New research into the "hobbit" fossils of now-extinct metre-tall humanoids discovered on the Indonesian island of Flores has found they most likely evolved from an ancient African ancestor and not from the taller Homo erectus found on Java, as widely believed.
Research led by Australian National University scientist Debbie Argue has said that Homo floresiensis was more likely a sister species of Homo habilis, which emerged up to 2 million years ago in Africa, rather than the larger and more recently evolved Homo erectus, which emerged 1.2 million years ago.
More Science Videos
Tracing the origins of 'hobbit' fossils
New research into the "hobbit" fossils has found they most likely evolved from an ancient African ancestor.
Dr Argue said: "It's possible that Homo floresiensis evolved in Africa and migrated, or the common ancestor moved from Africa then evolved into Homo floresiensis somewhere else."
Last year, research led by Gert van den Bergh from the University of Wollongong's centre for archaeological science claimed that the "hobbits" had evolved from the larger Homo erectus.
At the time, Dr van den Bergh said: "Homo floresiensis evolved its miniature body proportions during the initial 300,000 years on Flores and is thus a dwarfed side lineage that ultimately derives from Homo erectus.
"It is also possible that the dwarfism developed on a stepping-stone island between Asia and Flores, such as Sulawesi."
However, the ANU study, published on Friday in the Journal of Human Evolution, claims to have used a much wider number of data points on the "hobbit" fossils, ranging across the skull, jaws, teeth, arms, legs and shoulders.
Professor Colin Groves, a co-author of the study, said: "We looked at a lot of original fossils from African, Indonesia and Europe. We collated features by which they vary and performed a statistical analysis, which sorts out these features according to evolutionary status.
"We found that Homo floresiensis has quite a lot of derived characters with Homo habilis. These two species are clearly more primitive than Homo erectus."
Dr Argue said they looked at whether Homo floresiensis could be descended from Homo erectus.
"We found that if you try to link them on the family tree, you get a very unsupported result. All the tests say it doesn't fit – it's just not a viable theory."
Professor Groves said Homo habilis is known from sites in East Africa between 1.5 and 2 million years ago. Homo erectus – sometimes known as Java man – is known to have lived on the Indonesian island that gave the humanoid species its name from 1.2 million years ago.
"Some survived up to less than 500,000 years ago," Professor Groves said.
Scientists agree that the "hobbit" species lived on Flores until about 54,000 years ago. There is incomplete evidence that a 700,000-year-old jaw fossil from Flores could be from Homo floresiensis, Professor Groves said.
The original "hobbit" fossils were discovered in 2003 by a team led by the late University of New England archaeologist Mike Moorwood. They revealed a humanoid species about a metre tall with a brain likely the size of a grapefruit.
The pathway from Homo habilis to the "hobbit" remains unclear.
"Something like Homo habilis spread out of Africa 2 million years ago, changing as it went through tropical Asia to become a separate species on Flores," Professor Groves said. "We don't know when as we lack the fossils."
He said that Homo habilis was a little bit bigger than the "hobbit" but definitely smaller than Homo erectus.
Professor Groves said that our understanding of hominid evolution shows they were very diverse.
"We now know our line of evolution from chimpanzeess was diverse – there were lots of species running side by side that we just don't know about yet," Professor Groves said.
Rather than there being a straight branch from chimpanzee to human, Professor Groves said that "the family tree was a bush".
"It's an accident of history that we are the only species survived."
Fairfax Media was unable to reach University of Wollongong's Dr van den Bergh or Griffith University's Adam Brumm, who published the paper in Nature last year that argued Homo floresiensis evolved from Homo erectus.