Arrow-poison traces turn up on stone points from a 125,000-year-old Arabian lake

Stone points picked up on the shore of a vanished lake in southern Saudi Arabia carry chemical traces of the arrow-poison tree, a shrub that African hunters have long used to tip their weapons. If the dating holds, they are the oldest poisoned hunting weapons yet identified, roughly twice the age of the previous record from South Africa.

Poison changes what a hunter can do. A small wound from a poisoned tip can bring down an animal far larger than the person who threw it, as long as someone knows which plant to use and how to prepare it. And the animals in question lived in what is now the Empty Quarter, one of the driest places on Earth: hippos, buffalo, camels and ostriches gathered around a freshwater lake.

The site, called ‘Uruq Aljelibah, sits on the former shore of the Mundafan palaeolake. A team led by archaeologist Chris Clarkson of Griffith University in Australia, geologist Frans van Buchem of King Abdullah University of Science and Technology (KAUST) and the Saudi Heritage Commission dates the lake to about 125,000 years ago, a humid phase researchers call “Green Arabia”. Their study appears in the journal Science.

The surface of the old lake bed yielded 5,511 stone tools longer than two centimetres, more than 98% of them made of flint, together with over 500 fossil specimens from more than 130 locations. The animals include hippos, an extinct long-horned buffalo, camels, Arabian oryx, gazelles and ostriches. Satellite terrain models show that the lake covered about 340 square kilometres (130 square miles) and reached 29 metres deep.

“At 125,000 years old, these would be substantially older than the poisoned arrowheads recently identified in South Africa, which date to around 60,000 years ago,” Clarkson told Live Science. “But we haven’t demonstrated exactly how the poison was prepared or whether it was effective. So I’d describe these as potentially the oldest evidence for poisoned hunting weapons, rather than an unequivocally established first.”

How a stone tip records the weapon it belonged to

The team surveyed the site on a grid, dated the lake with two methods applied to sand and to fossil teeth, and fitted broken flakes back together in 3D to see where tools were made. The stone was knapped with the Levallois technique, in which a toolmaker shapes a core first and then strikes off a flake of a planned shape. Some of the flakes fit back together, which means the tools were made on the spot rather than carried in.

The decisive evidence came from microscopes and chemistry. Several points show impact damage of the kind a tip suffers when it hits a large animal. Traces of animal tissue survive near the tips, and plant resin and fibres cling to the bases, where a point would have been bound to a wooden shaft. That combination marks them as hafted weapon tips, parts of composite weapons made from stone, wood, resin and cord.

“The residues give us clues to parts of these weapons that rarely survive,” said Carney Matheson of Griffith University, a co-author. “Animal tissue near the tips and resin and fibres at the bases help us reconstruct how the points were mounted and used.”

On several points the chemists found biomarkers linked to Acokanthera schimperi, and other potentially toxic plant compounds turned up on weapon points as well. The stone tools cluster in the same places as the animal bones, which suggests the shore served as a place to hunt and butcher game.

What the arrow-poison tree does

Acokanthera schimperi grows in eastern and central Africa and in southern Arabia. Its toxins are cardiac glycosides, the same family of compounds as the heart drug digoxin, and in large doses they disrupt the heartbeat. African hunters have long coated arrowheads and spear tips with extracts of the plant to paralyse or kill game.

That raises an obvious question: how do hunters eat an animal they have poisoned? “A poison introduced directly into a wound doesn’t necessarily behave in the same way when swallowed,” Clarkson said, adding that hunters throughout history are known to cut away the meat around the wound before eating the rest.

The residues also point to a broad use of plants. Use-wear and chemical traces show that people at the lake processed tubers, shrubs and palm leaves, worked wood and twisted plant fibres. “These weren’t just people knocking stone flakes off cores,” Clarkson said. “They were combining stone with wood, plant fibers, resins and potentially toxic substances to make complex tools and weapons.”

Where the site fits in the story of early humans

The previous claim for the oldest poisoned weapons came from South Africa, where arrow tips dated to about 60,000 years ago carried plant poison. Wooden spears and stone-tipped weapons are far older than either site, but poison adds a layer of knowledge: which plant, which part, how to concentrate it and how to handle it safely.

No human bones were found at ‘Uruq Aljelibah, so the identity of the toolmakers is an inference. The team attributes the site to Homo sapiens. “We know that H. sapiens had already expanded out of Africa and were present in Southwest Asia during this period,” Clarkson said. “The age of the site fits with those early movements.”

The finding also sharpens a debate about where people could live in Arabia during wet phases. Earlier this month, a separate study in Nature by a team including Jeffrey Rose of the University of Algarve crossed the interior of the Empty Quarter, logged 55 ancient lake beds and found no stone tools at all, and argued that the sand sea lacked rock to make them. Mundafan lies on the southwestern edge of the desert, and nearly all of its tools are flint, so raw material was within reach there.

“What our site shows is that people weren’t necessarily just passing through these environments,” Clarkson said. “They were exploiting them in quite sophisticated ways, using a range of technologies and resources.”

What the study doesn’t settle

The biggest caveat concerns the date. Every stone artefact in the study was collected from the surface. “Not one was excavated from the layers that were dated,” Rose, who was not involved in the research, told Live Science. “The 125,000-year date belongs to the lake, not to the tools. For the tools, it is only a maximum age.” The tools could therefore be younger than the lake, which would weaken the claim to the poison record.

The poison itself is inferred from chemical biomarkers. The team has not shown how the toxin was prepared or whether it worked, as Clarkson acknowledged. Rose still called the site “valuable new data”, noting that Arabian sites with both stone tools and animal bones are rare.

Eleanor Scerri, an archaeologist at the Max Planck Institute of Geoanthropology who was also not involved, praised “the very convincing spatial relationship between hunting-related tools and the bones of animals such as hippos”. What the evidence cannot yet do is tie a single poisoned point to a single dated layer.

Common questions about the Arabian poisoned points

What is the oldest evidence of poisoned weapons?

The stone points from ‘Uruq Aljelibah in Saudi Arabia are the oldest candidates, at up to about 125,000 years old. Because the tools were collected from the surface, that figure is a maximum age. The previous oldest example is a set of arrow tips from South Africa dated to about 60,000 years ago.

What is the arrow-poison tree?

Acokanthera schimperi is a shrub or small tree native to eastern and central Africa and southern Arabia. It contains cardiac glycosides, toxins that affect the heart. Hunters in Africa have used extracts from it to coat arrowheads and spear tips.

Were there hippos in Saudi Arabia?

Yes. During humid phases such as the one about 125,000 years ago, lakes and grasslands spread across parts of Arabia. Fossils from the Mundafan lake include hippos, an extinct long-horned buffalo, camels, oryx, gazelles and ostriches.

Who made the tools at ‘Uruq Aljelibah?

No human remains were found, so the makers are not known for certain. The researchers attribute the tools to Homo sapiens, which had already spread from Africa into Southwest Asia by that time.

The fieldwork took place during the 2023 season with support from the Saudi Heritage Commission, and the paper was published in Science on 12 November 2026. The test that would settle the record is the one Rose describes: stone points found sealed inside the dated lake sediments, where the age of the layer would also be the age of the weapon.

Reference: Clarkson et al., “Poisoned hunting weapons, megafauna, and plant use at a 125,000-year-old Arabian paleolakeshore,” Science, 2026. DOI: 10.1126/science.aef9342

Tags: archaeology, Saudi Arabia, human evolution, Homo sapiens, Rub' al Khali, Acokanthera schimperi