In search of the first Asian hominins: excavations in the Soa Basin of Flores, Indonesia

Update September 2011

The search for Hobbit ancestors

A team of archaeologists from the University of Wollongong has just returned from a dig on the Indonesian island of Flores; their objective: to uncover fossil evidence of the early ancestors of the enigmatic Flores ‘Hobbits’.

The discovery in 2003 of the remains of a previously unknown hominin species, Homo floresiensis – the ‘Hobbit’ – in Liang Bua Cave in western Flores, led to major controversy over the origin of these tiny cave dwellers.  One theory is that Hobbits descend from Homo erectus, which became stranded on the island a million years ago and subsequently reduced in size.  A more contentious notion, however, is that Hobbits evolved from a more primitive and currently unknown group of diminutive hominins, that arrived on Flores at a much earlier point in time.  Whichever theory is correct, revealing the ancestry of the Hobbit will have profound implications for our understanding of early human evolution and migration in Asia and further afield.

With these problems in mind, the original discovery team has initiated a new project on Flores to recover the first fossil evidence of Hobbit ancestors.  Their focus has shifted from Liang Bua – which has a maximum age of 95,000 years, and thus is too young to register the arrival of hominins on the island – to much older localities in the So’a Basin, 50 km to the east.  Their previous digs in the basin unearthed fossils of an extinct elephant (Stegodon), and other fauna, dated to between 1 million and 650,000 years ago.  Hundreds of stone implements, clear evidence of humans, were also found, but no fossils of the early tool-makers.  Given the age of the So’a localities, a single tooth or other diagnostic hominin fossil could shed crucial light on the origin of Homo floresiensis.  This is the primary goal of the new project, funded by a five-year (2010-2015) grant from the Australian Research Council, and led by Mike Morwood, Adam Brumm and Gert van den Bergh from the Centre for Archaeological Science, University of Wollongong, in collaboration with palaeontologists from the Geological Survey Institute of Indonesia. 

The team returned this August from a three-month period of fieldwork in the So’a Basin.  Over the course of the season they excavated one site, Mata Menge, dated to between 880,000 and 800,000 years ago, on a scale that has not been seen in Indonesia since the ‘Golden Age’ of palaeoanthropology in the 1930s.  The team first used bulldozers to clear a 2000 m2 area of topsoil to expose the fossil layers.  Some 15 scientists working with 130 local people then painstakingly dug a series of trenches in cement-like deposits, using only hammers and chisels, in search of buried fossils.  The trenches uncovered a surface area of 380 m2 and yielded an extraordinary collection of 3000 fossils and 1500 stone artefacts, three times the amount of finds than the previous six field seasons at Mata Menge combined.  Among this rich haul were a 2.5 m long Stegodon tusk, the largest known from Flores, rare skull pieces from Komodo dragons, even rarer bird and amphibian remains, and abundant evidence for crocodiles and at least two distinct species of giant rats. 

The fossils of early hominins, however, proved elusive.  But there were some heart-stopping moments: one specimen, for a few glorious hours, appeared to be a hominin scapula (shoulder blade) fragment, but after careful cleaning turned out to be a piece of Stegodon skull.  Nonetheless, the team is confident that their efforts will be rewarded.  Louis Leakey dug in Olduvai Gorge for 30 years before he discovered a hominin fossil, Brumm points out, but another group working in the same area in the 1980s found one in three days.  Unlike in many arid sites in East Africa, where fossils are continuously exposed in the eroding landscape, in Indonesia, with its lush vegetation, it is necessary to move the maximum amount of earth to increase the chances of finding that rare hominin fossil fragment.

In the meantime, the Australian scientists are enjoying the adventure of fieldwork in the hot and often-trying conditions on Flores, with hazards ranging from scorpions in beds to actual rat-nests inside mattresses.  Extreme weight loss is also common, with one researcher, despite a steady diet of deep-fried vegetable fritters, shedding 11 kilos in three weeks.  ‘We all returned thin and dark’, says van den Bergh, noting that the team could host the Biggest Loser on their dig-site.

