On the Rat Trail in Near Oceania: Applying the Commensal Model to the Question of the Lapita Colonization1

E. Matisoo-Smith,2,4,7 M. Hingston,3,4 G. Summerhayes,5 J. Robins,2,4 H. A. Ross,3 and M. Hendy6

Abstract: Presented here are the most recent results of our studies of Rattus exulans, one of the main commensal animals transported across the Pacific by Lapita peoples and their descendants. We sampled several locations in Near Oceania to determine distribution of R. exulans mitochondrial DNA (mtDNA) haplotypes in the region. We also obtained data regarding distribution of other introduced Rattus species to several islands in the Bismarck Archipelago. Our results suggest that there were multiple introductions of R. exulans to the region, which may suggest a more complex history for Lapita populations in Near Oceania.

One of the greatest impacts of human ive fauna and numerous endemic species that arrival on previously uninhabited islands is cannot compete with the more recent intro- the introduction of human-associated plants ductions. and animals. The species introduced by hu- Despite the often negative impacts of in- mans can include not only intentional intro- troduced species on island ecosystems, there ductions such as domesticated plants and were clearly good reasons for people to trans- animals, but also a range of unintentionally port their familiar plants and animals to the transported species including weeds, insects, new environments they occupied. The trans- and other pests (Kirch 1982, Lee et al. ported landscapes (Kirch 1984) of Pacific 2007). Often island ecosystems contain a na- peoples also provide a valuable resource for prehistorians. Not only do they allow us to understand and appreciate how humans adapted to the various environments they en- countered, but understanding the history of 1 This research was funded by the Royal Society of the plants and animals that humans trans- New Zealand under Marsden Project UOA510: The ported can provide direct evidence regarding Rat’s Tale: Tracking Lapita Peoples through Phyloge- the history of the humans themselves: Where netic Analyses of Pacific Rats. Manuscript accepted 23 did they come from and when? How many January 2009. 2 Department of Anthropology, University of Auck- introductions and population arrivals were land, Auckland 1, New Zealand. there? Is the appearance of a particular spe- 3 School of Biological Sciences, University of Auck- cies associated with any particular archaeo- land, Auckland 1, New Zealand. logically definable culture? 4 Allan Wilson Centre for Molecular Ecology and Evolution, University of Auckland, Auckland 1, New Beginning in the 1990s a program was Zealand. developed at the University of Auckland fo- 5 Department of Anthropology, University of Otago, cused on determining if tracking the move- P.O. Box 56, Dunedin, Otago, New Zealand. ment of commensal animals introduced to 6 Allan Wilson Centre for Molecular Ecology and Pacific islands by initial human colonists, Evolution, Massey University, Palmerston North Cam- pus, Private Bag 11222, Palmerston North 444, New through analyses of mitochondrial DNA Zealand. (mtDNA) variation, might serve as a proxy 7 Corresponding author (e-mail: e.matisoo-smith@ for tracing human migration in the Pacific auckland.ac.nz). (Matisoo-Smith 1994, Matisoo-Smith et al. 1998). Pacific colonists transported, among Pacific Science (2009), vol. 63, no. 4:465–475 other things, dogs, pigs, chickens, and rats : 2009 by University of Hawai‘i Press when they settled the previously uninhabited All rights reserved islands of Remote Oceania. It is also gener-

465 . 466 PACIFIC SCIENCE October 2009 ally agreed that the Lapita colonists were the spheres in central East Polynesia (see Figure first to introduce dogs, pigs, chickens, and the 1), both most likely originating in a home- Pacific rat (Rattus exulans) to islands in Near land region centered in the southern Cook Oceania (Kirch 2000, Spriggs 1997), though and Society islands. A southern interaction debates continue around the possibility of sphere connected these populations with earlier, pre-Lapita introductions of pigs and New Zealand and the Kermadec Islands, and dogs to New Guinea (Bulmer 1982, 2001, a northern interaction sphere connected this Goreki et al. 1991, Allen 2000, Green 2000, homeland region to the Marquesas and Ha- Bellwood and White 2005, Larson et al. waiian islands. These results were consistent 2007). with archaeological and linguistic data as well It was decided, for a number of reasons, as oral traditions (Green 1966, Kirch 1986, that the Pacific rat, R. exulans, was the best Irwin 1992, Cachola-Abad 1993). Once it animal with which to first develop and test was shown that the Commensal Model did what we now refer to as the Commensal indeed provide a reasonable proxy for track- Model for human settlement of