Affinities among Melanesians, Micronesians, and Polynesians: A Neutral, Biparental Genetic Perspective

Affinities among Melanesians, Micronesians, and Polynesians: A Neutral, Biparental Genetic Perspective Lum, J. Koji. Jorde, Lynn B. Schiefenhovel, Wulf, 1943Human Biology, Volume 74, Number 3, June 2002, pp. 413-430 (Article)

Published by Wayne State University Press DOI: 10.1353/hub.2002.0031

For additional information about this article http://muse.jhu.edu/journals/hub/summary/v074/74.3lum.html

Access Provided by Universidad Nacional de Colombia at 07/14/11 9:48PM GMT

Affinities among Melanesians, Micronesians, and Polynesians: A Neutral, Biparental Genetic Perspective J. KOJI LUM,1 LYNN B. JORDE,2 AND WULF SCHIEFENHOVEL3

Abstract The human colonization of Remote Oceania, the vast Pacific region including Micronesia, Polynesia, and Melanesia beyond the northern Solomon Islands, ranks as one of the greatest achievements of prehistory. Many aspects of human diversity have been examined in an effort to reconstruct this late Holocene expansion. Archaeolinguistic analyses describe a rapid expansion of Austronesian-speaking “Lapita people” from Taiwan out into the Pacific. Analyses of biological markers, however, indicate genetic contributions from Pleistocene-settled Near Oceania into Micronesia and Polynesia, and genetic continuity across Melanesia. Thus, conflicts between archaeolinguistic and biological patterns suggest either linguistic diffusion or gene flow across linguistic barriers throughout Melanesia. To evaluate these hypotheses and the general utility of linguistic patterns for conceptualizing Pacific prehistory, we analyzed 14 neutral, biparental genetic (short tandem repeat) loci from 965 individuals representing 27 island Southeast Asian, Melanesian, Micronesian, and Polynesian populations. Population bottlenecks during the colonization of Remote Oceania are indicated by a statistically significant regression of loss of heterozygosity on migration distance from island Southeast Asia ( r = 0.78, p < 0.001). Genetic and geographic distances were consistently correlated ( r > 0.35, p < 0.006), indicating extensive gene flow primarily focused among neighboring populations. Significant correlations between linguistic and geographic patterns and between genetic and linguistic patterns depended upon the inclusion of Papuan speakers in the analyses. These results are consistent with an expansion of Austronesianspeaking populations out of island Southeast Asia and into Remote Oceania, followed by substantial gene flow from Near Oceanic populations. Although linguistic and genetic distinctions correspond at times, particularly between Western and Central-Eastern Micronesia, gene flow has reduced the utility of linguistic data within Melanesia. Overall, geographic proximity is a better predictor of biparental genetic relationships than linguistic affinities.

1 Department of International Affairs and Tropical Medicine, Tokyo Women’s Medical University, 8-1 Kawada-cho, Shinjuku-ku, Tokyo 162-8666, Japan. 2 Department of Human Genetics, Eccles Institute of Human Genetics, University of Utah Health Sciences Center, Salt Lake City, Utah 84112-5330. 3 Max-Planck-Institute for Behavioral Physiology, Von-der-Tann Str. 3 D, 82346 Andechs, Germany.

Human Biology, June 2002, v. 74, no. 3, pp. 413–430. Copyright © 2002 Wayne State University Press, Detroit, Michigan 48201-1309 KEY WORDS: MELANESIA, MICRONESIA, POLYNESIA, OCEANIA, STR, GENE FLOW, LANGUAGE

414 / lum et al. Human settlement of the Pacific began with the colonization of New Guinea 40–60,000 years before present (BP) (Groube et al. 1986). By 29,000 BP all the large, intervisible islands as far as the northern Solomons were inhabited (Wickler and Spriggs 1988). This western Melanesian region of Pleistocene colonization known as Near Oceania (Pawley and Green 1973; Green 1991; Green 1999) is also the eastern limit of Papuan-speaking people in the Pacific. Further expansion into the Pacific began within the last 3500 BP and is associated archaeologically with the Lapita cultural complex and linguistically with the Austronesian language family (Bellwood 1985; Kirch 1997; Kirch 2000). Lapita sites dated between 3500 and 3000 BP are found in the Bismarcks (Kirch 1987), the Solomons (Green 1976), Vanuatu (Hedrick 1971), Fiji (Davidson and Leach 1993), Tonga (Poulsen 1987), and Samoa (Green and Davidson 1974). The presence of pottery with nearly identical form and design motifs distributed on islands spread across 4500 kilometers of the Pacific dated within 500 years suggests an initial colonization of Remote Oceania by a single population or closely related populations with a shared culture (Kirch and Hunt 1988; Kirch 1997; Kirch 2000). By 1000 BP all of the major island groups of eastern Melanesia, Micronesia, and Polynesia, collectively known as Remote Oceania (Pawley and Green 1973; Green 1991; Green 1999), had been colonized (Spriggs and Anderson 1993; Intoh 1997). All of the populations of Remote Oceania speak Austronesian languages, and all except the people of the Marianas and Palau in Western Micronesia speak Oceanic languages, a lower-order branch of the Austronesian language family (Bender 1971; Pawley and Green 1973; Pawley and Ross 1993; Ross 1996). Thus, the restriction of Papuan languages within Pleistocene-settled Near Oceania and the distribution of Austronesian languages throughout Holocene-settled Remote Oceania suggest the colonization of these regions by distinct populations. The concordance among settlement time, geographic region, and language family coupled with the comprehensive classification of approximately 750 Papuan and 770 Austronesian languages (Foley 1992; Clark 1992) has resulted in the general use of linguistic patterns as the null hypothesis of Pacific human diversity. Much of the debate about the origins of Remote Oceanic islanders concerns the relative contributions to the region from Asia and Near Oceania. An intrusive expansion of people from Asia into Remote Oceania is suggested by both linguistic and archaeological patterns. The Austronesian languages of Remote Oceania are thought to have originated in Taiwan within the last 5000 to 6000 years (Blust 1981; Bellwood 1985; Blust 1995). Similarly, Kirch (2000) believes that although Lapita pottery originated in the Bismarcks, it was an outgrowth of similar ceramic traditions found in Indonesia and Taiwan dated to 4000 BP and 4500 BP, respectively. Thus, the languages and defining material culture of the initial settlement of Remote Oceania are both inferred to result from a recent cultural expansion out of Taiwan (Bellwood 1985; Bellwood 1987; Diamond 1988; Kirch 1997; Kirch 2000). If Remote Oceania was settled by Austronesian-speaking “Lapita people” originating in Taiwan, biological markers are expected to group Austronesian-

