Content uploaded by Jon M Olley
Author content
All content in this area was uploaded by Jon M Olley
Content may be subject to copyright.
(42) suggest that preindustrial levels were ⬍1.9 ppt.
Assuming a steady state and a 4.9-year lifetime, these
levels imply natural emissions of ⬍10 Gg year
⫺1
.
41. J. H. Butler et al., Nature 399, 749 (1999).
42. P. J. Fraser, unpublished data.
43. We estimate emissions of ⬃1 Gg year
⫺1
in Europe in
1999, 0.06 Gg year
⫺1
in Australia in 1998–99, and ⬍1
Gg year
⫺1
in the western United States in 1999 (5).
44. S. Karlsdottir, I. S. A. Isaksen, Geophys. Res. Lett. 27,
93 (2000).
45. M. Mayer, C. Wang, M. Webster, R. Prinn, J. Geophys.
Res. 105, 22869 (2000).
46. A. M. Thompson, Science 256, 1157 (1992).
47. Y. Wang, D. Jacob, J. Geophys. Res. 103, 31123
(1998).
48.
P. J. Crutzen, P. H. Zimmerman, Tellus 43AB, 136 (1991).
49.
The ALE, GAGE, and AGAGE projects involved substantial
efforts by many people beyond the authors of this paper
(5). In its latest phase (AGAGE), support came (and
comes) primarily from NASA, with important contribu-
tions from the Department of Environment, Transport
and the Regions (United Kingdom); Commonwealth Sci-
entific and Industrial Research Organization (Australia);
Bureau of Meteorology (Australia); and NOAA, among
others (5).
28 December 2000; accepted 17 April 2001
Published online 3 May 2001;
10.1126/science.1058673
Include this information when citing this paper.
R EPORTS
New Ages for the Last
Australian Megafauna:
Continent-Wide Extinction
About 46,000 Years Ago
Richard G. Roberts,
1
* Timothy F. Flannery,
2
Linda K. Ayliffe,
3
†
Hiroyuki Yoshida,
1
Jon M. Olley,
4
Gavin J. Prideaux,
5
Geoff M. Laslett,
6
Alexander Baynes,
7
M. A. Smith,
8
Rhys Jones,
9
Barton L. Smith
10
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 partic-
ular, 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 dis-
appearance of one species of giant bird (Ge-
nyornis 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, ob-
tained using optical and
230
Th/
234
U dating
methods, for the remains of several megafau-
nal 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.
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,
1
School of Earth Sciences, University of Melbourne,
Melbourne, Victoria 3010, Australia.
2
South Austra-
lian Museum, Adelaide, South Australia 5000, Austra-
lia.
3
Laboratoire des Sciences du Climat et de
l’Environnement, 91198 Gif-sur-Yvette, France.
4
Commonwealth Scientific and Industrial Research
Organization (CSIRO) Land and Water, Canberra, ACT
2601, Australia.
5
Department of Earth Sciences, Uni-
versity of California, Riverside, CA 92521, USA.
6
CSIRO Mathematical and Information Sciences, Mel-
bourne, Victoria 3168, Australia.
7
Western Australian
Museum, Perth, Western Australia 6000, Australia.
8
National Museum of Australia, Canberra, ACT 2601,
Australia.
9
Department of Archaeology and Natural
History, Research School of Pacific and Asian Studies,
Australian National University, Canberra, ACT 0200,
Australia.
10
Department of Earth Sciences, La Trobe
University, Melbourne, Victoria 3086, Australia.
*To whom correspondence should be addressed. E-
mail: rgrobe@unimelb.edu.au
†Present address: Department of Geology and Geo-
physics, University of Utah, Salt Lake City, UT 84112,
USA.
Fig. 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 Cor-
angamite; 15, Lake Weeranganuk; 16,
Lake Colongulac; 17, Warrnambool; 18,
Victoria Fossil Cave (Grant Hall); 19, Vic-
toria 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 ap-
proximate present-day boundary be-
tween 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.
R ESEARCH A RTICLES
8 JUNE 2001 VOL 292 SCIENCE www.sciencemag.org1888
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 X and 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, and Sarcophilus 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 as S. h. laniarius. Vombatus
hacketti and Wallabia kitcheneri belong to genera extant in eastern Australia but extinct in
Western Australia.