the Pacific. ing human migrations, the range was ex- The Commensal Model is based on the close panded beyond the Polynesian triangle, and relationship between human populations and a diachronic perspective was added through the plants and animals they transported across the application of ancient DNA (aDNA) the Pacific. By identifying the genetic rela- methods to archaeological remains of R. ex- tionship of the various populations of com- ulans (Matisoo-Smith et al. 1997, 1999, 2001). mensal plants and animals, we can model the To further expand this Commensal Model prehistoric migration and interaction patterns and address the issue of Lapita origins, of Pacific peoples. A primary reason for the Matisoo-Smith and Robins (2004) looked at choice of R. exulans for a commensal study is how Polynesian and other Remote Oceanic its near-ubiquitous distribution in the Pacific. samples related to those from Near Oceania In addition, although the dogs, pigs, and and Island Southeast Asia. The results of chickens carried by early Pacific colonists be- these analyses identified three major hap- long to the same species as those brought in logroups of R. exulans, each with a very dis- by European vessels from the 1700s onward, tinct geographic distribution. All three the rats introduced by these same historic haplogroups appeared to be ultimately de- vessels, Rattus rattus and Rattus norvegicus, are rived from Mainland Southeast Asian popu- different species and do not interbreed with lations. Haplogroup I was found only in R. exulans (Mayr 2000, Robins et al. 2007). populations from the Philippines, Borneo, Rattus exulans is not known to stow away in and Sulawesi. Although its presence in Bor- European shipping vessels, and therefore the neo may represent natural expansion of the populations living on Pacific islands today species across the Sunda Shelf during periods are the direct descendants of those intro- of lowered sea levels, Haplogroup I was most duced by prehistoric colonists. We can there- likely transported by humans to both the fore study extant populations as well as Philippines and Sulawesi, which were sepa- archaeological remains of R. exulans from rated from Sunda by deep undersea troughs. across the Pacific to model the prehistoric It was, however, the distribution of Hap- human colonization of Remote Oceania. logroups II and III that provided evidence The first major test of the Commensal of importance to issues of Lapita expansion. Model for human settlement was based on Given the archaeological evidence for a clear an analysis of mtDNA variation in Polynesian link between Lapita sites in Near and Remote populations of R. exulans (Matisoo-Smith Oceania, the results of this study were sur- 1994, Matisoo-Smith et al. 1998). Analyses prising. It was found that the Near and Re- of 132 R. exulans samples collected from mote Oceanic R. exulans populations sampled throughout Polynesia indicated that the were very different. In fact, there were no Commensal Model worked, with results sug- Near Oceanic mtDNA lineages in Remote gesting that there were two major interaction Oceania and vice versa, but both lineages . On the Rat Trail in Near Oceania Matisoo-Smith et al. 467

Figure 1. Map of Polynesia showing the interaction spheres identified through mtDNA analyses of Polynesian pop- ulations of Rattus exulans.

were present in Halmahera (Island Southeast followed by a Lapita introduction to small is- Asia/Wallacea) to the west, where they were lets, locations that have not been sampled in presumably introduced by humans at some any previous analyses, morphological or ge- point in prehistory. Studies of morphological netic. Our goal was therefore to test these variation of Pacific R. exulans (Tate 1935, two possibilities through more thorough and Motokawa 2004) have reported a similar lack specifically directed sampling of both modern of continuity between Near and Remote and archaeological R. exulans populations Oceania. This could be (1) real; or (2) due from throughout Near Oceania. to two introductions of R. exulans to Near Archaeological evidence suggests that the Oceania—an early one to the main islands, Lapita people generally targeted small, off- . 