Late Holocene Human Expansion into Near and …...Human Expansion into Oceania: A Bayesian Model 14Cyears;pooledvalue)andmarineinverte- brate taxa that inhabit reef or open marine

Full Terms & Conditions of access and use can be found athttp://www.tandfonline.com/action/journalInformation?journalCode=uica20

Download by: [66.66.217.214] Date: 07 June 2017, At: 09:52

The Journal of Island and Coastal Archaeology

ISSN: 1556-4894 (Print) 1556-1828 (Online) Journal homepage: http://www.tandfonline.com/loi/uica20

Late Holocene Human Expansion into Nearand Remote Oceania: A Bayesian Model of theChronologies of the Mariana Islands and BismarckArchipelago

Timothy M. Rieth & J. Stephen Athens

To cite this article: Timothy M. Rieth & J. Stephen Athens (2017): Late Holocene HumanExpansion into Near and Remote Oceania: A Bayesian Model of the Chronologies of theMariana Islands and Bismarck Archipelago, The Journal of Island and Coastal Archaeology, DOI:10.1080/15564894.2017.1331939

To link to this article: http://dx.doi.org/10.1080/15564894.2017.1331939

View supplementary material

Published online: 07 Jun 2017.

Submit your article to this journal

View related articles

View Crossmark data

The Journal of Island and Coastal Archaeology, 0:1–12, 2017Copyright C© Taylor & Francis Group, LLCISSN: 1556-4894 print / 1556-1828 onlineDOI: 10.1080/15564894.2017.1331939

Late Holocene Human Expansion intoNear and Remote Oceania: A BayesianModel of the Chronologies of the MarianaIslands and Bismarck Archipelago

Timothy M. Rieth and J. Stephen AthensInternational Archaeological Research Institute, Inc., Honolulu, Hawaii, USA

ABSTRACT

Since the investigations of Spoehr in the 1950s, most researchers haveaccepted a date of ∼3500 BP/1500 BC for the initial human settle-ment of the Mariana Islands in the western Pacific. The relationshipof this early expansion beyond Island Southeast Asia, characterizedby Lapita-like pottery, to the appearance of Lapita in Near Oceania,generally was either given little thought or largely ignored. TheLapita settlement of Near Oceania is almost universally regardedas the initial expansion of Austronesian speakers into the Pacific,followed a few centuries later by a rapid migration to the east intoRemote Oceania. More recently, however, radiocarbon evidence fromseveral sites suggests that initial late Holocene expansion into thePacific occurred in the Mariana Islands. This hypothesis needs criticalevaluation. To this end, we created site- and region-level Bayesiancalibration models. Results estimate that initial Lapita occupation ofthe Mussau Islands in the Bismarck Archipelago occurred between3535 and 3234 cal BP (95% probability), which is 50–385 years(95% probability) earlier than the initial settlement of the MarianaIslands, dated to 3230–3085 cal BP (95% probability). Additionally,settlement of the Mariana Islands was either coeval or later (−66 to254 years [95% probability]) than Lapita expansion out of Mussauinto the greater Bismarck archipelago between 3397 and 3115 calBP (95% probability). Radiocarbon datasets from these regions arehampered by problematic samples, and we anticipate that additionalreliable radiocarbon dates will refine these estimates.

Keywords Bayesian calibration, Bismarck Archipelago, chronology, Lapita, MarianaIslands

Received 10 January 2017; accepted 15 May 2017.Address correspondence to Timothy M. Rieth, International Archaeological Research Institute, Inc.,2081 Young St., Honolulu, HI 96826, USA. E-mail: [email protected] versions of one or more figures in this article are available online at www.tandfonline.com/uica.

