CHAPTER 5 ZOOARCHAEOLOGICAL ANALYSIS OF FAUNAL ASSEMBLAGES N.M. Howard and P. V. Kirch The rockshelter sites at Atiaoa, Kamaka, distilled vinegar and water. After air drying, the v ' ) and Nenega-Iti (Agakauitai Is.), as well clean faunal remains wvere fine sorted and iden- as the beach ridge midden site at tified to the lowest taxonomic categorv possible. Onemea on Taravai Island, all produced Shell identification and taxonomy follows t) assemblages of invertebrate and verte- Salvat and Rives (1975). Whole shells and most brate faunal remains.' In this chapter, larger fragments could be identified to genus and we present the results of a preliminary species, but smaller unidentifiable fragments were analysis of this material; a detailed analysis of designated as "miscellaneous shell." After sort- bird bones from these sites is provided in Chap- ing and identification, all of the specimens for ter 6. We stress that this is a preliminar) analysis each taxon were weighed to determine a total in that available reference collections have Jim- weight per excavation level. Shells were also ited to some degree the identifications possible, counted, with the number of identified speci- especiallt of the vertebrate remains. mens (NISP) determined by counting the total number of specimens present for each mollus- S1\AMP ING AND M\IETH( )DS can taxon in each level.2 The minimum number As noted in Chapter 3, (0).5 and 0.3 mm sieves of individuals (NINI) was estimated b) counting were used during all test excavations to recover specific diagnostic elements for each species. For faunal remains. In the field, faunal remains re- bivalves, the umbo (the apex or beak of the shell) covered at each site were loosely sorted into shell was used as the diagnostic element to determine and bone and bagged by layer and level; con- MINl. The apex (the tip or point) was used for tents of features such as hearths or pits were all gastropods (snails), except for Drupa, Nenlta, bagged individually as well. Each bag was as- ()ypraea, and those of the Cymatiidae family for signed a unique identification number. in the which the intact aperture was used instead. The laborator}; the contents of these field bags were aperture wxas selected as the diagnostic element rinsed with water to remove excess dirt and sand in the determination of MVNJ for these species and cleaned using a solution of equal parts of because the apertures were less fragile than the 107 ZOOARCHAEOLOGICAL ANALYSIS OF FAUNAL ASSEMBLAGES apex an(l therefore less likely to break. However, measure the maximum centrum diameters of the apertures had to be at least 75'Vo intact to be each intact vertebra, and these were recorded in counted as a viable individual representative. For millimeters. all L,tir1is nodatus specimens, the intact columella INVTERTEBRATF RlMALNS (the sQlid central column around whlich the gas- tropod shell spirals) was clhosen to determine B- veight,invertebrate remains constituted the bulk of the faunal materials recovered at all MNI, and for all operculae the centralportion three sites and these were dominated by marine wvas selected (although at least half of the nucleus . to be present to be counted). mollusks. Smaller quantities of echinoderms and had to be present to be counted). crustacea were also found. Several sites also Three shell species were abundant enough t mn tw yelded the shells of terrestrial gastropods; the to be measured for possible size changes between- . stratigraphic units: Gafrar7UMpecti natun-, Ce/lana latter were not food sources but do provide sig- taitensis, and.Turbo setosus as represented by nificant information on local environmental con- tailensis, and Turlho setosusf as represented bvT .1 ditions. operculae. All specimens of these three taxa were measured, provided that they were intact enough ALzARIjNE MoLLusKv to accurately record their size. Dial calipers were Salvat and Rives (1975:64) indicate that the used and the lengths recorded in millimeters. The Gambier archipelago is depauperate in mollus- Ga/i-arnum pectinatum shells were measured trans- can taxa when compared with the Society, verselv to the hinge (umbo). The maximum Tuamotu, and Marquesas islands of Eastern length diameter was measured for both Cellana Polvnesia. The decline in species richness from taitensis and TuIrbo operculae. west to east across French Polynesia reflects both All vertebrate specimens were first sorted increasing distance from the primar) Indo-Pa- into categories of mammal, bird, and fish. After cific source area and local ecololgical conditions, initial sorting to category, certain diagnostic speci- especially the cooler waters of Mangareva (aver- mens were identified to family and when pos- age seawater temperature in August 21.5", com- sible to genus and/or species. Most of the ver- pared with 26" in the Mlarquesas). Moreover, as tebrate material consisted of fishbone, wlhich was Richard (1974) demonstrates in his study of lit- identified with the aid of the comparative refer- toral species on Mangareva Island, the number ence collection of Pacific fishes in the Oceanic of molluscan taxa which are both abundant and Archaeology Laboratory at U. C. Berkeley and suitable for gathering as food is even more re- by consulting Fowler (1955) and Barnett (1978). stricted. Once identified to the lowest taxonomic level Table 5.1 lists the marine molluscan taxa rep- possible, the bone specimens were counted and resented in the faunal assemblages from the three the NISP established. Bone was not weighed nor sites, with notes as to habitat. Several taxa are did we attempt to determine the MINI, given that typical of rockv shores, either volcanic platforms a large number of specimens could not be