Volume 99 1990 > Volume 99, No. 3 > Recent archaeological research on Aitutaki, Southern Cooks: the Moturakau shelter, by M. S. Allen, p 265-296
RECENT ARCHAEOLOGICAL RESEARCH ON AITUTAKI, SOUTHERN COOKS: THE MOTURAKAU SHELTER
The Southern Cooks, lying near the western periphery of East Polynesia, occupy an important geographic position with respect to understanding both regional East Polynesian developments, and interaction between these islands and those further west (Figure 1). Traditionally, the Southern Cooks were believed to be settled comparatively late, probably from the Societies (e.g., Buck 1944; Duff 1968, 1974). Accumulating evidence (Bellwood 1978:60-80, 133-9; Walter 1987; this present report), however, indicates that the Southern Cooks shared in the “East Polynesian Archaic” previously described for the Societies and the Marquesas between c. A.D. 800 and 1200 (Suggs 1961; Emory and Sinoto 1964; Sinoto 1968a, 1979, 1983; Sinoto and McCoy 1975; Kirch 1986). The evidence also suggests that deposits related to initial human colonisation of the Southern Cooks have not yet been located. Excavations at the Ureia Site on Aitutaki in 1987 (Allen and Steadman 1990) located redeposited cultural materials below the 925 B.P. cultural stratum identified by Bellwood (1978). In addition, analysis of faunal materials associated with the intact 925 B.P. cultural layer show evidence of a native avifauna which is already greatly impoverished relative to that known for other Southern Cook Islands (Allen and Steadman 1990; Steadman 1985, 1987a, 1989). Presumably this depauperate avifauna is a reflection of human activities over a significant period of time. Thus, the possibility that the Southern Cooks were involved in the earliest period of East Polynesia settlement and subsequent diversification from the west cannot yet be discounted, particularly in light of their key geographical position.
The Southern Cooks may also have been an important crossroads for East and West Polynesia following initial colonisation. Traditions record contact with other archipelagos, including Samoa, the Societies, and Tonga (Gill 1880:1, 31, 36, 1856:2, 5; Buck 1944:484-8). Some of these traditions relate to the initial colonisation of the Southern Cooks, as, for example, the account of Makea Kariki's arrival from the Manu‘a Islands in Samoa (Gill 1876:23; 1880:1). Similarly, a Rarotongan chant indicates ties with Savai‘i, Upolu, Vava‘u, and Manu‘ a (Gill 1876:27). While a literal interpretation of these traditions may not be desirable, they minimally indicate inter-archipelago- 266 - 267
interaction of some kind. Buck (1944:484-9) notes still other traditions of continued contact between the Southern Cooks and the Societies, as well as describing material culture items shared between the Southern Cooks and more western islands. Recent archaeological evidence of contact with the west comes from Walter's excavations on Ma‘uke (Walter and Dickinson, 1989) and those of Sinoto and Kurashina on Atiu (Sinotoet al. 1988). In both cases, pottery from localities west of the Southern Cooks have been identified. Greater resolution on the nature, degree, and direction of interaction between the Southern Cooks and other Polynesian islands will be an important challenge for future archaeological research to which this paper makes a small contribution.
The degree of prehistoric interaction among islands within the Southern Cooks is not well understood at present. With the exception of the Ma‘uke-Atiu-Mitiaro group, allied into the Ngaputoru polity at contact (Mokoroa 1984), the Southern Cooks are not intervisible and the considerable distances between them might be expected to contribute to intra-archipelago diversification. Certainly, each island (with the above-mentioned exception) was politically independent at contact. Nevertheless, there is traditional, ethnohistoric, ethnographic, and archaeological evidence for communication between them (e.g. Gill 1876:23-7,1880:1, 14, 18-25; 1856:5, 21; Buck 1944:489-91; Walter 1987). The traditional evidence describes patterns of exogamous marriage between the high-ranking chiefly families of Rarotonga, Mangaia, Atiu, and Aitutaki, as well as interisland warfare (Gill 1876, 1880; Pakoti 1895;Buck 1944). The prehistoric control of Manuae (102km from Aitutaki) by Mangaians 385 km to the south, is also notable. Similarly, Walter (1987:248) has proposed that the raw materials of certain stone and shell artefacts represented at the Anai‘o site derive from localities outside of Ma‘uke. The recent archaeological finds from Aitutaki, described below, may be particularly relevant in this regard.
EARLIER ARCHAEOLOGICAL WORK
The Southern Cooks are among the more poorly known central East Polynesian archipelagos (Kirch 1986:33, 35-7). Most of the early work in this group consisted of surface surveys. Duff and his students (in Trotter 1974) investigated several traditional sites on Rarotonga, Atiu, and Aitutaki. On Rarotonga survey work concentrated along the Ara Metua and limited excavations were conducted in the valleys of Tupapa, Avana, and Maungaroa, and at Vaiakura. Bellwood (1969, 1971, 1978) subsequently undertook a more comprehensive survey of the late prehistoric settlements in Maungaroa Valley, as well as expanding upon the earlier excavations of Parker - 268 (1974). Surveys were also made in the valleys of Rutaki and Turangi, and the Avana Valley pondfield system was mapped (Bellwood 1978). At the early Ngati Tiare site on Rarotonga (RAR 40), Bellwood (1978:62-80) excavated approximately 42 square metres, providing important contextual information for the famous Ngati Tiare adze cache. Bellwood (1978:133-9) was also responsible for identification of the even earlier Ureia site on Aitutaki (AIT 10), where approximately 12 square metres was excavated. His work also included an inventory of religious structures on Aitutaki and Mangaia.
