Subpolar settlement in South Polynesia.

Anderson, Atholl

Introduction

The probable extent of Polynesian migration in prehistory reaches well beyond the conventional 'Polynesian Triangle', with its vertices at Hawai'i, mainland New Zealand and Easter Island. To the west of it, there were Polynesian outliers in Melanesia and a village site on Norfolk Island (Anderson & White 2001). Circular shell fish-hooks and associated subsistence changes along the east coast of Australia in contexts dating 1500-500 BP, together with the recovery there of stone adzes of Polynesian type (Thorpe 1929), have attracted conjecture about Oceanic influences (O'Connor & Chappell 2003). Similarly, the first-millennium AD appearance of planked canoes, and earlier of circular shell fish-hooks, in California and northern Chile has been associated with Polynesian or Oceanic influences (Heizer 1949; Heyerdahl 1952: 697-705), although American development of each is now argued respectively by Gamble (2002) and Rick et al. (2002). Further south, prehistoric colonisation seems to have been absent on the Juan Fernandez Islands (Anderson et al. 2002), but Amerindian architectural traits (Martinsson-Wallin 1994) and cultigens on Easter Island are held by Green (1998) to reflect substantial Polynesian voyaging along the coast of South America, and Ramirez-Aliaga (1992) has collated data suggestive of Polynesian contact with south-central Chile.

Two expeditions beyond the southern angle of the Polynesian Triangle have sought to elucidate the subpolar extent of Polynesian migration. The first expedition, to the Snares and Auckland Islands in 1998 (Figure 1), suggested that there had been prehistoric settlement on both (Anderson & O'Regan 2000), but key elements of the data remained uncertain. No artefacts were recovered in cultural stratigraphy, sparse shell and fishbone could have originated in seal scats, and questions remained about the extent of in-built age in radiocarbon-dated charcoal samples. A second expedition in 2003 sets those concerns to rest (Anderson 2003a). There is now unambiguous evidence of a thirteenth-fourteenth-century AD settlement on the Auckland Islands. The new data provide the first evidence of pre-European settlement on outlying islands in the Subantarctic zone (the other groups are Falklands and Gough Islands in the South Atlantic, Crozet and Prince Edward Islands in the South Indian Ocean and Bounty, Antipodes and Campbell Islands in the South Pacific). These results also complete a survey of the colonisation prehistory of the outlying archipelagos of South Polynesia (those lying at 500-800km around mainland New Zealand) and enable a review of some characteristics of that phase.

[FIGURE 1 OMITTED]

Auckland Islands archaeology

The Subantarctic islands lie between the sub-tropical and Antarctic convergence zones. Locations of these vary seasonally, but south of mainland New Zealand their mean positions are at about 47[degrees]S and 57[degrees]S, respectively. For this and other reasons outlined by Anderson (1981), Foveaux Strait and the Chathams, identified by Sutton and Marshall (1980) as 'Subantarctic', are not in fact subpolar. The largest subpolar archipelago (626[km.sup.2]) is the Auckland Islands, rising to about 650m asl, at 50[degrees]S. Cloudy (900 hours sunshine per annum), cool (mean annual temperature of 8[degrees]C) and humid (100-150cm annual precipitation), they support a narrow fringe of coastal forest and abundant marine life (Department of Conservation 1997).

The expedition in 2003 surveyed most of the east and north coast inlets and islands for signs of Polynesian occupation (the west and south coasts are cliffs up to 600m high). Shell and bird-bone deposits occurred in the only substantial cave encountered, at Tagua Bay, Carnley Harbour, but excavation showed them to be of natural origin and they were dated to 2555 [+ or -] 39 BP (Wk-13430). The other area was at Sandy Bay, Enderby Island. A boulder beach ridge there reaches about 2.0m above mean HTM, and it is overlain by a 470m-long foredune up to 1.2m deep, the only substantial area of sand dunes in the archipelago. The dune contains discrete deposits of blackened sand and cultural items which occur as a single layer, generally 0.25m thick, but up to 0.5m, enclosed by lower and upper palaeosols (Figure 2). The lower of these has a maximum age of about 2800 BP (McFadgen & Yaldwyn 1984), and the upper palaeosol, formed after abandonment of prehistoric occupation but before nineteenth-century deforestation and the introduction of domestic stock and rabbits, led to partial remobilisation of the dunes. That process allowed deposition of bottle glass and fragments of clay tobacco pipes on exposed parts of the prehistoric layer (Anderson 2003a).