1

Timothy M. Rieth and J. Stephen Athens

INTRODUCTION

Settlement of the Mariana Islands ∼3500 calBP has been posited as the first expansion ofAustronesian speakers beyond Island South-east Asia (ISEA) during the late Holocene(Carson 2014; Carson and Kurashina 2012;Hung et al. 2011). The Pre-Latte Period, asthis earliest archaeological manifestationis called in the Mariana Islands, is recog-nized by its distinctive pottery tradition ofred-slipped, impressed, lime-infilled wares(Spoehr 1957:117–124), a tradition gener-ally associated with the well-known Lapitaware of Melanesia (Bellwood 1975:13; Car-son et al. 2013). Spoehr’s (1957:66) originaldate on a large oyster shell from the ChalanPiao site on Saipan for the Pre-Latte Periodwas 1527 ± 200 BC. Marine shell was alsoused to date the Pre-Latte component of theTarague site on Guam, producing a roughlysimilar determination of 3435 ± 70 BP(Kurashina and Clayshulte 1983). Whilethese dates have been long-recognized asproblematic due to the metabolism of oceanreservoir 14C-depleted carbon in variablequantities (Athens 1986:116; also Petchey2009), two subsequent determinationson unidentified charcoal from reworkedshoreline sediments at the Achugao site onSaipan (Butler 1994:22–23) seem to haveenshrined the 3500 BP/1500 BC age asthe approximate date for the onset of thePre-Latte Period and the settlement of theMariana Islands (e.g., Butler 1994; Russell1998). Recent investigations carried outby Carson and colleagues, noted above,purportedly solidifies this early chronologyof initial Mariana Island settlement, butwith the added wrinkle of suggesting thatthe initial migratory thrust of Austronesiansout of ISEA was to the Mariana Islandsand not to Near Oceania (see also Rainbird2004:85). This position has been challengedbased on analyses of ceramic attributes andchronological data (Clark et al. 2010; Clarkand Winter n.d.; Petchey et al. 2016; Winteret al. 2012), as well as computer simula-tion modeling of voyaging (Fitzpatrick andCallaghan 2013; Montenegro et al. 2016).

Previously, the Bismarck Archipelagoheld priority as the initial area of settlement

in Oceania by migrants associated with theLapita Cultural Complex (Kirch 2000:88–93, 2010), which is materially identified byfinely dentate-stamped pottery and an asso-ciated suite of lithic and shell tools and shellornaments (see Specht et al. 2014 for an ar-gument that material and cultural changesoccurred variably across time and space inthe archipelago and not as a single Lapita“package”).

While the broad outline of this expan-sion out of ISEA is not in dispute, the ideathat the Mariana Islands were settled be-fore the Lapita settlement in the BismarckArchipelago is a hypothesis that needs to betested. We developed a series of Bayesiancalibration models using archaeologicaldata1 to clarify temporal trends of lateHolocene expansion and settlement inthese islands, and address the question:does initial human occupation of the Mar-iana Islands precede Lapita settlement ofthe Bismarck Archipelago? The robustnessand reliability of chronologies are then eval-uated, and we consider the use of variousplant materials and marine invertebratesfor radiocarbon dating in this region (Allenand Huebert 2014; Petchey 2009; Riethand Athens 2013). Methodological issues inmacro-scale chronological analyses such asthis are also evaluated.

METHODS

For the Mariana Islands, conventionalradiocarbon ages (CRA) were collectedfrom the academic and cultural resourcemanagement (CRM) literature (Supple-mental Material: Table 1; supplementalmaterial available online). Measured radio-carbon ages lacking correction for isotopicfractionation were excluded, as wereshell-derived ages unless locality-specificdeltaR (�R) values had been calculated.Research by Petchey et al. (2016) showsthat an archipelago-wide �R value, evenif taxon-specific, cannot be developed forthe Mariana Islands, due in part to a geo-graphical variation in hardwater effect. Ouranalysis uses locality- and taxon-specific �Rvalues for Anadara antiquata (218 ± 57

2 VOLUME 0 • ISSUE 0 • 2017

Human Expansion into Oceania: A Bayesian Model

14C years; pooled value) and marine inverte-brate taxa that inhabit reef or open marineenvironments (23 ± 37 14C years) for UnaiBapot, Saipan (Petchey et al. 2016), and A.antiquata (−44 ± 41 14C years; pooledvalue) for Ritidian, Guam (Carson 2010).

For the Bismarcks, we relied on syn-theses published by Specht and Gosden(1997), Spriggs (2003), Specht (2007), andDenham et al. (2012), although in lim-ited cases the primary references were re-viewed. Radiocarbon determinations wereobtained from unidentified and identifiedplant charcoal, wood, and marine shell(Supplemental Material: Table 2). Denhamet al. (2012) had excluded marine shelldates because of questionable �R valuesfor the region; however, Petchey and Ulm(2012) have since published values for sixtentative �R subregions for the Bismarcksalong with a new value for Mussau. Usingtheir �R values we have included calibratedshell-derived dates.

Oxcal 4.2 was used for Bayesian modelcalibration (Bronk Ramsey 2009, 2013).Bayesian calibration allows estimates of thestart, end, and duration of events, whichvisual inspection of calibrated dates andsummed probabilities cannot provide withthe same statistical rigor and certainty(Bayliss et al. 2007). The Intcal13 calibra-tion curve was used for terrestrial sampleswith the Marine13 curve used for marinesamples (Reimer et al. 2013). The NorthernHemisphere curve (Intcal13) was used forthe Bismarck samples, following McCormacet al. (2004:1088) since this region fallswithin the Intertropical Convergence Zone(ITCZ).