Taxa
Site
123456*6† 7 8 9, 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26J 26H 26D 27 28
Articulated remains represented X . XX .. .XX X X X X . XXX . X . X . XXX .. .X .
Reptiles
Meiolania sp. indet. . . . . . . x . . . . . ............. . . . ..
Megalania prisca ......x.. . ............... . . . ..
Wonambi naracoortensis ......... . .............x. . . . ..
Pallimnarchus sp. indet. . . . . x x . . . . . ............. . . . ..
Quinkana sp. indet. . . . . . . x . . . . . ............. . . . ..
Birds
Genyornis newtoni .....xx.. . ......(X) . . (x) X ..... . . ..
Progura naracoortensis ......... . ........x...... . . . ..
Mammals
Megalibgwilia ramsayi ......... . ........x....xx. . . ..
“Zaglossus” hacketti ......... . .............x. . . . ..
Sarcophilus harrisii laniarius x........ . x..x.x..x...... . . x ..
Diprotodon optatum X x XX .xx.. x .....x....X .... . . .X .
Diprotodon sp. indet. . . . . x . . . . . . . . x . . . . x . . .... . . . ..
Maokopia ronaldi ......... . ............... . . . .x
Zygomaturus trilobus ......... . .......xX ....XX ..x..
Zygomaturus sp. indet. . . . . . . . . . . . . . . . x . . . . . .... . . . ..
Palorchestes azael .x....x.. . ...... . .x...... . . . ..
Phascolonus gigas xx... . x .. XX..xX .....X ..... . . ..
Vombatus hacketti ......... . ...........xxx. . . x ..
Thylacoleo carnifex xx....... x . . .x.x.xX . .x.x. . . . ..
Propleopus oscillans ......... . ........x...... . . . ..
Borungaboodie hatcheri ......... . ............... . . x ..
Macropus ferragus ...... ... X ............... . . . ..
Macropus fuliginosus ......... . ...........xxXX x. ...
Macropus giganteus titan X x...xx.. . x . X x XX ..x.X .... . . . ..
Macropus sp. indet. . . . . x . . . . . . . .....xx...... . . . ..
Procoptodon goliah x..... .X . X x.......X ...... . . . ..
Procoptodon sp. indet. . . . . x . . . . . . . . . . x . . . . . .... . . . ..
Protemnodon anak ...... ... x . . . .X x......... . . . ..
Protemnodon brehus ...... ... X . Cf. ...cf.....Cf. ..x. . . x ..
Protemnodon hopei ......... . ............... . . . .x
Protemnodon roechus X ........ . ........x...... . . . ..
Protemnodon sp. indet. . x . . . . x . X . ...........x... . . . ..
Simosthenurus baileyi ......... . ........x...... . . . ..
Simosthenurus brownei ......... . ........X ...XXX ..x..
Simosthenurus gilli ......... . .......xX ...... . . . ..
Simosthenurus maddocki ......... . ........x...... . . . ..
Simosthenurus newtonae ......... . .......xx...... . . x ..
Simosthenurus occidentalis ......... . ........X ...xx.x x x..
Simosthenurus pales ......... . ........x...... . . x ..
Simosthenurus sp. indet. . . . . . . . . . . . . . . . x . . . . . x . . . . . . . .
Sthenurus andersoni .cf....cf.... x ........x.X .... . . . ..
Sthenurus atlas ......... x ....... ........ . . . ..
Sthenurus stirlingi ......... . ..........X .... . . . ..
Sthenurus tindalei ......... x ....... ...X .... . . . ..
Sthenurus sp. indet. . . . . . . x . x . . . . . x . . . . . . .... . . . ..
Wallabia kitcheneri ......... . .............x. . . x ..
*Site 6, units 5, 6a, and 6b. †Site 6, units 7 to 12.
R ESEARCH A RTICLES
www.sciencemag.org SCIENCE VOL 292 8 JUNE 2001 1889
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 max-
imize our prospects of encountering fossils
close in age to the terminal extinction event,
we chose sites that geomorphological and
stratigraphic evidence indicated were rela-
tively 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 (
14
C)
ages obtained for Australian megafauna be-
fore 1995 rejected the vast majority of ages as
being unreliable (16), including all those
younger than 28 ka before the present (B.P.).