[FIGURE 2 OMITTED]

Excavation of 5.0[m.sup.2] site at area X (Figure 1) disclosed a Polynesian earth oven, 2m in diameter, containing basalt cobbles, charcoal and abundant midden of mussel shell, and sea lion and bird bone. Midden was sparse nearby, but flakes and cores of chert and basalt were relatively numerous, suggesting an adjacent processing area. Material from the cultural layer was screened to 3mm where the matrix was sandy, but where it was sticky clay-loam, which resisted water sieving, the material was spread out on plastic sheets and picked through carefully by hand.

Test pits showed that this site represents a single phase of occupation on an area of 70-100[m.sup.2]. A similar site of comparable size and contents is largely eroded out at area S, 180m to the east. An uneroded midden there, S5, was test excavated (0.5[m.sup.2]) and the deflated area around it shows that the site comprised a cluster of at least seven ovens and associated middens and flake tools. On its periphery were the oven (area C) and midden scatter (area A), investigated in 1998 (Anderson & O'Regan 2000).

Combined, areas A, C, S and X extend over about 250[m.sup.2]. Systematic coring and spade pits along the entire foredune revealed no other cultural deposits of prehistoric provenance. It is possible that an originally extensive site has been largely lost by water, wind and bioerosion, but as the distribution of flaked chert in deflated dune swales along the foredune correlates closely with the location of the present prehistoric deposits, it is more probable that prehistoric occupation was limited and brief.

Chronology

Potential sources of error in radiocarbon dating include in-built age in charcoal samples, most of them of long-lived rata (Metrosideros umbellata); storage age from the use of driftwood and reservoir age for shell and bone from marine-feeding birds. To minimise these problems, multiple provenances were dated on diverse sample materials (Table 1). With the exception of one unidentified sample (ANU-12038), all dated charcoals were identified to taxa. Seven samples were of small-diameter inanga (Dracophyllum longifolium), which should be of short to medium lifespan, generally 50-80 years but possibly up to 220 years (Anderson & O'Regan 2000). Two samples also included Coprosma c.f. foetidissima (Wk-13652) and rata (ANU-11238), and one was exclusively of rata (ANU-11236A).

Four shell samples were of southern blue mussel (Mytilus edulis galliprovincialis), which contains outer layers of calcite that were ground off. Duplicate samples for Wk-13428 and Wk-13429 were tested for reproducibility and the results were within errors. Three samples were of ribbed mussel (Aulacomya atra maoriana), exclusively aragonite. Two samples of sooty sheatwater (Puffinus griseus) bone gelatine were also dated, and bone and shell conventional ages were corrected for the marine reservoir effect (OxCal v. 3.8, Bronk Ramsey 2001) using the Chatham Islands offset value of 140 [+ or -] 80 radiocarbon years (Waikato Radiocarbon Laboratory unpublished data).

The distribution of radiocarbon ages by site area (Figure 3) shows that areas A, C and S date entirely to an early occupation. In area X, there is a chronological difference between spit 1, which was only partially covered by the upper palaeosol, and spits 2-5 beneath. Shell samples in spit 1 date to the period of historical occupation of Sandy Bay, also represented by the midden in area Y. However, the radiocarbon dates from lower levels in area X indicate prehistoric occupation, contemporary with that in areas A, C and S during the early thirteenth to fourteenth centuries.

[FIGURE 3 OMITTED]

Artefactual and faunal remains

Artefacts from within prehistoric cultural levels included large basalt flakes of a form consistent with adze preform trimming, although neither preforms nor finished adzes have been found. More abundant are flakes and scrapers of a hydrothermal laminated chert which has been broken out of cobbles (Figure 4a and 4b). The basalt is identical to local material in hand specimen, and the source of the chert probably lies amongst the layers of indurated argillites and clays which occur between basalt flows near Sandy Bay.