Analysis was iterative. Single or multi-phase calibration models (see Bronk Ram-sey 2009) were created for the Marianasites, with single phase models createdfor the Mussau Islands and the BismarckArchipelago excluding Mussau. The individ-ual models for the Mariana Islands, Mussau,and greater Bismarcks were then combinedin an overlapping, multi-sequence modelto allow estimation of the differences insettlement dates. The models used Oxcal’sSequence, Phase, Boundary, and Difference

commands (code provided as Supplemen-tal Material). All individual models haveagreement indices above 60 (equivalent tothe 5% level of a χ2 test) and show goodagreement between the radiocarbon datesand phasing. Individual dates with lowagreement values were excluded if theycaused a model’s agreement value to fallbelow 60. The combined multi-sequencemodel has an agreement index below 60.However, the settlement estimates for eachregion remain unchanged from the indi-vidual modeled results, and therefore themulti-sequence model estimates are consid-ered valid. Modeled results are italicized todistinguish them from unmodeled dates.

SETTLEMENT OF THE MARIANAISLANDS

The Mariana Islands are approximately2,400 km due east of the northernPhilippines and roughly 1,700 km north-northwest of the Bismarck Archipelago.Eight deposits in the archipelago have pro-duced early assemblages and have associ-ated radiocarbon determinations (Figure 1).The number of determinations per site usedin our calibration models ranges from 1 (Ri-tidian, Guam) to 33 (Unai Bapot, Saipan).

Two site deposits were considered forGuam.2 One determination obtained froman A. antiquata shell was included for theRitidian site at the northern end of theisland. As a single age, it was simply cali-brated rather than calibrated in a Bayesianmodel structure. At 95% probability, occu-pation is estimated to have occurred 3501–3196 cal BP with a 68% probability rangeof 3435–3292 cal BP and 3283–3274 calBP. Six determinations from unidentifiedcharcoal from Mangilao were included ina multi-phase calibration model based onsite stratigraphy (Levels 10–13 of Layer IIIg,the basal cultural deposit). The oldest andyoungest determinations for level 12 wereflagged during the initial model iteration ashaving low agreement values and were re-moved, thus leaving the six determinations.

THE JOURNAL OF ISLAND AND COASTAL ARCHAEOLOGY 3

Timothy M. Rieth and J. Stephen Athens

Figure 1. Locations of early archaeological deposits in the Mariana Islands included in this study.

Initial occupation at Mangilao occurredsome time during 3631–3148 cal BP (95%probability) or more likely 3410–3207 calBP (68% probability).

The House of Taga and Unai Chuluwere included for Tinian. For the House

of Taga, two unidentified charcoal samplesincluded in a single phase model providean estimate for initial occupation sometimebetween 3943 and 3225 cal BP (95% prob-ability) or more likely 3443–3246 cal BP(68% probability). Thirteen determinations

4 VOLUME 0 • ISSUE 0 • 2017

Human Expansion into Oceania: A Bayesian Model

from Unai Chulu were included in a multi-phase model based on excavation units andsite stratigraphy. Initial occupation is esti-mated to have occurred sometime between3522 and 3215 cal BP (95% probabil-ity) or likely 3406–3246 cal BP (68%probability).

Achugao, Chalan Piao, and Unai Bapotwere included for Saipan. For Achugao, fiveages from unidentified charcoal were in-cluded in a single phase model, which es-timates initial occupation sometime during4894–3383 cal BP (95% probability) orlikely 4085–3513 cal BP (68% probabil-ity). Two unidentified charcoal dates forChalan Piao result in an even broader occu-pation estimate of 5833–3175 cal BP (95%probability) or likely 4117–3226 cal BP(68% probability).

Unai Bapot offers the best chronol-ogy for the archipelago. Thirty-four agesreported by Petchey et al. (2016) fromunidentified charcoal, identified short-livedplant charcoal, an extirpated rail (Galliral-lus cf. philippensis), Anadara sp. shell, andother marine invertebrates were organizedin a two-phase model based on stratigraphy.Determinations obtained by other investi-gators were excluded because they couldnot be correlated with the site stratigraphypresented by Petchey et al. (2016).3 Initialoccupation is estimated to have occurredsometime between 3224 and 3083 cal BP(95% probability) or likely 3208–3090 calBP (68% probability). Our model struc-ture follows Petchey et al. (2016) and, asexpected, our results are indistinguishablefrom their estimate (3200–3080 cal BP,68% probability).