The remaining
14
C 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 old-
er than 35 ka is problematic using conven-
tional sample pretreatments (17–19). Conse-
quently,
14
C ages were used in this study only
for comparison with ages of ⬍50 ka obtained
Table 2. Optical ages for burial sediments, supporting data, and sample contexts.
Site* Sample context†
Grain size
(m)
Dose rate‡
(Gy ka
⫺1
)
Paleodose‡
(Gy)
Optical age‡
(ka)
Queensland
1. Ned’s Gully¶ Megafaunal unit, sample 1 180–212 0.76 ⫾ 0.09 35 ⫾ 247⫾ 6
Megafaunal unit, sample 2 90–125 0.78 ⫾ 0.09 36 ⫾ 346⫾ 6
New South Wales
2. Mooki River Megafaunal unit 90–125 1.77 ⫾ 0.16 74 ⫾ 442⫾ 4
3. Cox’s Creek (Bando)¶ Megafaunal unit, sample 1 90–125 1.43 ⫾ 0.14# 75 ⫾ 353⫾ 5
Megafaunal unit, sample 1 180–212 1.38 ⫾ 0.14# 75 ⫾ 354⫾ 6
Megafaunal unit, sample 2 90–125 1.43 ⫾ 0.14# 72 ⫾ 650⫾ 6
4. Cox’s Creek (Kenloi)¶ 5 cm above megafaunal unit 90–125 0.93 ⫾ 0.06 47 ⫾ 251⫾ 4
30 cm below megafaunal unit 90–125 0.97 ⫾ 0.06 51 ⫾ 253⫾ 4
30 cm below megafaunal unit 180–212 0.94 ⫾ 0.06 54 ⫾ 358⫾ 4
5. Tambar Springs Megafaunal unit (spit 4) 90–125 1.43 ⫾ 0.09 2.9 ⫾ 0.2 2.0 ⫾ 0.2
6. Cuddie Springs Above main megafaunal unit (unit 4) 90 –125 2.47 ⫾ 0.15 ⬍41.3 ⫾ 1.2 ⬍16.7 ⫾ 1.2
Megafaunal unit (unit 5) 90–125 2.22 ⫾ 0.14 ⬍59 ⫾ 2 ⬍27 ⫾ 2
Megafaunal unit (unit 6a) 90–125 1.95 ⫾ 0.12 ⬍59 ⫾ 3 ⬍30 ⫾ 2
Megafaunal unit (unit 6b) 90–125 2.72 ⫾ 0.17 ⬍99 ⫾ 5 ⬍36 ⫾ 3
7. Lake Menindee (Sunset Strip)¶ Megafaunal unit 180–212 1.69 ⫾ 0.09 113 ⫾ 867⫾ 6
8. Willow Point¶ Attached to MV specimen 90–125 0.86 ⫾ 0.08 47 ⫾ 255⫾ 6
9. Lake Victoria (site 50)¶ Attached to MV specimen 180–212 0.59 ⫾ 0.07§ 30 ⫾ 352⫾ 8
10. Lake Victoria (site 51)¶ Attached to MV specimen 90–125 0.71 ⫾ 0.08 38 ⫾ 254⫾ 7
11. Lake Victoria (site 73)¶ Attached to MV specimen 90–125 1.71 ⫾ 0.18 165 ⫾ 597⫾ 11
Victoria
12. Montford’s Beach¶ Attached to MV specimen 90–125 0.83 ⫾ 0.10 ⬎49.7 ⫾ 1.2 ⬎60 ⫾ 7
13. Lake Weering¶ Attached to MV specimen 90–125 1.42 ⫾ 0.15# 117 ⫾ 382⫾ 9
14. Lake Corangamite Attached to MV specimen, sample 1 90–125 1.50 ⫾ 0.18 79 ⫾ 452⫾ 7
Attached to MV specimen, sample 1 180–212 1.47 ⫾ 0.18 78 ⫾ 453⫾ 7
Attached to MV specimen, sample 2 180–212 1.46 ⫾ 0.17 70 ⫾ 448⫾ 6
15. Lake Weeranganuk¶ Attached to MV specimen 180 –212 6.3 ⫾ 0.7# 437 ⫾ 18 70 ⫾ 8
16. Lake Colongulac¶ Attached to MV specimen 180 –212 1.59 ⫾ 0.16§ 131 ⫾ 10 82 ⫾ 10
17. Warrnambool¶ From MV sediment slab with footprint 180–212 0.61 ⫾ 0.08 37 ⫾ 360⫾ 9
South Australia
18. Victoria Fossil Cave (Grant Hall) 20 cm below top of megafaunal unit 90–125 1.28 ⫾ 0.08 107 ⫾ 984⫾ 8