Genetic Affinities in the Pacific / 415 speaking Melanesians, Micronesians, Polynesians, and Asians to the exclusion of Near Oceanic populations. Although analyses of morphological traits do group Asians, Micronesians, and Polynesians together, Oceanic Austronesian-speaking Melanesians cluster with Near Oceanic, Papuan-speaking Melanesians and Aboriginal Australians (Pietrusewsky 1990a; Pietrusewsky 1990b; Hanihara 1993). Likewise, analyses of mitochondrial DNA (mtDNA) polymorphisms reveal that the majority of lineages from Micronesian and Polynesian populations belong to a monophyletic clade characterized by the region V deletion (Cann and Wilson 1983) derived from East Asia (Hertzberg et al. 1989; Lum et al. 1994; Redd et al. 1995; Sykes et al. 1995; Hagelberg et al. 1999; Lum and Cann 1998; Lum et al. 1998; Lum and Cann 2000). Although ultimately of Asian origin, the most common region-V-deleted sequences of Remote Oceanic populations belong to lineage group I.1 (Lum et al. 1994; Lum and Cann 2000), also called the “Polynesian Motif” (Redd et al. 1995), thought to be derived proximally from Indonesia (Richards et al. 1998; Lum and Cann 2000; Oppenheimer and Richards 2001). The region V deletion cluster is also found within Austronesian-speaking Melanesian populations (and some of their Papuan-speaking neighbors), but in highly variable frequencies (Hertzberg et al. 1989; Sykes et al. 1995; Lum and Cann 1998; Lum and Cann 2000; Hagelberg et al. 1999; Merriwether et al. 1999). These data suggest extensive prehistoric gene flow among Austronesian- and Papuan-speaking Melanesian populations. Recent Y-chromosome analyses have distinguished Southeast Asia as the region of highest paternal genetic diversity. In these studies, North Asian and Pacific populations are seen as distinct subsets of Southeast Asian diversity (Su et al. 2000; Capelli et al. 2001). Moreover, eastern Indonesian, Micronesian, Polynesian, and both Papuan- and Austronesian-speaking Melanesian populations share a similar haplotype distribution (Kayser et al. 2000; Kayser et al. 2001). These data support a Melanesian (or eastern Indonesian) origin of Remote Oceanic Y chromosomes (Underhill et al. 2001), leading Capelli et al. (2001) to suggest diffusion of memes rather than genes as the origin of Austronesian languages within Melanesia and other regions of Remote Oceania. This assertion is contra maternally inherited mtDNA affinities that are generally correlated with linguistic relationships among Austronesian-speaking populations (Lum and Cann 1998), particularly those of Melanesia and Polynesia (Lum et al. 1998). These inconsistent associations between linguistic and paternally or maternally inherited genetic patterns can be reconciled either by linguistic diffusion or a male-biased “Gene Flow” model of Remote Oceanic prehistory (Lum et al. 1998; Delvin et al. 2001). Comparisons of archaeolinguistic and biological patterns reveal marked differences, particularly within Melanesia. Are the differences between linguistic and biological patterns the result of linguistic diffusion or gene flow across linguistic barriers? If they are due to gene flow, did it occur during or after settlement? What does the lack of correspondence among linguistic and maternal and paternal genetic data sets say about the utility of comprehensive analyses of Austronesian and Papuan languages in the conceptualization of Pacific prehistory? To

416 / lum et al. address these questions we analyzed 14 short tandem repeat (STR) loci from 965 individuals representing one Papuan-speaking population (Asaro Valley, Papua New Guinea) and 26 Austronesian-speaking populations. The five island Southeast Asian populations include three that are putatively involved directly in the colonization of Remote Oceania (Philippines, Java, and Borneo) and two that are peripheral to the Holocene expansion (Moken and Urak Lawoi). The six Austronesian-speaking Melanesian populations span the geographic range from the Trobriands off the northeastern coast of New Guinea, through Vanuatu (Torres and Banks Islands, Espiritu Santo, Maewo, and Tanna), to Fiji. The Polynesian representatives include western (Samoa), outlier (Kapingamarangi), and eastern (Hawai’i and Rapanui) populations. The 11 Micronesian populations represent all of the major islands and archipelagos including Yap and Palau proper, their respective outer island populations, and the linguistically related Chuukese.

