Holocene and recent aeolian reactivation of the Willandra Lakes lunettes,semi-arid southeastern Australia

Abstract

The Willandra Lakes in semi-arid southeastern Australia provide some of the most continuous combined palaeoenvironmental and archaeological records on the continent. These are best preserved within the transverse shoreline (lunette) dunes on their downwind margins. Following final lake retreat c. 15 ka avulsion of the dominant fluvial inflow eastwards, the Willandra lunettes periodically reactivated, experiencing erosion, aeolian redeposition and alluvial sheetwash. These reworked sedimentary archives reflect regional climatic conditions rather than those of the entire catchment. Yet the focus of most study in the region to date has remained on the late Pleistocene. The general paucity of Holocene data has contributed to a perception that people largely abandoned the area in favour of the perennial Murray and Darling Rivers to the south and west. Our study reconstructs past geomorphological conditions and patterns of human mobility in adjacent Lakes Mungo and Durthong over the last c. 15 ka subsequent to final lake retreat, including the most recent 150 years since Europeans established pastoralism in the region. Our data show that Indigenous people did not abandon the area as previously assumed, but developed effective strategies for responding to the changed environmental conditions. Final lake retreat transitioned into a phase of aeolian accumulation c. 15–12 ka, indicating locally dry conditions. Subsequent aeolian reactivation peaked during arid phases experiencing less rainfall in the early Holocene and twice in the most recent 1000 years prior to European settlement in the area. Alluvial sheetwash was deposited onto lake floors during the mid-Holocene, and again in the early decades of European settlement. Aeolian reactivation, likely driven by European pastoral activities, increases in the most recent 150 years. Our study underscores the necessity of integrating geomorphological and archaeological investigations over landscape scales in order to optimise our understanding of interactions between people and their environment through time.

Keywords

archaeology Australia Holocene Lake Mungo luminescence dating lunette

Introduction

The Willandra Lakes system in semi-arid southeastern Australia is well known for providing one of the most continuous, and longest, combined palaeoenvironmental and archaeological records on the continent (Bowler, 1998; Bowler et al., 2003). The lakes are a relict overflow system, fed by a distributary of the Lachlan River, which rises in the temperate subhumid southeastern Australian highlands. The lakes have been dry since c. 15 ka (Bowler et al., 2012; Fitzsimmons et al., 2014) following avulsion of the Willandra Creek drainage to form a new channel to the east. During the time that the Willandra Creek was active, the lakes experienced sediment deposition onto their downwind margins, forming transverse shoreline, or lunette, dunes. These so-called lunettes are composite aeolian shore-marginal landforms recording lake-fill conditions in sandy facies, and drying lake conditions through an increased input of clay (Bowler, 1973, 1983; Hill, 1940). Lunettes preserve the most conspicuous and abundant archaeological traces in the region (Stern et al., 2013) – including the oldest known ritual burials in Australia (Bowler et al., 1970, 1972, 2003; Bowler and Magee, 2000; Bowler and Thorne, 1976) – but also the best-preserved archives for past hydrological change (Bowler et al., 2012; Fitzsimmons et al., 2014, 2015). The lunette stratigraphy of the Willandra Lakes more closely reflects the environmental conditions of the headwaters – some 800 km to the east – rather than climate in the local area, which appears to have minimal input to Lake hydrology.

Following final lake retreat at c. 15 ka, the lunettes of the Willandra periodically reactivated, experiencing erosion, aeolian redeposition and alluvial sheetwash onto the lake floors. These more recent sedimentary archives consequently reflect local environmental conditions, in contrast to the preceding deposits (Fitzsimmons et al., 2014). It is assumed that erosion and aeolian reactivation are driven by increased surface availability of sandy sediment, most likely due to reduced vegetation cover associated with prolonged arid conditions. By contrast, episodes of alluvial sheetwash may be driven either by extreme rainfall events or generally wetter climates, although which of the two is unclear. The presence of incipient soils indicates reduced geomorphic activity under more stable conditions, sufficient to increase enrichment in organic matter close to the contemporary surface. Despite the opportunity provided by these reactivated lunette sediments to investigate local Holocene geomorphological and climatic history, the focus of scientific study to date in the Willandra has largely remained on the late Pleistocene (Bowler, 1998; Bowler et al., 2012; Fitzsimmons, 2017; Fitzsimmons et al., 2015; Long et al., 2014, 2018).

The nature of human occupation in the Willandra following final lake retreat has likewise received limited attention compared with the late Pleistocene (Barbetti and Allen, 1972; Barbetti and McElhinny, 1972; Bell, 1991; Bowler and Price, 1998; Bowler et al., 1970, 2003; Long et al., 2014, 2018; Olley et al., 2006; Shawcross, 1998; Shawcross and Kaye, 1980; Stern, 2015; Stern et al., 2013; Webb et al., 2006; Westaway et al., 2013, 2017). Traces of human activity post-15 ka have been observed on the Lake Mungo lunette, and elsewhere in the Willandra system, including hearths, chipped stone artefacts, faunal remains and grinding stones (Allen, 1972; Anderson et al., 1984; Dare-Edwards, 1979; Fitzsimmons et al., 2014; Fullagar et al., 2015a, 2015b). However, the general paucity of information about this later period has contributed to a perception that people largely abandoned the area in favour of the perennial Murray and Darling Rivers (Allen, 1972, 1974; Pardoe, 2015).

This study seeks to address the gap in knowledge about palaeoenvironmental conditions and settlement and land use in the Willandra Lakes area over the 15 ka subsequent to final lake retreat. Here, we use the term ‘palaeoenvironmental’ to describe past changes in the landscape in response to hydrological, climatic and anthropogenic influences. We examine the post-lake depositional history at the south-central Lake Mungo and adjacent Lake Durthong lunettes with respect to the nature of sediment reactivation, and provide a chronological framework based on optically stimulated luminescence (OSL) dating. These are integrated with data documenting the abundance and types of past activity traces preserved in the Mungo lunette over the past 15 ka, including more detailed information about the stone technology generated from clusters of chipped stone artefacts. Together, these lay a foundation for investigating the relationships between changes in land use, occupancy and mobility, and the geomorphic changes that took place after the inflow to the lake ceased.

Regional setting

The Willandra Lakes system lies in the southwest of Australia’s second largest catchment, the Murray-Darling Basin (MDB), located in the southeast of the continent. The lakes are a series of five larger, and several smaller, dry lakes (Figure 1) and are surrounded by linear and parabolic dune systems (Bowler and Magee, 1978). The lakes were filled at various times in the past by the Willandra Creek, a presently inactive channel of the Lachlan River which has its headwaters approximately 600 km to the east in the southeastern Australian highlands. The palaeohydrology of the Willandra lakes is recorded in their transverse lunette dunes, which lie on the downwind, eastward lake margins (Bowler, 1983). Until the creek ceased to flow c. 14–15 ka (Bowler, 1998; Fitzsimmons et al., 2014) because of drainage avulsion to a different channel (Kemp et al., 2017), lake hydrology and consequently, lunette sedimentation more closely reflected runoff from the distal uplands rather than local conditions.

Figure 1.

(a) Location of the Willandra Lakes system in the semi-arid zone of southeastern Australia, (b) the Willandra Lakes system and distribution of lunette dunes, with the Mungo and Durthong lunettes highlighted in orange, and (c) location of sampling sites along the Durthong and Mungo lunettes, colour-coded according to study.

In this study, we focus on the reactivation of the Lake Mungo and Durthong lunettes over the last c. 15 ka, subsequent to cessation of flow in the Willandra Creek. The larger lake, Mungo, is an overflow lake hydrologically linked to Lake Leaghur immediately to the north. The compound sand-clay lunette along its eastern margin (Figure 2) is c. 30 km in length. Lake Durthong is a through-flow lake located approximately 10 km west and downstream of Mungo. Its shore-marginal lunette extends c. 8 km from north to south (Fitzsimmons, 2017).

Figure 2.

Schematic overview of the stratigraphy and chronology of the main lunette units at Lake Mungo (after Fitzsimmons et al., 2014). Units A through E correspond to sediments deposited while the lake was active. Units F, G, H and I represent units deposited subsequent to final lake retreat.

Sample sites

Reactivation of the sedimentary system in the Willandra post-lacustrine system takes three main forms:

As aeolian erosion and deposition on the lunettes and immediately downwind thereof;

As sheetwash scour and alluvial fan deposition onto the lake floor;

As pedogenesis within these various phases.

These three phases are recorded over four-dimensional space, since erosion, deposition and preservation, are spatially variable across the lunettes (Fitzsimmons, 2017; Fitzsimmons et al., 2014; Stern et al., 2013). In this study, we sampled sites yielding representative information about post-lacustrine reactivation of the sedimentary system in the region and integrated these data with those from other studies of the central Mungo lunette (Barrows et al., 2014; Fitzsimmons et al., 2014).

At Lake Mungo, multiple phases of aeolian redeposition or alluvial fan deposition overlie one another in some places, separated by weakly developed humic paleosol horizons. Such sites give the clearest indication of palaeoenvironmental variability through time and therefore formed the focus of our sampling at this lake. Chronostratigraphic work focussed on Locality 973659, a small erosion basin in the central part of the lunette that lies just downwind of the present-day dune crest (Figure 1c, shown in green). Three small, vegetated hummocks representing preserved parts of the original aeolian landform (referred to here as residuals, see photos in Figure 3: Residuals 1–3), and a larger Residual 4 on the rim of the blowout, record the post-lake depositional history. The new geochronological data presented here comes from these four residuals (Figure 3a). We also revisited the sampling sites discussed by Barrows et al. (2014) in order to place their respective chronostratigraphies into the post-lake framework.

Figure 3.

