New Ages for the Last Australian Megafauna: Continent-Wide Extinction About 46,000 Years Ago

See allHide authors and affiliations

Science  08 Jun 2001:
Vol. 292, Issue 5523, pp. 1888-1892
DOI: 10.1126/science.1060264

This article has a correction. Please see:


All Australian land mammals, reptiles, and birds weighing more than 100 kilograms, and six of the seven genera with a body mass of 45 to 100 kilograms, perished in the late Quaternary. The timing and causes of these extinctions remain uncertain. We report burial ages for megafauna from 28 sites and infer extinction across the continent around 46,400 years ago (95% confidence interval, 51,200 to 39,800 years ago). Our results rule out extreme aridity at the Last Glacial Maximum as the cause of extinction, but not other climatic impacts; a “blitzkrieg” model of human-induced extinction; or an extended period of anthropogenic ecosystem disruption.

Twenty-three of the 24 genera of Australian land animals weighing more than 45 kg (which, along with a few smaller species, constituted the “megafauna”) were extinct by the late Quaternary (1–3). The timing and causes of this environmental catastrophe have been debated for more than a century (4, 5), with megafaunal extirpation being attributed to the impact of the first human colonizers (1, 5–8), who arrived 56 ± 4 thousand years ago (ka) (9–13), or climate change (4) [in particular, increased aridity at the Last Glacial Maximum (19 to 23 ka) (14)]. A resolution to this debate has been thwarted by the lack of reliable ages for megafaunal remains and for the deposits containing these fossils. The disappearance of one species of giant bird (Genyornis newtoni) from the arid and semi-arid regions of southeastern Australia has been dated to 50 ± 5 ka, on the basis of >700 samples of eggshell (8), but no secure ages for extinction have been reported for the giant marsupials or reptiles, which constitute 22 of the 23 extinct genera of megafauna weighing >45 kg. Here we present burial ages, obtained using optical and230Th/234U dating methods, for the remains of several megafaunal taxa (mostly giant marsupials; see Table 1) discovered at sites located in the humid coastal fringe and drier continental interior of Australia and in the montane forest of West Papua (Fig. 1), which was joined to Australia by a land bridge at times of lowered global sea level.

Figure 1

Map of the Australian region showing the megafauna sites dated in this study. Site numbers: 1, Ned's Gully; 2, Mooki River; 3, Cox's Creek (Bando); 4, Cox's Creek (Kenloi); 5, Tambar Springs; 6, Cuddie Springs; 7, Lake Menindee (Sunset Strip); 8, Willow Point; 9, Lake Victoria (site 50); 10, Lake Victoria (site 51); 11, Lake Victoria (site 73); 12, Montford's Beach; 13, Lake Weering; 14, Lake Corangamite; 15, Lake Weeranganuk; 16, Lake Colongulac; 17, Warrnambool; 18, Victoria Fossil Cave (Grant Hall); 19, Victoria Fossil Cave (Fossil Chamber); 20, Wood Point; 21, Lake Callabonna; 22, Devil's Lair; 23, Kudjal Yolgah Cave; 24, Mammoth Cave; 25, Moondyne Cave; 26, Tight Entrance Cave; 27, Du Boulay Creek; 28, Kelangurr Cave. The bold dashed line crossing the continent indicates the approximate present-day boundary between the zones dominated by summer rainfall from monsoonal activity (north of the line) and winter rainfall from westerly storm tracks (south of the line). The stippled area indicates the zone that receives less than 500 mm rainfall per year and where potential evapotranspiration exceeds mean monthly evapotranspiration year-round with negligible runoff. Climatic data are from (24, 38) and references therein.

Table 1

Megafaunal taxa represented at the study sites. The names of the numbered sites are given in Table 2 and Fig. 1. Taxa represented by articulated remains are indicated by Xand Cf., whereas x and cf. denote taxa represented by disarticulated remains or remains for which articulation is uncertain. Parentheses indicate that Genyornis newtoni is represented by a footprint at Warrnambool (site 17) and by eggshell at Wood Point (site 20). The extant Macropus giganteus, M. fuliginosus, andSarcophilus harrisii are included as they are represented by individuals up to 30% larger in dental dimensions than the living forms. The gigantic form of M. giganteus is referred to here as M. g. titan, and that of S. harrisii asS. h. laniarius. Vombatus hacketti andWallabia kitcheneri belong to genera extant in eastern Australia but extinct in Western Australia.