iden- or cliffs xvithin the surge zone, or coralline tified to a level other than basic class. However, rocks: Cellana taitensis and Nerta plicata with the vertebral centra of both Teleost fish and other nerites are the most typical of this zone. Elasmobranchii (sharks and rays) were measured Other mollusks are tvpicalhv found on the outer as a proxy for fish size, follorwing Reitz and Wing crests of fringing and barrier reefs: Turbo setosus (1999). Assuming that the available samples of and T argyrostornus, Drupa sp., Alorula ura are char- vertebrae come from a cross section of the iden- acteristic of this zone. Still other taxa, such as tified species, this allowvs one to determine ('onus and Cypraea spp., prefer the reef platform. whether size changes occurred across different With the exception of Tr/dacna mnax/mna and the stratigraphic layTers. Dial calipers xvere used to two Cbhama spp., both of which are sessile and 108 ARCHAEOLOGICAL INVESTIGATIONS IN THE MANGAREVA ISLANDS, FRENCH POLYNESIA TABLE 5.1 Molluscan taxa represented in the Mangareva sites. FAMILY GENUS AND SPECIES HABITAT COMMENTS GASTROPODS Patellidae Ce-ono toitens/s Exposed volcanic rocky shores. Prized for food. Turbinidae Turbo setosus Frontal zone of barrier reefs, on Prized for food; used in parts algal crests, surge zone. of Polynesia for fishhook manufacture. Turbo argyrostomus Algal crests of exterior reefs, Considerably larger than T. lagoons. setosus (up to 91 mm diameter). Neritidae Nerito plicato Rocky shores, littoral fringe. Nerito picea Same as N. plicato. Nertia morio Same as N. plicoto. Strombidae Lambis truncato Subtidal, in sand or gravel Large shell obtains up to 170 patches on lagoon floor. mm length. Cypraeidae Cyproea spp. Fringing and barrier reefs; in Several species represented, crevasses and under stones. including Erosaria moneta. Cymatiidae Cymotium spp. Fringing and barrier reefs. Choronio tritonis Deeper water, lagoon. Large shell up to 380 mm; used as a trumpet throughout Polynesia. Muricidae Drupo sp. Reef plafform. Morula uvo Exterior reefs near surge zone. Fasciolariidae Latirus nodatus Reef plafforms. Mitridae Mitra sp. Coral sand, lagoon floors or sandy Large family with many patches. species. Conidae Conus spp. Reef plafforms and sandy Some species highly toxic. patches. BIVALVES Arcidae Arco sp. Lagoon, sandy substrates. Pteriidae Pinctodo margaoitifero Lagoon. Shell provided primary material for fishhooks. Ostreidae Crossostreo cuculloto Rocky substrates. Lucinidae Codaki/ sp. Lagoon sediments, sandy patches. Chamidae Choma imbricoto Rocky substrates. Chamo pocifica Rocky substrates. Tridacnidae Tridocno maximo Reef plafforms. Prized for food. Veneridae Gafrorium pectinatum Sandy substrates, lagoon floors. Tellinidae Scutorcopagio Sandy substrates. scob/noto Tell/no spp. Sandy substrates. Psammobiidae Asoph/s violoseus Sandy substrates. 109 ZOOARCHAEOLOGICAL ANALYSIS OF FAUNAL ASSEMBLAGES require a hlard sutbstrate, most of the bivalv,es sentedl, again from a diversity of habitats. No listcd in Table 5.1 inhabit sandy or lagoon sedi- single species dominates as at Atiaoa, blut the fol- mentarv suLbstrates. Gaqfrarium pectil.atuul7 occurs lowing set of five mollusks accounts for more in quite shallow water where there are sand flats, than 75 o of the assemblage: Cellana taitensis, whereas Phnctada mal,-aiitifrra requires deeper la- Thrbo setosus, Nerita plicata, Latirms nodatus, and goon xvaters. Pinctada tuiaraal;tifera. Agakauitai is a small island Mlost of the taxa listed in Table 5.1 seem to with both exposed rocky, substrates on the south have been gathered for their food value. How- and east coasts (providing excellent habitats for ever, Pinctada malgarifet/a, the pearl oyster shell, Cellana and JNelita), and reef platforms and sandv yields not only edible meat but also large valves flats on the west, in the channel between which were the principal material used to manu- Agakauitai and Agakauiuta on Taravai Island. facture fishhooks. Much of the Pinctada material Most of the Pinctada shell in this rockshelter is in our sites shows signs of being worked for fish- presumably related to fishhook manufacture, as hook manufacture (see also (Chapter 7). the site also yielded a sizeable assemblage of The molluscan assemblage from the Atiaoa Acropora coral files and pearlshell fishhooks (see rockshelter site (1 9()-06-A\TA-1) is tabulated byr Chapter 7). The concentration indices in Nenega- NINI and weight in Table 5.2. Twelve taxa are Iti are similar to that of Layer lI at Atiaoa, with represented, deriving from several different habi- a slightly higher density (C.i. = 5.12 kg/m3) in tats. Howihever, the assemblage is overwhelmingly L ayer IIlA. There is no overall temporal trend dominated by one species, Gafra17riu pectinatum, evident. making up nearly 64') by weight. This domi- The Onemea site (190-12-TAR-6) assem- nance of c;. pectinaturn probably reflects the ex- bla,ges, recovered from two test pits, are tabu- tensive sandy, inter-tidal flats found at Atiaoa Bay lated in Tables 5.4 and 5.5. This site has both the an ideal habitat for this bivalve. In all, 1.9 kg of least taxonomic richness (7 taxa) and the lowest shellfish remains wTere reco)veredl from the single densityT values of the three sites studied. There I m2 test pit, but the density of mollusks varies are also considerable differences between the considerably b) stratigraphic layer. In Table 5.2 assemblages recovered from the two test pits. In we gQive concentration indices (C.I.) expressed as TP-1, Turbo setosus and Lairnis trunzcata dominate kilograms of shell midden per cubic meter (kg/ the assemblage, whereas in TP-2 most of the in3). From these it can be seen that the density weight