468 PACIFIC SCIENCE October 2009 shore islands for settlement, but the Near deed R. exulans was a Lapita introduction, we Oceanic R. exulans samples reported in would expect to find it associated with all Matisoo-Smith and Robins (2004) were pri- islands that have evidence of Lapita occupa- marily from larger islands (e.g., New Guinea, tion. In June and July 2006, our trapping ef- New Britain, and Bougainville), which were forts were focused on the islands in the New not typical early Lapita targets. Rattus exulans Ireland and Manus provinces, all of which ei- is generally assumed to be a Lapita introduc- ther had known Lapita sites or were likely tion to both Near and Remote Oceania, yet sites for Lapita settlement. Rats were trapped the precise dating of its introduction to Near on New Ireland, Lihir, Tatau and Simberi is- Oceania is problematic, because sites from lands in the Tabar Group, New Hanover, and the period just before Lapita arrival are rare. Manus. We returned to New Guinea in April Three R. exulans bones were found at Panaki- 2007 to trap rats and conduct archaeological wuk, New Ireland, in layers dating to 8,000– excavations on Emirau (sometimes also re- 13,000 B.P., but with such early dates, they ferred to as Emira) Island in the St. Matthias were assumed to be there as a result of site Group. We were also able to arrange for a disturbance (Marshall and Allen 1991). There local fisherman to distribute traps and 2 ml is much evidence of animal translocation in tubes of ethanol to nearby Tench Island. We the region from that period onward (Flan- requested that local villages simply remove a nery 1995, Grayson 2001), so it is possible small section of the tail of any rats trapped that R. exulans was introduced to Near Oce- and preserve them in the tubes until the sam- ania before Lapita. If one lineage of R. ex- ples were collected at a later date. ulans were introduced and established on the large islands earlier than 3,500 B.P. and Lap- materials and methods ita peoples then introduced a new lineage to the previously uninhabited islands, then R. ex- Rats were trapped using snap-type rattraps ulans from the smaller islets, like Mussau in (Victor). Various volunteers from the local the Bismarcks, should have Remote Oceanic communities were recruited to set traps in lineages distinct from those found on the their gardens and around the villages. Rats larger islands. If they do not, then Remote were brought back to the field base in the Oceanic R. exulans must be coming from early morning, and all samples were mea- elsewhere, and the settlement of the region sured, recording body length, tail length, is even more complex than we currently hind foot length, maximum ear length, and believe. nipple number. Tentative species identifica- To address this question, we decided to tions were made in the field based on undertake a major research project in Near morphological characteristics as described Oceania. This project involved targeted field- by Cunningham and Moors (1983) and work to trap extant samples of R. exulans from Matisoo-Smith and Allen (2001). A range of the more likely Lapita target locations such as tissue samples was collected from each ani- the small islands in the Admiralty and St. mal, and the samples were stored in 70% Matthias island groups. We also attempted ethanol to be transported to the laboratories to obtain additional archaeological samples at the University of Auckland. from the more western locations within Re- DNA was extracted from tail, liver, paw, mote Oceania to determine as precisely as and/or ear tissue in the modern DNA Labo- possible the mtDNA haplotypes transported ratory at the Department of Anthropology, into Remote Oceania. University of Auckland. A preparation kit In November 2005, the first trip to New (Roche’s High Pure PCR Template Prepara- Guinea was undertaken to trap rats and to de- tion Kit, Roche Applied Science, Switzerland) termine if there was any evidence of Lapita was used according to the manufacturer’s on the small island of Koil, located approxi- protocol for isolation of nucleic acids from mately 60 km off the coast of Wewak, East mammalian tissue. To generate genetic infor- Sepik Province, Papua New Guinea. If in- mation capable of revealing phylogenetic . On the Rat Trail in Near Oceania Matisoo-Smith et al. 469 structure within the species, the hypervariable TABLE 1 control region (D-loop) of mitochondrial Rattus Species Recovered in Near Oceania DNA was targeted for analysis. Primers used for amplification of a 583 base pair (bp) seg- R. R. R. R. ment from position 15358 to 15940 (based Location n exulans tanezumi rattus norvegicus on numbering according to Gadaleta et al. [1989]) were EGL4L (50-CCA CCA TCA Koil Island 21 0 21 0 0 New Ireland 4 4 0 0 0 ACA CCC AAA G-3 0) and RJ3R (50-CAT 0 Lihir 4 3 0 1 0 GCC TTG ACG GCT ATG TTG-3 ). Tatau 25 9 16 0 0 Polymerase chain reaction (PCR) amplifica- Simberi 2 2 0 0 0 tions were carried out in standard 30 mL reac- New Hanover 9 9 0 0 0 m Manus 7 5 1 0 1 tions in which 1 L genomic DNA was added Emirau 18 17 0 1 0 to a reaction mixture containing 50 mM KCl, Tench 8 8 0 0 0 10 mM Tris HCl pH 8.3 (10 buffer), 2 mM Total 98 57 38 2 1 MgCl , 150 mM dNTPs, 0.5 mM of each 2 Note: All species identified based on morphological and primer, and 0.5 U Taq-Polymerase (Ampli- mtDNA analyses. Taq, Applied Biosystems). PCR was per- formed in a thermal cycler (iCycler, Bio-Rad Laboratories, California) with a thermal pro- results file of 35 cycles of