Materials and Methods Nine hundred and sixty-five individuals representing 27 populations were examined for this study (Figure 1). STR data from 700 individuals were previously published (Lum 1998; Lum et al. 1998), while those from 265 individuals, primarily from six populations (Kapingamarangi, n = 34; Espiritu Santo, n = 23; Maewo, n = 23; Tanna, n = 23; the Trobriands, n = 79; Kiribati, n =28) are newly examined. DNA extractions and STR allele amplification and sizing were performed as previously described (Lum et al. 1998). The Genome Data Base (GDB) identification numbers for the 14 STR loci are D1S404, D1S407, D3S1545, D3S2322, D4S1530, D5S1347, D5S612, D6S400, D6S942, D7S623, D9S762, D12S297, D20S161, and D22S417. The frequencies of the alleles at each locus within populations are available upon request. The average locus diversity (Nei 1987) of each population and FST genetic distances between populations (Weir and Cockerham 1984; Michalakis and Excoffier 1996) were estimated from the number of different alleles between all pairs of individuals using Arlequin ver. 2.000 (Schneider et al. 2000). The correspondence of the loss of genetic diversity and an increased distance from island Southeast Asia was assessed using standard linear regression (Sokal and Rohlf 1995). The significance of the pairwise FST distances was evaluated by permuting individuals between pairs of populations 10,100 times and recording the proportion of the permutations that yielded an FST greater than the actual value (Schneider et al. 2000). The pairwise genetic distances between populations were scaled and plotted in two dimensions using NTSYSpc version 2.02j (Rohlf 1998), as described previously (Lum and Cann 2000). Linguistic distances among the 26 Austronesian-speaking populations were estimated from published subgroupings (Bender 1971; Pawley and Green 1973; Jackson 1983; Jackson 1986; Pawley and Ross 1993). In contrast to the other high island languages of Western Micronesia, Yapese is considered an Oceanic Aus-

Map showing the geographic origin of the populations examined. The three regions of Remote Oceania are delineated: Micronesia in the northwest Pacific, Polynesia in the east Pacific, and the Melanesian region in the southwest Pacific. The asterisk denotes the Polynesian outlier Kapingamarangi that is geographically within Micronesia. SWP = southwest island Palau; OIY = outer island Yap; PNG = Papua New Guinea Highlands; Banks = Banks and Torres islands; and Santo = Espiritu Santo.

Genetic Affinities in the Pacific / 417

Figure 1.

418 / lum et al. tronesian language as argued by Ross (1996). Since there is no consensus on the relationships between Austronesian and other language families, the distance between all the Austronesian languages and the Papuan language spoken in the Asaro Valley of New Guinea was estimated to be between 1.5 and 3 times the maximum distances within the Austronesian language family. Varying the distance to the Papuan outgroup, altering the relative depth of the Oceanic branch, or rooting the Greater Micronesian languages (Jackson 1986) within northern Vanuatu had no significant effect on the comparisons between linguistic and genetic or geographic distances. The estimated linguistic distances among the 27 populations used in the analyses are diagrammed in Figure 2.

Figure 2.

Estimated linguistic relationships among the populations examined. “A” includes the 19 Oceanic Austronesian-speaking populations. “B” contains the eight Greater Micronesian languages (Jackson 1986). The scale bar indicates the distances used for comparisons. For example, the maximum distance between any two Oceanic languages was estimated to be 14 units, and the distance between the Austronesian and Papuan language families was estimated to be 56 units.

Genetic Affinities in the Pacific / 419 Geographic distances among populations were calculated from their latitudes and longitudes assuming the earth to be a perfect sphere. Congruence among genetic, linguistic, and geographic distance matrices was evaluated using the Mantel permutation method extended to examine partial correlations of multiple matrices (Mantel 1967; Smouse and Long 1992) using Matrix Correlation Analysis version 1.0 (Long 1996). To determine significance levels, the observed correlations were compared to null distributions generated from 10,000 permutations.

Results The average locus diversities (ALD) within populations are listed in Table 1. In general, the lowest ALDs are found in the more remote island groups of both

Table 1. Average Locus Diversities (ALD) within Pacific Island and Asian Populations Population Borneo Tanna Urak Lawoi Philippines Java Marianas Espiritu Santo Moken Fiji Palau proper Maewo Trobriands Samoa Marshalls Nauru Kosrae Banks/Torres Papua New Guinea highlands Yap proper Chuuk Outer island Yap Pohnpei Hawai’i Kiribati Southwest island Palau Kapingamarangi Rapanui Total

n

Regiona

Languageb

ALD

7 23 8 29 25 60 23 9 13 119 23 79 17 30 29 25 23 11 55 34 127 56 17 45 7 61 10 965

ISEA Mel ISEA ISEA ISEA W Mic Mel ISEA Mel W Mic Mel Mel W Pol C-E Mic C-E Mic C-E Mic Mel Mel W Mic C-E Mic C-E Mic C-E Mic E Pol C-E Mic W Mic O Pol E Pol

AN OCAN AN AN AN AN OCAN AN OCAN AN OCAN OCAN OCAN OCAN OCAN OCAN OCAN NAN OCAN OCAN OCAN OCAN OCAN OCAN OCAN OCAN OCAN

0.749 ± 0.412 0.735 ± 0.379 0.735 ± 0.396 0.734 ± 0.377 0.733 ± 0.377 0.728 ± 0.371 0.727 ± 0.375 0.724 ± 0.388 0.719 ± 0.378 0.718 ± 0.364 0.715 ± 0.370 0.713 ± 0.363 0.709 ± 0.370 0.706 ± 0.363 0.703 ± 0.362 0.703 ± 0.363 0.701 ± 0.363 0.695 ± 0.371 0.687 ± 0.351 0.687 ± 0.353 0.683 ± 0.348 0.683 ± 0.350 0.674 ± 0.354 0.669 ± 0.344 0.646 ± 0.355 0.624 ± 0.321 0.623 ± 0.335

a. ISEA = island Southeast Asia; Mel = Melanesia; W Mic = Western Micronesia; W Pol = western Polynesia; C-E Mic = Central-Eastern Micronesia; E Pol = eastern Polynesia; and O Pol = outlier Polynesia. b. AN = Austronesian; OCAN = Oceanic Austronesian; and NAN = non-Austronesian.

