AUSTRALIAN ARCHAEOLOGY 2020, VOL. 86, NO. 2, 176–197 https://doi.org/10.1080/03122417.2020.1769250

ARTICLE Community research in a public place: Wangangarra 1 rockshelter, Mitchell River National Park, ()

Georgia L. Robertsa,b,c , Russell Mullettd, Bruno Davida,b , Joanna Fresløvd, Gunaikurnai Land and Waters Aboriginal Corporationd, Jerome Mialanesa,b , Fiona Petcheya,e , Matthew McDowella,f , Vanessa N. L. Wongg , Katherine Szabob,h , Daniel Stoesseli, Kyne Krusic-Golubj and Lynette Russella,b aARC Centre of Excellence for Australian Biodiversity and Heritage, Canberra, ACT, Australia; bMonash Indigenous Studies Centre, Monash University, Clayton, VIC, Australia; cDepartment of Archaeology and Natural History, College of Asia and the Pacific, Australian National University, Canberra, ACT, Australia; dGunaikurnai Land and Waters Aboriginal Corporation, Kalimna West, VIC, Australia; eRadiocarbon Dating Laboratory, University of Waikato, Hamilton, ; fSchool of Natural Sciences, Sandy Bay campus, University of Tasmania, Hobart, TAS, Australia; gSchool of Earth, Atmosphere and Environment, Monash University, Clayton, VIC, Australia; hMuseum of Archaeology and Anthropology, University of Cambridge, Cambridge, UK; iDepartment of Environment, Land, Water and Planning, Arthur Rylah Institute for Environmental Research, Heidelberg, VIC, Australia; jFish Aging Services Pty Ltd, Queenscliff, VIC, Australia

ABSTRACT ARTICLE HISTORY Southeastern Australia’s temperate East Gippsland region is a large and diverse landscape Received 12 February 2020 that spans from the coast to the Australian Alps. The region includes a number Accepted 11 May 2020 of national parks and reserves jointly managed by Aboriginal Traditional Owners, the KEYWORDS Gunaikurnai Land and Waters Aboriginal Corporation (the ‘Gunaikurnai Corporation’), and Parks . The Gunaikurnai Corporation recognises that archaeological research can be a Community archaeology; Gippsland; Gunaikurnai; fundamental tool in understanding relationships in past landscapes and managing places Mitchell River National Park; within Country. In 2017, the Gunaikurnai Corporation initiated a long-term collaborative partnership research; study of Gunaikurnai Country with university-based scientists, with the management of Traditional Owners Country through informed decision-making in mind. Here we present results from the first partnership research project in the Mitchell River National Park. A small-scale archaeological excavation of Wangangarra 1, a rockshelter that was not previously recognised to hold arch- aeological evidence, has revealed highly significant deposits spanning from before the Last Glacial Maximum to recent times, including evidence of occupation by the Old People. The results contribute to a better understanding of the Park as a cultural landscape, and demon- strate the success of respectful partnership research with local Indigenous groups as Traditional Owners.

Introduction the Old People1 on Indigenous lands (in local Aboriginal worldviews: ‘Country’), and thus In 2017, the Gunaikurnai Land and Waters approached the Monash Indigenous Studies Centre Aboriginal Corporation, representing the Aboriginal (Monash University) with a research proposal to Traditional Owners of large parts of East Gippsland investigate the archaeology of the Park and its sur- in southeastern Australia, became concerned that rounding areas. The Gunaikurnai Land and Waters while there was a wealth of cultural knowledge Aboriginal Corporation have since developed a for- about customary lands and waters in what is today mal collaborative research partnership with Monash ‘ known as the Mitchell River National Park ( the University and with the Australian Research Council ’ Park ), there was insufficient scientific information Centre of Excellence for Australian Biodiversity and ’ of any depth or value to help manage it in today s Heritage (CABAH), an intellectually independent corporate environment and to assess the risk to cul- but government-funded multi-university research tural values posed by recent and ongoing develop- group that works across disciplines, to investigate ment considerations and aspirations. The the Park among other locations within Gunaikurnai Gunaikurnai Traditional Owners realised that the lands and waters. The aim of this first partnership Park had great potential to tell the long (hi)story of project within the Park was community-driven, to

CONTACT Joanna Fresløv [email protected] Gunaikurnai Land and Waters Aboriginal Corporation, 27 Scriveners Road (Forestec), Kalimna West, VIC 3909, Australia Supplementary data for this article are available online at https://doi.org/10.1080/03122417.2020.1769250. 1The term ‘Old People’ is a respectful term commonly used by Gunaikurnai when referring to past/deceased ancestors. It is also a way of avoiding naming the recently deceased, as per respectful Gunaikurnai cultural ways. ß 2020 Australian Archaeological Association AUSTRALIAN ARCHAEOLOGY 177

