Geoarchaeological investigations of limestone caves along the Rove Peninsula, where several Lapita-era (1150-750 BC) sites dating from the earliest period of Fiji's human history have been found, was undertaken by a team from the University of the South Pacific and the Fiji Museum. Surface collection and excavation in the largest cave – Qaranibourewa – was hindered by large amounts of ceiling collapse and no trace of human occupation earlier than about AD 1000 was found. The second-largest cave – Qaramatatolu – had a cave fill 190 cm thick but this was determined to be all of recent origin, having accumulated as a result of being washed down through a hole in the cave roof from a settlement above that probably existed AD 750-1250. The shell faunal remains from the Qaramatatolu excavation all suggest an open-coast location, quite different from the mangrove forest that fronts the area today. This mangrove forest probably formed only within the last few hundred years.

Very little is known about the nature of the first humans to occupy the western South Pacific Islands, the so-called Lapita people. This is a final report on the osteological analysis of the skeleton named Mana, which was excavated at a Lapita Culture Complex site called Naitable on Moturiki Island in central Fiji in 2002. The Mana skeleton was reasonably well preserved. The skull is without doubt the best preserved of the Lapita-associated human skeletons ever described. Its major parts were nearly intact and reconstructed to an almost complete state. The skeleton proved to be an approximately 40-60 year old female. Radiocarbon dating of bone from the skeleton, and other archaeological considerations, place the burial around the middle of the first millennium BC (around 700 BC). In the present paper, osteological features of the cranium, mandible and infracranial skeleton of Mana are described very precisely for detailed comparative studies in the future.

Questions concerning the earliest human occupation of northern Fiji were addressed by geoarchaeological survey on the island of Yadua. Yadua lies at the entrance to an ocean passage that early seafarers might have followed into central Fiji where some early Lapita sites exist. Evidence for a Lapita presence was discovered on Yadua at a small coastal flat called Vagairiki, likely to have been occupied by Lapita people around 2600 cal yr BP because of available freshwater and one of the few fringing reefs existing in the area at the time. It is concluded that the Lapita people reached Yadua and other parts of northern Fiji in a post-founder phase of Fiji history.

In November–December 2004 a research team from the University of the South Pacific and the Fiji Museum undertook geoarchaeological investigations along the coast of the Rove Peninsula, part of southwest Viti Levu Island (Figure 1A) where evidence for Lapita-era occupation had been found on previous occasions (Kumar et al., 2004; Nunn et al., 2004). The main target was the extensive, early-period site at Bourewa but we were also shown a collection of pottery from nearby Qoqo Island (by owner Peter Jones) that included a dentate-stamped sherd that led to mapping and excavation of that island's coastal flat. Qoqo is a bedrock island (40,000 m²) reaching 32 m above sea level, located in the 7.3 km² mangrove swamp at the mouth of the Tuva River (Figure 1B). The island comprises two hills surrounded, particularly along their eastern side, by a 20-50 m broad coastal flat that also connects them (Figure 1C). At the time of Lapita arrival in Fiji, sea level was higher (+1.5 m, cal 3000 BP, Nunn 2005) and the hills on Qoqo are interpreted as recently-separated islands connected by a tombolo, the approximate form of which can be reconstructed today (Figure 1C).

In 2003 the authors discovered and excavated a Lapita site at Naitabale close to the southern end of Moturiki Island (central Fiji). Today the site is 350 m inland from the coast, but in Lapita times it was located behind the active beach ridge. A large collection of potsherds (including 92 dentate-stamped or incised Lapita sherds), shell, and animal bones was recovered, together with a human burial. Sherd decorations show affinities with the Western Lapita Province rather than the Eastern Lapita Province (which includes Fiji). Temper analyses of 45 Lapita sherds do not show any unmistakably exotic (to Fiji) pottery, but 29 percent are nonlocal to Moturiki and nearby islands. Fish bones are mostly from inshore species (dominated by Scaridae), while nonfish vertebrates are dominated by turtle and include dog and chicken. Shellfish remains are dominated by gastropods, mostly 'Strombus' spp. (43 percent of gastropod MNI). The surf clam ('Atactodea striata') accounts for 38 percent of bivalve MNI, with 'Anadara antiquata' and 'Gafrarium peetinatum' each representing 14 percent of the bivalve MNI. The skeleton is that of a woman (Mana) 161-164 cm tall who died at 40-60 years of age. Six radiocarbon dates from bones overlap 2740-2739 cal. years B.P. (790-789 B.C.). The mandible lacks antegonial notches but is not a proper rocker jaw. The cranium was better preserved than any Lapita-associated skeleton hitherto described, which allowed the head to be reconstructed. Stable-isotope analyses show that her diet contained significant amounts of reef foods but was probably dominated by terrestrial plants. The Lapita occupation of Naitabale is likely to have begun by 2850 cal. years B.P. (900 B.C.). Radiocarbon dates and pottery decorative styles both suggest Naitabale was first occupied within the early part of the Lapita history of Fiji.

In December 2003, with funding from the University of the South Pacific, a 17-person team led by the three authors conducted research into the early human occupation of the beach and the adjoining coastal flat at Rove, a few kilometres northwest of Natadola in the southwest part of Viti Levu Island. The site at Rove was not selected randomly. In the preceding 18 months, it had been visited twice by Roselyn Kumar who had recovered three pieces of Lapita pottery from the shore flat at low tide. The Lapita era is the earliest in the human history of Fiji (and many other western Pacific island groups), and is often recognized by the distinctively-decorated (so-called dentate-stamped) pottery that was made during that time. The three pieces of Lapita pottery from Rove were described by Kumar et al. (2004a) and were sufficient to make the area worth excavating. Yet the excavations we carried out in December 2003 at Rove were a little disappointing. There was certainly a Lapita settlement at Rove, and it was located on a tiny island off the main island at the time, but the radiocarbon dates showed that it was probably established only quite late in the Lapita history of Fiji, perhaps around 700 BC.