With the exception of Unai Bapot, theinitial settlement estimates for the MarianaIsland sites (Figure 2) are problematic forat least three reasons. First, sample size isaffecting the results. Simulation modelingconducted by one of us (TMR, along withD. Hamilton [Scottish Universities Environ-mental Research Centre]) indicates that forthe portion of the calibration curve rele-vant to the settlement of the Mariana Islandsand Bismarcks, ∼40 radiocarbon determi-nations are required to obtain sub-centuryprecision. Unai Bapot comes close to this

sample size and has the most precise esti-mates: sometime within a ∼150 year span at95% probability or ∼80 years at 68% proba-bility. In contrast, Chalan Piao and Achugaoare modeled on 2 and 5 dates, respectively,resulting in 95% probability estimates of1,500–2,700 years. Second, potential inbuiltage for unidentified charcoal may be affect-ing results to varying degrees. Inbuilt ageof decades to more than a century must beconsidered for all assemblages. However, alarge enough sample size should allow de-tection of anomalously old ages (e.g., resid-ual dating samples) as Petchey et al. (2016)demonstrate for Unai Bapot. Third, Achugaoand Chalan Piao have other problems inthe form of large error values and poorlyprovenienced (aggregated) dating samples,respectively.

LAPITA SETTLEMENT OF THE BISMARCKARCHIPELAGO

The Bismarck Archipelago is off the north-east coast of New Guinea, consisting oflarge, volcanically active New Britain, NewIreland, the Mussau Islands, the ManusIslands, and hundreds of smaller islands andisland groups. Humans initially occupiedmany areas of the archipelago during thePleistocene (Torrence et al. 2004). Figure 3displays the locations of 23 Lapita sites inthe Bismarck Archipelago and Nissan Island(sites are listed by the National Museum andArt Gallery of Papua New Guinea three orfour letter code).

The Mussau Group has been recog-nized as having the earliest Lapita sites inthe Bismarcks, designated ECA and ECBon Eloaua (Kirch 2001). This assessmentis supported by Denham et al.’s (2012)Bayesian analysis, which is based on char-coal and wood-derived radiocarbon deter-minations. We included charcoal and woodages along with marine shell determinationsfrom Kirch’s (2001) excavations, as well astwo determinations obtained more recentlyby Summerhayes et al. (2010) from Emi-rau. These ages (n = 26) were calibrated ina single-phase model for the island group.

THE JOURNAL OF ISLAND AND COASTAL ARCHAEOLOGY 5

Timothy M. Rieth and J. Stephen Athens

Figure 2. Settlement estimates for Guam, Tinian, and Saipan.

Shell dates were calibrated using Petcheyand Ulm’s (2012) �R of −293 ± 92. Ourresults indicate Lapita occupation begansometime during 3535–3234 (95% proba-bility) or likely 3453–3283 (68% probabil-ity) (Figure 4). At 68%, our results are es-sentially the same as Denham et al.’s favoredestimate (3470–3250 cal BP, 68.2% prob-ability), while our 95% probability resultsremove over 100 years from their estimate(3590–3110 cal BP, 95.4% probability).

We initially ran a single phase modelthat included charcoal and marine shelldates for the remaining Lapita sites in theBismarcks, but most of the shell-deriveddates were identified as having low agree-ment values. Therefore, we separated thesamples by material—charcoal or marineshell—and ran two single phase models. Ma-rine shell determinations were calibratedusing Petchey and Ulm’s (2012) �R of

38 ± 14, 40 ± 19, 273 ± 216, and314 ± 74, depending on the site locations.A cursory assessment of both models sug-gests that separate settlement events weredated: one which occurred sometime dur-ing 4018–3694 cal BP (95% probability)based on charcoal-derived determinationsand another, which post-dates the first bycenturies, between 3510 and 3094 cal BP(95% probability) based on shell-deriveddeterminations. However, this certainly isnot the case since this pattern is corre-lated with sample material. Rather, there arethree primary factors that may be at work.One, charcoal determinations with large er-ror values (170 or 190 years) are skewingthe model results to the left (older sideof the probability curve). Two, the char-coal model may be skewed by inbuilt agein some of the unidentified charcoal sam-ples. Or, three, the �R values for the region

6 VOLUME 0 • ISSUE 0 • 2017

Human Expansion into Oceania: A Bayesian Model

Figure 3. Locations of early Lapita deposits in the Bismarck Archipelago.

are incorrect and thus skew the results tothe right (younger side of the probabilitycurve).

To test factor one, we ran the char-coal model again after removing three de-terminations with error values of 170 or190 years. The settlement date estimate re-mained statistically the same. To test factortwo we ran the charcoal model with onlydates from nutshells. The initial occupa-tion estimate generated from the short-livednutshell dates is statistically indistinguish-able from the marine shell-derived estimate.When these datasets are combined, initialLapita occupation is estimated to have oc-curred sometime between 3397 and 3115(95% probability) or likely 3304–3177(68% probability) (see Figure 4). With re-gards to factor three—errors in the �Rvalues—we cannot evaluate this possibil-ity directly now. However, the concordance

between the nutshell and marine shell esti-mates supports the accuracy of Petchey andUlm’s (2012) �R values.