More recently Kurashina, Stevenson, and Sinoto (1987) have mapped and stabilised several marae on Rarotonga, Atiu, Aitutaki and, most recently, Mitiaro. It was in conjunction with this work that a single ceramic sherd was recovered from the vicinity of Marae Vairakaia on Atiu Island (Sinoto et al. 1988). Other recent work includes Walter's (1987) extensive excavations at the coastal Anai‘o site on Ma‘uke Island. The two occupations at Anai‘o, beginning in the 13th century A.D., showed evidence of a large permanent community with spatially differentiated activity areas reflecting cooking, adze production, and fishhook manufacture.
Despite over 20 years of archaeological interest in the Southern Cooks, albeit sporadic, there are few examples of well-stratified undisturbed deposits of significant temporal duration which might inform on trends in resource exploitation and material culture. The Ureia site, covering over 900 years of Southern Cook prehistory, is a notable exception (Bellwood 1978; Allen and Steadman 1990). This present paper summarises recent findings from a second Southern Cook site of lengthy temporal duration, the Moturakau Shelter, Aitutaki.
In 1987 the senior author was invited to join David Steadman of the New York State Museum in his interdisciplinary study of the biogeography of East Polynesian vertebrates in the Southern Cook Islands. Archaeological excavations were carried out principally on the island of Aitutaki, but also on Atiu. On Aitutaki, our efforts were focused on the Ureia site where Bellwood (1978) had previously identified abundant faunal remains in a stratified deposit. The primary purpose of this work was to recover a temporal sequence of Aitutaki's avifauna and other small vertebrates (i.e., geckos and skinks) which could be compared with Steadman's previous collections from Mangaia (Steadman 1987a, 1985) and other Polynesian islands (Steadman 1989, 1988, 1987b; Steadman and Olson 1985). Recovery of subfossil materials from dated contexts was especially important for understanding natural patterns of endemism and diversity and temporal patterns of extinction. In conjunction with this work, a small rockshelter on the islet of Moturakau was also tested.- 269
THE ENVIRONMENTAL SETTING
The Southern Cooks are a geologically diverse group of islands including one high island (Rarotonga), two atolls (Manuae and Palmerston), one almost-atoll (Aitutaki), and four makatea islands (Ma‘uke, Atiu, Mangaia, and Mitiaro). Aitutaki lies near the northern extreme of the group, over 200 km from the Atiu-Mitiaro-Ma‘uke cluster. The main island is a weathered remnant of the original volcanic cone around which the reef initially formed (Stoddart 1975a). The volcanic base rises from the narrow coastal plains fairly quickly, particularly on the western slope, reaching a maximum elevation of 119 m. A series of 13 coral detrital islands ring the eastern lagoon edge and two small volcanic islets, Moturakau and Rapota, lie within the lagoon near the southern end (Figure 2).
Figure 2. Map of Aitutaki Atoll, showing the locations of offshore islets mentioned in the text and the Ureia site (adapted from Rarotonga Survey Department map).- 270
The smaller offshore islands provide interesting settings for archaeological study. They are quite variable in area, ranging from 71 ha in the case of Tekopua to just under 4 ha for Moturakau (Stoddart 1975d:107). None of the islets has permanent water supplies. Five of the 15 (only one of which has been extensively surveyed) show evidence of human occupation of some sort. Bellwood (1978) reported a marae on Papau Islet, and Stoddart (1975b: Figure 22) provides an illustration of what may be an eroding cultural layer on the north shore. Bellwood (1978) also described eroding cultural deposits on Rapota. On Tapuaetai, Green (personal communication, 1987) observed a concentration of basalt flakes and preforms and in a wave-cut face along the south shore we ourselves located two distinct cultural layers and eroding basalt debitage (almost certainly the same locality seen by Green). We also noted eroding cultural deposits on the lagoon shore of Akitua Islet. These offshore islets undoubtedly functioned as temporary campsites, used in conjunction with reef exploitation, fishing, and (in the case of Rapota and Moturakau) quarrying activities. While their lack of fresh water would limit permanent habitation, the intensity of their use through time may provide an indirect measure of population growth and expansion on Aitutaki proper. Certainly the transport of lithic resources to Tapuaetai hints at the possibility of occupations of some duration.
