Stone tools from the ancient Tongan state revealprehistoric interaction centers in the Central PacificGeoffrey R. Clarka,1, Christian Reepmeyera, Nivaleti Melekiolab, Jon Woodheadc, William R. Dickinsond,and Helene Martinsson-Walline

aArchaeology and Natural History, College of Asia and the Pacific, Australian National University, Canberra, ACT 0200, Australia; bLapaha Town Council,Lapaha Village, Tongatapu, Kingdom of Tonga; cSchool of Earth Sciences, The University of Melbourne, Parkville, VIC 3010, Australia; dDepartment ofGeoscience, University of Arizona, Tucson, AZ 85721; and eDepartment of Archaeology and Ancient History, Gotland Campus, Uppsala University, 75105Uppsala, Sweden

Edited by Patrick V. Kirch, University of California, Berkeley, CA, and approved June 10, 2014 (received for review April 2, 2014)

Tonga was unique in the prehistoric Pacific for developing a mari-time state that integrated the archipelago under a centralizedauthority and for undertaking long-distance economic and politicalexchanges in the second millennium A.D. To establish the extentof Tonga’s maritime polity, we geochemically analyzed stone toolsexcavated from the central places of the ruling paramounts, par-ticularly lithic artifacts associated with stone-faced chiefly tombs.The lithic networks of the Tongan state focused on Samoa and Fiji,with one adze sourced to the Society Islands 2,500 km from Ton-gatapu. To test the hypothesis that nonlocal lithics were especiallyvalued by Tongan elites and were an important source of politicalcapital, we analyzed prestate lithics from Tongatapu and stoneartifacts from Samoa. In the Tongan state, 66% of worked stonetools were long-distance imports, indicating that interarchipelagoconnections intensified with the development of the Tongan pol-ity after A.D. 1200. In contrast, stone tools found in Samoa werefrom local sources, including tools associated with a monumentalstructure contemporary with the Tongan state. Network analysisof lithics entering the Tongan state and of the distribution ofSamoan adzes in the Pacific identified a centralized polity andthe products of specialized lithic workshops, respectively. Theseresults indicate that a significant consequence of social complexitywas the establishment of new types of specialized sites in distantgeographic areas. Specialized sites were loci of long-distance in-teraction and formed important centers for the transmission ofinformation, people, and materials in prehistoric Oceania.

Polynesian archaeology | geochemical sourcing | complex societies

Archaeological evidence for prehistoric interaction is criticalto understanding the role of intersocietal contact and the

power strategies used by elites in the formation of complex so-cieties. In the first half of the second millennium A.D., a pow-erful and complex society emerged in the Tonga Islands (Fig. 1)that was unique in the Pacific for the way it aggregated an entirearchipelago under a single political system. Considered a mari-time empire/chiefdom (1–3), Tonga has recently been categorizedas a primary/archaic state that, along with the late-prehistoricpolities of the Hawaiian Islands, were the most complex societiesin prehistoric Oceania (4, ref. 5, p. 146). The ancient Tongan state/chiefdom was headed by the paramount Tui Tonga (Lord ofTonga) and administered by closely related chiefly families, andit was exceptional in Polynesia for a network of political andeconomic relationships that extended to other islands and archi-pelagos (2, 6). The control and redistribution of exotic goods isposited as an important source of capital used to support polit-ical centralization (7, 8), but it has not been feasible to modelprehistoric interaction in the expansive Tongan state using ar-chaeological data because of the paucity of excavations at thecentral places of the chiefdom and the likelihood that mostimports were made in perishable materials that tend not topreserve in tropical contexts (9–11). As a result, it has been unclearhow far Tongan influence extended, whether the political economy

involved control and distribution of prestige exotic goods by elitesand whether the polity’s interaction sphere was only one of severalprehistoric networks responsible for the movement of people,goods, and ideas in the Central Pacific.This article reports the analysis of a significant lithic artifact

assemblage recovered during recent excavations of sites of theTongan polity, which was manifested by the construction ofreligo-political centers containing monumental architecture onthe island of Tongatapu (297 km2), where the political hier-archy was legitimized in ceremonial events, particularly chieflyfunerals and the regular presentation and redistribution of trib-ute from islands within and beyond Tonga (12). Geochemicalanalysis of lithics is used here to determine the spatial extent ofTongan interaction and to test a hypothesis that the acquisitionof nonlocal lithics—and by extension other exotic items that arepoorly represented in the archaeological record—was an im-portant source of power for Tongan elites.

Refining InteractionThe Tongan Islands consist of about 160 uplifted limestone andvolcanic islands with a total land area of 748 km2 (Fig. 1), whichwere first settled about 2,800 y ago by Lapita people (13). Thecore islands are divided into three groups spread over 330 km,comprising the southern Tongatapu Group, the Haapai Group,and the northern Vavau Group, with several small outlying is-lands located to the north (Tafahi–Niuatoputapu, Niuafoou) andsouth (Ata). The prehistoric peak population of Tonga is esti-mated to be around 30,000–40,000 people, about half of whom

Significance

The Tongan state was the only maritime polity in Oceania toencompass an entire archipelago and, through long-distancevoyaging, to extend its influence to other island groupsthrough political and economic exchanges. Stone tools re-covered from the central places of the Tongan state weregeochemically analyzed to provide the first archaeological as-sessment of maritime interaction in the Central Pacific, witha high proportion of tools (66%) identified as long-distanceimports from Fiji, Samoa, and the Society Islands. Exotic lithicswere an important source of political capital used by Tonganelites, and an important consequence of centralization was thedevelopment of interaction centers through which people,products, and information about political organizations reachedmany parts of the prehistoric Pacific.

Author contributions: G.R.C. and C.R. designed research; G.R.C., C.R., N.M., J.W., W.R.D.,and H.M.-W. performed research; G.R.C. and C.R. analyzed data; and G.R.C. and C.R.wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. Email: geoffrey.clark@anu.edu.au.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1406165111/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1406165111 PNAS Early Edition | 1 of 6

ANTH

ROPO

LOGY

EART

H,A

TMOSP

HER

IC,

ANDPL

ANET

ARY

SCIENCE

S

lived on the southern island of Tongatapu (Sacred Tonga),where the central places of the Tongan polity were raised (2, 14,15). Tongatapu is a limestone island and all volcanic rock arti-facts, including adzes, flakes, grindstones, hammer stones, andcooking stones found in archaeological contexts, were importedfrom volcanic islands within, or beyond, the Tongan archipelago.The first paramount center to contain chiefly stone architec-

ture, which signals increasing hierarchical organization, was lo-cated in eastern Tongatapu and built around A.D. 1300 beforebeing abandoned two to three generations later (16). AfterHeketa, the chiefdom relocated to Lapaha on the shores of theFanga Uta Lagoon around A.D. 1350–1400, where the Tonganstate reached its greatest extent. Manifested by a monumentalcentral place covering more than 50 ha, the principal monu-mental structures were stepped royal tombs of the paramountTui Tonga family, which were faced with slabs of beach rock andreef limestone, some weighing more than 20 tons. Lapaha has 27stone-faced burial structures that contain more than 2500 tons ofquarried and transported carbonate stone (Fig. 2). Radiocarbondates, architectonic features, and chiefly genealogies indicate thefirst royal tombs were built A.D. 1300–1400, with the last con-structed ∼A.D. 1760 (16, 17). Additional constructions markingthe chiefly center include ditch systems, roads, earth burialmounds, sitting platforms, bathing wells, standing stones, anda large area of reclaimed land containing a canoe harbor andwharf, which highlight the importance of maritime transport tothe polity (2, 15).Here, we discuss the geochemical analysis of a lithic assem-

blage associated with the central places of the Tongan state,particularly the monumental royal tombs. Lithic sourcing is anestablished method of revealing prehistoric long-distance voy-aging in Oceania (18, 19). As detailed in SI Appendix, chemicalanalysis of lithics is particularly applicable to the Central Pacificbecause of the significant geochemical differences in the com-position of volcanic rocks within the Tongan Island Arc, andbetween Tongan Arc rocks and intraplate volcanic islands, likeSamoa, located north and east of the “Andesite Line” (18). We

analyzed 599 rock samples composed of 567 artifacts and 32reference samples from northern Tonga, Samoa, Uvea (Wallis),and Rotuma. By island/island group, there are 196 lithic artifactsfrom Tongatapu and 371 from Samoa (Upolu, Savaii, Apolima).We used six techniques to identify artifacts to a potential source:thin-section petrology, pXRF, XRF, SEM–EDXA, LA–ICP–MS, and MC–ICP–MS, in addition to using reference sampledata and the extensive literature on the geology and geochemistryof the Central Pacific (SI Appendix). All samples were initiallycharacterized with pXRF (Ti, Rb, Sr, Y, Zr, Nb) to identifygroups and outliers (Dataset S1), followed by additional analysisof selected samples using major and trace elements (Dataset S2),radiogenic isotopes (SI Appendix, Table S1), and thin-sectionpetrography (SI Appendix, Table S2 and Fig. S7).Studies of prehistoric interaction are complicated by un-

certainty about the purpose, directionality, and significance oflong-distance contact (20–22). We refine our understanding ofinteraction in the Central Pacific by, first, constructing a geo-chemical-based network model of the Tongan state and com-paring it with the prestate network on Tongatapu. Second, wecontrast the movement of lithics in the Tongan state with Samoa,which in late prehistory was not centralized and contained sev-eral chiefdoms (23, 24), including lithics from a monumental siteon Savaii (see ref. 25).

ResultsInteraction in the Tongan State. There were 66 analyzed lithics fromLapaha (64) and Heketa (2), with “adzes-flakes” comprising 66%(n = 44) and decorative grave pebbles (kilikili) 34% (n = 22) of theassemblage. Artifact location and context is listed in Dataset S1and SI Appendix, Tables S3 and S4, with site locations shown

Fig. 1. (Upper) Location of Tonga and the main islands in the Central Pa-cific. (Lower) Tongatapu sites with significant lithic assemblages in italics andthe location of Lapaha and Heketa, the central places of the Tongan state.

Fig. 2. Plan view of Lapaha showing the location of the largest tombs as-sociated with the Tui Tonga chiefdom labeled by “J” structure number.Lithics associated with tomb wall excavations are listed in Dataset S1 and SIAppendix, Tables S3 and S4. Stone tomb walls are outlined in blue, and earthmounds are in green.

2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1406165111 Clark et al.

in Figs. 1 and 2. The majority (90%) of samples were directlyassociated with the stone-faced royal tombs, and the remainderwere from chiefly structures, including a ditch system, canoewharf, reclaimed land, and carbonate stone quarries. Geochemicalresults indicate that 44% (29/66) of all Lapaha–Heketa lithics areexotic to the Tonga archipelago, with a clear division betweengrave pebble manuports, which derive entirely from within thearchipelago, and worked adzes-flakes, which are predominantlyfrom outside Tonga (29/44, 66%). Local adzes-flakes and gravepebbles have a trace element geochemistry consistent with anorigin from Eua and central Tongan volcanics (Hunga Haapai,Hunga Tonga, Fonuafoou, Tofua, Kao, Late), but not volcanicsfrom northern Tonga (Fonualei, Tafahi–Niuatoputapu, andNiufoou). An adze quarry on Ata island in the south of the ar-chipelago (26) does not appear to have been an important sourceof adzes in late prehistory. Exotic adzes from Samoa made up55% of the adzes-flakes category (24/44), with trace elements in-dicating a source on Tutuila for eight (samples 36, 51, 53, 54, 241,678, 683, and 707) where significant adze quarries have beenreported (24). The Samoan adze category also includes artifactswith a geochemistry indicating currently undiscovered quarries inshield and posterosional volcanic settings on Upolu and Savaii(samples 31, 41, 42, 210–212, 681, and 688). Three lithics havemajor and trace element values indicating that they do not orig-inate in Tonga nor in Samoa–Uvea–Rotuma, and an east Fijiansource is likely for two samples (198 and 224) and possiblysample 708 (SI Appendix, Fig. S5). Interaction with west Fiji isdemonstrated by worked flakes of plutonic rock found at the J20tomb at Lapaha. A thin-section analysis identified the quartzosematerial (sample 692) as hornblende granodiorite, precludingan origin from the intraplate Polynesian islands, and the ma-terial most likely represents material collected from a shallowsubvolcanic intrusion on Viti Levu or Vanua Levu (SI Appendix,Table S2). Lithics from Fiji/?Fiji contribute 9% of the exotic as-semblage (4/44). Only two lithics were recovered from the earlychiefly center of Heketa, both exotic, with one adze flake fromTutuila (sample 707) and an adze blank (sample 708) potentiallyfrom Fiji. An adze fragment from Lapaha made of phonotephrite(203) was an extreme outlier in the pXRF analysis and was ex-amined further with XRF, LA–ICP–MS, and MC–ICP–MS. Thesample contained TiO2 ∼2.5wt%, indicating an intraplate originwith unusually high Sr (1330 ppm) and Nb (132 ppm) values thatexclude Samoa and point to the high field strength elements-enriched lavas from East Polynesia. Radiogenic Pb, Sr, and Ndisotope ratios measured with MC–ICP–MS (SI Appendix, TableS1 and Fig. S6) relate most closely with the Papenoo adze quarryin Tahiti, Society Islands (18), supported by alkali basalt resultsfrom Vallee de Punaruu (27).