Although hominin remains have so far not shown up, the Mata Menge site has already revealed a wealth of information on the fauna and local environment at the time the toolmakers left their stone implements in the area. The remains of at least 15 Stegodon individuals of all ages have been excavated. Many bones appear to have been transported by flowing water but locally bone accumulations pertaining to single carcasses appear to have been buried rapidly after the animals died. Such rapid burials were facilitated by large influxes of fine-grained ash to the lake margin following volcanic eruptions. It is because of these intermittent periods of rapid sedimentation after volcanic activity that the chances of finding hominin remains are significant. Under normal circumstances a dead carcass lying on the surface was likely scavenged completely by Komodo Dragons and/or crocodiles. The reptilian digestive system dissolves bone completely and thus the remains of smaller mammals, including hominins, would have disappeared completely before fossilization could start to take place. Unless volcanic eruption both caused mass death among the local fauna and at the same time provided large quantities of sediments to bury the dead creatures. So far the Mata Menge excavations have yielded abundant bones of small mammals and birds, which is encouraging for the search of hominin bones.

Analysis of the large amount of Stegodon fossils will probably also allow to assess the cause of death. If most deaths were the result of predation one would expect an over representation of juvenile individuals, but if the animals died of inhaling or eating large amounts volcanic ash, which mainly consists of tiny razor-sharp glass particles, then that would have affected all the individuals in a herd and you would expect to find animals of all ages. The fossils are currently stored in Bandung and many still need to be cleaned and conservated. In January Dr. Gert van den Bergh plans to visit Bandung and start with the analysis of the fossil material.

The sedimentary layers also frequently display signs of soft sediment deformation structures. Such structures may form during earthquakes when shockwaves cause unconsolidated sediments to “liquefy”. But there are indications that a lot of these structures were formed by Stegodonts wading around in the muddy shallows of the lake. This could well explain why some of the bones, including large Stegodon thigh bones, are fractured and some elongated bones are buried vertically in the sediments. Although the Mata Menge stegodonts were dwarfed, they still stood 2 meters at the shoulders and must have weighted around 1500 kg, enough to break bones when trampling them.

Although fractured bones are often found at close proximity to stone artifacts, there is so far no proof that bones were smashed by hominins. Percussion of bones with stone objects often leaves characteristic impact fractures on the bone surface, something that has not been observed during the excavations. But the large collection of bones (almost 3000 specimens ranging from tiny rat molars to Stegodon thigh bones) will be subjected to thorough scrutiny to look for impact or cut marks that could indicate a direct relation between animal remains and human activities at the former lake margin.

Other lines of evidence that are currently being investigated are the reconstruction of the vegetation and the timing of various environmental parameters, such as former lake-level fluctuations and major volcanic events. The palynologist Dr Sander van der Kaars took samples for fossil pollen analysis, not only at Mata Menge, but also from other stratigraphic sections in the So’a Basin. If the deposits will prove to contain well-preserved pollen signals, this will warrant a more thorough palynological study that could shed light on climatic fluctuations and the effects of major volcanic eruptions on the local vegetation during the time-span covered by the So’a Basin sequence (1 million to 700,000 years ago).

In order to get a better insight in the detailed chronology of the basin deposits, Dr Stephanie Flude from Quadlab at Roskilde University (Denmark) has taken volcanic tuff samples for Ar-Ar dating. So far most of the dates known from the So’a Basin are Fission-Track dates, which have an error of approximately 10%. Ar-Ar dating has the potential to refine the dating of various sites and events and will allow for a better correlation between sections and sites throughout the basin. She also plans to chemically fingerprint various volcanic products encountered at the sites and compare them with the volcanic centres surrounding the basin, in order to trace the origin of the volcanic materials.

Dida Yurnaldi, currently enrolled in a Master’s research project here at UOW, is working on paleomagnetic aspects of the So’a Basin deposits. This is expected to add to an even more refined timing and correlation of the basin infill. Ruly Setiawan, another student from the Geological Survey Institute in Bandung and currently gathering data for his PhD, focuses on the geochemical aspects and lateral correlation of the deposits. He also plans to learn more about the Ar-Ar dating techniques.

