A study that claims humans reached the Americas 130,000 years ago - much earlier than previously suggested - has run into controversy.
Humans are thought to have arrived in the New World no earlier than 25,000 years ago, so the find would push back the first evidence of settlement by more than 100,000 years.
The conclusions rest on analysis of animal bones and tools from California.
But many experts contacted by the BBC said they doubted the claims.
Thomas Deméré, Steven Holen and colleagues examined material from the Cerutti Mastodon site near San Diego. The site was originally uncovered in 1992, during highway construction work. Possible stone tools were discovered alongside the smashed up remains of a mastodon (Mammut americanum) - an extinct relative of mammoths and living elephants.
The researchers behind the latest study were unable to carry out radiocarbon dating on the remains, so they used a technique called uranium-thorium dating on several bone fragments, coming up with a date of 130,000 years.
The team members found that some of the bones and teeth bore a characteristic breakage pattern known as spiral fracturing, considered to occur when the bone is fresh. Additionally, some of the bones showed typical signs of being smashed with hard objects.
Rocks found alongside the mastodon remains show signs of wear and being struck against other surfaces, the researchers say. They conclude that these represent hammerstones and anvils - two types of stone tool used by prehistoric cultures.
Dr Deméré, curator of palaeontology at the San Diego Natural History Museum, said the totality of evidence at the site had led team members to the conclusion that "humans were processing [working on or breaking up] mastodon limb bones using hammerstones and anvils and that the processing occurred at the site of burial 130,000 years ago".
Dr Holen, co-director of the Center for American Paleolithic Research in South Dakota, commented: "We have conducted two experiments breaking elephant bones with large rock hammers and anvils. We produced exactly the same kind of fracture patterns as we found on the Cerutti mastodon limb bones."
He added: "We can eliminate all of the natural processes that break bones like this. These bones were not broken by carnivore chewing, or by other animals trampling on this bone... the distribution patterns of the fractured pieces of bone right around the anvils is fairly conclusive evidence because we see that experimentally also."
It's not entirely clear why early humans smashed up the mastodon bones.
"We have no evidence that this is a kill or butchery site, but what we do have evidence of is that people were here breaking up the limb bones of this mastodon, removing some of the big thick pieces - probably to make tools out of - and they may also have been extracting the marrow for food," said Dr Holen.
If the team's conclusions are correct, people could have reached the Americas from Asia via a land bridge across the Bering Strait. This bridge periodically emerged during cold periods - when ocean water was locked up as ice - and disappeared when the climate warmed again and sea levels rose.
The earliest widely accepted evidence for humans in the Americas dates to roughly 15,000 years ago. This is a field where fierce debate has raged over rolling back the ages of human occupation by one or two thousand years, let alone 100,000.
Dr Deméré and colleagues are not the first scientists to posit much earlier dates for people settling in the Americas. What distinguishes the latest work is that it has been published in one of the most prestigious peer-reviewed science journals in the world - Nature.
However, other experts remain unconvinced by the new evidence. Prof Michael R Waters, from Texas A&M University in College Station, described the new paper as "provocative".
He told BBC News the study "purports to provide evidence of human occupation of the Americas some 115,000 years before the earliest well established evidence".
Prof Waters explained: "I have no issues with the geological information - although I would like to know more about the broader geological context - and the likely age of the locality. However, I am sceptical of the evidence presented that humans interacted with the mastodon at the Cerutti Mastodon site."
"To demonstrate such early occupation of the Americas requires the presence of unequivocal stone artefacts. There are no unequivocal stone tools associated with the bones... this site is likely just an interesting paleontological locality."
Prof Tom Dillehay, from Vanderbilt University in Nashville, Tennessee, told BBC News the claim was not plausible.
Another authority on early American archaeology, Prof David Meltzer from Southern Methodist University in Dallas, Texas, said: "Nature is mischievous and can break bones and modify stones in a myriad of ways.
"With evidence as inherently ambiguous as the broken bones and non-descript broken stones described in the paper, it is not enough to demonstrate they could have been broken/modified by humans; one has to demonstrate they could not have been broken by nature.
"This is an equifinality problem: multiple processes can cause the same product."
Chris Stringer, from London's Natural History Museum, said that "if the results stand up to further scrutiny, this does indeed change everything we thought we knew about the earliest human occupation of the Americas," adding: "If true, the results may well mean that archaic people like the Denisovans or Neanderthals were the first colonisers of the Americas, rather than modern humans."
He explained that "extraordinary claims require extraordinary evidence - each aspect requires the strongest scrutiny," but Prof Stringer also observed: "High and concentrated forces must have been required to smash the thickest mastodon bones, and the low energy depositional environment seemingly provides no obvious alternative to humans using the heavy cobbles found with the bones."
The dating method used by the researchers to assign an age to this material works by measuring the radioactive decay of uranium that becomes incorporated into the bones over time.
"The type of samples that are most widely dated with this technique are ones that contain uranium as a primary substitution in their structure, such as inorganic carbonates, like cave carbonates, or corals, which take in uranium as they take calcium out of seawater," Dr Warren D Sharp, an expert in isotope dating from the Berkeley Geochronology Center in California, told BBC News.
"What they've done in this paper is applied it to bone. That can be challenging because bones don't contain significant amounts of primary uranium. They acquire the uranium when they become buried - they take it up from soil porewaters."
He added: "That said, I think the dating is sound. They have done a very careful job. They have dated multiple samples and obtained similar results. The systematics of the concentrations of uranium in profiles across the bones are what you'd expect for reliable dates. And the bones that they've dated seem to be an integral part of the site, so their age should be relevant to the rest of the observations."
Prof Meltzer said the history of the material from the site meant it would be difficult to prove that humans broke the bones. He explained: "[The evidence] comes from a site that was excavated [approximately] 25 years ago as a salvage project during a highway expansion.
"The kinds of detailed information necessary to understand how these bones and stones came to be... is simply not available. The authors do what they can with the extant collections, but they necessarily have to rely more on generalisations about what could (or could not) account for the evidence - which gets us back to the equifinality problem."
Follow Paul on Twitter.