Interaction in Prestate Tongatapu. Stone tools found at the centralplaces of the Tongan state/chiefdom are dominated by importsfrom outside the archipelago, indicating frequent long-distancevoyaging, but was interaction greater during the time of theTongan state than in earlier times? To examine prestate in-teraction, we initially analyzed 115 stone artifacts from ceramic-age sites on Tongatapu dated to B.C. 800–A.D. 200, as there arefew archaeological sites from the first millennium A.D. (28, 29).Seven adzes-flakes were removed from consideration, becausethey were surface finds from sites that had been occupied in thepostceramic era and the lithics were not associated with thesubsurface ceramic deposit, leaving a total of 108 prestate lithics(SI Appendix, Table S6). Adzes-flakes comprise 39% (42/108),and manuports 61% (66/108), of the assemblage. Most lithics(50%) were from the TO.6 site (Tufu Mahina), excavated byJens Poulsen (30), which has an estimated age of B.C. 350–100(31). Of the total lithic collection, 14% (15/108) were nonlocal,with the majority of lithics (86%) sourced to Tongan volcanics.Most prestate imports from outside the Tonga Group were

adzes-flakes with 33% (14/42) exotic, whereas almost all manu-ports (65/66) except for a possible import from Fiji (sample 703)are from local volcanic sources within the archipelago (centralTonga, Eua). Lithics associated with ceramic age deposits haveprobable sources in Samoa (2/42, 5%), Fiji/Rotuma (1/42, 2%),and Fiji/?Fiji (11/42, 26%) (Fig. 3). Lithics assigned to Fiji occurin both basal and upper levels of the TO.6 site (SI Appendix,Table S6), and early interaction between Tongatapu and Fiji isattested by lithics at three other ceramic sites (samples 5, 19, 21,and 694) that are sourced to Fiji (Datasets S1 and S2). In con-trast, only two adzes (samples 22 and 250) from excavated con-texts were sourced to Samoa, and both artifacts were from upperlevels of the primary ceramic deposit. The proportion of exoticadzes-flakes in the prestate assemblage (33%) doubles in theTongan state assemblage (66%) and is accompanied by markedchange in procurement with adzes from Fiji succeeded by im-portation of Samoan lithics.

Interaction in Samoa. Two lithic collections from Samoa were usedto examine whether the Tongan chiefdom acquired more exoticlithics than other stratified societies in the Central Pacific. Thefirst assemblage is essentially a surface collection of 330 adzes/adze fragments from Samoa (Upolu, Savaii, Apolima) (32). Thereis substantial continuity in the Samoan adze sequence, and althoughthe relative age of individual tools cannot be determined pre-cisely, the majority is associated with postceramic sites andprobably date, therefore, to the past 1,500 y (32, 33). To testthe proposition that exotic lithics were especially important tothe Tongan state, we analyzed a collection of 41 stone tools andmanuports from the Pulemelei Mound site. The Pulemelei Moundis a monumental stone mound with a volume of 17,000 m3, sur-rounded by more than 1,000 smaller stone platforms, indicatingit held an important position in the settlement hierarchy andwas coeval with the stone architecture of the Tongan state (34).Both assemblages are museum collections and could only beexamined with nondestructive pXRF. Lithics found on Upolu,Savaii, and Apolima (n = 371) originate overwhelmingly (99.2%)from Samoa. There are three possible imports (samples 144, 174,and 552) from Uvea in the Samoa adze-flake assemblage. Twoare adze-flakes from the Pulemelei Mound (samples 144 and178). However, reference samples and local manuports used inmound construction in the Pulemelei area are indistinguishablefrom reference material from Uvea, indicating, parsimoniously, alocal origin [Uvea (average of samples 633–639)/Pulemelei (averageof samples 663–667, 673); Ti = 22,866/21,943, Mn = 1,580/1,489,Rb = 29/32, Y = 25/26, Zr = 210/201, Nb = 35/43]. Sample 552 isa surface-collected adze from Upolu that has similar trace ele-ment values to a reference sample from Upolu (622) and is alsoprobably local.The proportion of exotic and local lithics in each of the three

assemblages is summarized in Fig. 3, with Tongatapu receivingsubstantial quantities of stone tools from other island groups—even though fine-grained adze rock was available in CentralTonga and on Eua—whereas Samoan lithics derive from localsources within the archipelago.

DiscussionBased on our geochemical analyses, there was extensive move-ment of lithics into the Tongan state, with stone tools sourcedfrom volcanic islands in the Tonga archipelago as well as Fiji,Samoa, and remarkably Tahiti, 2,500 km east of Tongatapu. Thecore area of the Tongan state covers an area of 500,000 squarekm and has a periphery zone up to seven times larger if theSociety Islands are the source of Lapaha sample 203. A largerinteraction zone is supported by traditional history of Tonganarrival on islands such as Tikopia and Anuta and linguistic evi-dence that Tongans traveled to Rotuma, Tokelau, and Tuvalu(35–37). Nonetheless, a comparison of lithic movement in prestate

Clark et al. PNAS Early Edition | 3 of 6

ANTH

ROPO

LOGY

EART

H,A

TMOSP

HER

IC,

ANDPL

ANET

ARY

SCIENCE

S

Tonga demonstrates that a significant proportion of stone tools(33% of adzes-flakes) came from outside the Tonga Islands duringthe first thousand years of human occupation (B.C. 800–A.D.200). Fiji appears to have been an important source of lithics inearly Tongatapu, and this matches linguistic data indicating an

early dialect association between east Fiji and southern Tonga(38–41).The oldest center of the Tongan state is Heketa, where the

first monumental stone architecture was built (16). It is notablethat interaction with Samoa and potentially Fiji is seen duringthe initial period of state formation (∼A.D. 1200–1300), al-though a larger lithic assemblage is needed to examine both earlystate, and immediately prestate state, interaction. Many of theLapaha adzes-flakes found in foundation trenches used to holdthe carbonate facings of the royal tombs were deposited whileshaping the stones and are associated with slab debitage. Thelargest Lapaha tombs have recently been placed in a chronolog-ical sequence using radiocarbon dates and architectonic features(17). The oldest tomb (J17), with an estimated age of A.D. 1350,has lithics from Samoa and central Tonga, whereas later tombsJ03 and J21 (A.D. 1550–1650) have lithics from Samoa (Fig. 2).The adze with a likely source in the Society Islands (sample 203)came from the foundation trench of the four-tiered J09 tombthat is notable for having a large average slab length (3.3 m). Thedating of this tomb is uncertain, but architectonic features in-dicate an age ∼A.D. 1550–1700 (15, 17), a century after long-distance interaction stopped in East Polynesia (42). The J20tomb has lithics from several sources including volcanic islandswithin Tonga (central Tonga, Eua), Samoa (Upolu–Savaii, Tutuila),and Fiji (west Fiji plutonic, east Fiji volcanic). Made of large reef–limestone blocks rather than beach rock slabs, the J20 tomb is thelargest worked stone structure in the Pacific (14) and containsmore than 500 tons of quarried limestone. In Tongan traditions,the J20 tomb is linked to the 29th Tui Tonga (A.D. 1550–1600),who had close connections with chiefly families in Fiji and Samoa(23, 43), suggesting that exotic lithics found in funerary contextsat the J20 tomb derive from the mobilization of valuables in theparamount’s extensive web of family/political connections (seerefs. 44, 45).In addition to chiefly funerals and marriages, political cen-

tralization was manifested in the first fruits (inasi) festival, whichwas a major ceremony attended by chiefs and people from allover Tonga. In the ceremony, the Tui Tonga mediated with thegod Hikuleo to ensure bountiful crop yields, and for this, theparamount was “... greatly reverenced throughout the island, andsupported in splendour and dignity by the contributions of thedifferent districts” (ref. 12, p. 91). An inasi witnessed in the 19thcentury when the influence of the paramount Tui Tonga wasseverely diminished noted tribute goods arriving from through-out the Tonga Islands as well as Uvea and Niuafoou, respec-tively, 870 km and 600 km distance from Tongatapu. Importeditems included yams, pearl shells, megapode eggs, fine mats,Touchardia latifolia fiber for attaching hooks to large trollinglures and making fishing nets, sea bird young, fish from a sacredlake, two kinds of iron wood, and arrowroot (46). The influx ofpeople to the central place at such times numbered in thethousands (ref. 12, p. 93; ref. 47, p. 343) and was accompanied byvast quantities of goods and manufactured products, including,we suspect, stone adzes.Recent work on adze production sites on Tutuila Island in

American Samoa has identified the emergence of nucleatedworkshops at ∼A.D. 1200 that specialized in the production ofbasalt tools (24). Our analysis of 330 Samoan stone tools revealsthat few lithics were imported into Samoa, and this is supportedby a study of lithics from the monumental Pulemelei Mound site.Clearly, the elites of Tonga and Samoa were using stone toolsdifferently, with a high proportion of exotic lithics imported tothe central place of the Tongan state, whereas parts of Samoa,especially Tutuila, focused on the specialized production of high-quality adzes. A Samoan tradition mentions a short-lived Tonganpresence on Tutuila that was repelled, suggesting that Tongansmay have attempted to control adze production (ref. 23, p. 480).We agree with Helms (ref. 48, p. 213), who notes both procurement

Fig. 3. Summary of lithic sourcing results for the Tongan state (Top), pre-state Tongatapu (Middle), and Samoa (Bottom). In the Tongan state,manuports (n = 22) are entirely local compared with adzes-flakes, where themajority (29/44) are nonlocal (66%). In prestate Tongatapu, manuports arealmost entirely local (65/66) compared with adzes-flakes, where a smallerproportion (14/42) are nonlocal (33%), including a number sourced to eastFiji (11/42, 26%). Samoan lithic collections differ from those of Tonga in thelarge number of local adzes-flakes (n = 356, 99.2%), demonstrating over-whelming use of local stone sources. Local production of Samoan lithicscontrasts with the extensive dispersal of Samoan adzes in Oceania (Fig. 4).

4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1406165111 Clark et al.

and production of nonlocal goods provides new forms of politicalcapital to leaders.The different strategies used by leaders in the Central Pacific

can be further visualized by showing the distribution of adzessourced to Samoa/Tutuila and the lithic network of the Tonganstate (Fig. 4). Samoan adzes occur on islands spread over 5,000km of the Pacific Ocean, with tools from workshops on Tutuilawidely esteemed for the high-quality basalt and artisanal skillinvolved in production. Although craft specialization and controlof adze circulation delivered chiefly wealth, it did not, appar-ently, result in the high political status of Tutuila in the Samoanarchipelago (23, 24). In short, the contrasting distribution oflithics in the Tongan state and Samoa–Tutuila correspondsneatly with archaeological expectation. The political center of anexpansive maritime polity in southern Tonga contains a highproportion of imported lithics, whereas the prestige products ofnucleated adze workshops on Tutuila were widely distributedoutside Samoa, including the transport of a large number to theTonga state. We anticipate that the lithic network in Fig. 4 willbe expanded and modified as evidence for prehistoric interactionderived from biological, linguistic, genealogical, and archaeologicalresearch, particularly that focused on specialized sites, is acquired.

ConclusionOnly in Tonga did areal integration under a ruling lineage extendto encompass an entire Pacific archipelago in prehistory. Geo-chemical analysis of lithics from the Tongan state on Tongatapudemonstrates that although local volcanic sources from withinthe archipelago were used in tool production, around two-thirdsof all adzes-flakes came from other island groups, particularlySamoa and Fiji. The close association of exotic lithics with theroyal tombs of the paramount Tui Tonga line illustrates thespatial extent of Tongan leaders’ biosocial networks. Elite net-works based in part on the exchange of high-status marriagepartners in the Central Pacific allowed chiefs to significantlyexpand their political and economic influence. Regular cere-monial events at Lapaha also concentrated staples and nonlocalvaluables that were redistributed as an important source ofcapital for Tongan elites to maintain and fund a centralizedpolitical system dependent on long-distance canoe voyaging. Ananalysis of prestate lithic transfer shows that interarchipelagoconnections were a feature of early Tongatapu that intensifiedwith the development of stratified chiefdoms in the second mil-lennium A.D. The import of lithics to the Tongan state and dis-persal of Samoan adzes provides archaeological signatures ofpolitical centralization and craft specialization, respectively, anddemonstrates that a significant consequence of social stratifica-tion is the formation of new types of specialized sites in distantgeographic areas. Most complex societies in the Pacific onlydeveloped in the past 1000 y, and external drivers such as climatechange have been proposed to explain the widespread emer-gence of hierarchically stratified societies (49). However, thelithic networks in Fig. 4 indicate that stratification was accom-panied by the development of significant interaction centers inthe Central Pacific that had the capacity to transmit informationabout political organization to many parts of Oceania.

Materials and MethodsArtifact location, site name, estimated site age, lithic type, and samplingare given in Dataset S1 and SI Appendix, Tables S3 and S4. “Adze/adzefragments” (n = 395) were identified from cross-section attributes or thepresence of a recognizable adze portion (no adze quarry production deb-itage was analyzed; see SI Appendix, Table S4). “Flakes” (n = 21) wereidentified by the presence of a bulb of percussion, and “adze-flakes” (n =49) had surface polish/hammer dressing, with the circular shape of severalpolished flakes indicating heating of adze/adze fragments. “Grave pebbles”(kilikili) (n = 23) are volcanic water-rolled stones placed on top of burials inTonga. The “manuport” category (n = 78) includes volcanic artifacts broughtto Tongatapu, including hammerstones, abraders, grindstones, gamingstones, and cooking stones. Samoan manuports include locally availablebuilding material and mound paving stones used in the construction ofthe Pulemelei Mound on Savaii Island. Reference volcanic samples (n = 32)from Samoa, Rotuma, Uvea, and northern Tonga are listed in SI Appendix,Table S5.

Sample characterization involved five levels of analysis. First, nonmuseumsamples were examined under low-power magnification and divided intoinformal groups based on grain size, color, and presence and size of inclu-sions, with 29 selected for thin-section analysis. Descriptions and photographsof thin sections are given in SI Appendix, Table S2 and Fig. S7. Second, thecomplete assemblage (n = 599, numbering 567 artifacts and 32 referencesamples) was analyzed with pXRF (Dataset S1). Third, outliers and referenceand representative group samples in the pXRF results were analyzed withXRF (n = 62) and SEM–EDXA (n = 25) (Dataset S2). Fourth, trace elementsfor 87 samples comprising outliers and reference and representativegroup samples were made with LA–ICP–MS (Dataset S2). Fifth, radiogenicPb, Sr, and Nd isotope ratios were obtained for two outliers with MC–ICP–MS (SI Appendix, Table S1). Description of the regional geology, charac-terization techniques, and the sample allocation procedure is detailed inSI Appendix.