Radiocarbon determinations suggest the coral detrital islets are relatively recent constructions. Tridacna maxima valves from the clastic conglomerate platforms of Akaiami and Muritapua Islets yielded dates of 2040 ± 90 and 160 ± 80, respectively (Stoddart 1975a). The dates suggest that the islets may not have been inhabitable at the time of initial human occupation. In terms of potential human resources, the islets were probably once important nesting grounds for seabirds and turtles. They are presently frequented by frigate birds (Fregata minor), fairy terns (Gygis alba), Tropic birds (Phaethon rubricauda), bristle-thighed curlews (Numerius tahitiensis), golden plovers (Pluvialis fulva) and reef herons (Egretta sacra), among others (C. Spaw, personal communication, 1988). The islets also provide ready access to shellfish beds along the outer reef, most notably Tridacna, and could have functioned as staging grounds for open sea fishing expeditions.
Moturakau Islet lies 5 km from the main island, at the south end of the lagoon. The islet consists of a spine of volcanic tuff, up to 9 m in height, which runs the length of the islet. To the lee of this spine is an apron of sandy sediments (Figure 3). The tuff is principally bedded palagonite ash (Wood and Hay 1970:38), but also contains dense basalt blocks of a quality suitable for adze manufacture. Not only does the rock-shelter deposit contain consid- - 271 erable flaking debitage, but also several flaked boulders are found along the shoreline.
Figure 3. Moturakau Islet and location of the Moturakau Shelter (after Stoddart and Gibbs 1975: 72).
The vegetation of the main Aitutaki island is entirely secondary, with no remaining areas of natural forest (Stoddart 1975e:117-122). The coral islets - 272 are characterised by coconut and other typical atoll species including Guettarda, Pandanus, Pemphis, Scaevola, Suriana, Euphorbia Tournefortia, Morinda, Casuarina (Stoddart 1975e:87-116). In contrast, the volcanic islets of Moturakau and Rapota are dominated by large stately Calophyllum inophyllum trees which are popular roosts for fairy terns and frigate birds; the name Moturakau literally translates as ‘islet of trees’.
The excavated shelter is one of two small rockshelters briefly mentioned by Bellwood (1978:117). Local informants told Bellwood that these shelters were traditional refuge places. Stoddart (1975c:70; see also Wood and Hay 1970:37) reports that Moturakau was used as a leper colony. Before excavations within the shelter began, an informal reconnaissance was conducted around the island and along the length of the volcanic spine. No structural remains or exposed deposits were located along the islet's shoreline and no additional rockshelters of any size were found in the interior. Along the water's edge, however, scatters of basalt flakes were noted and several examples of sizable boulders with large negative flake scars were found, particularly on the southern end of the islet.
THE MOTURAKAU SHELTER EXCAVATIONS
Moturakau Shelter (MR1) is a small cleft weathered into the islet's volcanic spine (Figure 4). The shelter measures 3.5 m wide by 2.8 m deep (to dripline) with a maximum of 1.0 m in height. To the south is a second shelter 3.2 m wide by 2.2 m deep to (dripline) and 97 cm in height. At one time the two shelters may have been a single unit, but sediments have filled them to the point where entry must now be made on hands and knees. The shelters are located in a well-vegetated area approximately 14 m inland of the present beach strand line.
In October 1987 the authors excavated a one-metre-square test in the centre of the northern shelter over the course of two and a half days. Limited access to the islet, intermittent heavy rains during the excavation work, and poor lighting within the shelter all worked to limit the area opened during this preliminary study. However, the density of cultural materials was such that, despite the small area of excavation, a considerable amount of information was recovered.
Stratigraphy and Features
The shelter deposit was composed of a sequence of sedimentological units which were sufficiently distinct as to allow excavation by natural layers; particularly thick layers were subdivided into arbitrary levels to increase stratigraphic control. All excavated sediments were passed through 1/8” and- 273
1/4” screens. The strata are summarised in Table 1 with respect to standard soil and sediment characteristics (see also Figure 5). Small scoop hearths were common, particularly in the upper layers.
TABLE 1. Description of Stratigraphic Layers
Figure 5. Stratigraphic profile of south face excavation unit; see Table 1 for detailed descriptions of the layers.
The sedimentological units indicate the interplay between: 1) organic deposition from the adjacent vegetation and human occupation; 2) physical and mechanical weathering of the shelter interior, the latter probably exacerbated by human activities; and 3) episodic deposition of marine sediments by storm events of variable magnitudes. In the upper portion of the - 276 deposit, terrigenous sediment sources dominate with mechanical weathering of the shelter walls, organic decomposition, and colluviation being important processes. The lower portion of the deposit is, in contrast, dominated by marine sediments and agents of deposition. Layer VII apparently represents a storm event of some magnitude which reworked cultural materials either within the shelter or nearby. Storm deposition is suggested both by the massive character of this sandy unit and by the upward-fining of the gravel and pebble components. Layer XII, a dark yellowish brown silt, is notably different from all other strata in texture, colour, and gravel content (see Table 1); both the origin and the sedimentary environment under which this layer was deposited are unclear.
The lack of associated structural features and the numerous small scoop hearths suggest occupations in the shelter were short-term and intermittent. More extensive excavations are needed for finer resolution on the duration of the shelter occupations and the range of associated activities.