19. Victoria Fossil Cave (Fossil Chamber)¶ 50 cm below top of megafaunal unit 90 –125 0.67 ⫾ 0.04 115 ⫾ 6 171 ⫾ 14
50 cm below top of megafaunal unit 180–212 0.65 ⫾ 0.04 102 ⫾ 8 157 ⫾ 16
20. Wood Point Unit containing eggshell 90–125 1.60 ⫾ 0.14# 88 ⫾ 355⫾ 5
21. Lake Callabonna¶ Attached to MV specimen 90 –125 0.62 ⫾ 0.07 46 ⫾ 275⫾ 9
Western Australia
22. Devil’s Lair Above main megafaunal unit (layer 28) 90–125 1.22 ⫾ 0.05 51 ⫾ 242⫾ 2
Megafaunal unit (layer 32) 90–125 1.71 ⫾ 0.07# 79 ⫾ 347⫾ 2
Megafaunal unit (layer 39) 90–125 1.35 ⫾ 0.06 65 ⫾ 248⫾ 3
23. Kudjal Yolgah Cave¶ Megafaunal unit (pit 2) 90–125 1.11 ⫾ 0.05 51 ⫾ 246⫾ 2
Attached to WAM specimen 125–250 1.22 ⫾ 0.14 56 ⫾ 246⫾ 6
24. Mammoth Cave¶ Upper megafaunal unit 90–125 0.72 ⫾ 0.09 40 ⫾ 455⫾ 9
Attached to WAM specimen, sample 1 90–125 1.04 ⫾ 0.15 66 ⫾ 263⫾ 9
Attached to WAM specimen, sample 2 90–125 0.90 ⫾ 0.12 67 ⫾ 374⫾ 10
25. Moondyne Cave¶ Megafaunal unit 90–125 0.68 ⫾ 0.05# 89 ⫾ 7 131 ⫾ 14
26. Tight Entrance Cave Megafaunal unit (unit J) 90–125 0.72 ⫾ 0.04 24 ⫾ 333⫾ 4
Megafaunal unit (unit H) 90–125 0.52 ⫾ 0.03 23 ⫾ 345⫾ 6
Megafaunal unit (unit D) 90–125 0.73 ⫾ 0.04 103 ⫾ 14 141 ⫾ 21
27. Du Boulay Creek¶ Attached to WAM specimen 90–125 2.7 ⫾ 0.3# ⬎215 ⫾ 22 ⬎80 ⫾ 12
West Papua
28. Kelangurr Cave Megafaunal unit 90–125 1.07 ⫾ 0.10§ 17.6 ⫾ 1.0 16 ⫾ 2
*Sites with taxa represented by articulated remains are marked by ¶. †MV and WAM indicate sediment removed from megafaunal collections at the Museum of Victoria and
Western Australian Museum, respectively. ‡Mean ⫾ 1 uncertainty. Samples with a significant deficit of
238
U compared to
226
Ra are marked by #, and those with a significant
excess of
238
U over
226
Ra are marked by §. Paleodose values include ⫾2% uncertainty associated with laboratory beta-source calibration. The Cuddie Springs (site 6) samples have
multiple paleodose populations, of which the highest are shown (29). The Montford’s Beach (site 12) and Du Boulay Creek (site 27) samples have paleodose distributions consistent
with partial bleaching (37), so the minimum paleodose values and age estimates [obtained using a minimum age model (35)] are shown. The high paleodose of the Lake Weeranganuk
(site 15) sample was obtained from aliquots with saturating exponential plus linear growth curves of luminescence intensity versus dose (36). This sample also has an unusually high
dose rate, which is due to concentrations of ⬎20 ppm of all radionuclides in the
238
U decay series. If these concentrations were lower in the past, then the optical age would be older.