Mitchell River National Park



a re P A A R ai rn ku ai n u G

N

100 km

Figure 1. Location of Gunaikurnai Land and Waters Aboriginal Corporation area and Mitchell River National Park. Base Map: Google Earth. Data LDEO-Columbia, NSF, NOAA Image Landsat/Copernicus. Spatial Data: Property of the State of Victoria used under approved licencing conditions. find out more about the past cultural occupation concern for the Traditional Owners regarding the and environmental properties of the Park, using security of the Park and its surrounding cultural careful and reliable scientific methods: how long ago context. Many undated archaeological sites had been did the Old People live along the Mitchell River located during several small-scale surveys of the (Gunaikurnai name: Wangangarra)? How did they Park, which have to date, due to their unsystematic live? Did they change, or maintain, the landscape nature, contributed very little to understanding the through traditional burning practices? What was the Park’s archaeological values or how they might be landscape like in the past and more recently, and affected by changes to the drainage of the Mitchell what kinds of creatures lived in the Park compared River catchment. A number of small rockshelters to now? What did they do when the Old People were known to be present downstream of the loca- camped or otherwise used individual sites? And tion of the proposed impoundment. None had been could recent advances in dating and science help thoroughly investigated, while some had not been Gunaikurnai tell the story of the Michell River and documented at all. All were, however, subject to its people? impacts from unregulated camping within the shel- The Park is heavily forested. It is located 40 km ters themselves. Importantly, while stone artefacts to the northwest of Bairnsdale in temperate East could be seen exposed in surface substrates of some Gippsland (Figure 1). The Park is an area of great of the rockshelters, they appeared to be absent, or at natural beauty and of high cultural significance to least undetected, from others. The presence of soft Traditional Owners, the Gunaikurnai who formally surface sediments, at times of considerable depth, share its management with Parks Victoria, a govern- indicated that these latter sites could still hold con- ment authority. Today it is surrounded by irrigated siderable archaeological deposits. How, then, were farmland, to which the supply of water was prob- these rockshelters, as unrecognised but likely sites, lematic during recent long-term droughts. As a to be addressed? Archaeological excavations were result, a proposal was made in 2017 to build an clearly needed for a better-informed understanding impoundment on the Park’s boundary, on a tribu- of this part of Gunaikurnai Country, including tary of the Mitchell River. This proposition, coupled for community wishes to better know the deeper with proposed sandmining in adjacent areas, caused (hi)story of the Old People within Country. 178 G. L. ROBERTS ET AL.

Gippsland comprises almost a fifth of the area of the laboratory work was undertaken at the Monash the state of Victoria, yet there has been little aca- University laboratories without hands-on commu- demic archaeological research beyond two excava- nity involvement due to community rangers being tions which occurred near the at unavailable at the time. The overseeing Gunaikurnai Cloggs and New Guinea II, 100 km to the Registered Aboriginal Party Manager (RM) and northeast (Flood 1974, 1980; Ossa et al. 1995). Both co-ordinating archaeologists for the Gunaikurnai excavations were poorly dated, were carried out dec- Land and Waters Aboriginal Corporation (JF) and ades ago, and had unreliable chrono-stratigraphic Monash and CABAH (BD) regularly reported to sequences. There is little information about the past Gunaikurnai representative committees such as the occupation of the East Gippsland region and no Elders Sub-Committee. Both in the field and at the indication of what the nature of occupation might Gunaikurnai Land and Waters Aboriginal look like for the Mitchell River valley. As Corporation headquarters, the Wangangarra 1 exca- Gunaikurnai Traditional Owners keep emphasising, vations were an opportunity to train community we cannot assume that the Snowy River valley’s two members in archaeological methods, with all aspects excavated are relevant to other river valleys to explained through two-way discussions in non-tech- the west, even if those excavated sequences had nical language, and for community members to been fine-grained and reliable. Each place has its teach the university-based researchers culturally own story to tell, for the Old People did specific appropriate ways of doing research on and about things in each place that cannot be air-brushed their homelands. It is important to stress the away by reference to distant places. The first stage Gunaikurnai Land and Waters Aboriginal of the research partnership between the Corporation explicitly requested the application of Gunaikurnai Land and Waters Aboriginal high quality and systematic scientific methods to the Corporation and trained archaeologists and related archaeological work, for previous cultural heritage specialists (see below) was therefore to focus on the management projects in Gunaikurnai Country had Park, with Gunaikurnai ancestral places of particular mostly employed very poor standards of investiga- interest to the Gunaikurnai community in mind, i.e. tion, including excavation by heavy machinery that with the aim of beginning to address Gunaikurnai was essentially ‘mining for artefacts’. The Traditional Owners’ questions about Country Wangangarra 1 excavations were small, because the through archaeological and palaeoecological aim was to determine if an archaeological record investigations. existed in the rockshelter, and if it did, what kinds of evidence did the Old People leave behind for cur- Partnership archaeology in rent generations to understand the past? With the Gunaikurnai Country individual archaeological and related scientific meth- odologies explained from the outset (and community Wangangarra 1 was chosen by the Gunaikurnai permissions granted to undertake each explained Land and Waters Aboriginal Corporation as the first method), the Gunaikurnai Land and Waters site to be excavated in the new research partnership Aboriginal Corporation has been in a position to with professional archaeologists, archaeozoologists interpret results for themselves, and to discuss these and geomorphologists because the rockshelter was interpretations with the scientists. Sometimes, where small compared to others nearby; it was in one of interpretations were uncertain or further details the most ‘at-risk’ areas in the Park; and it was an requested, additional analyses were undertaken. In opportunity to begin working together on Country short, the Gunaikurnai Land and Waters Aboriginal in a respectful manner endorsed by the local Corporation (representing the Aboriginal Traditional Aboriginal community. This first, small excavation Owners) embraces meticulous scientific research that in a small rockshelter was identified as both a test both sheds light on the Gunaikurnai past and that is of the relationship between community-based respectful to both Gunaikurnai Country and current (mainly Gunaikurnai) and university-based (non- community members. With these aims and commu- Gunaikurnai) researchers, and as a way of imple- nity processes in mind, we now present the results menting mutually appreciated fieldwork, laboratory of the excavation, and discuss their implications for analyses and overall research designs. All aspects of the project’s original aims. the research—site selection, aims, funding applica- tion, excavation, laboratory work, interpretations, The Wangangarra 1 rockshelter community reporting, publication, media coverage, etc.—were to be done through explicit mutual agree- Mitchell River National Park covers 142.5 km2 of ment and shared decision-making, although in the mainly heavily forested hills and mountains in the case of Wangangarra 1, and by mutual agreement, foothills of the Australian Alps, Australia’s highest AUSTRALIAN ARCHAEOLOGY 179