Two samples of wood charcoal were submitted to Beta Analytic Inc. for radiocarbon assay. Both samples were pretreated for carbonates and humic acids. Benzene synthesis and counting was reported as normal. The 13C/12C ratios were measured to establish a 13C-adjusted “conventional 14C age” (after Stuiver and Polach 1977). These conventional 14C ages were calibrated using the Stuiver and Reimer (1986) CALIB fortran program for terrestrial samples, based on work by Stuiver and Becker (1986).
One sample (Beta-25766) comes from Layer V, approximately midway up the stratigraphic column (see Figure 3), and yielded a conventional radiocarbon date of 540 ± 70 (delta 13C = -27.2 0/00). The calibrated A.D. range at one sigma is A.D. 1319 to 1433 with an intercept at A.D. 1410; the calibrated B.P. range at one sigma is 631 to 517 B.P. with an intercept at 540 B.P..
The second sample (Beta-25767) derives from the basal layer of the shelter (Layer XIII) and yielded a conventional radiocarbon age of 840 ± 80 (delta 13C = -26.4 0/00). The calibrated A.D. range at one sigma is 1043 to 1263 with an intercept at A.D. 1212; the calibrated B.P. range at one sigma is 907 to 670 with an intercept at 738.
These two dates fit well within the occupation span indicated by Bellwood's previous radiocarbon determinations from the Ureia site on the main island of Aitutaki. Using the same calibration procedures followed for the Moturakau samples, Bellwood's lowermost occupation layer dates to c. A.D. 1025 (A.D. range at one sigma is 987 to 1160); it should be noted, - 277 however, that Bellwood's dates were probably not 13C-adjusted. The Moturakau dates indicate a lengthy use of the shelter throughout the duration of Aitutaki's prehistory as presently known. The lowermost date may indicate initial use of the islet but, given the susceptibility of the shelter to storm waves, evidence of earlier occupations may have been eroded. With respect to the current chronological sequence for the Southern Cooks as a group, the Moturakau Shelter is quite early, more or less contemporaneous with the Ngati Tiare site on Rarotonga (Bellwood 1978) and the Anai'o Site on Ma'uke (Walter 1987).
Most of the recovered artefacts were basalt debitage flakes related to the manufacture of stone tools, presumably adzes (Table 2). In total, 2,225 pieces of lithic debitage weighing close to 7 kg were recovered from the 1 m2 excavation. No preforms or rejects were found and only one retouched flake was identified during preliminary counting. Clearly, Moturakau was an important basalt resource area throughout much of the island's prehistory. Buck (1927:208) mentions a second traditional basalt source at Black Rock in Aitutaki's Amuri District (there is also a Black Rock Quarry on Rarotonga). Nevertheless, to our knowledge, this is the first archaeologically identified quarry in the Southern Cooks.
Equally significant is the recovery of eight nearly complete one-piece pearl-shell fishhooks (Figures 6 and 7) and 11 smaller fragments (summarised in Table 2). Few fishhooks are known from the Southern Cooks, ethnographically or archaeologically. Captain Bligh (Lee 1920:134) reported collecting fishhooks made from turtle-shell from Aitutaki but the present whereabouts of these specimens is unknown. Gill (1880:212) recorded fishhooks fashioned from coconut and Turbo on Mangaia. Buck (1927:306, 1944:237) believed that pearl-shell was generally unavailable in the Southern Cooks and that prehistorically line fishing was the least important method of obtaining fish. Beasley (1928) attributes four one-piece pearl-shell fishhooks with Pandanus snoods to the Southern Cooks. These are all fairly robust specimens and include three rotating forms and one jabbing hook. Buck (1944:237), however, argues that Pandanus was not used as a fibre in the Southern Cooks and suggests these hooks are actually from Pukapuka. Bellwood (1978:76) illustrates a barbed one-piece pearl-shell hook in the Auckland Museum collections (Acc. No. 972) which is also reputed to be from Aitutaki.
Archaeologically, Bellwood (1978) recovered two one-piece pearl-shell fishhook fragments from the Ureia site. One specimen consists of a bend and suggests a rotating form. The second specimen is composed of a shank and head (Type HT4 according to Bellwood 1978:138, after Sinoto 1968b), and appears to be a jabbing hook. The jabbing hook is associated with Bellwood's date of A.D. 1432 (recalibrated using Stuiver and Reimer 1986) and the rotating hook is somewhat earlier. More recently, Walter (in press) recovered several pearshell hooks from his excavations at Anai‘o on Ma‘uke.
The Moturakau specimens are all pearl-shell (Pinctada), with the exception of Acc. 63 which is made from Turbo. Contrary to Buck's belief (1927, 1944), it appears that Pinctada margaritifera was available locally. Residents showed the authors several specimens which were collected from the Aitutaki- 279
Figure 6:. Fishhook specimens (left to right): Acc. 69 (Layer X), Acc. 77 (Layer XII), and Acc. 70 (Layer X).
