A mitochondrial stratigraphy for island southeast Asia

Abstract

Island Southeast Asia (ISEA) was first colonized by modern humans at least 45,000 years ago, but the extent to which the modern inhabitants trace their ancestry to the first settlers is a matter of debate. It is widely held, in both archaeology and linguistics, that they are largely descended from a second wave of dispersal, proto-Austronesian-speaking agriculturalists who originated in China and spread to Taiwan approximately 5,500 years ago. From there, they are thought to have dispersed into ISEA approximately 4,000 years ago, assimilating the indigenous populations. Here, we demonstrate that mitochondrial DNA diversity in the region is extremely high and includes a large number of indigenous clades. Only a fraction of these date back to the time of first settlement, and the majority appear to mark dispersals in the late-Pleistocene or early-Holocene epoch most likely triggered by postglacial flooding. There are much closer genetic links to Taiwan than to the mainland, but most of these probably predated the mid-Holocene "Out of Taiwan" event as traditionally envisioned. Only approximately 20% at most of modern mitochondrial DNAs in ISEA could be linked to such an event, suggesting that, if an agriculturalist migration did take place, it was demographically minor, at least with regard to the involvement of women.

Figure 1.

Map of Taiwan and Southeast Asia, showing both modern coastlines (darker shading) and the 120-m depth contour below sea level (lighter shading), indicating the extent of Sundaland at the Last Glacial Maximum. Sampling locations or populations are indicated by bold circles and are labeled with short codes if they were from a specific location or population. Taiwan: ATA = Atayal; BUN = Bunun; AMI = Ami; PAI = Paiwan. ISEA: MED = Medan; PAD = Padang; PEK = Pekanbaru; PLB = Palenbang; BGK = Bangka; KK = Kota Kinabalu; BAN = Banjarmasin; MND = Manado; PAL = Palu; TOR = Toraja; UJP = Ujung Padang; MTR = Mataram; WAI = Waingapu. Map outline kindly provided by H. Voris and C. Simpson, Field Museum of Natural History, Chicago.

Figure 2.

Schematic tree of East Eurasian mtDNA haplogroups, displaying the various branches of macrohaplogroup M (A) and macrohaplogroup N (B) and indicating diagnostic control-region positions and coding-region positions tested. We here rename R12 as R22, since R12 was used in a previous study. F1a3 is here defined solely by 16311 because the two other control-region sites mentioned as diagnostic by Kong et al. do not appear in our data set.

Figure 3.

PC analysis of mtDNA haplogroup frequencies in Taiwan and ISEA (Taiwanese data include that of Trejaut et al.41).

Figure 4.

Reduced median network of HVS-I sequences of haplogroup E in the region 16050–16400. The circles represent mtDNA HVS-I sequence types, shaded according to region and with an area proportional to their frequency in each region. Links are labeled with the nucleotide position of mutations; letters following positions indicate transversions, and the others are transitions. Mutations that have occurred more than once in the tree are underlined. Coding-region positions that were sequenced are indicated in italics. Subclades are labeled, and the M9 ancestor is indicated (arrow).

Figure 5.

Spatial frequency distributions, created using the Kriging algorithm of the Surfer package, of haplogroups E (A), M7c1c (B), D5 (C), and Y2 (D). Samples from each island were merged to a central location, indicated with a point, to reduce any effect of error in small samples.

Figure 6.

Reduced median network of HVS-I sequences of haplogroup M7c1c. They are labeled as in figure 4. The M7c1 ancestor is indicated (arrow).