196 The To'aga Site Table 13.7 Modern Samoan Fishing Methods (after Hill 1986) Line- Diving/ Gill- Throw- Taxa Gleaning fishing Spearing netting netting Holocentridae Serranidae Acanthuridae Mullidae Lutjanidae Muraenidae Letbrinidae Carangidae Scaridae Mugilidae Labridae three-fourths of the marine mammal bones were recovered from Layer HIC of Unit 15. Another concentration of marine mammal bones was associ- ated with the 'ili'ili paving found in Layers I and II of Unit 22. No fruit bat bones (Pteropus sp.) were identified from the site, although the fruit bat is present on Ofu Island today. Of the domesticated animals, only 1 pig (Sus scrofa) tooth and 15 chicken (Gallus gallus) bones were identified. The chicken bones are concentrated in Layer IIIB of Unit 20123. Generally, the non-fish vertebrate bones were very fragmented and difficult to identify to species or even class. Thus, 44 bones were placed in the "general vertebrate" category and 40 in the "general mammal" category. Many of the bones placed in the mammal category may be either pig or dog, but a distinction between the two could not be made. Invertebrate Remains As is true for most Pacific island faunal assem- blages, the invertebrate component dominated the To'aga faunal assemblage. The densest concentra- tions of shell midden were recovered from layers that dated to two periods of time and contained eitlr 'ili'ili paving or features inteipreted as food prepara- tion areas. For the period of 2500-1900 cal B.P., the instances of concentrated midden are dispersed across the site. Unit 28, Layer RC of Transect 5 had a shell density of over 7.8 kg/n3. Along Transect 9, about 27 kg of shell midden with a density of 11.9 kg/M3 were recovered from Layer III of Unit 20123 (table 13.10), along with one-third of the marine turtle remains for the site and half the chicken bones. The upper portion of this layer contained a large earth oven. The extension of Layer IIn into Unit 21 contained over 13 kg of shell midden (8.2 kg/M3). The densest midden in the 1987 Main Trench, in Layers IIB and UC, also dated to this time period (table 13.11). In Transect 5, Layer II of Unit 15129/30, inter- preted as a cookhouse activity area dating to the period 1641-1477 cal B.P., contained nearly 12 kilograms of midden, a density of 7.4 kg/M3 (table 13.12). This layer is contemporaneous with Layer I of Unit 16 of the same transect which was associ- ated with a dispersed distribution of 'ili'iMi gravel and Fauna! Assemblages 197 45- 40- 30-f .-- t E D~ 25- ------- f- 20f- - ------ t 15-t-------t- io-It-. . t 5- n . h- 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 is 19 20 21 22 Tuaa Figure 13.1 Relative frequency of fish taxa from the To'aga site, including Diodontidae. contained the densest concentration of midden in the site (13.4 kg/m3). A midden with 7.9 kg/n3 density was also present in Layer IIC of Unit 28, Transect 5. Only a few families make up the majority of the invertebrate assemblage. Over 76% of the 165 kilograms of identified shell consisted of three families, Turbinidae, Trochidae, and Tridacnidae, with Turbo setosus by far the most abundant species. Besides the shell, more than 14 kilograms of slate- pecil sea urchin (Heterocentronus mamilaus), comprising over 8.5% of the invertebrates, were recovered from the site. Most of the sea urchins were concentrated in Units 20-24 along Transect 9; about half were associated with the earth oven in Layer III, Units 20/23. The rank order of the invertebrate taxa varies little across time and space. Turbinidae is by far the major taxon in the assemblage with Echinoidea, Trochidae, Tridacnidae, Conidae, Cypraeidae, Muricidae, and Neritidae as secondary taxa. The remaining thirty-seven taxa are minor components, contributing less than 1% each to the assemblage. This high diversity may reflect both cultural and environmental factors. Food choice in foraging often reflects the natural abundance and distribution of resources. However, some of the most abundant taxa in the assemblage, such as Turbinidae, Tridacnidae, Echinoidea, and Conidae, were also used as raw material in the manufacture of artifacts. The abundance of these taxa therefore may reflect these dual uses and species may have been selected disproportionately to their natural distributions. A comparison of natural and archaeological inverte- brate distributions through modem marine survey infonmation would be useful in sorting out the influence of environment versus cultural effects on the invertebrate taxa represented archaeologically at To'aga. ANALYSIS OF BULK SAMPLES Archaeologists screen sediments in order to increase comparability within and between sites by systematically sampling the archaeological record. . 1 -- - I I . v , . . - I 1 I I I I I - -- ----------------------------------------------------------------------------------- rl rl ,mm I I I I I I rl ,? tl owu Oman  198 The To'aga Site zu- 18-4 f- 16-4 -----1-- 144- --------i 4-0 I........ . ." . .. .. .. .. ... . . . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . . 6X : :.......... : 1 : I : 1 1 1 " "' ................................................. 4 - I .-=.-. C 2 - I - - I - - I - - I - - I - I - 2 3 4 5 6 7 10 11 12 13 14 15 16 17- 18- 19- 20 21 - 22 Taxa Figure 132 Relative frequency of fish taxa from the Todaga site, excluding Diodontidae. Recovery methods can greatly influence the kinds and the amount of matenal retrieved from excava- tion (Grayson 1984). As with other sampling techniques, the size of the screen used is determined by the research problem. Pacific island archaeology has been oriented toward the recovery of artifacts, such as pottery, that can be readily recovered by 1/4" screens. Unfortunately, the consistent use of 1/4" screens does not always sufficiently sample other classes of archaeological material such as smaller faunal remains. Experiments on the differential recovery of faunal material show that screen size affects the sample size and the number and kind of taxa repre- sented (Thomas 1969; Casteel 1972; Butler 1987). Larger screen sizes bias the sample toward taxa with larger body sizes. The use of smaller screens increases the sample size and retrieves smaller taxa that would otherwise be lost though the larger screens. To determine how our recovery methods influenced the composition of the To'aga faunal assemblage, bulk samples of ca. 5 kilograms (ca. 500 cm3) was taken from Layer IIIA/B of Units 20123 and from Layer II of Unit 30. These layers were chosen because they contained dense midden concentrations. Thus the two samples may not represent the site in general since the recovery rate may be less for areas with a lower midden density. The bulk samples were wet-sieved thrugh window screen in the field to reduce the bulk of the sediment for shipping. They were dry-screened in the laboratory though nested -3 phi (8mm), -2 phi (4mm), -I phi (2mm), and 0 phi (Imm) geological sieves. The contents from each phi size were separated into gross categories (rocks, coral, shell, sea urchin, crab, and bone) and weighed (tables 13.13 and 13.14). The bone was then identified and quantified using NISP. The recovery of bone was affected by screen size more than shell. Almost all the bone was recovered in phi sizes -2 and smaller. The shell recovered by screen sizes less than -3 phi was difficult to identify. Furthennore, the weight of the ci) 01) 12-It 1-- 10-fr - . . 