Figure 7:. Fishhook specimens: Top row (left to right) Acc. 2 (Layer V), Acc. 63 (Layer VI), Acc. 3 (Layer VIIa). Bottom row (left to right) Acc. 64 (Layer VI) and Acc. 65 (Layer VI).- 280
lagoon. The common archaeological occurrence of pearlshell detritus and manufactured blanks, both at Moturakau and Ureia, are evidence of the local production of pearlshell artefacts and suggest the species was readily available in the past as well. Thus, as Bellwood (1978:137-8) previously argued, there is little reason to suspect that the archaeological specimens of Aitutaki represent imports from the atolls of the Northern Cooks. The same may not necessarily be true of other Southern Cook islands which lack lagoon environments, and the possibility that Aitutaki and Manuae Atoll were important source areas for pearlshell bears further consideration.
Various morphological dimensions of the Moturakau fishhooks are summarised in Table 3. Polynesian fishhook head forms have previously been demonstrated to have chronological significance by Sinoto (1968b, 1975), who analysed Hawaiian collections in detail. The Moturakau head forms were similarly tested for historical significance using a dimensional classification (after Dunnell 1971). Four dimensions of variability were considered, resulting in four filled paradigmatic classes (Table 3); the approximate equivalents in Sinoto's head typology are indicated in parentheses. Dimensional classification has the advantages of: 1) stipulating the necessary and sufficient criteria for class membership; 2) forming equivalent classes defined by the same set of criteria; and 3) accommodating, rather than suppressing, variability (Dunnell 1971).
The results indicate that fishhook head forms are also temporally sensitive in this assemblage. The four Moturakau hooks from Layers V and VI fall into a single class (2312). The head forms of specimens from the lower portion of the deposit are more diverse, representing three classes (1111, 1221, and 1121). In addition to the nearly complete specimens, two of the head fragments in Table 3 could also be classified. One of these from Layer IX belongs to Class 2312, while a second from Layer X belongs to Class 1111. The exceptionally limited evidence at hand suggests a trend from hooks with flat notched heads equal in width to their shanks but variable in other dimensions (Classes 1121, 1111, and 1221), to more morphologically uniform knobbed heads which are wider than their associated shanks (Class 2312). It should be noted, however, that Class 2312 is represented early in the sequence along with the variable flat notched forms. Based on the radiocarbon dates, the Class 2312 head type became the predominant form sometime before A.D. 1410.
The range of fishhook sizes in the collection, from 18 mm to 53 mm, is notable. The temporal distribution of hook size suggests this is largely a functional, rather than stylistic, attribute, as is the case with jabbing versus rotating hook forms (cf. Reinman 1970). Unfortunately, most of the Moturakau specimens lack points, making the differentiation of the rotating dimension somewhat tenuous. The absence of points suggests, however, that these collections probably represent spent hooks, rather than manufacturing rejects or accidental losses.
Further work at the Moturakau Shelter is planned with one primary objective being to obtain a larger sample of fishing gear. The association of these fishhooks with the excellent fishbone assemblages (see below) offers - 282 an unparalleled opportunity for analysis of long-term trends in fishing technology and patterns of faunal exploitation in the Southern Cook Islands. Larger collections are also needed to assess properly historical relationships with other island groups.
A small number of other kinds of artefacts were recovered from the Moturakau deposit and are summarised in Table 2. Specimens probably related to fishhook manufacture include four pearl-shell blanks (two quadrangular, one triangular, and one oval), three coral abraders, and several small, irregularly shaped, fragments of cut and ground pearlshell. Artefacts related to stone tool production were limited to a single hammerstone. Flat rusted metal fragments were common on the surface of the shelter and in Layers I to III; their presence in Layers IV and V is considered to be intrusive.
Faunal materials, primarily fish, were both abundant and well preserved in the Moturakau deposit. At least 14 fish families are represented (after Nelson 1984) in the 3569 gm of bone recovered. The number of naturally occurring inshore fish families for the nearby Society Islands is estimated at 83 (Springer 1982). Families identified during preliminary sorting include Acanthuridae, Balistidae, Belonidae, Carangidae, Diodontidae, Holocentridae, Labridae, Lethrinidae (including Monotaxis), Lutjanidae, Mullidae, Ostraciidae, Scaridae (including Scarus and Calatomus), Serranidae, and Tetradontidae. Elasmobranchimorphi (sharks and rays), and Anguilliformes (eels) are also present. These identifications are based primarly on dentaries and premaxillae, but also on distinctive spines (Acanthuridae), scales (Ostraciidae), vertebrae (sharks and rays), individual teeth (sharks), and fused tooth plates (rays).
Most of the foregoing families are well represented in lagoon environments, along reef edges, and in reef passes (see discussion in Green 1986; Kirch and Dye 1979; Kirch and Yen 1982). The carangids, belonids, sharks, and rays could represent open sea fishing but these taxa also occur with sufficient frequency within shallower waters that such an interpretation is not necessary. No distinctly pelagic taxa were identified in the Moturakau collections, as for example, Scombridae.