ACKNOWLEDGMENTS. We thank the Nobles of Lapaha the HonorableKalaniuvalu–Fotofili and Princess Mele Siu’ilikutapu Kalaniuvalu–Fotofilifor their assistance and Lord Vaea (Chair of the Tongan Traditions Committeeand Minister of Internal Affairs). We are grateful to the Auckland Institute

Fig. 4. Central Pacific interaction A.D. 1300–1750 visualized with the Gephi(0.8.1 beta) open-source network analysis program using the GeoLayoutfunction and Winkel tripel projection to illustrate spatial relations. Thepattern of adzes-flakes entering the Tongan state (red lines) contrasts withthe widespread dispersal of Samoan adzes, especially from quarries onSamoa–Tutuila (blue lines) representing archaeological signatures of a cen-tralized polity and the products of specialized lithic workshops, respectively.Edge weights are numbers of adzes-flakes sourced to a location with Tutuila(n = 8) and Samoa unlocalized (n = 14) combined (node, “Tutuila Samoa”).The distribution of Samoan adzes in the Pacific is based on previous work(20, 24, 38, 49–55).

Clark et al. PNAS Early Edition | 5 of 6

ANTH

ROPO

LOGY

EART

H,A

TMOSP

HER

IC,

ANDPL

ANET

ARY

SCIENCE

S

and Museum, especially Kath Prickett, for access to the Samoa adzes col-lected by the late Roger Green and Janet Davidson and The Museum ofSamoa for permission to analyze lithics from the Pulemelei site. RolandMaas assisted with the radiogenic isotope analyses, and we thank Marshall

Weisler and two anonymous reviewers for comments that helped us improvethe manuscript. Field work was supported by Australian Research CouncilFuture Fellowship Grant FT0990591 (to G.R.C.) and College of Asia andthe Pacific Grant (to C.R.).

1. Aswani S, Graves MW (1998) The Tongan maritime expansion: A case in the evolu-tionary ecology of social complexity. Asian Perspect 37(2):135–164.

2. Kirch PV (1984) The Evolution of Polynesian Chiefdoms (Cambridge Univ Press,Cambridge, UK).

3. Petersen G (2000) Indigenous island empires: Yap and Tonga considered. J Pac Hist35(1):5–27.

4. Kirch PV (2010) How Chiefs Became Kings. Divine Kingship and the Rise of ArchaicStates in Ancient Hawaii (Univ of California Press, Berkeley, CA).

5. Hommon RJ (2013) The Ancient Hawaiian State: Origins of a Political Society (OxfordUniv Press, New York).

6. Neich R (2006) Pacific voyaging after the exploration period. Vaka Moana. Voyages ofthe Ancestors. The Discovery and Settlement of the Pacific, ed Howe KR (DavidBateman Ltd., Auckland, New Zealand), pp 198–245.

7. Blanton RE, Feinman GM, Kowalewski SA, Peregrine PN (1996) A dual-processualtheory for the evolution of Mesoamerican civilization. Curr Anthropol 37(1):1–14.

8. Earle TK (1997) Exchange in Oceania: Search for Evolutionary Explanations. PrehistoricLong-Distance Interaction in Oceania: An interdisciplinary Approach, New ZealandArchaeological Association Monograph, ed Weisler MI (New Zealand ArchaeologicalAssociation, Auckland, New Zealand), Vol 21, pp 224–237.

9. Clark GR, Martinsson Wallin H (2007) Monumental architecture in Western Polynesia:Origins, chiefs and archaeological approaches. Archaeology in Oceania 42(Suppl):28–40.

10. Kirch PV, Green RC (2001) Hawaiki, Ancestral Polynesia. An Essay in Historical An-thropology (Cambridge Univ Press, Cambridge, UK).

11. Kaeppler AL (1978) Exchange patterns in goods and spouses: Fiji, Tonga and Samoa.Mankind 11(3):246–252.

12. Orange J (1998) Narrative of the Late George Vason of Nottingham. One of the Troopof Missionaries First Sent to the South Sea Islands by the London Missionary Society inthe Ship DUFF. Captain Wilson 1796 (Taulua Press, Nuku’alofa, Tongatapu).

13. Burley D, Weisler MI, Zhao J-x (2012) High precision U/Th dating of first Polynesiansettlement. PLoS One 7(11):e48769. doi: 10.1371/journal.pone.0048769.

14. Burley D (2007) Archaeological demography and population growth in the Kingdomof Tonga - 950 BC to the historic era. Long Term Demographic History in the PacificIslands, eds Kirch PV, Rallu J (Univ of Hawaii Press, Honolulu), pp 177–202.

15. Clark GR, Burley D, Murray T (2008) Monumentality in the development of theTongan maritime chiefdom. Antiquity 82(318):994–1008.

16. Clark GR, Reepmeyer C, Stone architecture, monumentality and the rise of the earlyTongan chiefdom. Antiquity, in press.

17. Clark GR Social Memory and Langi (Royal Tombs) of Lapaha, Tonga. Monuments andPeople in the Pacific, Studies in Global Archaeology, eds Martinsson Wallin H,Thomas T (Uppsala Univ, Uppsala, Sweden).

18. Collerson KD, Weisler MI (2007) Stone adze compositions and the extent of ancientPolynesian voyaging and trade. Science 317(5846):1907–1911.

19. Kirch PV, Mills PR, Lundblad SP, Sinton J, Kahn JG (2012) Interpolity exchange ofbasalt tools facilitated via elite control in Hawaiian archaic states. Proc Natl Acad SciUSA 109(4):1056–1061.

20. Clark GR (2002) Adzes of Interaction: Samoan Basalt Artefacts in Fiji. Fifty Years in theField: Essays in Honour and Celebration of Richard Shutler Jr’s Archaeological Career,New Zealand Archaeological Association Monograph, eds Bedford S, Sand C, Burley D(New Zealand Archaeological Association, Auckland, New Zealand), Vol 25, pp227–238.

21. Davidson J (1978) Western Polynesia and Fiji: The archaeological evidence. Mankind11(3):383–390.

22. Renfrew AC (1977) Alternative models for exchange and spatial distribution. Ex-change Systems in Prehistory, eds Earle TK, Ericson JE (Academic, New York), pp71–90.

23. Krämer A (1994) The Samoa Islands. An Outline of a Monograph with ParticularConsideration of German Samoa. Volume I. Constitution, Pedigrees and Traditions(Pasifika Press, Auckland, New Zealand).

24. Winterhoff EH (2007) The political economy of ancient Samoa: Basalt adze pro-duction and linkages to social status. PhD dissertation (University of Oregon, Eugene).

25. Hodder I (1982) Toward a contextual approach to prehistoric exchange. Contexts forPrehistoric Exchange, eds Ericson JE, Earle TK (Academic, New York), pp 1–12.

26. Burley D, Steadman DW, Anderson A (2004) The volcanic outlier of ’Ata in Tonganprehistory: Reconsideration of its role and settlement chronology. Journal of NewZealand Archaeology 25:89–106.

27. White WM, Duncan RA (2000) Geochemistry and geochronology of the Society Is-lands: New evidence for deep mantle recycling. Earth Process: Reading the IsotopicCode, eds Basu A, Hart S, Geophysical monograph (American Geophysical Union,Washington, DC), pp 183.

28. Davidson J (1979) Samoa and Tonga. The Prehistory of Polynesia, ed Jennings JD(Harvard Univ Press, Cambridge, MA), pp 82–109.

29. Sand C, Addison DJ (2008) Archaeology in the Polynesian Homeland: The State of Playat the Beginning of the 21st Century. Recent Advances in the Archaeology of the Fiji/West-Polynesia Region, Studies in Prehistoric Anthropology, eds Addison DJ, Sand C(Univ of Otago, Dunedin, New Zealand).

30. Poulsen J (1987) Early Tongan Prehistory (Department of Prehistory, Research Schoolof Pacific Studies, The Australian National University, Canberra), Vols I and II.

31. Spennemann DHR, Head MJ (1998) Tongan pottery chronology, 14C dates and thehardwater effect. Quat Geochronol 17(11):1047–1056.

32. Green RC, Davidson J (1969) Description and classification of Samoan adzes. Ar-chaeology in Western Samoa. Volume 1, Bulletin of the Auckland Institute and Mu-seum, eds Green RC, Davidson J (The Auckland Institute and Museum, Auckland,New Zealand), pp 21–32.

33. Green RC, Davidson J (1974) Archaeology in Western Samoa (The Auckland Instituteand Museum, Auckland, New Zealand), Vol 2.

34. Martinsson Wallin H, Wallin P, Clark GR (2007) Excavations at the Pulemelei moundsite. Archaeology in Oceania 42(Suppl):41–59.

35. Feinberg R (1998) Oral Traditions of Anuta: A Polynesian Outlier in the Solomon Is-lands (Oxford Univ Press, New York).

36. Geraghty P (1994) Linguistic Evidence for the Tongan Empire. Language Contact andChange in the Austronesian World, Trends in Linguistics, Studies and Monographs,eds Dutton T, Tryon D (Mouton de Gruyter, Berlin, Germany).

37. Schmidt H (2003) Loanword Strata in Rotuman. Language Contacts in Prehistory.Studies in Stratigraphy, Current Issues in Linguistic Theory, ed Andersen H (JohnBenjamins Publishing, Amsterdam), pp 201–240.

38. Best S (1984) Lakeba: The Prehistory of a Fijian Island. PhD thesis (Univ of Auckland,Auckland, New Zealand).

39. Clark GR, Anderson A (2009) Colonisation and Culture Change in the Early Prehistoryof Fiji. The Early Prehistory of Fiji, Terra Australis, eds Clark GR, Anderson A (Aus-tralian National University E Press, Canberra, Australia), pp 407–437.

40. Pawley A (1999) Chasing rainbows: Implications for the rapid dispersal of Austrone-sian languages for subgrouping and reconstruction. Selected Papers from the EighthInternational Conference on Austronesian Linguistics, eds Zeitoun E, Li PJ [Institute ofLinguistics (Preparatory Office), Academia Sinica, Taipei, Taiwan], pp 95–139.

41. Reepmeyer C, Clark GR, Sheppard PJ (2012) Obsidian source use in Tongan prehistory:New results and implications. Journal of Island & Coastal Archaeology 7(2):255–271.

42. WeislerMI (2002) Centrality and the collapse of long-distance voyaging in East Polynesia.Geochemical Evidence for Long-Distance Exchange, ed Glascock MD (Bergin and Garvey,Westport, CT), pp 257–273.

43. Bott E (1982) Tongan Society at the Time of Captain Cook’s Visits: Discussions with HerMajesty Queen Salote Tupou, Memoir 44 (The Polynesian Society, Wellington,New Zealand).

44. Gunson N (1979) The hau leadership in Western Polynesia. J Pac Hist 14(1):28–49.45. Herda P (2007) The political aspects of marriage in traditional Tonga. Tonga and the

Tongans. Heritage and Identity, ed Wood-Ellem E (Tonga Research AssociationBusiness Print Australia Print Group, Melbourne, Australia), pp 179–193.

46. Thomas J (n.d.) The Ranks of Chiefs, Microfilm MLMSS 6228, FM4/1439/48 (ML MSSA1961 Mfm. CY446) [State Library of New South Wales (Mitchell Library), Sydney,Australia].

47. Martin J (1991) Tonga Islands. William Mariner’s Account (Vava’u Press, Tonga).48. Helms MW (1993) Craft and Kingly Ideal. Art, Trade, and Power (Univ of Texas Press,

Austin).49. Nunn PD (2007) Climate, Environment and Society in the Pacific During the Last

Millennium (Elsevier, Amsterdam).50. Kirch PV (1988) Niuatoputapu: The Prehistory of a Polynesian Chiefdom (Burke Mu-

seum, Seattle).51. Best S, Sheppard PJ, Green RC, Parker R (1992) Necromancing the stone: Archae-

ologists and adzes in Samoa. J Polyn Soc 101(1):45–85.52. Allen MS, Johnson KTM (1997) Tracking ancient patterns of interaction: Recent ge-

ochemicl studies on the Southern Cook Islands. Prehistoric Long-Distance Interactionin Oceania: An Interdisciplinary Approach, New Zealand Archaeological AssociationMonograph, ed Weisler MI (New Zealand Archaeological Association, Otago,New Zealand), pp 111–133.

53. Sheppard PJ, Walter R, Parker RJ (1997) Basalt sourcing and the development of CookIsland exchange systems. Prehistoric Long-Distance Interaction in Oceania: An In-terdisciplinary Approach, New Zealand Archaeological Association Monograph, edWeisler MI (New Zealand Archaeological Association, Otago, New Zealand), pp85–110.

54. Di Piazza A, Pearthree E (2001) Voyaging and basalt exchange in the Phoenix and Linearchipelagoes: The viewpoint from three mystery islands. Archaeology in Oceania36(3):146–152.

55. Leach BF, Davidson J (2008) The Archaeology of Taumako: A Polynesian Outlier in theEastern Solomon Islands (New Zealand Journal of Archaeology Special Publication,Auckland, New Zealand).

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1406165111 Clark et al.

Clark et al. PNAS - SI Appendix

Regional Geology and Sample Allocation

Regional geology: The Fiji-West Polynesia region includes the island groups of Fiji, Tonga, Samoa,

Uvea (Wallis), Futuna and Rotuma. The area is conventionally subdivided by the 'Andesite Line' (1)

separating island arc-deriving igneous rocks of Fiji, Rotuma and Tonga from the intra-plate

volcanic islands of Samoa and Uvea further north. The Tongan Arc is subdivided by the Tofua

trough which separates the Tofua chain consisting of a series of north-south aligned basaltic

andesite to dacite islands (2) from low-lying limestone capped fore-arc islands on the Tongan

Platform (3). Like all subduction-related lavas, those of the Tongan Arc are characterized by an

enrichment of large-ion lithophile elements (LILE) in contrast to high field strength elements

(HFSE) and heavy rare earth elements (HREE) (4-6). The geology of the Fijian islands is complex,

involving both subduction related and intraplate components, and a detailed description is beyond

the scope of this paper. Arc-deriving igneous rocks are the oldest Late Eocene formations on the

main island of Viti Levu. Exposed lavas include tholeiitic basalts and boninites similar to Eua (7).