8- 8 9 OWN-01 Fawual Assemblages "4 5 t t N N Stn C4 m I I - I - " 1^ Il t- -4 0- "4 SrI S I I 1, t W 00 "-4 0o 00 0) X 00 V- [ " 4 5 S "4 C l41 C) L "I ! s !! !F- CZ> A4 (.- m o 0, Cu/ z 2 'Cu o a, C- o o - 4) 5- 4) E c O, Cu 0 .00 go I as 0 0% F- a fC Co Cl Cl N Il e4 v N Cl "4 A en e4 "4 40 "4 N " "4 No V" kn -4 -4 0 "_4 00 -4 -4 "-4 "-4 C4 Cl Cl4 (4 Cl1 C4 "-4 _4 N Ft q N _ t N " m _- so WI A F II~l I~" '0 %n C N : - t t "< ' C I I * % t : > i t 00 I I I t t N- I i 1 Cl4 I I 00 S "4 I0 I I I i Al 1 1 "4 1 I 1 C Clq I II 0 9-4 II I II I II I t- II I II I I II I II I II I II II I II I V-4 C! ON 0 en eq 00 "-4 "-4 I I Cl4 ON V-4 11 1 01 1 I en II I II I eq II I II I II a V-4 II I I I I I II I II I i I I 9-4 0% Cl4 -4 i II I II 0 V-4 I 1 7 1 00 t" M V4 "-4 c % x * Cs _F- _ %) Q ) bC - t - -s 14 to 2 2 m 2 0 199 ON "-4 Cl11 0 S N c 1 1 6 I t- e en _ " Cl- _l S "14 Is I N ' "-4 I I I I * I I 0 F- Cl V "4 a "4 0 N8 Nw V) 0 13 cn - a- co asd cd - 0 T-4 . 4. cuoo F* Cl- Cl Cl ,It o i V-4 II I II a i I II I I II I II I I II I I %O II I en i i I o II 200 The Todaga Site Table 13.9 Non-Fish Vertebrate Fauna from the To'aga Site A. 1987 Main Trench, Units 1, 4-9 Layers Taxa HA-1 hA HIB HIC III Total RaNus exulans 9 63 88 28 -- 188 Mamnmal -- -- 2 1 -- 3 Bird 5 13 21 14 1 54 Marine turtle -- 1 2 -- -- 3 Vertebrate -- 1 -- 2 -- 3 TOTAL 14 78 113 45 1 251 B. Transect 5, Units 15/29/30 Layers Taxa II HIA-1 IIIO HIC HID Total Rattus exulans 10 2 22 12 1 47 Mammal 8 -- -- -- -- 8 Marine mammal -- -- -- 18 2 20 Gallus gallus -- -- 2 -- -- 2 Bird 4 -- 1 -- 6 11 Marine turtle 3 -- -- 1 4 8 Vertebrate -- -- -- 11 -- 11 TOTAL 25 2 25 42 13 107 C. Transect 9, Units 20/23 Layers Taxa IIB IIIA HIB HIC IV Total Rattus exulans 5 2 5 20 2 34 Mammal -- 4 -- -- -- 4 Gallus gallus 1 -- 7 -- -- 8 Bird 6 2 4 2 1 15 Marine turtle -- 1 18 3 -- 22 Vertebrate -- 2 -- -- -- 2 TOTAL 12 11 34 25 3 85 Faunal Assemblages 201 Table 13.10 Invertebrate Fauna from Transect 9 (Units 20/23) (weight in grams) Layers Taxa I IUB HLiA HIB HIC IV Total GASTROPODA Patellidae Trochus maculatus Trochus niloticus Trochus spp. Tectus pyramis Turbo crassus Turbo setosus Turbo spp. Turbo operculae Astrea stellare Lunella cinereus Nerita albicilla Nerita picea Nerita plicata Nerita polita Nerita spp. Neritina spp. Tectarius grandinatus Cerithium nodulosum Cerithium spp. Clypeomorus spp. Strombus mutablis Strombus spp. Hipponix spp. Cypraea annulus Cypraea arabica Cypraea caputserpentis Cypraea mappa Cypraea moneta Cypraea tigris Cypraea spp. Natica spp. Tonna spp. Cassidae Cymatium nicobarium Cymnatium spp. Bursa granularis Bursa spp. Drupa ricina Drupa morum Drupa spp. Morula uva Nassa spp. Thais armigera Thais tuberosa (continued next page) 32.5 48.2 263.9 8.3 89.9 1.0 5.2 1.0 7.5 41.3 5.9 13.1 0.3 --- 0.2 0.4 118.4 235.8 642.6 394.6 --- 1.2 0.5 --- 3.9 23.8 4.7 --- 21.1 46.5 --- 276.4 932.9 1734.6 500.2 792.8 1561.0 3980.6 2275.0 114.6 140.8 135.1 125.3 138.1 228.7 1632.0 894.4 2.0 7.5 34.1 8.9 0.4 1.0 9.1 11.8 1.1 --- 0.9 2.8 2.7 0.6 1.2 23.2 29.3 3.3 1.5 47.4 44.9 2.2 4.3 38.0 11.4 --- --- 54.8 13.7 --- 51.5 19.0 0.6 --- --- 1.7 2.4 9.8 --- 21.9 1.3 9.4 37.1 15.1 12.7 --- 11.3 4.7 6.0 2.2 9.0 26.4 14.4 --- --- --- 19.2 7.7 9.2 16.0 68.7 9.0 6.5 89.9 66.3 3.1 29.9 62.0 80.4 --- --- 10.9 23.8 57.6 206.6 141.9 --- --- 0.5 6.3 6.6 30.2 7.5 --- --- 11.0 --- 2.5 3.6 4.2 4.0 1.0 --- 70.1 19.6 --- 3.3 17.1 10.2 40.3 --- 0.6 12.9 5.9 --- --- 6.9 2.4 15.9 25.4 1.4 --- --- --- 1.2 16.9 18.0 67.8 91.6 48.0 8.1 --- 29.6 --- 0.9 1.2 1425.1 1.7 32.4 --- 67.6 58.4 3550.7 100.9 8974.2 0.7 524.8 10.6 2993.7 52.5 --- 22.3 -- 1.1 1.2 7.6 8.0 63.3 5.6 107.9 56.9 0.1 0.1 1.6 1.6 --- 68.5 --- 78.6 4.1 33.0 74.3 --- 22.0 52.0 --- 19.2 --- 101.6 --- 171.7 175.4 10.9 2.0 473.2 0.5 --- 50.6 --- 11.0 2.5 --- 12.8 89.7 70.9 19.4 6.9 0.8 51.8 --- 1.2 16.9 - 238.5 37.7 202 The To'aga Site Table 13.10 (continued) Layers Taxa I IIB LiA fIB HIC IV Total Th'a-is spp. Cantharus undosa Nassarius spp. Vasum ceramicum Conus argus Conus chaldeus Conus eburneus Conus cf. maculifera Conus spp. Bulla spp. Dolabella spp. Melampus spp. Pythia spp. --- 6.3 46.9 43.4 17.1 113.7 1.3 5.1 10.9 16.7 2.2 36.2 --- 0.3 2.9 1.6 6.7 --- 11.5 27.0 12.6 22.4 40.9 14.5 --- 117.4 --- --- --- --- 38.2 --- 38.2 -- - --- 2.3 13.9 --- 16.2 --- --- 1.2 20.3 19.8 --- 41.3 0.5 -- 0.5 4.8 27.4 110.1 147.1 55.2 3A 348.0 --- 0.2 1.4 18.3 3.2 --- 23.1 --- --- 0.8 5.5 5.7 --- 12.0 0.4 1.6 11.9 21.3 28.0 3.6 66.8 --- --- --- 3.3 6.8 11.1 21.2 PELEYCYPODA Mytilidae Isognomon spp. Chama spp. Chlamys spp. Cardiidae Periglypta reticulata Tridacna maxima Quidnipagus palatam Scutarcopagia scobinata Trapezium oblongum Asaphis violaceus Pinna spp. --- 20.0 6.7 23.9 1.9 --- --- 0.7 --- 1.6 --- --- --- 1.1 7.2 --- 15.1 --- --- 1.8 14.9 4.7 --- --- 0.8 81.1 17.9 283.5 234.4 205.4 --- 10.6 18.1 70.3 31.8 2.2 9.0 5.1 75.4 29.8 13.7 --- - --- 3.4 12.5 35.4 2.6 0.3 0.2 52.5 2.3 8.3 16.9 14.9 5.5 --- 822.3 1.8 132.6 --- 121.5 13.7 7.6 61.5 0.5 ECHINOIDEA CRUSTACEA Unidentified TOTAL (g) VOLUME (m3) DENSrrY (kg/m3) 25.0 919.8 3052.9 3742.2 395.5 34.1 8169.5 --- 6.8 30.7 39.3 54.4 1.8 133.0 14.2 48.8 116.4 38.4 67.9 285.7 672.5 2676.5 7281.7 13703.0 5875.7 256.7 30466.1 1.90 0.85 0.40 0.95 0.90 0.30 5.30 0.35 3.15 18.20 14.24 6.53 0.86 5.75 Faunal Assemblages 203 Table 13.11 Invertebrate Fauna from the 1987 Main Trench, Units 1, 4-9 (weight in grams) Layers Taxa IC HA-I HA UB UC HI IV Total GASTROPODA Haliotis ovina Patellidae Trochus maculatus Trochus spp. Turbo crassus Turbo setosus Turbo spp. Turbo operculae Astrea stellare Astrea rhodostoma Lunella cinereus Nerita plicata Nerita polita Nerita spp. Cerithium nodulosum Cerithiidae Strombus cf. maculatus Strombus cf. mutablis Strombus spp. Hipponix conicus Cypraea arabica Cypraea annulus Cypraea caputserpentis Cypraea mappa Cypraea noneta Cypraea cf. tigris Cypraea spp. Policines spp. Cymatium nicobarium Cymatfidae Tonnidae Bursidae Drupa grossolaria Drupa ricina Drupa spp. Nassa spp. Thais armigera Thais spp. Muricidae Cantharus undosus Nassarius spp. Vasum ceramicum Conidae Bulla sp. Melampus fasciatus Melampidae Pythia scarabeus (continued next page) 36.7 5.0 20.0 498.0 60.0 2072.3 --- 215.2 347.3 --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- 1.0 --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- 3.2 0.3 6.7 20.0 2.0 15.0 1.2 11.0 11.6 2.6 1.0 0.4 53.4 5.9 25.2 2.3 1.6 0.9 79.2 1.6 11.5 1.1 0.5 15.0 -- --- 13.9 -- - - 2.0 66.8 145.4 136.9 20.0 65.0 71.3 383.1 1274.5 667.7 973.1 4646.3 2850.6 235.0 313.1 189.5 872.3 1116.1 2045.2 --- --- 4.3 7.6 12.9 0.7 0.9 4.3 7.9 9.0 65.4 --- 20.0 9.2 34.9 9.0 60.0 --- 2.2 3.0 15.3 --- 25.6 0.5 4.1 69.8 276.0 5.0 1.1 3.7 --- 24.1 1.5 4.2 15.2 19.5 6.5 2.7 5.9 1.4 13.7 48.8 