is made up byT Ce/lana taitensis and Pinctada of shellfish rises steadily throughout the deposit. margauitifera. Whether these differences reflect Layer 1, with a C.1. of 11.17 kg/mi3, has a density distinct activity areas within the site, or temporal more than four times greater than that of Layer shifts, is not clear (the TP- 1 deposits havre not II. This increased densityr could be the result of yTet been dated). The Turbo shells were probablyT several different factors, such as increased inten- obtained from the barrier reef lying to the west sityT of shellfish exploitation in later prehistory of Onemea BayT, whereas the (C/llana limpets or simplyT a higher rate o)f midden dumping or could have been readilyT collected on the volca- utilization within the shelter. The relative com- nc rock platforms found to either side of the position of the marine molluscan assemblages bayT. from the three sites, plotted byT major habitat Intense collecting pressures by human popu- zones, is shown graphically in Figure 5.1. lations have the potential to affect the popula- The molluscan assemblage from Nenea-iti tion structures of marine mollusks (e.g., Kay and rockshelter site (I 9()-()2-AGA-3) on Agakauitai Iagruder 1977), resulting in a reduction of older Island is tabulated in Table 5.3. This is the most (and larger) individuals relative to younger (and diverse assemblage analyzed with 19 taxa repre- smaller) individuals. The ethnohistoric literature 1 10 ARCHAEOLOGICAL INVESTIGATIONS IN THE MANGAREVA ISLANDS, FRENCH POLYNESIA TABLE 5.2 MOIIUSCan faUna frOm At;aOa rOCkShelter (S;te 190-06-ATA-1). I TaXOn I LaYer I I LaYer 11 | LaYer IIIA | TOtCl 1 % TOtOl | We;ght We;ght I MNI I Wel9ht I MNI I Wei9ht MNI I Wei9ht I (9) I I | Turbosetosus I 5 | 10.5 | 5 | 60.5 I I 71 | 3., | Turbo argyrostomus | 2 | 1 04. 1 I I | | 1 04. 1 1 5 Z 1 Turbo OPerCUIae 6 25.5 7 31 .3 56.8 2.9 | Neritaplicata I 14 | 7.79 | 18 | 14.5 2 | 0.9 | 23.2 | 1.Z X Cypraea SPP. 3 10.0 2 15.9 25.9 1 .3 Cymatium SPP. 5 24.0 2 5.3 29.3 1.5- Drupa sp. 1 2.3 4 6.5 8.8 0.4 Morula uva 0.6 0.6 <0.0 Pinctuda 2 97.0 3 66. 1 1 .6 1 64.7 8.5 margaritifera Crassostrea 2 1.6 1.6 <0.0 cucullata Gafrarium 152 775.1 87 43 1 .9 5 19.8 1 226.8 63.7 pectinatum Scutarcopogia 2 8.9 8.9 0.5 scobinata t Tel/inospp. l 5 | 19.5 | 9 | 77 1 | 3.9 | 100.4 | 5Z1 M;SCeliOneOUS 1 7 30.4 22 67.9 4 5.7 1 04 5.4 | TOtOl | 224 | 1.117.3 | 159 | 776.9 13 | 31.9 | 1.926.1 | | [ C.l.kg/m3 I | 11.17 I | 259 | 064 | I 2 fc)r WIangareva (see Chapter 2) suggests that in stratigraphic column, and therefore no indica- late prehistornT and protohistorT marine resources tion of significant human pressure on the popu- were extremelnr important in WIangarevan sub- lation structure of this bivalve. In the Nenega- sistence economr. In order to assess whether this Iti rockshelter, the most abundant mollusk is the heavwT reliance on marine resources mi,ht have limpet species Cellanataitensis. TEs species is closel+T had a statisticall+! detectable impact on the popu- related to the Hawaiian Cellana exarata, which has lation structures of gathered shellfish species, we been shown to respond dramaticallrr to oxrer-co]- measured the size ranges of the most abundant lecting through size reductions (lka:,r and WIagruder taxa in our assemblages. 1977). Table 5.7 presents the size data for C. For the Atiaoa rockshelter site, the most taitensis from the Nenega-lti TP-1 sample. While abundant mollusk species present is Gafra1zum there is some fluctuation in mean size through- pectinatBVm. Table 5.6 presents data on mean length out the stratigraphic column, no consistent pat- and standard deviation for five stratigraphic tern of size reduction occurs, and the variations subsamples of G. pertinatvas from the Atiaoa site. are not statisticall) significant given sample sizes. As can be seen, there is no statisticall+T signifi- We therefore conclude that there was no mea- cant difference from the top to bottom of this surable impact on the local population structure 11 1 ZOOARCHAEOLOGICAL ANAL YSIS OF FAUNAL ASSEMBLAGES Atiaoa Rockshelter Nenega-lti Rockshelter | oX ii Onemea TP-1 Onemea TP-2 F~IGURE5.1 Pi chart representing the| __/ t: l composition of molluscan fauna * rocky shore 2 sandy substrates from Mangareva sites i[ > reef crest * deeper lagoonl according to major <>r habitat zones. .ref platform ,otherl of C. ta/tens/s due to human gathering on in the assemblages, but it has not been possible Agakauitai. The O)nemea site assembulage did not as vet to have these identified to) taxon. in the pro)vide sufficientlyT large samples of any mol- Nenegwa -I ti site, 49 NISP of crustacea were col- luscan taxo-n for measurements to be taken. lected, with 21 of these in L,ayer IJIB. At Onemea LHIXOINRXIN AN) C~uioicE,~iTP-2, 48 NlSP of crustacea were found, and many of these tentatively appear to be pincer Small quantities of echino:derm (sea urchin) fragmnents of a land crab, po)ssibly cardisorna sp. spines and tests were found at all sites, but unlike This go,od sized terrestrial crab is common the situation in other E,astern PolyTnesian sites, these throug,hout much o)f .Eastern PolyTnesia, wvhere do not appear in sufficient qluantities to represent it inhabits sandy beach ridg,es and is often taken regular gathering for food. The large slate-p)encil as food. in MIangarevTa, howevTer, Gardisorna is not sea urchiin Heterocentrtu(ts mammnfi//atts was repre- present today (according to informants). If our sented by a fewt spines at Nenega-Iti, but none tentative identification of the specimens from of these showved any sigtns of use as abraders (as Onemea can be confirmed as representing is the case in the MIarquesas or Hawvaii). Cairdisoma, this mayr indicate a case of local ex- Small quantities of crustacea w^ere also found tinction wvithin the prehistoric period. 