Buang Merabak: Early Evidence For Human Occupation In The Bismarck Archipelago, Papua New Guinea

This paper reports new radiocarbon estimates for the age of human occupation of Buang Merabak, an archaeological site in central New Ireland, Papua New Guinea (Fig. 1). Previously, the oldest radiocarbon date for human occupation in New Ireland was 35,410 + 430 BP (Leavesley and Allen 1998:80). The radiocarbon determinations reported here, although preliminary, may extend the first evidence of human occupation in New Ireland to beyond 40,000 BP (uncalibrated) and indirectly support the evidence presented by Groube et al. (1986) and Chappell et al. (1994), for the occupation of the Huon Peninsula at a similar antiquity.

The radiocarbon determinations reported here, although preliminary, may extend the first evidence of human occupation in New Ireland to beyond 40,000 BP (uncalibrated) and indirectly support the evidence presented by Groube et al. (1986) and Chappell et al. (1994), for the occupation of the Huon Peninsula at a similar antiquity.

Background
The archaeological potential of the Buang Merabak cave site was first identified by Allen et al. (1984:13) and later excavated under the auspices of the Lapita Homeland Project (Allen and Gosden 1991). The mouth of the cave is ca.10 m wide and the roof is ca. 8 m high. The cave extends back ca. 30 m and is joined to a second larger chamber by a small passage.
It is situated at the base of a series of transgressive fringing coral terraces formed during the lower Miocene to Pliocene, known as the Lelet limestone (Hohnen 1978:2). The initial excavations indicated a substantial deposit of midden material and stone artefacts directly on top of the corroded limestone bedrock, 165 cm below the present ground surface (Rosenfeld 1997). There was no culturally sterile deposit between the cultural material and the bedrock. A radiocarbon determination on shell previously collected from the site, ANU-6614, returned an age of 31,990 + 830 BP for the lowest level of human occupation (Balean 1989). A subsequent conjoin analysis of midden bone fragments tested the integrity of the deposit and determined that, although the uppermost layer of the site was disturbed, the Pleistocene layers were largely intact (Leavesley and Allen 1998). The analysis concluded that much of the deposit remained relatively free of the postdepositional processes that might have caused the vertical redistribution of the cultural material.

me 2000 excavation
The site was re-excavated by Leavesley as part of a project to investigate the nature of Pleistocene colonisation and subsequent economic developments and interactions in the Bismarck Archipelago. The re-excavation was undertaken utilising 5 cm spits and bedrock was identified at a depth of 2 m. The excavation identified a deeper deposit and a potentially richer midden than the previous 1985 excavation. This allowed for the dating of cultural material at a depth not previously possible. N o shell samples collected from the prehistoric midden deposit were submitted for radiocarbon determination. The shells explicitly fit the criteria described elsewhere as shell midden material. Gosden and Robertson (1991) describe the nature of a southern New Ireland shell midden as containing larger individuals of the larger available taxa in the earliest layers of occupation (Gosden and Robertson 1991, Rosenfeld 1997). The pattern of midden shell deposition is repeated at both Matenkupkum and Buang Merabak (Spriggs 1997:37) and the shells utilised for this analysis were consistent with it. The determinations reported here were from shells excavated from Spit 40, immediately above the bedrock.

Samples for anafysis
The samples consisted of two complete shells, one Turbo argyrostoma and the other Purpurapersica. Prior to analysis, -500rng of carbonate was cut fiom the columella of each shell, and the surface cleaned by dissolving approximately 50% of the sample in dilute hydrochloric acid. The sample was then washed with milliQ water, dried, and loaded, along with -5mls of anhydrous phosphoric acid into the low-blank target preparation line at the Research School of Earth Sciences, ANU (Bird et al. 1999).
After evacuation of the extraction line, the phosphoric acid was added incrementally, and the carbon dioxide released fiom the sample was collected sequentially during the dissolution. Two graphite targets were made from each sample, one from the carbon dioxide liberated at -40-60% dissolution, and a second at -80-95% dissolution. The 14C/13C ratios of the graphite targets were measured by accelerator mass spectrometry on the 14UD accelerator at the ANU.

Results
The Purpurapersica sample returned ages of 33,270 k 560 and 32,440 * 570 (see Table 1).  Roberts et al. 1990), these are the oldest radiocarbon occupation dates obtained fiom sites in New Guinea and the Bismarck Archipelago region. The disparity between the two sets of determinations suggests the site was the subject of both natural and cultural formation processes. While any interpretation of the determinations is preliminary, there are a number of observations that support the conclusion that the dates represent human behaviour: (i) Humans were the only terrestrially based predators that fiequently prey on shellfish in New Ireland. Prior to the human colonisation, there weren't any natural, cave-dwelling, terrestrially based predators of shellfish. Before ca. 20,000 BP, of bird taxa have also been identified in New Ireland (Steadman et al. 1999). It is unlikely that any of these taxa would prey on shellfish, particularly shellfish of the sizes indicated in Table 1. (ii) Post-depositional processes almost certainly had a role in the formation of the site. There were no sterile deposits between the cultural layers and the bedrock. Therefore, the shell could not have moved fiom a non-cultural to a cultural layer by way of post-depositional processes. (iii) The possibility of a sub-fossil shell finding its way into the site must also be considered. The likelihood of a marine shell surviving in a coastal marine environment for ca. 5000 to 8000 years, fiom the time of the mollusc's death, until collection by humans, appears unlikely in the extreme although cannot ultimately be discounted. (iv) There is little evidence to support the suggestion that 'old shell' may have been brought to the site. During the Late Holocene, 'old shell' (specifically Tridacna sp.) has elsewhere been utilised for the manufacture of shell artefacts (Moir 1990). The shell specimens submitted to the radiocarbon laboratory in this study were collected fiom a matrix of food refuse debris consistent with other New Ireland shell middens (Gosden and Robertson 1991) and did not exhibit evidence of manufacture.
The presence of two specimens with divergent ages in the same excavation unit may be the result of one of three processes. First, the ages may reflect a slow rate of deposition at the site after its initial occupation around 40,000 BP. This is supported by the presence of relatively large individual shells in the lower levels of the site (Balean 1989). Second, the data may be indicative of some mixing of the archaeological deposits within the lower layers. Or, third, a shell in the order of ca. 4500 to 8000 years older than the rest of the deposit may have made its way into the midden as a result of complex and as yet unknown processes. While present evidence tends towards human causation, further investigations will clarifL the situation.

Conclusion
While the data presented here is problematic it presents an extremely interesting possibility of the first humans arriving in New Ireland by ca. 39,000-40,000 years ago. If so, it suggests that occupation of the Bismarck Archipelago may have been the result of deliberate voyaging (Allen and Gosden 1996: 184) and taken place without any significant pause once humans had colonised the Huon Peninsula. Alternatively, if hrther analyses conclude humans did not occupy Buang Merabak prior to ca.