70.0 183.0 0.5 9.3 64.4 14.5 48.7 1.4 9.5 --- 0.2 11.0 30.0 5.0 1.1 16.8 43.4 166.5 2.7 65.0 1.4 15.9 8.4 3.3 31.6 255.4 6.8 17.2 16.1 1.7 2.2 1.8 285.0 5.5 6.4 68.1 82.5 1.2 1.2 37.0 1.1 20.5 1.0 44.9 185.8 32.0 18.1 1.0 1.3 2.4 24.3 9.7 0.7 2.0 13.2 4.8 0.9 13.9 3.1 406.3 162.3 2892.8 10788.1 992.9 4403.5 24.8 5.0 13.0 1.9 51.6 139.1 186.5 48.8 149.0 5.0 2.2 45.2 25.6 20.5 10.0 24.3 42.3 0.4 660.2 5.0 19.5 49.3 22.9 64.0 1.7 8.8 10.2 1.4 64.7 2.7 617.2 7.6 6.4 141.8 162.0 13.2 2.8 93.0 1.5 4.6 8.1 4.5 8.7 14.9 3.9 1.5 204 The To'aga Site Table 13.11 (continued) Layers Taxa IC HA-1 HA IHB HC HI IV Total PELECYPODA Anadara sp. --- --- 1.0 --- --- --- 1.0 Arca spp. --- --- --- --- 6.0 --- --- 6.0 Mytilidae 2.0 75.4 69.5 38.5 40.2 1.0 --- 226.6 Isognomon spp. --- --- --- --- 0.7 --- --- 0.7 Chama spp. --- --- 30.0 1.1 --- --- 31.1 Codakia divergens --- --- --- --- 0.9 --- --- 0.9 Gafrarium spp. --- --- --- 65.0 --- --- --- 65.0 Lucinidae --- --- 3.0 7.0 --- --- --- 10.0 Periglypta reticulata --- --- --- 59.2 0.6 --- --- 59.8 Tridacna maxims --- 303.5 22.3 1320.4 1698.8 40.7 --- 3385.7 Hippopus hippopus --- --- --- 121.4 215.0 --- --- 336.4 Quidnipagus palatam --- 0.3 1.9 10.3 13.2 --- --- 25.7 Scutarcopagia scobinata --- 3.3 --- 7.5 10.9 4.8 --- 26.5 Tellinidae --- 8.0 10.0 57.0 75.0 --- --- 150.0 Asaphis violaceus --- 3.1 --- 0.9 --- 1.4 --- 5.4 ECHINOIDEA --- 21.6 32.8 27.5 52.2 17.9 --- 152.0 CRUSTACEA --- 4.8 5.1 28.4 1.9 0.3 5.3 45.8 TOTAL 88.0 3862.7 2949.7 10305.2 9255.7 429.8 51.5 26942.6 VOLUME (i3) 0.10 0.65 0.95 1.85 1.85 1.75 1.05 8.20 DENSITY (kg/m3) 0.88 5.94 3.10 5.57 5.00 0.25 0.05 3.29 shell recovered by the smaller screens added rela- tively little to the shell recovered from the larger screens. The bulk samples show that the size of the vertebrate sample greatly incmases as the screen size decreases. Only one unidentifiable bone was recovered from the -3 phi screen. The -2 phi screen recovered only a fraction of the material recovered from the -I and 0 phi screens (tables 13.15 and 13.16). Comparisons of the density of identifiable fish bone obtained from the bulk samples to that from the excavation unit illustrate the amount of material being lost trugh the 1/4" screens (tables 13.17 and 13.18). While the standardization of the volume to a cubic meter exaggerates the recovery rate for the bulk samples, it shows that a significant amount of bone may be lost through 1/4" screens. The smaller screen sizes also increase the sample's richness through the addition of new taxa. Balistidae and lizard were not recovered in either excavation unit. Along with Balistidae, four other fish families (Ostraciidae, Muraenidae, Carangidae, and Apogonidae) were added to the Unit 30 data through fine screening. Most of these are small- bodied taxa with small diagnostic skeletal elements that are less likely to be recovered by 1/4" screens. Although the To'aga fish sample is the largest in Westem Polynesia, the analysis of the bulk samples shows that sample size, taxonomic richness, and thus sample representativeness can greatly increase through the consistent use of smaller screens and bulk samples. This point is especially relevant for Pacific island archaeology where the vertebrate samples from most sites have been small. Because the representativeness of the 1/4" sample is suspect, the validity of interpretations based on measures of diversity, such as richness (the number of taxa present) and evenness (the distribution of abundance Faunal Assemblages 205 Table 13.12 Invertebrate Fauna from Transect 5, Units 15/29/30 (weight in grams) Layers Taxa H HIA-1 HIUB HID Total GASTROPODA Haliotis spp. --- --- --- 2.7 2.7 Patellidae 0.8 --- 2.2 0.4 3.4 Trochus maculatus 504.5 43.4 40.9 95.4 684.2 Trochus niloticus 42.5 --- --- 1.3 43.8 Tectus pyramis 12.4 --- --- 1.9 14.3 Turbo crassus 988.0 45.0 145.6 151.6 1330.2 Turbo setosus 4213.8 170.0 408.5 872.8 5665.1 Turbo spp. 172.8 13.6 47.4 138.9 372.7 Turbo operculae 2721.2 153.6 163.2 396.5 3434.5 Astrea stellare 21.3 3.2 1.2 57.4 83.1 Lunella cinereus 2.9 --- 0.4 7.8 11.1 Nerita albicilla 4.0 --- --- --- 4.0 Nerita picea 1.9 --- 0.3 8.5 10.7 Nerita plicata 29.1 --- 2.5 13.8 45.4 Nerita polita 36.0 1.1 3.9 82.6 123.6 Nerita spp. 23.7 --- 16.1 51.7 91.5 Cerithium nodulosum 56.8 --- 26.0 17.4 100.2 Cerithium columna --- --- --- 2.6 2.6 Cerithium spp. 5.0 --- 0.2 9.2 14.4 Clypeomorus spp. 19.9 0.5 6.5 8.6 35.5 Strombus mutablis --- --- 5.2 5.2 10.4 Strombus spp. 29.8 2.7 8.4 11.2 52.1 Hipponix conicus 5.4 --- --- 2.9 8.3 Hipponix sp. 7.8 --- 1.0 1.8 10.6 Cypraea annulus 1.7 --- 0.5 0.3 2.5 Cypraea caputserpentis 37.5 1.3 6.7 8.6 54.1 Cypraea eburneus 9.8 --- --- --- 9.8 Cypraea mappa 24.8 0.8 3.8 63.5 92.9 Cypraea mauritania 0.8 --- --- --- 0.8 Cypraea moneta 10.1 --- --- 17.5 27.6 Cypraea tigris --- --- 0.9 --- 0.9 Cypraea spp. 241.6 9.8 22.6 109.7 383.7 Policines spp. --- --- 2.3 0.8 3.1 Naticidae 0.3 --- 1.8 4.6 6.7 Tonna spp. 11.6 --- 0.1 5.4 17.1 Cymatium nicobarium 7.2 --- 7.3 3.5 18.0 Cymatium spp. 6.2 ----- 4.5 10.7 Bursa granularis 42.5 --- 42.5 Bursa spp. 5.8 --- --- --- 5.8 Drupa ricina 34.9 9.7 --- 9.7 54.3 Drupa morum 29.0 --- --- 6.8 35.8 Drupa rubusidaceus 13.6 --- --- --- 13.6 Drupa spp. 8.3 --- 1.1 4.5 13.9 Morula sp. 2.8 --- --- 1.2 4.0 (continued next page) 206 The To'aga Site Table 13.12 (continued) Layers Taxa II HIA-1 IB IIID Total Nassa sp. 2.3 --- 1.0 2.2 5.5 Thais armigera 88.2 7.6 --- 60.2 156.0 Thais tuberosa 111.7 --- 12.9 17.4 142.0 Thais spp. 124.5 10.0 25.4 33.6 193.5 Cantharus undosa 11.0 --- 1.9 21.4 34.3 Nassarius gaudiosis --- --- --- 0.6 0.6 Latirus filamentosa 66.8 --- --- --- 66.8 Vasum ceramicum 42.2 3.6 7.6 53.4 Conus chaldeus --- --- --- 0.5 0.5 Conus eburneus 11.2 --- --- 0.8 12.0 Conus spp. 349.5 0.4 15.0 36.6 401.5 Terebra sp. 5.7 --- --- --- 5.7 Bulla spp. 15.7 0.8 7.2 5.3 29.0 Dolabella spp. 1.4 --- --- --- 1.4 Pythia spp. --- 0.4 3.0 48.0 51.4 Melampus spp. 1.0 0.2 3.5 10.2 14.9 PELECYPODA Arca spp. 1.3 0.7 --- 1.6 3.6 Mytilidae 8.9 1.2 9.7 14.8 34.6 Isognomon spp. 0.3 --- --- 1.4 1.7 Chama spp. 2.7 --- 1.0 26.4 30.1 Codakia spp. --- --- 0.3 --- 0.3 Periglypta reticulata 26.9 --- 11.0 3.6 41.5 Tridacna maxima 1170.1 7.7 215.2 310.8 1703.8 Hippopus hippopus --- --- --- 13.6 13.6 Quidnipagus palatam 28.9 1.6 4.2 3.6 38.3 Scutarcopagia scobinata 51.2 2.4 13.1 10.3 77.0 Trapezium oblongum 4.2 --- 4.2 Asaphis violaceus 50.7 1.5 1.1 7.1 60.4 POLYPLACOPHORA --- --- --- 5.3 5.3 ECHINOIDEA 72.6 5.0 28.7 215.6 321.9 CRUSTACEA 17.7 1.0 11.1 8.8 38.6 Unidentified 198.4 19.4 30.5 162.4 410.7 TOTAL (g) 11853.2 514.6 1326.0 3212.5 16906.3 VOLUME (in3) 1.60 0.10 1.50 1.85 5.05 DENSITY (kg/m3) 7.41 5.15 0.88 1.74 3.35 Fauna! Assemblages 207 Table 13.13 Units 20/23 Bulk Sample Analysis (weight in grams) Screen Size Class >-3 -2 -10 Rock 522.5 30.7 Coral 234.4 34.1 133.6 103.3 Shell 56.6 19.6 1 Crab 8.6 --- Bone 0.4 1.0 2.9 2.5 TOTAL 950.2 104.5 136.5 105.8 Table 13.14 Unit 30 Bulk Sample Analysis (weight in grams) Screen Size Class >-3 -2 -1 0i Rock 138.4 49.7 Coral 102.2 68.5 10.9 30.4 Shell 57.8 24.5 30.4 