View this table:

Most major biogeographic and climatic regions, and all five main groups of fossil sites (14), are represented in our survey. Most of the sites in southwestern Australia are caves that have acted as pitfall traps, whereas the sites in eastern Australia consist mainly of aeolian deposits along the edges of former or present lake basins, river or swamp deposits, and coastal dune deposits. To maximize our prospects of encountering fossils close in age to the terminal extinction event, we chose sites that geomorphological and stratigraphic evidence indicated were relatively young. The most recent megafaunal site may not be included in our survey, but we consider that a sufficient number of sites (n = 28) have been dated to discern a clear pattern in the distribution of burial ages.

A review (15) of 91 radiocarbon (14C) ages obtained for Australian megafauna before 1995 rejected the vast majority of ages as being unreliable (16), including all those younger than 28 ka before the present (B.P.). The remaining14C ages were close to or beyond the practical limits of the technique, or were on materials that had ambiguous associations with the megafaunal remains. Radiocarbon dating of bone and charcoal older than 35 ka is problematic using conventional sample pretreatments (17–19). Consequently, 14C ages were used in this study only for comparison with ages of <50 ka obtained from optical dating of megafauna-bearing sediments and230Th/234U dating of flowstones formed above and below megafaunal remains. Optical dating is a luminescence-based method that indicates the time elapsed since the sediment grains were last exposed to sunlight (20–22). The optical age corresponds to the burial age of megafaunal remains in primary deposition, whereas 230Th/234U dating gives the crystallization age of the flowstone, and thus a constraining age for remains above or below the flowstone. Support for the optical ages reported here (Table 2) is provided by their consistency with the 14C and230Th/234U ages obtained at megafaunal sites where comparisons have been made (Table 3) (19, 23–25). All three methods yield concordant ages within the time range of 14C dating, and beyond this limit the optical and230Th/234U ages are in good agreement and correct stratigraphic order.

Table 2

Optical ages for burial sediments, supporting data, and sample contexts.

View this table:
Table 3

230Th/234U ages for Western Australian flowstones, supporting data, and sample contexts. The subscripts (t) and (0) denote the present and initial values of δ234U, respectively. All errors are 2σ. Ages for flowstones at Devil's Lair (site 22), Tight Entrance Cave (site 26), and Victoria Fossil Cave (Grant Hall, site 18, and Fossil Chamber, site 19) are reported elsewhere (19, 24, 25,28).

View this table:

Optical and 230Th/234U dating were conducted primarily on deposits containing the remains of megafauna in articulated anatomical position (Table 1) to avoid uncertainties introduced by post-depositional disturbance and reworking of fossils. This conservative approach is vital because the remains must be in primary depositional context to estimate the time of death from optical dating of the burial sediments or 230Th/234U dating of the enclosing flowstones. We also dated some deposits with disarticulated remains, but we recognize that these ages will be too young if the remains have been derived from older units. A sandstone slab bearing the impression of a Genyornis footprint and dune sands containing burnt fragments of Genyornis eggshell were also dated. Sediment samples for optical dating were collected on site from stratigraphic units that were clearly related to the megafaunal remains; in addition, lumps of sediment attached to megafaunal remains in museum collections were removed for dating (Table 2). We adopted a conservative approach to dating of museum samples (20), owing to their small size and the lack of an in situ dose rate measurement. Confidence in the age estimates for the museum specimens is given by the close agreement between the ages of the museum and field-collected samples from Kudjal Yolgah Cave (site 23; see Table 2). Our main conclusions, however, are based on field-collected samples, which yield the most reliable and precise ages. Calcite flowstones were prepared for230Th/234U dating using standard methods and were analyzed by thermal ionization mass spectrometry (19, 24,25), and the ages (Table 3) have been corrected for detrital 230Th contamination (26).