Figure 2. The Wangangarra 1 rockshelter (photo: Bruno David). mountain range. The Mitchell River runs through a shelter is uneven, mounded in the middle, and steep gorge and incised narrow valley for much of its slopes down towards the back wall. As the roof journey through the Park. Very little is known about slopes steeply downwards at the rear of the shelter, the Park’s archaeological sites, and until now none creating a pinch-point near the middle, only the has been professionally and systematically archaeo- easternmost 5 m of deposit, the area closest to the logically excavated. Over the past decade, preliminary mouth of the shelter, would be suitable for occupa- surveys by Traditional Owners and archaeologists tion today. The site has been recently disturbed by have inspected a series of rockshelters along the visitors to the Park and wild animals, and several southern section of the Park. Several of these were isolated, unburnt animal bones including the jaw of found to have soft deposits with good archaeo- a bettong and several lizard vertebrae could be seen logical potential. on the surface of the deposits. A Park visitor, most The Wangangarra 1 rockshelter is a small over- recently with his partner, was known to have lived hang located 115 m to the west of the Mitchell River in the shelter from time to time, even going to the (Figures 2 and 3). It is 102 m above sea level, facing extent of installing a mattress. They were evicted in the river 250 m south of its junction with Raymond 2019 when their activities caused damage to parts of Creek, a small tributary (Figure 4). The shelter the deposit, which the Gunaikurnai community formed in tilted layers of sandstone and conglomer- became increasingly concerned about given commu- ate. The conglomerate is a coarse sand matrix con- nity knowledge that the Old People had lived in the taining water-worn pebbles and cobbles ranging region deep into the past. Physical evidence of the from 0.5 to 10 cm in diameter. While a few blocks visitors’ presence could still be seen on the shelter of rock 17–84 cm long that dislodged from the roof floor at the time of the excavation, including cans, now lie on the shelter’s floor, evidence of roof-fall string and bread-bag clips. in the form of sand, broken rock, pebbles and cob- bles is abundant on, and in, the ground. The archaeological excavations Under the overhang, the shelter is 8 m deep (from the edge of the overhang to the back wall), The small-scale excavations consisted of two juxta- 9 m wide (across the dripline), 6 m high at the posed 50 50 cm squares aligned north-south, posi- entrance, and covers an area of 48 m2.No tioned in the north-central section of the shelter, in Aboriginal artefacts could be seen on the surface; at an area with good excavation potential (Figures 5 the time of excavation it was not, therefore, recog- and 6). The excavations were geared to addressing nised as an archaeological site. The floor of the the research questions, which did not require more 180 G. L. ROBERTS ET AL.

Figure 3. Soft sediment deposits inside the Wangangarra 1 rockshelter (photo: Bruno David).

Figure 4. View of Mitchell River from the Wangangarra 1 rockshelter (photo: Bruno David). than a small part of the site to be sampled, as the maximum depth of 31.3 cm in Square N12 and aim was not to tell the ‘full’ story of the site, but 24.3 cm in Square N13. XU27, the deepest rather to determine whether archaeological materials Excavation Unit (XU) of Square N12, was dug were present, and if so, their general nature and entirely into the disintegrating upper levels of the antiquity. The excavation proceeded to bedrock at a bedrock. Arbitrary XUs averaged 1.4 cm in thickness AUSTRALIAN ARCHAEOLOGY 181

DEFGHI JKLMNOPQRSTUVWXY

20

19

18

17

16

15

14

13 N13

12 N12

11

10

9

8

7

6

5

4

3

2 N 1

0

-1 01 -2 m

Figure 5. Plan of Wangangarra 1. and were mostly contained within, and followed, aside, the remaining sediments represent the over- visible Stratigraphic Units (SU) (Supplementary whelming majority of accumulated materials within Table SM12). the site, both from cultural and ‘natural’ processes. Animal bones, stone artefacts and other artefacts It was important to understand these sediments, to 30 mm long and smaller selected pieces of charcoal better determine the role that the Old People played were plotted in situ to 1 mm (vertical) and 10 mm in their build-up. Fifteen distinct SUs were identi- (horizontal) precision and individually bagged. Bulk fied. SU2N, the deepest, lies directly on bedrock. sediment samples were collected from each XU of The surface of the bedrock is burnt and fragmenting each square. The rest of the excavated material was into flat, sharp-edged platelets a few centimetres dry-sieved using 2 mm mesh in the field. The long, especially in Square N12. Each SU is described retained materials were later wet-sieved (again in in Supplementary Table SM2 (Figures 7 and 8). 2 mm-mesh), air-dried, sorted and analysed at the The SUs generally lie horizontally across the exca- Monash University archaeology laboratories. vated squares, except for three infilled animal burrows Following completion of the excavation, sediment (see below). Three SUs or sets of SUs are of particular samples were collected from each SU from the walls interest: SU2B, SU2C and SU2H–SU2I–SU2J. SU2B of the excavation pit, and from varied depths along and SU2C are thin lenses of ash-rich and charcoal- the hill-slope within 100 m of the shelter. rich sediment, with SU2C being fire-reddened. They are both found at the same depth and were therefore The sediments probably made at similar times (see below for dating). – – Critical to the interpretation of the site are its SU2H SU2I SU2J are three superimposed, variably buried sediments: the identified cultural materials fire-reddened and sometimes ashy layers near the mid-western section of the excavation at the junction

2To access the SM tables and Appendix A, go to of Squares N12 and N13. The combined Supplementary Materials. SU2H–SU2I–SU2J layers represent a > 70 cm-wide 182 G. L. ROBERTS ET AL.