[FIGURE 4 OMITTED]

Shell midden consisted mainly of southern blue mussel, ribbed mussel and the limpet, Cellana strigilis strigilis. The virtual absence of softshore or low-tidal species such as paua (Haliotis virginea huttoni) suggests a focus on the rocky mid-shore. The most abundant inshore fish locally are the Ice-cods, Paranotothenia spp. (Kingsford et al. 1989). Six P. microlepidota are represented in spit 1, area X, along with four conger eels, probably Conger verreauxi. However, this material is probably of historical age (above). In the lower spits, there is only one identified individual of fish (P. microlepidota) and another at area S, so fishing may have been relatively unimportant prehistorically.

The Sandy Bay prehistoric assemblages include 124 birds, by MNI, of which the top five taxa (MNI, per cent) are white-chinned petrel, Procellaria aequinoctinialis (28, 23), sooty shearwater, Puffinus griseus (26, 21), Auckland Island shag, Leucocarbo colensoi (19, 15), southern royal albatross, Diornedea epomophora (12, 10), and yellow-eyed penguin, Megadyptes antipodes (9, 7.3). As in the fish, there is a substantial difference between taxa in assemblages of different age. The main taxa in spit 1, area X, are white-chinned petrel (19, 31), Auckland Island shag (17, 28), sooty shearwater (8, 13) and white-headed petrel (5, 8). In spits 2-5, area X, plus area S5, the main taxa are sooty shearwater (16, 43), yellow-eyed penguin (7, 14), white-chinned petrel (6, 14) and southern royal albatross (5, 10).

The prehistoric material thus indicates stronger targeting of the muttonbird (sooty shearwater), the largest available taxa (albatross and penguin) and the facultatively flightless Auckland Island teal (Anas aucklandica), of which three individuals are recorded only in this assemblage. The general absence of local landbirds, such as tui (Prosthemadera novaeseelandiae), parakeets (Cyanorharnphus spp.) and a rail (Rallus muelleri), suggests a bias against fowling in the wet coastal forest. The prominence of the shearwaters, petrels and albatross indicate fowling during the September to May breeding period of these species in the Auckland Islands, but occupation during the winter cannot be ruled out.

Mammal bone assemblages are split between Hookers sea lion (Phocarctos hookeri, MNI = 7), which breeds at Sandy Bay, and New Zealand fur seal Arctocephalus forsteri, MNI = 7), which breeds elsewhere on Enderby Island. Sea lion pup bone indicates local breeding some 650 years ago, and capture during the summer months. Fur seal bone is dominant in spits 1 and 2 of area X, but it is replaced by sea lion bone in the lower spits. This might reflect either a cultural impact on the Sandy Bay sea lion colony or a seasonal effect, since fur seals remain once sea lions have largely departed following breeding. Some pieces of bone had been chewed in patterns characteristic of dogs (Canis familiaris), the first evidence that dogs reached the Subantarctic islands prehistorically (Figure 5). Remains of 14 seals and 124 large birds, amongst those of fish and shellfish, within an excavated volume of 1.6[m.sup.3], typify the high faunal density in colonisation sites.

[FIGURE 5 OMITTED]

Subpolar settlement and South Polynesia

About 650 years ago, Polynesians and their dogs settled at Sandy Bay in the subpolar Auckland Islands, during at least one summer--autumn period, but probably for only a few years at most. They hunted sea lions, fur seals and nesting seabirds, but there was little foraging in the sea or coastal forest. Chert cobbles were fashioned into flake tools and basalts tested for adze manufacture. Once habitation ceased (there is no indication yet of whether the people left or died out), there was no further settlement until after the European discovery of the islands in AD 1807. When settlement resumed, it was by a group of Maori and Moriori in 1842-1856. They, and British colonists, 1849-1852, abandoned the islands in the face of harsh environmental conditions (Dingwall et al. 1999). Further exploration is planned to determine whether prehistoric Polynesians reached other subantarctic islands, notably Campbell Island and the Antipodes.

However, the 2003 expedition to the Auckland Islands completes the first archaeological survey of prehistoric colonisation in each of the outlying archipelagos of South Polynesia: the Chathams, Kermadecs, Norfolk, Lord Howe and Subantarctic groups. As such, it invites a brief consideration of initial colonisation patterns in the region.