Bone --- 0.8 3.9 2.5 TOTAL 299.4 143.5 14.8 32.9 values) is also questionable (Gordon 1991). Ideally, a faunal assemblage should reflect the larger target population of the archaeological record, not simply te archaeological recovery techniques used. TEMPORAL TRENDS IN THE TO'AGA ASSEMBLAGE One goal of faunal analysis is the description and interpretation of temporal change in prehistoric subsistence patterns. A pattern of subsistence change which has been described for some Pacific island sites is a quantitative shift from the exploita- tion of wild or naturally occuning resources to a dominance on horticultural production (e.g., the Tikopia case documented by Kirch and Yen [1982] or the Mangaia case described by Steadman and Kirch [1990]). Temporal increases in the frequency of pig, dog, and chicken and decreases in wild vertebrate taxa such as birds, turtle, and marine mammal are taken to indicate this trend. In contrast, the character of the To'aga assemblage changes little over time and does not strongly reflect this kind of shift. Wild taxa are found thmughout the site, and the sample of domesticated animals is too small to draw any firm conclusions. Although much of the wild taxa (especially the birds) are represented in early contexts, most of the chicken bone is also found in those early layers. Thus, there is no clear cut shift from one type of resource use to the other. A corollary of the wild to domesticated fauna hypothesis is the reduction of marine resources with increasing reliance on horticulture (e.g., Janetski 1976, 1980; Kirch 1982, 1988). Resource exploita- tion and environmental degradation by humans are also suggested to contribute to the decline in marine resources, with decreases in the density of shellfish used to support this hypothesis. Invertebrate density varies at To'aga, increasing then decreasing over time (tables 13.10-13.12). However, the use of density measures may be misleading since changes in density may result from other factor, such as changing rates of sedimentation or shifts in settle- ment pattern. In sum, the composition of the To'aga assem- blage changes little over time. The invertebrate assemblage best illustrates this with a few taxa dominating the assemblage across time and space. A similar trend appears to be evident for the fish assemblage as well. At present, the cause of this pattern is not evident. Some possible causes include the exploitation of naturally abundant taxa from a temporally stable environment, a lack of change in subsistence practices, or a combination of both factors. REGIONAL COMPARISONS Comparisons of faunal assemblages from different areas or islands allow for the assessment of 208 The To'aga Site Table 13.15 Vertebrate Taxa Represented in the Bulk Sample From Units 20/23 (NISP) Taxa Balistidae Ostraciidae Serranidae Labridae Holocentridae Diodontidae Muraenidhae Acanthuridae Scaridae Rattus sp. Bird Screen Size -24 -1 0i 7 5 3 4 3 2 1 1 1 1 1 1 2 1 Lizards TOTAL IDENTIFIED UNIDENTIFIED TOTAL 1 18 10 196 11 214 * Not found in regular 0.25 inch screened samples from this excavation unit. regional trends. To'aga may be compared with other faunal assemblages from well-documented sites in Western Samoa (Green and Davidson 1969, 1974; Janetski 1976, 1980; Lohse 1980; Smith 1976), Tonga (Kirch 1988; Poulsen 1987), and Fiji (Best 1981, 1984; Hunt 1980; Kay 1984). First, the issue of data comparability is addressed to determine the quality of the regional data base. Differences in recovery, identification, and quantification tech- niques can seriously affect the comparability of data across assemblages (Butler 1988; Nagaoka 1988). If data sets are not comparable, differences between them may reflect methodological rather than re- gional differences. Once these issues have been addressed, the faunal data are then examined for the invertebrate, fish, and non-fish vertebrate categories. For Western Polynesian faunal assemblages, recovery and quantification techniques vary consid- erably across sites (table 13.19). As was shown in the analysis of the bulk samples from To'aga, screen size influences the kind and the size of the faunal sample. Screen size differences can even change data at a nominal level since smaller screen sizes add taxa. Quarter-inch screens have been the most commonly used although for some sites screen size was not reported. In other cases, several screen sizes were used, but when and where the different sizes were used was not reported. This lack of informa- tion makes it difficult to evaluate the comparability of the data. 2 2 1 19 456 475 Faunal Assemblages 209 Table 13.16 Vertebrate Taxa Represented in the Bulk Sample from Layer II, Unit 30 (NISP) Screen Size -20 -10 0 6 7 8 Labridae Ostraciidae Serranidae Muraenidae Scaridae Lutjanidae Carangidae Apogonidhae Rattus sp. Lizard TOTAL IDENTIFIED UNIDENTIFIED TOTAL 2 4 5 3 2 1 1 1 1 2 3 3 1 4 34 13 18 246 532 22 280 546 * Not found in 0.25 inch screened samples from this excavation unit. NISP was the common technique for quantifica- don of the vertebrate component, except for the Fiji sites where MNI was used. For the invertebrates, weight was used except for Lakeba and Naigani. These differences in quantification may not be as severe as screen size differences since, in many cases, there is little difference between quantification techniques at an ordinal level (Grayson 1984; Jaretski 1980). If the data are considered in terms of the rankings of taxa, it may still be possible to make valid comparisons. In the identification process, differences in the reference collections and the diagnostic elements used can also affect the data. Kirch (1988) and Best (1984) noted that their fish reference collections were inadequate, limiting the number of possible taxa that could be identified. This problem also exists for the To'aga assemblage. Publication of reference collections and the elements used would help evaluate how these factors have biased the data. Some differences among the faunal samples may be due to the range of diagnostic elements used to identify the assemblages, especially for the fish assemblages. Kirch, Poulsen, and Best used mainly the premaxilla, dentary, and special bones for their fish identifications. For the To'aga assemblage three additional elements, the articular, maxilla, and quadrate, were used. This increased the size of the To'aga sample about fifteen percent and added two taxa. Taxa Balistidae Diodontidae - 210 The To'aga Site Table 13.17 Density of Identified Fish Bone from Layer III A/B, Units 20/23 No. of Sample Density Identified Fish Volume (m3) (NISP/m3) Excavation Unit 221 1.3 170 Bulk Samunple 34 0.0005 68,000 Given the problems in data comparability of the Western Polynesian faunal assemblages, only general comparisons between the data sets can be made. The issue of comparability is important for future faunal work in the area. Ideally, a faunal data base would be created in which data from different sites could be easily assimilated into one body of knowledge with new data continually adding to our knowledge