Figure 6. Excavation in progress. Left to right: Max Solomon, Roberts, Bradley Hood (photo: Bruno David).

WEST NORTH N12 N13 N13 cm XU1 0 modern XU2A SU2A XU1 SU1 SU1 XU3 XU2 XU3A XU4C SU2C XU4A XU4 XU1 XU5 XU5 SU2A/SU2G XU6 1669±28 XU2 SU2G XU6A XU4 XU3 XU7 XU7 XU5 XU8 XU8 XU6A 10 XU9 SU2H SU2I XU9A SU2E SU2H XU8 XU7 SU2D XU9A XU14 disturbance XU10 XU9B XU9A disturbance SU2K SU2J XU10 SU2F XU10 XU15 XU11 XU11 XU17 XU16 XU12 XU12 disturbance XU18 XU14 XU13 XU13 XU13 XU19 XU15 XU16 XU14 XU11 XU15 XU20 XU17 SU2M 20 SU2M XU16 XU21 12,555±40 XU18 XU17 XU12 XU22 XU19 XU18 21,867±87 XU23 XU20 XU19 24,226±115 SU3 (bedrock) XU20 SU2N XU24 XU21 XU21 XU20 SU2N XU25 not excavated SU3 (bedrock) not excavated 30 XU27

40 EAST SOUTH N13 N12 N12

XU1 XU2A XU1 SU2A XU3A 0 SU1 XU1 XU2A XU2B XU2B XU4A XU2 XU3B SU2B XU3 1265±28 XU3A SU2A XU3B XU5 XU4 XU4A XU4B XU4B XU6 XU5 XU5 942±20 873±28 XU7 XU6A XU6 SU2A/SU2G XU7 XU7 XU8 XU8 1492±28 XU8 1320±28 SU2G SU2D XU9A XU9 XU10 XU14 XU9 SU2G XU11 1539±29 disturbance XU12 XU15 XU10 10 XU16 XU15 XU13 SU2F XU16 XU17 SU2L XU14 XU17 XU18 XU11 disturbance XU15 5099±29 XU18 XU19 XU16 XU19 XU20 XU17 XU20 XU21 XU18 SU2M XU12 XU19 XU21 XU22 XU22 XU20 20 XU23 XU23 5117±30 XU13 XU24 XU21 XU24 SU2N XU25 SU3 (bedrock) not excavated XU25 SU2N XU26 SU2N XU26 XU27 SU3 XU27 30

Rock Leaf or twig Root 40 Figure 7. Wangangarra 1 section drawings with radiocarbon dates projected. hearth that extends into the unexcavated north wall of must have burnt on top of the bedrock. No vegetation Square N13. The shape of the visible parts of the fea- grows in the shelter today, nor could it have grown turesuggestsitisabout1mwide. there in the past. The rockshelter is well recessed, and The bedrock at the bottom of both excavation the bedrock could not have burnt through bushfires squaresisextensivelyburnt,indicatingthatafire outside the shelter. The H/C atomic ratios of AUSTRALIAN ARCHAEOLOGY 183

Figure 8. West wall of the excavated squares after completion of excavation. At the base of the pit in Square N12 (left-hand half of photo), the bedrock is slightly deeper than in Square N13 because XU27 consisted of the fragmented bedrock (photo: Bruno David). sediments also indicate more intensive levels of burn- Wangangarra 1, deposition of sediments that have ing immediately on the bedrock than on the adjacent been weathered and eroded from the surrounding hill-slope (see below). The most parsimonious explan- rocks inside the rockshelter. Sediment particle sizes ation for the burnt deposits is that they are the result outside the shelter were found to be different from of campfires, but this needs to be tested through fur- those inside the shelter (Figure 9; for methods used, ther excavation in other parts of the shelter, both to see Supplementary Appendix A), with samples determine whether in situ hearths are preserved, or if derived from the hill-slope having lower proportions identifiable artefacts, food refuse or other cultural of sand and higher proportions of silt and clay. materials are present just above the bedrock. Conversely, those from inside the shelter have much The intrusive, and mixed, sediments of three ani- higher proportions of sand, which suggests that the mal burrows were located by excavation, and could depositional environments inside and outside the in most cases be separated from surrounding sedi- shelter were quite different. ments during excavation, but at times they were so Similarly, carbon concentrations are much higher faint that they could only be made out from the on the hill-slope than inside the shelter (Figure 9). walls of the pit after excavation (for further details, This is because there has been continual input of see Supplementary Appendix A). Given their strati- plant litter and other organic matter from surround- graphic positioning, all three burrows date to about ing vegetation on the hill-slope, but not inside the the same time, and could have been made by the shelter. Within the shelter, a notable peak in carbon same animal. It is difficult to determine what animal concentrations is nevertheless present in the SU2B species made them. However, the volume of dis- hearth layer, which suggests an accumulation or placed sediment suggests they were made by bet- higher concentration of organic matter or other tongs or potoroos (rabbit-sized marsupials of the material high in carbon such as charred organic Potoroidae family). Each of these taxa digs for much material or charcoal, as expected of a hearth. of their food and still occupies parts of East Furthermore, the lower atomic H/C ratios in the Gippsland today (Bilney et al. 2010). shelter as opposed to those in hill sediments, indi- cates that there has been more burning under the Sediment analyses cover of the rockshelter than on the surrounding hill-slope (Figure 9). Atomic H/C ratios can be used Particle size distributions can provide information to indicate the temperature at which organic on the environment and processes that caused soils materials were heated under pyrolysis (decompos- to form. Generally, finer particles such as clays can ition under high temperatures) to form biochar travel greater distances than sands, and can also (carbon-rich materials caused by burning). In gen- indicate low-energy depositional environments. eral, lower atomic H/C ratios of sediments suggest Larger particles such as sands can indicate higher- evidence of burning or deposition of ash or charcoal energy depositional environments, or, such as at (Xiao et al. 2016). It is significant that the lowest H/ 184 G. L. ROBERTS ET AL.