Photographs of Locality 973659 in the central Lake Mungo lunette: (a) Looking north across the erosion basin showing the four residuals sampled, including the location and age of the OSL sample collected from Residual 3. (b–d) Photographs of Residuals 4, 1 and 2, respectively (displayed as observed from left to right in photo a), showing sampling locations and ages.

Locality 973659 is of particular interest for reconstructing human-environmental interactions, because the artefacts lying on the surface of the erosion hollow were part of a broader investigation of the stone technology employed in this area during the post-lake era (Spry, 2014). The surface of the erosional hollow surrounding Residuals 1–3 contains several groups of refitting artefacts demonstrating sequences of stone tool production, as well as clusters of artefacts that were arguably struck from the same block of raw material (Figure S7, available online). Several sets of refitting artefacts also lie on the surface of upper part of Unit E (Zanci), which is well exposed in the area surrounding Residual 4 (Spry, 2014).

At Lake Durthong, aeolian sediment overlying lake-phase lunette stratigraphy is not as extensive or as thickly deposited as at Lake Mungo. Consequently, investigation of post-lacustrine stratigraphy was limited to single OSL samples from four exposures (Table 1): two in the southern part of the lunette (DTH26, 27), one in the central part of the lunette adjacent the inflow channel (DTH07) and one in the central-northern lunette (DTH31) (Figure 1). Two additional samples were collected from linear dunes on the north-northwestern margins of the lake (DTH14, 15), and are interpreted to correspond to activation of those dunes.

Table 1.

OSL sample codes, locations and stratigraphic units.

Laboratory code	Location along lunette	Field code	Description
Lake Mungo, Locality 697626 (central lunette)
EVA1100	Residual 1	697626-1	Immediately overlying Arumpo/Unit E sediments
EVA1101	Residual 1	697626-2	Middle of brown unit overlying Arumpo/Unit E sediments
EVA1103	Residual 2	697626-4	Upper part of uppermost reworked unit
EVA1104	Residual 2	697626-5	Lower part of uppermost reworked unit
EVA1105	Residual 3	697626-6	Lower laminated reworked sands (no soils identified)
EVA1106	Residual 2	697626-7	Base of residual, lowermost of three units exposed. Likely age of adjacent blowout surface which has refitting sets of artefacts and clusters of artefacts struck from the same core
EVA1107	Residual 4	697626-8	Lower part of residual, laminated sands, equivalent to surface with refitting sets of artefacts and clusters of artefacts struck from the same core
EVA1108	Residual 4	697626-9	Middle part of residual and pale massive sands
Lake Durthong
EVA1069	DTH27: southeastern lunette	DTH27-1	Reworked brown aeolian sand
EVA1072	DTH26: southeastern lunette	DTH26-1	Reworked fine aeolian sand
EVA1092	DTH07: central lunette and proximal to inflow channel	DTH07-1	Unconsolidated red-brown aeolian sand with carbonate lag; possible incipient soil development
EVA1094	DTH31: central lunette and north of inflow channel	DTH31-2	Unconsolidated red-brown aeolian sand
EVA1097	DTH14: North lunette and western extremities of lunette sediments	DTH14-3	Silty red sand with weak soil development; possible alluvial or dust component
EVA1099	DTH15: North lunette and western extremities of lunette sediments	DTH15-2	Unconsolidated silty and dark red aeolian sand

OSL: optically stimulated luminescence.

In addition, we incorporate recently available data from previous studies at Lake Mungo (Figure 1; Barrows et al., 2014; Fitzsimmons et al., 2014). This includes two dates from a chronostratigraphic transect in the central part of the lunette, several hundred metres south of Locality 973659 that sampled superimposed outwash fans on the lake floor, and one that sampled reactivated aeolian sediment near the lunette crest (Fitzsimmons et al., 2014). Barrows et al. (2014) sampled residuals in three erosional basins on the lunette (Figure 1): Locality 969660 (2 samples), Locality 940691 (4 samples; Tumney, 2011) and Locality 966617 (2 samples; Tumney, 2011). Most of these were collected from unconsolidated, reactivated aeolian sediment overlying the final lake-phase of the lunette sequence, but one was collected from a reactivated aeolian sample underlying a weakly developed brown paleosol at Locality 969660, and two were collected from outwash fans, one each from 966617 and 940691 (Figure 5). Two samples were also collected from a low elevation dune bounding a soak on the now-dry lake floor c. 2 km from 940691 (Figure 5), and two from a linear dune on the western shoreline of Lake Mungo (Barrows et al., 2014).

Methods

Field methods

Tight integration of geological and archaeological data is needed to build an understanding of the strategies people developed in response to the landscape changes that followed the drying of the Willandra Lakes. We undertook this by systematically documenting the distribution and three-dimensional stratigraphy of post-lacustrine sedimentary units and the archaeological features they contain. The study area is a 3 km × 600 m swath of the central Mungo lunette that incorporates Localities 973659 and 969660, as well as the chronostratigraphic section (Figure 4 and Figure S6, available online). It extends from the lake floor immediately adjacent the lunette, including low slope-angle fans of redeposited lunette material, as well as the lee slope of the lunette, much of which is overlain by reactivated aeolian sediment (Fitzsimmons et al., 2014; Stern, 2015; Stern et al., 2013).

Figure 4.

Geological map of the central Mungo lunette showing exposures of Units E (Arumpo/Zanci), F (post-lake aeolian) and G (post-lake alluvium) draped over the project’s digital air photographs, and the distribution of hearths and clusters of stone tools in each unit.

Over three field seasons, the three-dimensional locations of stratigraphic boundaries, and archaeological features, were plotted onto georectified aerial photographs for geographic information systems (GIS)-based analysis of the number and type of activity traces preserved in the post-lake strata. The content and context of all traces of past human activity which were at least partially embedded in the sediment were documented during systematic foot survey (for further information, see Supplementary material, available online). In addition, discrete clusters of surface artefacts were included when their stratigraphic origin could be inferred, as well as the presence of refitting artefacts, and artefacts struck from the same block of raw material (Foley et al., 2017). In addition, information was recorded about the types of hearths encountered, including any associated food remains or tools (Stern, 2015). This was followed by more detailed studies of particular activity traces, with an initial focus on chipped stone artefacts (Spry, 2014), grindstones (Fullagar et al., 2015a) and shell tools (Weston et al., 2017).

The Lake Durthong lunette is substantially smaller and more subdued than that of Mungo, with fewer stratigraphic exposures. Work undertaken at this site was solely geological in nature (e.g. Fitzsimmons, 2017). The length of the Durthong lunette was traversed to locate suitable sedimentary exposures for stratigraphy and OSL dating. Seven sites were sampled, of which six (Figure 1) exposed sediment representing aeolian reactivation or linear dune activity post-dating the final retreat of the lake. Although hearths were observed on the lunette and lake floor, as yet no archaeological survey has been undertaken of this basin.

The OSL sampling strategy was designed to provide bracketing age estimates for all identified strata representing lunette reactivation. OSL sampling involved driving 4 cm diameter, 10-cm-long stainless steel tubes horizontally into cleaned and exposed surfaces. The sample holes were then widened and deepened to fit a 3-inch diameter portable sodium iodide gamma spectrometer for dosimetric analyses. These measurements were made in each hole for 30 min. The sediment removed was collected in a sealed plastic bag for additional laboratory measurements of moisture content, beta and gamma spectrometry.

Luminescence dating

We applied single grain quartz OSL dating to all 14 samples collected from the post-lacustrine lunette sediments at Lakes Mungo and Durthong. Single aliquot OSL measurements were additionally undertaken on small aliquots (<100 grains) of sand-sized quartz on all 8 of the Mungo samples, and 4 of the 6 Durthong samples (because of limited datable sample).

The OSL samples were opened and processed under dim red light conditions in the luminescence dating laboratory of the Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany. Quartz was extracted from the raw dateable sample using published methods (Fitzsimmons et al., 2014; Supplementary material, online available). For each sample, 24 small (1 mm; c. 100 grains) aliquots and 6 single grain discs were prepared for equivalent dose (D_e) measurement. Six single grain discs (600 grains) were measured for each sample.

D_e measurements on both single aliquots and single grains were undertaken using an automated Risø TL-DA-20 reader equipped with a single grain laser attachment using the single aliquot regenerative dose (SAR) protocol of Murray and Wintle (2000, 2003); Preheat plateau tests were undertaken on small aliquots of representative sample EVA1100 using preheat temperatures between 180°C and 280°C. Since the preheat plateau extended to 260ºC (Figure S1, available online) despite the young ages of the samples, preheat and cutheat temperatures of 260°C and 220°C, respectively, were used in the measurement protocols. For the majority of samples, both aliquots and single grains yielded broadly normal distributions, Figure 2 so justifying application of the Central Age Model (CAM; Galbraith et al., 1999). Generally, the very youngest samples gave indication of incomplete bleaching of the OSL signal – a trait which may broadly characterise modern, unconsolidated sediments deposited rapidly and over short distances as is likely the case here based on strong resemblance between the modern sands and the immediately underlying units. For these samples (Table 2), where appropriate, the Minimum Age Model (MAM; Galbraith et al., 1999) was used to calculate D_e.

Table 2.

Equivalent dose (D_e), dose rate data (attenuated) and OSL age estimates for the Lake Mungo and Durthong lunette samples. Single aliquot D_es and age estimates are shown in plain text; single grain results are in italics.