There have been several episodes of uplift that have exposed plutonic intrusions and sedimentary

rocks that are otherwise uncommon in Western Polynesia (8). The islands of the Lau Group are

situated on the Lau Ridge, a remnant arc, separated from the active Tonga-Kermadec Arc system by

the opening of the Lau Basin spreading centre. The volcanic rocks of the Lau Group are subdivided

into three discrete groups: Lau, Korobasaga and Mago (9). The oldest lavas of the Lau group are

basalt to rhyolite flows, the Korobasaga group has a basalt-to-basaltic andesite spread with dacites

exposed on Lakeba and Vanau Balavu. The youngest volcanics are the Mago group consisting of

alkalic basalts with strongly enriched LILE and HFSE contents similar to ocean island basalts

(10).The lavas of Rotuma appear to have a primitive island arc geochemistry (11). They consist of

an older basalt formation, associated with dolerite dykes and olivine gabbros, and a later formation

of basaltic andesites. Both formations have a tholeiitic composition with low potassium values, but

slightly enriched TiO2 and FeO contents indicative of back-arc basin basalts (12). The Samoa

islands are products of the Samoan hotspot (13) with the island of Uvea located on the westernmost

end of the archipelago (14). The islands contain a series of shield volcanoes and post-erosional lava

flows. Shield volcanic deposits consist of tholeiitic to alkalic basalt sequences with a distinctively

enriched TiO2 content and slightly enriched alkali content, even for tholeiites. These older

sequences are overlain by highly undersaturated post-erosional volcanics (15) giving rise to the

complex chemistry of exposed surface rocks of the Samoan islands. Igneous rocks exposed on Uvea

are significantly younger than the islands of Samoa (14). Quaternary volcanism comprises two

sequences of older alkaline basalts and hawaiites, and younger tholeiites. The young age of Uvea

igneous formations forbids association with the Samoan hot spot trail and is more likely associated

with the melting of subducted oceanic crust, although Uvea lavas appear to have a similar isotopic

composition to some Samoan shield lavas (14:283). Figure S1 shows the results of two discriminant

function analyses using geological sample major and trace element values to successfully

distinguish samples from Fiji-Lau islands, Tonga-Kermadec Arc and Samoa islands.

Sample allocation: A five step process was used to provenance lithic artifacts to a source area.

First, a geological dataset of samples (n=193) from Fiji islands-Lau Arc, Tonga-Kermadec Arc and

Samoa islands was examined with discriminant function analysis (DFA) to separate island arc

igneous rocks from ocean island volcanics (Figure S1). Second, a DFA of geological reference data

(n=138) from the Tongan Arc and intra-plate volcanics north and east of the Andesite Line (Samoa,

Uvea, Rotuma) was made to classify archaeological samples using the pXRF results (Figure S2 and

Figure S3). As expected the DFA (Figure S3) showed clear separation between rocks deriving from

island arc and ocean island contexts (see also XRF, SEM-EDXA results in Figure S4). Third,

source allocation of 26 archaeological samples identified as from the Tongan Arc with pXRF were

examined by DFA of 247 reference samples from central Tonga, northern Tonga and Eua and

classifying archaeological samples using 17 trace and rare earth elements obtained with LA-ICP-

MS. Fourth, samples with a composition indicating an origin in the Fiji islands (or archipelagos

west of Fiji), were examined with a DFA (Figure S1) using XRF, SEM-EDXA and LA-ICP-MS

archaeological sample data (Figure S5). Fifth, radiogenic results for two samples (203, 204)

identified with LA-ICP-MS as probable imports from the Society Islands and Samoa respectively

were compared with isotopic results from the potential source areas (Figure S6).

pXRF (Archaeology and Natural History, ANU)

A Bruker Tracer III-V PXRF equipped with a rhodium tube, peltier-cooled Si-PIN detector at a

resolution of approximately 170eV FWHM at the Mn Kα peak (5.9keV at 1000 counts per second)

and a 1024 channel configuration multichannel analyzer was used. Instrument parameters were 40

keV, 15 µA, using a 0.1524 mm Cu, 0.0254 mm Ti and 0.3048 mm Al filter in the x-ray path and a

100 second live-time count at 185 FWHM. Interferences from air were minimized by placing the

instrument as close as possible to the flat surface of a sample. Net values of the samples were

calculated with the Bruker ARTAX Spectra 7.1 package. Nine correction cycles were run for

background stripping and peak deconvolution. Net values were calibrated by linear regression

against 14 international standards (AGV-1, BCR-1, BCR-2, BHVO-1, BHVO-2, BIR, CRPG-BR,

DNC-1, JB-1, NIST1633a, NIST1646, NIST2704, NIST278, NIST27D, RGM-1, WSE) and six in-

house standards (GC-006, GC-11, GC-188, GC-200, KILAUEA 93-1489, TAFAHI) in Microsoft

Excel 2010 (see Figure S2 for calibration algorithms).

Discriminant function analysis on the data set using log10-transformed values and absolute counts

of Ti, Rb, Sr, Y, Zr, Nb was successful in providing initial artifact groups (Figure S3). There is

clear separation between arc deriving material from Tonga and intra-plate hotspots such as the

Samoan hotspot trail (39). Rotuma and Uvea plot discrete from both arc and hotspot deriving lavas

in the 95% confidence interval ellipses, but with a strong overlap in the 50% normal distribution

ellipses. The overlap in the 50% mean ellipses between Samoa, Rotuma and Uvea source materials

indicates that the unambiguous separation of these source areas is, as expected, difficult with pXRF

alone. Samoan samples tend to have higher Ti and Fe, Rotuma samples display lower

concentrations of Ti around 10,000 ppm and have higher Rb/Sr ratios than lavas from Uvea (Data

S2). Samples identified as unlikely to derive from the Samoan islands or the Tonga-Kermadec Arc

were distinguished by Ti values of <10,000 ppm, Rb values of 20-50 ppm and Nb values of <10

ppm. Low Nb values indicate an Arc origin and the parsimonious location is Fiji, although islands

further to the west along the New Hebrides Arc cannot be ruled out. The majority of samples

allocated to the Fiji islands were examined further with XRF, SEM-EDXA and LA-ICP-MS data

(Figure S5).

XRF and SEM-EDXA, LA-ICP-MS, MC-ICP-MS

A Philips (PANalytical) PW2400 X-ray spectrometer was used to determine major elements (Na,

Mg, Al, Si, P, S, K, Ca, Ti, Mn and Fe analysis. Lithium borate discs were prepared by fusion of

0.27 g of dried sample powder and 1.72 g of "12-22" eutectic lithium metaborate-lithium

tetraborate. The major elements were calibrated against 28 international standard rock powders.

Analytical procedures for SEM-EDXA and LA-ICP-MS are given in Ambrose et al. (40), and for

MC-ICP-MS in Woodhead (41) and Hergt & Woodhead (42).

XRF and SEM-EDXA (Research School of Earth Sciences and Centre for Advanced Microscopy,

ANU)

Traditional XRF and SEM-EDXA (n=87) was employed to verify the pXRF dataset and further

characterize artifact and reference material. Figure S4 shows the Total Alkali Silica (TAS)

classification of 68 artifacts found in Tongan archaeological sites with basic basalts (alkalic and

tholeiitic), hawaiites (trachy-basalts), phonotephrites, andesites, dacites and rhyolites represented.

The TAS distribution supports the pXRF classifications with more acidic basaltic andesites,

andesites and dacites characteristic of the Tongan Arc (18) clearly separated from alkalic, more

mafic Samoan hot spot trail volcanics. All basaltic andesites and andesites are likely to originate

from the Tongan Arc. Rhyolites are relatively uncommon in the region with high silica dacites

reported from Fonualei (2), rhyolitic flows on Eua (43) and obsidian deposits from Tafahi in

northern Tonga (1, 44, 45). Sample 690 was classified as rhyolite, but with a distinctive low CaO

concentration and very low Sr and Ba values, ruling out the northern Tongan islands and Fonualei.

Felsic lavas from nearby Eua have similar major element values to 690 indicating a potential origin

on that island (43). All alkalic basalts and hawaiites show distinctively enriched TiO2 and Fe2O3

values common to the Samoan hot spot trail. Due to the complex geology of Samoa it is not

possible to identify individual Samoan islands with major element geochemistry alone. Three

artifacts (samples 688, 698, 701) have unusually high SiO2 values for Samoan rock suits.

Mugearites have been reported from the Samoan islands on Tutuila (27) and Upolu (28), however,

none of the adze quarries on Tutuila have an SiO2 >50wt% (34, 35). High Sr values of close to

1,000 ppm have only been reported from young, post-erosional lava flows on Upolu (15) suggesting

undetected adze quarries (samples 334, 390, 558, 559). An adze fragment found at Lapaha (sample

203), identified as an outlier in the pXRF analysis was classified as phonotephrite with TiO2

~2.5wt% indicating an intra-plate origin. A wide spread of igneous rocks from mafic composition

to phonolites and quartz trachytes is exposed in the Samoan islands (23), but Sr values >1,100 ppm

and Nb >110 ppm have not been reported from Samoa, even in the high-Sr volcanic suits on Upolu.

Shoshonite (sample 694), Hawaiite (sample 695) and Mugearite (sample 696) with low TiO2

content but higher K2O are known from Fiji (see below), but have not been recorded in Tongan Arc

lavas.

LA-ICP-MS (Research School of Earth Sciences)

LA-ICP-MS analysis of 87 samples was made to investigate artifact sources using trace and rare

earth elements (S3) due to the overlap of igneous rock geochemistry in the study area. Only one

adze quarry in the Tongan Arc has been geochemically analyzed (46) and we employ published

data from geological studies to assist identification of artifact provenance. Ewart et al. (5) recorded

important latitudinal geochemical variation along the Tongan Arc. There is a marked changed in

Zr/Ba and Nb/Yb ratios, element abundances of HFSE while element ratios of Zr/Sm decreasing in

the northern island of Niuafoou suggesting increasing magma source depletion and element ratios

of Sc/Y increasing in the north of the Tongan Arc. Trace elements from 247 subaerial geological

samples from islands in the Tonga Arc listed in the online GEOROC database (http://georoc.mpch-

mainz.gwdg.de/georoc/) were analyzed with discriminant function analysis (Rb, Sr, Y, Zr, Nb, Ba,

Ce, Nd, Sm) to determine if northern volcanic islands (Niuafoou, Tafahi, Niuatoputapu, Fonualei)

could be differentiated from central Tongan volcanics (Hunga Haapai-Hunga Tonga, Tofua, Kao,

Late) and volcanics on Eua in the south. As expected, individual islands in the central Tonga group

have an overlapping geochemistry, but there was excellent separation between central Tongan

volcanics (n=163, 99%) and northern islands (Fonualei n=27, 93%, Tafahi n=13, 92%,

Niuatoputapu n=11, 91%, Niuafoou n=21, 100%) and Eua (n=12, 100%). The discriminant function

correctly grouped 97.6% of cases with the first two eigenvalues contributing 93% of between group

variance. We tested the efficacy of the discriminant function by entering three reference samples

collected on Niuafoou (samples 234, 237, 238) as ungrouped cases and all were correctly placed

with Niuafoou geological samples. A further 26 archaeological samples were identified as deriving

from the Tonga Arc based on trace element values and ratios were entered as ungrouped cases with

the majority (n=22) placed in the central Tonga volcanic group (samples 26, 27, 28, 29, 33, 34, 35,

39, 43, 44, 46, 48, 49, 206, 213, 217, 228, 233, 235, 242, 244, 705) and four with Eua (samples 193,

195, 685, 690). Volcanic pebbles used to decorate graves at Lapaha today are often collected on

Niuafoou, but in the past grave pebbles were collected exclusively from volcanics in central Tonga

and no adze-flakes nor manuports indicate the importation of lithics from northern Tonga.

Archaeological literature on basaltic adze sources in Samoa has focused on the small island of

Tutuila (52 square km) where 17 large and small adze quarries have been recorded (34, 35, 47).

Trace elements can distinguish lithics from Samoan island sources as incompatible element-

enrichment increases with decreasing age along the Samoan hot spot trail (33:27). Using the trace

element data published by Collerson and Weisler (34) several adzes-flakes found in Tongan sites

likely derive from Tutuila (samples 36, 51, 53, 54, 208, 241, 243, 678, 683). Slightly lower

incompatible element concentrations (Rb, Th, Ba) and more depleted REEs (La-Lu) indicate

samples 31, 41, 42, 210, 211, 212, 679 and 681 are not from Tutuila and are from the older islands

of Savaii or Upolu (33), with older shield volcanic settings rather than post-erosional volcanics

from Nb/U and Ba/Sm ratios. XRF sourcing of artifact sample 688 with high Sr-values to the post-

erosional lavas on Upolu (15), is supported by its trace element values (Rb, Th, Ba). Sample 204

was identified as Samoan, but it had depleted incompatible element values (Rb, Th, Ba) and very

low REEs, which suggest one of the western islands of the Samoan chain as the source. The low

TiO2 value of sample 204 indicates Uvea as a potential source (14), and the sample was further

analyzed with MC-ICP-MS (see below). Sample 697 has a trace element composition similar to

reference samples from Rotuma (samples 130, 134, 135, 137, 140, 142), although a source in Fiji is

also possible.

Due to the geochemical variability in Fijian volcanic rocks we distinguish between samples of

probable Fijian origin ('Fiji') and those which are less securely sourced labeled as '?Fiji', as they

might originate from the Tongan Arc. An unusual combination of depleted HFSE and enriched

REEs was noted for adze samples 694, 695 and 696, and the element signatures indicate possible

mantle-mixing of post-subduction ocean island basalts with arc-deriving magmas (9). High Ce/Yb

ratios and enriched LILEs have been found in the Mago volcanic group in the Lau Islands (9, 42).

The geochemistry of samples 695 and 696 supports an origin in the Lau Group while sample 694 is

also likely to derive from east Fiji, although a source in the large islands of Fiji cannot be excluded.

