The Genetic Structure of Pacific Islanders

Six hundred and eighty-seven microsatellites were spread across nine hundred and fifty-two people from forty-one Pacific populations. That number was not built for abstraction. It was built to settle a fight — a decades-long argument about where the Polynesians actually came from and what the genetic landscape of Melanesia could possibly tell us about the deep history of the Pacific. Friedlaender and colleagues set out to answer both questions at once. The geography matters before anything else does. The region they call Near Oceania, which includes New Guinea and the island chains immediately to its east — the Bismarcks and Solomons — was settled between roughly 50,000 and 30,000 years before present. Those early arrivals stayed put at the edge of the human world for something like 25,000 years. Sporadic contact between New Guinea and the Bismarck archipelago is visible in the archaeological record from about 22,000 years ago, but contact with Bougainville and Buka in the Solomons only appears around 3,300 years before present. Pre-Neolithic carrying capacities were tiny; New Ireland, which is more than 300 kilometers long, may have supported about 1,200 people or fewer. Small groups, long isolated, on islands surrounded by ocean. That combination is a genetic pressure cooker. The second pulse came much later. Around 3,300 years ago, Austronesian-speaking peoples from Island Southeast Asia, associated with the Lapita Cultural Complex, began moving east through the Bismarcks and into Remote Oceania. Within a few hundred years, Lapita-associated peoples had reached Tonga and Samoa. The languages they carried trace back linguistically to Taiwan, where Austronesian diversity is greatest and where the family probably originated four thousand to five thousand years ago. That timeline sets up the core tension the genetic data would eventually resolve: did Polynesian ancestors sweep rapidly through Melanesia without mixing much — the "Express Train to Polynesia" hypothesis — or did they absorb substantial Melanesian ancestry along the way, as the "Slow Boat" model proposed? Earlier studies couldn't cleanly answer that. Mitochondrial DNA, particularly a haplotype called B4a1a1, seemed to support the Express Train: a precursor form exists in Taiwan, and the haplotype dominates Central and Eastern Polynesia. But Y-chromosome distributions pointed toward significant Melanesian ancestry, favoring the Slow Boat. The problem, as Friedlaender and colleagues point out, is that mitochondrial DNA and the non-recombining Y chromosome are single-locus systems. Each has only one-quarter the effective sample size of autosomal markers — the kind found on the non-sex chromosomes — making them especially sensitive to the noise of random genetic drift. A drift event in one generation can erase or amplify a lineage signal, leading to contradictions that look like biology but are really statistics. The team's solution was to move to six hundred and eighty-seven microsatellites, which are short, tandemly repeated DNA sequences used as genetic signposts, plus two hundred and three insertion-deletion polymorphisms, for eight hundred and ninety markers in total. Hundreds of independent autosomal loci give a biparental picture of ancestry that no single chromosome can match. The sampling strategy was deliberate: fifteen to twenty-five unrelated individuals per locale, drawn from populations across Northern Island Melanesia, New Guinea, Micronesia, Polynesia, and Taiwanese Aboriginal groups, were then placed in direct comparison with the global Human Genome Diversity Project — Centre d'Etude du Polymorphisme Humain reference panel. The team also tracked a diversity measure called h-hat — scaled to effective population size, so that h-hat rising or falling tells you roughly how large or small the breeding population has been over time. Pacific populations fell toward the low end globally, with h-hat values ranging between roughly 4.8 and 2.9, confirming that reduced internal diversity is a Pacific-wide signature, not an artifact of one island or one group. And that low diversity within groups is exactly what makes the between-group differences so striking. This is the paradox at the heart of the Melanesian findings. Within any given Melanesian community, people are genetically similar to each other — homogeneous by almost any measure. But between communities, the differences are among the largest recorded anywhere on Earth. This is not a contradiction. It is a mathematical consequence of isolation. As internal heterozygosity — the probability that two randomly chosen alleles differ — drops within a population, the standard measure of between-population differentiation, FST, rises. Small, long-isolated groups become internally uniform and externally distinctive at the same time. Geography drives the pattern. Large, rugged islands produce the most isolated interior groups, and those groups show the strongest differentiation. New Britain, the largest and most topographically complex island in the sample, produces five distinct inferred genetic clusters at greater than 50 percent probability across its populations. Bougainville shows two common cluster assignments, and New Ireland shows one. The team's analysis of molecular variance — a method that partitions genetic variation across hierarchical levels — shows that among Papuan-speaking populations, the fraction of genetic variation explained by differences between groups is five point six percent, compared to three point four percent among Oceanic-speaking groups. Those numbers sound small, but in population genetics they represent enormous real-world divergence. Traditional marital migration distances for many inland communities were roughly one to two kilometers. Generation after generation, those short distances compounded into populations that are genetically worlds apart from people living on the same island. Language tracks this, but imperfectly. Papuan-speaking interior groups are the most differentiated. Oceanic-speaking coastal groups are more intermixed. That much fits the historical narrative. But here is where the data get genuinely surprising. When the team ran STRUCTURE — a clustering algorithm that infers ancestry proportions from genetic markers — across a combined East Asia and Pacific dataset, the Austronesian genetic signal in Melanesia turned out to be faint and patchy. It appeared in fewer than half of the Melanesian groups that speak Oceanic languages. Where it did appear, it never exceeded about 20 percent of ancestry. The largest values were just under 20 percent in the Kove and Saposa groups, 15 percent in the Mussau, and lower proportions in several others, including Teop, Mangseng, and Nakanai. In Papuan-speaking groups, the Austronesian signal is entirely absent. Some Oceanic-speaking groups are genetically indistinguishable from their Papuan-speaking neighbors — the Mamusi with the Ata, the Nalik and Notsi and Madak with the Kuot. Let that land for a moment. These communities speak Austronesian languages. Their genes say mostly Papuan. Language arrived; a large-scale movement of people did not. Friedlaender and colleagues are direct about the implication: rates of language borrowing and language adoption in Near Oceania have been faster and more pervasive than rates of genetic admixture. A language can spread through trade, prestige, and contact. DNA requires actual reproduction across community lines. In Melanesia, apparently, the words traveled further than the people. This finding reframes the Polynesian origins debate. In the STRUCTURE analyses, Polynesians and Micronesians cluster tightly with Taiwanese Aborigines and East Asians — the Māori, Samoans, and Micronesians all show primary affinity to Taiwanese Aboriginal groups, not to Melanesians. The ancestors of Polynesians did pass through Melanesia, and they left a small autosomal imprint in some coastal Oceanic groups along the way. But Polynesian genomes remain predominantly Taiwan and East Asia-derived. The Express Train, then, is still the best large-scale description of what happened — the route went through Melanesia, movement was relatively rapid, and admixture with indigenous Melanesians was modest and often barely detectable. What the autosomal data add, which single-locus studies could not, is the confidence to say this clearly. The contradictions between the mitochondrial and Y-chromosome evidence dissolve when you move to hundreds of independent loci. The genome-wide signal is consistent: Polynesians are not primarily Melanesian. What the Pacific ultimately shows us is what geography and isolation do to human populations when both are extreme. The differentiation documented here — within individual groups very low, between groups exceptionally high — reflects random genetic drift operating on very small founding populations over very long timescales. The same processes that shape variation everywhere are simply louder here because the islands enforce them. Language and genes can decouple entirely. Culture can move without people moving. A dataset large enough to cut through single-locus noise can resolve arguments that decades of earlier work could only deepen. The Pacific, precisely because it was settled in such distinct waves under such severe geographic constraints, gives us one of the clearest windows we have onto how the human species actually spread across the world. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.