420 / lum et al. Polynesia and Micronesia, while the highest diversities are found in island Southeast Asian and Austronesian-speaking Melanesian populations. The regression of the decrease of ALD on the straight-line distances from island Southeast Asia (Borneo) was statistically significant (r = 0.475, p < 0.013). Archaeological and linguistic analyses indicate that although both Micronesia and Polynesia were settled ultimately from island Southeast Asia, many of the archipelagos were not settled directly. Specifically, some of the languages of Central and Western Micronesia belonging to a dialect chain called the Trukic continuum (Chuuk, outer island Yap, and southwest island Palau) that cluster with Pohnpeian to the East (Bender 1971; Jackson 1986). Similarly, there is linguistic (Pawley and Green 1973; Pawley and Ross 1993) and archaeological (Spriggs and Anderson 1993; Intoh 1997) evidence indicating that the islands of eastern Polynesia and the Polynesian outlier Kapingamarangi (geographically within Micronesia) were settled from western Polynesia. When the geographic distances between island Southeast Asia and the archipelagos of the Trukic continuum are routed through Pohnpei and the outlier and eastern Polynesian islands are routed through Samoa, the significance of the regression of ALD loss on migration distance increased (r = 0.775, p < 0.001). Correlations between genetic, linguistic, and geographic distance matrices were calculated for two sets of populations. The first set of comparisons included the 26 Austronesian-speaking populations, and the second also included the Papuan-speaking population from the Asaro valley of New Guinea. In both sets of comparisons the genetic and geographic distances were significantly correlated (r > 0.37, p < 0.003), even when linguistic distances were controlled (r > 0.35, p < 0.006; Table 2). The significance of the correlations between genetic and linguistic distances and between linguistic and geographic distances depended upon whether the Papuan speakers were included in the analyses. When only Austronesian-speaking populations were considered, the geographic and linguistic distances were significantly correlated (r = 0.35, p < 0.008) and largely independent of genetic distances. A significant correlation between genetic and linguistic distances was observed when all 27 populations are examined (r = 0.30, p < 0.037; Table 2). This correlation becomes marginal when geographic distances are controlled (r = 0.27, p < 0.062).

Table 2. Mantel Correlations among Genetic, Geographic, and Linguistic Distances

Pairwise and Partial Correlationsa STR × geography STR × geography (language) STR × language STR × language (geography) Language × geography Language × geography (STR) a. Distance in parentheses is controlled.

Austronesian-speaking Populations

All Populations

r

p

r

p

0.442 0.418 0.158 0.003 0.351 0.317

0.001 0.001 0.252 0.985 0.007 0.016

0.374 0.351 0.302 0.272 0.140 0.031

0.002 0.005 0.036 0.061 0.271 0.839

Genetic Affinities in the Pacific / 421

Figure 3.

Nonmetric multidimensional scaled plot of the FST genetic distances between populations. Regions are identified by symbols: triangles = island Southeast Asia; solid squares = Melanesia; open squares = Polynesia; solid circles = Western Micronesian high islands; open circles = Central-Eastern Micronesia. Abbreviations are as in Figure 1.

Figure 3 shows the pairwise FST distances between populations plotted in two dimensions. The stress or “badness of fit” (Rohlf 1998) of the original distance matrix and the scaled plot is 0.25. A general correspondence between genetic and geographic relationships is observed. The island Southeast Asian populations and the Marianas are clustered in the lower left quadrant, most of the Melanesian populations and Palau are in the upper left quadrant, the CentralEastern (C-E) Micronesian populations are all in the extreme right, while Fijian, Polynesian, and Western Micronesian populations tend towards the center of the plot. STR allele distributions of 27 pairs of populations were not statistically different at the 0.01 level. These pairs include 70% (19/27) of the populations examined. The levels of inferred gene flow were grouped into three classes according to the proportion of permutations of individuals between populations resulting in a higher FST value than actually observed: high ( p ≥ 0.10), moderate (0.10 > p ≥ 0.05), and low (0.05 > p ≥ 0.01) (Table 3, Figure 4).

Discussion Debate on the colonization of the Pacific centers on the relative contributions from Asia and Near Oceania to Remote Oceania. Archaeolinguistic models

422 / lum et al. Table 3. Inferred Gene Flow among 21 Pairs of Populations Gene Flow Class

Population Pairs

p

High Java Espiritu Santo Espiritu Santo Espiritu Santo Fiji Outer island Yap Maewo Tanna

Borneo Fiji Banks/Torres Tanna Samoa Samoa Banks/Torres Fiji

0.591 0.511 0.286 0.160 0.141 0.113 0.112 0.100

Moken Pohnpei Chuuk Borneo Espiritu Santo

Philippines Kosrae Samoa Philippines Maewo

0.090 0.086 0.065 0.053 0.050

Urak Lawoi Samoa Banks/Torres Java Banks/Torres Papua New Guinea Highlands Fiji Maewo Samoa Borneo Moken Moken Borneo Hawai’i

Borneo Pohnpei Tanna Philippines Fiji Espiritu Santo Kosrae Tanna Kosrae Maewo Marianas Borneo Espiritu Santo Rapanui

0.047 0.047 0.037 0.036 0.033 0.033 0.033 0.030 0.028 0.023 0.022 0.019 0.014 0.012

Moderate

Low

suggest an initial settlement of Remote Oceania by populations expanding out of Taiwan (Blust 1981; Bellwood 1985; Blust 1995; Kirch 1997; Kirch 2000). These findings prompted Diamond (1988) to suggest the colonization of Polynesia via an “express train” from southern China. In contrast to this view of a rapid, intrusive expansion of Asian populations through Near Oceania into Remote Oceania are studies of biological markers that generally cluster Austronesian-speaking Melanesians with Papuan-speaking Melanesians to the exclusion of Austronesian-speaking Micronesian and Polynesians (O’Shaughnessy et al. 1990; Pietrusewsky 1990a; Pietrusewsky 1990b; Hanihara 1993; Martinson 1996). Furthermore, studies of alpha globin haplotypes (O’Shaughnessy et al. 1990; Martinson 1996) and mtDNA (Lum et al. 1994; Sykes et al. 1995; Lum and Cann 2000) reveal haplotypes derived from Near Oceania within Polynesian and Micronesian

Map showing the 21 pairs of populations that are not significantly different at the 0.01 level. Inferred levels of gene flow are grouped into three classes, High, Moderate, and Low (Table 3), which correspond to the thickness of the lines connecting pairs of populations. Abbreviations are as in Figure 1.

