Quaternary Science Reviews 238 (2020) 106338

Contents lists available at ScienceDirect

Quaternary Science Reviews

journal homepage: [Link]/locate/quascirev

Late Pleistocene lake level history of Lake Mungo, Australia

Timothy T. Barrows a, b, *, Kathryn E. Fitzsimmons c, Stephanie C. Mills a, Jacqui Tumney d,
Daryl Pappin e, Nicola Stern d
a
School of Earth, Atmospheric and Life Sciences, University of Wollongong, Australia
b
School of the Environment, Geography and Geosciences, University of Portsmouth, United Kingdom
c
Research Group for Terrestrial Palaeoclimates, Max Planck Institute for Chemistry, Hahn-Meitner-Weg 1, 55128, Mainz, Germany
d
Department of Archaeology and History, La Trobe University, Bundoora, VIC, 3086, Australia
e
Heritage Division, New South Wales Office of Environment and Heritage, Buronga, Australia

a r t i c l e i n f o a b s t r a c t

Article history: Lake Mungo is a currently dry lake basin in the semi-arid zone of southeastern Australia. The transverse
Received 7 November 2019 dune system on the downwind side contains a record of human occupation of international importance.
Received in revised form It also contains one of the most continuous records of climate change over the last glacial cycle in the
9 April 2020
Australia desert. In this paper we provide a framework for the interpretation of lake level history from
Accepted 23 April 2020
before the arrival of people (>41 ka) until after the establishment of the pastoral industry in the area. We
Available online 18 May 2020
present 83 optically stimulated luminescence ages from the Lake Mungo lunette. The lake level history is
reconstructed from 34 stratigraphic sections along three transects through the lunette. The dating re-
Keywords:
Lake Mungo
veals considerable lake level fluctuations through time which occur over a depth range of ~10 m in the
Optically stimulated luminescence basin. At its height, probably at multiple times before 20 ka, the lake held more than 1 km3 of water and
Lake levels at its final level at ~19 ka, contained only 0.03 km3. The inception of Lake Mungo appears to have taken
Australia place during the mid-Pleistocene between ~256 and 369 ka. During the last glacial cycle, Lake Mungo
Arid zone was almost continuously wetter than present from shortly after 60 ka until ~19 ka. The Upper Mungo,
Pleistocene Arumpo and Zanci units represent a succession of lake filling and drying events, briefly interspersed by
soils. The final Zanci unit does not represent a single high lake phase, but an initial lake filling followed by
a series of short-lived lake level events within a brief period of a few thousand years. At the conclusion of
this event, the lake remained dry until the present day. Four OSL ages from a linear dune upwind of Lake
Mungo indicates regional aeolian activity from ~30 ka until present. Widespread erosion that produces
the characteristic topography of the lunette began after the arrival of British pastoralists and traditional
aboriginal ways of life overlapped briefly with this erosion. The presence of water in Lake Mungo closely
corresponds to periods when regional surface temperature was colder than present during the late
Pleistocene. Our new data supports a model that decreased evaporation and increased runoff were
primarily responsible for increased availability of surface water in the hydrological cycle.
© 2020 Elsevier Ltd. All rights reserved.

1. Introduction human remains were dated to 41 ± 4 ka (Olley et al., 2006), rep-

resenting one of the earliest forays of anatomically modern humans
Lake Mungo, a lake basin in the semi-arid zone of eastern out of Africa. The timing of human arrival in Australia remains
Australia, contains a valuable record of environmental change and highly controversial (Clarkson et al., 2017; O’Connell et al., 2018) as
some of the earliest archaeological traces on the continent, does the impact of the arrival on the megafauna and the environ-
including the world’s oldest known cremation and ritual ochre ment (Hamm et al., 2016; Miller et al., 1997, 2016). These changes
burial (Bowler et al., 1970, 1972, 2003). Australia’s oldest recorded occurred against a backdrop of considerable continental climatic
variability over the last 100,000 yr (Reeves et al., 2013; Williams
et al., 2009).
Lake Mungo lies within a chain of playas in the Willandra Lake
* Corresponding author. School of Earth, Atmospheric and Life Sciences, Univer- system (Fig. 1 inset). These lakes are presently dry but were
sity of Wollongong, Australia.
formerly fed by overflow of the Willandra Creek, a distributary of
E-mail address: [Link]@[Link] (T.T. Barrows).

[Link]
0277-3791/© 2020 Elsevier Ltd. All rights reserved.
2 T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338

the Lachlan River which has its headwaters in the eastern Austra- staggered filling and progressive salinisation of water through
lian highlands. As a playa, the Lake Mungo floor frequently deflated evaporation down the system from a sandy lunette at Lake
and the sediment accumulated as an arcuate transverse dune sys- Mulurulu in the north to a gypsum-dominated lunette at Chibnal-
tem (lunette sensu Hills (1940)) on the downwind, eastern margin. wood Lakes in the south (Fig. 1; inset; Bowler (1998); Bowler et al.
The lunette has preserved an almost continuous record of lake level (2012)).
change during the late Pleistocene (Bowler, 1998). The lake level The climate contrast between Lake Mungo and its eastern
record is important in the context of the general absence of highland catchment is illustrated through the comparison of
continuous records of climate change in the arid zone of Australia climate stations of nearby Mildura, downstream of where the
(Fitzsimmons et al., 2013). The archaeological record accompanying Lachlan joins the Murray River, and Goulburn at the top of the
the lake level changes provides an opportunity to study human catchment (Bureau of Meteorology, 2020). Goulburn has a mean
responses to large-scale environmental changes over long time- summer/winter temperature of 20.0/7.2
C and Mildura 23.8/
scales (Stern et al., 2013). 10.6
C. Mean annual precipitation at Goulburn is 618 mm/year and
The focus of most previous studies has been the antiquity of the Mildura is half that at 286 mm/year. Mean evaporation at Goulburn
human remains and activity traces, initially entirely using radio- is 1263 mm/year and Mildura is almost twice that at 2188 mm/year.
carbon dating (Barbetti and Polach, 1973; Bowler et al., 1972; The lunette is composed of sandy sediments interbedded with
Gillespie, 1997, 1998)). Thermoluminescence (TL) dating subse- evaporite-rich deflationary sediments (Bowler, 1976). During high
quently provided a means of directly dating Aboriginal fireplaces lake levels at Lake Mungo, waves and currents driven by prevailing
and sediment deposition (Adams and Mortlock, 1974; Bell, 1991; westerly winds washed sandy sediments to the eastern lake edges,
Huxtable and Aitken, 1977; Oyston, 1996)). More recently, uranium- forming beaches and source bordering dunes. During periods of
series and electron-spin resonance techniques provided a means of fluctuating low lake levels, efflorescence of salts on the exposed
directly dating human remains, but these techniques are subject to lake floors resulted in the formation of clay pellets that were blown
large errors (Thorne et al., 1999). Increased availability and into crescentic transverse dunes (lunettes). Thus, beach gravels and
improved protocols for optically stimulated luminescence (OSL) sandy dunes reflect periods of persistent high-water levels, and
dating, coupled with the ideal suitability of this technique of clay-rich sediments reflect low lake conditions and higher salinity.
aeolian sediments such as those forming the Lake Mungo lunette Soils developed during periods of landscape stability, associated
(e.g. Fitzsimmons et al. (2010)), has led to better chronologies with lake dry phases or temporary cut-off in sediment supply, such
(Bowler et al., 2003; Fitzsimmons et al., 2014; Olley et al., 2006). as from lake filling (Bowler, 1998). Severe erosion since the last lake
This technique provides more reliable luminescence ages than TL filling has exposed the internal stratigraphy of the lunette and has
and on a much longer timescale than radiocarbon. Optical dating provided a means for studying its depositional history.
also dates sediment deposition, and therefore landscape evolution, Inflow into Lake Mungo is a channel in the north that accepted
directly. overflow from Lake Leaghur at a sill depth of ~67 m above sea level
The Willandra Lakes is one of the most intensively dated Qua- (Fig. 2). The lowest point on the Lake Mungo floor is in the southern
ternary locations in Australia, with more than 200 radiocarbon basin at 61 m. The lunette rises typically 20e40 m above the lake
dates published and more than 100 ages from other techniques floor. There are four main shorelines on the margin of Lake Mungo.
such as TL, OSL, uranium series and ESR. However, radiocarbon The earliest and highest at ~74 m is a cliff cut into the oldest soils of
dating of charcoal is practically limited to about 40 ka BP because the lunette. This wave erosion is most likely the result of lake full
contamination is common in lunette settings (Gillespie, 1997, 1998). status (~71 m) and wave action associated with the long fetch
Additionally, older luminescence ages show a great deal of incon- across the lake basin from the west. At a lake level depth of 11 m
sistency between workers for the same unit, such as those sum- (72 m), the lake would overflow back into Willandra Creek, rep-
marised in Bowler (1998). Despite the earlier work, there are still resenting the maximum level the lake can obtain. At this level, a
significant spatial and temporal gaps in the stratigraphic record, minor new channel forms from the overflow of the north basin of
providing an incomplete environmental history. To help contribute Lake Leaghur (Fig. 2). The lake also overflowed at an early stage
to closing these gaps, we present 83 OSL ages for five locations on south of Joulni into two smaller playas. At levels lower than 71 m,
the lunette, providing one of the largest chronological datasets for the margins have been significantly modified by the deposition of
the Australian arid zone. By documenting 34 sediment sections beach and lunette sediments. Two further high-water marks are
along three transects we reassess the stratigraphy of the lunette, present at ~69 m and ~66 m, stratigraphically formed much closer
focussing on the eastern margin where the sedimentary record is together in time compared to the earlier shoreline. The 69 m
most complete. Lastly, we combine the dating and stratigraphy to shoreline is represented by gravel beaches on the western margin,
review the chronology of sediment deposition and reconstruct the and carbonate tubules on the eastern margin of the lake at the
lake level history in the context of regional climate change. The Walls of China (Fig. 1). The 66 m shoreline is the break in slope at
hydrological record provides an opportunity to link the arid zone the base of the lunette. A ridge at 64 m marks the last major
records with the more numerous records in the wetter margins of shoreline and then a series of regressive shorelines lie below this
temperate eastern Australia and provides a more robust context for level to 62 m, representing the last significant water level in the
studies of human responses to changing environments in the lake (Fig. 2).
region. Fig. 2 shows the area of Lake Mungo under the scenarios asso-
ciated with the four main lake shorelines. Lake area and volume for
2. Geomorphology and setting each scenario are presented in Table 1.
Deflationary sediments associated with the lowest shorelines
Lake Mungo is the second southernmost of five major lakes cover the highest shoreline in most places. Silty dune sediments
which formed a terminal lake playa system of the Willandra Creek forming an irregular surface with relief varying by 2 m cover the
during dry phases but overflowed to the Murray River through the modern lake floor. Linear and irregular dune sand sheets border the
Prungle lakes during the wettest phases (Fig. 1 inset). The lake western margin of Lake Mungo (and encroach the upper shoreline
represents a shallow kidney-shaped basin cut into sediments on the western margin) but do not encroach the youngest shore-
deposited within a ~6 million year old marine incursion (Bowler lines, indicating most of the sand movement was contemporary or
et al., 2006). Sediments deposited by the Willandra Lakes reflect pre-dates its formation. The last deposition in the lake consists of
 T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338 3

Fig. 1. Map of the Lake Mungo area. Sampling sites BLW1-4 ¼ Blowouts 1 to 4. MSH ¼ Mungo sandhill sampling site. WOC ¼ “Walls of China” Tourist site. PM ¼ “Palaeomagnetic
site”. Triangles indicate the lowest point on the lake floor and the highest point on the lunette. Also indicated are the position of significant geomorphic features and sites. Inset
shows the position of Fig. 1 within the Willandra Lakes chain. LM ¼ Lake Mulurulu, WC¼Willandra Creek, LG ¼ Lake Garnpung, LL ¼ Lake Leaghur, OA¼Outer Arumpo,
CL¼Chibnalwood Lakes, PL¼Prungle Lakes.

filling of small basins or soaks adjacent to the base of the lunette originally recognised and dated by Bowler (1971), Bowler et al.
(Fig. 1). The modern lunette serves as a catchment for run-off that (1972) and Polach et al. (1970). These stratigraphic subdivisions
accumulates in soaks along the margin of the lake floor. were revised by subdividing the Mungo unit into an Upper and
Three stratigraphic units, the Golgol, Mungo and Zanci, were Lower and introducing the Arumpo unit (Bowler, 1998), dividing
4 T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338

Fig. 2. Lake area at Lake Mungo for the area the lake would occupy at the 62, 64, 66, 69 and 71 m contours. Arrows indicate potential flow directions at sills between Lake Mungo
and Joulni (JS), Willandra Creek (WS) and Lake Leaghur (LS).

the lunette into five main units. These units are defined by sedi- 2.1. Golgol Member
mentology, matching to lake levels, and are usually separated by
soils. Here we informally group the units into the Lake Mungo The Golgol is the oldest unit and comprises the core of most of
Formation, comprised of five members, otherwise we continue to the main lunette. At the Palaeomagnetic Site (Bowler (1998); Fig. 1)
refer them as “units”. dune sands lie at the base with overlying pelletal clays, as an upper
and lower pair, indicating a major and prolonged high lake level
 T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338 5