112 ARCHAEOLOGICAL INVESTIGATIONS IN THE MANGAREVA ISLANDS, FRENCH POLYNESIA TABLE 5.3 Molluscan fauna from Nenega-Iti rockshelter (Site 190-02-AGA-3). Taxon Layer I Layer II Layer IIIA Layer IIIB Total % Wt. MNI Wt. MNI Wt. MNI Wt. MNI Wt. (g) g) (g) (9) (g) ( g) Cellona taitensis 13 14.2 7 6.6 390 440.0 70 67.6 528.4 20.4 Turbo setosus 4 50.3 2 43.5 10 120.0 2 45.2 259.0 10.0 Turbo argyrostomus 2 30.2 30.2 1.2 Turbooperculae 3 9.6 17 101.0 5 54.3 164.9 6.4 Neritoplicato 17 20.7 4 5.3 220 396.4 9 8.7 431.1 16.7 Cyproeaspp. 1 2.1 4 13.4 1 3.2 18.7 0.7 Charonio tritonis 1 28.4 1 4.7 33.1 1.3 Druposp. 3 6.4 7 19.4 3 11.0 36.8 1.4 Lotirusnodatus 7 48.2 10 44.2 22 135.6 5 50.3 278.3 10.8 Mitra sp. 3 0.9 0.9 <0.0 Conus spp. 1 1.2 1.2 <0.0 Arco sp. 2 0.6 2 0.6 1.2 <0.0 Pinctodo 1 54.3 1 1.3 4 315.1 2 135.2 505.9 19.6 margoritifero Crossostreo 2 0.9 0.9 <0.0 cuculloto Codoklo Sp. 2 0.8 1 0.3 1 0.2 1 2.9 4.2 0.2 Chomo pocifico 1 20.1 20.1 0.8 Tridocno maximo 1 6.4 6.4 0.2 Gafrorium 2 7.7 16 59.0 1 4.2 70.9 2.7 pectinotum Tellino spp. 2 4.1 7 44.9 2 17.5 66.5 2.6 Asophis violosceus 3 1.0 1.0 <0.0 Miscellaneousshell 14 18.1 8 5.0 21 92.7 6 10.6 126.4 4.9 Total 79 239.9 36 135.2 724 1,793.9 108 417.1 2,586.1 C.I. kg/m3 2.39 1.35 5.12 2.08 TERRESTRIAL GA4sVTR0P0D.S> on the available terrestrial biological inventories Two inter-related biogeographic character- (Cochereau 1974). Thanks to the extensive col- istics of Pacific island faunas are their dishar- lecting efforts of the 1934 MIangarevan Expedi- monic nature with respect to higher-order taxa tion (Cooke 1935), however, followed by, addi- (i.e., absence of many groups) and remarkable tional collecting in 1997 by Philippe Bouchet radiation at the species level. The MIangareva Is- (Bouchet and Abdou 2001, 2003; Abdou and lands appear to have followed this pattern, al- Bouchet 2000), we have some idea of the en- thou>gh the severe degradation of the terrestrial demic land snail fauna which formerly existed environment-and probable extinction of many on the MIangareva higzh islands and islets. This species-makes this more difficult to ascertain fauna included: six endemic species or subspe- 113 ZOOARCHAEOLOGICAL ANALYSIS OF FAUNAL ASSEMBLAGES TABLE 5.4 Molluscan fauna from Onemea, TP-1 (Site 190-12-TAR-6). Taxon Layer IA Layer |I Total %Weight MNI Weight (9) MNI Weight (g) Weight (g) Turbo setosus 6 260.9 2 12.5 273.4 47.1 Turbo 7 37.2 1 11.7 38.9 6.6 operculae Nerita plicata 1 1.8 1 1.0 2.8 0.5 Lambis 1 91.1 1 129.9 221.0 37.4 truncata Cypraea spp. 2 6.1 6.1 1.0 Latirus 3 15.0 15.0 2.5 nodatus Miscellaneous 5 8.1 3 15.1 23.2 3.9 shell Total 25 420.2 8 170.2 590.4 C. kg/M3 1.40 0.61 Table 5.5 Molluscan fauna from Onemea, TP-2 (Site 190-12-TAR-6). Taxon Layer 11 % Weight MNI Weight (g) Cellana taitensis 47 42.0 36.9 Turbo setosus 2 5.3 4.7 Nerita plicata 9 6.0 5.3 Nerita morio 1 0.6 0.5 Cypraea sp. 1 1.3 1.1 Drupa sp. 4 6.5 5.7 Pinctada margaritifero 3 38.2 33.6 Miscellaneous shell 9 13.8 12.1 Total 76 113.7 C.l. kg/m3 0.28 cies of the genus Tubuaja in the family species in the family Assimineidae (Bouchet and Achatinellidae (Kondo 1962); three endemic gen- Abdou 2003). era (Anceyodonta, Rikitea, and Gambiodonta) and at Between the 1934 and 1997 malacological least 24 endemic species in the familyT expeditions, more than 50,00() specimens of Endodontidae (Solem 1976; Abdou and Bouchet Mlangarevan land snails have been assembled in 200); one endemic species in the famil thecollections of the Bernice P. Bishop useum Punctidae (Abdou and Bouchet 200)0); two en- and the MIus%um National d'Histoire Naturelle demic species in the familyr Euconulidae (Abdou and Bouchet 2000:691). What is truly (Bouchet and Abdou 2001); and two endemic striking about these collections is that-with the 114 ARCHAEOLOGICAL INVESTIGATIONS IN THE MANGAREVA ISLANDS, FRENCH POLYNESIA Table 5.6 Mean lengths of Gafrarium pectinatum from the Atiaoa Rockshelter (Site 1 90-06-ATA-1). Layer and Level Mean Standard N Length Deviation Layer 1, Level 1 22.57 3.73 49 Layer 1, Level 2 23.24 2.80 49 Layer 11 Level 1 22.72 3.92 40 LayerII,Level2 21.43 3.10 30 Layer III 22.62 3.68 8 Table 5.7 Mean diameters of Cellana taitensis from the Nenega-Iti Rockshelter (Site 190-06-ATA-1). Layer and Level Mean Standard N Length Deviation Layer 1 Level 1 18.80 2.43 13 Layer IIIA. Level 3 19.44 3.63 36 Layer IIIA, Level 4 22.38 3.15 226 Layer IIA. Level 5 20.07 2.60 30 Layer IIIA, Level 6 19.62 2.96 82 Layer IIIB, Level 7 19.90 3.61 61 Layer IIIB, Level 8 20.44 2.74 8 exception of historically introduced taxa (such and causes of this "environmental crisis," and as Bradaybaena similarii or Subulina octona) -none here the recovery of land snail shells in datable of the endemic taxa are represented b living archaeological contexts ma) be of much value. specimens. With the exception of a very few Given Kirch's prior research on land snails British Mluseum specimens dating to collections in Pacific island archaeological sites (e.g., made bv Lesson in 1842 (Bouchet and Abdou Christensen and 1Iirch 1981), particular attention 2003:169), all of this material consists of sub- was paid during our excavations to the recovery of fossil specimens, much of it obtained from re- snail shells. In all, 116 specimens were recovered cent sedimentary deposits. As Solem noted of from four site contexts, as enumerated in Table the 1934 Mangarevan Expedition collections, "no 5.8. These represent eight species in six families, living material of endodontids was obtained, but including both endemic and introduced taxa. specimens proved to be quite abundant in sev- Several taxa are illustrated in Figure 