CHAPTER 8 EMERGING PATTERNS OF MANGC,AREVAN PREHISTORY P.1'. Kirch and E. Conte The Mlangareva Islands have for too \ long remained a significant lacuna in the emerging picture of Polvnesian culture historv. In spite of Roger Green's pioneering excavations in 1959, little work had been done in subsequent decades. At the interna- tional conference on Eastern Polynesian archae- ology held at MIo'orea in 200() (see Preface), participants signaled their view that Mangarevra was a key area for renewed archaeological in- vestigations. Thanks to significant support from the French Polvnesian Ministry of Culture, our team has been able to take up the problems of MNangarevan archaeology and prehistorn with field seasons in 2001 and 2003. As reported in the preceding chapters of this monograph, a number of key localities and sites have been discovered and investigated, and the excavated materials have been studied in the laboratorv. In this concluding chapter, we assess these re- sults in terms of the four specific objectives laid out at the commencement of our project (see Chapter 1). We then briefly consider several re- search directions that we feel may reward fur- ther investigation in these fascinating islands. RESU!LTS ACHIE\TED To DATE THLI! IAJAN(-A REI LA NA RCHAEOIO)G7GAL RFC( )RD The first of four objectives that we laid out at the commencement of field research in 2001-4"to contribute to the inventorv of ar- chaeological sites"-reflects the relatively un- developed state of MIangarevan archaeology. Initial assessments of MIangarevan archaeologv were not promising (see C hapter 1). Emory (1939) thought that virtually all of the impor- tant sites (especially marae) had been destroyed and that MIangareva did not present a produc- tive area for research. Although he found and dug in rockshelter sites, Emory lacked an ap- preciation of stratigraphy and failed to recog- nize that these sites contained a diachronic record of cultural change. AppWing the more advanced methods of stratigraphic excavation which E. WV. Gifford had introduced to Pacific archaeologav just after World War 11 (Kirch 2000a:27-29), Roger Green in 1959 demon- strated that these rockshelter sites, relatively common in MIangareva, contained well-strati- fied cultural deposits. With his emerging inter- est in Poly nesian settlement patterns, Green also 150 ARCHAEOLOGICAL INVESTIGATIONS IN THE MANGAREVA ISLANDS, FRENCH POLYNESIA mapped a settlement complex at Tokani Bav on Akamaru Island (Green and Weisler 2000, fig. 2) and drew upon Mangarevan ethnohistoric sources to interpret contact-period settlement patterns (Green 1967). Unfortunately, full pub- lication of Green's promising results was de- laved nearly fort)' years, and in the interim the archipelago returned to its status as an archaeo- logical backwater, even as research was advanc- ing in other Eastern Polynesian venues such as the MIarquesas, Societv Islands, Cook Islands, Rapa Nui, Hawaii, and Aotearoa. Our two field seasons to date have con- vinced us that the Mangarevan archaeological record is indeed rich, both in stratified sites with good potential to vield a chronologically con- trolled record of cultural change, and in stone structural remains which are amenable to settle- ment-pattern and landscape analytical ap- proaches. Although it is true that the impres- sive architectural works initiated by Pere Laval and his missionary colleagues resulted in much destruction to the most important indigenous monuments, Emorn (1939:5) was misguided in his claim for "the complete disappearance of all important structures in the Mangarevan group." Indeed, even parts of the foundation of the great Mlarae Te Kehika are extant to this day (see Fig. 3.5). We suspect that the ruins of other marael, such as Mlarae Mlata-o-Tu at Atituiti Raro, may have considerable excavation poten- tial, even if portions of their superstructures were robbed of stone during the missionary zeal to build cathedrals and parish houses. Mlore importantly, our work at Atituiti Ruga and in the Atiaoa Valley have demonstrated that in- tact archaeological landscapes have escaped destruction even on the most heavily populated island of NMangareva itself. Indeed, at Atituiti Ruga the large ATU-1A paepae site adds a new dimension to Mangarevan settlement archaeol- ogy, with the possibilitv that this unique struc- ture represents a class of sites used by the Mangarevan priests for solstitial observation, as described in ethnohistoric sources (Laval 1938). In addition to these twvo localities, reconnais- sance survey has shown that there are exten- sive stone structural complexes on the collu- vial slopes inland of Rikitea Village, and at Rauriki surrounding the Paepae o Uma site. In short, there is much potential for applying a settlement-pattern approach in Mangarevan ar- chaeology, as has been so fruitful in other parts of Eastern Poliynesia. In Appendix B, we provide a checklist of 79 archaeological sites recorded to date in the Mlangareva Islands. Several of these sites incor- porate large numbers of individual stone struc- tural features, such as platforms and terraces. However, it is clear that this list represents onlyr a fraction of the archaeological record still ex- tant in the archipelago. A high priority for con- tinued work in Mlangareva should be to add to this inventorv, through reconnaissance survey as well as detailed mapping in all of the princi- pal valleys and coastal plains of the high islands. We would urge the Service de la Culture et du Patrimoine of the government of French Polynesia to allocate resources towards such continued survey and inventory, work, so that the rich cultural patrimony of Mlangareva can be recorded, studied, and protected. C1O ruT1T1RAI CHRONLO(GY Our second objective also reflects the lack of definitive knowledge on Mangarevan prehis- tor); specificall) the timing of initial Polynesian discovery and settlement of the archipelago. Green and Weisler (2000) had