denaturation at 94C for 30 sec, annealing at 60C for 30 sec, and ex- Trapping during the three field seasons re- tension at 72C for 1 min, preceded by an sulted in the collection of 98 rats. The specif- initial denaturation at 94C for 2 min and fol- ics regarding numbers caught, locations, and lowed by a final extension for 5 min at 72C. species are shown in Table 1. In his review Negative controls were always included to of New Guinea mammals, Flannery (1995) check for contamination. For verification of had no record of Rattus species present on successful amplification, 5 mL of PCR prod- Koil, New Hanover (Lavongai), Lihir, the uct was visualized in ethidium bromide– Tabar Group, Emirau, or Tench. Of the lo- stained 1% agarose gels. All PCR products cations included in our study, Flannery re- were purified (QIAquick PCR Purification corded Rattus species only on Manus and Kit, QIAgen, Hilden, Germany) and quanti- New Ireland, with Manus having three spe- fied in comparison with a low mass ladder in cies (R. exulans, R. praetor, and R. rattus)and ethidium bromide–stained 1% agarose gels. New Ireland only two species (R. exulans and Cycle sequencing thereafter was carried out R. praetor). at the Allan Wilson Centre Genome Service A total of 57 R. exulans were caught in our at Massey University, Albany, New Zealand, study, and the species was present on all is- using a sequencing kit (BigDye Terminator lands we visited except for Koil Island. All 21 v 3.1 Cycle Sequencing Kit, Applied Biosys- rats trapped on Koil were identified by their tems, Foster City, California) with 10 ng tem- mtDNA as Rattus tanezumi, an Asian rat com- plate per microliter and 1 mL5mM primer. monly referred to as the ‘‘Asian house rat’’ Capillary separation was carried out on a ge- and formerly considered a subspecies of R. netic analyzer (ABI3730, Applied Biosystems, rattus (Musser and Carleton 2005). Rattus Foster City, California). tanezumi was also abundant on Tatau Island For identification of rat species, the se- and was present on Manus. We have also pre- quences thus obtained were compared against viously identified it on the mainland of New the D-loop reference sequence alignment Guinea (E.M.-S., unpubl. data). of the DNA Surveillance phylogenetic tool We did not trap or encounter any New ‘‘What rat is that?’’ (www.dna-surveillance Guinea native rats on the islands we visited. .auckland.ac.nz) using the simple cluster We were also surprised to find that we search (Ross et al. 2003, Ross and Murugan trapped very few European rats, given the 2006, Robins et al. 2007). substantial number of U.S. military ships in . 470 PACIFIC SCIENCE October 2009 the region during World War II. We found those belonging to Haplogroup III, on only two R. rattus in our collections: one Tench, Manus, and New Hanover. All R. ex- from Lihir and one from Emirau. Only one ulans trapped on Tench and Manus belonged R. norvegicus was trapped on any of the is- to Haplogroup III, but those trapped on New lands we visited. This was most likely because Hanover belonged to both Haplogroups II our trapping focused on villages and in gar- and III. dens rather than in larger port cities/towns where R. rattus and R. norvegicus are most discussion likely to be found. It is not surprising that the only R. norvegicus was trapped in Loren- The main point of the rat trapping study in gau, the main town on Manus and an admin- the Bismarck Archipelago was to address the istrative base for both the Japanese and the question regarding the apparent discontinuity U.S./Australian forces during World War II. between the R. exulans populations in Near Of the R. exulans obtained in our field and Remote Oceania. Is there a real differ- study, we found that populations from New ence between the two regions or is the appar- Ireland, Lihir, Tatau, Simberi, and Emirau ent lack of connection merely the result of a contained lineages belonging exclusively to sampling problem in Near Oceania? In this mtDNA Haplogroup II, also referred to in regard, perhaps the most important finding Matisoo-Smith and Robins (2004) as the of our study is that we have now shown that Near Oceanic Type (Figure 2). We did, how- the lack of Remote Oceanic lineages in Near ever, find the Remote Oceanic lineages, or Oceania was indeed due to sampling error.

Figure 2. Distribution of R. exulans haplotypes in Near Oceania. Circles represent locations with R. exulans Hap- logroup II lineages, and triangles represent locations with Haplogroup III lineages. . On the Rat Trail in Near Oceania Matisoo-Smith et al. 471

