• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of ajhgLink to Publisher's site
Am J Hum Genet. Jan 10, 2008; 82(1): 194–198.
Published online Jan 4, 2008. doi:  10.1016/j.ajhg.2007.09.010
PMCID: PMC2253960

Genome-wide Analysis Indicates More Asian than Melanesian Ancestry of Polynesians


Analyses of mitochondrial DNA (mtDNA) and nonrecombining Y chromosome (NRY) variation in the same populations are sometimes concordant but sometimes discordant. Perhaps the most dramatic example known of the latter concerns Polynesians, in which about 94% of Polynesian mtDNAs are of East Asian origin, while about 66% of Polynesian Y chromosomes are of Melanesian origin. Here we analyze on a genome-wide scale, to our knowledge for the first time, the origins of the autosomal gene pool of Polynesians by screening 377 autosomal short tandem repeat (STR) loci in 47 Pacific Islanders and compare the results with those obtained from 44 Chinese and 24 individuals from Papua New Guinea. Our data indicate that on average about 79% of the Polynesian autosomal gene pool is of East Asian origin and 21% is of Melanesian origin. The genetic data thus suggest a dual origin of Polynesians with a high East Asian but also considerable Melanesian component, reflecting sex-biased admixture in Polynesian history in agreement with the Slow Boat model. More generally, these results also demonstrate that conclusions based solely on uniparental markers, which are frequently used in population history studies, may not accurately reflect the history of the autosomal gene pool of a population.

Main Text

Previous analyses of mtDNA and NRY variation in Polynesians have demonstrated a dramatic discrepancy in the origins of the Polynesian mtDNA and NRY gene pools:1–4 about 94% of Polynesian mtDNAs are of East Asian origin and only 6% are of Melanesian origin (consisting of mainland New Guinea and surrounding islands, also known as Near Oceania), whereas 66% of Polynesian Y chromosomes are of Melanesian origin and only 28% are of East Asian origin.4 To explain this discrepancy, the Slow Boat hypothesis for Polynesian origins was formulated, according to which the ancestors of Polynesians migrated from East Asia/Taiwan through New Guinea and Island Melanesia, mixing extensively with the local populations as they did so, before continuing eastward to colonize Remote Oceania/Polynesia.2,4 This mixing was sex biased, involving many more Melanesian males than females admixing with East Asian newcomers, consistent with the view that ancestral Polynesian society may have been matrilocal in residence (men move to their wife's land) and matrilineal in descent (clans are inherited through the mothers line).4–6 This large discrepancy between the mtDNA and NRY genetic contributions of East Asian and Melanesian ancestors to Polynesians raises the question as to the constitution of the autosomal gene pool of Polynesians. To address this question, we genotyped 377 genome-wide distributed autosomal short tandem repeat (STR) loci in 47 Polynesians (10 Cook Islanders, 10 Tongans, 18 Samoans, 5 Tokelau Islanders, and 4 Nuie Islanders), 44 Han Chinese from Beijing, and 24 Papua New Guineans from the interior highlands (15 from the Eastern Highlands, 9 from the Southern Highlands). Genotyping was performed essentially as described elsewhere.7 Additional STR loci were chosen to reduce the average distance between markers to 8.8 cM; the genotype data are available from the authors.

Some basic statistics of genetic diversity in these three groups, based on the 377 STR loci, are presented in Table 1 and indicate that diversity in the Polynesians is intermediate between that of Chinese and New Guineans. Previous evidence from NRY and mtDNA data indicates that Polynesians have NRY and mtDNA haplogroups of both East Asian and Melanesian origins.1–4,8–12 We therefore hypothesized that East Asians and Melanesians represent the parental populations of Polynesians, and we estimated the genome-wide autosomal proportion of the East Asian (represented by the Chinese samples) and Melanesian (represented by the PNG samples) contribution to Polynesians. However, the estimation of the East Asian and Melanesian components in the current Polynesian autosomal gene pool is potentially complicated by the severe bottlenecks that accompanied the migrations to and through the Pacific (Remote Oceania), as well as the relatively long period since the admixture occurred, approximately 3500 years ago.13