Sample code	D_e (Gy)	K (%)	U (ppm)	Th (ppm)	Cosmic dose rate (Gy/ka)	Total dose rate (Gy/ka)	Age (ka)
Lake Mungo
EVA1100	10.6 ± 0.5^a 5.7 ± 0.2^a	0.42 ± 0.02	0.53 ± 0.03	2.65 ± 0.13	0.17 ± 0.02	0.87 ± 0.04	12.2 ± 0.7 6.5 ± 0.4
EVA1101	3.8 ± 0.2^a 3.1 ± 0.1^a	0.45 ± 0.02	0.62 ± 0.03	2.61 ± 0.13	0.19 ± 0.03	0.93 ± 0.04	4.0 ± 0.3 3.3 ± 0.2
EVA1103	0.05 ± 0.00^b 0.04 ± 0.01^b	0.24 ± 0.01	0.44 ± 0.02	1.55 ± 0.08	0.13 ± 0.01	0.56 ± 0.02	0.09 ± 0.01 0.06 ± 0.02
EVA1104	0.06 ± 0.01^b 0.04 ± 0.00^a	0.19 ± 0.01	0.43 ± 0.02	1.38 ± 0.07	0.13 ± 0.01	0.49 ± 0.02	0.12 ± 0.01 0.07 ± 0.01^c
EVA1105	0.04 ± 0.00^b 0.03 ± 0.02^b	0.52 ± 0.03	0.81 ± 0.04	2.99 ± 0.15	0.12 ± 0.01	0.99 ± 0.04	0.04 ± 0.00 0.03 ± 0.02
EVA1106	8.3 ± 0.3^a 9.4 ± 0.4^a	0.55 ± 0.03	0.78 ± 0.04	3.33 ± 0.17	0.12 ± 0.01	1.03 ± 0.04	8.1 ± 0.5 9.1 ± 0.5
EVA1107	13.1 ± 0.4^a 14.4 ± 0.4^a	0.47 ± 0.02	0.85 ± 0.04	3.18 ± 0.16	0.15 ± 0.01	1.00 ± 0.04	13.1 ± 0.7 14.4 ± 0.7
EVA1108	19.2 ± 0.9^a 19.0 ± 0.6^a	0.67 ± 0.03	0.79 ± 0.04	3.78 ± 0.19	0.17 ± 0.02	1.23 ± 0.05	15.6 ± 1.0 15.5 ± 0.8
Lake Durthong
EVA1069	0.84 ± 0.16 ^b	0.69 ± 0.04	2.61 ± 0.13	0.95 ± 0.05	0.20 ± 0.03	1.37 ± 0.08	0.61 ± 0.12
EVA1072	0.15 ± 0.01 ^a	1.30 ± 0.07	8.20 ± 0.41	1.94 ± 0.10	0.20 ± 0.03	2.42 ± 0.13	0.06 ± 0.01 ^c
EVA1092	2.10 ± 0.6^a 0.84 ± 0.13^a	0.49 ± 0.03	4.20 ± 0.21	0.69 ± 0.04	0.20 ± 0.04	1.12 ± 0.06	1.87 ± 0.55 0.75 ± 0.13
EVA1094	1.00 ± 0.03^a 0.97 ± 0.04^a	0.43 ± 0.02	3.20 ± 0.16	0.74 ± 0.04	0.20 ± 0.02	0.99 ± 0.05	1.01 ± 0.06 0.98 ± 0.07^c
EVA1097	13.1 ± 7.5^a 1.80 ± 0.10^b	0.43 ± 0.02	4.31 ± 0.22	1.01 ± 0.05	0.19 ± 0.02	1.14 ± 0.06	11.5 ± 6.6 1.58 ± 0.12
EVA1099	16.5 ± 1.9^a 3.80 ± 0.21^b	0.49 ± 0.03	3.86 ± 0.19	0.98 ± 0.05	0.19 ± 0.02	1.14 ± 0.06	14.4 ± 1.8 3.33 ± 0.25

OSL: optically stimulated luminescence.

Calculated using the central age model of Galbraith et al. (1999).

Calculated using the minimum age model of Galbraith et al. (1999).

Samples yielding thermal transfer exceeding 5%.

Sample dose rates were determined by in situ gamma spectrometry and laboratory beta counting, with dose rate attenuation calculated using published factors (Guérin et al., 2011). In-situ moisture content was calculated by weighing the raw and oven-dried weight of material both from the ends of the tubes and bulk sediment; the average was taken as the final value (Table S2, available online). The cosmic ray component of the dose rates was following Prescott and Hutton (1994). Although assuming constant depth may reduce the reliability of the dose rate calculations and therefore ages in these sediments with very low dose rates, we decided against modelling variability in cosmic ray dose rates through time (e.g. Burrough et al., 2007). We argue that the assumption of gradual accumulation of overburden is unrealistic in this dynamic environment.

Analysis of previously published chronological data

In addition to incorporating the luminescence ages for sediment records from previous studies (Barrows et al., 2014; Fitzsimmons et al., 2014), we generated a dataset of published radiocarbon dates based on archaeological traces in the vicinity of the Lakes Mungo and Durthong for the most recent 15 ka (Table S4, Supplementary information, available online). The majority of published legacy data precede radiocarbon measurements using accelerator mass spectrometry (AMS) and modern pretreatment methods (Allen, 1972; Allen, 1998; Clark, 1987; Johnston and Clark, 1998), and must therefore be treated with caution. Dates from material deemed unsuitable for dating using earlier methods based on the current state of the art (including burnt earth and soil carbonates) were removed from the dataset as unreliable; those derived from skeletal remains (bone collagen and amino acids) are not included since permission has not been granted to use these. Radiocarbon dates were calibrated based on SHCal13 (Hogg et al., 2013); the statistically most likely dates are listed in Table S4, available online. Given the limitation to reliable radiocarbon dating for very young samples (Fitzsimmons et al., 2007; Walker, 2005), the youngest date in the dataset exceeds 500 cal. yr BP. Despite the limitations of these legacy data, this metadataset provides an additional chronological overview regarding human presence on the landscape against the backdrop of the palaeoenvironmental information generated in this study.

Results

Stratigraphy

The stratigraphic contexts and locations of the samples collected for this study are listed in Table 1. The residuals studied include the full range of sediment types observed for material deposited following lake retreat, including profiles that preserve the stratigraphic contact between the sediments representing the final oscillating lake (Unit E or Zanci unit) and overlying material (Figure 1).

Residual 1 is approximately 1.5 m high and 3 m wide and capped by an eastward-sloping bench representing the surface of a brownish soil, which to the north is overlain by mobile, aeolian sands (Figure 3). The stratigraphy of Residual 1 is exposed to the south towards the blowout surface, and comprises a brownish paleosol with minor ped development, which overlies the final lacustrine phase (Fitzsimmons et al., 2014; known as Unit E or the Zanci Unit) without visible unconformity. Two OSL samples were collected from residual 1 (Table 1, Figure 3c). One sample (EVA1100) was collected at the base of the paleosol and residual, immediately above the laminated clay-sand sediments that are characteristic of the final lake-phase of sediment accumulation on the lunette (Unit E or Zanci; Figure 1). The other (EVA1101) was collected from the middle of the soil profile, within massive brown sands containing some roots; visible bioturbation was avoided during sampling.

Residual 2 is the largest of the three residuals within the blowout (Figure 3a). It rises c. 3 m above the floor of the erosional hollow and contains a bench of brownish sandy soil slightly lower in altitude than that occurring on top of Residual 1. Pale, unconsolidated aeolian sands overlie the brown sandy soil. Three OSL samples were collected from Residual 2 (Table 1, Figure 3d): one from the base of the residual (EVA1106), assumed to represent the age of the blowout floor and the artefacts scattered across its surface, and two from upper and lower parts of the reworked unit, respectively (EVA1103 and EVA1104).

Residual 3 lies on the edge of the hollow, above mobile aeolian sands to the north of Residual 2 (Figure 3a). The pale sands exposed in this residual are laminated, especially towards the base; such laminations are common in recently deposited dune sediments (Bagnold, 1941). The sediments of Residual 3 comprise well sorted fine to medium pale sand, and most likely correspond to the uppermost pale sands of residual 2. One sample (EVA1105; Table 1, Figure 3a) was collected from the lower laminated exposures of Residual 3.

Residual 4 lies on the present-day lunette crest, at the southern end of an extensive exposure of sediments representing the final phase of lacustrine deposition, at the top of Unit E (Figure 3a and b). The northern flank of Residual 4 provides the best exposure of the stratigraphy which comprises massive yellowish sands with carbonate concretions. Towards the base of the residual, these sands are laminated, and grade upwards into more massive sands with no noticeable unconformity. Two OSL samples were collected from Residual 4 at Locality 697626: one (EVA1107) from the lower, laminated sands, stratigraphically equivalent to the blowout surface that contains the refitted artefacts; and the other (EVA1108) from pale, massive sands in the middle part of the residual (Figure 3b).

The crest of the dune surrounding the sampled residuals at Lake Mungo comprises an expansive surface of pale yellow, laminated sand. This closely resembles, and appears to grade upwards from, the lacustrine Unit E (Zanci) but is distinct from it in having no clay enrichment. We refer to this transitional stratigraphic layer as Unit F (Figure 1). This laminated sand is exposed in Residual 4 and the base of Residual 1, can be traced towards the lake floor, and appears to conformably overlie Unit E (Zanci). At the base of Residual 1, and also within a residual at 969660 (Barrows et al., 2014), this unit is overprinted by a weakly developed, brownish and humic paleosol. We propose that this laminated sand represents a gradational change from lake drying conditions to deflation of the completely dried lake bed. This interpretation contrasts with previous work proposing an abrupt end to the final lacustrine phase followed by superposition of a weakly developed soil (Bowler, 1998), but may be explained by variability in stratigraphic preservation along the length of the lunette (Fitzsimmons, 2017) and the geographical focus of earlier studies.

At Residual 1, and at Locality 969660, Unit F is capped by massive, unconsolidated, brownish sands that lack the laminations characteristic of the lower part of the unit. The relative intensity of the brownish sediment suggests increased humic content. This brown unit outcrops at the base of Residual 2 and Residual G of a previous study (Fitzsimmons et al., 2014), although both of these appear to have been truncated. It can also be traced near to residuals A and B (Fitzsimmons et al., 2014), where it contains traces of burnt sediment hearths. Despite its colour, it otherwise lacks pedogenic structure and may reflect coeval vegetation cover and relatively high rates of aeolian accumulation.