Table 1 unrectified aerial photographs of the entire lunette, 1:10,000

Lake area and volume for lake levels at Lake Mungo, “þ” includes Joulni playas. georectified aerial photographs of the central lunette, 1:5,000
Lake level Volume (km3) Area (km2) georectified aerial mosaic along 20 km of the lunette and a 1 m
71 m 1.145 142
gridded elevation data for essentially the same area as the 1:5,000
71 m (þ) 0.083 22 mosaic(~26 km2). For the remainder of the Lake Mungo area, a 5 m
69 m 0.865 136 LiDAR DEM (DFSI Spatial Services, NSW Government) was used to
66 m 0.467 129 study the geomorphology of the Lake Mungo basin and calculate
64 m 0.221 116
lake levels (Fig. 1). Lake Level reconstructions were created
62 m 0.028 59
following similar methods to Barrows et al. (2014). Lake volume
was calculated by creating a triangulated irregular network (TIN)
surface from the digital elevation model. The ArcGIS Polygon Vol-
phase followed by lake drying and deflation of the exposed lake
ume tool was then used to calculate volume below a given lake
bottom. Significant pedogenesis over the upper few metres of the
level, and surface area. All heights referred to in the text are based
unit, which would require much more than ~20,000 years
on the LiDAR digital elevation model (DEM) with a horizontal da-
compared to the modern soil which has only nodular caliche, has
tum of GDA94, a vertical datum of AHD71 (local geoid model) in
resulted in the formation of thick rhizomorphic calcrete. The period
MGA zone 54H.
of deposition is not well constrained by dating, but TL and OSL
Four transects across erosional blowouts in the lunette (BLW1 to
dating indicates an age greater than 100 ka (Bowler and Price, 1998;
4) were selected for study (Fig. 1), of which three have accompa-
Fitzsimmons et al., 2014; Oyston, 1996).
nying detailed archaeological studies (Blowouts 1 (¼Locality
966617) and 2 (¼Locality 940691) (Tumney, 2018) and Blowout 3
2.2. Lower Mungo Member (¼Locality 969660) (Stern, 2015). A small blowout on the lake floor
(BLW4; Locality 950673), was investigated for its potential as an
This unit is composed of lakeshore beach and dune sands and archaeological survey site. In addition, a road cut through a linear
represents the first significant lake-full phase during the last glacial dune to the west of Lake Mungo (MSH) was also sampled (Fig. 1).
cycle. A soil usually separates the Upper from the Lower Mungo The sedimentology at each of the sampling sites was logged. Beds
Member. Extensive dating of this unit in association with the oldest were correlated between sites based on soil stratigraphy and sim-
human remains places its inception before 55 ka (Bowler et al., ilarity in texture and other physical properties. Prominent soils and
2003), lasting until ~39 ka (Fitzsimmons et al., 2015). marker beds were used to trace the major units laterally, using
erosional ‘residual’ sediment sections in the blowouts. Major units
2.3. Upper Mungo Member were then sampled for OSL to constrain their age. The number of
samples was biased towards the younger sediments to provide a
A pelletal clay unit marks a lake level drop and deflation from chronostratigraphic context for ongoing archaeological studies and
the lake floor subsequent to the Lower Mungo high lake level phase. because the oldest units have the best exposure at the Joulni sec-
The basal age of this unit is dated at approximately 40 ka with tions (Fig. 1) where considerable dating has already focused (see
deposition continuing until 35 ka (Bowler, 1998). Bowler et al. Bowler et al. (2003)).
(2012) and Fitzsimmons et al. (2015) revised the conclusion of
this unit closer to 30 ka. 3.1. Blowout 1 (Locality 966617)

2.4. Arumpo Member This is the southernmost site located in the section of the lunette
with the highest relief, where the sedimentary units are stacked up
This unit is comprised of alternating sandy and pelletal clay to 40 m thick. BLW1 is to the north of the “Palaeomagnetic Site”
units representing deposition over a substantial period. The vari- studied by Barbetti and Allen (1972) and Bowler and Price (1998)
able sediments reflect oscillating lake levels with intermittent lake (Fig. 1). Thirty-two OSL ages were determined in 11 sections from
full conditions interspersed with short deflation periods. Bowler a near shore blowout and across the main lunette, with additional
(1998) dated this unit from near 35 to 22 ka and Bowler et al. sites to the north (BLW1; Table 2; Figs. 3 and 4).
(2012) placed this unit in a narrower period of 34e24.5 ka or
closer to 30 ka ((Bowler et al., 2012) Fig. 12). Lakeshore features 3.2. Blowout 2 (Locality 940691)
correlated within this unit are associated with carbonate deposi-
tion, which coats gravel and silcrete outcrops (that served as a stone The northernmost site contrasts with the southernmost because
tool source) on the western margin of the lake. the shape of the basin has resulted in a separation of the older and
younger components of the lunette system by ~600 m (Fig. 1).
2.5. Zanci Member Twenty-five OSL ages were determined from seven sections along a
transect extending from the younger lunette to the outer edge of
This is the last Pleistocene unit deposited on the lunette and the older lunette (BLW2; Table 2; Figs. 5 and 6).
consists of laminated sandy clays. The unit has a well-developed
red-brown earth soil reflecting a long period since deposition 3.3. Blowout 3 (Locality 969660)
ceased. This event is relatively short-lived, placed by Bowler (1998)
and Bowler et al. (2012) between ~23 ka and ~19 ka. This location represents an intermediate site, with a discontin-
uous inner lunette and a more prominent main lunette 200 m to
3. Site selection the east (Fig. 1). Wave erosion from lake highs has benched the
Golgol unit, removing some of the older units present at BLW1.
Sites for sampling for luminescence dating were initially Twenty-two OSL ages were determined from 14 sections at the
assessed using aerial photography and a digital elevation model front and back of the lunette (BLW3; Table 2; Figs. 7 and 8). Two
collected for the project. Imagery and LiDAR were collected by sites were sampled at the front of the lunette together with sections
Fugro Spatial Analysis (Melbourne) in 2007, including 1:10,000 across the thickest sediments of the main lunette.
6 T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338

Table 2
Optically stimulated luminescence data and ages for samples from Lake Mungo.

Lab ID Sample Depth Water DE (Gy) Radionuclide concentrations External g dose rate Cosmic ray dose rate Total dose rate Age estimate
ID (m) content (%) (Gy/ka) (Gy/ka) (Gy/ka) (ka)
K (%) Th (ppm) U (ppm)

Blowout 1
K2137 BLW1- 1.11 5±3 20.5 ± 0.7 0.78 ± 0.04 3.22 ± 0.16 0.48 ± 0.02 0.44 ± 0.00 0.18 ± 0.01 1.29 ± 0.05 15.9 ± 0.8
01
K2138 BLW1- 1.33 3±2 17.1 ± 0.4 0.18 ± 0.01 1.15 ± 0.06 0.26 ± 0.01 0.31 ± 0.00 0.18 ± 0.01 0.67 ± 0.02 25.7 ± 0.9
02
K2139 BLW1- 1.53 3±2 13.7 ± 0.3 0.24 ± 0.01 1.26 ± 0.06 0.24 ± 0.01 0.21 ± 0.00 0.17 ± 0.01 0.61 ± 0.02 22.6 ± 0.9
03
K2140 BLW1- 1.95 5±3 14.2 ± 0.4 0.23 ± 0.01 1.53 ± 0.08 0.26 ± 0.01 0.19 ± 0.00 0.16 ± 0.01 0.57 ± 0.02 25.0 ± 1.1
04
K2141 BLW1- 2.32 5±3 14.2 ± 0.9 0.31 ± 0.02 1.57 ± 0.08 0.28 ± 0.01 0.20 ± 0.00 0.16 ± 0.01 0.62 ± 0.02 22.8 ± 1.6
05
K2142 BLW1- 2.54 15 ± 5 >102 0.74 ± 0.04 3.31 ± 0.17 0.55 ± 0.03 0.31 ± 0.00 0.15 ± 0.01 1.02 ± 0.05 >100
06
K2147 BLW1- 3.43 5±3 27.8 ± 0.9 0.68 ± 0.03 6.29 ± 0.31 0.73 ± 0.04 0.37 ± 0.00 0.14 ± 0.01 1.17 ± 0.05 23.7 ± 1.2
11
K2148 BLW1- 3.56 5±3 19.6 ± 1.0 0.28 ± 0.01 2.10 ± 0.11 0.26 ± 0.01 0.26 ± 0.00 0.13 ± 0.01 0.67 ± 0.02 29.5 ± 1.8
12
K2149 BLW1- 4.14 5±3 29.4 ± 1.8 0.55 ± 0.03 2.79 ± 0.14 0.30 ± 0.02 0.33 ± 0.00 0.12 ± 0.01 0.93 ± 0.04 31.5 ± 2.3
13
K2150 BLW1- 3.71 5±3 42.2 ± 2.0 0.58 ± 0.03 3.34 ± 0.17 0.39 ± 0.02 0.50 ± 0.00 0.13 ± 0.01 1.15 ± 0.04 36.7 ± 2.1
14
K2152 BLW1- 4.22 5±3 40.5 ± 1.8 0.65 ± 0.03 2.92 ± 0.15 0.36 ± 0.02 0.49 ± 0.00 0.12 ± 0.01 1.17 ± 0.04 34.5 ± 2.0
16
K2153 BLW1- 4.53 5±3 63.3 ± 2.9 1.36 ± 0.07 5.54 ± 0.28 0.70 ± 0.04 0.68 ± 0.00 0.12 ± 0.01 1.92 ± 0.09 33.0 ± 2.1
17
K2154 BLW1- 4.56 5±3 80.8 ± 5.0 1.83 ± 0.09 10.15 ± 0.51 0.94 ± 0.05 0.90 ± 0.01 0.12 ± 0.01 2.54 ± 0.11 31.8 ± 2.4
18a
K2155 BLW1- 4.56 5±3 97.0 ± 5.0 1.90 ± 0.09 8.85 ± 0.44 0.96 ± 0.05 0.90 ± 0.01 0.12 ± 0.01 2.56 ± 0.12 37.9 ± 2.6
18 b
K2156 BLW1- 5.21 5±3 98.6 ± 3.8 2.13 ± 0.11 9.46 ± 0.47 1.11 ± 0.06 0.81 ± 0.01 0.11 ± 0.01 2.65 ± 0.13 37.2 ± 2.4
19a
K2157 BLW1- 5.21 5±3 107 ± 6 2.05 ± 0.10 9.00 ± 0.45 1.21 ± 0.06 0.81 ± 0.01 0.11 ± 0.01 2.60 ± 0.13 41.2 ± 3.2
19 b
K2158 BLW1- 0.71 5±3 0.07 ± 0.01 0.33 ± 0.02 1.84 ± 0.09 0.24 ± 0.01 0.19 ± 0.01 0.19 ± 0.02 0.68 ± 0.03 0.10 ± 0.02
20
K2159 BLW1- 1.16 3±2 0.12 ± 0.01 0.18 ± 0.01 1.12 ± 0.06 0.20 ± 0.01 0.12 ± 0.01 0.18 ± 0.01 0.48 ± 0.02 0.25 ± 0.02
21
K2160 BLW1- 1.37 5±3 8.7 ± 0.6 0.29 ± 0.01 1.68 ± 0.08 0.19 ± 0.01 0.10 ± 0.01 0.18 ± 0.01 0.54 ± 0.02 16.2 ± 1.4
22
K2161 BLW1- 1.7 3±2 11.3 ± 0.5 0.17 ± 0.01 1.22 ± 0.06 0.20 ± 0.01 0.08 ± 0.00 0.17 ± 0.01 0.42 ± 0.02 27.0 ± 1.6
23
K2162 BLW1- 0.88 5±3 15.0 ± 0.6 0.34 ± 0.02 1.88 ± 0.09 0.23 ± 0.01 0.12 ± 0.01 0.19 ± 0.01 0.60 ± 0.03 24.8 ± 1.5
24
K2163 BLW1- 4.71 5±3 57.9 ± 2.5 0.40 ± 0.02 2.64 ± 0.13 0.32 ± 0.02 0.28 ± 0.00 0.17 ± 0.01 0.77 ± 0.03 75.3 ± 4.2
25
K2164 BLW1- 5.09 3±2 61.7 ± 2.4 0.28 ± 0.01 2.10 ± 0.11 0.29 ± 0.01 0.25 ± 0.00 0.11 ± 0.01 0.63 ± 0.02 97.5 ± 5.0
26
K2165 BLW1- 6.05 3±2 144 ± 9 0.67 ± 0.03 3.88 ± 0.19 0.53 ± 0.03 0.40 ± 0.00 0.10 ± 0.01 1.10 ± 0.04 130 ± 10
27
K2166 BLW1- 5.48 3±2 70.9 ± 3.1 0.79 ± 0.04 4.76 ± 0.24 0.60 ± 0.03 0.51 ± 0.00 0.11 ± 0.01 1.33 ± 0.05 53.4 ± 3.1
28
K2167 BLW1- 4.88 3±2 14.0 ± 0.5 0.12 ± 0.01 1.17 ± 0.06 0.22 ± 0.01 0.22 ± 0.00 0.11 ± 0.01 0.47 ± 0.01 29.5 ± 1.3
29
K2168 BLW1- 3.3 5±3 15.1 ± 0.7 0.21 ± 0.01 1.63 ± 0.08 0.20 ± 0.01 0.28 ± 0.00 0.14 ± 0.01 0.62 ± 0.02 24.3 ± 1.3
30
K2169 BLW1- 4.6 5±3 14.1 ± 0.4 0.25 ± 0.01 1.95 ± 0.10 0.29 ± 0.01 0.23 ± 0.00 0.12 ± 0.01 0.59 ± 0.02 23.8 ± 1.0
31
K2170 BLW1- 0.92 3±2 17.6 ± 0.6 0.33 ± 0.02 1.94 ± 0.10 0.29 ± 0.01 0.29 ± 0.00 0.19 ± 0.01 0.78 ± 0.03 22.6 ± 1.1
32
K2171 BLW1- 1.55 5±3 9.2 ± 0.2 0.18 ± 0.01 1.11 ± 0.06 0.20 ± 0.01 0.19 ± 0.00 0.17 ± 0.01 0.53 ± 0.02 17.3 ± 0.7
33
K2172 BLW1- 2.26 5±3 15.8 ± 0.5 0.41 ± 0.02 1.93 ± 0.10 0.27 ± 0.01 0.22 ± 0.00 0.16 ± 0.01 0.74 ± 0.03 21.5 ± 1.1
34
K2173 BLW1- 3.22 5±3 25.6 ± 1.0 0.46 ± 0.02 2.10 ± 0.11 0.27 ± 0.01 0.38 ± 0.00 0.14 ± 0.01 0.92 ± 0.03 27.9 ± 1.5
35
Blowout 2
K2175 BLW2- 0.50 3±2 0.13 ± 0.01 0.51 ± 0.03 2.86 ± 0.14 0.33 ± 0.02 0.30 ± 0.00 0.20 ± 0.02 0.96 ± 0.04 0.14 ± 0.01
02
K2176 BLW2- 0.15 3±2 0.09 ± 0.01 0.44 ± 0.02 2.58 ± 0.13 0.33 ± 0.02 0.32 ± 0.00 0.21 ± 0.03 0.93 ± 0.04 0.10 ± 0.01
03
K2177 BLW2- 0.27 5±3 0.09 ± 0.00 0.71 ± 0.04 3.33 ± 0.17 0.44 ± 0.02 0.37 ± 0.00 0.20 ± 0.02 1.18 ± 0.05 0.07 ± 0.00
04
K2178 0.80 3±2 21.9 ± 0.53 0.53 ± 0.03 2.60 ± 0.13 0.35 ± 0.02 0.35 ± 0.00 0.19 ± 0.01 1.01 ± 0.04 21.8 ± 1.0
 T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338 7