Table 1
Genetic Diversity in the Chinese, New Guinean, and Polynesian Samples, Based on 377 Autosomal STR Loci

Several methods have been proposed for estimating the amount of admixture of a hybrid population,14 of which two seem to be more robust to departures from model assumptions. These are the maximum likelihood method proposed by Wang15 as implemented in the program LEADMIX and the coalescent approach proposed by Bertorelle and Excoffier.16 The latter approach has been recently incorporated into an Approximate Bayesian Computation (ABC) framework,17 which seems to be especially suited for STR data.18 The model underlying these two methods assumes that an ancestral population splits into the two parental populations, which diverge for some period of time and then meet once to form the admixed population (Figure 1).

Figure 1
Model of Admixture Assumed in This Study, as Proposed Elsewhere16

We performed LEADMIX and ABC simulations on the total data set from 377 genome-wide autosomal STR loci. The ABC program was kindly provided to us by L. Excoffier. In the case of LEADMIX, we used the default parameters suggested previously;15 results are presented in Table 2. For the ABC approach, we used the same prior distributions and summary statistics as suggested previously18 with the exception of the statistic that computes the average extent of linkage disequilibrium D′ between independent markers in the admixed population. Because the time for computing this statistic increases to square with the number of loci, it turned out to considerably increase the computational time of each simulation with our current data set of 377 STRs. The proposed statistics try to capture different aspects of the demographic processes shaping the genetic diversity of the populations, including the presence of bottlenecks (i.e., the M statistics19) and the amount of admixture between the two parental populations (i.e., the my statistic). We performed 500,000 simulations with the program SIMCOAL 2.0.20 The ABC program18 was then used to estimate the parameters by means of local and weighted linear regression on the best 1000 simulations after transformation of the data.21 We increased the number of simulations suggested by Excoffier et al.18 by factor 5 to achieve reliable results given the relatively large number of genetic markers used and the expected complexity of the demographic history of the populations analyzed. Results of the ABC approach are presented in Table 3. An additional analysis considering 100,000 simulations with ascertaining the best 1000 simulations produced similar results although with larger credible intervals (data not shown).

Table 2
Maximum Likelihood Estimations of the Parameters of Interest from LEADMIX
Table 3
Results from the ABC Simulations

The resulting admixture estimates from the two methods were highly concordant (Tables 2 and 3). Also, the posterior distribution of the amount of admixture as estimated with the ABC approach showed a clear improvement of the uncertainty about the parameters when compared with the uniform prior distribution (see Figure 2). Both methods indicate that 79% of the autosomal gene pool of the Polynesian sample is of East Asian origin and about 21% is of Melanesian origin, which is more similar to the mtDNA estimates than to the NRY estimates of admixture in these very same samples (Figure 3).

Figure 2
Prior and Posterior Distribution of the Amount of Asian Admixture
Figure 3
Estimated Proportion of Asian and Melanesian Ancestry in 47 Polynesians

Our admixture estimates could be distorted by European admixture, which is prevalent in some Polynesian groups22 but is much less in the sample set from which the 47 Polynesians used here were ascertained.4 We therefore had selected Polynesian samples for autosomal STR genotyping that do not carry European mtDNA or NRY haplogroups and whose self-described ancestry does not include any Europeans for at least two generations. Nonetheless, we cannot exclude some small amount of European admixture in these samples.