Our results indicate that Lapita set-tlement of Mussau probably preceded ex-pansion into other parts of the BismarckArchipelago, though contemporaneous set-tlement cannot be discounted (−80 to334 years [95% probability], or 29–234 years [68% probability]). It is impor-tant to note that the Mussau estimate, un-like the greater-Bismarck estimate, includesa number of determinations obtained fromunidentified charcoal. Based on our evalua-tion of the greater-Bismarck suite, it is likelythat some of the Mussau charcoal dates in-clude a degree of inbuilt age that is affect-ing the results. The sample size for eacharea is moderate (n = 26 for each model)though future reliable dates should improvethe precision of the settlement estimates.

THE JOURNAL OF ISLAND AND COASTAL ARCHAEOLOGY 7

Timothy M. Rieth and J. Stephen Athens

Figure 4. Settlement estimates for the Mussau Islands and greater Bismarck Archipelago.

DISCUSSION: SEQUENTIAL ORCONTEMPORANEOUS POPULATION

EXPANSION IN THE WESTERN PACIFIC?

Our results do not support the contentionthat the initial occupation of the MarianaIslands preceded Lapita settlement of theBismarck Archipelago. Colonization of theMariana Islands occurred approximately50–385 years (95% probability) afterLapita appearance in Mussau. However, thisfinding is based on Unai Bapot—whichis well-dated, contains early ceramics andother artifacts, and extirpated avifauna—being representative of earliest settlementin the archipelago. Our results indicatethat Lapita settlement of Mussau occurredsometime between 3535 and 3234 cal BP(95% probability), with contemporaneousor centuries-later (−80 to 334 years [95%probability]) expansion into the greater Bis-marck area between 3397and 3115 cal BP(95% probability) (Figure 5). Settlementof the Mariana Islands between 3230 and3085 cal BP (95% probability) may havebeen contemporaneous with this expansioninto the greater Bismarck Archipelago, or it

may have succeeded it by centuries (−66to 254 years [95% probability]). How-ever, colonization of the Mariana Islandsdoes represent initial dispersal into RemoteOceania, as noted by previous researchers(Carson 2014; Hung et al. 2011; Petcheyet al. 2016; see also Sheppard et al. 2015).

It is equivocal whether robust and re-liable chronologies have been establishedfor these regions. Unai Bapot is a well-datedsite deposit, assuredly one of the best inthe western Pacific. This only highlightsthe poor state of chronology building forthe rest of the Mariana Islands. At a macroscale, the Mariana dataset has to contendwith a majority of dates from unidentifiedcharcoal, samples from problematic prove-niences, potential residual dating samples,and the need for the development of lo-calized �R values. Re-dating key sites maybe possible using curated samples, whichfor certain sites such as Chalan Piao andAchugao may be the only option since thedeposits have been largely destroyed bydevelopment.

The Mussau and greater BismarckArchipelago datasets are somewhat better,

8 VOLUME 0 • ISSUE 0 • 2017

Human Expansion into Oceania: A Bayesian Model

Figure 5. Estimated settlement dates for the Bismarck Archipelago and Mariana Islands.

particularly due to the suite of nutshelldates and improved �R estimates. The Mar-iana issues are just as pertinent for the Bis-marcks. As with the Mariana Islands, an in-creased number of dates from short-livedmaterials reliably associated with human ac-tivity will improve the precision and accu-racy of the colonization estimates.

While it is possible that the initialsettlers of the Mariana Islands could havecome from the Bismarck region, potterydifferences with Lapita suggest it is morelikely they came directly from ISEA (Clarkand Winter n.d.). An origination point(s)in the Bismarcks, coastal New Guinea, orthe Maluku Islands is supported by sea-faring simulation models (Fitzpatrick andCallaghan 2013; Montenegro et al. 2016;also Winter et al. 2012). Interestingly, thesemodels provide no support for either down-wind (drift) or directed (planned) voyages

to the Mariana Islands from the Philippines,but show that between the months of Julyand October there was a modest probabil-ity for successful voyages originating fromthe south (Halmahera, New Guinea, andthe Bismarck Archipelago). Montenegroet al. (2016), using several other modelingparameters (ENSO events, shortest-hop tra-jectories), show that successful downwindand directed voyages can occur to westernMicronesia via Palau and Yap from thesouthern Philippines and Maluku Islandsduring ENSO conditions, and from NewGuinea and the Bismarcks during La Niñaconditions.