of subsistence patterns. Fish For many of the Western Polynesian sites either little faunal material was recovered or the data are poorly reported. The data on fish bones from Western Samoa consists of brief notes on their presence in the sites. Janetski (1976, 1980) mentions 10 fish bones identified from Potusa and an un- known quantity from Falemoa and Jane's Camp. Over 174 grams of fish bone were recovered from Lotofaga (Davidson 1969), but no other data are presented. The best reported and largest samples of fish come from Tongatapu, Niuatoputapu, Lakeba, and To'aga. Kirch (1988) recovered a sample of 231 NISP across 11 taxa from Niuatoputapu. From Tongatapu, Poulsen (1987) identified 15 fish taxa containing 179 NISP. Lakeba produced 323 MI or 1782 NISP, and 21 taxa from the four sites for which the fish component was analyzed (Best 1984). The To'aga assemblage is comparable in size to Lakeba with 2196 NISP and 22 taxa represented. The most abundant fish taxa are inshore/reef fishes, and a few taxa make up the majority of the assemblage (fig. 13.3). The cause of this distribution of fish taxa may be cultural (fishing strategies, food preferences) or environmental (natural abundances and distributions). Unfortunately, the biases created by the recovery techniques, differential preservation, and identifiability may have influenced these distributions. The most common fish families across sites are Scaridae, Lethrinidae, Serranidae, Acanthuridae, and Diodontidae. The dominance of taxa, such as Scaridae, Lethrinidae, and Diodontidae, may be due to preservation and identification bias as much as subsistence patterns. The diagnostic elements of these taxa are very robust and easily identified. It is Table 13.18 Density of Identified Fish Bone from Layer II, Unit 30 No. of Sample Density Identified Fish Volume (m3) (NISP/M3) Excavation Unit 29 0.8 36 Bulk Samunple 52 0.0005 104,000 Faunal Assemblages 211 -a riranidae (1 7.7%) -Acanthuridae (18.2%) Hobcentridae (8.1 %)- Muraenidae (6.0%)I Carangidae (5.2%)- Acanthuridae (10.7%)- '-Serranidae (10.4%) Scaridae (112%) Lakeba 101/7/196 Fblocentridae (13.7%) Lakeba 101/7/197 Other (6.6%) Balistidae (5.3%) Serranidae (6.6%%) Diodontidae (6.6% Lethrinidae (35.5%) Other (1 6.7/o)- (39.5%) Acanthuridae (6.7%)- Labridae (10.0%) Diodontidae (1 1.7%)- Scaridae (25.0%) Lethrinidae (15.0%) LBalistidae (15.0%) Tongatapu Scaridae (24.6%) Niuatoputapu Other (1 7.7h)- Acanthuridae (7.8?h)- -Lethrinidae (17.3%) Serranidae (8.4%)- Elasmobranchii (10.0?h)- Diodontidae (9.5%)- '-Gireilidae (10.1%) Diodontidae (22.5%)-' Percentage composition of fish faunal assemblages from major Western Polynesian and Fijian sites; the two diagrams for the To'aga site are with and without Diondontidae. To'aga Other (22.0%)- To'aga Scaridae (6.6%)- (42.1%) Other (28.0%). Other (22.3?h)- Labridae (7.8h)- (42.0%) Figure 13.3 212 The To'aga Site Table 13.19 Summary of Recovery and Quantification Techniques for Western Polynesian Faunal Analyses Vertebrate Invertebrate Site Screen Size Quantification Quantification Reference SAMOA Potusa N/A P/A, NISP Weight Jennings et al. 1980 Falemoa 1/4" P/A, NISP Weight Jennings et al. 1980 Jane's Camp 1/2", 3/16" P/A, NISP Weight Jennings et al. 1976 Lotofaga 1/4" P/A, Weight Weight Green & Davidson 1969, 1974 TONGA Niuatoputapu 1/4" NISP Weight Kirch 1989 Tongatapu 1/4" NISP Weight Poulsen 1987 FUI Lakeba 2.5, 5, 9mm MNI MNI Best 1984 Naigani 2.5, 3.5, 7.1mm MNI NISP Best 1981; Kay 1984 Yanuca N/A MNI --- Hunt 1980 N/A, information not available P/A, presence/absence therefore more likely that these taxa will be pre- served and identified than taxa with less robust and distinctive skeletal elements. The quality of the fish reference collections has also influenced the presence or absence of taxa in the assemblages. Based on edthoarchaeological data, Mullidae and Pomacentridae are among the most abundant fish caught on Niuatoputapu, but none are recorded archaeologically (Kirch and Dye 1979). Kirch (1988:225) suggests that the differences between the modem and archaeological assemblages may be due to the poor quality of the reference collection. The lack of an adequate reference collection is also a factor in the composition of the Lakeba (Best 1981:497) and To'aga data. Finally the use of 1/4" screens may have resulted in the absence of Pomacentridae from te Niuatoputapu archaeo- logical assemblage since these fish have small diagnostic elements. Non-Fish Vertebrates Compared to the fish, the non-fish vertebrate sample is smaller, but better reported (table 13.20). Rat, bird, and marine turtle are found at most sites. Marine mammal was identified at only two sites, NISP, number of identified specimens MNI, minimum number of individuals Tongatapu and To'aga. Fruit bat was recovered from the Fijian sites, from Falemoa, and from Niuatoputapu. The 'wild' vertebrate fauna tend to be from earlier instead of later sites. Some of the largest amounts of turtle and bird were from the Lapita sites, TO-2, NT-90, 101M7/196 and 1017/197. Chicken is the most common of the three domesticated animals. Dog and pig are less abun- dant, possibly due to the problem of distinguishing between tex two species when the bone is frag- mented. The evidence for the presence of the pig, dog, and chicken from initial colonization is scant. Pig is present throughout the Lotofaga sequence, but the basal date of the site is about A.D. 1000. At Tongatapu, only chicken is found in the early site, TO-2. Dog and chicken, as well as a bone tenta- tively identified as pig, were recovered from the Lapita layers of Yanuca. The best evidence for dte early presence of all three domesticated animals comes from NT-90, on Niuatoputapu. Invertebrates The invertebrate component comprises a large proportion of Western Polynesian faunal assem- blages. The most abundant taxa vary across sites, Fawsal Assemblages 213 Table 13.20 Summary of Western Polynesian Vertebrate Faunal Assemblages (NISP) Site Pig Dog Chicken Rat Fruit Marine Bird Marine Fish Bat Turtle Mammal To'aga 1 Jane's Camp --- Falemoa 2 Potusa 37 Lotofaga Locus A Locus B Locus C Tongatapu TO-I TO-2 TO-3 TO-4 TO-5 TO-6 Total Niuatoputapu NT-90 NT-91 NT-93 NT-100 NT-l10 NT-112 NT-113 NT-135 NT-163 Total Lakeba 1o0n/196 1o0n1/97 ioin/47 olin/i32 ioin/135 iOin/2b 1o0n//66 Total Naigani Yanuca p p p 10 3 189 202 2 3 3 19 1 33 1 3 4 4 P = present in unknown quantities 15 p p 380 p 56 139 25 p p p 27 p p p 2196 87 p 10 p p p p p p >39g >74g >61g 47 7 1 3 16 74 294 9 2 2 198 505 17 404 19 1 18 15 474 125 109 42 2 73 167 504 2 1 1 4 72 43 1 27 36 179 1 --- --- --- --- --- --- --- --- --- 4 7 3 14 12 6 7 1 37 3 26 92 1 16 17 71 6 10 2 3 1 93 31 1 3 1 7 4 5 52 10 1 15 46 --- 11 4 1 12 34 --- 231 19 2 2 1 1 1 5 1 2 28 2 8 69 22 1 3 1 20 20 3 4 11 11 5 2 21 2 76 60 180 7 323 18 23 61 12 9 72. 