5.2. eral cave deposits or road cuts" (1983:280; see Four endemic species are present in the as- also Kondo and Clench 1952:18). All of the en- semblages. The most frequent, present at three demic species of terrestrial gastropods formerly sites, is Omphal/tropi.s matgarita, an endemic present in the Mangareva Islands are thus thought assimineid formerly distributed throughout the is- to be extinct, the result of an "environmental lands and exhibiting remarkablemicrogeogtraphical crisis that has affected the native land snail fauna variation, as shown byT Bouchet and Abdou (200)3, of this island group" (Bouchet and Abdou fig. 3). This species was particularly common in 2003:169). The question arises as to the timing the Nenega-Iti and Atiaoa rockshelters and is also 115 ZOOARCHAEOLOGICAL ANALYSIS OF FAUNAL ASSEMBLAGES TABLE 5.8 Distribution of terrestrial gastropods in Mangarevan sites. Family Genus/species ATA-1 TAR-6 AGA-3 GAE-1 _______________ ~~~~ ~~~~(TP-2) _ _ _ _ Tornatellinidae Lame//idea oblonga 7 Endodontidae Gambiodonta cf. grandis 4 13 | Minidontasp.? 1 Punctidae Punctum sp. 1 Subulinidae Allopeas gracile 26 2 Subulina octona 1 Bradybaenidae Bradybaena similaris 1 Assimineidae Omphalotropis margarita 10 46 4 present in the erosional sediments at Gaeata (site do not occur in the uppermost levels of either GAE-3). In Nenega-Iti, more than 14 individu- rockshelter. This might suggest that the forest habi- als were found at the contact between Layers IIIB tat preferred by this large endemic snail was disap- and IV, probably representing the original land- pearing in the later prehistoric period. scape surface prior to human occupation. In the The onlT other endemic taxa recovered are a Atiaoa site, it was present in both Layers I and single specimen each of an endodontid tenta- 11. The fact that this species persists throughout tively identified as a species of Alinidonta, and of the stratigraphic sequences in these two a punctid tentatively identified as belonging to rockshelter sites and is present at Gaeata, which the genus Punctum. Abdou and Bouchet (2000) is 14C dated to cal ,\.D. 1650-1670, 1770-1800, describe an endemic Punctum mokotoense, from suggests that it persisted throughout much of Mlangareva Island, but our specimen comes from the period of human occupation in the Agakauitai Island. Mlangareva Islands. The introduced land snails are also of con- Also present in the Atiaoa and Nenega-Iti siderable interest. A number of anthropophilic sites is an unusually large endodontid snail, ten- snails are known to have been transported be- tatively identified as Gambiodontagrandis (Solem tween islands and archipelagoes by the 1976: 441-44, fig. 189), shown here in Figure 5.2 Poly nesians and other Pacific peoples, probably a and b. Solem reports this unusually large adhering to crop plants or in soil with crop plants endodontid as being present (based on the 1934 during inter-island voyages (Christensen and MIangarevan Expedition collections) on only Kirch 1981; Kirch 1984:136-37). Two such an- Aukena and Agakauitai islands, but it (or a closely thropophilic taxa are present in our assemblages: related species, or subspecies) must also have Lamellidea oblouga and Allopeasgra/le. The former formerlv existed on Mlangareva Island, based on has a geographic distribution over much of the the material from Atiaoa. The specimens from central eastern Pacific, and Cooke and Kondo Nenega-iti have diameters consistent with the state that: "There is little doubt that the wide range given by Solem (average 12.3 mm). How- distribution of this species is due, mainly, to the ever, a specimen from the base of Layer II in frequent voyrages of the Polynesians, who trans- the Atiaoa rockshelter exceeds this size consid- ported food plants on their travels, especiallyr erabl}; with a diameter of 16.5 mm. The specimens between islands onlyr a few hundred miles apart" of Gamnbiodonta cf. gr^andis are more heavilyr concen- (1960:201, fig. 85). This snail was found onlyr at trated in the lower levels of the Nenega-iti site and the GAE- 1 site, where its presence suggests that 116 ARCHAEOLOGICAL INVESTIGATIONS IN THE MANGAREVA ISLANDS, FRENCH POLYNESIA FIGURE 5.2 Terrestrial gastropods from Mangareva sites: a, b, large endodontids, Gambiodonta cf. grandis (a, Nenega-Iti; b, Atiaoa); c, large subulinid, Subulina octona (Nenega-Iti); d, Allopeas gracile (Onemea site, TP-2); e, assimineid, Omphalotropis margarita (Atiaoa). All scale bars = 5mm. d W e-...Z.... < --t000:0000;-- t t | l~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~...... ......~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~~~. i. .... ...... ......................................... ....:--0-t:-; D:0: 2 _ : X 0 0 .:: 0 0 . f~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.... . 2!41 %~~~~~~~~~~~~~~~~~~~~~~~~~~~~ the local environment had already becen modi- a sin,gle individual and come from the uppermost fled for a,gricultural purposes, as the snail is found stratigraphic co)ntexts at Nenega Iti and O)nemea, primarily in asso)ciation with economic plants. thus dating to the po)st-contact period. Both taxa MSore interesting, perhaps, is Ailopear varaile have been widely dispersed as a result of Euro- (formerly named lamellaxisgraci/is), a small snail pean commerce and the spread o)f plants and in the Suboulinidlae also knoxNvn to have been soil in the past two hundred yTears- widelyT transported byT Pacific islanders in prehis- toryT (Christensen andl Kirch 1981). This snail is VERTEBRATE REMIAINS also present at GlAB-I, but more significanth} is Vertebrate remains havTe been anal zed from relatively abundant at the base of the TP-2 de- the Nenega-Iti rockshelter site and Agakauitai posits in