Table 2 (continued )

Lab ID Sample Depth Water DE (Gy) Radionuclide concentrations External g dose rate Cosmic ray dose rate Total dose rate Age estimate
ID (m) content (%) (Gy/ka) (Gy/ka) (Gy/ka) (ka)
K (%) Th (ppm) U (ppm)

BLW2-
05
K2179 BLW2- 1.14 3±2 23.4 ± 0.56 0.58 ± 0.03 3.07 ± 0.15 0.40 ± 0.02 0.41 ± 0.00 0.18 ± 0.01 1.10 ± 0.04 21.2 ± 0.9
06
K2180 BLW2- 1.51 3±2 26.5 ± 0.61 0.59 ± 0.03 2.98 ± 0.15 0.41 ± 0.02 0.45 ± 0.00 0.17 ± 0.01 1.14 ± 0.04 23.2 ± 1.0
07
K2181 BLW2- 2.23 3±2 12.0 ± 0.38 0.22 ± 0.01 1.29 ± 0.06 0.21 ± 0.01 0.23 ± 0.00 0.16 ± 0.03 0.60 ± 0.02 20.0 ± 0.9
08
K2182 BLW2- 2.51 10 ± 5 16.0 ± 0.86 0.44 ± 0.02 2.57 ± 0.13 0.37 ± 0.02 0.25 ± 0.00 0.15 ± 0.01 0.77 ± 0.03 20.8 ± 1.4
09
K2183 BLW2- 2.80 5±3 14.4 ± 0.3 0.31 ± 0.02 2.01 ± 0.10 0.26 ± 0.01 0.19 ± 0.00 0.15 ± 0.01 0.62 ± 0.02 23.3 ± 1.0
10
K2184 BLW2- 0.42 3±2 0.12 ± 0.01 0.72 ± 0.04 3.89 ± 0.19 0.46 ± 0.02 0.37 ± 0.00 0.20 ± 0.02 1.20 ± 0.05 0.10 ± 0.01
11
K2185 BLW2- 0.71 3±2 21.1 ± 0.5 0.56 ± 0.03 2.69 ± 0.13 0.34 ± 0.02 0.35 ± 0.00 0.19 ± 0.02 1.03 ± 0.04 20.5 ± 0.9
12
K2186 BLW2- 1.13 3±2 19.4 ± 0.8 0.46 ± 0.02 2.26 ± 0.11 0.49 ± 0.02 0.36 ± 0.00 0.18 ± 0.01 0.98 ± 0.03 19.9 ± 1.1
13
K2187 BLW2- 1.33 5±3 37.7 ± 1.9 1.62 ± 0.08 6.20 ± 0.31 0.79 ± 0.04 0.63 ± 0.00 0.18 ± 0.01 2.12 ± 0.10 17.8 ± 1.2
14
K2188 BLW2- 1.44 5±3 41.7 ± 1.7 1.61 ± 0.08 6.55 ± 0.33 0.84 ± 0.04 0.73 ± 0.00 0.17 ± 0.01 2.21 ± 0.10 18.9 ± 1.2
15
K2189 BLW2- 1.98 3±2 14.3 ± 0.6 0.34 ± 0.02 1.80 ± 0.09 0.27 ± 0.01 0.25 ± 0.00 0.16 ± 0.01 0.72 ± 0.03 19.9 ± 1.1
16
K2190 BLW2- 2.70 3±2 11.6 ± 1.2 0.18 ± 0.01 1.38 ± 0.07 0.31 ± 0.02 0.18 ± 0.00 0.15 ± 0.01 0.52 ± 0.02 22.3 ± 2.5
17
K2191 BLW2- 1.93 3±2 12.8 ± 0.5 0.16 ± 0.01 1.28 ± 0.06 0.30 ± 0.02 0.14 ± 0.00 0.16 ± 0.01 0.48 ± 0.02 26.5 ± 1.4
18
K2192 BLW2- 2.73 5±3 21.7 ± 0.8 0.37 ± 0.02 1.91 ± 0.10 0.25 ± 0.01 0.34 ± 0.00 0.15 ± 0.01 0.82 ± 0.03 26.6 ± 1.3
19
K2193 BLW2- 3.13 5±3 11.4 ± 0.6 0.14 ± 0.01 1.19 ± 0.06 0.16 ± 0.01 0.13 ± 0.00 0.14 ± 0.01 0.41 ± 0.01 27.9 ± 1.7
20
K2194 BLW2- 3.68 3±2 18.2 ± 0.6 0.23 ± 0.01 1.37 ± 0.07 0.23 ± 0.01 0.21 ± 0.00 0.13 ± 0.01 0.55 ± 0.02 33.0 ± 1.6
21
K2196 BLW2- 3.22 3±2 140 ± 6 0.23 ± 0.01 1.76 ± 0.09 0.26 ± 0.01 0.26 ± 0.00 0.14 ± 0.01 0.55 ± 0.02 256 ± 14
23
K2197 BLW2- 4.51 3±2 152 ± 10 0.16 ± 0.01 1.30 ± 0.07 0.23 ± 0.01 0.23 ± 0.00 0.12 ± 0.01 0.41 ± 0.02 369 ± 27
24
K2198 BLW2- 0.97 3±2 12.9 ± 0.6 0.20 ± 0.01 1.67 ± 0.08 0.24 ± 0.01 0.24 ± 0.00 0.19 ± 0.01 0.54 ± 0.02 23.7 ± 1.4
25
K2199 BLW2- 2.17 5±3 12.7 ± 0.4 0.20 ± 0.01 1.36 ± 0.07 0.17 ± 0.01 0.17 ± 0.00 0.16 ± 0.01 0.56 ± 0.02 22.8 ± 1.0
26
K2200 BLW2- 3.59 5±3 12.0 ± 0.5 0.23 ± 0.01 1.48 ± 0.07 0.21 ± 0.01 0.21 ± 0.00 0.13 ± 0.01 0.57 ± 0.02 21.0 ± 1.1
27
K2201 BLW2- 5.43 3±2 13.7 ± 0.4 0.28 ± 0.01 1.46 ± 0.07 0.24 ± 0.01 0.24 ± 0.00 0.11 ± 0.01 0.53 ± 0.02 25.9 ± 1.3
28
Blowout 3
K2202 BLW3- 0.86 3±2 17.1 ± 0.6 0.39 ± 0.02 2.42 ± 0.12 0.32 ± 0.02 0.39 ± 0.00 0.19 ± 0.01 0.94 ± 0.03 18.3 ± 0.9
01
K2203 BLW3- 1.17 3±2 17.6 ± 0.9 0.32 ± 0.02 2.07 ± 0.10 0.29 ± 0.01 0.33 ± 0.00 0.18 ± 0.01 0.81 ± 0.03 21.6 ± 1.3
02
K2204 BLW3- 1.43 5±3 11.6 ± 0.4 0.30 ± 0.02 1.86 ± 0.09 0.26 ± 0.01 0.36 ± 0.00 0.17 ± 0.01 0.81 ± 0.02 14.3 ± 0.6
03
K2205 BLW3- 2.25 5±3 25.4 ± 0.9 0.62 ± 0.03 3.00 ± 0.15 0.37 ± 0.02 0.25 ± 0.00 0.16 ± 0.01 0.95 ± 0.04 26.8 ± 1.5
04
K2206 BLW3- 2.85 5±3 17.0 ± 1.1 0.23 ± 0.01 1.72 ± 0.09 0.20 ± 0.01 0.38 ± 0.00 0.15 ± 0.01 0.75 ± 0.02 22.8 ± 1.5
05
K2207 BLW3- 4.45 5±3 19.4 ± 1.2 0.32 ± 0.02 2.17 ± 0.11 0.25 ± 0.01 0.20 ± 0.00 0.12 ± 0.01 0.62 ± 0.02 31.4 ± 2.2
06
K2208 BLW3- 2.96 5±3 16.5 ± 1.1 0.29 ± 0.01 1.74 ± 0.09 0.24 ± 0.01 0.25 ± 0.00 0.14 ± 0.01 0.65 ± 0.02 25.2 ± 1.8
07
K2209 BLW3- 3.99 3±2 9.7 ± 0.5 0.09 ± 0.00 0.89 ± 0.04 0.17 ± 0.01 0.17 ± 0.00 0.13 ± 0.01 0.40 ± 0.01 24.5 ± 1.6
08
K2210 BLW3- 1.08 3±2 13.8 ± 0.5 0.30 ± 0.01 1.95 ± 0.10 0.28 ± 0.01 0.20 ± 0.00 0.18 ± 0.01 0.67 ± 0.02 20.6 ± 1.0
09
K2211 BLW3- 1.55 5±3 16.7 ± 0.7 0.36 ± 0.02 1.83 ± 0.09 0.22 ± 0.01 0.24 ± 0.00 0.17 ± 0.01 0.73 ± 0.03 22.8 ± 1.3
10
K2212 BLW3- 2.79 5±3 14.2 ± 0.6 0.28 ± 0.01 1.70 ± 0.09 0.23 ± 0.01 0.27 ± 0.00 0.15 ± 0.01 0.67 ± 0.02 21.1 ± 1.1
11
K2213 BLW3- 4.30 5±3 22.6 ± 1.1 0.78 ± 0.04 3.62 ± 0.18 0.47 ± 0.02 0.42 ± 0.00 0.12 ± 0.01 1.21 ± 0.05 18.7 ± 1.2
12
K2214 BLW3- 3.46 5±3 20.0 ± 1.5 0.88 ± 0.04 3.14 ± 0.16 0.46 ± 0.02 0.37 ± 0.00 0.14 ± 0.01 1.24 ± 0.06 16.1 ± 1.4
13
(continued on next page)
8 T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338

Table 2 (continued )

Lab ID Sample Depth Water DE (Gy) Radionuclide concentrations External g dose rate Cosmic ray dose rate Total dose rate Age estimate
ID (m) content (%) (Gy/ka) (Gy/ka) (Gy/ka) (ka)
K (%) Th (ppm) U (ppm)