There is now a clear connection between fishbone, so we cannot yet be sure that the the R. exulans populations found in Remote Haplotype II R. exulans we found on Emirau Oceania and those found in the Bismarck represents an introduction by the first Lapita Archipelago and a path linking the Remote colonists. However, given the relatively close Oceanic rats back to the island of Halmahera association between Lapita settlements and in the Moluccas, the most westerly location in the appearance of R. exulans, coupled with which we have found Haplogroup III. the fact that there is no indication of a pre- It is tempting, given the near-ubiquitous Lapita settlement phase on the island, we distribution of Haplogroup III in Remote suggest that, at least in Near Oceania, La- Oceania and its now-known distribution in pita settlement was associated with both the Bismarck Archipelago, to suggest that Haplogroups II and III. The fact that very Haplogroup III R. exulans is indeed a marker few island populations actually have both of Lapita migration. The presence of Hap- haplogroups may indicate that there were logroup III R. exulans on Manus is also inter- two Lapita introductions of R. exulans to esting given that island’s importance in the Near Oceania, with the earliest arrivals bring- Lapita-associated obsidian exchange system ing Haplogroup II. (Spriggs 1997, White and Harris 1997, Sum- The introduction of Haplogroup III into merhayes 2003). However, if we consider the Near Oceania and its dispersal into and known distribution of early Lapita sites in the throughout Remote Oceania may be associ- Bismarck Archipelago, we find that the ma- ated with a slightly later or additional group jority of those islands sampled so far have R. of Lapita colonists. Alternatively, the pre- exulans belonging exclusively to Haplogroup dominance of one haplogroup over another II—these include New Ireland and adjacent on the islands could be the result of genetic islands, New Britain, and Emirau—located drift or bottlenecks. Only analyses of archae- just a few kilometers from Mussau and the ological R. exulans remains can resolve this location of some of the earliest dated Lapita question. We are currently planning further deposits (Kirch 2000, Summerhayes 2007). archaeological excavations of Lapita sites or The Erarae archaeological site, located dur- closely associated natural deposits in the re- ing our time on Emirau, has now been se- gion to recover R. exulans bones for aDNA curely dated and represents one of the analyses to address this issue. earliest Lapita sites (G.S., E.M.-S., J. Specht, One island on which we did find the Re- K. Amanga, K. Thomas, and J. Ridges, un- mote Oceanic, Haplogroup III, R. exulans publ. data). It is interesting that the one loca- that does not yet have any indication of Lap- tion in Near Oceania in which we previously ita settlement is Tench Island. Tench is a identified Haplotype II was the Reef/Santa small, isolated atoll in the St. Matthias Group Cruz islands, just over the boundary between and is located approximately 70 km east of Near and Remote Oceania (Matisoo-Smith Emirau. Unlike Emirau Island, Tench is and Robins 2004). The sample consisted of a only a few meters above current mean sea bone recovered from the SE-RF-3 site, and level and would have been submerged during although the site is clearly post-Lapita (and the mid-Holocene high sea stand (i.e., the pe- therefore, again, does not provide any direct riod of Lapita occupation). As on Emirau and association between Lapita and Haplotype Mussau, the current populations had historic II), there is archaeological, linguistic, and connections with the eastern Carolines in genetic evidence for direct Lapita as well as Micronesia to the north, and this can be seen post-Lapita links between the Reef/Santa today in the presence of backstrap weaving Cruz islands and the Bismarck Archipelago loom, which was a Micronesian introduction (Green 1997, Friedlaender et al. 2002, Shep- (see Parkinson 1999:143, 148). Yet Tench pard and Walter 2006, Ross and Naess 2007). was culturally isolated from both Emirau and Unfortunately our preliminary test excava- Mussau to the west, and R. exulans on Tench tions of the Erarae site did not result in the may well have been introduced from Micro- recovery of any faunal remains other than nesia to the north. . 472 PACIFIC SCIENCE October 2009

The only island on which we trapped but Ireland, Manus, or Koil, may indicate that did not find any R. exulans was Koil, located the distribution of both of these species is some 60 km off the north New Guinea coast- now being impacted by more recently intro- line. The island is composed of raised lime- duced species. stone, some 4 km long and only 2 km at its widest point. The archaeological survey and conclusions test pitting on Koil Island (G.S., M. Leaves- ley, A. Fairbairn, and G. Hope, unpubl. data) Our focused sampling of Rattus species on indicated that there was also no evidence of the islands of Near Oceania has solved one Lapita settlement on the island, with perma- mystery identified by our previous genetic nent settlement probably no earlier than 600 