the O)nemea site, extending well down Island, and the Onemea site on Taravai Island; into the basal Layer IIl, where it wi&as recox ered unfortunatelyT, the vrertebrate remains from Atiaoa in direct association with the assemblarc of indig- rockshelter were lost in transit to the UniversityT enous bird bones. As a human-introduced species, of Florida in 200(2. The v ertebrate assemblag:e the presence of A//opea,rac/rie in direct association from Nenega-Iti is enumerated bx bJasic faunal with the now extinct or extirpated birds indirectly categories in Tabule 5.9, and that from the two poinlts to the presence of humans in the immedi- Onemea site test pits in Table 5.10. In both sites, ate vicinity of the Onemea site at this earls time fish bones make up the majoritx of the material, period (appro)ximately cal AX.D. 1000)). althougth TP-2 at Onemea also yielded a signifi- Two other introduced species, S'ubu/ma octona cant quantityT of bird bones from the lowest strati- and Bradybaena simni/aris, are represented each byT graphic levels. 117 ZOOARCHAEOLOGICAL ANALYSIS OF FAUNAL ASSEMBLAGES TABLE 5.9 Vertebrate remains (NISP) from Nenega-Iti rockshelter (Site 190-02-AGA-3). Faunal Layer I Layer 11 Layer IIIA Layer IIIB Total NISP % Total Category Sus scrofa 1 1 <0.0 Medium 4 1 5 0.2 mammal Rattus exulans 38 3 40 1 2 93 4.2 Rattus sp. 2 4 6 0.3 Bird 3 10 13 0.6 Fish 201 243 1,116 527 2,087 94.6 Totals 245 251 1 1,154 549 2,205 TABLE 5.10 Vertebrate remains (NISP) from the Onemea site (Site 190-12-TAR-6). Faunal Category Test Pit 1 Test Pit 2 Layer I Layer Total % Total Layer |1 Layer Ill Total % Total 11 NISP NISP Homo sapiens 6 6 0.7 Medium mammal 2 2 0.4 Raffus exulans 5 5 1.1 Bird 21 132 153 33.8 Fish 551 252 803 99.3 177 115 292 64.6 Totals 557 252 809 205 247 452 AL4ALVIALS from the two sites were identified as "medium Ethnographicallh, the onlyT mammals known mammal" and most likelyT also represent either to be present in Mlangareva were the rat and the pigs or dogs. The low frequency of pig in our pig, but the latter had become extinct prior to sites thus matches that of Green's earlier exca- European contact (Hiroa 1938a:194-95). Green vations, and indicates that while pigs were for- and Weisler (2004), reporting on the faunal re- merly present, they were never abundant. mains recovered from Green's 1959 excavations The bones of the Pacific rat (Rattus exulans) (see Chapter 1), indicate that dog (Canisfamiliais), were quite common in the Nenega-Iti rockshelter pig (Sus scrofa), and rat (Rattus exulans) were all but rare at Onemea. At Nenega-Iti, we also found present at several sites, but that the first two had 6 NISP of a larger species of rat in tlhe two upper- indeed gone locaHlv extinct prior to European most levels; this probably represents a European contact. Whereas the rat was fairly common in introduction (possibly R. rattus). Unlike the situ- Green's samples (total of 99 NISP),3 the pig is rep- ation in the Tangatatau rockshelter on Mangaia resented by only 11 bones and the dog by a mere 5 Island, where the numerous rat bones showed bones (these from a total of 13,598 NISP). evidence of burning and chewing (Kirch et al. We recovered only a single unambiguous 1995), none of the rat bsones from Nenega-iti specimen of Suis scrofa, a premolar tooth from displayed such taphonomic characteristics. This LayTer II of the Nenega-Iti site. However, a few supports Hiroa's contention that rats were not other fragmnentaryT pieces of post-cranial bone eaten in MIangareva (1 938a: 194), which is some- 118 ARCHAEOLOGICAL INVESTIGATIONS IN THE MANGAREVA ISLANDS, FRENCH POLYNESIA what surprising considering the abundant indi- pharTngeal grinding apparatus), or of distinctive cations of terrestrial food stress in late prehis- spines (as with the Balistidae) or tangs tory. It may be, however, that the Mangarevan (Acanthuridae). We consider this a preliminar) marine resources were sufficient to provide for analy,sis onl1, and doubtless additional taxa will protein needs, and that food stress was primarily be recognized when the collection is analyzed in the realm of terrestrial carbohydrates. by, a zooarchaeologist specializing in Pacific Six NISP of Hotmo sapiens were also recov- fishes. ered from the uppermost stratigraphic level of Table 5.11 lists the identified fish remains TP-I at the Onemea site. It is uncertain whether from the Nenega-Iti site. Mlost prevalent are these derived from a disturbed burial context, scarids (parrotfish), probably representing more or represent food remains. than one genus, followed by balistids (trigger- BIRDS fish). Also present in large quantities are serranids A total of 166 NISP of bird bones was re- (groupers). Fourmanoir et al. (1974) list 19 spe- cles of Scaridae (in the genera Scarus, Bolbometopon, covered from the tvo sites, pnrmarily at TP-2 of and Hipposcarus), six species of Bastidae, and Onemea. No bones of the domestic jungle fowl .. 