K2215 BLW3- 5.25 5±3 10.8 ± 0.8 0.13 ± 0.01 1.22 ± 0.06 0.16 ± 0.01 0.13 ± 0.00 0.11 ± 0.01 0.38 ± 0.01 28.3 ± 2.4
14
K2216 BLW3- 6.37 5±3 13.1 ± 0.6 0.22 ± 0.01 1.21 ± 0.06 0.18 ± 0.01 0.25 ± 0.00 0.10 ± 0.01 0.55 ± 0.02 23.9 ± 1.3
15
K2217 BLW3- 7.77 3±2 11.9 ± 0.5 0.13 ± 0.01 0.90 ± 0.05 0.17 ± 0.01 0.10 ± 0.00 0.08 ± 0.01 0.31 ± 0.01 38.6 ± 2.1
16
K2218 BLW3- 8.07 5±3 11.9 ± 0.6 0.24 ± 0.01 1.49 ± 0.07 0.22 ± 0.01 0.26 ± 0.00 0.08 ± 0.01 0.57 ± 0.02 20.9 ± 1.2
17
K2219 BLW3- 3.48 5±3 0.13 ± 0.02 0.22 ± 0.01 1.27 ± 0.06 0.16 ± 0.01 0.18 ± 0.00 0.13 ± 0.01 0.51 ± 0.02 0.25 ± 0.04
18
K2220 BLW3- 1.06 5±3 8.3 ± 0.5 0.31 ± 0.02 1.87 ± 0.09 0.21 ± 0.01 0.20 ± 0.00 0.18 ± 0.01 0.66 ± 0.02 12.6 ± 0.9
19
K2221 BLW3- 1.41 5±3 12.3 ± 0.4 0.24 ± 0.01 1.38 ± 0.07 0.17 ± 0.01 0.23 ± 0.00 0.17 ± 0.01 0.62 ± 0.02 19.9 ± 0.9
20
K2222 BLW3- 3.03 5±3 15.7 ± 1.5 0.25 ± 0.01 1.83 ± 0.09 0.22 ± 0.01 0.26 ± 0.00 0.14 ± 0.01 0.64 ± 0.02 24.5 ± 2.5
21
K2223 BLW3- 3.54 5±3 13.8 ± 0.9 0.32 ± 0.02 1.99 ± 0.10 0.23 ± 0.01 0.28 ± 0.00 0.13 ± 0.01 0.71 ± 0.02 19.5 ± 1.5
22
K2224 BLW3- 4.29 5±3 13.5 ± 0.5 0.27 ± 0.01 1.66 ± 0.08 0.20 ± 0.01 0.27 ± 0.00 0.12 ± 0.01 0.64 ± 0.02 21.0 ± 1.0
23
Blowout 4
K2225 BLW4- 0.45 5±3 0.10 ± 0.01 0.41 ± 0.02 2.15 ± 0.11 0.22 ± 0.01 0.31 ± 0.00 0.20 ± 0.02 0.86 ± 0.03 0.12 ± 0.01
01
K2226 BLW4- 1.12 5±3 0.16 ± 0.03 0.50 ± 0.02 2.35 ± 0.12 0.24 ± 0.01 0.33 ± 0.00 0.18 ± 0.01 0.93 ± 0.03 0.17 ± 0.03
02
Linear dune
K2227 MSH-01 0.38 5 ± 3 1.20 ± 0.12 0.18 ± 0.01 1.46 ± 0.07 0.16 ± 0.01 0.09 ± 0.00 0.20 ± 0.02 0.46 ± 0.02 2.6 ± 0.3
K2228 MSH-02 0.82 5 ± 3 6.5 ± 0.52 0.18 ± 0.01 1.48 ± 0.07 0.20 ± 0.01 0.09 ± 0.00 0.19 ± 0.02 0.46 ± 0.02 14.0 ± 1.3
K2229 MSH-03 1.27 5 ± 3 9.6 ± 1.12 0.18 ± 0.01 1.58 ± 0.08 0.19 ± 0.01 0.10 ± 0.00 0.18 ± 0.01 0.46 ± 0.02 21.0 ± 2.6
K2230 MSH-04 1.95 5 ± 3 13.3 ± 0.66 0.18 ± 0.01 1.53 ± 0.08 0.20 ± 0.01 0.10 ± 0.00 0.16 ± 0.01 0.44 ± 0.02 30.0 ± 1.9

3.4. Blowout 4 (Locality 950673) common in Zanci and Arumpo soils and care was taken to avoid
nests and bioturbation (Williams, 2019) by only sampling undis-
Stratigraphically the youngest site, this blowout occurs in an turbed bedding. However, BLW1-01 was sampled between nests
isolated dune of the frontal lunette between blowouts 2 and 3 on because of a lack of exposure and this sample gave an anomalously
the lake floor (Fig. 1). Erosion of the dune has created a small young age. Each sample was collected using a 10 cm long stainless-
catchment that can hold a shallow basin of water after heavy rain. steel tube with a diameter of 4 cm hammered into a cleaned sur-
Around the shore of this basin (63.5 m) are many hearths which face. We consider that in all cases that this depth interval records a
post-date the widespread erosion of the lunette. Two OSL ages very brief depositional period, well within measurement un-
were determined from a pit dug in the adjacent deflated sediments certainties. Depths and unit thicknesses were surveyed using a total
to constrain the timing of the waterhole formation (BLW4; Table 2; station relative to a survey monument placed on the lake floor in
Fig. 9). 2007 with a position determined by differential GPS.

3.5. Linear dune 4.2. OSL methodology

Lake Mungo is surrounded by linear dune field and sediment Samples were opened and processed under low intensity red
mobilised from these dunes contributed to sediment deposition in and yellow-orange light at the luminescence dating laboratory,
the lake, either windblown or through encroachment. The top Research School of Earth Sciences, Australian National University.
2.4 m of a dune cut by the road to the east of the lake was dated Only the sediment in the central section of the tubes was processed
using four OSL ages to investigate the timing of dune construction for dating to minimise the risk of potential exposure to light during
(MSH; Table 2; Fig. 9). collection.
Sand-sized (120e210 mm) quartz was isolated according to
4. Stratigraphy and sampling strategy published methods (Fitzsimmons et al., 2007) including sieving and
density separation in sodium polytungstate solution, digestion in
4.1. Sampling dilute hydrochloric acid, hydrogen peroxide and hydrofluoric acid.
The resulting clean quartz grains were then mounted onto the
Sampling for OSL was undertaken at erosional residuals or central 1 mm of 10 mm diameter stainless steel discs using silicone
gullies where stratigraphy was suitably exposed. Zones of pedo- oil; 18 aliquots were prepared from each sample for equivalent dose
genesis were avoided which were identified by either a red, oxi- (De) measurement.
dised zone at the top of the unit, organic staining and the presence Equivalent dose measurements were undertaken using auto-
of charcoal, loss of sedimentary structure through bioturbation or mated Risø TL-DA-12 and TL-DA-15 readers equipped with blue
decalcification and the presence of a Bca horizon. Termites are light-emitting diodes for light stimulation and calibrated 90Sr/90Y
 T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338 9

Fig. 3. Stratigraphic logs and dating results at Blowout 1 (BLW1). The same legend applies for Figs. 5, 7 and 9. Colour scheme is indicative only. (For interpretation of references to
colour in these figures, the reader is referred to the Web version of this article.)
10 T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338

Fig. 4. 3D view of Blowout 1 (BLW1) showing relative position of sampling sites to lunette morphology. S1 is positioned on an inner lunette in front of the main lunette.

beta sources for irradiation (Bøtter-Jensen et al., 2000). Lumines- incorporating the moisture content to account for dose rate
cence signals were detected by EMI 9235QA photomultiplier tubes attenuation (Mejdahl, 1979). The external dose rate for MSH and
with coated Hoya U-340 filters (Bøtter-Jensen, 1997). Samples were BLW1-20 to BLW1-24 are derived from radioisotope concentra-
measured using the single-aliquot regenerative-dose (SAR) proto- tions. The doses for all other samples are calculated from the
col (Murray and Wintle, 2000, 2003), incorporating an infrared gamma spectrometry.
stimulated luminescence (IRSL) wash prior to each OSL signal In situ moisture content of some samples was calculated by
measurement (Murray and Wintle, 2003). Following a preheat weighing the raw and oven-dried weight of material from the ends
plateau test on sample BLW1-1 (Suppl. Fig. 1A), a preheat of 260
C of the tubes, averaged with the moisture content from samples
and cutheat of 220
C for 10 s was used for all samples. In most containing the surrounding sediment. Because erosion has
instances, the sample aliquots yielded Gaussian distributions, dramatically increased drainage, exposed sections were much drier
indicating complete bleaching of the signal, and therefore the than experienced during most of their buried history. Water con-
Central Age Model of Galbraith et al. (1999) was used to calculate tents were therefore corrected by relating samples from exposed
the equivalent dose for those samples. Supplementary Table 1 lists sections with deeply buried samples of the same sediment type,
the samples for which “over-dispersion” was observed and shows characterised by grain size and trace element content. All samples
where the Minimum Age Model was most appropriate in prefer- had a water content of 3e5%, except BLW2-09 (10%) and BLW1-06
ence to the Central Age Model. (15%).
The cosmic ray component of the dose rate was calculated based
on sample latitude, longitude, altitude and depth, and assuming a
4.3. Dose rates sediment overburden density of 1.9 g cm3, using published equa-
tions (Prescott and Hutton, 1994). Due to the widespread deflation
Environmental radiation dose rates from the quartz-rich sands of the lunette, some sites were formerly buried up to 5 m below the
of Lake Mungo lunette sediments are typically very low (Table 2), surface, resulting in an overestimate of cosmic ray dose if the
yielding attenuated values on the order of 0.5e1.0 Gy/ka with the modern depth is used. To improve the accuracy of the cosmic ray
exception of the laminated clay-bearing units (such as the Upper dosage, original burial depth was estimated by trigonometry using
Mungo) which yield total dose rates at least twice as great. This survey points on the surface of the soil on adjacent residuals.
characteristic can often result in relatively wide dose distributions Where the soil was partially eroded, total soil depth was estimated
due to low counts and dose rate heterogeneity (Lomax et al., 2007), by measuring the depth from the soil surface to the Bca horizon on
and the cosmic ray component of the dose rate can be a significant adjacent residuals. This depth correction may slightly underesti-
proportion of the total dose rate (Burrough et al., 2007; Telfer et al., mate the cosmic ray dose where there were large gaps in time
2017). This means care must be taken to use realistic burial depths between the deposition of the dated sediment and that of subse-
for the samples. In situ gamma spectrometry was conducted at all quent sediment cover. Based on the dating results, erosion was very
sites in the same sample holes using a portable sodium iodide recent and therefore the increase in cosmic ray dose since exposure
gamma spectrometer with a three-inch crystal detector. Counting is negligible.
was undertaken for 30 minutes per sample. All OSL ages in the text are reported in thousands of years (ka)
Sediment collected from around the sample holes was homo- with 1s uncertainty and are relative to the processing date (2008
genised, split, and subsampled for elemental analyses in order to CE). Where radiocarbon dates have been referred to in the text, they
calculate beta dose rates. Subsamples were ground in a ball mill and have been calibrated to calendar years using the SHCal13 calibra-
analysed using induction-coupled plasma mass spectrometry (ICP- tion (Hogg et al., 2013; Reimer et al., 2013) in conjunction with
MS) for U and Th, and induction-coupled plasma atomic emission CALIB7.0 (Stuiver and Reimer, 1993) without any reservoir effect on
spectrometry (ICP-AES) for K, undertaken at Genalysis Laboratories, biogenic carbonates (Gillespie et al., 2009). Radiocarbon dates are
Perth, Western Australia. These concentrations were converted to expressed as radiocarbon years (14C yr BP) and calibrated dates as
dose rates using the factors of Adamiec and Aitken, (1998),
T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338 11

Fig. 5. Stratigraphic logs and dating results at Blowout 2 (BLW2).

12 T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338

Fig. 6. 3D view of Blowout 2 (BLW2) showing relative position of sampling sites to lunette morphology. S1-4 are positioned on an inner lunette in front of the main lunette. Key as
for Fig. 4.