The data show considerable differences in settlement duration. The Chathams (700km east of New Zealand) were inhabited continuously, there was relatively substantial settlement, marked by the existence of small villages, in the Kermadec and Norfolk Islands (800kin north-east and north-west of New Zealand, respectively) but only brief habitation in the Subantarctic and none on Lord Howe Island (1200km west of New Zealand; Anderson 2003c). Leaving aside contingencies of survival in small colonies, both demographic and social, it might be thought that differences in available resources had been an important factor, especially since the South Polynesian islands are distributed over 21 degrees of latitude (29-50[degrees]S). Yet, foraging regimes were remarkably similar everywhere, with muttonbirding prominent in every case (Anderson 1996), and sealing evident even in the subtropical archipelagos. It was too cold to grow Polynesian cultigens such as sweet potato in the Chathams or Subantarctic, but there is scarcely any evidence to suggest that horticulture was available to the early subtropical colonists either; no remains of cultigens or associated artefacts or structures are known from the archaeology of the Kermadecs or Norfolk. In the Subantarctic, the miserable climate and virtual absence of plant foods to alleviate a diet of seals and birds, must have been a discouragement to long-term habitation. Otherwise, island size may have been important demographically. Excepting the Auddands, the Chathams are 30 to 60 times the size of each of the other outlying groups.

The pattern of initial colonisation in South Polynesia can be inferred from the distribution of sourced obsidians (Anderson 2000). These show that the Chathams and Kermadecs were settled directly from New Zealand, while Norfolk Island was settled from New Zealand via the Kermadecs. The Auckland Islands were probably settled from southern New Zealand via the Snares, from which an adze of early type has been recovered (Anderson & O'Regan 2000). Rather than progressive colonisation from the more congenial north to the cold south, as might have been expected in the light of tropical East Polynesian origins, the pattern was therefore essentially radial, expanding in all directions from mainland New Zealand.

Colonisation occurred at virtually the same time everywhere and at a period indistinguishable from that of the initial settlement of New Zealand (Anderson 1991). Prehistoric dates for Sandy Bay are virtually the same as those for the earliest known sites in the Kermadecs (Higham & Johnson 1996) and Norfolk Island (Anderson & White 2001). Current earliest dates for the Chathams are later, c. 450 BP, but there are unexcavated sites which contained artefacts indicative of settlement several hundred years earlier (Duff 1956: 118). At an archaeological timescale, the South Polynesian dispersal was thus almost instantaneous. In this, it conforms to the pattern of very rapid dispersal evident both in East Polynesia, c. 1100-900 BP (Anderson & Sinoto 2002), and the c. 3300-2800 BP Lapita expansion in the west Pacific (Anderson 2003b).

Colonisation in the Pacific sector of the subpolar zone is not as unexpected as it would seem elsewhere. Polynesian sailing technology was more advanced than that in southern South America, for example, so that although the Falklands are a larger and closer target, they were probably out of reach until the advent of European ships. Nevertheless, they lie in the circumpolar West Wind Drift and it is not impossible that they were reached accidentally; it would be much more surprising if prehistoric seafarers had reached subpolar islands in the Indian Ocean.

Acknowledgements

Thanks to Nga Runanga o Murihiku and Te Runanga o Ngai Tahu for approval to work in the Subantarctic islands, and to Rachel Egerton, Paul Dingwall, Jeremy Carroll, and Andy Cox of the Department of Conservation (New Zealand), who arranged the 2003 expedition. I thank Katherine Szabo, Alan Tennyson, Ian Smith, Richard Walter and Rod Wallace for their analytical reports and Fiona Petchey for help with radiocarbon matters. The Centre for Archaeological Research (ANU) and the Department of Conservation provided financial assistance with radiocarbon dating. Lyn Schmidt assisted with the illustrations.

Received: 17 May 2004; Accepted: 12 October 2004; Revised: 13 October 2004

References

ANDERSON, A.J. 1981. The value of high-latitude models in south Pacific archaeology: a critique. New Zealand Journal of Archaeology 3: 143-60.

--1991. The chronology of colonization in New Zealand. Antiquity 65: 767-95.

--1996. Origins of Procellariidae hunting in the southwest Pacific. International Journal of Osteoarchaeology 6: 1-8.

--2000. Implications of prehistoric obsidian transfer in South Polynesia. Bulletin of the Indo-Pacific Prehistory Association 20: 117-23.