The admixture estimates also depend critically on the choice of “parental” populations. Linguistic evidence strongly suggests that Taiwan was the ancestral homeland of the proto-Austronesians23 who ultimately colonized Polynesia, which in turn suggests that Taiwan Aborigines may be a more appropriate “Asian parental” population than Han Chinese. However, mtDNA and NRY evidence suggests further bottlenecks in aboriginal Taiwanese1,24 and that Han Chinese and other East Asian groups are highly similar to one another and are as similar (or even more similar) to Polynesians than are aboriginal Taiwanese.4 Thus, the use of a different East Asian group is unlikely to change the admixture estimate. With regard to the “Melanesian parental” population, the admixture between Polynesian ancestors and Melanesians took most likely place somewhere in coastal/island New Guinea (probably the Bismarck Archipelago9,10) because Austronesians arrived in Melanesia by boat, so it may be argued that a coastal/island New Guinea population would be more appropriate than a highland New Guinea population as the parental Melanesian population. However, coastal/island New Guinea populations usually exhibit some proportion of Asian mtDNA and/or NRY types4,9 as result of genetic admixture, whereas non-Austronesian-speaking highland New Guinea groups usually lack Asian-specific markers resulting from a lack of such admixture.4,25 Therefore, Highland New Guineans provide the best available estimate of a nonadmixed Melanesian population.

To investigate any possible influence of selection on these admixture estimates obtained from the autosomal STR data, we repeated the simulations for a subset of 107 putatively neutral loci that were selected to be at least 100 kb away from any known genes. The admixture and other demographic parameter estimates did not differ significantly between this subset of putatively neutral loci and the full data set (results not shown), indicating that selection on genes is not influencing the admixture estimates obtained here.

We also attempted to obtain individual-based estimates of admixture by using the program STRUCTURE,26 but the output results were highly sensitive to the model assumptions in that different results were obtained in different runs of the program. We speculate that this failure to obtain consistent results with STRUCTURE could be due to the high mutation rate of the STR loci and/or the impact of bottlenecks on the autosomal allele frequencies. This is supported by the fact that the parental populations and the Polynesians carry a relatively large number of nonshared alleles per locus (Table 1), as a result of new mutations and/or alleles lost because of drift.

Overall, the genetic data suggest a dual origin of Polynesians, with a mostly East Asian but also considerable Melanesian genetic component. This is in keeping with models of Polynesian origins, such as the Triple-I27 and Slow Boat2,4 models, which propose substantial admixture between Polynesian ancestors and Melanesians before the colonization of the Pacific (Remote Oceania). Models that do not allow for both a primarily East Asian source of Polynesians as well as considerable Melanesian genetic input, such as the Express Train28 and Entangled Bank29 models, can be rejected based on the genetic evidence. Models that propose an older origin of Polynesians, such as the Slow Boat model of Oppenheimer and Richards,30 which postulates an origin in island Southeast Asia some 17,000 years ago, are also not supported by the genetic evidence.

In addition and more generally, our results suggest that single-locus studies of population histories (in particular, mtDNA and/or the Y chromosome) should be interpreted cautiously, because they may not be representative of the entire genome. In particular, in the present case, estimates of the Melanesian contribution to Polynesians based on NRY data differ dramatically from estimates based on genome-wide STR data (Figure 3) and are thus not representative of the total (autosomal) Polynesian gene pool. However, although the autosomal STR data provide more reliable admixture estimates than can be obtained from mtDNA or NRY data (because they are based on many loci and are representative of the entire genome), the mtDNA and NRY data nonetheless provide insights into the admixture process that the autosomal data cannot provide. In particular, the large discrepancy in the estimated Asian and Melanesian contributions to Polynesians for mtDNA versus the Y chromosome suggests sex-biased genetic admixture during Polynesian history. This admixture mostly involved Melanesian men (as evidenced by the high proportion of Melanesian Y chromosomes in Polynesia) and Asian women (as evidenced by the high proportion of Asian mtDNAs in Polynesia), an interesting finding that cannot be observed from the autosomal data. This scenario is supported by suggestions of matrilineal descent and matrilocal residence in the ancient Polynesian society,5,6 which would therefore favor incorporation of Melanesian men rather than women into the ancestral Polynesian groups. Thus, although it is entirely appropriate (and necessary) that genetic studies of human population history should turn to autosomal data, in particular to take full advantage of the increasingly sophisticated demographic models as well as of increasingly automated genotyping technologies that are now becoming available, studies of mtDNA and NRY variation can still provide invaluable insights into the history of human populations.