calibrated years (cal yr BP). Unless otherwise noted, means are error-weighted, and the un-
certainty is the larger of the internal or external error on the mean.
4.4. Bayesian statistics
5. Results
Age inversions are often observed in luminescence chronologies
derived from rapidly accumulated sand deposits with very low dose Ages calculated for the OSL samples are presented in Table 2.
rates (e.g., Bowler (1998)). Age inversions can be interrogated, and Table 3 presents the results of the Bayesian analysis and gives the
chronological precision improved, using Bayesian statistical ana- likely unit ages, durations and interval lengths together with the
lyses. The Bayesian approach takes into account the probability respective uncertainties for the younger three units. The OSL
distributions of the ages and prior knowledge such as stratigraphic behaviour of the samples from Lake Mungo indicate that these
information (ie. a sample stratigraphically above another must be sediments are generally well suited to OSL dating using the SAR
younger). protocol (see supplementary information). All samples exhibit
We conducted a Bayesian analysis for the three best represented bright OSL signals with rapid decay indicative of sensitive quartz
stratigraphic units using OxCal (Bronk Ramsey, 2009). Where which is dominated by the fast component and contains negligible
possible, ages with poor agreement (A<60%) with the rest of the feldspar (Suppl. Fig. 1B). The high sensitivity of these samples is
sequences were iteratively removed until the overall agreement attributed to sensitisation through multiple cycles of reworking
index (Ac) was ~>60%. Where an age with poor agreement was within the Quaternary sedimentary system (Pietsch et al., 2008).
excluded from the analysis, a percentage chance that the sample IRSL signals, measured prior to the OSL signal in order to identify
should be in this position in the sequence was calculated (in potential contamination by feldspar (Murray and Wintle, 2003), are
brackets in the following). For the Zanci unit, we used the sections low to non-existent. Thermal transfer is generally <2%, with
BLW1 S1 (minus BLW1-01, 5.8%; BLW1-02, 0%), BLW2 S3 and S2 exception of several very young samples, and recycling values were
(which were averaged) and BLW3 S1þ2, S4þS5 (minus BLW3-03, all within 10% of unity. Most samples yield dose response charac-
0%) and S6eS8. For the Arumpo unit, we used BLW1 S2eS7 teristics corresponding to a simple exponential function
(minus BLW1-11, 0%; BLW1-33, 0%; BLW1-35, 0.1%) and BLW2 (Suppl. Fig. 1C), with the exception of BLW1-06 from the Golgol
S6þS4. BLW3 was excluded because soils were not present. For the unit, which was saturated with respect to the quartz OSL signal.
Upper Mungo unit we only used BLW1 S2eS7 with duplicates of
BLW1-18 and 19 combined. We calculated the likely boundary ages 5.1. Stratigraphy
and duration for each sampled sequence. These boundary ages
provide a good indication of the limits of deposition where samples 5.1.1. Golgol Member
were taken close to the beginning and end of unit. The estimation of This unit was present at all three sites along the main lunette as
boundary ages relies heavily on the assumption that the distribu- the basal unit. This is the thickest unit at the sampled locations, at
tion of the ages is representative of the unit and the full age range is least 12 m thick at BLW1 and displays varying degrees of pedo-
covered. For our sites, this is likely in most cases where we have genesis. At both BLW1 and 2, where it is made more resistant by
multiple ages and because sedimentation rates are high. Lastly, the calcrete, it is benched by an undated high lake phase. At BLW2,
interval between depositional periods and its uncertainty was also caliche is up to 3 m thick, in comparison to the Bca of the Post-Zanci
independently estimated. Only BLW1 S2eS7 was used to calculate soil which is only carbonate nodules about 1 cm in diameter. Below
this for the Zanci and Upper Mungo units. Examples of plots the pedogenic zone, the unit consists of well-sorted light brown
showing prior and posterior distributions, together with boundary sand with shell fragments. This foredune facies represents a rela-
age estimates, are given in the supplementary information tively high lake level given its coarse sediment and elevation. The
(Suppl. Fig. 3aeb, 4). Normalised chi-squared tests (Х2/n) were two ages for the top and bottom of the unit differ by more than 100
conducted to assess the appropriateness of averaging multiple ka (256 ± 14 ka and 369 ± 27 ka) despite there being no obvious
sections together. A value close to unity indicates a single popula- stratigraphic break. This most likely reflects variation in dose rate
tion where scatter in the data can be accounted for solely by measurements in this very low dose rate environment, since the
random error. Where Х2/n[1, the average contains unexpected equivalent doses of the two samples are comparable (Suppl. Fig. 2)
scatter and where Х2/n≪1, the average is unexpectedly precise. whereas the calculated dose rates are 25% different from one
T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338 13

Fig. 7. Stratigraphic logs and dating results at Blowout 3 (BLW3).

14 T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338

Fig. 8. 3D view of Blowout 3 (BLW3) showing relative position of sampling sites to lunette morphology. S1-5 are positioned on an inner lunette in front of the main lunette. Key as
for Fig. 4.

Fig. 9. Stratigraphic logs at Blowout 4 (BLW4) and Mungo Sandhill 1 (MSH).

another (c. 0.4 vs. 0.55 Gy/ka; Table 2). 141 ± 35 ka.
Where exposed at the thickest part of the lunette (BLW1) the
Golgol consists of two units with pedogenesis to several metres 5.1.2. Lower Mungo Member
deep. Previous TL ages in this section are younger than the new This unit consists of a thin, discontinuous sand sheet that can be
ages, the lower unit having two ages of 113 ± 13 ka and 126 ± 16 ka traced laterally stratigraphically on top of, and behind, the wave-
and the upper unit dated at 98 ± 30 ka and 99 ± 17 ka (Bowler and cut cliff in the Golgol unit. The sand is reddish, light brown or
Price, 1998). In other sections of the lunette, Oyston (1996) dated a pink depending on the erosional source material, sourced from
unit at Joulni under the Lower Mungo that may be correlative and is thick, well-developed soils, beach sands or a mixture respectively.
also younger at 144 ± 15 to 178 ± 29 ka (J4 and J5). Finally, An incomplete soil profile at our sites, combined with the incor-
Fitzsimmons et al. (2014) dated one section between BLW1 and 3 at poration of eroded soils, makes it difficult to determine the length
 T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338 15

Table 3
Sedimentary unit boundary ages, duration, interval lengths and their uncertainty at Lake Mungo.

Member Age (yr) ± (yr) Х2/n

Zanci duration 3500 3900 553

upper 19800 1700 11.2
lower 23400 2800 4.4
interval 800 300
Arumpo duration 8900 600 0.1
upper 24100 1100 1.1
lower 32800 500 0.1
interval 1200 1200
Upper Mungo duration 2500 2500
upper 33900 1400
lower 37100 1700

of time of pedogenesis relative to the other units. At BLW1 there are at a site east of the burials on sediments correlated with the site of
two ages from one unit of 75.3 ± 4.2 and 97.5 ± 5.0 ka, with the the burials and obtained ages of 59 ± 3 ka and 63 ± 3 ka. These are
variation probably due to dose rate measurments. A single large significantly older than the previously reported ages but are
macropod skull was found at the top of the Lower Mungo at BLW1, consistent with subsequent extensive dating of the unit by Bowler
but no other large vertebrate fossils were found at any site in this et al. (2003). These authors found an age range of 62.2 ± 1.8 to
study. Further north, the unit shows poor outcrop and is not pre- 41.8 ± 2.4 ka. The presence of six soils and three distinct gravel
sent at BLW2. At BLW3, the unit is dated to 38.6 ± 2.1 ka (BLW3-16). bands indicating high lake phases indicates this unit was deposited
The unit is continuously exposed at the north end of the Tourist Site over a substantial period. These ages indicate that the Lower Mungo
(Fig. 1) where it is dated at 39.0 ± 3.2 to 51.1 ± 2.7 (Fitzsimmons lake level high was a significant period which lasted at least 20 ka.
et al., 2014).
Extensive hearths at BLW1 in the Lower Mungo unit provide 5.1.3. Upper Mungo Member
minimum limiting ages for its formation. This zone is a shallow soil This unit is not a major unit at the three sites investigated. At
containing reworked and rounded nodules of calcrete from the BLW1 this unit conformably overlies the Lower Mungo and is about
Golgol underneath. Also present are rare rounded cobbles of sil- 1 m thick (Fig. 3). It is a pelletal clay sediment, almost entirely clay
crete, often reddish in colour, which cannot be in situ because of at the base of unit but grades to sand rich with shell fragments at
their elevation (>81 m). These are probably emu or Genyornis the top. The clay unit has an age of 37.2 ± 2.4/41.2 ± 3.2 ka (BLW1-
gastroliths, the latter have been recorded up to 400 g in weight by 19) at the base and 37.9 ka ± 2.6 ka (BLW1-18; a duplicate has an
Stirling and Zeitz (1900). The sediment overlying the Golgol unit anomalous De) at the top of the clay rich unit, indicating it could
has an age of 130 ± 10 ka (BLW1-27; Fig. 3) but is highly likely to be have been deposited in a short period of time. There are 2 weakly
a mixed age sample. Overlying the Golgol soil is a grey soil with no developed soils in this unit and a well-developed soil (40 cm thick)
sedimentary structure, probably reworked by human activity, at the top. Three ages in the overlying sand-rich sub-units overlap
containing abundant hearths on its surface. A single OSL age at the at their 1 s errors and range from 34.5 to 36.7 ka. The unit is not
base of 54.3 ± 3.1 ka represents a minimum age for this unit in the present at BLW2 and the thin outcrop at BLW3 (Fig. 7) was not
area sampled. This soil correlates with the ash-rich zone identified dated.
by Barbetti and Allen (1972) at the nearby Palaeomagnetic Site. The At the Palaeomagnetic Site TL ages of 24.5 ± 3.7 (MWC3) and
OSL age is older than the radiocarbon ages on charcoal from hearths 29.3 ± 3.2 (MCW4) (Readhead, 1988) for the Upper Mungo are
at the top of the unit which range from 31,340 ± 250 (ANU-682) to much younger than our ages for this unit. Given that the thickness
34,700 ± 480 cal yr BP (ANU-680) (Barbetti and Allen, 1972). of the unit in that paper is around 3 m compared to the thin deposit
Thermoluminescence dating was applied to baked oven stones at we observed, it is possible that Arumpo aged sediments were
the same site which ranged between 29.5 and 38.6 ka, with one inadvertently sampled. Further north, the upper boundary of the
younger outlier (Huxtable and Aitken, 1977). Bell (1991) rein- unit lies above a fish hearth at the Tourist Site which was dated at
vestigated the site by dating baked sediment at the same fireplaces 38,860 þ 1390/-1270 cal yr BP (ANU-2964) (Clark in (Johnston and
and obtained TL ages of 31.4 ± 2.1 to 36.4 ± 2.5 ka. The sediment Clark, 1998) and (Bowler, 1998). Another hearth dated at the level of
itself had depositional TL ages of 29.3 ± 3.2 ka (Readhead, 1988) and the Lower Mungo soil (ANU-687) provides a maximum age
16.4 ± 1.2 ka (Bowler and Price, 1998). The radiocarbon ages and the 39,650 þ 1720/-1250 cal yr BP for the Upper Mungo (Barbetti and
TL ages scatter beyond expected for a single unit. These ages Polach, 1973). Both radiocarbon ages are consistent with the
together with the new OSL ages above may indicate a long period of inception of the unit as dated here (37.1 ± 1.7 ka).
exposure on this section of the Lower Mungo unit on the lunette. In the south, the equivalent of the Upper Mungo unit at the
Subsequently, the Arumpo unit was deposited at the Palae- Joulni burial site was dated using TL at 24.6 ± 2.4/34.0 ± 3.9 ka
omagnetic Site (see next section) with a hearth near the base dated (Oyston, 1996). The former selective bleach age being much
at 30,460 ± 300 cal yr BP (ANU-667) (Barbetti and Allen, 1972). younger than the total bleach age or the ages here. Bowler and Price
In contrast to the south-eastern margin, the Lower Mungo unit (1998) re-dated this unit to 29.3 ± 3.1 ka (W1798). Bowler et al.
is well exposed at Joulni where the burial sites of Mungo I and III are (2003) subsequently also re-dated the unit with ages of 37.8 ± 1.9
located. Initial TL dating indicated an age of 43.3 ± 3.8/43.1 ± 6.7 ka to 42.0 ± 1.7 ka. A third single age of 30.6 ± 1.1 near the top was
(J3) (Oyston, 1996). Bowler and Price (1998) obtained a statistically used to put an upper limit of 30 ka on the unit (Bowler et al., 2003),
identical age of 41.4 ± 6.7 ka (W1799) on the same section. At the which is 4000 years younger than the dated end of this unit (33.9
Mungo I burial site, ages in the Lower Mungo unit range from ~48 ka), which was followed by a hiatus of at least 1000 years.
ka at the top of an augur hole to ~51 ka at the base with large errors,
assuming cosmic ray doses decreasing with depth (Bowler and 5.1.4. Arumpo Member
Price, 1998). Thorne et al. (1999) subsequently applied OSL dating Arumpo sediments conformably overlie the Upper Mungo
16 T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338