--2003a. Prehistoric archaeology in the Auckland Islands, New Zealand Subantarctic region. Report to Department of Conservation, Wellington.

--2003b. Initial human dispersal in remote Oceania: pattern and explanation, in C. Sand (ed.) Pacific archaeology: assessments and prospects: 71-84. Noumea: Service des Musees et du Patrimoine.

--2003c. Investigating early settlement on Lord Howe Island. Australian Archaeology 57: 98-102.

ANDERSON, A.J., S. HABERLE, G. ROJAS, A. SEELENFREUND, I.W.G. SMITH & T. WORTHY. 2002. An archaeological exploration of Robinson Crusoe Island, Juan Fernandez Archipelago, Chile, in S. Bedford, C. Sand & D. Burley (ed.) Fifty years in the field: essays in honour and celebration of Richard Shutler Jr's archaeological career. 239-49. New Zealand Archaeological Association Monograph 25.

ANDERSON, A.J. & G. O'REGAN. 2000. To the final shore; prehistoric colonisation of the Subantarctic islands in South Polynesia, in A.J. Anderson & T. Murray (ed.) Australian archaeologist: collected papers in honour of Jim Allen: 440-54. Canberra: Coombs Academic Publishing.

ANDERSON, A.J. & Y.H. SINOTO. 2002. New radiocarbon ages of colonization sites in East Polynesia. Asian Perspectives 41: 242-57.

ANDERSON, A.J. & J.P. WHITE (ed.). 2001. The prehistoric archaeology of Norfolk Island, southwest Pacific. Records of the Australian Museum, Supplement 27.

BRONK RAMSEY, C. 2001. Development of the radiocarbon calibration program OxCal. Radiocarbon 43: 355-63.

Department of Conservation. 1997. Subantarctic islands heritage. Wellington.

DINGWALL, P.R., C. FRASER, J.G. GREGORY & C.J.R. ROBERTSON (ed.). 1999. Enderby settlement diaries: records of a British colony at the Auckland Islands 1849-1852. Wellington: Wild Press and Wordsell Press.

DUFF, R. 1956. The Moa-hunter period of Maori culture. Wellington: Government Printer.

GAMBLE, L.H. 2002. Archaeological evidence for the origin of the plank canoe in North America. American Antiquity 67: 310-15.

GREEN, R.F. 1998. Rapanui origins prior to European contact, the view from eastern Polynesia, in P.V. Casanova (ed.) Easter Island and East Polynesian prehistory: 87-110. Santiago: University of Chile.

HEIZER, R.F. 1949. Curved single-piece fishhooks of shell and bone in California. American Antiquity 15: 89-97.

HEYERDAHL, T. 1952. American Indians in the Pacific: the theory behind the Kon-Tiki expedition. London: Allen & Unwin.

HIGHAM, T.F.G. & L. JOHNSON. 1996. The prehistoric chronology of Raoul Island, the Kermadec group. Archaeology in Oceania 31: 207-13.

KINGSFORD, M.J., D.R. SCHIEL & C.N. BATTERSHILL. 1989. Distribution and abundance of fish in a rocky reef environment at the subantarctic Auckland Islands, New Zealand. Polar Biology 9: 179-86.

MARTINSSON-WALLIN, H. 1994. Ahu--the ceremonial stone structures of Easter Island. Uppsala: Aun 19.

McFADGEN, B.G. & J.C. YALDWYN. 1984. Holocene sand dunes on Enderby Island, Auckland Islands. New Zealand Journal of Geology and Geophysics 27: 27-33.

O'CONNOR, S. & J. CHAPPELL. 2003. Colonisation and coastal subsistence in Australia and Papua New Guinea: different timing, different modes? in C. Sand (ed.) Pacific archaeology: assessments and prospects: 17-32. Noumea: Service des Musees et du Patrimoine.

RAMIREZ-ALIAGA, J.-M. 1992. Contactos transpacificos: un aceramiento al problema de los supuestos rasgos polinesicos en la cultura mapuche. Clava 5: 41-74.

RICK, T.C., R.L. VELLANOWETH, J.M. ERLANDSON & D.J. KENNETT. 2002. On the antiquity of the single-piece shell fishhook: AMS radiocarbon evidence from the southern California coast. Journal of Archaeological Science 29: 933-42.