11 species of Serranidae as being present in (Gallussgallus) were included in this sample, al- .lnaea Th ihfeunyo hs aai though Green and Weisler (2004:36) report a Mnreva.pTes hig frequency theseotaxaein total of four NISP chicken bones from Green's these fishes in the inshore and benthic habitats 1959 excavations. Our sample consists predomi- of the Mangarevan marine environment. Sev- nantly of several species of sea birds, although eral other taxa of inshore reef fishes are also an extinct or extirpated pigeon is also repre- present, including moray eels (Muraenidae), con- sented. It iS significant that these bird bones de- vic tag Aathrde prbb. inldn at rived from the deepest stratigraphlc contexts in both Nenega-Iti and Onemea (especially the least one species of Naso), wrasses (Labridae), Layer IIl deposit in TP-2 at Onemea). A full re- emperors (Lethriridae), and the spiny puffer (Diodo;; port on the bird bones is provided in Chapter 6 ~!sttix). Only one specimen of a pelagic fish was represented, a fragment of dentarT tentatively iden- byr Worth y and TennITSon, who undertook the .t of this collection. tified as Acanthocybium solandi, the wahoo. Sharks identification of this collection. and/or rays were represented by some small teeth FISH and byT the distinctive vertebrae. As was the case with Green's 1959 faunal The fish bone assemblage from the Onemea samples (Green and Weisler 2004), the majority site is both smaller and less diverse than that from of bone from our sites consists of the bones of Nenega-Iti (fable 5.12). The samples from the teleost fishes or elasmobranches (rays and two test pits are also very different in composi- sharks). Fullyr 94()/o of the Nenega-Iti bone tion (as were the molluscan assemblages). The sample and 640 o of that from Onemea consist TP-1 sample is heavily dominated by, parrotfish, of fish bone. MIuch of this material consists of which are indeed abundant in the inshore waters fragmentaryT cranial bones, along with abundant along the western coastline of Taravai Island. In post-cranial spines and vertebrae which we have TP-2, the sample is dominated by elasmobranch not attempted to identify to a lower taxonomic teeth and vertebrae. Mluch of this material de- level. However, using reference collections and rives from the Layer III deposit also containing published sources available to us, we have been a higth density of bird bones. The 98 elasmo- able to identifyT a proportion of the assemblage branch vertebrae and 23 teeth from LayTer III in to family level. MIost of the identified material TP-2 probably derive from a single individual of consists of mouth parts (dentaries, premaxilaries, very small shark.4 119 ZOOARCHAEOLOGICAL ANALYSIS OF FAUNAL ASSEMBLAGES TABLE 5.11 Identified fishbone from Nenega-Iti rockshelter (Site 1 90-02-AGA-3). Taxon Layeri Layeril LayerlilA LayeriliB Total NISP %Total Elasmobranchii 1 1 0.7 vertebrae Lamiformes 2 2 1.4 (shark) teeth I Muraenidae 3 1 4 2.9 Acanthocybiidae 1 1 0.7 Serranidae 2 4 1 1 7 24 17.5 Lethrinidae 1 1 0.7 Labridae 2 2 4 3 11 8.0 Scaridae 2 2 24 23 51 37.2 Acanthuridae 6 6 4.4 Diodontidae 7 2 9 6.6 Balistidae 3 15 8 26 19.0 Ostraciidae 1 1 0.7 Totals 9 12 70 46 137 TABLE 5.12 Identified fishbone from the Onemea site (Site 190-12-TAR-6). Taxon Test Pit 1 Test Pit 2 Layer I Layer II Total %Total Layer I| Layer Ill Total %Total NISP NISP Elasmobranchii 1 2 1 2 16.9 21 98 119 78.8 vertebrae Lamniformes 2 23 25 16.5 teeth Serranidae 2 4 6 8.4 1 1 0.7 Lethrinidae 3 3 4.2 Labridae 1 1 2 2.8 2 2 1.3 Scaridae 32 14 46 64.8 2 1 3 2.0 Diodontidae 1 1 1.4 Balistidae 1 1 1.4 1 1 0.7 Totals 51 20 71 29 122 151 In order to assess whether there were an) being taken.) Results are given in Table 5.13. Al- significant changes in the sizes of fish being taken though there appears to be a slight increase in the by the inhabitants of the Nenega-Iti and Onemea size of fish in Layer I of the Nenega-Iti site rela- sites over time, w^e measured the vertebral cen- tive to the lower layers, the sample size is small tra from these assemblages. (Eollowing Reitz and and statisticallyT not sig>nificant. There is no sta- Wing [1999], this assumes that the samples of ver- tisticallvr significant difference betwTeen the two tebrae provide a representative cross section of fish layers at Onemea TP- 1. There is, however, a sta- 120 ARCHAEOLOGICAL INVESTIGATIONS IN THE MANGAREVA ISLANDS, FRENCH POLYNESIA tistically significant difference between the mean L)omestic pigs and dogs were clearlv present vertebral sizes in the two sites, with the fish at in Mlangareva prehistorically but, on the avail- Onemea being generally, larger. This mav reflect able evidence, were nevTer abundant and seem to the marine habitats in proximity to the two sites, have been eliminated prior to European contact. with Onemea having access to deeper waters and Kirch, drawing upon comparative cases includ- the Nenega-Iti site being fronted by a large ex- ing Tikopia, Mlangaia, and Mlangareva, suggests panse of shallow water. In any event, there is no that under conditions of small-island resource indication from this admittedly small sample for limitation, high human population densit, and any size reduction in fish over time at either site, competition for resources, "pigs and humans which is consistent with the evidence from the eventually came into a situation of direct trophic molluscan assemblages. competition" (2000b:438). On small islands where horticulture is a critical component of the sub- TABLE 5.13 Diameters of fish vertebral centra sistence economN, pigs must be kept penned so from Nenega-Iti and Onemea sites. that thev do not devastate gardens and will need to be fed a certain quantity of carbohydrate foods Site and Mean Standard N Siterand Meane Standard Nthat otherwise could be consumed by their hu- Layer Diameter Deviation . man keepers. The fact that more than half of the pig bones recovered by Green in 1959 came Layer I 6.96 3.40 19 from a marae site on Kamaka Island suggests that Layer IilA 5.14 3.33 33 pork was an elite or rituallRy charged food, and Layer IIIB 5.30 2.86 35 not an item of daily consumption. Nonetheless, Onemea TP-1 the ethnographic record makes it clear that pigs Layer I 8.61 2.38 64 had been eliminated from NMangareva by the time of European arrival. Layer II 8.57 1.86 105 At the same time that the faunal record speaks to the extreme limitation of terrestrial CO)NCLUJSIONS protein sources, there is no indication in our data While the faunal samples analyzed here are thus far