Genetic Affinities in the Pacific / 423

Figure 4.

424 / lum et al. populations. These data suggest that if ever there was a “train” to Polynesia, it stopped often enough to pick up passengers along the way and continued to transport substantial numbers of people within Melanesia. There are also those who argue that current linguistic patterns have little to say about the colonization of the Pacific (Terrell 1988; Terrell et al. 1997; Terrell and Welsch 1997; Terrell et al. 2001). Recently, Capelli et al. (2001) have argued, based on Y-chromosome diversities, that the presence of Austronesian languages in Melanesia is the result of linguistic diffusion. Our analyses of neutral, biparental genetic patterns suggests a rapid initial colonization of Remote Oceania from island Southeast Asia followed by substantial gene flow among Austronesian- and Papuan-speaking populations of Melanesia. The statistically significant (r = 0.78, p < 0.001) decrease of ALD regressed on the migration distance from island Southeast Asia suggests a series of genetic bottlenecks during Oceanic settlement. This loss of STR diversity during colonization is consistent with that seen in studies of both mitochondrial DNA (Redd et al. 1995; Melton et al. 1995; Sykes et al. 1995; Lum et al. 1998; Richards et al. 1998; Oppenheimer and Richards 2001) and Y-chromosome haplotypes (Su et al. 2000; Kayser et al. 2000; Capelli et al. 2001). The populations from Melanesia do not fit this general pattern of diversity loss during colonization. Of the Melanesian populations, the Papua New Guinea highlands have the lowest ALD, while the Austronesian-speaking Melanesian populations, especially Tanna and Espiritu Santo, have among the highest ALDs (Table 1). When the Melanesian populations are excluded, the correspondence of ALD loss and migration geographic distance increases (r = 0.83, p < 0.001). These data suggest that the higher diversity in Austronesian-speaking Melanesian populations results from the fusion of Near Oceanic and island Southeast Asian gene pools. This interpretation is supported by comparisons of distance matrices. The robust correlation between genetic and geographic distances (Table 2) indicates that substantial gene flow among neighboring populations rather than isolation within the Pacific has occurred. Excluding and including the Papuan speakers from New Guinea is particularly informative. Excluding the Papuan speakers yields a significant correlation between geographic and linguistic distances (r > 0.31, p < 0.017), consistent with an origin of the Austronesian language family in island Southeast Asia followed by a dispersal through Melanesia and into Micronesia and Polynesia. In these comparisons, however, the genetic and linguistic distances are not significantly correlated. The inclusion of the Papuan speakers in the analyses yields a significant correlation between genetic and linguistic distances (r = 0.30, p < 0.037). About 10% of this correlation is due to geography, however, and when its influence is removed by partial correlation analysis the correlation between genetic and linguistic distances is only marginally significant (r = 0.27, p < 0.062). Perhaps the correlation would be more robust if multiple Papuan-speaking populations were included. This suggests that although there has been gene flow among Near and Remote Oceanic Melanesians (Table 1, Figure 3 discussed below), most of the

Genetic Affinities in the Pacific / 425 gene flow has been among Austronesian speakers, and thus the Papuan speakers remain the genetic outgroup. Because the Papuan speakers, the linguistic outgroup, are located in the middle of the geographic distribution of Austronesianspeaking populations, it is not surprising that when all 27 populations are compared linguistic and geographic distances are not significantly correlated (r < 0.15, p > 0.270). The plot of the genetic distances between populations reveals a cluster of Melanesian populations in the upper left quadrant (Figure 3). This cluster includes the Papuan speakers from the New Guinea highlands, the Austronesian speakers from the Trobriands, and all four of the populations from Vanuatu. This pattern suggests past gene flow among Papuan- and Austronesian-speaking populations within Melanesia. The position of Fiji in the center of the plot near Samoa rather than with other Melanesian populations suggests a more limited participation in gene flow from Near Oceania, consistent with studies of Fijian mtDNA (Hertzberg et al. 1989; Lum et al. 1994; Lum and Cann 2000). The moderate ALD of Fiji (Table 1) indicates that this interpretation is not the result of diversity lost during settlement. In Figure 3, the outlier Polynesians from Kapingamarangi are found near the Micronesians of Pohnpei, their geographic neighbors, rather than the other Polynesians with whom they share language and culture. A severe drought between 1916 and 1918 resulted in the death of nearly a third of the population of Kapingamarangi. At this time about half the remaining population moved to Pohnpei (Alkire 1978), and a small Kapingamarangi-derived village still exists on Pohnpei. It is unclear whether the genetic affinity between Kapingamarangi and Pohnpei results from gene flow associated with the drought and consequent residence shift or prehistoric contact. In contrast to the similarity of Kapingamarangi and Pohnpei, the populations from outer island Yap and the southwest island Palau are closer to the linguistically related Chuukese than to the high island Yapese or Palauans, respectively. These data suggest that although they are geographically near each other, there was little gene flow from the high islands into the outer islands. This barrier to gene flow is apparent even though the Yapese and the outer islanders were linked via the prehistoric Sawei Empire (Lingenfelter 1975; Alkire 1978). The Yapese were the rulers of this empire, and as such they owned and exacted regular tribute from the outer islands. The transport of goods and travel within the Sawei Empire was accomplished primarily by the outer islanders, however, and so gene flow is more likely to have been in the other direction, as evidenced by the establishment of at least one outer island maternal clan and corresponding mtDNA lineage on Yap proper (Lum and Cann 2000). The positions of the Western Micronesian archipelagos of Palau near the Melanesian cluster and the Marianas near the island Southeast Asian cluster are consistent with previous analyses of their mtDNA lineages (Lum and Cann 1998; Lum and Cann 2000). While approximately 5% of the mtDNA lineages from most Micronesian populations are inferred to be derived from Papua New