SUTTON, D.G. & Y.M. MARSHALL 1980. Coastal hunting in the Subantarctic zone. New Zealand Journal of Archaeology 2: 25-49.

THORPE, W.W. 1929. Evidence of Polynesian culture in Australia and Norfolk Island. Journal of the Polynesian Society 38: 123-26.

Atholl Anderson, Centre for Archaeological Research, Australian National University, Canberra ACT 0200, Australia (Email: aja@coombs.anu.edu.au)

Table 1. Auckland Islands archaeological radiocarbon dates

Provenance/Type Lab. No. CRA

Charcoal samples
Location A: Midden Site
 Loc.A/1 ANU-11088 780 [+ or -] 60
Location C: Oven site
 Loc.C/C6/1 ANU-11085 840 [+ or -] 60
 Loc.C/C4/2 ANU-11086 660 [+ or -] 70
 Loc.C/C4/2 ANU-11087 620 [+ or -] 60
 Loc.C/C4/2 ANU-11236A 800 [+ or -] 50
 Loc.C/Section ANU-11238 770 [+ or -] 70
Location X: Occupation site
 AA1/2/4 ANU-12038 1030 [+ or -] 70
 AA1/2/5 Wk-13651 701 [+ or -] 46
 AB2/2/4 Wk-13652 658 [+ or -] 47
 55/1/2 Wk-13653 720 [+ or -] 43
Location Y: Midden site
 Loc.Y/1 ANU-11089 190 [+ or -] 60
 Marine shell samples
Location S: Midden site
 S5/1/2 ANU-12035 1200 [+ or -] 70
Location X: Occupation site
 AA/2/4 ANU-12039 1280 [+ or -] 60
 T/pit DD Wk-13426 676 [+ or -] 30
 AC1/2/1 Wk-13427 661 [+ or -] 42
 AA1/2/3 Wk-13428 1093 [+ or -] 48
 AAI /2/3 Wk-13428 1126 [+ or -] 43
 A132/2/4 Wk-13429 1174 [+ or -] 43
 AB2/2/4 Wk-13429 1115 [+ or -] 43
 AB1/2/1 Wk-13431 581 [+ or -] 42
Bird bone samples
Location S: Midden site
 S5/1/2 Wk-13441 1216 [+ or -] 43
Location X: Occupation site
 AA1/2/4 Wk-13440 1289 [+ or -] 43
 * estimated value

 [[partial
 derivative]
Provenance/Type .sup.13] 68% 95%

Charcoal samples
Location A: Midden Site
 Loc.A/1 -24 * 1221-1288 1165-1303
Location C: Oven site
 Loc.C/C6/1 -24 * 1165-1276 1036-1289
 Loc.C/C4/2 -24 * 1286-1398 1248-1422
 Loc.C/C4/2 -24 * 1297-1405 1283-1431
 Loc.C/C4/2 -27.4 1211-1279 1159-1291
 Loc.C/Section -24 * 1215-1291 1070-1385
Location X: Occupation site
 AA1/2/4 -24 * 904-1034 887-1126
 AA1/2/5 -25.7 1280-1390 1260-1400
 AB2/2/4 -25.4 1295-1395 1280-1410
 55/1/2 -25.7 1270-1390 1250-1400
Location Y: Midden site
 Loc.Y/1 -24 * 1657-1954 1638-1955
 Marine shell samples
Location S: Midden site
 S5/1/2 0.0 * 1253-1414 1136-1469
Location X: Occupation site
 AA/2/4 0.0 * 1180-1326 1048-1421
 T/pit DD 0.7 1680-1880 1644-1950+
 AC1/2/1 1.7 1690-1890 1651-1950+
 AA1/2/3 1.3 1320-1460 1250-1550
 AA1/2/3 1.1 1310-1440 1230-1510
 A132/2/4 1.2 1280-1420 1230-1510
 AB2/2/4 1.2 1310-1415 1240-1520
 AB1/2/1 1.2 1803-1950+ 1710-1950+
Bird bone samples
Location S: Midden site
 S5/1/2 -15.4 1110-1230 1060-1270
Location X: Occupation site
 AA1/2/4 -15.7 1030-1160 1010-1200
 * estimated value