0.5

0.5

Figure 9. Sediment analyses on Wangangarra 1 SUs and surrounding hill-slope.

C ratios inside the shelter are found in the three boundaries representing a gap in the temporal hearth layers SU2B, SU2C and SU2I, and on the sequence. The internal consistency of the calibrated burnt bedrock (Figure 9). These results support the ages was then assessed using an outlier detection stratigraphic interpretation of SU2A, SU2B and method that enables a probabilistic measure of the SU2H–SU2I–SU2J as hearths, and of the bedrock as degree to which samples appear to be outliers, and having been burnt (see below for their dating). then calculates an offset relative to the context within which it is found (Bronk Ramsey et al. 2010). A General t-type Outlier Model was assigned Radiocarbon dating to all ages with a prior outlier probability of 0.05. Charcoal was present in all SUs and all XUs except The impact of outliers on the model can be assessed the basal 60 mm (XU25–XU27) of Square N12. by the convergence values generated (Table 2). Twelve accelerator mass spectrometry (AMS) radio- These should be >95%. Lower values indicate many carbon dates were obtained on individual pieces of different incompatible solutions to the model at charcoal, and one from an otolith. All radiocarbon these points. dates were calibrated to 95.4% probability using This model identified four minor outliers (<10% OxCal v4.3 (Bronk Ramsey 1995) with the IntCal13 probability of being an outlier); Wk-48194, Wk-48200, curves (Reimer et al. 2013)(Tables 1 and 2). Wk-48202 and Wk-48199. The modelled result for To provide a temporal outline, we used Bayesian ‘Transition SU2M early/late’ returned a low conver- Sequence Analysis, whereby radiocarbon ages are gencevalueof91.3%.Theextremeagedifference constrained by prior stratigraphic information between the early and late ages in SU2M requires (Bronk Ramsey 2009). In this model (Figure 10), the further investigation, but clearly a significant amount dates were grouped into three phases (Phase of time has lapsed within this stratigraphic unit. SU2A–2G, Phase SU2M, and Phase SU2N), sepa- Overall, the remaining model returned good results rated by contiguous boundaries because they share a with convergence values of >97%. Table 2 shows the common border. SU2H, SU2I, SU2J, SU2K and ageofgroupedSUsasmodelledbytheBayesian SU2L have been modelled with sequential sequencing, and indicates that burning began within AUSTRALIAN ARCHAEOLOGY 185

Table 1. Radiocarbon determinations, Wangangarra 1. Calibrations undertaken in OxCal v4.3 with IntCal13 curve selection (Bronk Ramsey 1995; Reimer et al. 2013). Sample Wk-48192 calibration undertaken in Calibomb with SHZ1-2 curve selection (Hua et al. 2013). Sample Wk-49083 is a fish otolith, all other samples are single pieces of charcoal. Unmodelled Calibrated Age (95.4% Median Depth Below probability) Calibrated Age Ground (cm) SU XU Laboratory Code d13C(&) 14C Age (BP) (cal BP) (cal BP) Square N12 1.0 1 1 48192 24.7 ± 0.5 122.8 ± 0.5% AD 1961–1985 4.6 2B 3B 48193 25.5 ± 0.5 873 ± 28 680–790 740 4.8–6.1 2A/2G interface 5 49083 5.4 ± 0.5 942 ± 20 730–910 790 4.3 2A/2G interface 5 48194 26.8 ± 0.5 1,265 ± 28 1,060–1,270 1,130 5.2 2G 7 48195 26.6 ± 0.5 1,320 ± 28 1,090–1,280 1,220 9.8 2G 9 48196 26.4 ± 0.5 1,539 ± 29 1,300–1,450 1,370 25.7 2M 24 48199 25.5 ± 0.5 5,117 ± 30 5,730–5,920 5,820 25.9 2N 23 48198 23.7 ± 0.5 21,867 ± 87 25,850–26,250 26,040 26.4 2N 22 48197 26.1 ± 0.5 24,226 ± 115 27,895–28,554 28,220 Square N13 5.7 2G 6A 48201 27.1 ± 0.5 1,492 ± 28 1,290–1,400 1,340 6.3 2G 4 48200 28.5 ± 0.5 1,669 ± 28 1,420–1,590 1,530 13.8 2M 15 48202 25.4 ± 0.5 5,099 ± 29 5,710–5,910 5,820 22.8 Base of 2M 19 48203 26.1 ± 0.5 12,555 ± 40 14,370–15,090 14,780