Much of the crest and lee of the Lake Mungo lunette surface comprises pale, unconsolidated aeolian sediment. At times, this uppermost aeolian sediment preserves laminations indicative of relatively recent deposition. This unit, H, overlies the brown humic unit at Residual 2, comprises the entirety of Residual 3, as well as previously sampled residuals (Barrows et al., 2014) at 966617, 940691 and BLW4. We assume Unit H to represent the most recent period(s) of aeolian reactivation.

Towards the lake floor, just to the north of the Walls of China tourist area at Lake Mungo, several generations of alluvial fan or sheetwash activity superpose one another. The lowermost layer observed in outcrop is a brownish, poorly sorted, relatively clay-rich sand, identified as Unit G in Fitzsimmons et al. (2014) and containing in situ hearths and stone artefacts. In this earlier study, Unit G was interpreted as a single unit representing more humid environmental conditions, although it may have deposited over multiple episodes. Unit G is unconformably overlain by the pale clay-rich sands of Unit I, which contains little organic material and was inferred to represent recent sheetwash accumulation during intense rainfall events (Fitzsimmons et al., 2014).

At Lake Durthong, sediments clearly post-dating lacustrine phases are comparatively sparse, thin and variable in character compared with those at Lake Mungo. In the southern part of the lunette, reworked, pale aeolian sediments, less than one metre thick, unconformably overlie laminated pelletal clay layers interpreted to represent the Durthong equivalent to Unit E/Zanci (Fitzsimmons, 2017). In the central part of the lunette, sediments occupying the same post-lacustrine stratigraphic context are reddish in colour. At the more northerly of the two sites sampled, the red reactivated aeolian sand is unconformably overlain by a pale, unconsolidated sediment, a few tens of centimetres thick. At the northern end of the lunette, reddish unconsolidated sand contains a substantial silt component, and may represent part of an incipient linear dune overlying the abandoned lunette. The stratigraphic equivalence of the Durthong reactivated sites to one another, and to the sites at Mungo, is unclear based on sedimentological characteristics alone.

OSL dating

The OSL characteristics of the quartz samples from both Lakes Mungo and Durthong, even the very youngest, are generally well suited to dating with OSL using the SAR protocol. Proportions of aliquots or grains passing criteria for age analysis are generally high (Table S1, available online). OSL signals of datable aliquots or grains are bright relative to background, even in the case of very young samples (Figure S4, available online); generate simple exponential dose-response curves; and with only a few exceptions, yield recycling ratios within 10–15% of unity (Table S1, available online). Thermal transfer in single grains, known to be common within the SAR protocol for very young sediments (Li and Li, 2006; Pietsch, 2009), varied between 0–8% for those samples measured (Table S1, available online). Thermal transfer generally decreased with increasing equivalent dose, although two samples from Lake Durthong of similar age (EVA1092 and EVA1094) yielded substantially different thermal transfer values of 0.1% and 7.7%, respectively. Three samples for which the thermal transfer exceeded 5% are listed in Table 2 require interpretation with caution. Dose recovery of single grains of the Lake Durthong sample EVA1069 lies at unity (1.00 ± 0.02; Figure S5, available online). Dose recovery of samples from Lake Mungo, when given an applied dose comparable with the equivalent dose, also yield ratios close to unity (Doerschner et al., 2016). The preheat plateau results of sample EVA1100 (Figure S1, available online) are comparable with tests run on other samples from Lake Mungo (Fitzsimmons et al., 2014, 2015) from which the reactivated sediments in this study are presumed to have been derived. Single grain and single aliquot dose distributions from Lake Mungo (but not Durthong), and consequently equivalent dose values, generally lie within error of one another (Figure S2, available online). The comparability of OSL characteristics between the younger samples from this study and older samples from earlier work reinforces the argument for inheritance of luminescence characteristics over depositional generations (Fitzsimmons, 2011).

Some variability was observed in the luminescence characteristics between sites. Generally, luminescence characteristics of the Lake Mungo samples are better than for those from Lake Durthong (Table S1, available online). In part this may be due to the fact that the Durthong sample set comprises a larger number of younger samples. Substantial discrepancies between the single aliquot and single grain equivalent doses were observed in the Durthong samples. Aliquot De values yielded higher overdispersion values and were calculated using the CAM, resulting in much older ages than for the younger single grain D_es, which were generally calculated using the MAM on the basis of likely incomplete bleaching of the very young signal. The prevalence of older D_e values in the calculated aliquots versus single grains of given samples (Table S2, available online) is attributed to the homogenisation of older, incompletely bleached signals within aliquots. Single grain ages are preferred to the single aliquot values since incomplete bleaching is more reliably identified in single grains than in aliquots.

A statistically robust number of single grains from even the youngest samples yielded sufficient signal for dating (Figure S4, available online). The very young ages provide a reliable lower limit to OSL dating for the Willandra Lakes area of c. 30 years (EVA1105: 0.03 ± 0.02 ka). In part this is assisted by the generally very low dose rates, which in the case of the youngest samples are well below 1 Gy/ka (Table 2). Data relating to OSL age calculations for this study are listed in Table 2. Subsequent discussion of the chronology will be based on the single grain ages, given in italics.

Chronology for palaeoenvironmental change

The OSL chronology generated by this study (Table 2) was integrated with published ages (Barrows et al., 2014; Fitzsimmons et al., 2014), divided into age groups, and summarised in Table 3.

Table 3.

Synthesis of post-lacustrine OSL ages and proposed scheme for aeolian reactivation. Ages in italics (marked with asterisk) represent linear dune activity surrounding the lake basins.

Depositional phase	Aeolian deposition		Alluvial fan/sheetwash deposition
	Ages (ka)	Mean (ka)	Ages (ka)	Mean (ka)
Transitional post-lake	15.6 ± 1.0^a 13.1 ± 0.7^a 12.7 ± 1.0^b 12.2 ± 0.7^a 14.2 ± 1.3^b,*	13.5 ± 1.5
Phase I aeolian	8.3 ± 0.6^c 8.1 ± 0.5^a	8.2 ± 0.6
Phase A fans(mid-Holocene aeolian)	4.0 ± 0.3^a		5.2 ± 0.3^c 3.4 ± 0.3^c	4.3 ± 1.3
Local linear dune activity	3.33 ± 0.3^a,* 2.62 ± 0.3^b,* 1.58 ± 0.12^a,*
Phase II aeolian	0.98 ± 0.07^a 0.75 ± 0.13^a 0.61 ± 0.12^a	0.75 ± 0.19
Phase III aeolian	0.26 ± 0.04^b 0.25 ± 0.02^b 0.18 ± 0.03^b	0.23 ± 0.04
Phase B sheetwash			0.14 ± 0.01^b 0.10 ± 0.01^b	0.12 ± 0.01
Phase IV aeolian	0.12 ± 0.01^b 0.12 ± 0.01^a 0.10 ± 0.02^b 0.10 ± 0.01^b 0.09 ± 0.02^a 0.07 ± 0.01^a 0.07 ± 0.00^b 0.06 ± 0.01^a 0.04 ± 0.00^a	0.10 ± 0.01

OSL: optically stimulated luminescence.

This paper.

Barrows et al. (2014).

Fitzsimmons et al. (2014).

The chronology of post-lake reactivation is presented visually in Figure 5. On this map, sediment type and antiquity are differentiated by colour. Alluvial fans and sheetwash are shown in green, aeolian reactivation of the lunettes is indicated by yellow-brown and linear dune activity surrounding the lakes is shown in red.

Figure 5.

Spatial distribution of sediment ages from the Lake Durthong and Mungo lunettes, colour-coded according to depositional phase and type. Depositional phases are summarised in the legend, together with the study from which they derive.

Figure 5 provides a spatial and chronological overview of sedimentary reactivation of the post- lake lunettes and provides a first organisation of the synthesised chronology which is discussed in more detail in the ‘Discussion’ section.

Archaeological survey and artefact analysis

Archaeological traces are present in the post-lake sedimentary units on the Lake Mungo lunette, attesting to continued land-use and occupancy after the overflow system dried out. However, the activity traces that accumulated in the post-lake aeolian sands (Unit F) and alluvial fans (Unit G) are much less abundant than predicted on the basis of their areas of exposure (Table S5, available online). This is consistent with the observation that they make up < 9% of the activity traces preserved in the central Mungo lunette study area (Table S6, available online). After the lake dried out, there would have been a significant change in the distribution and abundance of surface water and staple plant foods in this region and it is reasonable to assume that the accumulation of fewer activity traces on the lunette during the post-lake period reflects a change in patterns of mobility and land use in response to this.

Clusters of chipped stone artefacts are by far the most abundant trace of past human activity in the post-lake lunette sediments (63%), although hearths (~17%) and isolated finds (~20%) are also part of the record (Table S7, available online). Most hearths consist of heat retainers, occasionally made from carbonate nodules but mostly from material collected from termite mounds. Two consist of a lens of disseminated charcoal, representing the final remnants of hearths eroded away in antiquity. Food remains and/or tools are rarely found in association with these hearths. Isolated finds include one shell tool, two grindstones and some manuports (unworked blocks of sandstone and quartzite and nodules of silcrete that were carried to the lunette). In view of the materials preserved, initial efforts to understand how people responded to past environmental changes have focused on information that can be gleaned from the chipped stone artefacts found in the post-lake aeolian sediments. Analysis of these assemblages interrogated whether the accumulation of fewer activity traces in the post-lake sediments was related in any way to increased mobility, given that water sources would have become smaller, more dispersed, and generally more ephemeral.

Only 1% of the clusters of stone artefacts in these sediments preserve in situ artefacts. The rest are surface scatters made up either of artefacts struck from the same block of raw material (19%) or refitting sets of artefacts (80%). When the individual sets of knapping debris are analysed as an aggregated assemblage, they elucidate the technological system of which they were once a part (Spry, 2014).