Marine molluscs are similarly well represented with over 14.5 kg of material recovered. Table 4 summarises the 29 taxa recovered by layer. All water-worn shell was excluded from consideration and only materials larger than 1/4”screen were identified. Six taxa, including Turbo setosus, Scutarcopagia scobinata, Modiolus auriculatus, Asaphis violascens, and Corculum fragum, dominate the assemblage.- 283
Three primary shellfish habitats are represented in Aitutaki: the intertidal reef flat, the lagoon floor, and the reef rim (Gibbs 1975). Of these, the reef rim environment, approximately one kilometre from Moturakau, has the greatest number of molluscan taxa (Gibbs 1975:131). Tridacna and Turbo are two economically important taxa which are common on the reef edge; Tridacna are also found on coral heads within the lagoon. Many of the bivalve species dwell on the sandy lagoon floor. Asaphis, Modiolus, Quidnipagus, and Gafarium are reported in the broad sand flats exposed along the eastern shore of the mainland at low tide (Gibbs 1975:124). Gibbs notes that Corculum, a lagoon bottom species, is presently rare in Aitutaki, but is a common species in other atoll environments. Archaeologically this taxon is quite common and his suggestion (Gibbs 1975:128) that its current low numbers in Aitutaki might be related to human exploitation is well taken (see Table 4).
The molluscan assemblage suggests that both the shallow lagoon environments adjacent to the islet and the more distant reef edge were being exploited by the shelter occupants. There is a notable temporal trend from predominantly lagoon-floor dwelling bivalves to predominantly gastropods (i.e., Turbo). In Layer XIII, bivalves (excluding Tridacna) compose 97 per cent of the assemblage, while in Layers I to III they average only 26 per cent (Table 4).
This temporal trend in shell composition is graphically illustrated in Figure 8, an ordered similarity matrix (OSM) which compares each layer with every other layer in terms of the relative proportions of different shell taxa (by weight) (see more detailed discussions in Pielou 1979, 1983; Kirch and Yen 1982:318-23; Hunt 1986). OSMs provide a basis for visually assessing the degree and character (i.e, gradual, stepped, or cyclical) of change within a temporal sequence. For Figure 8, similarity was assessed using the Proportional Similarity index (Pielou 1979, 1983). The degree of patterning within an OSM can also be evaluated statistically through Q/Qm values. 2 The stronger the patterning within an OSM, the lower the Q/Qm value will be. Computer simulated primitively random OSMs are then used as comparative standards of randomness (Pielou 1983). Figure 8 reflects a moderately graded sequence, both graphically and in terms of the test statistic (Q/Qm=.328).
As previously suggested, Layers VIIa and VIIb appear to represent a high energy depositional event (see Stratigraphy and Features, above). In terms of their shell composition, they are distinctive in: 1) the large quantities of Tridacna they contain; 2) their taxonomic richness, which is in part a reflection of the volume of sediment they represent relative to most of the other layers; and 3) the differing proportions of other molluscan taxa. Their- 284
For each layer the six most abundant taxa, as determined by per cent weight, are rank-ordered. Other taxa which are present but not among the six most abundant are indicated by “X”.- 285
Figure 8. An ordered similarity matrix comparing each stratigraphic layer with every other stratigraphic layer in terms of the relative proportions of 29 shell taxa.
distinctiveness is graphically apparent in Figure 8. When Layers VIIa and VIIb are removed from the matrix, the grading of the sequence becomes even stronger and Q/QM decreases to .217. Given that there is no readily available source of Tridacna in the adjacent waters, coupled with the number of burned specimens, it is suggested that these shells were redeposited from a midden area outside of the shelter.
Also notable is the similarity of Layer XI with those of the upper portion of the sequence. This is in large part because of the abundance of Turbo, but may also be an artefact of the taxonomic poverty of this layer (see Table 4). - 286 Larger samples are needed to properly evaluate the similarity of this lower layer with those from the upper portion of the deposit.
The sequence outlined here forms a hypothesis for future work. Expanded excavations will not only increase the size of the molluscan sample, but will also provide a firmer foundation for understanding the depositional history of the shelter. More attention will also need to be paid to differentiating midden from naturally deposited shell, particularly in the lower layers where marine deposition has been more important.
Little mammal or bird bone was recovered during this preliminary excavation. The bird bones (identified by David Steadman, of the New York State Museum) include chicken or Gallus gallus (4 specimens), sooty crake or Porzana tabuensis (1 specimen), and brown noddy or Anous stolidus (1 specimen). The four chicken bones were distributed throughout the deposit, from Layer I to Layer XIII. The paucity of domesticates further strengthens the argument that the cultural deposits represent short-term occupations. Rat bone was present throughout. More specific identifications of the vertebrate materials are currently under way.
The Moturakau deposit attests to the long-term importance of this islet to the prehistoric inhabitants of Aitutaki Atoll. Although the degree to which the intensity and duration of usage of Moturakau is tied to its basalt resources, this is a matter for future study. Preliminary evidence from several of the coralline reef islets suggests that they too were used prehistorically. In the case of Tapuaetai, the possibility of a more permanent habitation is intimated by the scatters of basalt flakes which were clearly transported from one of the three volcanic islets.