426 / lum et al. Guinea, these lineages account for over 30% of those from Palau and none from the Marianas. Thus, the analyses of STR and mtDNA variation support prehistories of these two archipelagos distinct from other Micronesian island groups, consistent with linguistic classifications of their languages outside the Oceanic subgroup of Austronesian languages (Bender 1971; Pawley and Ross 1993). The genetic distances discussed above reflect the overall relationships among all populations. Permuting individuals between populations allows one to statistically assess the specific relationships of pairs of populations. Figure 4 shows connections between 21 pairs of populations that are not significantly different at the 0.01 level. The populations of Hawai’i and Rapanui form an isolated duet in the eastern Pacific, joined by low levels of inferred gene flow. Both of these populations have low ALDs (Table 1), suggesting a common loss of diversity during the settlement of eastern Polynesia. Although the 20 remaining pairs of populations are all interconnected, there are two foci of interaction characterized by moderate-to-high levels of inferred gene flow. One group of populations includes all of the island Southeast Asian populations and centers on Borneo. The Marianas join this cluster by low levels of gene flow with the Moken. Since the Moken speak Austronesian languages but are thought to be only tangential to the colonization of Remote Oceania, this linkage suggests that the Marianas were settled early in the expansion of Austronesian-speaking people throughout island Southeast Asia. Borneo is also linked to several islands of Vanuatu, which are the westernmost populations of the other interaction sphere. The Papuan speakers from New Guinea are also connected to the second cluster by means of low levels of gene flow with Espiritu Santo. Moderate-to-high levels of inferred gene flow extend from Vanuatu through Fiji to Samoa and then out to C-E Micronesia. The connections extending from Vanuatu to Samoa are consistent with linguistic analyses (Pawley and Green 1973; Pawley and Ross 1993) and with gene flow from Near Oceania extended through Melanesia towards Polynesia. In contrast, the connections between C-E Micronesia and Samoa, and, to a lesser extent, Fiji are not consistent with linguistic expectations (Figure 2) and provide insight into the timing of gene flow within Melanesia. The similarity of STR alleles (Figures 3 and 4), mtDNA lineages (Sykes et al. 1995; Lum and Cann 1998; Lum and Cann 2000), and morphological traits (Pietrusewsky 1990a; Pietrusewsky 1990b; Hanihara 1993) of C-E Micronesians and Samoans is consistent with their derivation from a common source population. According to linguistic analyses, the languages of C-E Micronesia are alternatively derived from the southern Solomons, northern Vanuatu, or the Admiralty Islands (Blust 1984; Pawley and Green 1984; Jackson 1986). Thus, the common ancestral population of both Polynesians and C-E Micronesians is expected, on linguistic grounds, also to be ancestral to the people of Fiji and the Vanuatu archipelago (Figure 2). One explanation for the primary connections among C-E Micronesians and Samoans to the exclusion of populations from Vanuatu is postsettlement gene flow from Near Oceania into Austronesian-speaking populations of

Genetic Affinities in the Pacific / 427 Melanesia that diluted the original affinities. If this gene flow had occurred during the Holocene expansion, then both C-E Micronesians and Polynesians would be derived independently from a diverse Melanesian population. As described above, both regions show reduced genetic diversity relative to Melanesia; therefore, their resemblance to each other based on multiple biological traits is not parsimonious. In contrast, postsettlement gene flow largely restricted to Melanesia does not require a coincidental retention of similar markers after separate bottlenecks in the settlements of C-E Micronesia and Polynesia. Although there are instances of correspondence of genetic and linguistic distinctions, particularly between Western and C-E Micronesia, correlations between genetic and geographic relationships are statistically significant and robust. Furthermore, inferred gene flow has rendered questionable the utility of linguistic patterns within Melanesia. These findings are in sharp contrast to maternally inherited mtDNA relationships from Pacific populations, which tend to be correlated to linguistic rather than geographic affinities (Lum and Cann 1998; Lum et al. 1998). These data suggest that interisland travel and gene flow after settlement was primarily undertaken by males, supporting the “Gene Flow” model of Remote Oceanic prehistory (Lum et al. 1998; Delvin et al. 2001). Since Melanesia has been inhabited since the Pleistocene and is the most linguistically diverse region of the planet, more research including a better sampling of the genetic variability of this region must be undertaken to confirm these conclusions. Acknowledgments We thank all the people who made this study possible by donating DNA samples. We also express our appreciation to those who facilitated the collection of samples, particularly the traditional leaders of the Palau (Rubekul Belau) and of Yap State (Council of Pilung), M. Andrew, R.K. Blaisdell, C. Ching, A. di Piazza, J. Filefney, S. George, Y. Gibbons, J. Gilmatam, J. Haleyalmang, G. Heathcote, N. Hinshiranan, Y. Hsia, J.J. Martinson, D. Otobed, A. Polloi, E. Petrick, R. Puri, S. Serjeantson, D. Sol, J. Winter, and J. Wozniak. We also thank R. Blust and K. Rehg for their insights into Austronesian linguistics. This work was supported by the Wenner-Gren Foundation for Anthropological Research through grant 5640, the National Science Foundation (NSF)/Sloan through grant BIR-9510748, the National Science Foundation through grant SBR-9818215, the National Institutes of Health through grant GM-59290, and the Ministry of Education, Culture, Sports, Science, and Technology of Japan through grant 13576030. Received 5 September 2001; revision received 7 January 2002.

Literature Cited Alkire, W.H. 1978. Coral Islanders. In Worlds of Man: Studies in Cultural Ecology, W. Goldschmidt, ed. Arlington Heights: AHM Publishing. Bellwood, P. 1985. Prehistory of the Indo-Malaysian Archipelago. Honolulu, HI: University of Hawai’i Press.