Samples are identified as deriving from 'Fiji' (e.g. samples 10, 21, 198, 245, 694, 695, 696) if they

have a composition that rules out an origin in the Tongan Arc (e.g. K2O >2wt%, Rb >30 ppm) and

Rotuma-Uvea-Samoa (e.g. TiO2 <1.0), and samples have a comparable geochemistry to volcanic

rocks in Fiji (9, 16, 21, 32, 42). Three samples may derive from Fiji, but have major and/or trace

element values that are potentially within the range of the Tongan Arc (e.g. K2O <1.5wt%, Ti

<6,000 ppm, Rb <30 ppm); these are labeled as '?Fiji' (samples 3, 20, 708). Figure S5 shows the

results of a discriminant function analysis of geological samples from Fiji, Tonga and Samoa

(Figure S1) with archaeological lithics (S3) entered as ungrouped samples. Results support the

identification of lithic imports from Fiji and allocation of samples to the Tongan Arc and Samoa

islands. Sample 203 has very high Sr (1331 ppm), Rb (113 ppm) and Nb (132 ppm) values

indicating it does not originate in Samoa. Strongly enriched lavas in HFSE have been reported from

the Marquesas and the Society islands (48-50) and the sample was analyzed with MC- ICP-MS.

MC-ICP-MS (School of Earth Sciences, University of Melbourne)

Radiogenic Pb, Sr and Nd isotope ratios were assessed for two samples (203, 204) which plotted as

outliers in previous analyses. Several studies have investigated the isotope signature of adze

quarries and natural outcrops in East Polynesia (34, 47, 51-53). The isotope ratios of sample 203 are

most like the Papenoo adze quarry on Tahiti, Society Islands (34). Adze quarries from the

Marquesas have different 87Sr/86Sr and 208Pb/204Pb ratios, although Legrende et al. (54) reported

several closely related basanites, with only slightly higher 208Pb/204Pb ratios, from Taiohae, Nuku

Hiva, Marquesas. The radiogenic signature together with major element distribution of sample 203

indicates a Tahitian source as the most likely origin. Sample 204 has a radiogenic isotope ratio

indicative of the Samoan hot spot trail (33), supporting the initial allocation of the adze flake to one

of the older, western Samoan islands rather than Uvea. Figure S6 shows radiogenic isotope values

for archaeological samples 203 and 204 plotted against isotope values (87Sr/86Sr x 143Nd/144Nd, 87Sr/86Sr x 207Pb/204Pb, 87Sr/86Sr x 208Pb/204Pb) indicating the lithics are probable imports from the

Society Islands and Samoa respectively.

References for SI Appendix 1. Smith IEM, Ward GK, & Ambrose WR (1977) Geographic distribution and the

characterization of volcanic glasses in Oceania. Archaeology in Oceania 12(3):173-201. 2. Ewart A, Brothers RN, & Mateen A (1977) An outline of the geology and geochemistry,

and the possible petrogenetic evolution of the volcanic rocks of the Tonga-Kermadec-New Zealand Island Arc. Journal of Volcanology and Geothermal Research 2:205-250.

3. Bloomer SH, Ewart A, Hergt JM, & Bryan WB (1994) Geochemistry and origin of igneous rocks from the outer Tonga forearc (Site 841). Proceedings of the Ocean Drilling Program. Scientific Results, eds Hawkins JW, Parson LM, Allan JF, Abrahamsen N, Bednarz U, Blanc G, Bloomer SH, Boe R, Bruns TR, Bryan WB, et al.College Station, TX [Ocean Drilling Program]), Vol 135, pp 625-646.

4. Ewart A, Bryan WB, Chappell BW, & Rudnick RL (1994) Regional geochemistry of the Lau-Tonga arc and back-arc systems. Proceedings of the Ocean Drilling Program. Scientific Results, eds Hawkins JW, Parson LM, Allan JF, Abrahamsen N, Bednarz U, Blanc G, Bloomer SH, Boe R, Bruns TR, Bryan WB, et al.College Station, TX (Ocean Drilling Program)), Vol 135, pp 385-425.

5. Ewart A, Collerson KD, Regelous M, Wendt JI, & Niu Y (1998) Geochemical Evolution within the Tonga-Kermadec-Lau Arc-Back-arc Systems: The Role of Varying Mantle Wedge Composition in Space and Time. Journal of Petrology 39(3):331-368.

6. Ewart A & Hawkesworth CJ (1987) The Pleistocene-Recent Tonga-Kermadec Arc Lavas: Interpretation of New Isotopic and Rare Earth Data in Terms of a Depleted Mantle Source Model. Journal of Petrology 28(3):495-530.

7. Hall R (2002) Cenozoic geological and plate tectonic evolution of SE Asia and the SW Pacific: computer-based reconstructions, model and animations. Journal of Asian Earth Sciences 20:353-431.

8. Neall VE & Trewick SA (2008) The age and origin of the Pacific islands: a geological overview. Philosophical Transactions of the Royal Society 363:3293-3308.

9. Cole JW, Graham IJ, & Gibson IL (1990) Magmatic evolution of Late Cenozoic volcanic rocks of the Lau Ridge, Fiji. Contributions to Mineralogy and Petrology 104:540-554.

10. Gill JB & Whelan P (1989) Early Rifting of an Oceanic Island Arc (Fiji) Produced Shoshonitic to Tholeiitic Basalts. Journal of Geophysical Research 94(B4):4561-4578.

11. Grzesczyk A, et al. (1987) Pétrographie et minéralogie des îles Futuna et Alofi, TOM de Wallis et Futuna (Pacifique Sud-Ouest). Comptes Rendus de l'Académie des Sciences - Series II 305:93-97.

12. Sinton JM, Johnson KTM, & Price RC (1985) Petrology and geochemistry of volcanic rocks from the Northern Melanesian Borderland. Investigation of the Northern Melanesian Borderland, ed Brocher TM (Circum-Pacific Council for Energy and Mineral Resources, Houston, Texas), pp 35-65.

13. Koppers AAP, et al. (2008) Samoa reinstated as a primary hotspot trail. Geology 36:435-438.

14. Price RC, Maillet P, McDougall I, & Dupont J (1991) The geochemistry of basalts from the Wallis Islands, Northern Melansian Borderland: Evidence for a lithosperic origin for Samoan-type basaltic magmas? Journal of Volcanology and Geothermal Research 45:267-288.

15. Wright E & White WM (1987) The origin of Samoa: new evidence from Sr, Nd, and Pb isotopes. Earth and Planetary Science Letters 81:151-162.

16. Cole JW, Gill JB, & Woodhall D (1985) Petrologic history of the Lau Ridge, Fiji. Geology and offshore resources of Pacific island arcs - Tonga region, eds Scholl DW & Vallier TL (Circum-Pacific Council for Energy and Mineral Resources, Houston, US), pp 379-414.

17. Ewart A & Bryan WB (1972) Petrography and Geochemistry of the Igneous Rocks from Eua, Tongan Islands. Geological Society of America Bulletin 83(11):3281-3298.

18. Ewart A, Bryan WB, & Gill JB (1975) Mineralogy and Geochemistry of the Younger Volcanic Islands of Tonga S.W. Pacific. Journal of Petrology 14(3):429-465.

19. Farley KA, Natland JH, & Craig H (1992) Binary mixing of enriched and degassed (primitive?) mantle components (He, Sr, Nd, Pb) in Samoan lavas. Earth and Planetary Science Letters 111:183-199.

20. Gill JB (1976) Composition and age of Lau Basin and Ridge volcanic rocks: Implications for evolution of an interarc basin and remnant arc. Geological Society of America Bulletin 87:1384-1395.

21. Gill JB & Whelan P (1989) Postsubduction Ocean Island Alkali Basalts in Fiji. Journal of Geophysical Research 94(B4):4579-4588.

22. Hawkins JWJ & Falvey DA (1985) Petrology of andesitic dikes and flows from 'Eua, Tonga. Geology and offshore resources of Pacific Island Arcs - Tonga Region, eds Scholl DW & Vallier TL (Circum-Pacific Council for Energy and Mineral Resources, Houston, US), Vol 2, pp 269-279.

23. Hawkins JW & Natland JH (1975) Nephelites and basanites of the Samoan linear volcanic chain: their possible tectonic significance. Earth and Planetary Science Letters 24:427-439.

24. Hubbard NJ (1971) Some chemical features of lavas fro the Manua Islands, Samoa. Pacific Science 25(2):178-187.

25. Jackson MG, Kurz M, Hart SR, & Workman RK (2007) New Samoan lavas from Ofu Island reveal a hemisperically heterogenous high 3He/4He mantle. Earth and Planetary Science Letters 264:360-374.

26. Jackson MG, et al. (2008) Globally elevated Titanium, Tantalum, and Niobium (Titan) in Ocean Island Basalts with 3He/4He. Geochemistry, Geophysics, Geosystems 9.

27. Natland JH (1980) The progression of volcanism in the Samoan linear volcanic chain. American Journal of Science 280-A:709-735.

28. Natland JH & Turner DL (1985) Age progression and petrological development of Samoan shield volcanoes: evidence from K-Ar ages, lava compositions, and mineral studies. Investigation of the Northern Melanesian Borderland, ed Brocher TM (Circum-Pacific Council for Energy and Mineral Resources, Houston), pp 139-171.

29. Palacz ZA & Saunders AD (1986) Coupled trace element and isotope enrichment in the Cook-Austral-Samoa islands, southwest Pacific. Earth and Planetary Science Letters 79:270-280.

30. Turner S, et al. (1997) 238U-230Th disequilibria, magma petrogenesis, and flux rates beneath the depleted Tonga-Kermadec island arc. Geochimica et Cosmochimica Acta 61(22):4855-4884.

31. Wendt JI, Regelous M, Collerson KD, & Ewart A (1997) Evidence for a contribution from two mantle plumes to island-arc lavas from northern Tonga. Geology 25(7):611-614.

32. Wharton MR, Hathway B, & Colley H (1991) Volcanism associated with extension in an Oligocene - Miocene arc, southwestern Viti Levu, Fiji. Geological Society, London, Special Publications:95-114.

33. Workman RK, et al. (2004) Recycled metasomatized lithosphere as the origin of the Enriched Mantle II (EM2) end-member: Evidence from the Samoan Volcanic Chain. Geochemistry, Geophysics, Geosystems 5(4):1-44.

34. Collerson KD & Weisler MI (2007) Stone Adze Compositions and the Extent of Ancient Polynesian Voyaging and Trade. Science 317(5846):1907.

35. Winterhoff EH (2007) The political economy of ancient Samoa: basalt adze production and linkages to social status. PhD Dissertation (University of Oregon, Eugene).

36. Le Bas MJ, Le Maitre RW, Streckeisen A, & Zanettin B (1986) A Chemical Classification of Volcanic Rocks Based on the Total Alkali-Silica Diagram. Journal of Petrology 27(3):745-750.

37. Irvine TN & Barager WRA (1971) A Guide to the Chemical Classification of the Common Vocanic Rocks. Canadian Journal of Earth Sciences 8:523-548.

38. Hémond C, Devey CW, & Chauvel C (1994) Source compositions and melting processes in the Society and Austral plumes (South Pacific Ocean): Element and isotope (Sr, Nd, Pb, Th) geochemistry. Chemical Geology 115:7-45.

39. Gorton MP & Schandl ES (2000) From continents to island arcs: A geochemical index of tectonic setting for arc-related and within-plate felsic to intermediate volcanic rocks. The Canadian Mineralogist 38:1065-1073.

40. Ambrose WR, et al. (2009) Possible obsidian sources for artefacts from Timor: narrowing the options using chemical data. Journal of Archaeological Science 36(3):607-615.

41. Woodhead JD (2002) A simple method for obtaining highly accurate Pb isotope data by MC-ICP-MS. Journal of Analytical Atomic Spectrometry 17:1381-1385.

42. Hergt JM & Woodhead JD (2007) A critical evaluation of recent models for Lau–Tonga arc–backarc basin magmatic evolution. Chemical Geology 245:9-44.

43. Cunningham JK & Anscombe KJ (1985) Geology of 'Eua and other islands, Kingdom of Tonga. Geology and offshore resources of Pacific island arcs - Tonga region, eds Scholl DW & Vallier TL (Circum-Pacific Council for Energy and Mineral Resources, Houston, US), pp 221-257.

44. Reepmeyer C, Clark GR, & Sheppard PJ (in press) Obsidian source use in Tongan prehistory: New results and implications. Journal of Island & Coastal Archaeology.

45. Rogers GA (1974) Archaeological discoveries on Niuatoputapu Island, Tonga. Journal of the Polynesian Society 83:308-348.

46. Burley D, Steadman DW, & Anderson A (2004) The volcanic outlier of 'Ata in Tongan prehistory: Reconsideration of its role and settlement chronology. Journal of New Zealand Archaeology 25:89-106.

47. Weisler MI & Sinton JM (1997) Towards identifying prehistoric interaction in Polynesia. Prehistoric long-distance interaction in Oceania: an interdisciplinary approach, New Zealand Archaeological Association Monograph, ed Weisler MI (New Zealand Archaeological Association, Otago), pp 173-193.

48. Keating BH, Mattey DP, Naughton J, & Helsley CE (1984) Age and origin of Truk Atoll, eastern Caroline Islands: Geochemical, radiometric-age, and paleomagnetic evidence. Geological Society of America Bulletin 95(3):350-356.

49. Legendre C, et al. (2005) Origin of exceptionally Abundant Phonolites on Ua Pou Island (Marquesas, French Polynesia): Partial Melting of Basanites Followed by Crustal Contamination. Journal of Petrology 46(9):1925-1962.

50. White WM & Duncan RA (2000) Geochemistry and Geochronology of the Society Islands: New Evidence for Deep Mantle Recycling. Earth Process: Reading the Isotopic Code, Geophysical monograph, eds Basu A & Hart S (American Geophysical Union, Washington, US), p 183.

51. Weisler MI ed (1997) Prehistoric long-distance interaction in Oceania: an interdisciplinary approach (New Zealand Archaeological Association, Auckland), Vol 21.

52. Weisler MI (1998) Hard evidence for Prehistoric Interaction in Polynesia. Current Anthropology 39(4):521-532.

53. Weisler MI & Woodhead JD (1995) Basalt Pb isotope analysis and the prehistoric settlement of Polynesia. Proceedings of the National Academy of Science USA 92:1881-1885.

54. Legendre C, et al. (2005) The origin of intermediate and evolved lavas in the Marquesas archipelago: an example from Nuku Hiva island (French Polynesia). Journal of Volcanology and Geothermal Research 143:293-317.