4 11 214 The To'aga Site but the dominant taxa tend to reflect the marine environment near the site. For example, the exploi- tation of a sheltered lagoon is reflected in the abundance of bivalves in the Tongatapu assemblage. Other sites contain mainly Turbo or other taxa reflecting the exploitation of a fringing reef environ- ment. Changes in the dominant taxa at Niuatoputatpu, Tongatapu, and Lakeba are also used to indicate changing marine environments. While it appears that the most abundant taxa are good indicators of environment, the influence of environmental versus cultural factors still needs to be determined. As at To'aga, the bulk of the inverte- brate assemblages is concentrated in a few taxa. Whether this uneven distribution reflects the exploi- tation of naturally abundant taxa or cultural prefer- ences will need to addressed in future studies. CONCLUSIONS The analysis of the To'aga faunal assemblage has generated a number of new questions. Despite the time depth represented at the To'aga site, there is a striking lack of change in the resources exploited. A few taxa comprise a large percentage of the fish and invertebrate components of the assemblage. The overall pattem of high diversity may reflect the exploitation of naturally abundant taxa or culturally preferred taxa. Population studies of marine envi- ronments off To'aga would be useful for creating a baseline of natural distributions which could then be compared to the archaeological data. Addressing these and other faunal questions requires data robust enough to compile into a cumulative data base and to use at a level higher than nominal. The To'aga analysis has shown that methodological biases introduced during excavation and analysis can severely affect the data, reducing its robustness. Thus, interpretations must be made cautiously. The problems created by these biases are compounded when data are compiled from different sites into a regional data base. Variability in and among data sets may be attributed to differences in recovery or analytical techniques rather than prehis- toric cultural patterning. Faunal analysis can be a useful tool for understanding subsistence practices, an important aspect of prehistoric culture. Its utility in future studies, however, depends upon the com- mitment of faunal analysts and archaeologists to create quality faunal data. ACKNOWLEDGEMENTS M. S. Allen, P. V. Kirch, and the Bishop Museum kindly allowed the use of their reference collections. V. L. Butler assisted in the fish identifi- cations. E. Marshall at the Burke Museum and R Kawamoto at the Bishop Museum assisted in confirming shell identifications. E. Gordon, K. Stark, and P. Kirch read the numerous versions of this chapter. Funding was provided by Sigma Xi Grants-in-Aid of Research for the development of the author's fish reference collection. REFERENCES CITED Abbot, R. T., and S. P. Dance 1986. Compendium of Seashells. Amen can Malacologists, Melboume, Florida Allen, M. S. 1990. Excavations at the Ureia site, Aitutaki, Cook Islands: Preliminary results. Archaeology in Oceania 25:24-37. Anderson, A. 1986. Mahinga ika o te moana: Selection in the pre-European fish catch of southeem New Zealand. In A. Anderson, ed., Traditional Fishing in the Pacific: Ethno- graphical and Archaeological Papersfrom the 15th Pacific Science Congress, pp. 151-66. Pacific Anthropological Records 37. Honolulu: Bernice P. Bishop Museum. Best, S. 1981. Excavations at Site VL21/5, Naigani Island, Fiji, a preliminary report. Department of Anthropology, University of Auckland. -. 1984. Lakeba: The prehistory of a Fijian Island. Ph.D. dissertation. University of Auckland. Butler, V. 1987. Fish remains. IN D. E. Lewarch, ed., The Duwamish No. I Site: 1986 Data Recovery, pp. 1-37. URS Corporation and BOAS Inc., Seattle. 1988. Lapita fishing strategies: The faunal analysis. IN P. V. Kirch and T. L. Hunt, eds., Archaeology of the Lapita Cultural Complex: A Critical Review, pp. 99-116. Thomas Burke Memorial Washington State Museum Research Report No. 5, Burke Museum, Seattle. Casteel, R. 1972. Some of the biases in recovery of archeological faunal remains. Proceedings of the Prehistoric Society 36:382-88. Faunal Assemblages 215 Chaplin, R. E. 1971. The Study of Animal Bones from Archaeological Sites. New York: Seminar Press. Eisenberg, J. M. 1981. A Collector's Guide to Seashells of the World. New York: McGraw- Hill. Gordon, E. 1991. Differential faunal recovery and models of subsistence change: A Hawaiian example. Paper presented at the 1991 Annual Meeting, Society for American Archaeology, New Orleans, Louisiana Grayson, D. K. 1979. On the quantification of vertebrate archaeofaunas. IN M. B. Schiffer, ed., Advances in Archaeological Method and Theory, vol 2, pp. 199-237. New York: Aca- demic Press. Grayson, D. K. 1984. Quantitative Zooarchaeology. New York: Academic Press. Green, R. C. 1986. Lapita fishing: The evidence from sites in the Reef/Santa Cruz group, southeast Solomons. IN A. Anderson, ed., Traditional Fishing in the Pacific: Ethno- graphical and Archaeological Papersfrom the 15th Pacific Science Congress, pp. 1 19-36. Pacific Anthropological Records 37. Bishop Museum, Honolulu. Green, R. C., and J. Davidson 1969. Archaeology in Western Samoa, Vol. 1. Auckland Institute and Museum Bulletin 6. 1974. Archaeology in Western Samoa, Vol. 2. Auckland Institute and Museum Bulletin 7. Hill, H. B. 1978. The use of nearshore marine life as afood resource by American Samoans. Unpub- lished MA. thesis, University of Hawaii, Honolulu. Hinton, A. G. 1972. Shells ofNew Guinea and the Central Indo-Pacific. Rutland (Vermont): Tuttle. Hunt, T. L. 1980. Toward Fiji's past: Archaeologi- cal research on southwestern Vidt Levu. Unpublished M.A. thesis. University of Auckland. Janetski, J. C. 1976. Dietary remains from Jane's Camp-a midden site. IN Jesse Jennings, ed., Excavations on Upolu Western Samoa, pp. 75- 82. Pacific Anthropological Records 25, Department of Anthropology, Bishop Museum, Honolulu. 1980. Analysis of dietary remains from Potusa and Falemoa. IN J. Jennings and R. Holmer, eds., Archaeological Excavations in Western Samoa, pp. 117-22. Pacific Anthropological Records 32, Department of Anthropology, Bishop Museum, Honolulu. Kay, R. M. A. 1984. Analysis ofarchaeological materialfroom Naigani. Unpublished M.A. thesis. University of Auckland. Kirch, P. V. 1976. Ethno-archaeological investiga- tions in Futuna and 'Uvea (Western Polynesia): A preliminary report. Journal of the Polynesian Society 85:27-69. 