R EPORTS
8 JUNE 2001 VOL 292 SCIENCE www.sciencemag.org1890
from optical dating of megafauna-bearing
sediments and
230
Th/
234
U dating of flow-
stones 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 corre-
sponds to the burial age of megafaunal re-
mains in primary deposition, whereas
230
Th/
234
U dating gives the crystallization age of
the flowstone, and thus a constraining age for
remains above or below the flowstone. Sup-
port for the optical ages reported here (Table
2) is provided by their consistency with the
14
C and
230
Th/
234
U 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
14
C dating, and beyond this
limit the optical and
230
Th/
234
U ages are in
good agreement and correct stratigraphic
order.
Optical and
230
Th/
234
U dating were con-
ducted primarily on deposits containing the
remains of megafauna in articulated ana-
tomical position (Table 1) to avoid uncer-
tainties 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
230
Th/
234
U dating of the enclosing flow-
stones. 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 impres-
sion of a Genyornis footprint and dune
sands containing burnt fragments of Geny-
ornis 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 giv-
en 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 for
230
Th/
234
U 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
230
Th 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 ⫾ 2kafor
Kudjal Yolgah Cave (site 23) in Western
Australia. This result implies broadly syn-
chronous extinction across the continent.
Claims have also been made (28) for articu-
lated remains of Simosthenurus 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 con-
trast, much younger apparent burial ages
were obtained for some sites containing dis-
articulated remains (Table 2) (27); the
youngest such age is 2.0 ⫾ 0.2 ka for frag-
mented 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 evi-
dence that the remains are not in their prima-
ry 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. Ac-
cordingly, 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 estimat-
ed the time of extinction by maximum like-
lihood, confining attention to articulated
remains with optical ages of ⱕ55 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 extinc-
tion 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 depos-
its 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, Pro-
coptodon, Protemnodon, and Simosthenurus)
are represented at the two sites dated to
around 46 ka. These data indicate that a
relatively diverse group of megafauna sur-
vived until close to the time of extinction.
Further sites are needed to test this proposi-
tion and to identify the cause(s) of megafau-
nal extinction.
Table 3.
230
Th/
234
U ages for Western Australian flowstones, supporting data,
and sample contexts. The subscripts (t) and (0) denote the present and initial
values of ␦
234
U, respectively. All errors are 2. 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).
Site* Sample context
Detrital Th
correction†
U
(ppm)
230
Th/
238
U
activity ratio
␦
234
U
(t)
(‰)
␦
234
U
(0)
(‰)
230
Th/
232
Th
activity ratio
230
Th/
234
U
age (ka)
23. Kudjal Yolgah Cave Above megafaunal unit,
sample 1
U 0.008 0.302 ⫾ 0.006 102 ⫾ 12 112 ⫾ 14 13.9 ⫾ 0.2 34.7 ⫾ 0.9
C 0.294 ⫾ 0.008 102 ⫾ 31 112 ⫾ 34 33.6 ⫾ 1.6
Above megafaunal unit,
sample 2
U 0.008 0.331 ⫾ 0.003 105 ⫾ 7 117 ⫾ 7 5.37 ⫾ 0.05 38.5 ⫾ 0.6
C 0.308 ⫾ 0.005 105 ⫾ 18 116 ⫾ 19 35.4 ⫾ 1.0
24. Mammoth Cave Above upper megafaunal unit U 0.025 0.375 ⫾ 0.003 49 ⫾ 357⫾ 3 5.95 ⫾ 0.05 47.9 ⫾ 0.6
C 0.353 ⫾ 0.006 49 ⫾ 16 56 ⫾ 18 44.4 ⫾ 1.3
Below upper megafaunal unit U 0.056 0.457 ⫾ 0.005 73 ⫾ 486⫾ 5 4.49 ⫾ 0.05 60.0 ⫾ 1.0
C 0.429 ⫾ 0.009 72 ⫾ 22 84 ⫾ 25 55.2 ⫾ 2.2
25. Moondyne Cave Above megafaunal unit U 0.061 0.341 ⫾ 0.004 40 ⫾ 445⫾ 4 17.7 ⫾ 0.3 43.1 ⫾ 0.7
C 0.335 ⫾ 0.008 41 ⫾ 24 46 ⫾ 27 42.2 ⫾ 1.8
*All three sites have taxa represented by articulated remains. †Data corrected (C) and uncorrected (U) for detrital
230
Th contamination (26). The detritally corrected ages are
considered more reliable.