The youngest optical ages obtained for deposits with articulated megafaunal remains (Table 2) (27) are 47 ± 4 ka for Ned's Gully (site 1) in Queensland and 46 ± 2 ka for Kudjal Yolgah Cave (site 23) in Western Australia. This result implies broadly synchronous extinction across the continent. Claims have also been made (28) for articulated remains ofSimosthenurus occidentalis of similar age from Tight Entrance Cave (site 26, unit H or below) in Western Australia, and several sites (3, 4, 8, 9, and 10) in New South Wales produced slightly older ages (50 to 55 ka) for articulated megafauna. In contrast, much younger apparent burial ages were obtained for some sites containing disarticulated remains (Table 2) (27); the youngest such age is 2.0 ± 0.2 ka for fragmented remains at Tambar Springs (site 5). Optical dating of individual grains from the Cuddie Springs deposit [site 6 (23)] indicates that some sediment mixing has occurred (29). We interpret the young ages obtained for disarticulated remains and the indication of sediment mixing at Cuddie Springs as evidence that the remains are not in their primary depositional setting, but have been eroded from older units and redeposited in younger units with contemporaneous sediment and charcoal.

The youngest measured burial age for articulated remains may be older than the terminal extinction event, unless the most recent burial site is fortuitously included in our survey. But each optical age has a relative standard error of 5 to 15%, so the measured age could by chance be less than the true extinction age at some sites. Accordingly, we built a statistical model of the data under the assumption that the true burial ages are a realization of a Poisson process of constant intensity up to the time of extinction. That is, we assumed that the true burial ages are distributed randomly through time, with equal numbers per unit time, on average. The optical age is the true burial age plus a Gaussian error with a mean of zero and a standard deviation equal to the reported standard error. We estimated the time of extinction by maximum likelihood, confining attention to articulated remains with optical ages of ≤55 ka (30). This avoids a potential difficulty caused by the undersampling of sites much older than the extinction event. Using this model, the maximum likelihood estimate of the extinction time is 46.4 ka, with 68% and 95% confidence intervals of 48.9 to 43.6 ka and 51.2 to 39.8 ka, respectively.

Our data show little evidence for faunal attenuation. Twelve of the 20 genera of megafauna recorded from Pleistocene deposits in temperate Australia (1, 2) survived to at least 80 ka, including the most common and widespread taxa, and six of these genera (Diprotodon, Phascolonus, Thylacoleo,Procoptodon, Protemnodon, andSimosthenurus) are represented at the two sites dated to around 46 ka. These data indicate that a relatively diverse group of megafauna survived until close to the time of extinction. Further sites are needed to test this proposition and to identify the cause(s) of megafaunal extinction.

The burial ages for the last known megafaunal occurrence suggest that extinction occurred simultaneously in eastern and western Australia, and thus probably continent-wide, between 51 and 40 ka (95% confidence interval), at least 20 ka before the height of the Last Glacial Maximum. We estimate that the megafauna had vanished within 10 ± 5 ka of human arrival [56 ± 4 ka (9–13)] across a wide range of habitats and climatic zones. Megafaunal extinction in Australia occurred tens of millennia before similar events in North and South America, Madagascar, and New Zealand, each of which was preceded by the arrival of humans (31). A prediction of the “blitzkrieg” model of human-induced extinction [as proposed first for North America (32) and later for New Zealand (33)] is that megafaunal extinction should occur soon after human colonization, and that extinction is followed by widespread ecosystem disruption (1). Alternatively, human arrival may first have triggered ecosystem disruption, as a result of which the megafauna became extinct (8). The latter sequence of events allows for a substantial time interval between human colonization and megafaunal extinction, so that climatic factors may also be involved (34). There is sufficient uncertainty in the ages for both human colonization and megafaunal extinction that we cannot distinguish between these possibilities, but our data are consistent with a human role in extinction. Resolving this debate would require more precise ages for human colonization and megafaunal extinction, as well as an improved understanding of human interactions with the Australian landscape and biota during the earliest period of human occupation.

  • * To whom correspondence should be addressed. E-mail: rgrobe{at}

  • Present address: Department of Geology and Geophysics, University of Utah, Salt Lake City, UT 84112, USA.


View Abstract

Stay Connected to Science

Navigate This Article