428 / lum et al. Bellwood, P. 1987. The Polynesians. London, UK: Thames and Hudson. Bender, B.W. 1971. Micronesian languages. In Current Trends in Linguistics, T.A. Sebeok, ed. The Hague, Netherlands: Mouton, 426–465. Blust, R. 1981. Variation in retention rate among Austronesian languages. Paper presented at the Third International Conference on Austronesian Linguistics, Bali, 1981. Blust, R. 1984. Malaita-Micronesia: An eastern Oceanic subgroup? J. Polynesian Soc. 93:99–140. Blust, R. 1995. The position of Formosan languages: Methods and theory in Austronesian comparative linguistics. In Austronesian Studies relating to Taiwan, P. J.-K. Li, ed. Symposium Series of the Institute of History and Philology. Academia Sinica, No. 3, Taipei, 585–650. Cann, R.L., and A.C. Wilson. 1983. Length mutations in human mitochondrial DNA. Genet. 104:699– 711. Capelli, C., J.F. Wilson, M. Richards et al. 2001. A predominantly indigenous paternal heritage for the Austronesian-speaking peoples of insular Southeast Asia and Oceania. Am. J. Hum. Genet. 68:432–443. Clarke, R. 1992. Austronesian Languages. In International Encyclopedia of Linguistics, W. Bright, ed. Vol. 1. New York, NY: Oxford University Press, 142–145. Davidson, J.M., and F. Leach. 1993. The chronology of the Natunuku site, Fiji. New Zealand. J. Archeol. 15:99–105. Delvin, B., K. Roeder, C. Otto et al. 2001. Genome-wide distribution of linkage disequilibrium in the population of Palau and its implications for gene flow in Remote Oceania. Hum. Genet. 108:521–528. Diamond, J.M. 1988. Express train to Polynesia. Nature 336:307–308. Foley, W.A. 1992. New Guinea Languages. In International Encyclopedia of Linguistics, W. Bright, ed. Vol. 3. New York, NY: Oxford University Press, 87–90. Green, R.C. 1976. Lapita sites in the Santa Cruz group. In Southeast Solomon Islands Cultural History: A Preliminary Survey, R.C. Green and M.M. Cresswell, eds. Royal Society of New Zealand Bulletin 11:245–265. Wellington, NZ: Royal Society of New Zealand. Green, R.C. 1991. Near and Remote Oceania: Disestablishing “Melanesia” in cultural history. In Man and a Half: Essays in Pacific Anthropology and Ethnobotany in Honour of Ralph Bulmer, A. Pawley, ed. Auckland, NZ: The Polynesian Society Inc., 491–502. Green, R.C. 1999. Integrating historical linguistics with archaeology: Insights from research in Remote Oceania. Indo-Pacific Prehistory Association Bulletin 18:3–16. Green, R.C., and J.M. Davidson. 1974. Archaeology in Western Samoa. Vol. 2. Bulletin of the Auckland Institute and Museum. Groube, L., J. Chappell, J. Muke et al. 1986. A 40,000 year-old human occupation site at Huon Peninsula, Papua New Guinea. Nature 324:453–455. Hagelberg, E., M. Kayser, M. Nagy et al. 1999. Molecular genetic evidence for the human settlement of the Pacific: Analysis of mitochondrial DNA, Y chromosome and HLA markers. Phil. Trans. R. Soc. Lond. B 354:141–152. Hanihara, T. 1993. Population prehistory of East Asia and the Pacific as viewed from craniofacial morphology: The basic populations in East Asia, VII. Am. J. Phys. Anthropol. 91:173–187. Hedrick, J.D. 1971. Lapita style pottery from Malo Island. J. Polynesian Soc. 80:15–19. Hertzberg, M., K.N.P. Mickleson, S.W. Serjeantson et al. 1989. An Asian-specific 9-bp deletion of mitochondrial DNA is frequently found in Polynesians. Am. J. Hum. Genet. 44:504–510. Intoh, M. 1997. Human Dispersal into Micronesia. Anthropol. Sci. 105:15–28. Jackson, F.H. 1983. The internal and external relationships of the Trukic languages of Micronesia. Ph.D. thesis, University of Hawai’i at Manoa, Honolulu. Jackson, F.H. 1986. On determining the external relationships of the Micronesian languages. In FOCAL II: Papers from the Fourth International Conference on Austronesian Linguistics, P.L. Geraghty and S.A. Wurm, eds. Pacific Linguistics C-94. Australian National University, Canberra, 201–238. Kayser, M., S. Brauer, G. Weiss et al. 2000. Melanesian origin of Polynesian Y chromosomes. Curr. Biol. 10:1237–1246.