R EPORTS
www.sciencemag.org SCIENCE VOL 292 8 JUNE 2001 1891
The burial ages for the last known
megafaunal occurrence suggest that extinc-
tion occurred simultaneously in eastern and
western Australia, and thus probably conti-
nent-wide, between 51 and 40 ka (95% con-
fidence 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 ⫾ 4ka
(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 extinc-
tion should occur soon after human coloniza-
tion, and that extinction is followed by wide-
spread ecosystem disruption (1). Alternative-
ly, human arrival may first have triggered
ecosystem disruption, as a result of which the
megafauna became extinct (8). The latter se-
quence of events allows for a substantial time
interval between human colonization and
megafaunal extinction, so that climatic fac-
tors may also be involved (34). There is
sufficient uncertainty in the ages for both
human colonization and megafaunal extinc-
tion 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 extinc-
tion, as well as an improved understanding of
human interactions with the Australian land-
scape and biota during the earliest period of
human occupation.
References and Notes
1. T. F. Flannery, Archaeol. Oceania 25, 45 (1990).
2. P. Murray, in Vertebrate Palaeontology of Australasia,
P. Vickers-Rich, J. M. Monaghan, R. F. Baird, T. H. Rich,
Eds. (Pioneer Design Studio, Melbourne, 1991), pp.
1071–1164.
3. T. F. Flannery, R. G. Roberts, in Extinctions in Near
Time: Causes, Contexts, and Consequences,R.D.E.
MacPhee, Ed. (Kluwer Academic/Plenum, New York,
1999), pp. 239–255.
4. C. S. Wilkinson, Proc. Linn. Soc. New South Wales 9,
1207 (1884).
5. R. Owen, Researches on the Fossil Remains of the
Extinct Mammals of Australia (Erxleben, London,
1877).
6. D. Merrilees, J. R. Soc. West. Austr. 51, 1 (1968).
7. R. Jones, Archaeol. Phys. Anthropol. Oceania 3, 186
(1968).
8. G. H. Miller et al., Science 283, 205 (1999).
9. R. G. Roberts, R. Jones, M. A. Smith, Nature 345, 153
(1990).
10. R. G. Roberts et al., Quat. Sci. Rev. 13, 575 (1994).
11.
㛬㛬㛬㛬 , Ancient TL 16, 19 (1998).
12. J. M. Bowler, D. M. Price, Archaeol. Oceania 33, 156
(1998).
13. A. Thorne et al., J. Hum. Evol. 36, 591 (1999).
14. D. R. Horton, in Quaternary Extinctions: A Prehistoric
Revolution, P. S. Martin, R. G. Klein, Eds. (Univ. of
Arizona Press, Tucson, AZ, 1984), pp. 639–680.
15. A. Baynes, Rec. West. Austr. Mus. (suppl. 57), 391
(1999).
16. D. J. Meltzer, J. I. Mead, in Environments and Extinc-
tions: Man in Late Glacial North America, J. I. Mead,
D. J. Meltzer, Eds. (Center for the Study of Early Man,
Univ. of Maine, Orono, ME, 1985), pp. 145–173.
17. J. P. White, T. F. Flannery, Austr. Archaeol. 40,13
(1995).
18. S. Van Huet, R. Gru¨n, C. V. Murray-Wallace, N. Red-
vers-Newton, J. P. White, Austr. Archaeol. 46,5
(1998).
19. C. S. M. Turney et al., Quat. Res. 55, 3 (2001).
20. Optical ages were calculated from the burial dose
(paleodose), measured using the photon-stimulated
luminescence (PSL) signal, divided by the dose rate
due to ionizing radiation (21, 22). The portion of each
sample exposed to daylight was first removed under
dim red illumination and discarded. Quartz grains in
three ranges of diameter (90 to 125 m, 180 to 212
m, and 125 to 250 m) were extracted from the
remaining sample using standard procedures (22) and
etched in 40% hydrofluoric acid for 45 min. As a test
of internal consistency, some stratigraphic units were
dated using more than one sample or grain-size
fraction. Paleodoses were obtained using single-ali-
quot regenerative-dose protocols, statistical models,
and experimental apparatus as described (35–37).