We are grateful to all individuals who provided samples for this study. We thank Inka Szangolies for excellent technical assistance, Gunter Weiss for his contribution to a previous version of this work, and Guillaume Laval for help with the program SIMCOAL2. Financial support was provided by the Max Planck Society.

Web Resources


1. Melton T., Peterson R., Redd A.J., Saha N., Sofro A.S., Martinson J., Stoneking M. Polynesian genetic affinities with Southeast Asian populations as identified by mtDNA analysis. Am. J. Hum. Genet. 1995;57:403–414. [PMC free article] [PubMed]
2. Kayser M., Brauer S., Weiss G., Underhill P.A., Roewer L., Schiefenhovel W., Stoneking M. Melanesian origin of Polynesian Y chromosomes. Curr. Biol. 2000;10:1237–1246. [PubMed]
3. Capelli C., Wilson J.F., Richards M., Stumpf M.P., Gratrix F., Oppenheimer S., Underhill P., Pascali V.L., Ko T.M., Goldstein D.B. A predominantly indigenous paternal heritage for the Austronesian-speaking peoples of insular Southeast Asia and Oceania. Am. J. Hum. Genet. 2001;68:432–443. [PMC free article] [PubMed]
4. Kayser M., Brauer S., Cordaux R., Casto A., Lao O., Zhivotovsky L.A., Moyse-Faurie C., Rutledge R.B., Schiefenhoevel W., Gil D. Melanesian and Asian origins of Polynesians: mtDNA and Y chromosome gradients across the Pacific. Mol. Biol. Evol. 2006;23:2234–2244. [PubMed]
5. Hage P. Was Proto-Oceanic society matrilineal? J. Polynesian Soc. 1998;107:365–379.
6. Hage P., Marck J. Matrilineality and the Melanesian origin of Polynesian Y chromosomes. Curr. Anthropol. 2003;44:121–127.
7. Saar K., Geller F., Ruschendorf F., Reis A., Friedel S., Schauble N., Nurnberg P., Siegfried W., Goldschmidt H.P., Schafer H. Genome scan for childhood and adolescent obesity in German families. Pediatrics. 2003;111:321–327. [PubMed]
8. Trejaut J.A., Kivisild T., Loo J.H., Lee C.L., He C.L., Hsu C.J., Li Z.Y., Lin M. Traces of archaic mitochondrial lineages persist in Austronesian-speaking Formosan populations. PLoS Biol. 2005;3:e247. [PMC free article] [PubMed]
9. Scheinfeldt L., Friedlaender F., Friedlaender J., Latham K., Koki G., Karafet T., Hammer M., Lorenz J. Unexpected NRY chromosome variation in Northern Island Melanesia. Mol. Biol. Evol. 2006;23:1628–1641. [PubMed]
10. Friedlaender J.S., Friedlaender F.R., Hodgson J.A., Stoltz M., Koki G., Horvat G., Zhadanov S., Schurr T.G., Merriwether D.A. Melanesian mtDNA complexity. PLoS ONE. 2007;2:e248. [PMC free article] [PubMed]
11. Pierson M.J., Martinez-Arias R., Holland B.R., Gemmell N.J., Hurles M.E., Penny D. Deciphering past human population movements in Oceania: provably optimal trees of 127 mtDNA genomes. Mol. Biol. Evol. 2006;23:1966–1975. [PMC free article] [PubMed]
12. Hurles M.E., Nicholson J., Bosch E., Renfrew C., Sykes B.C., Jobling M.A. Y chromosomal evidence for the origins of oceanic-speaking peoples. Genetics. 2002;160:289–303. [PMC free article] [PubMed]
13. Kirch P.V. Blackwell; Oxford: 1997. The Lapita Peoples: Ancestors of the Oceanic World.
14. Choisy M., Franck P., Cornuet J.M. Estimating admixture proportions with microsatellites: comparison of methods based on simulated data. Mol. Ecol. 2004;13:955–968. [PubMed]