Table 2. Results of Bayesian sequence model. Rows highlighted in grey represent boundary ages. Modelled (cal BP) 95% probability Name from to median outlier probability convergence Phase SU2A–SU2G End 730 −3,340 −360 99.3 End SU2A 800 80 640 99.9 SU2B firescoop Wk-48193 (873, 28) 800 680 740 96.2 99.9 SU2A–SU2G interface Wk-49083 (942,20) 910 730 800 96.5 99.9 Wk-48194 (1,265, 28) 1,190 990 1,100 93.9 99.9 Transition SU2G–SU2A 1,250 1,060 1,150 100 Wk-48194 (1,265, 28) 1,270 1,090 1,180 95 99.9 Wk-48195 (1,320, 28) 1,290 1,110 1,230 96.5 99.9 Wk-48200 (1,669, 28) 1,590 1,400 1,490 93.8 99.8 Wk-48196 (1,539, 29) 1,450 1,300 1,370 96.6 100 Wk-48201 (1,492, 28) 1,410 1,290 1,340 96.7 99.9 Start SU2G 1,820 1,380 1,550 99.7

End SU2H–SU2I–SU2J–SU2K–SU2L 4,290 1,440 2,220 99.4 Start SU2H–SU2I–SU2J–SU2K–SU2L 5,810 1,740 4,140 94.9

Phase SU2M End SU2M 11,380 2,880 5,630 96.9 Wk-48202 (5,099, 29) 11,950 5,650 5,820 90.7 98.0 Wk-48199 (5,117, 30) 12,760 5,650 5,820 90.9 98.8 Transition SU2M early/late 14,840 5,730 6,150 91.3 Wk-48203 base of SU2M (12,555, 40) 15,090 14,330 14,770 95.2 99.5 Start of SU2M 23,210 14,450 17,100 95.8 Transition SU2M/SU2N 26,160 16,840 23,220 98.7 Phase SU2N Wk-48198 (21,867, 87) 26,280 25,840 26,040 95.5 99.3 Wk-48197 (24,226, 115) 28,560 27,870 28,210 95.5 99.1 Start – interface with bedrock 43,880 27,970 31,220 97.6 the shelter sometime before 27,970 cal BP, with mul- different from the numbers produced by tiple phases of burning and increased sedimentation Supplementary Table SM3 because on Figure 11 occurring after 5,810 cal BP (the SU2H–SU2I–SU2J values have been standardised per cubic metre of hearth) with noticeable further increases after 1,820 cal excavated material for better comparability, as not BP (incorporating the SU2B and SU2C hearths). all XUs were dug to the same thickness, nor did all XUs cover the entire square.

Distribution of cultural materials The stone artefacts Supplementary Table SM3 lists the distribution of excavated cultural materials by XU. Figure 11 shows The main challenge of the Wangangarra 1 stone the density of cultural materials by XU; it is assemblage concerns the exfoliation of the 186 G. L. ROBERTS ET AL.

Figure 10. Bayesian model for radiocarbon ages from Wangangarra 1 grouped by SU. Black distributions represent boundary ages after Bayesian modelling. Red distributions represent the modelled radiocarbon ages for each date. The notation [O:2/5] indicates a 2% posterior probability of being an outlier in the model. rockshelter’s walls, ceiling and bedrock. The con- To help distinguish stone artefacts definitely stant cycle of expansion/contraction through diurnal flaked by people from exfoliated and heat-fractured and seasonal temperature and moisture variations rocks, the upper levels of the >27,970 cal BP burnt has resulted in the rock fracturing into thin, sharp fragmented bedrock at the base of Square N12 ‘flakes’, sometimes called ‘onion peeling’ (Figure 12). (XU27) were excavated. This enabled the establish- Stress fractures from temperature changes can also ment of the characterisation of thousands of sharp be caused by strong heat from fires (either campfires stones that came from the fragmenting bedrock but against the rock, or bushfires) (David et al. that were not flaked by people, for comparison with 2017:395). The extreme heat of fires causes rocks to the sharp stones that were found in the layers of break in a variety of ways, including ‘potlidding’. soft deposit above the bedrock. Some of these sharp The recovery of many potlids, rocks with potlid stones, both from the bedrock and buried in the scars, and fire-cracked rocks from Wangangarra 1 overlying soft sediments, have distinctive recurring indicates that fires affected many of the now-buried elongated (blade-like) shapes such as L-bends stones during the period that the soft sediments (Figure 13). Such naturally produced blade-like built up. L-bend stones with sharp edges have not previously Fine-grained pebbles (e.g. mudstone, slate, chert) been documented in Australian archaeology, and at are abundant in the shelter’s conglomerate matrix some sites have probably been automatically (Easton 1938; McRae-Williams et al. 1981:270). included in artefactual assemblages by virtue of their Exfoliation by the heat of the sun, frost and high elongation and sharp edges. Since in archaeology temperatures from fires, can cause cracks in the fine- ‘flaked pieces’ are defined as artefacts lacking ventral grained pebbles along their fracture planes, creating (inner) surfaces and displaying only flake scars on sharp-edged broken rocks (‘pseudo-artefacts’). one or more surfaces, while scars from the ‘natural’ AUSTRALIAN ARCHAEOLOGY 187

N12 Charcoal Bone Flaked stone Shell Fish scales (g/m3) (g/m3) artefacts (#/m3) (g/m3) (g/m3) 0 30000 60000 0 2000 4000 0 100 200 0 5000 10000 0 2000 4000 1 2A 2B 3A 3B 4A 4B 4C 5 6 7

t 8 9 10 11 12 13 14

xcavation Uni 15

E 16 17 18 19 20 21 22 23 24 25 26 27

N13 Charcoal Bone Flaked stone Shell Fish scales (g/m3) (g/m3) artefacts (#/m3) (g/m3) (g/m3) 0 2000 4000 03006000 200 400 0 500 1000 02040 1 2 3 4 5