Each aliquot was illuminated for 100 to 125 s at
125°C, and paleodoses were calculated from the first
3to5sofPSLarising from the burial, regenerative,
and test doses, using the final 20 s as background.
Each sample was given a preheat plateau test (22)
using aliquots composed of ⬎100 grains, and a re-
peat regenerative dose was given to verify that the
protocol yielded the correct (known) dose (35, 36).
The paleodoses in Table 2 were obtained from ali-
quots typically composed of ⬍10 grains to permit
detection of insufficient bleaching before burial from
examination of the paleodose distribution (37). For
samples with clearly asymmetrical (positively
skewed) distributions, mean paleodoses were calcu-
lated using the minimum age model (35); the central
age model (35) was used for other samples. For some
samples, paleodoses were also obtained using the
standard multiple-aliquot additive-dose method (22).
The dose rates due to
238
U and
232
Th (and their
daughter products) and due to
40
K were calculated
from a combination of high-resolution gamma and
alpha spectrometry (to check for disequilibria in the
238
U and
232
Th decay series), thick-source alpha
counting, x-ray fluorescence of powdered samples,
and field measurements of the gamma dose rate.
Cosmic-ray dose rates were estimated from pub-
lished data [J. R. Prescott, J. T. Hutton, Radiat. Meas.
23, 497 (1994)] and an effective internal alpha dose
rate of 0.03 Gy ka
⫺1
was assumed for all samples.
Gamma and beta dose rates were corrected for the
estimated long-term water content of each sample
and for beta-dose attenuation [V. Mejdahl, Archae-
ometry 21, 61 (1979)]. We used the sediment far-
thest from the bone to date the museum specimens
to minimize any dose rate heterogeneity in the sed-
iments adjacent to the bone. The gamma dose rates
for the museum specimens were estimated from
the attached lumps of sediment and from sedi-
ment-bone mixtures, using an uncertainty of
⫾20% to accommodate any spatial inhomogeneity
in the gamma radiation field. This uncertainty was
also applied to field samples collected without
measuring the in situ gamma dose rate; in situ
measurements had uncertainties of less than ⫾5%.
Some samples had a significant deficit or excess of
238
U with respect to
226
Ra (see Table 2). The
optical ages for these samples were calculated
using the measured radionuclide concentrations,
but any error due to post-burial uranium migration
should be accommodated within the age uncer-
tainties.
21. D. J. Huntley, D. I. Godfrey-Smith, M. L. W. Thewalt,
Nature 313, 105 (1985).
22. M. J. Aitken, An Introduction to Optical Dating: The
Dating of Quaternary Sediments by the Use of Pho-
ton-Stimulated Luminescence (Oxford Univ. Press,
Oxford, 1998).
23. J. Field, J. Dodson, Proc. Prehist. Soc. 65, 275 (1999).
24. L. K. Ayliffe et al., Geology 26, 147 (1998).
25. K. C. Moriarty, M. T. McCulloch, R. T. Wells, M. C.
McDowell, Palaeogeogr. Palaeoclimatol. Palaeoecol.
159, 113 (2000).
26. Ages corrected for detrital
230
Th contamination were
obtained by determining the thorium and uranium
isotope compositions for different splits of the same
flowstone (each split containing different proportions
of the detrital end member and the pure authigenic
calcite phase). Mixing line plots of
230
Th/
232
Th versus
238
U/
232
Th, and of
234
U/
232
Th versus
238
U/
232
Th,
provide estimates of the detritally corrected
230
Th/
238
U and
234
U/
238
U ratios as well as the isotope
ratios of the detrital end member phases:
230
Th/
232
Th ⫽ 0.37 ⫾ 0.04,
234
U/
232
Th ⫽ 0.00 ⫾ 0.03
(Kudjal Yolgah Cave);
230
Th/
232
Th ⫽ 0.36 ⫾ 0.02,
234
U/
232
Th ⫽ 0.00 ⫾ 0.02 (Mammoth Cave, upper
flowstone);
230
Th/
232
Th ⫽ 0.28 ⫾ 0.03,
234
U/
232
Th ⫽ 0.01 ⫾ 0.03 (Mammoth Cave, lower flow-
stone); and
230
Th/
232
Th ⫽ 0.29 ⫾ 0.04,
234
U/
232
Th ⫽ 0.00 ⫾ 0.01 (Moondyne Cave, using the
average detrital end member isotope ratios from five
nearby sites). The half-lives of
234
U and
230
Th used in
the age calculation are 244,600 ⫾ 490 years and
75,381 ⫾ 590 years, respectively.