15. Wang J. Maximum-likelihood estimation of admixture proportions from genetic data. Genetics. 2003;164:747–765. [PMC free article] [PubMed]
16. Bertorelle G., Excoffier L. Inferring admixture proportions from molecular data. Mol. Biol. Evol. 1998;15:1298–1311. [PubMed]
17. Beaumont M.A., Zhang W., Balding D.J. Approximate Bayesian computation in population genetics. Genetics. 2002;162:2025–2035. [PMC free article] [PubMed]
18. Excoffier L., Estoup A., Cornuet J.M. Bayesian analysis of an admixture model with mutations and arbitrarily linked markers. Genetics. 2005;169:1727–1738. [PMC free article] [PubMed]
19. Garza J.C., Williamson E.G. Detection of reduction in population size using data from microsatellite loci. Mol. Ecol. 2001;10:305–318. [PubMed]
20. Laval G., Excoffier L. SIMCOAL 2.0: a program to simulate genomic diversity over large recombining regions in a subdivided population with a complex history. Bioinformatics. 2004;20:2485–2487. [PubMed]
21. Hamilton G., Stoneking M., Excoffier L. Molecular analysis reveals tighter social regulation of immigration in patrilocal populations than in matrilocal populations. Proc. Natl. Acad. Sci. USA. 2005;102:7476–7480. [PMC free article] [PubMed]
22. Hurles M.E., Irven C., Nicholson J., Taylor P.G., Santos F.R., Loughlin J., Jobling M.A., Sykes B.C. European Y-chromosomal lineages in Polynesians: a contrast to the population structure revealed by mtDNA. Am. J. Hum. Genet. 1998;63:1793–1806. [PMC free article] [PubMed]
23. Blust R. Subgrouping, circularity and extinction: some issues in Austronesian comparative linguistics. Symposium Series Inst. Ling. Acad. Sinica. 1999;1:31–94.
24. Kayser M., Brauer S., Weiss G., Schiefenhovel W., Underhill P.A., Stoneking M. Independent histories of human Y chromosomes from Melanesia and Australia. Am. J. Hum. Genet. 2001;68:173–190. [PMC free article] [PubMed]
25. Kayser M., Brauer S., Weiss G., Schiefenhovel W., Underhill P., Shen P., Oefner P., Tommaseo-Ponzetta M., Stoneking M. Reduced Y-chromosome, but not mitochondrial DNA, diversity in human populations from West New Guinea. Am. J. Hum. Genet. 2003;72:281–302. [PMC free article] [PubMed]
26. Pritchard J.K., Stephens M., Donnelly P. Inference of population structure using multilocus genotype data. Genetics. 2000;155:945–959. [PMC free article] [PubMed]
27. Green R.C. The Lapita cultural complex: current evidence and proposed models. Indo-Pacific Prehist. Assoc. Bull. 1991;11:295–305.
28. Diamond J.M. Express train to Polynesia. Nature. 1988;336:307–308.
29. Terrell J.E. History as a family tree, history as an entangled bank: constructing images and interpretations of prehistory in the South Pacific. Antiquity. 1988;62:642–657.
30. Oppenheimer S.J., Richards M. Polynesian origins. Slow boat to Melanesia? Nature. 2001;410:166–167. [PubMed]

Articles from American Journal of Human Genetics are provided here courtesy of American Society of Human Genetics
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • MedGen
    Related information in MedGen
  • PubMed
    PubMed citations for these articles

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...