To investigate changes in the frequency and/or distance and/or duration of moves between residential campsites, and between campsites and activity loci, archaeologists try to identify the strategies people used to ensure that they had raw materials and/or tools, where and when they were needed (Kuhn, 1995). Inferences are usually based on the form in which raw materials of different origin and knapping quality were introduced to, and taken away from, locations on the landscape, as well as the types of stone working techniques and tools produced. We investigated the question of mobility by comparing the chipped stone artefacts found in the post-lake aeolian sediments (Unit F) with those found in the underlying Last Glacial Maximum (LGM) age sediments (Unit E/Zanci; Table S8, available online; Figure S7, available online). The only sources of raw material in the Willandra Lakes region suitable for tool-making originate from numerous outcrops of silcrete located within 5–80 km of the central Mungo lunette and 2 of quartzite located 30–80 km away; thin section, trace element and Pb-isotope analyses have been used to characterise the silcrete sources (Kurpiel, 2017; Kurpiel et al., 2019). During both time periods, most of the material for making tools was obtained from the silcrete outcrops, but after the lake dried out, greater use was made of the higher-quality quartzite, suggesting that these localised and more distant sources of material were visited more often. Following final lake retreat, a smaller proportion of the cores carried onto the lunette were made from cobbles or slabs and a greater proportion were made from flakes or tools. This suggests that once the lakes dried out, raw material was more often carried around the landscape in the form of smaller, lighter and more prepared cores.

The post-lake stone clusters contain more elongate flakes than those dating to the LGM, indicating more systematic and efficient reduction of cores. A greater proportion of the flakes with edge modification are elongate with a thick cross-section, suggesting that tools with greater length of cutting edge/unit mass and with robust edges were more likely to be made and carried around after the lake dried out. Although the composition of the stone clusters indicates that some tools were made, used and discarded on the lunette during the post-lake period, the presence of small, backed artefacts points to the fact that people were also replacing these small, highly portable tools during visits to the lunette (Spry, 2014).

Taken together, these features of the Unit F chipped stone assemblage suggest that once the lakes dried out, people made greater use of raw material from relatively distant sources and placed greater emphasis on provisioning individuals with tools and highly portable cores than provisioning the landscape with raw material. In contrast, the LGM assemblages from the same part of the lunette suggest that provisioning the landscape with raw material was the predominant strategy employed at that time (Kurpiel, 2017). suggests that people were more mobile after the lakes dried out, and that a change in stone technology was one of the ways in which they responded to the altered environmental conditions.

Discussion

Holocene palaeoenvironments on the southeastern Australian desert margins

Subsequent to final lake retreat around 15 ka, the Mungo and Durthong lunettes experienced multiple phases of enhanced aeolian reactivation, and two phases of water-lain erosion and sheetwash (Table 3, Figure 5). The most recent phase of aeolian reactivation, spanning the last c. 130 years and corresponding to the recent period following European settlement of and intensified pastoralism in the area, is discussed in Section ‘Recent impact of post-colonial land use on desert margin landscapes associated with the introduction of pastoral activities in the 1860’.

Figure 6 provides a timeline for palaeoenvironmental change in the Willandra over the post-lake period, based on a dichotomy between aeolian conditions interpreted to reflect relative aridity, and wetter conditions represented by alluvial fan and sheetwash deposition. Where ages cluster together, these are also shown in the form of a probability plot based on the normalised sum of probability distributions of individual ages. The probability plot is included solely for the purpose of visualising age groupings, since the dataset lacks sufficient ages to truly test these clusters statistically. Alluvial fan and sheetwash may have been deposited under one of two sets of conditions: either sustained wetter climates or catastrophic precipitation events.

Figure 6.

Probability distribution functions highlighting peaks in aeolian reactivation phases as well as individual ages for coeval linear dune activity, contrasting with periods of outwash fan activity onto the Lake Mungo lake floor. Probabilities are irrelevant and the probability plots are included solely as a visual aid to identifying age groupings; the ages and their uncertainties included in the plots are shown in black, and the mean ages, incorporating a 1σ uncertainty, for the age clusters are highlighted in light orange. Ages for local linear dune activity beyond the lunettes are shown in red; the single c. 4 ka lunette reactivation age, which does not fall within the age clusters, is shown in orange; each of these are highlighted in a paler orange than the age clusters since they represent individual ages. Ages for wetter conditions (green and blue) are illustrated on the left of the age bar and highlighted in blue. The aeolian stratigraphy illustrated corresponds to that observed in the lunettes. Climatic phases and possible relevant events are listed in black text above the graph: Holocene Climatic Optimum (HCO), ‘Medieval Warm Period’ (MWP), ‘Little Ice Age’ (LIA), arrival of Europeans and establishment of pastoral activity (1860s–), 1937–1947 drought, introduction of myxoma virus to rabbit populations resulting in reduced numbers (1950s).

The earliest period of post-lake lunette reactivation appears to be a transitional phase of laminated aeolian sediments grading without unconformity upwards from the Unit E (Zanci) oscillating lake event. This period provides an aggregate age of 13.5 ± 1.5 ka (number of ages, n = 4) and is coeval with linear dune accumulation on the western shoreline dating to 14.2 ± 1.3 ka (Barrows et al., 2014). The gradation from laminated Unit E pelletal clay-quartz sand lunette couplets into a post-lake aeolian depositional regime indicates a smoother transition from lake drying conditions to complete cessation of the lake hydrological system and reactivation of the lunette sediments than has previously been assumed (Bowler, 1998). The dominance of aeolian activity suggests persistently dry climatic conditions in the region which is supported by a coeval peak in nearby desert dune activity from 14.5–12 ka in the western MDB (Lomax et al., 2011), as well as inferred aridity from linear dune reactivation in the Strzelecki Desert to the northwest (Fitzsimmons et al., 2007). It is likely that the c. 13.5 ka aeolian event postdates a slightly wetter period as demonstrated by accelerated speleothem growth from 20–15 ka in the Flinders Ranges, several hundred kilometres to the west (Cohen et al., 2011), indicating increased effective precipitation associated with temporary northward penetration of the westerlies into inland southern Australia (Barrows and Juggins, 2005). Palynological records from the nearby Darling River Anabranch lakes suggest that vegetation during the 18–9 ka period was a heterogeneous mosaic of tall and low shrublands (Cupper, 2005), not inconsistent with semi-arid conditions conducive to aeolian activity.

Aeolian reactivation peaked again around 8.2 ± 0.6 ka (n = 2), following a break in the stratigraphic record which cannot clearly be associated with stability in the form of pedogenesis or erosion and unconformity. We interpret this phase to correspond to the subsequent phase of relative aridity in the region. Additional indications for local aridity at this time are given by similar ages from desert dunes in the western MDB (Lomax et al., 2011), and a peak in aeolian dust input to the southeastern Australian highlands sourced from the MDB (Marx et al., 2009).

Two mid-Holocene ages correspond to alluvial fan deposition onto the Mungo lake floor and indicate relatively wetter conditions in the Willandra region. The average of the two ages (5.2 ± 0.3 ka and 3.4 ± 0.3 ka) is 4.3 ± 1.3 ka, and may represent either a sustained phase of wetter climates or two phases of shorter duration; this cannot be distinguished since no clear unconformity can be identified between the two samples dated (Fitzsimmons et al., 2014). Irrespective of single or multiple events, our interpretation of mid-Holocene moister climates is consistent with evidence for more humid conditions across southeastern Australia at this time. These include increased lake recharge in a number of basins (Bray et al., 2012; Cohen et al.,2011, 2012; Fitzsimmons and Barrows, 2010; Gell et al., 2005; Gouramanis et al., 2010; Jones et al., 1998; Magee et al., 1995) and increased river discharge (Cohen and Nanson, 2007; Gingele et al., 2007).

Linear dune activity persisted in the area throughout the period c. 4–1 ka (also Lomax et al., 2011). However, without more detailed sedimentological analyses (e.g. Fitzsimmons et al., 2009), this activity cannot be inferred to represent any more than intermittent partial reactivation, as is common in a semi-arid environment (Hesse and Simpson, 2006). The most recent 4 ka oversaw a general decline in woodland vegetation in the Willandra/Darling River anabranch region (Cupper, 2005; Cupper et al., 2000), interpreted to represent the establishment of semi-arid conditions that have persisted to the present day (Bray et al., 2012). The transition to semi-arid climate at this time is likely to be associated with the intensification of El Niño-Southern Oscillation (ENSO) variability (Moy et al., 2002; Quigley et al., 2010; Shulmeister and Lees, 1995) and overall aridification of the southern half of the continent (Moros et al., 2009).

The most recent millennium records two apparently distinct phases of aeolian reactivation of the Mungo and Durthong lunettes, defined by weighted averages as 0.75 ± 0.19 ka (n = 3) and 0.23 ± 0.04 ka (n = 3). The earlier phase appears to immediately postdate a brief period of pluvial conditions, recorded by lake filling in the Frome-Callabonna basin to the northwest (Cohen et al., 2011, 2012); and flooding in the Barrier Ranges north of the Willandra (Jansen and Brierley, 2005). The wet conditions have been interpreted to correspond to a brief phase of intensified high-pressure systems over the Indian Ocean and Tasman Sea propagating northward penetration of a trough into inland Australia (Cohen et al., 2012). It is unclear what mechanisms may have driven the aeolian reactivation in the Willandra shortly thereafter, nor the arid peak c. 230 years ago.