analyses of R. exulans as a proxy for tracking years in age (G.S., M. Leavesley, A. Fairbairn, human migrations. Although we now have and G. Hope, unpubl. data). This result adds evidence of R. exulans Haplogroup III in strength to the argument that R. exulans was Near Oceania and thus no major discontinu- dispersed primarily, if not exclusively, by ity between Near and Remote Oceanic R. ex- Lapita peoples in Near Oceania. ulans populations, it now appears that we may In addition to trapping R. exulans, we also have more than one R. exulans lineage associ- encountered a number of other Rattus species ated with Lapita dispersal. that were introduced to the islands we visited. The distribution of R. exulans Haplogroup These introduced species not only impact II does appear to be associated with islands the distribution of R. exulans, but also New associated with early Lapita colonization in Guinea native species. We have no idea how Near Oceania. Whether these rats were in long R. tanezumi (sometimes referred to as the region before the arrival of Lapita is not Rattus rattus mansorius) has been in New yet clear. Haplogroup III, however, was Guinea or on the islands of Near Oceania, clearly the main lineage taken out to most of but it is clearly not a native species to New Remote Oceania. Because there is no way of Guinea (Musser and Carleton 2005). Flan- identifying morphologically which rats carry nery (1995) suggested that it is a relatively re- which mtDNA lineage, intentional human se- cent introduction to the region. It is present lection is unlikely to have determined which in much of western Micronesia, where it pre- haplotypes were transported. The distribu- dates the appearance of R. exulans in the ar- tion does suggest that the source population chaeological record (Wickler 2004). for the settlement of Remote Oceania must The fact that we trapped no New Guinea have come from a location where Hap- native rats on any of the islands was surpris- logroup III was present, if not the predomi- ing. Flannery (1995) recorded that R. praetor nant lineage. Does this then suggest two is/was present on New Ireland, Manus, and Lapita dispersals, with the first arrivals in Blup Blup, an island located quite close to Near Oceania carrying Haplogroup II rats, Koil within the Schouten Group. Another dispersing perhaps as far as the Reef/Santa New Guinea rodent, Melomys rufescens, be- Cruz islands and possibly northern Vanuatu, lieved to be native to New Ireland, is also and later arrivals heading out farther east recorded on Blup Blup. Flannery (1995:145) through Remote Oceania carrying Hap- stated that Melomys rufescens is ‘‘a very com- logroup III R. exulans? Clearly only aDNA mon species, particularly in disturbed habitats analyses of well-dated archaeological rat re- throughout its range. It is often found in mains will solve this question. houses, particularly where introduced murids Unfortunately, to date, we do not have are rare or absent, and it can often be seen many archaeological collections in Near Oce- climbing in low vegetation in gardens or ania that include large numbers of rat bones near villages at night.’’ This is exactly where and date between 3,000 and 4,000 B.P. Un- and when we were trapping, so the fact that less we can find rat bones in well-protected we did not encounter any Melomys on New sites such as rock shelters or caves, our Ireland or Koil, and no R. praetor on New chances of being able to obtain DNA from . On the Rat Trail in Near Oceania Matisoo-Smith et al. 473 those 3,000þ year old faunal remains using and Kelly Amanga, Kenneth Thomas, Pastor traditional DNA techniques are not high be- Wilson, and the staff and students at Rongoe cause of DNA degradation in tropical cli- School on Emirau. mates (Robins et al. 2001). However, we are continuing to look for such archaeological Literature Cited samples, and recent developments in DNA sequencing technology may provide the tools Allen, J. 2000. From beach to beach: The to obtain reliable DNA sequences from development of maritime economies in highly degraded samples (Millar et al. 2008). prehistoric Melanesia. Mod. Quat. Res. The results of this study combined with (Southeast Asia) 16:137–176. those of our previous work show the poten- Bellwood, P., and J. P. White. 2005. Domes- tial for using commensal animals to address ticated pigs in eastern Indonesia. Science issues of Pacific prehistory. Given that at least (Washington, D.C.) 309:381. two other rat species were introduced and Bulmer, S. 1982. Human ecology and cultural transported through Near and Remote Oce- variation in prehistoric New Guinea. ania during prehistory, analyses of R. praetor Pages 169–206 in J. L. Gressitt, ed. Bio- and R. tanezumi can perhaps provide further geography and ecology of New Guinea. evidence of human mobility and migration D. R. Junk, The Hague. in the region. In addition, analyses of the ———. 