surface or lie on an erosional surface of either of the older two units. highest site is at BLW3 where the dune is situated at the top of the
The unit is present at the surface of the main lunette at BLW1, and break in slope above 75 m, to the north and south, the dune is at the
the eastern sections of the lunette at BLW2 and 3 (Figs. 6 and 8). A bottom of the break of slope at 65.5 m and 67.5 m at BLW2 and
diverse sedimentology is displayed within this unit representing a BLW1 respectively. No depositional shoreline is exposed at any of
dynamic range of lake level conditions over a considerable period. the sites, but the lowest elevation of these deposits indicates that
The unit consists of sand, laminated clayey sands and sandy clay, the lake had only 3.5 m water in it when it finally emptied. At BLW1
with interbedded clay units. At all sites the deflated sediments are and 3, the base of the dune is located on a Golgol soil, dated at the
dissociated from shoreline sediments leaving no clear link with former site at >100 ka. Despite the differences in elevation, the
lake level height. The maximum number of layers occur at BLW1 three sites all have similar ages and indicate that this final lake level
where two minor soils and three pedogenic zones indicating age phase was brief (23.4 ± 2.8 ka to 19.8 ± 1.7 ka), lasting only a few
gaps in deposition of sediment. None of the soils are as well thousand years (3.5 ± 3.9 ka).
developed as the soil on the Upper Mungo. The unit ranges in age At all three localities, the facies is similar representing a sandy
from 32.8 ± 0.5 to 24.1 ± 1.1 ka at the top of the lunette with some beach or foredune succeeded by a clay dune and sandy pelletal clay
age inversions. At the Palaeomagnetic site, a previous age of units. The most detail is recorded at BLW2. Two clay lenses indicate
30,460 þ 350/-260 cal yr BP (ANU-667) constraining the start of deposition of clay dunes during drying phases alternating with
sedimentation (Barbetti and Polach, 1973) is consistent with this foredune quartz sands. At BLW3 two clay layers are preserved in the
interval. thickest dune section (section 2). At BLW1, the sequence is split
This long period of fluctuating lake level included numerous between the clay dune at the front and a capping on the main
periods of high lake level of long enough duration for mussels and lunette (Fig. 3).
fish to establish in the lake. At BLW1 there are at least six major lake This unit correlates with the lowest shoreline mapped by
drying phases recorded by clay units like the Upper Mungo, four of Bowler (1998) at the front of the Tourist Site. As the unit with the
these concentrated into the final depositional phase (Fig. 3, S2). best exposure and the last deposited, it has been the focus of a
Only one unit is exposed at BLW3 and none at BLW2 indicating relatively large amount of dating. The best constraints on its for-
considerable heterogeneity in sediment deposition along the mation are supplied by radiocarbon ages on fish otoliths dated by
lunette. Quartz-rich dunes were deposited over the main lunette at Kalish et al. (1997) and by Long et al. (2014). These ages range from
all locations during intermittent high lake levels. At the back of the 19,370 þ 200/-200 cal yr BP (OZB-623) to 20,450 þ 160/-170 cal yr
main lunette at BLW2, there is a thick unit of rhythmically depos- BP (OZB-626) and 19,340 þ 140/-130 cal yr BP (SANU-8817) to
ited pelletal clay layers where individual layers are 1e2 mm thick. 21,100 þ 170/-170 cal yr BP (SANU-8814) respectively. One anom-
Therefore a 1 m section may be composed of 500e1000 cycles. The alous age of 13,630 þ 80/-60 cal yr BP (SANU-27810) was found at a
large number of cycles and the preservation indicates rapid filling stratigraphically lower level (Long et al., 2014). The final phase of
and drying of the lake over a few years at most. drying is possibly recorded by unionid shells in growth position
This unit is correlated with a well-defined shoreline along the from the Lake Mungo lakebed (17,580 þ 370/-370 cal yr BP; N-
front of the lunette associated with precipitation of microbialites 1656) (Gillespie, 1997; Yamasaki et al., 1977) but this is younger
(similar to microbialites in southern Victoria (Last and Last, 2015);) than all the other ages here. The Zanci unit has a well-developed
and tubular carbonate (Bowler, 1998). Direct radiocarbon ages on soil indicating a considerable lapse of time consistent with the
tubules places growth at 26,150 þ 400/-420 (N-1660) to final ages of deposition (~19 ka). There is no significant difference
27,490 þ 530/-510 cal yr BP (ANU-310) (Polach et al., 1970; between the soils developed on the Zanci or on uncovered Arumpo
Yamasaki et al., 1977). sediments. Elsewhere, Fitzsimmons et al. (2014) and Fitzsimmons
Although this is a major depositional unit, there are relatively et al. (2015) did not separate the Zanci and Arumpo where there
few ages from elsewhere. At Joulni, Oyston (1996) dated a unit at was no soil separating them. They refer to these laterally equivalent
19.5 ± 2.3/31.6 ± 4.0 ka (J1) remapped by Bowler and Price (1998) as sediments dated from 25 to 14 ka as Unit E. Sediments at the border
the Arumpo unit. The range of ages from selective to total bleach of their Upper Mungo/Arumpo boundary at ~23e25 ka are referred
bracket most of the ages presented here. Bowler and Price (1998) to as the Red Lunette Unit by Fitzsimmons et al. (2014).
obtained an age of 22.4 ± 2.1 ka (W1800) at the younger end of
our ages. Subsequent dating by Bowler et al. (2003) at that site 5.1.6. Walls Member
suggests the unit began deposition shortly after 30.3 ± 1.0 ka (M3T A new unit is proposed here, which is the youngest unit at Lake
73.3 m), consistent with our dated interval. Mungo, to accommodate the large-scale remobilisation and re-
deposition of sediment associated with erosion of the lunette.
5.1.5. Zanci Member The unit is named after the ‘Walls tank’ waterhole which derives its
This unit comprises the front section of the lunette at BLW1 and name from the adjacent ‘Walls of China’ Tourist site (Fig. 1), an area
the lake-ward parts of the lunette complex at BLW2 and 3 of distinctive erosion. The unit consists of slope wash units on the
(Figs. 3e8) whereas BLW4 is entirely eroded from the Zanci unit. flanks on the lunette, alluvial fans at the front of the lunette,
The most pronounced feature in the deposition of this unit is a extensive mobile dunes at the rear of the lunette and low “coppice”
change in lake geometry due to a decrease in lake volume (Fig. 2). dunes on the lake floor and near the tops of the lunette. The unit
The result is that most of the deflated sediment is in a series of has no association with lake deposits and is entirely deposited after
discontinuous clay dunes along the shore-face (Fig. 1). Similar to the final lake drying and deposition of the Zanci Unit. In
the Arumpo unit, there is considerable heterogeneity in sediment Fitzsimmons et al. (2014) these sediments are referred to as Units F
deposition along the lunette, with areas of 6 m sand dunes adjacent to J.
to sediment-free areas such as at BLW3 (Fig. 8). The greatest sep- The origin of the Walls unit can be determined from its associ-
aration of the clay dunes from the main lunette at our sites occurs ation with the Zanci soil at the three transects. The Zanci soil is
in the north at BLW2 at 600 m, due to the low slope angle of the sandy and friable once decalcified and where carbonate has leached
lake basin, which reduces south to 200 m at BLW3. At BLW3, the to the Bca horizon. The Zanci soil has been eroded over almost the
inner lunette is irregular with the presence of an embayment and entire lunette, frequently to the Bca which forms a resistant layer.
the Zanci unit is also deposited on the main lunette. However, the The Walls unit frequently begins as steeply dipping reddish slope
altitude of the Zanci dunes varies between the three localities. The wash layers on the front and back of the lunette, deposited over the
 T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338 17

soil downslope (e.g. Fig. 3; section 10; Fig. 7. Section 13). The unit agents.
forms alluvial fans at the front of the lunette out onto the lake floor
(e.g. Fig. 5; section 1,2). The sediment lightens to pink as the 5.2. Facies model
mixture of reworked soil decreases with dilution of white sand
from subsoil sandy clays. The soil is eroded as planar reddish sands Regardless of the age, the sediments in the Lake Mungo lunette
at front and back of the lunette, decreasing in a mix of blowout and display a recurring depositional pattern related to sediment supply
wash sands until pure white sand from blowouts. In places the unit as a function of lake level. The pattern is best typified in sections
has been stabilised by vegetation and moss/lichen cover as sedi- BLW1-S1 and BLW2-S3 and 4. Elements from these sites are used to
ment supply has decreased. At BLW3, dunes that formed during the construct a facies model (Fig. 10). This model is used to interpret the
1980s are now stabilised. The alluvial fan sands are often covered stratigraphy of the lunette to construct a lake level history.
with fine silt stabilised as a coppice dune (e.g. Fig. 5; section 2,4). The base of the model lies on a soil (Fig. 10). A well-developed
Some coppice dunes (mostly silt-fine sand) are stabilising the sur- soil on a pelletal clay sediment indicates the lake is dry, unless
face and thick sand sheets are being incised and gullied. At the rear there is a lateral equivalent sand dune unit. The first unit is a poorly
of the lunette, an extensive series of dunes remains mobile, con- sorted gravelly sand layer which heralds the first arrival of water in
sisting of sand winnowed from eroded sediments on the lunette the lake. This unit is rarely thick, probably because early in the
and deflated to the lee of the lunette. The mobile dune field is up to wetting cycle, groundwater levels are low, and salts have not yet
200 m wide and encroaches on the hinterland. This unit is actively accumulated in the system. There is probably limited sediment
forming and up to 6 m of dunes have been deposited at the back of availability and insufficient time for clay to accumulate on the floor
the lunette in the time between the two DEMs used in this paper or sand along the shore. Later low lake level phases in the cycle are
being collected. associated with high water tables and high salinities in the
The Walls unit is much younger than the Pleistocene units groundwater window (Bowler, 1971).
below and has an active depositional surface. At BLW1 at the back The next unit, dune sand, indicates a lake full phase, since sand
of the lunette (Fig. 3), slope-wash sands overlying the Zanci soil is not produced without wash action at the shore. The environment
date to 250 ± 20 yr and the base of the overlying coppice dune is of deposition is a foredune on the lake edge and grades down slope
100 ± 20 yr (Fig. 3). At BLW2 on the alluvial fan at the front of the into a massive or gravelly sand representing beach sediment.
inner lunette, the sand sheet ranges in age from 140 ± 10 to Foredune sand units are typically steeply bedded (dip angle of
100 ± 10 yr and the base of the coppice dune at the front of the repose) with up to 30 m wavelength dunes. They can contain fine,
inner lunette is 70 ± 10 yr (Fig. 5). At BLW3 at the rear of the main broken shell fragments (Velesunio), and rare fish otoliths. The sand
lunette there is a well-defined contact between erosion of the is usually well sorted, washed and winnowed from older units
lunette and the dunefield behind. The base of the reworked sedi- (such as the benched shore on the Golgol unit). The sediment
ments is dated at 260 ± 40 yr (Fig. 7). A short distance away along sometimes coarsens laterally towards the beach from massive fine
the contact with the Zanci soil, a fully intact sheep skeleton was sand to coarse sand and at the shoreline into pebbles and gravel.
found eroding out on top of the soil. Beach units (not shown) are steeply dipping sand interbedded with
It is important to note that the scale of modern erosion of the gravel, often lagged on an older surface. The gravel is sometimes
lunette is unparalleled in the last 40,000 years. Relict cliffing and silcrete but usually rounded caliche. The unit may contain weakly
gullying are present on the Golgol unit, but there is no equivalent to developed soils as sand supply along the beach varies. Sand is
the mobile dune exposed in the lunette. Significant erosion was transported along the shoreface by longshore currents and also
only recorded at one site within the lunette sediments. Below the deposited as spits. The lack of a delta between Lake Leaghur and
soil at BLW3, reworked sediments eroding from the Arumpo unit Lake Mungo indicates that sediment transport via overflow was
are rich in charcoal hearths and date to 12.7 ± 1.0 ka (Fig. 7, S14), but suspension load.
this erosion was limited when it occurred and subsequently sta- Long periods of lake highs result in clay deposition on the lake
bilised. Given the age of the reworked sediments of the Walls unit, floor. Upon drying of the lake, the abundant clay is then deflated as
the widespread erosion is most likely a result of pastoral practises clay pellets, accumulated mostly in the back dune setting in the lee
after the arrival of British settlers to the area in the 1860s to 1870s of the shoreface dunes. These are typically thin units that conform
and the arrival of rabbits in the last 150 years. The ages are to pre-existing topography. These units are often interbedded with
consistent with the clear association of the sheep skeleton with the sandy clay or clayey sand. The dunes have a shorter wavelength
soil contact. The young age is supported by the high observed rates (3e4 m) than the sandy dunes.
of erosion (e.g., 60 cm in 25 years in Bowler and Magee (2000), and The formation of clay pellets requires lake drying and so where
also the absence of soil developed on depositional surfaces. the units are rhythmic, they indicate intermittent lake filling.
Pelletal clays often form largely flat lying to shallow dipping dunes.
5.1.7. Linear dunes Minor pedogenic zones and clay interbedded subunits indicating
A linear dunefield extends up to the western shore of the lake some periods of variable lake filling may be interspersed. Sandy
and is eroded by the lowest shoreline, indicating that it pre-dates, lenses often contain shell or otolith fragments indicating lake
or is at least contemporary with the Zanci lake level high. North phases deep and long enough for mussels and fish to survive. The
of the visitors centre, dunes cover the higher shorelines indicating conclusion of the lake filling and drying phase is marked by the
they post-date the Arumpo high lake levels (Fig. 1). The basal date development of another soil once sediment transport ceases. An
for the linear dune above the highest shoreline was 30 ± 1.9 ka. indication of age is the degree of decalcification of the soil and the
Above this level, the dune sand is mixed showing continuous re- development of caliche (Fig. 10).
deposition from 21 ± 2.6 through to 2.6 ± 0.3 ka. The crests of
the dunes are currently active and reworked sand overlies a soil on 6. Discussion
the dune which presumably is pre-British colonisation in age
(Fig. 9, S1). In a similar setting, Readhead (1988) found ages ranging 6.1. Lake level history
from 41.5 ± 4.4 (MASH6) to 6.9 ± 2.3 (MASH2) in a longitudinal
sand dune to the west of the lake. The unexpected linear age trend Previous lake level curves for the Willandra Lakes have focused
in the section is suggestive of reworking by wind or biological on relative changes and relied on estimating lake levels based on
18 T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338

Fig. 10. Facies model for lunette sediments at Lake Mungo and the basis for interpreting lake levels. Depth is for illustrative purposes only.