These studies, of course, do not addressthe issue of chronology that is the subjectof this paper, but they do concur that anorigin for initial Mariana Islands settlementfrom the northern Philippines is unlikelyas posited by Hung et al. (2011). Further,

THE JOURNAL OF ISLAND AND COASTAL ARCHAEOLOGY 9

Timothy M. Rieth and J. Stephen Athens

contra Hung et al. (2011) and Carson et al.(2013), even the best candidate site fordemonstrating a Philippine connectionthrough its pottery, Nagsabaran, locatedin the northern Philippines, shows con-siderable significant differences with earlyPre-Latte Period pottery (Clark and Winterin press; Winter et al. 2012). Similarly, thepottery evidence does not support a disper-sal of early Austronesians from the MarianaIslands to either Near or Remote Oceania.

It is apparent that the period ∼3500–3000 cal BP was one of dynamic populationmovement in the western Pacific (Kirch2010), with continued expansion into Re-mote Oceania in the following centuries(Sheppard et al. 2015). Movement bymultiple human groups from numerouslocations within a relatively short timeperiod, and across the large ISEA/westernPacific region, may be explicable through ageneral population-level mechanism, suchas the agriculturally fueled demographicexpansion argued by Bellwood (2011), orselection associated with changes in cli-mate and maritime technology as argued byCochrane (in press). A reliable chronologyis essential for distinguishing between theseand other possibilities.

Chronological modeling for the west-ern Pacific needs to be improved with ad-ditional quality dating samples. We suggestthat archaeologists exercise greater care insample selection, choosing short-lived sam-ples confidently associated with human ac-tivity, while controlling for residual sam-ples, and date more high-quality samplesfrom individual sites. Further research isalso needed to accurately determine lo-calized �R values throughout regions ofinterest. This work has been ongoing inHawai‘i and central Polynesia for the lastseveral decades, with the result that settle-ment dates are often determined to be laterthan previously thought (e.g., Athens et al.2014; Dye 2015; Rieth and Cochrane 2017;Wilmshurst et al. 2008). A similar reevalu-ation is already underway in the MarianaIslands and Near Oceania, and we antici-pate a slightly later chronology for Lapita ex-pansion to the Bismarck Archipelago whenprobability estimates for dating in these re-gions are narrowed.

ACKNOWLEDGEMENTS

An earlier version of this article was pre-sented at the 2016 Society for AmericanArchaeology meeting in a session chairedby Robert Dinapoli and Thomas Leppard.Geoff Clark, Tom Dye, Jennifer Huebert,Fiona Petchey, and Jim Specht providedhelpful comments on a draft of the article,while Derek Hamilton and Tony Krusgraciously answered questions about theBayesian modeling. Lastly, we would liketo thank Tony and Victor Thompson forinviting us to submit this article.

END NOTES

1. We do not consider paleoenvironmental evi-dence for initial Mariana Islands colonizationat or slightly prior to 3500 BP (Athens et al.2004; Athens and Ward 1995, 2004) due tothe qualitatively different kind of data usedin such studies. They suggest a human pres-ence going back to the mid- or late fourth mil-lennium BP; however, experience in Hawai‘ihas shown the need for caution when mak-ing fine-resolution dating inferences based onbulk sediment samples obtained from coastalbrackish or saltwater environments as in theMariana Islands (Athens et al. 2014:152).

2. The single determination on “marine shells,primarily limpets” (Kurashina and Clayshulte1983:118) from Tarague was not considereddue to uncertainties about the proper �Rvalue for this location.

3. Inclusion of Carson’s (2008) A. antiquata-derived ages from his basal cultural stra-tum (Layer IV-A) in our Layer VII modelphase did not produce a significantly dif-ferent settlement estimate. Carson’s oldestdetermination (Beta-216616, 3710 ± 50 BP)had a poor agreement value.

REFERENCES

Allen, M. S. and J. M. Huebert. 2014. Short-livedplant materials, long-lived trees, and Polyne-sian 14C dating: Considerations for 14C sam-ple selection and documentation. Radiocar-bon 56(1):257–276.

10 VOLUME 0 • ISSUE 0 • 2017

Human Expansion into Oceania: A Bayesian Model

Athens, J. S. 1986. Archaeological Investigationsat Tarague Beach, Guam. Report preparedfor Base Civil Engineering, 43D Strategic Wing,Andersen Air Force Base (available at Interna-tional Archaeological Research Institute, Inc.,Honolulu).

Athens, J. S., M. F. Dega, and J. V. Ward.2004. Austronesian colonization of the Mar-iana Islands: The palaeoenvironmental evi-dence. Indo-Pacific Prehistory AssociationBulletin 24(2):21–30.

Athens, J. S., T. M. Rieth, and T. S. Dye.2014. A paleoenvironmental and archaeolog-ical model-based age estimate for the col-onization of Hawai‘i. American Antiquity79(1):144–155.