Figure S1. Discriminant function analysis (DFA) of geological reference data from Tonga-Kermadec Arc (n=66), Fiji islands and Lau Arc (n=53) and Samoa islands (n=74). (A) DFA of absolute wt(%) values for TiO2 and K2O resulted in the correct classification of 94.1% of samples. Function 1 accounted for 92.0% of between group variance and Function 2 accounted for 8.0% of the variance. Ellipses indicate source areas (blue '+': from Tonga, red '○': from Samoa; brown '∆': from Fiji). Small ellipses are 95% confidence intervals of source mean and large ellipses are 50% Gaussian normal distribution of mean. (B) DFA of absolute wt(%) of TiO2 and absolute ppm counts of Nb and Rb resulted in the correct classification of 93.8% of samples. Function 1 accounted for 92.5% of between group variance and Function 2 accounted for 7.5% of variance. The geochemistry of source areas was taken from the GEOROC database (n=193) compiled from analyses published in (16), (17), (18), (5), (19), (20), (21), (11), (22), (23), (24), (25), (26), (27), (28), (29), (14), (30), (31), (32), (33), in addition to sample data recorded in (34) and (35).

Figure S2. HHpXRF calibration algorithms. Net values (ppm) were calibrated by linear regression against 16 international standards (AGV-1, BCR-1, BCR-2, BHVO-1, BHVO-2, BIR, CRPG-BR, DNC-1, JB-1, NIST1633a, NIST1646, NIST2704, NIST278, NIST27D, RGM-1, WSE) and six in-house standards (GC-006, GC-11, GC-188, GC-200, KILAUEA 93-1489, TAFAHI) in Microsoft Excel™ 2010.

Figure S3. Sample allocation of the pXRF dataset using a discriminant function analysis (DFA) of geological reference data from the Samoan islands (n=74), Uvea (n=19), Tonga-Kermadec Arc (n=39) and Rotuma (n=6). Ellipses are the confidence intervals of the source samples. Small ellipses are 95% confidence interval of the source mean and large ellipses are 50% of the source mean (Gaussian normal distribution of mean). Tonga 'blue' ellipses, Samoa 'red' ellipses, Rotuma 'orange' ellipses, Uvea 'green' ellipses. (A/B) DFA using absolute ppm values of Ti, Rb, Sr, Y, Zr and Nb resulted in the correct classification of 83.4% of reference samples. Function 1 accounted for 93.7.% of between group variance, Function 2 accounted for 4.1% of variance, and Function 3 accounted for 2.2% of variance. Archaeological samples were entered into the DFA as ungrouped cases with group classification indicated by the position of coloured symbols with reference sample ellipses (Tonga, blue '+' and Samoa, red '○'). Reference material from Rotuma and Uvea is marked respectively with orange '□' and green 'x'. Note the clear separation between archaeological lithics deriving from the Tonga Arc and lithics from islands north and east of the Andesite Line (associated with Samoa, Rotuma, Uvea). (C/D) DFA of the same data sets using log10 transformed values of Ti, Rb, Sr, Y, Zr and Nb resulted in the correct classification of 88.1% of reference samples. Function 1 accounted for 98.1% of between group variance, Function 2 accounted for 1.75% of variance, and Function 3 accounted for 0.15% of variance. The position of archaeological lithics is shown by blue '+' for stone tools found in Tonga and red '○' for stone tools found in Samoa.

Figure S4. Total Alkali Silica (TAS, 36) classification (alkaline/tholeiitic boundary, 37) of archaeological samples obtained with XRF and SEM (n=68) showing separation of lithics from island arcs and ocean island volcanics.

Figure S5. Discriminant function analysis of the geological reference dataset (see Figure S1, absolute wt(%) of TiO2 and absolute ppm counts of Nb and Rb) with archaeological samples entered as ungrouped cases (Figure S3). Ellipses indicate geological source areas (Tonga blue '+', Samoa red '○', Fiji brown '∆',); small ellipses are 95% confidence intervals of source mean and large ellipses are 50% confidence intervals of source mean (Gaussian normal distribution of mean). Archaeological samples are marked by infilled circles '�' with lithics identified as imports to Tonga from Fiji indicated by sample number (10, 21, 245, 198, 694, 695, 696, 708). Samples 203 and 204 were analysed with MC-ICP-MS (see Figure S6).

Figure S6. Radiogenic isotope values (Table S1) of an adze fragment (sample 203, blue triangle) excavated from Lapaha tomb J09 and an adze flake excavated from the J20 tomb (sample 204, blue triangle) compared with selected isotope results from the Society Islands (red diamonds) and Samoa (green circles) (15, 33, 34, 38, in addition to the GEOROC database).

Table S1. Radiogenic isotope results for samples 203 and 204. Sample 87Sr/86Sr 143Nd/144Nd 206Pb/204Pb 207Pb/204Pb 208Pb/204Pb 203 0.704324 0.512896 19.062 15.574 38.799 204 0.705291 0.512849 18.848 15.573 38.777

 

Table  S2.  Thin-­‐section  description  of  archaeological  and  reference  samples.  Image  and  analysis  number  identify  samples  in  Dataset  1  and  Dataset  2.  

   

 Analysis  number    

 Sample    

 Location  and  type    

PPL/XPL    

Mag    

Image  number  and  Description    

026  GC1/  TON-­‐7  

Tongatapu,  Lapaha  J20,  TP1,  150cm,  Archaeological  adze  

PPL   40x   3435:  Fine  grained  feldspars,  phenocrysts,  glassy  crystals,  plagioclase  needles.  XPL   40x   3437:  Microcrystals  with  high  birefringence  

XPL   100x  

3443:  Phenocrysts  composed  of  highly  birefringent  (coloured)  ferromagnesium  minerals  (pyroxene?)  and   low   birefringence   (pale)   plagioclase   set   in   mostly   amorphous   groundmass   containing  microcrystalline  needles  of  plagioclase  

XPL   40x  3447:  Phenocrysts  comprise  a  small  proportion  of  the  rock  and  the  microcrystals  have  a  subparallel  alignment  with  a  variable  quantity  of  microcrystalline  material  in  the  amorphous  groundmass  

XPL   40x  3449:  Two  clusters  of  plagioclase  phenocrysts  set  in  subparallel  aligned  microcrystals  of  plagioclase  and  ferromagnesium  minerals  

027   GC2/  TON-­‐2  

Tongatapu,  Lapaha  J02,  TP2,  140cm.  Archaeological  kilikili  stone  

XPL   40x  3395:  Andesite,  orthopyroxene,  phenocrysts  of  plagioclase  and  a  high  birefringence  mineral.  Vesicles  set  in  a  partly  microcrystalline  groundmass  with  no  obvious  orientation  of  microcrystals  

PPL   40x  3401:  Similar  view  in  plane  light  shows  a  circular  vesicle  on  the  right  and  scattered  fine  grains  of  a  black  opaque  mineral  

XPL   200x  3403:   Close   up   of   pyroxene   crystal   with   resorbed   margins;   opaque   mineral   grains   forming   black  areas  above  pyroxene  and  elsewhere  in  groundmass  

XPL   100x  3404:   Cluster   of   plagioclase   showing   clean   interiors   and   faces;   much   of   groundmass   is  microcrystalline  plagioclase  

028   GC3/  TON-­‐3  

Tongatapu,  Lapaha  J02,  TP2,  140cm.  Archaeological  kilikili  stone  

PPL   40x   3406:  Andesite,  vesicles  and  colour  variation  show  lineation.  Matrix  of  plagioclase  needles  XPL   40x   3409:  A  few  small  phenocrysts  of  plagioclase  XPL   200x   3412:  Possible  ferromagnesium  mineral  in  amorphous  groundmass  

029  GC7/  TON-­‐1  

Tongatapu,  Lapaha  J02,  TP2,  62cm.  Archaeological  flake  

PPL   x40  3364:   Andesite,   porphyritic   plagioclase,   clinopyroxene.   Fine   grained   feldspars,   phenocrysts,   glassy  crystals,  plagioclase.  Plagioclase  phenocrysts  in  groundmass  containing  microcrystals  showing  some  alignment  along  long  axes  

XPL   x40   3366:  Microcrystals   composed  of  plagioclase  and  a  high  birefringent  mineral   (pyroxene)  with  dark  voids  

XPL   x100   3389:   Twinned   pyroxene   crystal   near   centre   and   phenocrysts   of   plagioclase   set   in   a   partly  microcrystalline  groundmass  of  plagioclase  needles  and  pyroxene  

XPL   x40   3394:  Plagioclase  phenocrysts  showing  clean  interiors  and  forming  clusters  

031   GC8A   Tongatapu,  Lapaha  J20,  north,  surface.  Archaeological  adze  flake  

PPL   40x  370:   Dark   microcrystalline   rock   with   granular   groundmass   and   microcrystalline   material   showing  variation  in  abundance;  some  orange  coloured  areas  possible  alteration  

XPL   40x   375:  Microcrystals  appear  to  be  plagioclase  needles  with  no  obvious  orientation  

XPL   100x  378:  Fine  opaque  mineral  grains  are  visible  through  specimen;  dark  horizontal  band  forms  boundary  between  microcrystal  poor  and  rich  areas  

XPL   100x  380:   Dark   orange   alteration   band   parallel   to  margin;   area   rich   in  microcrystals   of   plagioclase   and  opaque  mineral  grains  

033   GC8B   Tongatapu,  Lapaha  J20,  north,  surface.  Kilikili  stone  

PPL   40x  383:  Dark  amorphous  rock  with  phenocrysts,  scattered  opaque  mineral  and  a  dark  brown  coloured  irregular  lineation,  possibly  a  fracture  line  with  alteration  

XPL   40x   384:  Phenocrysts  are  pyroxene  and  plagioclase  in  fine  granular  groundmass  

XPL   100x  386:  Many  fine  opaque  mineral  grains  visible;  cluster  of  plagioclase  and  high  birefringence  mineral  phenocrysts;  groundmass  shows  low  and  highly  birefringent  microcrystals  

XPL   40x  389:  ~2mm  cluster  of   interlocking  plagioclase  crystals  with  clean  interiors  and  sharp  margins;  voids  present  

 

Table  S2  continued.  Thin-­‐section  description  of  archaeological  and  reference  samples.  Image  and  analysis  number  identify  samples  in  Dataset  1  and  Dataset  2.  

       

034  GC9/  TON-­‐8    

Tongatapu,  Lapaha  J20,  TP1,  150cm.  Archaeological  adze  flake  

PPL   40x  3463:   Andesite,   porphyritic   plagioclase,   clinopyroxene.   Sharply   defined   phenocrysts,  microcrystals  and  granular  groundmass  

XPL   40x  3465:   Phenocrysts   of   plagioclase   and   pyroxene  with   clean   crystal   faces   and   interiors   set   in   partly  microcrystalline  groundmass  with  a  high  proportion  of  pyroxene  to  plagioclase  crystals  

XPL   100x  3466:   Central   vesicle   and   rounded   plagioclase   crystal   set   in   groundmass   with   microcrystals   of  pyroxene  and  plagioclase  

XPL   200x  3474:  Groundmass  showing  needles  of  plagioclase  and  pyroxene  with  poor  crystal  outline  and  finely  scattered  opaque  mineral  grains  

035   GC10   Tongatapu,  Lapaha  J15,Tier  1,  surface.  Kilikili  stone  

PPL   40x  322:  0.5mm  plagioclase  phenocryst  in  dark  partly  microcrystalline  groundmass  showing  alignment  of  elongate  axes  of  crystalline  material  

XPL   100x  326:  Unusual  2mm  cluster  of  large  plagioclase,  pyroxene  and  opaque  mineral  crystals  with  resorbed  edges  

XPL   200x  328:   Groundmass   showing   subparallel   alignment   of   aligned   needles   of   plagioclase   in   dark  groundmass  

036   GC12   Tongatapu,  Lapaha  J21,  Tier  2,  east.  Archaeological  adze  flake  

PPL   40x  336:   Dark   crystalline   rock   crystals  more   tabular   than   needle   shaped  with   amorphous   groundmass  and  scattered  opaque  mineral  grains    

XPL   40x   340:  Most  crystals  appear  to  be  plagioclase  with  some  pyroxene  

XPL   40x  343:   Pyroxene   crystal   showing   inclusions   and   resorbed   margins;   plagioclase   and   pyroxene  microcrystals  scattered  through  granular  groundmass  

PPL   100x   346:  Pyroxenes  showing  yellow  alteration;  opaque  mineral  grains  and  granular  groundmass  

041   GC14A   Tongatapu,  Lapaha  J03,  TP2,  158cm.  Archaeological  flake  

XPL   40x   303:  Fine  grained  dark  rock  with  microcrystals  of  plagioclase  in  dark  amorphous  groundmass  

PPL   40x  304:   Cluster   of   plagioclase   and   brown   coloured   mineral   set   in   dark   partly   microcrystalline  groundmass;  some  opaque  mineral  grains  

046   GC168C   Tongatapu,  Lapaha  J26,  surface.  Kilikili  stone  

PPL   40x   351:  Rounded  ~2mm  vesicles  and  phenocrysts  in  microcrystalline  groundmass  XPL   40x   353:  Phenocrysts  of  plagioclase  and  pyroxene;  microcrystals  include  ferromagnesium  minerals  XPL   200x   358:  Birefringent  minerals  as  abundant  as  plagioclase  needles;  set  in  dark  amorphous  groundmass  XPL   40x   360:  Large  cluster  of  plagioclase  phenocrysts  surrounding  an  irregular  pyroxene  crystal  

048   GC168D   Tongatapu,  Lapaha  J26,  surface.  Kilikili  stone  PPL   40x  

362:   Dark   microcrystalline   rock   with   granular   groundmass   and   microcrystalline   material   showing  parallel  to  subparallel  alignment  

XPL   40x   364:  Crystalline  material  includes  plagioclase  and  a  coloured  birefringent  minerals  

XPL   100x  367:  Plagioclase  and  smaller  more  equidimensional  birefringent  mineral  grains  are  visible   in  a  dark  amorphous  groundmass  

051   GC173   Tongatapu,  Lapaha  J20,  TP.1,  surface.  Archaeological  adze  flake  

XPL   40x  310:  Dark  microcrystalline  rock  with  glassy  groundmass  showing  subparallel  alignment  of  crystalline  material;  some  brown  coloured  minerals  

XPL   40x  317:  Distinct  variation  in  texture  from  abundant  microcrystalline  plagioclase  and  lighter  'intrusive'  in  area  of  darker  and  less  crystalline  groundmass  

PPL   40x  318:   Alteration   (brown   colouring)   of   rock   between   a   lower   and   higher   proportion   of   crystalline  material  

140   RT-­‐7   Rotuma,  Unlocalised,  surface.  Adze  

PPL   40x   402:  Dark  microcrystalline  rock  with  glassy  groundmass  and  some  alignment  of  crystalline  material  XPL   40x   405:  Crystals  include  plagioclase  and  ferromagnesium  minerals  (equidimensional)  

XPL   100x  409:   Pyroxene   crystal   with   loss   of   material;   surrounding   plagioclase   needles   aligned   within   dark  groundmass  

XPL   100x   415:  Large  pyroxene  with  opaque  mineral  grains  and  resorbed  margins  

 

Table  S2  continued.  Thin-­‐section  description  of  archaeological  and  reference  samples.  Image  and  analysis  number  identify  samples  in  Dataset  1  and  Dataset  2.  