While our study provides the most substantial record yet for past environmental change on the semi-arid southeastern Australian desert margins, distinct from the main lunette archives that represent more distal climatic influence, it is subject to several limitations. The limited sample number reduces confidence in identifying peak phases in aeolian or sheetwash accumulation, although the pattern is consistent with other archives across southeastern Australia. Luminescence dating of the aeolian sediments only provides the most recent age of activity and cannot inform us about the duration of depositional events. This study may incorporate some spatial and temporal bias, since sampling at Lake Mungo focuses on the central portion of the lunette, and therefore does not consider aeolian reactivation and spatially variable sediment supply along the length of this landform (Fitzsimmons, 2017). The limited degree of correlation between aeolian reactivation events between the Mungo and Durthong lunettes, the latter of which was sampled at multiple points along the lake margins, may underscore this limitation.

Implications for human adaptation to landscape change in desert margins

The archive of activity traces preserved by the post-lake reactivation of the Mungo lunette potentially fills a long-standing gap in our knowledge of early Holocene settlement on Australia’s desert margins (Hiscock, 2008; Smith, 2013). However, our understanding of post-lake settlement history is constrained by the accumulation of sediment during the post-lake period, and the ways in which archaeological traces have been studied. At present, integrated archaeological and geological data are available only for the central Mungo lunette, which represents a 2 km² sample from one 33 km-long landform, in a semi-arid landscape in which foraging ranges may have been extensive (perhaps up to 4500 km²; Marlowe, 2005). Together with legacy data from the 1970s which provide a less systematic record of 359 sites scattered across the Willandra, these data highlight the need to re-evaluate long-standing interpretations about the post-lake settlement history in the Willandra (e.g. Allen, 1974, 1990); Furthermore, they provide a basis for comparing continent-wide models of demographic change (e.g. Williams et al., 2013, 2015a, 2015b) with the documented settlement history of one, climatically sensitive locale.

Initially it was argued that people largely abandoned the area after the lakes dried out, shifting to the better-watered corridors bounding the Murray or Darling Rivers (Allen, 1974, 1990); This assertion was supported by an apparent absence of hearths at Lake Mungo during 19–5 ka (Barbetti, 1973). Ethno-historic information about the subsistence activities of the Barkindji/Paakantyi, who lived along the Darling River during the late 19th century, provided further rationale for arguing that the Willandra Lakes had become part of the riverine hinterland, visited only following good winter rain when seeds provided a predictable food staple (Allen 1974). At the time, the idea of seasonal transhumance as fundamental to hunter-gatherer adaptation was pervasive (Higgs et al., 1967); since then, however, studies worldwide have shown that most social groups ranged across interior ranges or coastal plains (e.g. Sealy, 2006), riverine corridors or the arid plains beyond (Pardoe, 2006).

Our systematic data identify three aspects of this initial interpretation of settlement history that need re-evaluating:

People did not abandon the Willandra Lakes region when the overflow system ceased to operate: traces of their activities are found not only in the post-lake sediments on the lunettes (Figure S6, available online), but also in other landforms within the overflow system (Clark, 1987).

Seeds were not incorporated into the diet to replace the loss of aquatic resources after the lakes dried out (Allen, 1974, 1990). Seed-processing was part of the subsistence repertoire during the LGM, although which seeds were being ground, the contribution they made to the overall diet, and how this may have changed over time, is not yet known (Fullagar et al., 2015a, 2015b). Furthermore, evidence for the exploitation of fish and shellfish is largely restricted to the LGM (Bowler et al., 2012) and when the overflow system was active, these were probably fall-back foods rather than a focus of the diet (Stern et al., 2013).

Although the activity traces found in the post-lake sediments on the Mungo lunette are much less abundant than those found in the underlying strata, at this stage there is no rationale for attributing this to a demographic change rather than a change in land use and occupancy. Analysis of the stone artefacts certainly indicates an increase in relative mobility, although it is not known whether this was associated with shifts in diet and foraging activities, social networks or a pattern of intermittent occupancy. At present, the Holocene activity traces cannot be parsed into clusters (Figure 7) connected with episodes of relatively wetter or drier conditions because the incipient soil horizons which would otherwise provide a means of achieving this, are not laterally extensive. Only direct dating of a large sample of hearths preserved in different landforms would permit a finer-scale analysis of human responses to fluctuating local conditions over shorter time spans (e.g. Fanning et al., 2008; Holdaway et al., 2005).

Figure 7.

Holocene palaeoenvironmental history at Lake Mungo prior to European settlement (from Figure 6), compared with calibrated radiocarbon dates for archaeological traces (fireplaces, middens and isolated finds) derived from legacy data. The climatic phases HCO, MWP and LIA are shown in black text on the right.

Models of demographic history, based on analysis of the geographic and temporal distribution of archaeological sites across the continent, suggest an overall decline in population size during the LGM, as well as abandonment of areas without reliable surface water (Williams et al., 2013, 2015a, 2015b). The Willandra Lakes overflow system, which received large volumes of spring melt-water from the southeast highlands at this time, is identified as a LGM refugium (Williams et al., 2013). It is argued that during the early Holocene, populations increased and people re-colonised areas that had been uninhabitable previously (Williams et al., 2013). Some subsequent changes coincide with environmental shifts and some do not (Williams et al., 2015a). At present, there is insufficient detail in the Willandra’s Holocene record to attempt a comparison with the Holocene time-series data (Williams et al., 2015a).

The settlement history of the Willandra Lakes region is based on substantially different data: the number of discrete activity traces, like cooking hearths or patches of tool-making debris, contained within distinct sedimentary envelopes whose bracketing ages have been established. As such, it presents an internally consistent dataset for comparing the density of activity traces in sediments representing different time intervals and correspondingly different climatic and hydrologic conditions. Site density encapsulates variables including the number and the sizes of social groups visiting the area, as well as the frequency and duration of their visits. To establish whether there were changes in all or some of these variables, and whether those changes were influenced by demographic or behavioural changes or both, requires additional information about diet and foraging activities, technologies and social networks.

The stone artefact assemblages studied up to now indicate that people were relatively more mobile following final lake retreat. However, mobility also encapsulates a number of variables, including the frequency, distance and duration of moves between campsites, and between camps and activity loci. The relative influence of these variables cannot be disentangled without more detailed information about diet and foraging strategies, and the use made of materials from different geographic locations. It is premature to conclude that a higher density of activity traces and evidence for reduced mobility in the LGM indicates that Lake Mungo was a refugium, or that a lower density of sites together with evidence for greater mobility in the post-LGM record reflects the movement of people out of the Willandra Lakes region once surface water was more dispersed and less abundant.

Recent impact of post-colonial land use on desert margin landscapes associated with the introduction of pastoral activities in the 1860s

The first Europeans to arrive in the Willandra belonged to the exploration party of Charles Sturt (1833, 1984), who described a barren country of sandy plains following expeditions in 1833 and 1844–1845. Based on the chronology presented in our study, these decades would most likely correspond to the tail end of the dry phase III, thus providing confidence in our age estimates and interpretation of the post-lacustrine sediment records. Despite the pessimistic accounts, European settlers brought cattle and sheep to the region from the 1860s (Withers, 1989). Two ages for alluvial fans on the Mungo lake floor averaging 0.12 ± 0.02 (at 95% confidence), indicating wetter conditions than experienced by Sturt’s company c. 1870–1910 CE, might provide some basis for their optimism. In addition, documentary sources indicate that southeastern Australia experienced a phase of wetter conditions from 1841–1860 (Gergis and Ashcroft, 2013), while station records in the vicinity of the Willandra Lakes (available from the 1870s onwards) indicate wetter conditions from the 1870s to mid-1890s (Bureau of Meteorology). Wetter conditions would very likely have encouraged the European settlers to bring cattle and sheep to the area, initially on a seasonal, and subsequently on a permanent basis.

The intense drought in the years around Australia’s federation (c. 1901 CE) brought about a return to arid conditions in the Willandra area. During this period, however, aeolian reactivation accompanying the dry years (0.10 ± 0.01 ka; n = 8) was exacerbated by erosion through grazing of stock and the introduction of rabbits, as well as the removal of stabilising vegetation (Cupper, 2005; Cupper et al., 2000). European land use in the MDB over the last 150 years has affected enormous changes across the landscape, including increased salinity and sedimentation in lagoons associated with the MDB rivers (Gell et al., 2005, 2009)

Reorganisation of the sedimentary system in response to the intense impact of European land use on this vulnerable desert marginal environment continues to the present day in the form of accelerating erosion of the lunette sediments and their redeposition by wind on the crest and lee slopes, as well as catastrophic sheetwash onto the lake floors following even moderate rainfall events. This legacy has substantial implications for land management in the Willandra Lakes area, even subsequent to cessation of pastoral activities and the establishment of a National Park.

Conclusion

The reactivation of lunette sediments in the Willandra Lakes region following final lake retreat provides a valuable insight into the response of desert marginal systems to changing climatic conditions over the Holocene. Our chronology generally corresponds to broader evidence for Holocene environmental change across southeastern Australia. The archaeological traces preserved in the post-lake sediments demonstrate the continued use and occupancy of Lake Mungo area after the overflow system dried out. Although these activity traces are less abundant than those found in the sedimentary envelopes dating to the LGM, it is premature to attribute this to demographic rather than behavioural change, particularly as the artefact analysis indicates that people were relatively more mobile at this time. This study rectifies long-standing mis-perceptions about the history of human settlement in the Willandra Lakes region and lays an important foundation for future research aimed at generating more detailed, landscape-scale information about changes in economy, technology and social networks over the past 15 ka.

