A thousand years of lunch has been sitting off the north coast of Fiji’s second-largest island, quietly passing itself off as scenery.

The patch of ground off Culasawani, on the northern shore of Vanua Levu, covers about 3,000 square metres and rises to as much as 60 centimetres above the high-tide mark. Mangroves crowd it. A creek slips past. Push an auger into it, though, and there is almost no soil down there. Between 70 and 90 per cent of the deposit is shell, in a layer 20 to 40 centimetres thick.

Nearly all of it came out of somebody’s dinner.

Crabs got there first

No trench, no lab breakthrough. The site gave itself away in 2017 because burrowing mud crabs had been hauling shell fragments and scraps of pottery up from 30 to 50 centimetres down and piling them on the surface.

Two seasons of fieldwork in 2024 followed: mapping, 20 hand-auger cores, four test pits of one square metre. Geographer Patrick Nunn and colleagues lay out the results in a paper published in Geoarchaeology.

Their title asks the only question that matters: midden, or muddle?

What the shells rule out

Could a tsunami have piled all this up instead? Northern Vanua Levu sits in a seismically active part of Fiji, so the possibility needed ruling out. But tsunami deposits tend to carry stones, seafloor rubble and plenty of marine life nobody would eat, and they thin out as they push inland rather than stopping dead at a shoreline.

Culasawani turned out to be fussy. Anadara clams dominate the deposit, alongside other edible bivalves and sea snails. The shell layer is confined to the island itself, with no tapering trail running across the mainland. Ten radiocarbon dates cluster around 760 CE, with the full spread running from roughly 420 to 1040 CE. Plain potsherds appear throughout, matching the later pre-modern Fijian earthenware tradition, which suggests people were using vessels right there while the pile grew.

Nunn and his co-authors call the midden reading their preferred interpretation, not a settled conclusion. One site, one paper.

Same centuries, other continents

760 CE is a handy anchor for anyone who likes their history stacked side by side. In the Petén lowlands of Guatemala, Tikal was near the height of its power. Britannica places the city’s Late Classic period at 600 to 900 CE, when its great plazas, pyramids and hieroglyphic monuments went up and its population grew into the tens of thousands.

Up in the North Atlantic, Norse sailors were still the better part of a century from reaching the Faroes, and further again from settling Iceland. That timeline keeps getting untidier. Sedimentary DNA and faecal biomarkers pulled from a Faroese lake bed, published in Communications Earth & Environment by Lorelei Curtin and colleagues, put sheep on the islands around 500 CE, some three centuries before any Viking longship arrived. Sheep are not native and cannot swim there. That’s a single lake core from a single valley, and whether earlier settlement reached beyond it is still unknown.

Three unconnected coastlines, all busy, none of them aware of the others.

What people were doing there

So who lived on the shell island? Going by the evidence, nobody. Excavation turned up no animal bones, no fish remains and no stone tools, which is a strange shortlist for a settlement. The likelier scenario, set out in coverage by Arkeonews, is an open-air processing station: gather shellfish from the shallows, prise them open on the spot, drop the shells, carry the meat home in a pot.

Do that on a Tuesday and you have made rubbish. Do it every week for three hundred years and you have made real estate.

Seafood has never been a sideline in this part of the world. Shellfish still accounts for around 15 per cent of what some western Pacific communities eat, as reported by Phys.org, and middens across the region are the main record archaeologists have of how coastlines and diets shifted together.

Vanua Levu has had far less archaeological attention than the rest of Fiji, which is partly why a whole island made of shells went unexamined until crabs started redecorating. If the interpretation holds, this is the first landform of its kind described anywhere in this stretch of the Pacific. Similar deposits may simply be sitting elsewhere in these waters, unnoticed, because nobody has drilled them yet.