the sedimentology of deflated sediments associated with lake from the presence of woody vegetation (Gillespie, 1998). Interpre-
shorelines (e.g. Bowler et al. (2012)). We follow the same approach tation of relative probability plots are subject to several problems,
here and use the facies model in conjunction with the sediment especially oversampling of surficial units and uneven sampling of
logs and surveyed shorelines to attempt a finer resolution, absolute units of interest. More precise dating produces relatively higher
lake level curve. We used the results from the Bayesian analyses to peaks. For Fig. 11c, areas have been normalised to a maximum
determine the duration of lake filling events and intervening lake height of 1 and then added together. The presence of fish and shell
dry periods (Table 3). We supplemented this dating with the pub- fossils through all units demonstrates at least brief periods of deep
lished radiocarbon ages from fossil shells and fish otoliths water and low solute content throughout the sequence.
described above. The lake level history of Lake Mungo (Fig. 11a) is Determining lake level history using deflated sediments in the
likely to be slightly shorter than lakes further up the Willandra lunette dune is problematic, since shorelines are usually not pre-
chain, but longer than lakes further down the chain (Bowler et al., sent in the sequence. Therefore, estimates of lake level have been
2012). Because the lake level records are likely to be diachronous made based on the sediment types and their association with
and the geometry and interlinking of the lake basins varies, the lake shoreline altitudes. The stratigraphic sections reveal six major
level history here is specific to Lake Mungo. drying phases during the Arumpo and two within the Zanci. The
Deposition of lunette sediments from the time of the Upper highest lake level during the Arumpo (~27 ka) was associated with
Mungo Unit to the Zanci Unit was continuous except for brief in- microbialites in the shallows of the shore edge (Fig. 1). Between
tervals when the lake was dry and not deflating. It is impossible to inferred lake level highs, levels were fluctuating, possibly on an
subdivide the lunette units lithologically (as shown by the facies interannual basis. Therefore, absolute heights and durations are
model) and this demonstrates that the style of lake filling and indicative only. For the older part of the curve (>37.1 ka), we used
emptying was similar throughout most of this period. The lake the reconstruction of Bowler et al. (2012) except that we used the
levels attained during the period are necessarily approximate, since measured maximum gravel height at the Joulni exposure (71 m) to
none of our ages were specifically on shoreline sediments. The lake approximately scale lake level depth during this period.
has occupied the same shorelines multiple times through its
history.
We updated the database of Bowler et al. (2012) for direct ages 6.2. Regional climate change
on the sediments using TL and OSL and Velesunio shell, and Mac-
quaria otolith ages (Fig. 11b and c). The luminescence dating in- Transport of water from the highlands in the east to the western
dicates the bulk of the sampled lunette is of Zanci or Arumpo age plains of southeastern Australia is recorded in fluvial sedimenta-
and that the errors overlap between the younger units. Shell and tion. Late Pleistocene rivers in the Murray-Darling Basin were
otolith dates are reliable indicators of multi-year water presence, substantially larger and carried coarser bedload (Hesse et al., 2018;
whereas charcoal does not indicate environmental conditions apart Kemp and Spooner, 2007). OSL ages from along the Lachlan River
indicate that large sinuous channels were fully established from
 T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338 19

Fig. 11. Panel diagram showing a) Lake level reconstruction, b) Relative probability plot of all luminescence ages from Lake Mungo. White area are previously published ages
(n ¼ 101; see text) and black area are new ages c) Relative probability plot of all otolith (n ¼ 41; white area) and unionid radiocarbon ages (n ¼ 10; black area), d) Sea-surface
temperature reconstruction for the Southern Ocean (Barrows et al., 2007).

before 67 ka until before 34 ka (Kemp and Rhodes, 2010; Kemp and discharge. Discharge declined to 915 m3s1at 24.2 ka then to 772
Spooner, 2007), covering most of the Lower Mungo phase. From 34 m3s1by 18.3 ka and to 219 m3s1by 16.1 ka. Kemp et al. (2017) have
ka to 20 ka bankfull discharges declined (Kemp and Rhodes, 2010). established that at 18.4 ka, or shortly before, drainage into the
This time period covers most of the Upper Mungo, Arumpo and Willandra Creek at least partly avulsed to a more southerly channel.
Zanci phases. Hesse et al. (2018) reviewed discharge estimates for It remains unclear whether this was due to silting up of the Will-
Murray-Darling Basin rivers. In the Lachlan River basin, discharge andra Creek channel or due to the dramatic drop-off in discharge to
was highest at 66.6 ka (2512 m3s-1) with about 20 times modern near modern levels. However, avulsion explains low lake levels in
20 T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338

the Willandra Lakes when cold conditions and glaciers persisted in 4) Most of the sediment on the lunette was deposited from 37.1
the Snowy Mountains (Barrows et al., 2001). Overall, higher until 19.8 ka during a period of drying in southeastern Australia.
discharge in the Pleistocene explains why Lake Mungo filled in an There were brief hiatuses during this period, but the production
otherwise low precipitation environment. of sediment was almost continuous. Adjacent linear dunes also
Lake level filling events at Lake Mungo are restricted to the developed in this interval.
coldest phases of the last glacial cycle (Fig. 11d). No continuous 5) The timing of lake filling supports previous observations of
temperature curve is available for mainland Australia, but offshore increased discharge and sediment supply from the eastern
sea-surface temperatures are similar to Antarctic air temperature highlands during the coldest periods of the last glacial cycle.
reconstructions (Barrows et al., 2007) and probably reflect the This was most likely a result of the combination of higher runoff
pattern of change in southern Australia. Initial lake filling at around coefficients combined with lower vegetation cover, increased
60 ka is similar to the time glaciers appeared in the Snowy spring melt from greater snowpack and periglacial activity.
Mountains during the last glacial cycle (Barrows et al., 2001, 2002). 6) Widespread erosion of the lunette accompanied the arrival of
The most significant production of sediment onto the lunette British pastoralists in the 19th century. Although some surfaces
occurred during the period when glaciers were present, from the have stabilised, erosion continues through to the present day.
Headley Tarn glaciation through to the Blue Lake glaciation
(~32e19 ka). Maximum cooling occurred when periglacial land-
forms were most active at 21.9 ± 0.5 ka (Barrows et al., 2004) and Author statement
this would have been the time of lowest treeline and highest pro-
portion of precipitation stored as snowpack in the mountains. After TB Investigation (Fieldwork), Formal analysis, Supervision,
19 ka, the lake underwent final drying, and the area remained arid Writing - Original Draft, KF Investigation (OSL), Writing - Review &
through the rest of the late Pleistocene and the Holocene, corre- Editing, SM Visualization (GIS), Writing - Review & Editing, JT
lating with rapid SST increase (Fig. 11d). Galloway (1965) estab- Investigation (Fieldwork). Writing - Review & Editing, DP Investi-
lished that the late Pleistocene water balance in the eastern gation (Fieldwork). Writing - Review & Editing, NS Funding
highlands was mostly controlled by temperature. Hesse et al. (2018) acquisition. Writing - Review & Editing.
have re-examined this finding using palaeochannel characteristics
in the Murray-Darling Basin. They found that discharge through Declaration of competing interests
time was sensitive to temperature change and higher runoff effi-
ciency, which to some extent is also a function of temperature The authors declare that they have no known competing
through vegetation cover, snowpack storage and periglacial limit. financial interests or personal relationships that could have
However, they could not quantify the effect of precipitation and this appeared to influence the work reported in this paper.
remains a largely unknown variable. It therefore appears likely that
temperature had the greatest impact on the hydrological cycle of Acknowledgements
the Lachlan River and Willandra Creek.
It was wetter than present at Lake Mungo from ~60 ka until We are grateful to the Three Traditional Tribal Groups from the
about 19 ka. The Willandra Lakes have a surface area of ~1088 km2, Willandra Lakes region, the Paakantyi/Barkindji, the Mutthi Mutthi,
and under the present evaporation-precipitation regime would and the Ngiyampaa, for permission to work at Lake Mungo. The late
require the total discharge of the modern Lachlan River to fill Ray Lawson, a Paakantyi Elder, provided valuable guidance in the
(Bowler, 1971). Many of the clayey sand pelletal clay sediments in field. Tegan Kelly kindly calculated dose rates from the gamma
the Arumpo and Zanci units are laminar and this reveals that there spectrometric data for KF. Rudy Frank conducted the surveying and
was a cyclical nature to the lake emptying and refilling on a very participated in many useful discussions in the field. We thank Eva
short time scale, either seasonal or on a longer interannual scale, Reynolds and Wayne Cook for help in the Research School of Earth
such as in the El Nin ~ o-Southern Oscillation. Flow was likely to be Sciences, Australian National University OSL laboratory. We also
very seasonal because of water storage as snow in the upper thank Jim Bowler for useful discussions in the field. This research
headwaters of the catchment. Reinfelds et al. (2014) showed the was funded as part of Australian Research Council Linkage grant
potential sensitivity of runoff in the Snowy Mountains to changes in LP0775058, in partnership with the Elders Council of the Three
temperature. Each 1
C decrease corresponds to a 17% increase in Traditional Tribal Groups and the Lower Darling Branch of the New
the runoff coefficient, possibly increasing runoff by a factor of four. South Wales Office of Environment and Heritage. Insightful com-
Otoliths indicate longer wet periods where the drying process took ments from Prof. Martin Williams and reviews from two anony-
place over many years after a lake filling (Long et al., 2014). mous reviewers improved the manuscript.

Appendix A. Supplementary data

7. Conclusions
Supplementary data to this article can be found online at
1) We used 83 OSL ages to define a new lake level history for Lake [Link]
Mungo. The presence of water, although intermittently,
throughout much of the period from 60 to 19 ka provides a
References
context from which to interpret human occupation of the area.
2) The inception of Lake Mungo was in the mid-Pleistocene prob- Adamiec, G., Aitken, M.J., 1998. Dose-rate conversion factors: update. Ancient TL 16
ably between ~256 and 369 ka. A scatter of ages older than 100 (2), 37e50.
Adams, G., Mortlock, A.J., 1974. Thermoluminescent dating of baked sand from fire
ka may indicate discontinuous deposition over several glacial
hearths at Lake Mungo, New South Wales. Archaeol. Phys. Anthropol. Ocean. 9
cycles. (3), 236e237.
3) Lake level fluctuated over a range of about 10 m and at the lake’s Barbetti, M., Allen, H., 1972. Prehistoric man at Lake Mungo, Australia, by 32,000
deepest level, it would have overflowed sills south to the Joulni years BP. Nature 240 (5375), 46e48.
Barbetti, M., Polach, H., 1973. ANU radiocarbon date list V. Radiocarbon 15 (2),
depression and north back into the Willandra Creek. The area of 241e251.
the lake varied from 164 km2 to 59 km2. Barrows, T.T., Juggins, S., De Deckker, P., Calvo, E., Pelejero, C., 2007. Long-term sea-
 T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338 21