     

195   T10   Tongatapu,  Lapaha  J20,  TP4,  42cm.  Archaeological  adze  flake  

XPL+PPL   40x  190:  Plagioclase  phenocrysts   in  crystalline  groundmass.  Some  opaque  minerals  present  and  brown  discolouration  

XPL   100x   193:  Fine-­‐grained  opaque  grains  through  groundmass,  possible  alteration  

XPL   200x  195:   Large  plagioclase  phenocryst  with  high  birefringent  mineral   and  opaque  mineral   at   centre  of  image  

XPL   200x   198:  Plagioclase  with  pyroxene  phenocrysts  and  amorphous  opaque  rich  groundmass    XPL   40x   201:  Irregular  opaque  minerals  surrounded  by  high  birefringent  minerals  and  some  plagioclase  

196   T101   Tongatapu,  Lapaha  Lakafue,  Area  1,  0-­‐30cm.  Archaeological  adze  flake  

XPL   40x  3517:   Subparallel   alignment   of   microcrystals   set   in   granular   to   amorphous   groundmass   with  scattered  opaque  grains  and  small  areas  of  brown  alteration    

XPL   100x  3525:   Plagioclase   phenocryst   showing   resorbtion   and   brown   mineral   (alteration   around  ferromagnesium  mineral?)  

198   T103   Tongatapu,  Lapaha  north  quarry,  Poulahi,  20cm.  Archaeological  adze  

XPL   40x  3545:  Abundance  of  pyroxene   (more  elongate)  and  olivine  crystals   forming  visible  crystal  grains   in  dark  to  granular  groundmass  

XPL   40x   3551:  Abundance  of  high  brirefringent  minerals  and  lack  of  plagioclase  PPL   40x   3552:  Some  brown  alteration  around  phenocrysts  and  dark  opaque  areas  

XPL   100x  3556:  Rock  appears   completely   crystalline  with  abundant  opaque  mineral  grains   in  an   interlocking  fabric  

XPL   40x   3561:  Well  shaped  (euhedral)  high  birefringent  phenocrysts  some  containing  opaque  mineral  grains  

199   T104  Tongatapu,  Lapaha  J28,  Aponima,  10-­‐15cm.  Archaeological  kilikili  stone  

PPL   40x  3562:   Dark   microcrystalline   rock   with   plagioclase   needles   showing   alignment;   vesicles   present   as  white  oval  bodies  

XPL   40x  3566:   Small   phenocrysts   and   dominant   amorphous   groundmass;   microcrystals   include   high  birefringence  ferromagnesium  mineral(s)  

XPL   100x   3567:  View  of  a  phenocryst  showing  clean  crystal  faces  and  interior  

XPL   100x  3568:   Cluster   of   plagioclase   crystals   with  microcrystals   aligned   around   it   unlike   the  more   general  parallel  alignment  

200   T105  Tongatapu,  Lapaha  J28,  Aponima,  10-­‐15cm.  Archaeological  kilikili  stone  

PPL   40x   2:  Circular  vesicles  with  dark  and  clear  infill,  interlocking  microcrystals,  clear  small  voids  or  vesicles  XPL   100x   5:  Circular  vesicles  with  infilling;  interlocking  needles  of  plagioclase,  opaque  minerals  present  XPL   100x   7:  Alteration  forming  dark  mass  and  brightly  coloured  pyroxene  phenocryst  XPL   200x   11:  Possible  mica  mineral  with  pleochroism  XPL   200x   16:  Plagioclase  needles  show  dark  cores;  significant  birefringence  

201   T106   Tongatapu,  Lapaha  Talasiu  reclaimed  land,  well  feature.  Adze  

PPL   40x  18:   Dark   fine-­‐grained   rock   containing   angular   to   subangular   crystals   of   equidimensional   to  rectangular  form  in  a  dark  amorphous  groundmass  

XPL   40x   20:  Crystals  are  mostly  low  birefringence(plagioclase?)  with  a  few  higher  birefringence  XPL   100x   23:  Possible  lithic  fragment  containing  aligned  needles  of  plagioclase  and  larger  crystal  forms    

203   T11  Tongatapu,  Lapaha  J09,  SW  TP,  20-­‐30cm.  Archaeological  adze  

XPL+PPL   40x   209:  Flow  texture  in  plagioclase  microcrystals  around  pyroxene  phenocryst  

XPL   40x  212:   Flow   texture   in   needle   shaped   plagioclase   microcrystals   around   euhedral   zoned   pyroxene  phenocryst,  some  high  birefringent  microcrystals  

XPL   40x  219:   Flow   texture   in   microcrystals   parallel   to   long   axis   of   thin   section   specimen,   phenocryst   on  margin  

XPL   40x   223:  Hornblende(?)  and  pyroxene  crystals  in  groundmass  with  parallel  alignment  of  microcrystals  

206   T13   Tongatapu,  Lapaha  Ditch  (south),  excavation.  Archaeological  flake  

PPL   40x  228:   Euhedral   well-­‐formed   phenocrysts,   fine-­‐grained   opaque   grains,   amorphous   to   granular  groundmass  

XPL   40x   231:  Plagioclase  and  pyroxene  phenocrysts  XPL   40x   238:  Groundmass  showing  strong  birefringence,  opaques  

 

Table  S2  continued.  Thin-­‐section  description  of  archaeological  and  reference  samples.  Image  and  analysis  number  identify  samples  in  Dataset  1  and  Dataset  2.  

           

208   T14  Tongatapu,  Fafa  Island,  north  beach,  beachrock  quarry  surface.  Adze  flake  

XPL+PPL   40x  

247:   Scatter   of   fine   grained   opaques;   subparallel   alignment   of   larger   needle   to   lath   shaped  plagioclase;   ferromagnesium   minerals   visible   by   birefringence;   fine   granular   to   amorphous  groundmass  

XPL   100x   252:  Opaques  and  plagioclase  laths;  granular  texture  groundmass  

XPL   200x  258:   Plagioclase   laths,   abundant   fine-­‐grained   and   larger   opaque   grains,   ferromagnesium  minerals;  brown  alteration  mineral;  granular  groundmass  

210   T17   Tongatapu,  Lapaha  Loamanu,  Faletuipapai,  surface.  Adze  

XPL+PPL   40x  59:   Fine-­‐grained   rock  with  microcrystals   of   plagioclase   and   opaque  mineral   grains;   distinct   colour  variation  with  large  areas  of  brown  discolouration  suggesting  alteration  

  100x  67:  Opaque  grain  and  altered  ferromagnesium  mineral  set   in  granular  groundmass  with  needles  of  plagioclase  and  areas  of  distinct  colour  difference  

233   T23   Tongatapu,  Lapaha  J17,  TP3,  L1.  Archaeological  kilikili  stone  

XPL   200x  271:   Sharp   pyroxenes   with   clean   interiors   set   in   dark   groundmass   with   plagioclase   microcrystal  needles  

PPL   40x   276:  Dark  fine  grained  rock  with  microcrystals  and  phenocrysts  

XPL   100x  277:   Phenocrysts   are   shown   as   plagioclase;   microcrystal   needles   of   plagioclase   show   subparallel  alignment  along  long  axes  

XPL   100x  283:   Opaque   mineral   grains   in   cluster,   high   birefringent   mineral,   microcrystals   in   surrounding  groundmass  

237   T26   Tongatapu,  Lapaha  J10,  surface.  Kilikili  stone  PPL   40x  

290:  Very  fine-­‐grained  amorphous  rock  with  some  variation  in  texture  and  vesicles  up  to  2mm  and  needle  shaped  plagioclase  microcrystals  

XPL   40x  291:   Microcrystals   have   some   birefringence   suggesting   ferromagnesium   minerals   in   addition   to  plagioclase  

XPL   40x   297:  Larger  plagioclase  crystal  in  very  fine-­‐grained  groundmass  

678  GC6/  TON-­‐11   Tongatapu,  Lapaha  J20,  surface.  Adze  flake  

PPL   40x  3494:  Very   fine-­‐grained  banded  rock  with  alternating  glassy  and  microcrystalline  texture.  Trachytic  texture,  feldspar,  glass,  micro-­‐crystals  of  titanium  oxide,  micro-­‐olivine  

XPL   40x   3496:  Clear  banding  with  microcrystals  of  plagioclase  and  opaque  mineral  grains  PPL   40x   3499:  Brown  staining  (alteration)  

XPL   100x  3501:  Opaque  mineral  grains  and  a  ferromagnesium  mineral  in  a  mix  of  microcrystalline  plagioclase  and  granular  groundmass  with  some  brown  alteration  

680   GC11/  TON-­‐6  

Tongatapu,  Lapaha  J21  TP1,  midlayer.  Archaeological  adze  flake  

PPL   40x   3426:  Very  fine  grained  granular  texture  with  micro-­‐feldspars,  micro-­‐titanium  oxides  XPL   100x   3432:  Crystalline  and  fine  granular  texture  XPL   200x   3433:  Ferromagnesium  minerals  set  in  granular  groundmass  

688   T102   Tongatapu,  Lapaha  Lakufue,  surface.  Adze  flake  

XPL   40x  3526:   Brown   discolouration   on   margins   suggesting   alteration   rind;   subparallel   alignment   of  plagioclase  microcrystals  

XPL   40x  3528:   Random   orientation   of   plagioclase   microcrystal   in   granular/amorphous   groundmass   with  brown  alteration  

XPL   40x   3532:  Opaque  mineral  grains;  birefringent  ferromagnesium  minerals  and  plagioclase  in  glassy  matrix  XPL   100x   3535:  Granular/glassy  groundmass  and  opaque  mineral  grain  

692  T107    (same  rock  as  T15)    

Tongatapu,  Lapaha  J20  Test  pits.  Adze/tool  flakes.  

PPL   40x   32:  Plain  light  view  of  coarse  grained  igneous  rock  containing  large  opaque  mineral  grains  

XPL   40x  

34:   Plutonic   rock,   amphibole,   feldspars,   quartz,   clinopyroxene,   microtitanium   oxides.   Plagioclase  with   a   high   birefringent   mineral   -­‐   hornblende;   replacement   of   original   crystalline   material   with  radiating  needle-­‐shaped  mineral  aggregates    

 

Table  S2  continued.  Thin-­‐section  description  of  archaeological  and  reference  samples.  Image  and  analysis  number  identify  samples  in  Dataset  1  and  Dataset  2.  

 

692  T15  (same  rock  as  T107)  

Thin  section  description  of  plutonic  rock  (692)  by  W.R.Dickinson  

Derivation  of  the  plutonic  rock  from  Eua  can  be  ruled  out  on  the  following  grounds:  (a)  no  relicts  of  pyroxene  can  be  identified  within  it;  (b)  although  some  hornblende  in  the  manuport  has  a  coarsely  fibrous  character,  abundant  blocky  to  prismatic  green  hornblende  undescribed  from  Eua  gabbros  is  also  abundant;  (c)  ferromagnesian  minerals  form  ~40%  of  typical  Eua  gabbros,  but  only  ~20%  of  the  manuport,  which  is  not  gabbro;  (d)  blocky  crystals  of  partly  perthitic  alkali  feldspar  with  a  refractive  index  lower  than  plagioclase  feldspar  are  present  sparingly  among  the  plagioclase  laths  of  the  manuport  but  does  not  occur  in  gabbros;  (e)  equant  quartz  crystals  are  common  among  the  plagioclase  feldspar  laths  in  the  manuport  but  occur  in  only  trace  amounts  (<1%)  in  Eua  gabbros.The  manuport  can  best  be  termed  hornblende  granodiorite,  for  which  an  origin  from  Fiji  is  most  likely.  Strong  oscillatory  zoning  in  many  of  the  plagioclase  crystals  suggests  derivation  from  a  comparatively  shallow  subvolcanic  intrusion  of  the  type  abundant  on  Viti  Levu  and  Vanua  Levu.  An  origin  from  elsewhere  in  Melanesia  cannot  be  excluded  on  petrographic  grounds,  but  origins  in  Polynesia  are  precluded  by  the  quartzose  character  of  the  rock.    

na   TON-­‐4   Tongatapu,  Lapaha  J21,  TP1,  65  cm.  Adze  flake  

PPL   40x  3416:  Fine-­‐grained,  microcrystals,  clinopyroxene,  plagioclase.  Microcrystalline  texture  with  granular  groundmass  showing  colour  textural  banding  and  scattered  fine  angular  opaque  mineral  grains    

XPL   40x  3418:   Opaque   grains   show   clearly   as   do   the   predominantly   plagioclase   microcrystals   without  significant  alignment  of  long  axes  

XPL   200x   3419:  Granular  texture  with  needles  of  plagioclase  and  less  common  ferromagnesium  minerals  

PPL   100x  3421:  Granular  texture  is  mottled  (0.3-­‐0.4mm  scale)  with  some  clusters  of  yellow-­‐coloured  minerals  (possibly  altered)  

XPL   200x   3425:  Ferromagnesium  mineral  and  opaque  mineral  grain  in  granular  groundmass  

na   TON-­‐9   Tongatapu,  Lapaha  J20,  TP1,  150cm.  Adze  flake  XPL   40x  

3475:   Microplagioclase,   feldspar,   small   pyroxene,   many   glass   crystals   in   groundmass   Dark  microcrystalline  rock  with  granular  groundmass;  microcrystals  show  some  alignment  of  long  axes.    