to support an interpretation of increas- admittedly small in size, they do begin to pro- ing stress on marine resources over time. Samples vide some indication of certain aspects of the of measured mollusks and fish vertebrae from pre-contact Mlangarevan subsistence economy our sites do not display statistically significant and environment. One salient conclusion dernv- size reductions over time, as would be expected ing from our analysis, as well as that of Green if there had been appreciable resource depres- and Weisler (2004), is the overwhelming empha- sion. It seems likely that the extensive reefs and sis on marine as opposed to terrestrial resources. lagoon of Mlangareva-which are vastly greater Only in the deepest lavers at Onemea and than the small area of land, especially arable Nenega-Iti are any appreciable quantities of wild land were more than sufficient to provide pro- terrestrial resources represented, in this case by tein resources to the Mlangarevan population, a number of indigenous seabirds and a native pi- even at its maximum level. Thus the small spa- geon. As was the case in many other Pacific islands tial extent of the high islands was most likely the (Steadman 1989, 1995, 1997), nesting populations key limiting factor to TMangarevan population o)f seabirds in MIangareva were probably decimated growth and density,. within a few decades after the arrival of humans, The zooarchaeological assemblages have also as a consequence of direct predation combined provided important new data on environmental with the effects of human-introduced rats.4 chan$ges during the period of human occupation 97 RADIOCARBON DATING AND SITE CHRONOLOGY Of the 24 dated samples, five were considered Waikato for independent dating. All three dates to have low potential for inbuilt age (<50 years), are ANIS dates, and pretreatment methods were seven were judged to have potential for a me- comparable with acid/alkali washes. dium degree of inbuilt age (>50-10()() years), and The results of AMS dating on the 24 samples seven were considered to have significant po- are provided in Table 4.2. Somewhat surpris- tential for inbuilt age (>100 years). Two inglv-, five samples yielded ages which are re- samples (GANI-7 and GANI-20) were difficult ported in Table 4.2 as 'pMC' or 'percent mod- to evaluate based on their unusual nature (given ern carbon'. These samples are <50 vTears old, as "'unknown" in Table 4.1). These should con- meaning that there xvas a 'greater concentration servatively be considered as potentially contain- of "C in the sample than in the A.D. 1950 refer- ing a high degree of inbuilt age. These estima- ence standard (95'} (-of the "4C content of the tions need be taken into account when inter- National Bureau of Standards Oxalic Acid). preting the calibrated date ranges. For samples Remaining fragments of these samples were re- with "medium," "high," or "unknown" inbuilt turned by Beta Analytic to U.C. Berkeley for age potential, the dates obtained should be seen reexamination after AMNS dating. One of these as providing a terminus ante quenm, or "date be- (GANI-7), a sample that originallv appeared to be fore which," cultural events of site formation semi-carbonized, mayr represent stratigraphicallT occurred. It is also possible, however, that intrusive modern root material. The initial iden- samples with potential medium, high, or un- tification of the other two samples, originally known inbuilt age are actually free of such bias identified as carbonized seed tissue (GANI-2) and that their dates do in fact accurately- reflect and unknown dicotyledonous wood charcoal the calendar period when the wood was burned. (GAN-13), were reconfirmed. The reasons that these two samples returned modern ages remains DATING MIETHODt)S AND RESULTS unclear. Samples GAN1-1 to -15 and -17 to -22 were For the 19 other samples listed in Table 4.2, submitted to Beta Analytic Inc. for pretreatment we have provided the results in terms of the and ANIS radiocarbon dating. The same pretreat- measured 14C age (calculated using the Libbi, ment procedure was applied to all charcoal 14C half-life of 5568 yrs), the ratio (613C) be- samples in order to eliminate contaminants such tween 13C and 1kC (calculated relative to the as carbonates and secondary organic acids, PDB-1 international standard), the "conven- along with modern rootlets. The samples were tional radiocarbon age" (as defined by Stuiver gently crushed and dispersed in de-ionized wa- and Polach 1977), and the calibrated age range ter, folloxved byT hot HCI acid washes and alkali at I standard deviation (68' oprobability). Cali- (NaOH) washes; this was followed by, a final bration follows the calibration database and acid rinse to neutralize the solution prior to dry- methods of Stuiver et al. (1998) and of Talma ing (Darden Hood, pers. comm., Feb. 4, 2003). and Vogel (1993). All charcoal samples were For GANI-22, a sample of bird bone, bone col- calibrated using the atmospheric calibration lagen was extracted with alkali pretreatment. database INTCAL98, while a sample of sea- Three additional samples (GAMI-16a, b, c) bird bone from the Onemea site was calibrated consist of subsamples ("splits") from a single using the marine calibration curve MtARINE98 bulk sediment sample taken from a buried gleved with a AR value of 0 ? 0. We now turn to a clayr horizon in Rikitea Village. Three separate brief discussion of the various dates reported subsamples were sent to Beta Analytic, to the ra- in Table 4.2 in terms of their stratigraphic and diocarbon dating laboratory at the Australian archaeological contexts. The first five localities National University7, and to the UnivTersity of discussed below are situated on Mlan?gareva Is-