2001. Lapita dogs and singing dogs other commensal animals, including not only and the history of the dog in New Guinea. those associated with Lapita expansion, but Pages 183–201 in G. R. Clark, A. J. Ander- also those moved to islands by earlier inhabi- son, and T. Vunidilo, eds. The archaeol- tants of Near Oceania, could also be valuable. ogy of Lapita dispersal in Oceania: Papers All of these future studies of commensal from the Fourth Lapita Conference, June plants and animals, however, need to be con- 2000, Canberra, Australia. Terra Australis ducted carefully with full cooperation and 17. Pandanus Books, Australian National collaboration between archaeologists, faunal University, Canberra. experts, and biologists to fully interpret and Cachola-Abad, K. C. 1993. Evaluating the make the most of the resulting data. In addi- orthodox dual settlement model for the tion, we need good chronological control Hawaiian Islands: An analysis of artefact based on analyses of well-dated archaeologi- distribution and Hawaiian oral traditions. cal remains as well as extant samples to fully Pages 13–32 in M. W. Graves and R. C. understand the biogeography and genetic Green, eds. The evolution and organisa- variation of commensal animals and how that tion of prehistoric society in Polynesia. relates to past human behavior. New Zealand Archaeological Monograph 19. Auckland. acknowledgments Cunningham, D. M., and P. J. Moors. 1983. A Guide to the identification and col- We acknowledge and thank the following lection of New Zealand rodents. New people who assisted with collections and Zealand Wildlife Service Occasional Pub- made the fieldwork possible, not to mention lication No. 4. Department of Internal Af- enjoyable: Herman Mandui, Geoff Hope, fairs, Wellington, New Zealand. Andy Fairbairn, Jim Specht; Chief Matthew Flannery, T. 1995. Mammals of the South- and the people of Wanap village on Koil Is- West Pacific and Moluccan Islands. Cor- land; Jimmy Peter, Andy Holding, and Lihir nell University Press, Ithaca, New York. Gold on Lihir; Ken, King, Elsie, Jessie, Se- Friedlaender, J. S., F. Gentz, K. Green, and lina, Dennis, and all at Andi’s Guest House D. A. Merriwether. 2002. A cautionary on Tatau; Bernard Minol and family, Father tale on ancient migration detection: Mito- Paul Mohe, and the congregation of Loren- chondrial DNA variation in Santa Cruz gau Catholic Church on Manus; Shawn Yip, Islands, Solomon Islands. Hum. Biol. John Aini, and Jim Ridges on New Ireland; 74:453–471. . 474 PACIFIC SCIENCE October 2009

Gadaleta, G., G. Pepe, G. De Candia, C. P. J. Piper, R. J. Rabett, P. Shooter, G. Quagliariello, E. Sbisa`, and C. Saccone. Van den Bergh, E. West, S. Wickler, and 1989. The complete nucleotide sequence J. Yuan. 2007. Phylogeny and ancient of the Rattus norvegicus mitochondrial ge- DNA of Sus provides insights into Neo- nome: Cryptic signals revealed by compar- lithic expansion in Island Southeast Asia ative analysis between vertebrates. J. Mol. and Oceania. Proc. Natl. Acad. Sci. Evol. 28:497–516. U.S.A. 104:4834–4839. Gorecki, P., M. Mabin, and J. Campbell. Lee, T., J. B. Burch, T. Coote, B. Fontaine, 1991. Archaeology and geomorphology of O. Gargominy, P. Pearce-Kelly, and the Vanimo Coast, Papua New Guinea: D. O. Foighil. 2007. Prehistoric inter- Preliminary results. Archaeol. Oceania archipelago trading of Polynesian tree 26:119–122. snails leaves a conservation legacy. Proc. Grayson, D. K. 2001. The archaeological rec- R. Soc. Lond. B Biol. Sci. 274:2907–2914. ord of human impacts on animal popula- Marshall, B., and J. Allen. 1991. Excavations tions. J. World Prehist. 15:1–68. at Panakiwuk Cave, New Ireland. Pages Green, R. C. 1966. Linguistic subgrouping 59–91 in J. Allen and C. Gosden, eds. within Polynesia: The implications for Report of the Lapita Homeland Project. prehistoric settlement. J. Polynesian Soc. Occasional Papers in Prehistory 20. De- 75:6–38. partment of Prehistory, Research School ———. 1997. Linguistic, biological and cul- of Pacific Studies, Canberra. tural origins of the initial inhabitants of Matisoo-Smith, E. 1994. The human coloni- Remote Oceania. N. Z. J. Archaeol. 17:5– sation of Polynesia. A novel approach: 27. Genetic analyses of the Polynesian Rat ———. 2000. Lapita and the cultural model (Rattus exulans). J. Polynesian Soc. for intrusion, integration and innovation. 103:75–87. Pages 372–392 in A. Anderson and T. Matisoo-Smith, E., and J. S. Allen. 2001. Murray, eds. Australian archaeologist— Name that rat: Molecular and morpholog- Collected papers in honour of Jim Allen. ical markers in Pacific rodent remains. Int. Coombs Academic Publishing, Canberra. J. Osteoarchaeol. 11:34–42. Irwin, G. J. 1992. The prehistoric exploration Matisoo-Smith, E., J. S. Allen, T. N. Lade- and colonisation of the Pacific. Cambridge foged, R. M. Roberts, and D. M. Lambert. University Press, Cambridge. 