27. See the supplementary figure at Science Online
(www.sciencemag.org/cgi/content/full/292/5521/
1888/DC1).
28. G. J. Prideaux, G. A. Gully, L. K. Ayliffe, M. I. Bird, R. G.
Roberts, J. Vertebr. Paleontol. 20 (suppl. to no. 3),
62A (2000).
29. Single-grain optical dating and finite mixture models
[R. G. Roberts, R. F. Galbraith, H. Yoshida, G. M.
Laslett, J. M. Olley, Radiat. Meas. 32, 459 (2000)]
were used to distinguish paleodose (and hence age)
populations in the sediment samples. Multiple dis-
crete populations were identified, which we attribute
to the mixing of grains with different burial histories.
The populations with the highest paleodoses ( Table
2) yielded optical ages consistent with the
14
C ages
obtained from pieces of charcoal (23).
30. We cannot be certain that articulated remains were
recovered from the upper megafaunal unit at Mam-
moth Cave (site 24), so the optical age of 55 ⫾ 9ka
was not included in the data set.
31. P. S. Martin, D. W. Steadman, in Extinctions in Near
Time: Causes, Contexts, and Consequences, R. D. E.
MacPhee, Ed. (Kluwer Academic/Plenum, New York,
1999), pp. 17–55.
32. P. S. Martin, Science 179, 969 (1973).
33. R. N. Holdaway, C. Jacomb, Science 287, 2250
(2000).
34. Much of the 60- to 40-ka interval was marked by
generally wetter conditions than at present in both
eastern Australia (24, 38) [G. C. Nanson, D. M. Price,
S. A. Short, Geology 20, 791 (1992); J. M. Bowler,
Archaeol. Oceania 33, 120 (1998)] and southwestern
Australia [J. Balme, D. Merrilees, J. K. Porter, J. R. Soc.
West. Austr. 61, 33 (1978), using the revised chro-
nology for Devil’s Lair (19)]. But monsoonal activity
may have been variable with short-lived climatic
oscillations (38), in keeping with evidence from deep-
sea cores of climate instability [J. P. Sachs, S. J.
Lehman, Science 286, 756 (1999); S. L. Kanfoush et
al., Science 288, 1815 (2000)].
35. R. F. Galbraith, R. G. Roberts, G. M. Laslett, H. Yoshida,
J. M. Olley, Archaeometry 41, 339 (1999).
36. H. Yoshida, R. G. Roberts, J. M. Olley, G. M. Laslett,
R. F. Galbraith, Radiat. Meas. 32, 439 (2000).
37. J. M. Olley, G. G. Caitcheon, R. G. Roberts, Radiat.
Meas. 30, 207 (1999).
38. B. J. Johnson et al., Science 284, 1150 (1999).
39. We thank S. Eberhard, J. Field, G. Gully, L. Hatcher, R.
McBeath, D. Merrilees, G. Miller, R. Molnar, K. Mori-
arty, A. Ritchie, I. Sobbe, the late G. van Tets, J.
Wilkinson, D. Witter, T. Worthy, and R. Wright for
sample collection, field assistance, and discussions;
the Western Australian Museum and Museum of
Victoria for permission to access their collections; M.
Olley for preparing the gamma spectrometry sam-
ples; R. Galbraith for mixture modeling; and R.
Gillespie and O. Lian for comments. Supported by a
Large Grant and a Queen Elizabeth II Fellowship from
the Australian Research Council (R.G.R.).
27 February 2001; accepted 25 April 2001
R EPORTS
8 JUNE 2001 VOL 292 SCIENCE www.sciencemag.org1892