t 6A i 6B 7A 7B 8 9A 9B 10 xcavation Un 11 E 12 13 14 15 16 17 18 19 20 21

Figure 11. Density of cultural materials by cubic metre, Wangangarra 1. detachment of broken rocks also regularly form on indicates that the Wangangarra 1 sharp rocks exfoliated and heated-fractured rocks, the potential resulted from heat fractures rather than flaking to mistake flaked pieces (made by people) from by people. heat-fractured sharp rocks (not necessarily made by Only two definite flakes were found in the exca- people) was high. vation (Figure 14). An absence of small ‘shatter Exfoliated and heat-cracked sharp rock fragments flakes’ produced during stone tool manufacture, and lack the characteristic bulbs of percussion emanating an absence of cores, suggest that the two flakes were from impact points on striking platforms, and the brought to the site rather than knapped on-site. The large numbers of sharp rock fragments at absence of tiny flake fragments between 2 and 5 mm Wangangarra 1 all lacked bulbs of percussion and long confirms off-site manufacture. The raw materi- impact points. Detachment scars on the surfaces of als from which these two artefacts were made is fur- rocks also lacked negative bulbs of percussion. This ther evidence of manufacture away from the shelter; 188 G. L. ROBERTS ET AL.

Figure 12. Example of ‘onion peeling’ on the Wangangarra 1 rockshelter ceiling above the excavation squares (photo: Bruno David).

Figure 13. Examples of ‘pseudo-flakes’, some with L-shape bends, caused by exfoliation of the Wangangarra 1 rockshelter ceil- ing above the excavation squares. A: N12 XU11. B: N12 XU20. C: N12 XU15. D: N12 XU19. E: N12 XU2. F: N13 XU2. G: N12 XU2A. Scale in 1 cm units (photos: Steve Morton). these raw materials are not found within the shel- catchment, as many sites with chert have been ter’s rock matrix. Both are complete flakes, manu- found at Wire Plain and the Victoria River south of factured using unipolar (freehand) percussion. In Hotham, and at Talbotville and the Square N12, the complete flake from XU9 is made in the Wonnangatta Valley (Fresløv 2005:82–83; of a dark grey (Munsell colour: GLEY 1 N4) to olive Fresløv et al. 2004:56, 157). The Wangangarra 1 grey (Munsell colour: 5Y 4/2) chert. The source of chert flake is small, measuring 13.3 mm long, the chert is unknown. Lower Ordovician chert 18.9 mm wide and 6.7 mm thick. It has a single- occurs in sediments to the northwest of the Park at flaked platform with evidence of preparation (over- Tabberabbera (Fergusson 2003:479). However, chert hang removal), and several flake scars on its dorsal is uncommon in sites in the lower Mitchell River surface. It also displays the remnant of an old strik- catchment, and mudstone may often have been mis- ing platform along its lateral left margin, indicating identified as chert. It is nevertheless likely that the that the core from which this flake originated was source of the chert artefacts is in the upper rotated at least once during its use-life. In Square AUSTRALIAN ARCHAEOLOGY 189

Figure 14. Wangangarra 1 excavated stone artefacts. A: N13 XU17. B: N12 XU9. Scale in 1 cm units (photos: Steve Morton).

N13, the complete flake from XU17 is made of a The terrestrial fauna light brownish grey (Munsell colour: 10YR 6/2) Eight hundred and seventy-five bones, most fine-grained quartzite. Although the source location from small (<1 kg) to medium-sized (1–5 kg) is unknown, it is also thought to lie in the Mt mammals (Supplementary Tables SM3 and SM4), Hotham-Dargo region, again because of the abun- were recovered from each XU, with small taxa dance of stone artefacts made on this kind of raw making up 97% of the total excavated bone by material in this region (Fresløv et al. 2004:58). It has weight. The high pH (range: 7.7–8.9) of the sedi- a single flake scar originating from the proximal ments has been conducive to good bone end on its dorsal surface. The flake was used as a tool, as evidenced by the presence of fine edge dam- preservation. The animal bones were examined under 10–40 age along its distal end. The Square N12 XU9 flake came from SU2I, part magnification and identified using published of the SU2H–SU2I–SU2J hearth that dates to descriptions and animal bone reference collections between 1,440 and 5,810 cal BP, indicating working held at the Tasmanian Museum and Art Gallery. with stone tools around the hearth. Upper and lower jaws, teeth and some postcranial The Square N13 XU17 flake came from near the bones were used for identification. All bones were top of SU2M, from a level dated to sometime identified to the lowest taxonomic level possible (i.e. between 2,880 and 14,840 cal BP; the broad range is species where possible), though due to a high degree due to an insufficient number of dates to refine the of fragmentation, many could not be identified chronology further in this part of the sequence. This beyond Class. The zoological notation ‘cf.’ was used represents the oldest recognisable stone artefact to indicate a species identification that was uncer- from the two excavation squares (see above). While tain but probably correct. Numbers of identified there are only two definite stone artefacts from the specimens (NISPs) by taxon were identified, rather Wangangarra 1 excavation, both of which originated than minimum numbers of individuals (MNIs), from more distant parts of Gunaikurnai Country or because of the small sample size and research ques- beyond, it is possible that people also used exfoli- tions asked of the assemblage (Lyman 2008; ated and/or heat-fractured ‘pseudo-flakes’ as tools. McDowell 2014). 190 G. L. ROBERTS ET AL.