1988. Niuatoputapu: The Prehistory of a Polynesian Chiefdom. Thomas Burke Memo- rial Washington State Museum Monograph No. 5. Burke Museum, Seattle. Kirch, P. V., and T. S. Dye 1979. Edmoarchaeology and the development of Polynesian fishing strategies. Journal of the Polynesian Society 88:53-76. Kirch, P. V., and D. E. Yen 1982. Tikopia: The Prehistory and Ecology of a Polynesian Outlier. Bishop Museum Bulletin 238. Bishop Museum, Honolulu. Leach, F. 1986. A method for the analysis of Pacific Island fishbone assemblage and an associated database management system. Journal of Archaeological Science 13:147-59. Lohse, E. 1980. Excavations on Manono Islet. IN J. Jennings and R. N. Holmer, eds., Archaeologi- cal Excavations in Western Samoa, pp. 21-32. Pacific Anthropological Records 32, Depart- ment of Anthropology, Bishop Museum, Honolulu. Masse, B. 1989. The archaeology and ecology of fishing in the Belau Islands, Micronesia. Unpublished Ph.D. dissertation, Department of Anthropology, Southern Illinois University, Carbondale. Nagaoka, L. 1988. Lapita subsistence: The evidence of non-fish archaeofaunal remains. IN P. V. Kirch and T. L. Hunt, eds., Archaeology of the Lapita Cultural Complex: A Critical Review, pp. 117-134. Thomas Burke Washington State Memorial Museum Research Report No. 5. Seattle. Payne, S. 1972. On the interpretation of bone samples from archaeological sites. IN E. S. Higgs, ed., Papers in Economic Prehistory, pp. 216 The To'aga Site 65-81. Cambridge: Cambridge University Press. Poulsen, J. 1987. Early Tongan Prehistory: The Lapita Period on Tongatapu and its Relaion- ships, Vols. 1 and 2. Terra Ausais 12. Department of Prehistory, Research School of Pacific Studies, Australia National University, Canberra. Smith H. L. 1976. Jane's Camp (SUFl-I). IN J. Jennings, ed., Excavations on Upolu, Western Samoa, pp. 61-74. Pacific Anthropological Records 25. Department of Antrpology, Bishop Museum, Honolulu. Thomas, D. H. 1969. Great Basin hunting pattems: A quantitative method for treating faunal remains. American Antiquity 34:393401. 13 FAUNAL ASSEMBLAGES FROM THE TO'AGA SITE LISA NAGAOKA THE CALCAREOUS SAND depositional environ- ment at To'aga favored the preservation of faunal matenial (table 13.1). Over 165 kilograms of invertebrate remains were recovered, represening more than forty families. The To'aga fish-bone sample-the largest in Western Polynesia contains 2,196 identified bones across twenty-two taxa. Pig, chicken, rat, marine mammal, urle, and bird comprse the 687 bones of the non-fish vertebrate sample. Each component of the faunal assemblage is described in detail below. The problem of recovery bias is addressed here through the analysis of bulk sediment samples from the To'aga excavation. Generally, the use of smaller- sized screens increases the size of the faunal sample and the number of taxa recovered. The To'aga bulk samples were sieved through different screen sizes to determine the effects of screen size on the composi- tion of the faunal assemblage. Cuwent knowledge of Western Polynesian subsistence practices is limited since few zooarcheological studies have been conducted in the region. In this context, the To'aga faunal assemblage is important for adding new information to our understanding of regional subsistence trends. The long temporal sequence at To'aga allows for an assessment of changing subsistence patterns. Despite the small sample of Western Polynesian sites, comparisons of the To'aga assemblage with other regional faunal assemblages may yield infor- mation about subsistence patterns. METHODS The faunal remains were recovered by dry- screening all excavated earth (except the clayey colluvial sediment) through 1/4" mesh. To deter- mine the feasibility of screening the colluvium, Layer I of Unit 20 was screened rough 1/4" screens. Only one poody preserved Turbo shell was recovered from the 0.8 m3 sieved. A decision was made not to screen the colluvial layer in other units because of the difficulty in dry-screening the matrix, and because of the low density and poor preservation of faunal and other material in this clayey deposit (See chapter 7 for further discussion of the pH and other aspects of the To'aga site sediments.) During the 1987 field season, most of the shell recovered was identified, weighed, and discarded in the field. Voucher samples and the remaining unidentified shell and bone were shipped back to the laboratory. All faunal materials recovered from the 1989 field season were washed in the field and returned to the laboratory for identification and analysis. Fish remains were identified to the family level using reference collections from the Bishop Mu- seum, and the personal collections of Patrick Kirch (U.C. Beikeley), Melinda Allen (University of 190 The To'aga Site Table 13.1 Summary of To'aga Site Faunal Remains Excavation Units Faunal Class 1-14 15-30 Total Tota Shel (kg) 50.291 118.367 168.658 Identified Shell (kg) 50.291 115.669 165.960 Tota Non-fish (NISP) 322 365 687 Tota Fish (NISP) 3462 6062 9524 Identified Fish (NISP) 723 1473 2196 Washington) and the author. Reference collections from the Bishop Museum were used to identify rat, dog, pig, manine mammal, and marine turtle. The bird component was identified by David Steadman of the New York State Museum (see chapter 14). Although many Pacific faunal analysts use MNI (minimum number of individuals) to quantify vertebrate remains (e.g., Leach 1986; Anderson 1986; Green 1986), we use NISP (number of identified specimens) for the To'aga vertebrate assemblage. The problems with both measures have been discussed extensively elsewhere (see Chaplin 1971; Grayson 1979, 1984; Payne 1972). NISP was chosen here because the effects of aggregation make MNI an inconsistent measure. Although the prob- lem of interdependence affects NISP, the measure is constant across aggregation units. The invertebrate faunal component was identi- fied using standard shell identification guides (Abbot and Dance 1986; Hinton 1972; Eisenberg 1981). Tentative identifications were confirmed using reference collections at the Burike Museum in Seattle and at the Bishop Museum in Honolulu. Inverte- brate remains were quantified by weight. As with MNI and NISP, use of weight has its drawbacks because of variations in size and density of different shell taxa. For example, Tridacna maxima, the giant clam, has a very dense shell, and one large individual may weigh more than 1 kg. On the other hand, shells such as limpets (Patellidae) are very light, so that many individuals may account for a small amount of weight. As a result, heavy shells may be overrepresented and light shells undernepresented in any sample. Post-depositional alterations such as leaching and fossilization also may distort shell weight RESULTS The To'aga faunal data are presented in three categories: the vertebrate component, which is subdivided into fish and nonfish, and the invertebrate component Of the thirty excavation units, complete data by stratigraphic layer from thre areal excava- tions (1987 Main Trench, Units 20/23, and Units 15/ 29/30) are presented in the text for comparison. These units were chosen to represent the site because they comprise a larger sample than the individual units. Data for the remaining excavation units are presented only in the summary tables for the separate faunal categories. The complete faunal data by stratigraphic layers from all excavation units are available from the author on request. Fish Remains The To'aga excavations yielded 2,196 identified fish bones representing twenty taxa (tables 13.2-6). Although this is the largest archaeological fish bone assemblage from Western Polynesia, an average of only 73 bones were identified for each excavation unit. Acanthuridae (surgeonfish), Diodontidae FawuaIAsbages 191 Table 13.2 Fish Fauna Recovered from the 1987 Excavations (NISP) Excavations Units Taxa 1, 4-9 Diodontidae Holocentridae Acanthuridae Seffanidae Scaridae Carangidae Balistidae Muraenidae Labridae Ostaciidae Lutjanidae Aulostomidae Congridae Elasmobranchii Lethrinidae 289 64 48 47 24 22 24 15 9 5 9 8 5 3 2 2 Belonidae Kyphosidae Sphyraenidae Scombridae 2 3 10 11 12 13 14 1 -- 4 15 -- -- 19 3 4 9 1 -- -- 4 -- 1 13 -- -- 1 2 -- 2 15 1 -- 1 6 -- 1 5 -- -- 1 3 -- -- 6 -- -- -- 2 -- -- 2 -- -- -- -- -- 1 1 1 1 -- -- -- -- 2 -- - - -- -- -- 7 -- -- -- ~ 1 -- -- 1 -- -- 2 -- - -- 2 -- -- -- -- -- -- 1 1 Total 328 80 67 67 38 32 28 17 13 12 11 8 7 5 3 2 1 1 1 -- -- -- -- -- 1 -- -- -- -- -- -- -- 1 Bothidae TOTAL IDENTIFIED UNIDENTIFIED TOTAL 577 2003 2580 4 16 20 1 1 7 45 52 38 190 228 3 2 5 1 10 19 29 82 464 546 1 722 2739 3461 (spiny puffers), Holocentridae (squinelfish), The stucture and composition of the Toaga Serranidae (groupersfcods) and Scaridae (paotfish) fish-bone data pwbably reflect a conbination of comprise appoximately 78% of the identified fish methodological, environmental, and cultuml factors. remains (fig. 13.1). These taxa are usually the most Methodological factors include recovely bias and almiant across time and space at the To'aga site. problems in identification and quantification. Bias - 192 The To'aga Site ) oo >! 8 x _- xo x vo ~o o _- O N o o vo (71, Ch %D on o m m m N _ _ _ _ 4 n N m xn w m 2 It N o N - : C- ', N X 0 e ct 4 U) O F 0 m _ m C4 N V-4 V4 -_ _ 9- V- - t I I I o Wo : a : - : N : - _-4 _- - - - a "-4 0 T-4 e> - : : a - a - 4 a a - a a a a a a a~ ~ a I I I I I I I I I I ON I I I I 00 - V-4 en Cf) dl~t - a - t Q _4 m 7 _; t n I _ 2 1. Q t S: a _ : : : a ar - a C - -- . -4 _- 1. !4 - t ,m A : a - a a : - : a -4 a en - a a en oo V- 00 "-4 "I a - - 0 ; "-4 tn r | 3V3 4 ] 8 I U 0~ * ~ . ~ ~ . I 3m a r )o 00 00 "t tn (I _ la C 6 ;O. ON m 9 e4 A kn e4 A0 "4 v W 0 Cie ;I- CU aC Cu 4 ) e !s 0 ) *;0 0x 00 "14 00 I I I I I I I I I I V-4 en II II en o C4 11 lqt C4 V-4 II It Faunal Assemblages 193 Table 13.4 Fish Fauna from the 1987 Main Trench (Units 1, 4-9) Layers Taxa HA-I HA HB HC Total Diodontidae 4 25 154 106 289 Holocentridae 3 17 25 19 64 Acanthuridae 3 13 20 12 48 Serranidae 5 11 17 14 47 Scaridae 2 1 21 1 25 Balistidae 4 8 10 2 24 Carangidae 2 4 10 6 22 Muraenidae --- 4 4 7 15 Labridae 2 3 2 2 9 Lutjanidae 2 1 5 1 9 Aulostomidae --- 1 4 3 8 Ostraciidae --- 1 3 1 5 Congridae --- 1 1 3 5 Elasmobranchii --- 2 --- 1 3 Lethrinidae --- --- 2 --- 2 Belonidae 1 1 --- --- 2 Kyphosidae --- --- 1 --- 1 TOTAL IDENTIFIED 28 93 279 178 578 UNIDENTIFIED 191 535 827 450 2003 TOTAL 219 628 1106 628 2581 in the recovery process is shown to affect the sample size and the taxa represented in the assemblage (see "Bulk Samples" section). Problems in the identifica- ion process include the quality of the reference collection, which can limit the accuracy of the identifications and the number of taxa represented. For the To'aga assemblage, several distinctive mouth parts could not be identified using the reference collection at hand. With a better reference collec- ton, subfamily identifications may also be possible. Another methodological problem is the inclu- sion of "special bones" in the NISP count. A few taxa are identified mainly by special bones that can number up to 300 per individual, thus greatly inflating he NISP count. This is especially true for Diodontidae, which can have more than 250 spines per individual, and to a lesser extent for Ostraciidae, Elasmobanchii, and Balistidae. Of the 923 Diodontidae bones identified at To'aga, 901 were spines and only 22 were mouth parts, most being concentrated in Unit 21 and in the 1987 main excavation trench. If the Diodontidae spines are removed from the NISP count, the ranking of diodonts drops from one to thirteen, and the shape of the graph changes (fig. 13.2). Although the presence of this poisonous fish may seem odd, its remains are common in middens across the Pacific (e.g., Allen 1990; Butler 1987; Masse 1989). Moreover, the fish is still eaten by some modem Pacific populations (Bagnis 1972, Masse 1986). 194 The To'aga Site Table 13.5 Fish Fauna from Transect 5, (NISP) Units 15/29/30 Layers Taxa H HA-1 IIIB HID Total Diodontidae 9 1 9 16 35 AcanthuridM 9 --- 8 15 32 Serranidae 11 1 4 7 23 Holocentridae 3 --- 10 5 18 Scaridae 3 1 3 8 15 Lutjanidae 3 --- 2 5 10 Labridae 6 --- 2 1 9 Ostraciidae --- --- 1 3 4 Lethrinidae 2 --- --- --- 4 Muraenidae --- --- --- 3 3 Carangidae --- --- 1 1 2 Mullidae --- --- 2 2 Balistidae --- --- 2 --- 2 Kyphosidae 1 --- --- 1 Bothidae 1 --- --- --- 1 Congridae --- --- --- 1 1 TOTAL IDENTIFIED 48 3 42 67 160 UNIDENTIFIED 201 16 250 297 764 TOTAL 249 19 292 364 924 The structure of the To'aga fish-bone data also may reflect natural distributions and abundances of fish taxa. Most of the To'aga assemblage can be classified as inshore fishes, although a few families such as Serranidae, Lutjanidae, and Carangidae cover a wide range of habitats. Reef ecosystems are generally more diverse and have a higher pwductiv- ity rate than open ocean environments; therefore, the abundance of inshore versus pelagic fish may reflect the natural diversity of the different environments. Fishing strategies may also be reflected in the fish-bone data. The fishhooks recovered from the site (see Kirch, chapter 1 1) may have been used to catch serranids, holocentrids, and lutjanids, but probably not scarids or acanthunds which are more likely to be caught by netting or spearing. A com- parison of modem Samoan reef exploitation (Hill 1986) and the To'aga fish data shows that the most abundant taxa in the archaeological assemblage can be caught by several fishing techniques (table 13.7). These taxa may have had more opportunity to be caught than taxa for which only one stsegy was used. FawalAssemblages 195 Table 13.6 Fish Fauna from Transect 9 (Units 20/23) (NISP) Layers Taxa IIB HIA HIIB uic IV Total Acanthuridae Serranidae Diodontidae Muraenidae Holocentridae Scaridae Labridae Ostraciidae Carangidae Lutjanidae Congridae Aulostomidae Lethrinidae Balistidae Scombridae Elasmobranchii Bothidae Kyphosidae Mullidae TOTAL IDENTIFIED UNIDENTIFIED TOTAL 3 2 30 2 5 37 3 14 13 3 3 23 1 3 14 --- 1 14 --- 1 15 --- 3 1 2 9 1 2 9 2 1 5 7 5 1 2 3 - 1 16 34 191 63 161 610 79 195 801 18 --- 53 9 --- 53 12 2 44 9 --- 38 8 1 27 9 1 25 6 --- 22 6 13 22 2 --- 14 --- --- 12 2 --- 10 7 1 6 4 --- 5 2 --- 4 3 1 1 1 --- 1 1 90 17 348 414 46 1294 504 63 1642 Non-Fish Vertebrate Remtains Trench. Bird bones were the second most abundant, with 139 bones. Steadman presents an analysis of The non-fish vertebrate sample of 687 bones is the 72 identified bird bones in chapter 14. Fifty-six small, averaging only 23 bones per excavation unit marine turtle-bone fragments were scattered (tables 13.8, 13.9). About half the sample consists of throughout the excavations with one-third of the Rattus exulans, the Pacific Rat, with nearly half the sample concentrated in Layer IIIB of Unit 20, dating rat bones coming from Layer II of the 1987 Main to about 2900-2400 B.P. From this same time period, --- --- --- ---