surface temperature and climate change in the Australian-New Zealand region. Levchenko, V.A., Foley, E.C., Worthy, T.H., Stephenson, B., Coulthard, V.,
Paleoceanography 22, 2215, 2210.1029/2006PA001328. Coulthard, C., Wilton, S., Johnston, D., 2016. Cultural innovation and megafauna
Barrows, T.T., Stone, J.O., Fifield, L.K., 2004. Exposure ages for Pleistocene periglacial interaction in the early settlement of arid Australia. Nature 539, 280.
deposits in Australia. Quat. Sci. Rev. 23 (5e6), 697e708. Hesse, P.P., Williams, R., Ralph, T.J., Fryirs, K.A., Larkin, Z.T., Westaway, K.E.,
Barrows, T.T., Stone, J.O., Fifield, L.K., Cresswell, R.G., 2001. Late Pleistocene glacia- Farebrother, W., 2018. Palaeohydrology of lowland rivers in the Murray-Darling
tion of the kosciuszko massif, Snowy Mountains, Australia. Quat. Res. 55 (2), Basin, Australia. Quat. Sci. Rev. 200, 85e105.
179e189. Hills, E.S., 1940. The lunette, a new land form of aeolian origin. Aust. Geogr. 3 (7),
Barrows, T.T., Stone, J.O., Fifield, L.K., Cresswell, R.G., 2002. The timing of the last 15e21.
glacial maximum in Australia. Quat. Sci. Rev. 21 (1e3), 159e173. Hogg, A.G., Hua, Q., Blackwell, P.G., Niu, M., Buck, C.E., Guilderson, T.P., Heaton, T.J.,
Barrows, T.T., Williams, M.A.J., Mills, S.C., Duller, G.A.T., Fifield, L.K., Haberlah, D., Palmer, J.G., Reimer, P.J., Reimer, R.W., Turney, C.S.M., Zimmerman, S.R.H., 2013.
Tims, S.G., Williams, F.M., 2014. A White Nile megalake during the last inter- SHCal13 southern hemisphere calibration, 0e50,000 Years cal BP. Radiocarbon
glacial period. Geology 42 (2), 163e166. 55 (4), 1889e1903.
Bell, W.T., 1991. Thermoluminescence dates for the Lake Mungo aboriginal fire- Huxtable, J., Aitken, M.J., 1977. Thermoluminescent dating of Lake Mungo
places and the implications for radiocarbon dating. Archaeometry 33 (1), geomagnetic polarity excursion. Nature 265 (5589), 40e41.
43e50. Johnston, H., Clark, P., 1998. Willandra lakes archaeological investigations 1968e98.
Bøtter-Jensen, L., 1997. Luminescence techniques: instrumentation and methods. Archaeol. Ocean. 33 (3), 105e119.
Radiat. Meas. 27 (5e6), 749e768. Kalish, J.M., Miller, G.H., Tuniz, C., Pritchard, J.C., Rosewater, A., Lawson, E., 1997.
Bøtter-Jensen, L., Bulur, E., Duller, G.A.T., Murray, A.S., 2000. Advances in lumines- Otoliths as recorders of palaeoenvironments: comparison of radiocarbon age
cence instrument systems. Radiat. Meas. 32 (5e6), 523e528. and isoleucine epimerization in Pleistocene golden perch Macquaria ambigua
Bowler, J.M., 1971. Pleistocene salinities and climatic change: evidence from Lakes otoliths from Willandra Lakes. In: Conference Handbook, Sixth Australasian
and Lunettes in southeastern Australia. In: Mulvaney, D.J., Golson, J. (Eds.), Archaeometry Conference, Australian Museum, Sydney, 10-13 February 1997.
Aboriginal Man and Environment. ANU Press, Canberra, pp. 47e65. Kemp, J., Pietsch, T., Gontz, A., Olley, J., 2017. Lacustrine-fluvial interactions in
Bowler, J.M., 1976. Aridity in Australia: age, origins and expression in aeolian Australia’s riverine plains. Quat. Sci. Rev. 166, 352e362.
landforms and sediments. Earth Sci. Rev. 12 (2e3), 279e310. Kemp, J., Rhodes, E.J., 2010. Episodic fluvial activity of inland rivers in southeastern
Bowler, J.M., 1998. Willandra Lakes revisited: environmental framework for human Australia: palaeochannel systems and terraces of the Lachlan River. Quat. Sci.
occupation. Archaeol. Ocean. 33 (3), 120e155. Rev. 29 (5e6), 732e752.
Bowler, J.M., Gillespie, R., Johnston, H., Boljkovac, K., 2012. Wind v water: glacial Kemp, J., Spooner, N.A., 2007. Evidence for regionally wet conditions before the
maximum records from the Willandra lakes. In: Haberle, S., David, B. (Eds.), LGM in southeast Australia: OSL ages from a large palaeochannel in the Lachlan
Peopled Landscapes: Archaeological and Biogeographical Approaches to Land- Valley. J. Quat. Sci. 22 (5), 423e427.
scape. Terra Australis ANU E Press, Canberra, pp. 271e296. Last, F.M., Last, W.M., 2015. An Overview of Microbialite Genesis and Composition in
Bowler, J.M., Johnston, H., Olley, J.M., Prescott, J.R., Roberts, R.G., Shawcross, W., Saline Lakes of Southern Australia, Sixth International Limnogeology Congress
Spooner, N.A., 2003. New ages for human occupation and climatic change at June 15e19, 2015dAbstract Volume. U.S. Geological Survey Open-File Report
Lake Mungo, Australia. Nature 421 (6925), 837e840. 2015-1092, Reno, Nevada, p. 244.
Bowler, J.M., Jones, R., Allen, H., Thorne, A.G., 1970. Pleistocene human remains from Lomax, J., Hilgers, A., Twidale, C.R., Bourne, J.A., Radtke, U., 2007. Treatment of broad
Australia: a living site and human cremation from Lake Mungo, western New palaeodose distributions in OSL dating of dune sands from the western Murray
South Wales. World Archaeol. 2 (1), 39e60. Basin, South Australia. Quat. Geochronol. 2 (1e4), 51e56.
Bowler, J.M., Kotsonis, A., Lawrence, C.R., 2006. Environmental evolution of the Long, K., Stern, N., Williams, I.S., Kinsley, L., Wood, R., Sporcic, K., Smith, T., Fallon, S.,
mallee region, western Murray basin. Proc. R. Soc. Victoria 118 (2), 161e210. Kokkonen, H., Moffat, I., Grün, R., 2014. Fish otolith geochemistry, environ-
Bowler, J.M., Magee, J.W., 2000. Redating Australia’s oldest human remains: a mental conditions and human occupation at Lake Mungo, Australia. Quat. Sci.
sceptic’s view. J. Hum. Evol. 38 (5), 719e726. Rev. 88, 82e95.
Bowler, J.M., Price, D.M., 1998. Luminescence dates and stratigraphic analyses at Mejdahl, V., 1979. Thermoluminescence dating: beta-dose attenuation in quartz
Lake Mungo: review and new perspectives. Archaeol. Ocean. 33 (3), 156e168. grains. Archaeometry 21 (1), 61e72.
Bowler, J.M., Thorne, A.G., Polach, H.A., 1972. Pleistocene man in Australia: age and Miller, G., Magee, J., Smith, M., Spooner, N., Baynes, A., Lehman, S., Fogel, M.,
significance of the Mungo skeleton. Nature 240 (5375), 48e50. Johnston, H., Williams, D., Clark, P., Florian, C., Holst, R., DeVogel, S., 2016.
Bronk Ramsey, C., 2009. Bayesian analysis of radiocarbon dates. Radiocarbon 51 (1), Human predation contributed to the extinction of the Australian megafaunal
337e360. bird Genyornis newtoni ~47 ka. Nat. Commun. 7, 10496.
Bureau of Meteorology, 2020. Climate Data Online. [Link] Miller, G.H., Magee, J.W., Jull, A.J.T., 1997. Low latitude glacial cooling in the Southern
climate/data/. Hemisphere from amino acids in emu eggshells. Nature 385, 241e244.
Burrough, S.L., Thomas, D.S.G., Shaw, P.A., Bailey, R.M., 2007. Multiphase quaternary Murray, A.S., Wintle, A.G., 2000. Luminescence dating of quartz using an improved
highstands at Lake ngami, kalahari, northern Botswana. Palaeogeogr. Palae- single-aliquot regenerative-dose protocol. Radiat. Meas. 32 (1), 57e73.
oclimatol. Palaeoecol. 253 (3), 280e299. Murray, A.S., Wintle, A.G., 2003. The single aliquot regenerative dose protocol:
Clarkson, C., Jacobs, Z., Marwick, B., Fullagar, R., Wallis, L., Smith, M., Roberts, R.G., potential for improvements in reliability. Radiat. Meas. 37 (4e5), 377e381.
Hayes, E., Lowe, K., Carah, X., Florin, S.A., McNeil, J., Cox, D., Arnold, L.J., Hua, Q., O’Connell, J.F., Allen, J., Williams, M.A.J., Williams, A.N., Turney, C.S.M., Spooner, N.A.,
Huntley, J., Brand, H.E.A., Manne, T., Fairbairn, A., Shulmeister, J., Lyle, L., Kamminga, J., Brown, G., Cooper, A., 2018. When did <em>Homo sapiens</em>
Salinas, M., Page, M., Connell, K., Park, G., Norman, K., Murphy, T., Pardoe, C., first reach Southeast Asia and Sahul? Proc. Natl. Acad. Sci. Unit. States Am. 115
2017. Human occupation of northern Australia by 65,000 years ago. Nature 547, (34), 8482e8490.
306. Olley, J.M., Roberts, R.G., Yoshida, H., Bowler, J.M., 2006. Single-grain optical dating
Fitzsimmons, K.E., Cohen, T.J., Hesse, P.P., Jansen, J., Nanson, G.C., May, J.-H., of grave-infill associated with human burials at Lake Mungo, Australia. Quat.
Barrows, T.T., Haberlah, D., Hilgers, A., Kelly, T., Larsen, J., Lomax, J., Treble, P., Sci. Rev. 25 (19e20), 2469e2474.
2013. Late Quaternary palaeoenvironmental change in the Australian drylands. Oyston, B., 1996. Thermoluminescence age determinations for the Mungo III human
Quat. Sci. Rev. 74, 78e96. burial, Lake Mungo, southeastern Australia. Quat. Sci. Rev. 15 (7), 739e749.
Fitzsimmons, K.E., Rhodes, E.J., Barrows, T.T., 2010. OSL dating of southeast Pietsch, T.J., Olley, J.M., Nanson, G.C., 2008. Fluvial transport as a natural lumines-
Australian quartz: a preliminary assessment of luminescence characteristics cence sensitiser of quartz. Quat. Geochronol. 3 (4), 365e376.
and behaviour. Quat. Geochronol. 5 (2e3), 91e95. Polach, H.A., Lovering, J.F., Bowler, J.M., 1970. Anu radiocarbon date list IV. Radio-
Fitzsimmons, K.E., Rhodes, E.J., Magee, J.W., Barrows, T.T., 2007. The timing of linear carbon 12 (1), 1e18.
dune activity in the Strzelecki and Tirari Deserts, Australia. Quat. Sci. Rev. 26 Prescott, J.R., Hutton, J.T., 1994. Cosmic ray contributions to dose rates for lumi-
(19e21), 2598e2616. nescence and ESR dating: large depths and long-term time variations. Radiat.
Fitzsimmons, K.E., Stern, N., Murray-Wallace, C.V., 2014. Depositional history and Meas. 23 (2), 497e500.
archaeology of the central Lake Mungo lunette, Willandra Lakes, southeast Readhead, M.L., 1988. Thermoluminescence dating study of quartz in aeolian sed-
Australia. J. Archaeol. Sci. 41, 349e364. iments from southeastern Australia. Quat. Sci. Rev. 7 (3e4), 257e264.
Fitzsimmons, K.E., Stern, N., Murray-Wallace, C.V., Truscott, W., Pop, C., 2015. The Reeves, J.M., Barrows, T.T., Cohen, T.J., Kiem, A.S., Bostock, H.C., Fitzsimmons, K.E.,
Mungo mega-lake event, semi-arid Australia: non-linear descent into the last Jansen, J.D., Kemp, J., Krause, C., Petherick, L., Phipps, S.J., 2013. Climate vari-
ice age, implications for human behaviour. PloS One 10 (6), e0127008. ability over the last 35,000 years recorded in marine and terrestrial archives in
Galbraith, R.F., Roberts, R.G., Laslett, G.M., Yoshida, H., Olley, J.M., 1999. Optical the Australian region: an OZ-INTIMATE compilation. Quat. Sci. Rev. 74, 21e34.
dating of single and multiple grains of quartz from Jinmium rock shelter, Reimer, P.J., Bard, E., Bayliss, A., Beck, J.W., Blackwell, P.G., Ramsey, C.B., Buck, C.E.,
northern Australia. Part 1, Experimental design and statistical models. Cheng, H., Edwards, R.L., Friedrich, M., Grootes, P.M., Guilderson, T.P.,
Archaeometry 41 (2), 339e364. Haflidason, H., Hajdas, I., Hatte , C., Heaton, T.J., Hoffmann, D.L., Hogg, A.G.,
Galloway, R.W., 1965. Late quaternary climates in Australia. J. Geol. 73 (4), 603e618. Hughen, K.A., Kaiser, K.F., Kromer, B., Manning, S.W., Niu, M., Reimer, R.W.,
Gillespie, R., 1997. Burnt and unburnt carbon: dating charcoal and burnt bone from Richards, D.A., Scott, E.M., Southon, J.R., Staff, R.A., Turney, C.S.M., van der
the Willandra lakes, Australia. Radiocarbon 39 (3), 239e250. Plicht, J., 2013. IntCal13 and Marine13 radiocarbon age calibration curves
Gillespie, R., 1998. Alternative timescales: a critical review of Willandra. Lakes 0e50,000 Years cal BP. Radiocarbon 55 (4), 1869e1887.
Dating Archaeol. Ocean. 33 (3), 169e182. Reinfelds, I., Swanson, E., Cohen, T., Larsen, J., Nolan, A., 2014. Hydrospatial
Gillespie, R., Fink, D., Petchey, F., Jacobsen, G., 2009. Murray-Darling basin fresh- assessment of streamflow yields and effects of climate change: Snowy Moun-
water shells: riverine reservoir effect. Archaeol. Ocean. 44 (2), 107e111. tains, Australia. J. Hydrol. 512, 206e220.
Hamm, G., Mitchell, P., Arnold, L.J., Prideaux, G.J., Questiaux, D., Spooner, N.A., Stern, N., 2015. The Archaeology of the Willandra its empirical structure and
22 T.T. Barrows et al. / Quaternary Science Reviews 238 (2020) 106338

narrative potential. In: McGrath, A., Jebb, M.A. (Eds.), Long History, Deep Time. Taylor, L., Curnoe, D., 1999. Australia’s oldest human remains: age of the Lake
Deepening Histories of Place. ANU Press, pp. 221e240. Mungo 3 skeleton. J. Hum. Evol. 36 (6), 591e612.
Stern, N., Tumney, J., Fitzsimmons, K.E., Kajewski, P., 2013. Strategies for Investi- Tumney, J., 2018. Identifying and characterising different types of stone artefact
gating Human Responses to Changes in Landscape and Climate at Lake Mungo accumulations on the surface of the Lake Mungo lunette, southwest New South
in the Willandra Lakes, Southeast Australia, Archaeology in Environment and Wales, Australia. J. Archaeol. Sci.: Report 21, 380e388.
Technology: Intersections and Transformations, pp. 31e50. Williams, M., Cook, E., van der Kaars, S., Barrows, T., Shulmeister, J., Kershaw, P.,
Stirling, E.C., Zeitz, A.H.C., 1900. Fossil remains of lake callabonna. 1. In: Genyornis 2009. Glacial and deglacial climatic patterns in Australia and surrounding re-
Newtoni. A New Genus and Species of Fossil Struthious Bird, vol. 1. Royal Society gions from 35[punctuation space]000 to 10[punctuation space]000 years ago
of South Australia, Memoirs, pp. 41e80. reconstructed from terrestrial and near-shore proxy data. Quat. Sci. Rev. 28
Stuiver, M., Reimer, P.J., 1993. Extended 14C data base and revised CALIB 3.0 14C age (23e24), 2398e2419.
calibration program. Radiocarbon 35 (1), 215e230. Williams, M.A.J., 2019. Termites and stone lines - traps for the unwary archaeologist.
Telfer, M.W., Hesse, P.P., Perez-Fernandez, M., Bailey, R.M., Bajkan, S., Lancaster, N., Quat. Sci. Rev. 226, 106028.
2017. Morphodynamics, boundary conditions and pattern evolution within a Yamasaki, F., Hamada, C., Hamada, T., 1977. Riken natural radiocarbon measure-
vegetated linear dunefield. Geomorphology 290, 85e100. ments IX. Radiocarbon 19 (1), 62e95.
Thorne, A., Grün, R., Mortimer, G., Spooner, N.A., Simpson, J.J., McCulloch, M.,