The basal rockshelter date of 738 B.P. also raises the issue of whether these offshore islets were inhabitable before that time. In the case of Moturakau and Rapota, their volcanic formation dates to the Pleistocene (Wood and Hay 1970:38). In contrast, Stoddart's radiocarbon dates for three coral islets suggest a relatively recent emergence and/or stabilisation. There is accumulating evidence from several South Pacific localities which indicates that sea-level may have been as much as one metre higher 5,000 to 6,000 years ago (e.g., Schofield 1970; Clark and Lingle 1979, Ash 1987, Pirazzoli and Montaggioni 1986; Montaggioni and Pirazzoli 1984). More controversial is the slope of sea-level fall since that time. Clark and Lingle (1979) suggest a gradual fall, while Pirazzoli and Montaggioni (1986) argue, on evidence from the Tuamotus, that sea-level remained high until 1,200 years ago. Although precise measurements were not made during the present study, it - 287 is doubtful that the base of Moturakau Shelter is much above present sea-level; thus sea-level must have been at near-present levels by at least 738 years B.P.
The identification of Moturakau as an important quarry site with a lengthy history of use is quite significant. Undoubtedly the volcanic islet of Rapota was also similarly exploited for its stone resources. The number of additional localities on the main island of Aitutaki with suitable rock materials for adze manufacture has not yet been explored, but will be an important consideration for future field work. The possibility that Aitutaki supplied adzes or raw materials to other islands in the Southern Cooks remains open, particularly in light of Walter's (1987) suggestion that some of the Anai‘o lithic artefacts are imported.
The Moturakau fishhooks constitute one of the larger collections from the Southern Cooks. Buck (1944:237) once argued that line fishing was not an important technique in Southern Cooks. Archaeological evidence, however, suggests that line fishing may have been more important earlier in the prehistoric sequence (Bellwood 1978; Walter 1987; this present report) than at European contact.
As with most other central East Polynesian localities, pearl-shell appears to have been the preferred material for fishhooks. A single Turbo hook was recovered from the Moturakau excavations, midway through the sequence. The use of Turbo in fishhook manufacture has a long history and was the predominant material in the both the Anuta and Tikopia collections (Kirch and Rosendahl 1973; Kirch and Yen 1982). Several one-piece Turbo fishhooks were recently recovered from the early To‘aga site on Olosega, Samoa (Hunt and Kirch, personal communication, 1988). Turbo fishhooks are also known in late prehistoric contexts on Mo‘orea in the Societies (Rappaport et al. 1967:184-92).
While there is evidence for the local occurrence of pearl-shell on Aitutaki, the species is no longer abundant and it was not recorded in the 1969 molluscan survey of Gibbs et al. (1975; see also Gibbs 1975). Three factors may be responsible for the local decline of pearl-shell: 1) lowering sea-levels would have reduced the suitable habitat for Pinctada which prefers deep lagoon waters; 2) over-exploitation by human populations; 3) lagoon infilling via increased terrigenous sedimentation arising from deforestation and subsequent erosion on the Aitutaki proper. With respect to the latter, it is significant that the deepest portions of the lagoon are adjacent to the eastern coast of the mainland (Stoddart 1975b: Figure 13) and would have been particularly susceptible to intensified terrigenous sedimentation.
Stylistically, the Moturakau fishhooks are similar to those recorded from - 288 other East Polynesian sites of comparable antiquity (e.g., Suggs 1961; Emory and Sinoto 1964; Sinoto 1966, 1970; Sinoto and McCoy 1975; Chikamori 1987). Interestingly, the head morphology of the Moturakau fishhooks follows the general temporal pattern identified by Sinoto (1968b) for South Point, Hawai‘i, and for the Society Islands (Sinoto, personal communication). Unfortunately, the South Point assemblage is the only large collection which has been analysed for stylistic frequency distributions. Although Suggs (1961) attempted to seriate his Marquesan fishhooks, most of his types occur in numbers too small to meet the necessary conditions for the seriation model (cf. Dunnell 1970, 1981; Green 1971). It is notable, however, that in the Marquesan assemblages flat, notched heads are predominantly late rather than early.
While head features are apparently stylistic, other morphological attributes (as, for example, shank and bend shapes) appear to have a functional basis (Reinman 1970). In these other dimensions of morphology the Moturakau hooks are fairly uniform. This contrasts markedly with the highly diverse collections from the Marquesas (Suggs 1961; Sinoto 1970) and Mangareva (Green, MS n.d.). Analyses of the Marquesan collections point to a clear reduction in morphological variability through time (Suggs 1961:78-82; Kirch 1980; Dye, in press). Kirch (1980) suggests that the temporal trend of increasing dominance of simple jabbing hooks represents adaptation to local conditions (but see also Dye, in press). Jabbing hooks, Kirch (1980:45) argues, were more effective in rocky coastal regions which lacked developed reefs and lagoons. Given that Marquesan coastal environments are unlike most others in central East Polynesia, the degree to which the Marquesan trends are representative of the region at large is uncertain. Indeed, evidence from other central East Polynesian islands suggests that both jabbing and rotating hooks were important late in prehistory (e.g., Beasley 1928; Rappaport et al. 1967). The Southern Cooks, with their varied coastal environments, may ultimately provide an important case for differentiating functional traits (i.e. those affected by selection) from stylistic ones (i.e., those which are selectively neutral) (cf. Dunnell 1978).