Athens, J. S. and J. V. Ward. 1995. Paleoenvi-ronment of the Orote Peninsula, Guam. Mi-cronesica 28(1):51–76.

Athens, J. S. and J. V. Ward. 2004. Holocene vege-tation, savanna origins and human settlementof Guam. Records of the Australian Museum,Supplement 29:15–30.

Bayliss, A., C. Bronk Ramsey, J. van der Plicht,and A. Whittle. 2007. Bradshaw and Bayes:towards a timetable for the Neolithic. Cam-bridge Archaeological Journal 17(1):1–28.

Bellwood, P. 1975. The prehistory of Oceania.Current Anthropology 16(1):9–28.

Bellwood, P. 2011. Holocene population historyin the Pacific region as a model for worldwidefood producer dispersals. Current Anthropol-ogy 52(S4):S363–S3678.

Bronk Ramsey, C. 2009. Bayesian analysis ofradiocarbon dates. Radiocarbon 51(1):337–360.

Bronk Ramsey, C. 2013. Oxcal 4.2. http://c14.arch.ox.ac.uk/oxcal.

Butler, B. M. 1994. Early prehistoric settlementin the Mariana Islands: New evidence fromSaipan. Man and Culture in Oceania 10:15–38.

Carson, M. T. 2008. Refining earliest settlementin Remote Oceania: Renewed archaeologicalinvestigation at Unai Bapot, Saipan. Journal ofIsland and Coastal Archaeology 3:115–139.

Carson, M. T. 2010. Radiocarbon chronologywith marine reservoir correction for the Ri-tidian archaeological site, northern Guam. Ra-diocarbon 52(4):1627–38.

Carson, M. T. 2014. First Settlement of RemoteOceania: Earliest Sites in the Mariana Is-lands. New York: Springer.

Carson, M. T., H.-C. Hung, G. Summerhayes,and P. Bellwood. 2013. The pottery trail fromSoutheast Asia to Remote Oceania. Journal ofIsland and Coastal Archaeology 8:17–36.

Carson, M. T. and H. Kurashina. 2012. Re-envisioning long-distance Oceanic migration:Early dates in the Mariana Islands. World Ar-chaeology 44(3):409–435.

Clark, G., F. Petchey, O. Witner, M. Carson, and P.O’Day. 2010. New radiocarbon dates from theBapot-1 site in Saipan and Neolithic dispersalby stratified diffusion. Journal of Pacific Ar-chaeology 1(1):21–35.

Clark, G. and O. Winter. In press. The ceramictrail: Evaluating the Marianas and Lapita WestPacific connection. In Lapita Colonisation ofthe Western Pacific (S. Bedford and M. Spriggs,eds.). Canberra: ANU E Press.

Cochrane, E. E. In review. The evolution of mi-gration: The case of Lapita in the southwest Pa-cific. Journal of Archaeological Method andTheory.

Denham, T., C. Bronk Ramsey, and J. Specht.2012. Dating the appearance of Lapita potteryin the Bismarck Archipelago and its dispersalto Remote Oceania. Archaeology in Oceania47:39–46.

Dye, T. S. 2015. Dating human dispersal in Re-mote Oceania: A Bayesian view from Hawai‘i.World Archaeology 47(4):661–676.

Fitzpatrick, S. M. and R. T. Callaghan. 2013.Estimating trajectories of colonisation to theMariana Islands, western Pacific. Antiquity87:840–853.

Hung, H., M. T. Carson, P. Bellwood, F. Z. Cam-pos, P. J. Piper, E. Dizon, M. J. L. A. Bolunia, M.Oxenham, and Z. Chi. 2011. The first settle-ment of Remote Oceania: The Philippines tothe Marianas. Antiquity 85(329):909–926.

Kirch, P. V. 2000. On the Road of the Winds: AnArchaeological History of the Pacific Islandsbefore European Contact. Berkeley: Univer-sity of California Press.

Kirch, P. V. (ed.). 2001. Lapita and its Trans-formations in Near Oceania: ArchaeologicalInvestigations in the Mussau Islands, PapuaNew Guinea, 1985-88. Berkeley: Contribu-tions of the Archaeological Research Facility,Number 59.

Kirch, P. V. 2010. Peopling of the Pacific: A holis-tic anthropological perspective. Annual Re-view of Anthropology 39:131–148.

Kurashina, H. and R. N. Clayshulte. 1983. Siteformation processes and cultural sequence atTarague, Guam. Indo-Pacific Prehistory Asso-ciation Bulletin 4:114–122.

McCormac, F. G., A. G. Hogg, P. G. Blackwell, C.E. Buck, T. F. G. Higham, and P. J. Reimer. 2004.SHCal04 southern hemisphere calibration 0-11.0 cal kyr BP. Radiocarbon 46(3):1087–1092.