And then there are of course the hundreds of stone artifacts that will need to be analysed. This task will be coordinated by Dr Adam Brumm in the near future. In the mean time the team will soon have to start with preparations for next years excavations. When working with such a large group of people it is almost inevitable that things go wrong. So next year the team will be better prepared and able to anticipate a number of shortcomings that were experienced this year. The search for fossil hominins will soon be continued!

In search of first Asian Hominids 4
A Stegodon florensis mandible with molar. These extinct elephants stood around 2 metres at the shoulders.

In search of first Asian Hominids 5
Local Ngadapeople at work at one of the Mata Menge trenches (photo: Kerrie Grant).

In search of first Asian Hominids 6
An overview of the Mata Menge site halfway the excavation. (photo: Gert van den Bergh)

In search of first Asian Hominids 7 
A step trench dug at Wolo Sege to expose the stratigraphic sequence. (photo: Gert van den Bergh)

In search of first Asian Hominids 8
Geochronologist Stephanie Flude from the Quadlab Dating Facility of Roskilde University in Denmark sampled the archaeological layers for Ar-Ar dating purposes (photo: Gert van den Bergh)

In search of first Asian Hominids 9
Archaeologist Kerrie Grant measuring 3D coordinates of the excavated finds. (photo: Gert van den Bergh) 

In search of first Asian Hominids 10

An accumulation of Stegodon florensis bones from a single individual. (photo: Gert van den Bergh)

In search of first Asian Hominids 11

Attempts to wet-sieve the excavated sediments were made to collect small finds, but generally the deposits proved to be too cemented to allow for wet-sieving and the sediment blocks had to be crushed with hammers. (photo: Gert van den Bergh)

In search of first Asian Hominids 12

The fossil and artifact-bearing layers exhibit abundant soft-sediment deformation structures. Some of these structures may have been formed by stegodonts walking through shallow water on the muddy lake floor. (photo: Gert van den Bergh)

In search of first Asian Hominids 13

A beautiful flake made of green tuff. (photo: Adam Brumm)

In search of first Asian Hominids 14

An upper jaw of Komodo dragon, Varanus komodoensis.  The Mata Menge dragons were of similar size as modern ones that survive on a few small islands west of Flores. The prey of modern dragons consists mainly of pigs, deer and even water buffaloes, all animals that were not yet around on the island 800,000 years ago. During those times they must have prayed on Stegodon (photo: Gert van den Bergh)

In search of first Asian Hominids 15

Not ivory poachers but fossil hunters! (photo: Gert van den Bergh)

In search of first Asian Hominids 16

Adam Brumm recording finds with a Trimble Total Station. In the background the magnificent Ambulobo Volcano that dominates the So’a Basin landscape (photo: Kerrie Grant).

In search of first Asian Hominids 17

Gert van den Bergh explains to local junior high-school students the importance of the archaeological site in their backyard (photo: Kerrie Grant).

In search of first Asian Hominids 18

At the end of each working week a sports event was organized, one way to shed some weight. Here Archaeologist Michael Marsh in action against a team of Ngada diggers (photo: Erick Setiabudi)

In search of first Asian Hominids 19

Mike Morwood standing amidst Ngada diggers. (photo: Kerrie Grant)

In search of first Asian Hominids 20

Hanny Oktaviani from the Geology Museum in Bandung carefully exposing fossils in one of the trenches  (photo: Kerrie Grant).

In search of first Asian Hominids 21

Some of the team members discussing the finds from a rich fossil-bearing surface, including two well-preserved Stegodon tusks and several stone artifacts. From left to right: Iwan Kurniawan, Dr Adam Brumm, Agus Rozak and Tatang Suryana (photo: Kerrie Grant).


This 5-year Discovery Project (2010–2014) to Mike Morwood and Adam Brumm, and involving Gert van den Bergh, has been funded by the Australian Research Council (ARC) to establish fossil and behavioural evidence of early hominins in the Soa Basin on the eastern Indonesian island of Flores. 