The taxonomic richness of the Moturakau fish assemblage is similar to that from other Polynesian localities (e.g., Kirch 1973; Dye, in press; Leach et al. 1984; Leach and Intoh 1984: Best 1984),. In the Marquesas, Dye (in press) recorded four additional families not represented at Moturakau: Cirrhitidae, Kyphosidae, Polynemidae, and Scombridae. The fish bone collections from Site 197 in Lakeba, Fiji (Best 1984) also provide an interesting comparison given that: 1) this site is also a rockshelter; and 2) collections from site 197 were made with comparable size screens (1.5 mm). At Lakeba, six additional - 289 families were recorded, including Kyphosidae, Nemipteridae, Scombridae, Sphyraenidae, Syngnathidae, and Theraponidae. The last family has apparently not been recorded on the Pacific Plate (Springer 1982). Ostraciidae is the only family represented at Moturakau that was not present at site 197. Taking into account the natural fall-off in species diversity from west to east (Springer 1982), the two assemblages are fairly similar at the family level. The Vitaria, Rurutu assemblages are less rich than those from Moturakau, but the two assemblages have many inshore taxa in common (Leach and Intoh 1984).
These assemblages contrast with those from Fa‘ahia, Huahine, Societies where the emphasis was on pelagic species (Leach et al. 1984). At Fa‘ahia, Scombridae (identified as Thunnidae/Katsuwonidae) is the predominant taxon (based on counts of minimum number of individuals), closely followed by Carangidae. The inshore taxa at Fa‘ahia are similar to those described above for Moturakau and other Polynesian localities. The Moturakau molluscan assemblage is also taxonomically similar to those of other central East Polynesian localities (e.g., Vaito‘otia, Societies — Sinoto and McCoy 1975:181-2; Huahine, Societies — Pigeot 1986:15; Mo‘orea, Societies — Rappaport and Rappaport 1967:201-15; Marquesas — Kirch 1973; Rarotonga, Southern Cooks — Bellwood 1978:34, 72, 137, 148). Turbo and Tridacna are particularly well represented in most of these sites. Pigeot (1986:15) reports that Asaphis was the second most important taxon in their excavations. These three taxa were also important in the Moturakau assemblage.
The Moturakau Shelter represents a specialised activity area, one used for both basalt quarrying and for initial stages of stone tool manufacture, and possibly as a base for procurement of marine resources, most notably fish. As work progresses on the main island of Aitutaki, this site will form an important comparative data base. For the present, however, it provides an unparalleled diachronic record of over 700 years of technology and subsistence in the Southern Cooks. The islet is the first archaeologically recorded quarry for the Southern Cooks and should figure importantly in future technological studies of East Polynesian stone tools (cf. Cleghorn 1984). The association of well-preserved and abundant fish remains with fishing gear and the evidence for on-site fishhook manufacture are equally significant. With larger samples from expanded excavations, this collection will greatly aid our understanding of both functionally based changes in fishing gear and historical relationships between the Southern Cooks and other archipelagos.- 290
We are especially indebted to the late Mr Roy Joseph (former director), Mr Tuakura Tuakura (current director), Mrs Ruru Carl, and Mr Mikaela Tumu, of the Aitutaki Ministry of Internal Affairs, for arranging permissions and logistics for the Moturakau field work, as well as providing Upu Daniel, Teokotai Rio, Tiare Ta‘i, and Kaitai Putu to assist in boat operations. Mrs Carl's generous loan of her family's boat is particularly appreciated. Mr Tony Utanga, Secretary of the Ministry of Internal Affairs (Rarotonga) greatly facilitated the overall project. Dr David Steadman's NSF grant (BSR-8607535) funded the Cook Island field work in general. Mr Dyer Moore made a generous private contribution which aided the Aitutaki work in particular. Sigma Xi, the Scientific Research Society, provided funds to Allen for radiocarbon dating. The Burke Museum provided laboratory space and supplies for analyses. Virginia Butler aided the senior author in preliminary identification of the fish bone and Elsie Marshall shared her expertise in the identification of the mollusc voucher specimens. Patrick Kirch and Terry Hunt are thanked for their critical comments and continuous encouragement. Finally, we thank Dave Steadman for making it all possible.
- 296 Page is blank
1 Metal flakes in Layers IV (n=1) and V (n=4) are probably intrusive.
2 Q is defined as the total number of different pairs of elements for which the larger member of the pair is in a lower-ranking diagonal than the smaller member of the pair. Qm is the maximum possible value of Q. With a sample of 14 cases, as in Figure 8, any Q value below. 425 is considered nonrandom with a probability of error that is less than 10 percent (Pielou 1983).