Supplemental Material

Fitzsimmons_et_al_Holocene_Supplementaryrevision29012019 – Supplemental material for Holocene and recent aeolian reactivation of the Willandra Lakes lunettes, semi-arid southeastern Australia

Supplemental material, Fitzsimmons_et_al_Holocene_Supplementaryrevision29012019 for Holocene and recent aeolian reactivation of the Willandra Lakes lunettes, semi-arid southeastern Australia by Kathryn E Fitzsimmons, Caroline Spry and Nicola Stern in The Holocene

Footnotes

Acknowledgements

This research was undertaken with permission from the Elders’ Council and the Technical and Scientific Advisory Committee of the Willandra Lakes Region World Heritage Area. We thank the Paakantyi/Barkindji, Ngiyampaa and Mutthi Mutthi Elders for welcoming us into their Country and for their support of this work. We are especially indebted to Daryl Pappin, the Mungo Archaeology Project’s Cultural Heritage Officer, for his dedicated assistance during fieldwork. Rudy Frank, Paul Penzo-Kajewski and students from the Department of Archaeology and History at La Trobe University provided invaluable assistance in the field. We are grateful to Ian Wakefield for permission to work on Top Hut station and to the staff of Mungo National Park for facilitating fieldwork on the Lake Mungo lunette. We thank Paul Penzo-Kajewski from the Department of Archaeology and History at La Trobe University for his work on the MapInfo files, Steffi Albert, from the Max Planck Institute for Evolutionary Anthropology in Leipzig, for preparing the samples in the OSL laboratory and Elizabeth Foley and Katherine Thomas for discussions of an earlier draft of this paper.

Funding

This work was funded by grants from the Australian Research Council (DP1092966 and LP0775058) and by a Higher Degree Research Grant and Post-doctoral Write-up Grant awarded to Caroline Spry by the Faculty of Humanities and Social Sciences at La Trobe University.

Supplemental material

Supplemental material for this article is available online.

ORCID iDs

Kathryn E Fitzsimmons

Nicola Stern

References

Allen

(1972) Where the crow flies backwards: Man and land in the Darling Basin. PhD Thesis, Australian National University.

Allen

(1974) The Bagundji of the Darling Basin: Cereal gatherers in an uncertain environment. World Archaeology 5: 309–322.

Allen

(1990) Environmental history in southwestern New South Wales during the Late Pleistocene. In: Gamble

Soffer

(eds) The World at 18000 BP: Low Latitudes, Vol. 2. London: Unwin Hyman, pp. 296–321.

Allen

(1998) Reinterpreting the 1969–1972 Willandra Lakes archaeological surveys. Archaeology in Oceania 33: 207–220.

Anderson

Capel

Galloway

et al . (1984) Aboriginal occupation of the Lake Mungo region during mid–late Holocene Times. Quaternary Australasia 2: 16–23.

Bagnold

(1941) The Physics of Blown Sand and Desert Dunes. London: Methuen and Co.

Barbetti

(1973) Archaeomagnetic and radiocarbon studies of aboriginal fireplaces. PhD Thesis, Australian National University.

Barbetti

Allen

(1972) Prehistoric man at Lake Mungo, Australia, by 32,000 years BP. Nature 240: 46–48.

Barbetti

McElhinny

(1972) Evidence of a geomagnetic excursion 30,000 yr BP. Nature 239: 327–330.

10.

Barrows

Juggins

(2005) Sea-surface temperatures around the Australian margin and Indian Ocean during the Last Glacial Maximum. Quaternary Science Reviews 24: 1017–1047.

11.

Barrows

Fitzsimmons

Stern

et al . (2014) Environmental change at Lake Mungo, Willandra Lakes World Heritage Area, semi-arid Australia. In: Australasian quaternary association biennial meeting, Mildura, VIC, Australiam, 29 June–4 July.

12.

Bell

(1991) Thermoluminescence dates for the Lake Mungo Aboriginal fireplaces and the implications for radiocarbon dating. Archaeometry 33: 43–50.

13.

Bowler

(1973) Clay dunes: Their occurrence, formation and environmental significance. Earth Science Reviews 9: 315–338.

14.

Bowler

(1983) Lunettes as indices of hydrologic change: A review of the Australian evidence. Proceedings of the Royal Society of Victoria 95: 147–168.

15.

Bowler

(1998) Willandra Lakes revisited: Environmental framework for human occupation. Archaeology in Oceania 33: 120–155.

16.

Bowler

Magee

(1978) Geomorphology of the Mallee region in semi-arid northern Victoria and western New South Wales. Proceedings of the Royal Society of Victoria 90: 5–25.

17.

Bowler

Magee

(2000) Redating Australia’s oldest human remains: A sceptic’s view. Journal of Human Evolution 38: 719–726.

18.

Bowler

Price

(1998) Luminescence dates and stratigraphic analyses at Lake Mungo: Review and new perspectives. Archaeology in Oceania 33: 156–168.

19.

Bowler

Thorne

(1976) Human remains from Lake Mungo: Discovery and excavation of Lake Mungo III. In: Kirk

Thorne

(eds) The Origin of the Australians. Canberra, ACT, Australia: Australian Institute of Aboriginal studies, pp. 127–138.

20.

Bowler

Gillespie

Johnston

et al . (2012) Wind v water: Glacial maximum records from the Willandra Lakes. In: Haberle

David

(eds) Peopled Landscapes: Archaeological and Biogeographic Approaches to Landscapes. Canberra, ACT, Australia: The Australian National University, pp. 271–296.

21.

Bowler

Johnston

Olley

et al . (2003) New ages for human occupation and climatic change at Lake Mungo, Australia. Nature 421: 837–840.

22.

Bowler

Jones

Allen

et al . (1970) Pleistocene human remains from Australia: A living site and human cremation from Lake Mungo, western New South Wales. World Archaeology 2: 39–60.

23.

Bowler

Thorne

Polach

(1972) Pleistocene man in Australia: Age and significance of the Mungo skeleton. Nature 240: 48–50.

24.

Bray

Jones

Fallon

et al . (2012) Radiocarbon analysis of halophilic microbial lipids from an Australian salt lake. Quaternary Research 77: 104–109.

25.

Burrough

Thomas

DSG

Shaw

et al . (2007) Multiphase Quaternary highstands at Lake Ngami, Kalahari, northern Botswana. Palaeogeography, Palaeoclimatology, Palaeoecology 253(3–4): 280–299.

26.

Clark

(1987) Willandra Lakes World Heritage Area Archaeological Resource Study. Report for the New South Wales, Department of Planning and the Western Lands Commission of NSW, Sydney, NSW, Australia.

27.

Cohen

Nanson

(2007) Mind the gap: An absence of valley-fill deposits identifying the Holocene hypsithermal period of enhanced flow regime in southeastern Australia. The Holocene 17: 411–418.

28.

Cohen

Nanson

Jansen

et al . (2011) Continental aridification and the vanishing of Australia’s megalakes. Geology 39: 167–170.

29.

Cohen

Nanson

Jansen

et al . (2012) Late Quaternary mega-lakes fed by the northern and southern river systems of central Australia: Varying moisture sources and increased continental aridity. Palaeogeography, Palaeoclimatology, Palaeoecology: 356–357: 89–108.

30.

Cupper

(2005) Last glacial to Holocene evolution of semi-arid rangelands in southeastern Australia. The Holocene 15: 541–553.

31.

Cupper

Drinnan

Thomas

(2000) Holocene palaeoenvironments of salt lakes in the Darling Anabranch region, south-western New South Wales, Australia. Journal of Biogeography 27: 1079–1094.

32.

Dare-Edwards

(1979) Late quaternary soils on Clay Dunes of the Willandra Lakes, New South Wales. PhD Thesis, Australian National University.

33.

Doerschner

Hernandez

Fitzsimmons

(2016) Sources of variability in single grain dose recovery experiments: Insights from Moroccan and Australian samples. Ancient TL 34: 14–25.

34.

Fanning

Holdaway

Rhodes

(2008) A new geoarchaeology of Aboriginal artefact deposits in western NSW, Australia: Establishing spatial and temporal geomorphic controls on the surface of the archaeological record. Geomorphology 101: 524–532.

35.

Fitzsimmons

(2011) An assessment of the luminescence sensitivity of Australian quartz with respect to sediment history. Geochronometria 38: 199–208.

36.

Fitzsimmons

(2017) Reconstructing palaeoenvironments on desert margins: New perspectives from Eurasian loess and Australian dry lake shorelines. Quaternary Science Reviews 171: 1–19.

37.

Fitzsimmons

Barrows

(2010) Holocene hydrologic variability in temperate southeastern Australia: An example from Lake George, New South Wales. The Holocene 20: 585–597.

38.

Fitzsimmons

Gell

Bickford

et al . (2007) The OZPACS database: A resource for understanding recent impacts on Australian ecosystems. Quaternary Australasia 24: 2–6.

39.

Fitzsimmons

Magee

Amos

(2009) Characterisation of aeolian sediments from the Strzelecki and Tirari Deserts, Australia: implications for reconstructing palaeoenvironmental conditions. Sedimentary Geology 218: 61–73.

40.

Fitzsimmons

Stern

Murray-Wallace

(2014) Depositional history and archaeology of the central Lake Mungo lunette, Willandra Lakes, southeast Australia. Journal of Archaeological Science 41: 349–364.

41.

Fitzsimmons

Stern

Murray-Wallace

et al . (2015) The Mungo mega-lake event, semi-arid Australia: Non-linear descent into the last ice age, implications for human behavior. PLoS ONE 10: e0127008.

42.

Foley

Spry

Stern

(2017) Establishing the integrity and stratigraphic origin of stone artefact scatters on the surface of the Lake Mungo lunette in south-eastern Australia. Journal of Archaeological Science: Reports 13: 547–557.

43.

Fullagar

Hayes

Stephenson

et al . (2015a) Evidence for Pleistocene seed grinding at Lake Mungo, south-eastern Australia. Archaeology in Oceania 50(S1): 3–19.

44.

Fullagar

Hayes

Stephenson

et al . (2015b) The scale of seed grinding at Lake Mungo. Archaeology in Oceania 50: 177–179.

45.

Galbraith

Roberts

Laslett

et al . (1999) Optical dating of single and multiple grains of quartz from Jinmium rock shelter, northern Australia. Part 1, Experimental design and statistical models. Archaeometry 41: 339–364.

46.