THE JOURNAL OF ISLAND AND COASTAL ARCHAEOLOGY 11

Timothy M. Rieth and J. Stephen Athens

Montenegro, Á., R. T. Callaghan, and S. M. Fitz-patrick. 2016. Using seafaring simulations andshortest-hop trajectories to model the prehis-toric colonization of Remote Oceania. Pro-ceedings of the National Academy of Sciences113(45):12685–12690.

Petchey, F. 2009. Dating marine shell in Ocea-nia: Issues and prospects. In New Directionsin Archaeological Science (A. Fairbairn andS. O’Connor, eds.):157–172. Canberra: ANU EPress.

Petchey, F., G. Clark, O. Winter, P. O’Day, andM. Litster. 2016. Colonisation of Remote Ocea-nia: New dates for the Bapot-1 site in theMariana Islands. Archaeology in Oceania.doi:10.1002/arco.5108.

Petchey, F. and S. Ulm. 2012. Marine reservoirvariation in the Bismarck region: An evalua-tion of spatial and temporal change in �Rand R over the last 3000 years. Radiocarbon54(1):45–58.

Rainbird, P. 2004. The Archaeology of Microne-sia. Cambridge: Cambridge University Press.

Reimer, P. J., E. Bard, A. Bayliss, J. W.Beck, P. G. Blackwell, C. Bronk Ramsey,C. E. Buck, et al. 2013. IntCal13 and Ma-rine13 radiocarbon age calibration curves 0–50,000 years cal BP. Radiocarbon 55(4):1869–1887.

Rieth, T. and J. S. Athens. 2013. Suggested bestpractices for the application of radiocarbondating to Hawaiian archaeology. Hawaiian Ar-chaeology 13:3–29.

Rieth, T. and E. Cochrane. 2017. The chronol-ogy of colonization in Remote Oceania. In TheOxford Handbook of Prehistoric Oceania (E.Cochrane and T. Hunt, eds.). Oxford: OxfordUniversity Press.

Russell, S. 1998. Tiempon I Manmofo’Na: An-cient Chamorro Culture and History of theNorthern Mariana Islands. Micronesian Ar-chaeological Survey Report No. 32. Saipan:Commonwealth of the Northern Mariana Is-lands, Division of Historic Preservation.

Sheppard, P. J., S. Chiu, and R. Walter. 2015. Re-dating Lapita movement into Remote Ocea-

nia. Journal of Pacific Archaeology 6(1):26–36.

Specht, J. 2007. Small islands in the big picture:The formative period of Lapita in the BismarckArchipelago. In Oceanic Explorations: Lapitaand Western Pacific Settlement (S. Bedford, C.Sand, and S. P. Connaughton, eds.):51–70. Can-berra: ANU E Press.

Specht, J., T. Denham, J. Goff, and J. E. Terrell.2014. Deconstructing the Lapita Cultural Com-plex in the Bismarck Archipelago. Journal ofArchaeological Research 22(2):89–140.

Specht, J. and C. Gosden. 1997. Dating Lapitapottery in the Bismarck Archipelago, PapuaNew Guinea. Asian Perspectives 36(2):175–194.

Spoehr, A. 1957. Marianas Prehistory: Archae-ological Survey and Excavations on Saipan,Tinian and Rota (Fieldiana: Anthropology48). Chicago: Chicago Natural History Mu-seum.

Spriggs, M. 2003. Chronology of the Neolithictransition in Island Southeast Asia and thewestern Pacific: A view from 2003. The Re-view of Archaeology 24(2):57–80.

Summerhayes, G. R., E. Matisoo-Smith, H. Man-dui, J. Allen, J. Specht, N. Hogg, and S. McPher-son. 2010. Tamuarawai (EQS): An early Lapitasite on Emirau, New Ireland, PNG. Journal ofPacific Archaeology 1(1):62–75.

Torrence, R., V. Neall, T. Doelman, E. Rhodes,C. McKee, H. Davies, R. Bonetti, et al. 2004.Pleistocene colonization of the BismarckArchipelago: New evidence from West NewBritain. Archaeology in Oceania 39(3):101–130.

Wilmshurst, J. M., A. J. Anderson, T. F. G. Higham,and T. H. Worthy. 2008. Dating the late prehis-toric dispersal of Polynesians to New Zealandusing the commensal Pacific rat. Proceed-ings of the National Academy of Sciences105(22):7676–7680.

Winter, O., G. Clark, A. Anderson, and A. Lin-dahl. 2012. Austronesian sailing to the north-ern Marianas, a comment on Hung et al. An-tiquity 86:898–914.

12 VOLUME 0 • ISSUE 0 • 2017