1997. Ancient DNA from Polynesian Rats: Kirch, P. V. 1982. The impact of the prehis- Extraction, amplification and sequence toric Polynesians on the Hawaiian ecosys- from single small bones. Electrophoresis tem. Pac. Sci. 36:1–14. 18:1534–1537. ———. 1984. The evolution of Polynesian Matisoo-Smith, E., K. A. Horsburgh, J. H. chiefdoms. Cambridge University Press, Robins, and A. J. Anderson. 2001. Genetic New York. variation in archaeological Rattus exulans ———. 1986. Rethinking East Polynesian remains from Emily Bay, Norfolk Island. prehistory. J. Polynesian Soc. 95:9–40. Pages 81–84 in A. Anderson and P. White, ———. 2000. On the road of the winds: An eds. The prehistoric archaeology of Nor- archaeological history of the Pacific islands folk Island, Southwest Pacific. Rec. Aust. before European contact. University of Mus., Suppl. 27. California Press, Berkeley. Matisoo-Smith, E., R. M. Roberts, J. S. Allen, Larson, G., T. Cucchi, M. Fujita, E. Matisoo- G. J. Irwin, D. Penny, and D. M. Lambert. Smith, J. Robins, A. Anderson, B. Rolett, 1998. Patterns of human colonisation in M. Spriggs, G. Dolman, T.-H. Kim, Polynesia revealed by mitochondrial DNA N. T. D. Thuy, E. Randi, M. Doherty, from the Polynesian Rat. Proc. Natl. Acad. R. A. Due, R. Bollt, T. Djubiantono, B. Sci. U.S.A. 95:15145–15150. Griffin, M. Intoh, E. Keane, P. Kirch, Matisoo-Smith, E., and J. Robins. 2004. Ori- K.-T. Li, M. Morwood, L. M. Pedrina, gins and dispersals of Pacific peoples: Evi- . On the Rat Trail in Near Oceania Matisoo-Smith et al. 475

dence from mtDNA phylogenies of the M. Goode, G. Ewing, P. McLaren, A. G. Pacific rat. Proc. Natl. Acad. Sci. U.S.A. Rodrigo, S. Lavery, and C. S. Baker. 101:9167–9172. 2003. DNA surveillance: Web-based mo- Matisoo-Smith, E., D. G. Sutton, T. N. La- lecular identification of whales, dolphins, defoged, D. M. Lambert, and J. S. Allen. and porpoises. J. Hered. 94:111–114. 1999. Prehistoric mobility in Polynesia: Ross, H. A., and S. Murugan. 2006. Using MtDNA variation in Rattus exulans from phylogenetic analyses and reference data- the Chatham and Kermadec islands. Asian sets to validate the species identities of Perspect. 38:186–199. cetacean sequences in GenBank. Mol. Mayr, E. 2000. The biological species con- Phylogenet. Evol. 40:866–871. cept. Pages 17–29 in Q. D. Wheeler and Ross, M., and A. Naess. 2007. An oceanic or- R. Meier, eds. Species concepts and phylo- igin for Aiwoo, the language of the Reef genetic theory: A debate. Columbia Uni- Islands? Oceanic Ling. 46:456–498. versity Press, New York. Sheppard, P. J., and R. Walter. 2006. A re- Millar, C. D., L. Huynen, S. Subramanian, E. vised model of Solomon Islands culture Mohandesan, and D. M. Lambert. 2008. history. J. Polynesian Soc. 115:47–76. New developments in ancient genomics. Spriggs, M. J. T. 1997. The island Melane- Trends Ecol. Evol. 23:386–393. sians. Blackwell, Oxford. Motokawa, M. 2004. Geographic variation in Summerhayes, G. R. 2003. The rocky road: cranial features of the Polynesian Rat Rattus The selection and transport of Admiralties exulans (Peale, 1848) (Mammalia: Rodentia: obsidian to Lapita communities. Aust. Ar- Muridae). Raffles Bull. Zool. 52:653–663. chaeol. 57:135–143. Musser, G. G., and M. D. Carleton. 2005. ———. 2007. The rise and transformation of Family Muridae. Pages 894–1531 in D. E. Lapita in the Bismarck Archipelago. Pages Wilson and D. M. Reeder, eds. Mammal 129–172 in S. Chiu and C. Sand, eds. species of the world: A taxonomic and From Southeast Asia to the Pacific: Ar- geographic reference. The Johns Hopkins chaeological perspectives on the Austrone- University Press, Baltimore. sian expansion and the Lapita Cultural Parkinson, R. 1999. Thirty years in the South Complex. Academia Sinica, Taipei. Seas. 2nd ed. Crawford House Press, Tate, G. H. H. 1935. Rodents of the genera Bathurst. (Original in German.) Rattus and Mus from the Pacific islands. Robins, J. H., M. Hingston, E. Matisoo- Bull. Am. Mus. Nat. Hist. 68:145–178. Smith, and H. A. Ross. 2007. Identifying White, J. P., and M.-N. Harris. 1997. Chang- Rattus species using mitochondrial DNA. ing sources: Early Lapita obsidian in the Mol. Ecol. Notes 7:717–729. Bismarck Archipelago. Archaeol. Oceania Robins, J. H., E. Matisoo-Smith, and L. 32:97–107. Furey. 2001. Hit or miss? Factors affecting Wickler, S. 2004. Modelling colonisation and DNA preservation in Pacific archaeologi- migration in Micronesia from a zooarch- cal material. Australasian Connections and aeological perspective. Pages 28–41 in M. New Directions: Proceedings of the 2001 Mondini, S. Mun˜ oz, and S. Wickler, eds. Australasian Archaeometry Conference. Colonisation, migration, and marginal Res. Anthropol. Ling. 5:295–305. areas: A zooarchaeological approach. Ox- Ross, H. A., G. M. Lento, M. L. Dalebout, bow Books, Oxford.