First Hominid AB 1First Hominid AB 2

Left: Map of Flores, eastern Indonesia, showing the location of key early hominin localities in the Soa Basin (Wolo Sege and Mata Menge) and Late Pleistocene Liang Bua Cave in western Flores, where skeletal remains of Homo floresiensis were recovered in 2003-2004. Right: Femur, tibia and foot of LB1, Homo floresiensis.

The discovery in 2003 of the small-bodied hominin Homo floresiensis in Late Pleistocene Liang Bua Cave on Flores has led to considerable debate about the origin of this new species.  Initial interpretations suggested that H. floresiensis descended from H. erectus that became isolated on Flores ~880 ka and underwent a process of insular dwarfism.  Alternatively, however, the origin of this species may lie in a much older lineage of diminutive and small-brained hominins. If the latter proposition is correct, there are fundamental implications for our understanding of the evolution and dispersal of early human populations in Asia.  Testing this hypothesis, however, requires fossil hominin evidence from the earliest periods of human occupation of Flores.

This ARC Discovery Project is a collaborative effort with researchers from the Indonesian Centre for Geological Survey, (CGS), and targets key archaeological and paleontological sites in the Soa Basin, where there are well preserved faunal remains associated with evidence of Early and Middle Pleistocene hominin occupation (~1.0 to 0.65 Ma), and where distribution of early hominin activities across the basin provides insights into their adaptive behavior. Although there is a long history of research in the basin, we are currently undertaking a level of inter-disciplinary investigation and scale of excavation not attempted previously on Flores, and seldom in Southeast Asia.  The principal aim of the project is to recover diagnostic hominin skeletal material, which will establish the identity of the earliest tool-workers in the Soa Basin and resolve the debate about the origin of Homo floresiensis.  Our investigations will also greatly increase the resolution of the Pleistocene hominin behavioural record from Flores and provide significant insight into the palaeo-environmental contexts of these early humans.

First Hominid AB 3  First Hominid AB 4

Left: The Soa Basin of Flores, Indonesia, is ringed by volcanoes. Right: Faunal sequence and Pleistocene record of hominins on Flores (G.D. van den Bergh and M.J. Morwood).

The first season of fieldwork took place in June-August 2010.  Excavations were conducted at two key Middle and Early Pleistocene localities: Mata Menge (~880-800 ka), and Wolo Sege (~1 Ma). The former site is situated on the margins of an ancient lake and excavations yielded fossil Stegodon, Komodo dragon, murine rodents and crocodile remains, along with in situ flaked stone artefacts. The latter site, Wolo Sege, provides the earliest-known record of hominins from the Soa Basin (or elsewhere on Flores), and the excavations recovered an important collection of stone tools from stratified streambed and floodplain sediments. The first season of investigations was focused primarily on recording the wider geological and stratigraphic contexts of these sites, and taking further samples for dating and palaeo-environmental analyses. Subsequent seasons will involve the use of heavy machinery and large teams of fieldworkers to significantly expand the scale of excavations at Mata Menge and Wolo Sege, and at other Soa Basin sites with potential for revealing hominin fossils.

First Hominid AB 5  First Hominid AB 6

Left: Excavations at Wolo Sege (~1 Ma) (Photo: A. Brumm). Right: Geological step-trench at Mata Menge to record the stratigraphy in detail (Photo: G.D. van den Bergh).

Main collaborators
  • Fachroel Aziz: CGS, Indonesia
  • Matt Tocheri: Smithsonian Institution, USA
  • Bill Jungers: Stony Brook University, USA
  • Michael Storey: QuadLab, Roskilde University, Denmark
Key publications

Brumm, A., Jensen, G., van den Bergh, G.D., Morwood, M.J., Kurniawan, I., Aziz, F. & Storey, M. (2010). Hominins on Flores, Indonesia, by one million years ago. Nature 464, 748-752.

Print pdf version of this page