Gell

Bulpin

Wallbrink

et al . (2005) Tareena Billabong: A palaeolimnological history of an ever-changing wetland, Chowilla Floodplain, lower Murray-Darling Basin. Marine & Freshwater Research 56: 441–456.

47.

Gell

Fluin

Tibby

et al . (2009) Anthropogenic acceleration of sediment accretion in lowland floodplain wetlands, Murray-Darling Basin, Australia. Geomorphology 108: 122–126.

48.

Gergis

Ashcroft

(2013) Rainfall variations in south-eastern Australia part 2: A comparison of documentary, early instrumental and palaeoclimate records, 1788–2008. International Journal of Climatology 33: 2973–2987.

49.

Gingele

De Deckker

Norman

(2007) Late Pleistocene and Holocene climate of SE Australia reconstructed from dust and river loads deposited offshore the River Murray mouth. Earth and Planetary Science Letters 255: 257–272.

50.

Gouramanis

Wilkins

De Deckker

(2010) 6,000 years of environmental changes recorded in Blue Lake, South Australia, based on ostracod ecology and valve chemistry. Palaeogeography, Palaeoclimatology, Palaeoecology 297: 223–237.

51.

Guérin

Mercier

Adamiec

(2011) Dose-rate conversion factors: Update. Ancient TL 29: 5–8.

52.

Hesse

Simpson

(2006) Variable vegetation cover and episodic sand movement on longitudinal desert sand dunes. Geomorphology 81: 276–291.

53.

Higgs

Vita-Finzi

Harris

et al . (1967) The climate, environment, and industries of Stone Age Greece. Part 3. Proceedings of the Prehistoric Society 33: 1–29.

54.

Hill

(1940) The lunette: A new land form of aeolian origin. Australian Geographer 3: 5–21.

55.

Hiscock

(2008) Archaeology of Ancient Australia. Oxford: Routledge.

56.

Hogg

Hua

Blackwell

et al . (2013) SHCal13 Southern Hemisphere calibration, 0–50,000 years cal BP. Radiocarbon 55(4): 1889–1903.

57.

Holdaway

Shiner

Fanning

(2005) Absence of evidence or evidence of absence? Understanding chronology of indigenous occupation of western New South Wales. Archaeology in Oceania 40: 33–49.

58.

Jansen

Brierley

(2005) Pool-fills: A window to palaeoflood history and response in bedrock-confined rivers. Sedimentology 51: 901–925.

59.

Johnston

Clark

(1998) Willandra Lakes archaeological investigations 1968–98. Archaeology in Oceania 33: 105–119.

60.

Jones

Bowler

McMahon

(1998) A high resolution Holocene record of P/E ratio from closed lakes in Western Victoria. Palaeoclimates 3: 51–82.

61.

Kemp

Pietsch

Gontz

et al . (2017) Lacustrine-fluvial interactions in Australia’s Riverine Plains. Quaternary Science Reviews 166: 352–362.

62.

Kuhn

(1995) Mousterian Lithic Technology: An Ecological Perspective. Princeton, NJ: Princeton University Press.

63.

Kurpiel

(2017) Stone sources in the Willandra Lakes Region, in South-Eastern Australia: A study of source characterisation and raw material procurement. PhD Thesis, La Trobe University.

64.

Kurpiel

Pickering

Maas

et al . (2019) Lead (Pb) isotope signatures for silcrete sources from the Willandra Lakes region, Australia: A pilot study of a new method for provenancing silcrete artefacts. Journal of Archaeological Science: Reports 23: 62–71.

65.

(2006) Correcting for thermal transfer in OSL measurements of young sediment samples. Radiation Measurements 41: 855–861.

66.

Lomax

Hilgers

Radtke

(2011) Palaeoenvironmental change recorded in the palaeodunefields of the western Murray Basin, South Australia: New data from single grain OSL dating. Quaternary Science Reviews 30: 723–736.

67.

Long

Stern

Williams

et al . (2014) Fish otolith geochemistry, environmental conditions and human occupation at Lake Mungo, Australia. Quaternary Science Reviews 88: 82–95.

68.

Long

Wood

Williams

et al . (2018) Fish otolith microchemistry: Snapshots of lake conditions during early human occupation of Lake Mungo, Australia. Quaternary International 463: 29–43.

69.

Magee

Bowler

Miller

et al . (1995) Stratigraphy, sedimentology, chronology and paleohydrology of Quaternary lacustrine deposits at Madigan Gulf, Lake Eyre, South Australia. Palaeogeography Palaeoclimatology Palaeoecology 113: 3–42.

70.

Marlowe

(2005) Hunter-gatherers and human evolution. Evolutionary Anthropology 14: 54–67.

71.

Marx

McGowan

Kamber

(2009) Long-range dust transport from eastern Australia: A proxy for Holocene aridity and ENSO-type climate variability. Earth and Planetary Science Letters 282: 167–177.

72.

Moros

De Deckker

Jansen

et al . (2009) Holocene climate variability in the Southern Ocean recorded in a deep-sea sediment core off South Australia. Quaternary Science Reviews 28: 1932–1940.

73.

Moy

Seltzer

Rodbell

et al . (2002) Variability of El Niño/Southern oscillation activity at millennial timescales during the Holocene epoch. Nature 420: 162–165.

74.

Murray

Wintle

(2000) Luminescence dating of quartz using an improved single-aliquot regenerative-dose protocol. Radiation Measurements 32: 57–73.

75.

Murray

Wintle

(2003) The single aliquot regenerative dose protocol: Potential for improvements in reliability. Radiation Measurements 37: 377–381.

76.

Olley

Roberts

Yoshida

et al . (2006) Single-grain optical dating of grave-infill associated with human burials at Lake Mungo, Australia. Quaternary Science Reviews 25: 2469–2474.

77.

Pardoe

(2006) Becoming Australian: Evolutionary processes that structure biological variation from origin to modern times. Before Farming 2006(1): 4.

78.

Pardoe

(2015) Riverine, biological and cultural evolution in southeastern Australia. Antiquity 69: 696–713.

79.

Pietsch

(2009) Optically stimulated luminescence dating of young (<500 years old) sediments: Testing estimates of burial dose. Quaternary Geochronology 4: 406–422.

80.

Prescott

Hutton

(1994) Cosmic ray contributions to dose rates for luminescence and ESR dating: Large depths and long term variations. Radiation Measurements 23: 497–500.

81.

Quigley

Horton

Hellstrom

et al . (2010) Holocene climate change in arid Australia from speleothem and alluvial records. The Holocene 20: 1093–1104.

82.

Sealy

(2006) Diet, mobility, and settlement pattern among Holocene hunter-gatherers in Southernmost Africa. Current Anthropology 47: 569–595.

83.

Shawcross

Kaye

(1980) Australian archaeology: Implications of current interdisciplinary research. Interdisciplinary Science Reviews 5: 112–128.

84.

Shawcross

(1998) Archaeological excavations at Mungo. Archaeology in Oceania 33: 183–200.

85.

Shulmeister

Lees

(1995) Pollen evidence from tropical Australia for the onset of an ENSO-dominated climate at c. 4,000 BP. The Holocene 5: 10–18.

86.

Smith

(2013) The Archaeology of Australia’s Deserts. New York: Cambridge University Press.

87.

Spry

(2014) Refitting a past: A comparison of late Pleistocene and terminal Pleistocene/early Holocene stone tool technology at Lake Mungo, Southwestern New South Wales, Australia. PhD Thesis, La Trobe University.

88.

Stern

(2015) The archaeology of the Willandra: Its empirical structure and narrative potential. In: McGrath

Jebb

(eds) Long History, Deep Time. Canberra, ACT, Australia: ANU Press, pp. 221–240.

89.

Stern

Tumney

Fitzsimmons

et al . (2013) Strategies for investigating human responses to changes in landscape and climate at Lake Mungo in the Willandra Lakes, southeast Australia. In: Frankel

Webb

Lawrence

(eds) Archaeology in Environment and Technology: Intersections and Transformations. New York: Routledge, pp. 31–50.

90.

Sturt

(1833) Two Expeditions into the Interior of Southern Australia. London: Smith, Elder & Co.

91.

Sturt

(1984) Journal of the Central Australian Expedition, 1844–6 (ed. Waterhouse

). Dover: Caliban Books.

92.

Tumney

(2011) Environment, landscape and stone technology at Lake Mungo, Southwest New South Wales, Australia. PhD Thesis, La Trobe University.

93.

Walker

(2005) Quaternary Dating Methods. Chippenham: John Wiley & Sons.

94.

Webb

Cupper

Robins

(2006) Pleistocene human footprints from the Willandra Lakes, southeastern Australia. Journal of Human Evolution 50: 405–413.

95.

Westaway

Cupper

Johnston

et al . (2013) The Willandra Fossil Trackway: Assessment of ground penetrating radar survey results and additional OSL dating at a unique Australian site. Australian Archaeology 76: 84–89.

96.

Weston

Szabó

Stern

(2017) Pleistocene shell tools from Lake Mungo lunette, Australia: Identification and interpretation drawing on experimental archaeology. Quaternary International 427: 229–242.

97.

Williams

Ulm

Cook

et al . (2013) Human refugia in Australia during the Last Glacial Maximum and terminal Pleistocene: A geo-spatial analysis of the 25–12ka Australian archaeological record. Journal of Archaeological Science 40: 4612–4625.

98.

Williams

Ulm

Turney

CSM

et al . (2015a) Holocene demographic changes and the emergence of complex societies in prehistoric Australia. PLoS ONE 10: e0128661.

99.

Williams

Veth

Steffen

et al . (2015b) A continental narrative: Human settlement patterns and Australian climate change over the last 35,000 years. Quaternary Science Reviews 123: 91–112.

100.

Withers

(1989) Bushmen of the Great Anabranch. Wentworth, NSW, Australia: Maxine Withers.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.95 MB