XPL   100x  3478:   Needles   of   plagioclase   have   greater   alignment   with   accompanying   brown   tinted   mineral;  opacity  suggests  high  quantities  of  fine  opaque  minerals.  Elsewhere  the  groundmass  is  granular    

XPL   100x   3479:  Brown-­‐tinted  minerals  do  not  have  significant  birefringence  

na   TON-­‐10  Tongatapu,  Lapaha  J20,  TP.1,  1.50cm.  Adze  flake  

PPL   40x  3481:   Groundmass   is   very   fine   grained   with   clusters   and   phenocrysts   of   clinopyroxene   and  orthopyroxene  

XPL   40x   3485:  Phenocrysts  include  plagioclase  and  voids  and  with  a  brown  coloured  area  of  alteration    XPL   100x   3487:  Pyroxene  crystals  among  the  phenocrysts;  opaque  grains  visible  in  groundmass  

Figure S7. Thin-section images (polarized and plain light) of archaeological and reference samples. Image and analysis number identify samples in Table S2 and Dataset 1 and Dataset 2.

   

Figure S7 continued. Thin-section images (polarized and plain light) of archaeological and reference samples. Image and analysis number identify samples in Table S2 and Dataset 1 and Dataset 2.  

         

Figure S7 continued. Thin-section images (polarized and plain light) of archaeological and reference samples. Image and analysis number identify samples in Table S2 and Dataset 1 and Dataset 2.  

         

Figure S7 continued. Thin-section images (polarized and plain light) of archaeological and reference samples. Image and analysis number identify samples in Table S2 and Dataset 1 and Dataset 2.  

 

Table S3. Site, age and association of analysed lithic artifacts. Lithics from sites marked with an asterisk are not mentioned in the text as their archaeological context is uncertain. All archaeological lithics found in excavations at Lapaha were analyzed except for one collection from the J20 tomb which produced multiple flakes (>35). The J20 TP.4 flake assemblage was sub-sampled because of the possibility that multiple flakes from a single exotic/local adze could be analyzed and distort the sourcing results. The TP.4 assemblage was divided into four groups based on flake color and texture (light microscopy). Each group was sub-sampled based on flake frequency and size as some flakes were too small for pXRF analysis with 16 flakes selected for geochemical analysis. Results (pXRF, LA-ICP-MS, thin section petrography) identified multiple stone artifact sources (Central Tonga, n=5; Eua, n=2; Samoa unlocalized, n=4; Upolu-Savaii, n=2; west Fiji, n=1 (coarse-grained hornblende granodiorite); ?east Fiji, n=1 (fine-grained volcanic). The diversity of stone sources in the TP.4 flake assemblage indicates that a variety of different stone tools were used in tomb construction consistent with the observed variation in the lithic assemblage. Site Analysed

Artifacts Site Age Association Investigator/

Collector Tongatapu, Vuki's Mound

17 early, B.C. 800-A.D. 200

ceramics in shell midden

L. Groube

Tongatapu, TO.1 12 early, B.C. 800-A.D. 200

ceramics in shell midden

J.Poulsen

Tongatapu, TO.2 18 early, B.C. 800-A.D. 200

ceramics in shell midden capped by recent burial mound, surface material

J.Poulsen, G. Clark

Tongatapu, TO.5 2 early, B.C. 800-A.D. 200

ceramics in shell midden

J.Poulsen

Tongatapu, TO.6 56 early, B.C. 800-A.D. 200

ceramics in shell midden

J.Poulsen

Tongatapu, TO.7 3 early, B.C. 800-A.D. 200

surface ceramics J.Poulsen

Tongatapu, Talasiu 7 early, B.C. 800-A.D. 200

ceramics in shell midden

G. Clark

Tongatapu, Lapaha 64 late, A.D. 1300-1750 stone architecture G. Clark Tongatapu, Heketa 2 late, A.D. 1300-1750 stone architecture G. Clark *Tongatapu, Hihifo 1 unknown na J.Poulsen *Tongatapu, Moungatapu Island

1 late, A.D. 1300-1750 ?stone architecture G. Clark

*Tongatapu, Monuafe Island

2 late, A.D. 1300-1750 ?stone architecture J.Poulsen

*Tongatapu, Fafa Island 4 late, A.D. 1300-1750 ?stone quarry G. Clark *Tongatapu, Pea/Tokomololo

4 unknown na J.Poulsen

*Tongatapu, unlocalized 3 unknown na J.Poulsen

Table S4. Site and artifact type. 'Adze/Adze fragment' includes complete or partial adzes recognized from the mid-section or identified by portion (blade, butt) if no mid-section was present. The use life of an adze can involve the following stages; blank, preform, primary, refurbished, recycled and broken/discarded. Identifying the life stage of an adze fragment is difficult, particularly in the case of Samoan adzes which typically have numerous flake scars that are progressively removed during the life of the adze by grinding. Similarly, refurbishment and recycling of adzes can lead to the creation of a new adze blank or preform while refurbishment and use of adzes typically creates adze flakes (identified by surface polish, friction gloss, grinding) as well as flakes that lack diagnostic features of adze derivation. This is probable at Lapaha where adze flakes and flakes made in the same material occur together in association with tomb slabs and likely result from adze breakage while trimming slabs as stone tool fragments (polished and unpolished) are found with carbonate stone debitage. The 'adzes-flakes' category refers to stone artifacts that we believe represent adzes, and we did not analyse adze quarry production debitage. The Lapaha 'Adze flake' assemblage includes a hornblende granodiorite adze flake (sample 692). Lithics from sites marked with an asterisk are not mentioned in the text as their archaeological context is uncertain. Note that seven samples from the TO.6 site in the 'Adze/Adze fragment' category were surface finds that were not included in the pre-state lithic sample (Table S6) because they likely derive from post-ceramic site use. Site Number Adze/Adze

Fragment Adze blank Adze flake Flake Manuport/Grave

Pebble Manuport/

various Tongatapu, Vuki's Mound

17 9 - - - - 8

Tongatapu, TO.1 12 2 - - - - 10 Tongatapu, TO.2 18 7 - 1 2 - 8 Tongatapu, TO.5 2 1 - - - - 1 Tongatapu, TO.6 56 19 - 2 - 35 Tongatapu, TO.7 3 2 - - - - 1 Tongatapu, Talasiu

7 1 - 3 3

Tongatapu, Lapaha

64 6 - 26 10 22 -

Tongatapu, Heketa

2 - 1 1 - - -

*Tongatapu, Hihifo

1 1 - - - - -

*Tongatapu, Moungatapu Island

1 - - 1 - - -

*Tongatapu, Monuafe Island

2 2 - - - - -

*Tongatapu, Fafa Island

4 - - 3 1 - -

*Tongatapu, Pea/ Tokomololo

4 4 - - - - -

*Tongatapu, unlocalized

3 3 - - - - -

Samoa, Pulemelei

41 8 - 14 6 - 13

Samoa, Upolu 149 149 - - - - - Samoa, Savaii 165 165 - - - - - Samoa, Apolima 16 16 - - - - - Total 567 395 1 49 21 22 79

Table S5. Reference volcanic samples (n=32) from Samoa (n=16), Uvea (n=7), Rotuma (n=6) and northern Tonga (Niuafoou, n=3). Analysis Number Analysis Code Island Group and Location GPS coordinates

669 Savaii_Letui_5 Samoa, Savaii, Letui S 13° 28' 33.43", W 172° 27' 40.35"

670 Savaii_Matava Samoa Savaii, Matava

S 13° 31' 38.10", W 172° 23' 51.46"

671 Savaii_mod.quarry_2 Samoa Savaii, modern quarry

S 13° 33' 12.61", W 172° 14' 16.63"

672 Savaii_Papalaule_1 Samoa Savaii, Papalaule

S 13° 34' 42.40", W 172° 13' 20.66"

673 Savaii_Pulemelei_Stream Samoa Savaii, Pulemelei stream pebble

S 13° 44' 8.5878", W 172° 19' 24.729"

668 Savaii_LavaField Samoa Savaii, recent lava field

S 13° 27' 44.15", W 172° 19' 08.00"

674 Savaii_Roadside_3 Samoa Savaii, outcrop

S 13° 31' 09.13", W 172° 15' 47.09"

675 Savaii_Sili Samoa Savaii, Sili

S 13° 45' 21.57", W 172° 02' 39.21"

676 Savaii_Viaata_Road_7 Samoa Savaii, Viaata

S 13° 37' 18.33", W 172° 15' 43.47"

621 UPO_Aleisa_01 Samoa Upolu, Aleisa

S 13° 51' 43.0554", W 171° 51' 26.568"

622 UPO_Aleisa_02 Samoa Upolu, Aleisa

S 13° 51' 43.0554", W 171° 51' 26.568"

623 UPO_Malaefao_01 Samoa Upolu, Malaefono Plantation

S 13° 49' 15.168", W 171° 53' 46.788"

624 UPO_Malaefao_02 Samoa Upolu, Malaefono Plantation

S 13° 49' 15.168", W 171° 53' 46.788"

625 UPO_MF_AG area Samoa, Upolu, Mulifanua area S 13° 49' 37.5162", W 172° 01' 33.8556"

626 UPO_Vaiele_01 Samoa, Upolu, Vaiele, Fagali'i Stream S 13° 51' 4.176", W 171° 44' 6.828"

627 UPO_Vaiele_02 Samoa, Upolu, Vaiele, Fagali'i Stream S 13° 51' 4.176", W 171° 44' 6.828"

633 ALE_Lutunupule Clark Uvea, Latunipule, surface ?S 13° 14' 49.79", ?W 176° 10' 43.25"

634 NUK_SW Clark Uvea, Nukufetau, surface S 13° 21' 28.18", W 176° 11' 35.69"

635 UTU_Lapita Clark Uvea, Utuleve, surface S 13° 18' 05.08", W 176° 14' 52.37"

636 UTU_LomipQuarry Clark Uvea, Utuloko, surface S 13° 18' 22.58", W 176° 10' 52.84"

637 NUK_NBeach Clark Uvea, Nukufetau, surface S 13° 21' 28.18", W 176° 11' 35.69"

638 NUK_top Clark Uvea, Nukufetau, surface S 13° 21' 28.18", W 176° 11' 35.69"

639 TAL-Kolonui Clark Uvea, Talietemu/Koloniu, surface S 13° 20' 12.11", W 176° 12' 31.56"

130 RT-2 68.106 Parke Rotuma, Ututu Village S 12° 30' 50.05", W 177° 08' 13.89"

134 RT-4 Parke Rotuma, Kalvaka Village S 12° 31' 16.94", W 177° 07' 16.43"

135 RT-5 68.107 Parke Rotuma, Malhaha District S 12° 28' 58.32", W 177° 04' 32.20"

137 RT-6 68.106 Parke Rotuma, Kalvaka Village S 12° 31' 16.94", W 177° 07' 16.43"

140 RT-7 Parke Rotuma, Unlocalised na 142 RT-8 Parke Rotuma, Unlocalised na

234 T24 Mateni Niufoou, pebble (kilikili) collected on Niufoou and taken to Lapaha

S 15° 35' 33.86", W 175° 40' 30.07"

237 T26 Mateni Niufoou, pebble (kilikili) collected on Niufoou and taken to Lapaha

S 15° 35' 33.86", W 175° 40' 30.07"

238 T27 Mateni Niufoou, pebble (kilikili) collected on Niufoou and taken to Lapaha

S 15° 35' 33.86", W 175° 40' 30.07"

Table S6. Distribution of adzes excavated from site TO.6 (Tufu Mahina) by Jens Poulsen in basal (Horizons IB and IT) and upper sediments (Horizons II and III). Lithics sourced to Samoa derive from surface and upper levels (Horizon III) of the site suggesting that increased importation of Samoan adzes dates to the post-ceramic era. As a result seven surface adzes-flakes collected from the vicinity of early sites which had also been occupied in the post-ceramic period were removed from the pre-state assemblage. These lithics (samples 58, 192, 215, 249, 631, 698, 701) were sourced to Samoa, but likely date to the period of the Tonga state. ID Sample Horizon Layer Sourcing 14 E16 TO6-25 Poulsen Horizon IB Layer 5, bottom midden spit Tonga

105 OT-57 TO6-215 Poulsen Horizon IB Layer 8 Tonga

15 E19 TO6-33 Poulsen Horizon IB Layer 8, bottom spit Tonga

695 TO6-20wr Poulsen Horizon IB Layer 8, bottom spit MVG-Lau

18 E23 TO6-30 Poulsen Horizon IB/IT Layer 5-6, bottom midden spit Tonga

19 E4 TO6-29 Poulsen Horizon IB/IT Layer 8, bottom spit Fiji

20 E6 TO6-102 Poulsen Horizon IT Layer 7 ?Fiji

696 TO6-170wr Poulsen Horizon IT Layer 6 MVG-Lau

697 TO6-1514 Poulsen Horizon IT Layer 6 Fiji/Rotuma

16 E22 TO6-171 Poulsen Horizon IT/II na Tonga

67 OT-19 TO6-2193 Poulsen Horizon II Layer 4 Tonga

12 E13 TO6-108 Poulsen Horizon II Layer 6 Tonga

9 E10 TO6-167 Poulsen Horizon II Layer 7 Fiji

13 E14 TO6-27 Poulsen Horizon II/III Layer 3, Posthole BH Tonga

10 E11 TO6-158 Poulsen Horizon III Pit AN Fiji

250 TO6-109 Poulsen Horizon III Layer 3 Tutuila

73 OT-25 TO6-3375 X-6 Poulsen ?Horizon Collected during infilling Tonga

21 E9 TO6-134 Poulsen ?Horizon Posthole Y Fiji

56 No.2 Poulsen Surface Surface collection Tonga

58 No.5 Poulsen Surface Surface collection Samoa

701 No.4 Poulsen Surface Surface collection Tutuila