argued that the earliest sites in Mlangareva were vet to be dis- covered; determining the age of earliest Polvnesian settlement in these islands was re- garded as critical to resolving the on-going de- bate about "long" and "short" chronologies in Eastern Polynesian generally (see Chapter 1). MIoreover, if-as some have claimed Mangareva was the "gatewav" to the discovery and settlement of that perennially enigmatic island, Rapa Nui, then determining when people first established a foothold in MIangareva 151 EMERGING PATTERNS OF MANGAREVAN PREHISTORY will also be critical to resolving, the question of xvhen Rapa Nui was settled. Our excavations at the ()nemea site ofn Taravai Island have now provided important newr evidence for the timing of early Polynesian presence in Manficant human-induced impacts to the terrestrial ecosystem of Mlangareva, we have been unable to find evidence for similar effects on the marine ecosystem. Based on the limited samples of marine invertebrates and fish that we have been able to study thus far, there is no sign of size reductions, or of reduction in rep- resentation of large and more highIl prized prev, either of which would potentially signal resource depression as a consequence of predation pres- sure (see, for example, Butler [2001] on the MIangaia case). Indeed, we would hypothesize that given the vast extent of the Mlangarevan reef-lagoon ecosystem in contrast to the avail- able area of arable land, it has alway,s been ter- restrial resources which are human population- limiting in Mlangareva, not marine resources. This hypothesis fits well with the available ethnohistoric evidence (see Chapter 2), which indicates that the sea provided the bulk of pro- tein in the traditional NMangarevan diet, while carbohydrate food sources were limited, with control over agricultural lands and crops being the cause of intense competition and warfare. PRO)BLEMS F(0)R CO)NTINUED INVESTIGATION Having summarized some of the ke) results emerging from our first two field seasons in Mlangareva, as well as formulating several hy- potheses arising from these findings, we turn now to a brief consideration of research ques- tions that we believe are deserving of further investigation. We hope to be able to address these research problems through continued work in the archipelago over the next few years. TI-/i: \A/n,iti ;l () wL1).7-DIS-I I 'IIRACTIO\ As reviewed in Chapter 1, an important ad- vance in Eastern Polvnesian prehistorv over the past two decades has been the demonstration that the earlv communities who emplaced them- selves on the scattered islands and archipela- goes xvere not immediately isolated following initial settlement. Rather, an emerging body of evidence increasingly demonstrates that inter- island and interarchipelago contacts continued after colonization, in some cases for several centuries or even up until the time of Euro- pean contact (as certainly was the case between the Societv Islands and Tuamotu group). On the other hand, some islands (such as Rapa Nui and the Hawaiian chain) clearly did become cut off from contact with other populations after a pe- riod of time. Essential to this emerging picture of long-distance interaction has been the de- velopment of geochemical methods (particu- larly XRF) of characterization and sourcing of basalt artifacts, especially adzes. As discussed above, the limited results obtained by our project have added to the evidence for down-the-line transfer of basalt adzes from the Nlarquesas and Society Islands into Mlangareva, and for direct contact betveen MIangareva and the Pitcairn- Henderson group. As work continues in Iangareva, it will be essential to continue to apply the best analyti- cal tools to the problem of tracing the move- ment of materials into and out of Mlangareva. In addition to work on basalt artifacts, charac- terization and sourcing analytical techniques need to be developed for other classes of mate- rial, such as artifacts of pearlshell. Research on long-distance interaction networks in the west- ern Pacific, particularly during the Lapita pe- iod, has shown that such networks exhibit con- siderable complexitv, as well as changing con- figurations over time (Green and Kirch 1997). Admittedly, tracking such changes through the archaeological record is facilitated in the case- of Lapita by a more diverse set of material types, including pottery. In Eastern Polynesia, 156 ARCHAEOLOGICAL INVESTIGATIONS IN THE MANGAREVA ISLANDS, FRENCH POLYNESIA the range of materials potentially amenable to sourcing analysis is more restricted, and this poses challenges to our ability to reconstruct interaction networks. Nonetheless, this line of research is critical, because understanding to wvhat degree and in what ways the early popula- tions of southeastern Polynesia were in contact with each other, to what extent they were able to share cultural innovations, and why and when they became isolated and cut-off from external contacts, are fundamental to explaining the course of Eastern Polynesian culture history. D)YN 4W\1ICs ovF CCuL7I TRAL CHAI,'\c- To date, most of our effort (as with that of Green before us) has gone into the tedious but essential tasks of defining the Mangarevan ar- chaeological record in time and space. This is as it must be in any area or region where there has been little prior research, and where the basic parameters of local culture history must be es- tablished. Defining basic variability in the ar- chaeological record (site types, settlement dis- tribution, artifact sequences), establishing when Mlangareva was first inhabited by humans, and constructing a well-dated cultural chronology are fundamental tasks that must be accom- plished before other kinds of research questions can ever be posed. Fortunateh; this kind of basic archaeological work is now approaching the point in Mangareva where, xve believe, it is possible to address questions of broader and more theoreti- cal interest. Foremost amon