic speakers, such as the H’tin and Plang of Thailand. Since Austro-Asiatic languages are nearly absent from ISEA today, this suggests either a lost Austro-Asiatic presence in the islands or that Austronesian migrants admixed on the mainland before reaching western Indonesia.
Indigenous Layers: Negrito and Melanesian
The study identifies two distinct indigenous components: a "Melanesian" source related to Papuans (dominant in eastern Indonesia) and a "Negrito" source (dominant in the Philippines). This confirms that the Austronesian expansion was a process of continuous mixing with diverse local groups.
Results and Admixture Timing
The researchers found that 14 populations were best modeled as two-way admixed, while 11 (mostly from western ISEA) required a three-way admixture model. Using ALDER, admixture was estimated to have occurred roughly 30–65 generations ago (approx. 0.9–1.8 kya). These dates likely reflect the most recent major pulses of mixing rather than the initial arrival of migrants.
Discussion and Future Directions
While the study provides a robust framework, some populations (like the Batak Karo and Malay) did not fit the MixMapper models perfectly, suggesting additional unmodeled ancestry, possibly from South Asia. Furthermore, the exact location of Austro-Asiatic mixing remains an open question. The authors note that future ancient DNA studies will be crucial to pinning down the precise geography and timing of these migration pulses with higher certainty.
Summary updated