Genetic Affinities in the Pacific / 429 Kayser, M., S. Brauer, G. Weiss et al. 2001. Independent histories of human Y chromosomes from Melanesia and Australia. Am. J. Hum. Genet. 2001 68:173–190. Kirch, P.V. 1987. Lapita and Oceanic cultural origins: Excavations in the Mussau Islands, Bismarck Archipelago, 1985. J. Field Archaeol. 14:163–180. Kirch, P.V. 1997. The Lapita Peoples: Ancestors of the Oceanic World. Oxford, UK: Blackwell Publishers. Kirch, P.V. 2000. On the Road of the Winds: An Archaeological History of the Pacific Islands before European Contact. Berkeley, CA: University of California Press. Kirch, P.V., and T.L. Hunt. 1988. Radiocarbon dates from the Mussau Islands and the Lapita colonization of the Southwest Pacific. Radiocarbon. 30:161–169. Lingenfelter, S.G. 1975. Yap: Political Leadership and Culture Change in an Island Society. Honolulu, HI: University of Hawai’i Press. Long, J.C. 1996. Matrix Correlation Analysis, version 1.0. Distributed by the author. Section on Population Genetics and Linkage, Laboratory of Neurogenetics, NIAAA/NIH, Rockville, MD. Lum, J.K. 1998. Central and Eastern Micronesia: Genetics, the overnight voyage, and linguistic divergence. Man and Culture in Oceania 14:69–80. Lum, J.K., and R.L. Cann 1998. MtDNA and language support a common origin of Micronesians and Polynesians in island Southeast Asia. Am. J. Phys. Anthropol. 105:109–119. Lum, J.K., and R.L. Cann. 2000. mtDNA lineage analyses: Origins and migrations of Micronesians and Polynesians. Am. J. Phys. Anthropol. 113:151–168. Lum, J.K., O. Rickards, C. Ching et al. 1994. Polynesian mitochondrial DNAs reveal three deep maternal lineage clusters. Hum. Biol. 66:567–590. Lum, J.K., R.L. Cann, J.J. Martinson et al. 1998. Mitochondrial and nuclear genetic relationships among Pacific island and Asian populations. Am. J. Hum. Genet. 63:613–624. Mantel, N. 1967. The detection of disease clustering and a generalized regression approach. Canc. Res. 27:209–220. Martinson, J.J. 1996. Molecular perspectives on the colonization of the Pacific. In Molecular Biology and Human Diversity, A.J. Boyce and C.G.N. Macie-Taylor, eds. London, UK: Cambridge University Press, 171–195. Melton, T., R. Peterson, A.J. Redd et al. 1995. Polynesian genetic affinities with Southeast Asian populations as identified by mtDNA analysis. Am. J. Hum. Genet. 57:403–414. Merriwether, D.A., J.S. Friedlaender, J. Mediavilla et al. 1999. Mitochondrial DNA variation is an indicator of Austronesian influence in Island Melanesia. Am. J. Phys. Anthropol. 110:243–270. Michalakis, Y., and L.A. Excoffier 1996. Generic estimation of population subdivision using distances between alleles with special reference for microsatellite loci. Genetics 142:1061–1064. Nei, M. 1987. Molecular Evolutionary Genetics. New York, NY: Columbia University Press. O’Shaughnessy, D.F., A.V.S. Hill, D.K. Bowden et al. 1990. Globin genes in Micronesia: Origins and affinities of Pacific Island peoples. Am. J. Hum. Genet. 46:144–155. Oppenheimer, S.J., and M. Richards. 2001. Polynesian origins. Slow boat to Melanesia? Nature 410:166–167. Pawley, A., and R.C. Green. 1973. Dating the dispersal of the Oceanic languages. Oceanic Linguistics 12:1–67. Pawley, A., and R.C. Green. 1984. The Proto-Oceanic language community. J. Pacific History 19:123–146. Pawley, A., and M. Ross. 1993. Austronesian historical linguistic and cultural history. Ann. Rev. Anthropol. 22:425–459. Pietrusewsky, M. 1990a. Craniofacial variation in Micronesia and the Pacific: A multivariate study. Micronesia Suppl. 2:373–402. Pietrusewsky, M. 1990b. Craniofacial variation in Australasian and Pacific populations. Am. J. Phys. Anthropol. 82:319–340. Poulsen, J. 1987. Early Tongan Prehistory. 2 vols. Terra Australis 12. Canberra, Australia: Australian National University. Redd, A.J., N. Takezaki, S.T. Sherry et al. 1995. Evolutionary history of the COII/tRNALys intergenic

430 / lum et al. 9 base pair deletion in human mitochondrial DNAs from the Pacific. Mol. Biol. Evol. 12:604– 615. Richards, M., S. Oppenheimer, and B. Sykes. 1998. MtDNA suggests Polynesian origins in eastern Indonesia. Am. J. Hum. Genet. 63:1234–1236. Rohlf, F.J. 1998. NTSYSpc: Numerical taxonomy and multivariate analysis system version 2.02j. New York, NY: Applied Biostatistics Inc. Ross, M. 1996. Is Yapese Oceanic? In Reconstruction, Classification, Description. Festschrift in Honor of Isidore Dyen, B. Nothofer, ed. Hamburg, Germany: Verlag Meyer & Co., 121–166. Schneider, S., D. Roessli, and L. Excoffier. 2000. Arlequin: A software for population genetics data analysis Ver. 2.000. Genetics and Biometry Laboratory, Department of Anthropology, University of Geneva. Smouse, P.E., and J.C. Long. 1992. Matrix correlation analysis in anthropology and genetics. Yrbk. Phys. Anthropol. 35:187–213. Sokal, R.R., and F.J. Rohlf. 1995. Biometry. New York, NY: W.H. Freeman and Company. Spriggs, M., and A. Anderson. 1993. Late colonization of East Polynesia. Antiq. 67:200–217. Su, B., L. Jin, P. Underhill et al. 2000. Polynesian origins: Insights from the Y chromosome. Proc. Natl. Acad. Sci. USA 97:8225–8228. Sykes, B., A. Leiboff, J. Low-Beer et al. 1995. The origins of the Polynesians: An interpretation from mitochondrial lineage analysis. Am. J. Hum. Genet. 57:1463–1475. Terrell, J.E. 1988. History as a family tree, history as an entangled bank: Constructing images and interpretations of prehistory in the South Pacific. Antiq. 62:642–657. Terrell, J.E., T.L. Hunt, and C. Gosden. 1997. The dimensions of social life in the Pacific. Cur. Anthropol. 38:155–195. Terrell, J.E., and R.L. Welsch. 1997. Lapita and the temporal geography of prehistory. Antiq. 71:548– 572. Terrell, J.E., K.M. Kelly, and P. Rainbird. 2001. Foregone conclusions? An analysis of the concepts of “Austronesians” and “Papuans.” Curr. Anthropol. 42:97–124. Underhill, P.A., G. Passarino, A.A. Lin et al. 2001. Maori origins, Y-chromosome haplotypes and implications for human history in the Pacific. Hum. Mutat. 17:271–280. Weir, B.S., and C.C. Cockerham. 1984. Estimating F-statistics for the analysis of population structure. Evol. 38:1358–1370. Wickler, S., and M. Spriggs. 1988. Pleistocene human occupation of the Solomon Islands, Melanesia. Antiq. 62:703–706.