Preservation white and changing its structure dramatically Bones were generally well-preserved but highly frag- (Fernandez-Jalvo et al. 2018). mented. With the exception of burnt specimens, Despite the excavation of three hearths (SU2B, they were typically a uniform light brown colour SU2C, SU2H–SU2I–SU2J), only 21 bones show evi- and showed no evidence of subaerial weathering, dence of burning. Alterations to bone colour and suggesting that they had not been exposed to the structure that signal a variety of temperatures were open environment for prolonged periods of time observed throughout the excavation, but represent and therefore that they had been buried rapidly. only 2.4% of the excavated bones. The burnt bones were not concentrated together. Instead, they were Breakage mixed with many other, unburnt bones Bones were checked for break angles, type of break- (Supplementary Table SM4). This suggests that at age, evidence of depredation from other animals, least some of the bones were burnt by hearths built and post-depositional modification such as rodent on top of them, rather than by being deliberately gnawing (Supplementary Table SM4). Three bones thrown into a fire e.g. during or after a meal. The had multiple small point compressions consistent implication is that the excavated animal bones did with chewing by a mammalian carnivore. In add- not result from food refuse that was systematically ition, four bones had crenulated breaks and three thrown into fires (as regularly done ethnographically showed evidence of acid digestion and edge-round- in many contexts across Aboriginal Australia). Two ing. Each of these features indicates that the bones bones show evidence of point burning, where part were chewed and/or digested by mamma- of the bone was charred but the majority remains lian predators. unburnt, consistent with these interpretations. Several other bones had spiral or obtuse breaks, typically created by the application of rotational Body size force to fresh bone and usually associated with Animal body size was divided into three types cause of death. In fresh/living bone collagen, a long, (small ¼ <1 kg, medium-sized ¼ 1–5 kg, and large fibrous tri-helical protein wraps around bone, mak- ¼ >5 kg animals) (see Supplementary Table SM4 ing it flexible, resilient and difficult to break. for details). While bones from each size category However, if enough force is applied, fresh bone will were observed, those from medium-sized and large break in a spiral or helical pattern along the colla- animals were rare. Small mammal bones predomin- gen fibres. After death collagen breaks down, mak- ate. Only one, very small unburnt bone fragment ing bone much easier to break, usually doing so from near the base of the excavation (Square N12 perpendicular to the bone shaft. However, the high XU22) came from a large animal (a macropod); it degree of bone fragmentation, much of which is shows no signs of being food refuse, although the associated with independent evidence of people at remaining parts are likely to occur elsewhere in the Wangangarra 1, suggests that some of the bones site and could yield further information. may have been crushed by treadage. We note that small owls such as boobooks Taxonomy (Ninox boobook) inflict more damage on bones than The diagnostic bones were dominated by mammals, larger owls, but both cause spiral or acute-angled with fish bones, fish scales and a few bird and rep- breaks. Most of the bones we encountered had per- tile bones also present. Due to the high degree of pendicular breaks, indicating that bones were bro- bone breakage, only a few bones could be identified ken long after death; broken bones do not always to genus or species (Supplementary Table SM4). indicate the collection agent. The characteristics of The faunal remains did not reveal any trends in species in the assemblage are much more inform- species abundance, taphonomy or taxonomy. Body ative, and most of the animals recovered are noctur- size analysis showed that small mammals occurred nal small mammals frequently hunted by owls, in almost every XU. Burnt bone was rare compared suggesting that owls may have been involved in the to unburnt bone, and breakage, while extreme, did deposition of some of the bones. not appear to change with time nor appear deliber- ate. Despite independent evidence (e.g. stone arte- Burning facts, hearths) that the site was on occasions Bone burnt at low temperatures (180–200 C) takes occupied by people, much (but not all, see below) of on a deep brown colour. When heated to about the animal bone assemblage is more characteristic of 300 C, it blackens. When heated to 400–650 C, it an owl roost, though bones bearing quoll (Dasyurus becomes grey and begins to change chemically. And sp.) or dunnart (Antechinus sp.) tooth marks show when it is heated to more than 700 C, it becomes that a mammalian predator also used the shelter calcined, a process that oxidises the bone, turning it (there are no signs of rodent tooth marks). None of AUSTRALIAN ARCHAEOLOGY 191 the mammal remains is likely to have been depos- impacts including competition with, and destruction ited by the Old People. They do, however, reveal by, exotic herbivores and predators such as rabbits information of considerable palaeoeco- and foxes. However, its distribution is reasonably logical interest. well understood from Late Holocene fossils (Medlin Most of the bones identified to species level are 2008), and its presence in the Mitchell River region from animals known to have lived in and around is of little surprise. the Park over the past 20 years (Atlas of Living The remaining two rodents (Pseudomys australis Australia 2019). These include Sminthopsis sp. (pos- and Pseudomys shortridgei) recorded from sibly Sminthopsis leucopus, the white-footed dunnart; Wangangarra 1 appear to be rare in the East Gunaikurnai name: jaiung), Pseudocheirus peregrinus Gippsland region. The plains mouse (Pseudomys (common ringtail possum; Gunaikurnai name: australis) is commonly recorded from Australia’s blang), Cercaretus nanus (eastern pygmy possum), arid interior but has been collected from Victoria Mastacomys fuscus (broad-toothed mouse) and near Cambarville. When heavy rains fall in the arid Rattus fuscipes (bush rat; Gunaikurnai name: baiuk). zone, plains mice are capable of irrupting in enor- In addition, several species of bat are known from mous numbers to exploit food resources that grow the area. A partial dentary identified as a potoroid following rain (Pavey et al. 2014). It is possible that probably represents Bettongia gaimardi (eastern bet- the plains mouse whose remains were found at tong) or Potorous tridactylus (long-nosed potoroo; Wangangarra 1 migrated south following one such Gunaikurnai name: mandu), both of which have event. The heath rat (Pseudomys shortridgei) is com- been recorded in East Gippsland since European monly recorded in Gariwerd and western Victoria, colonisation. In contrast, Thylogale billardierii and may have had a larger range prior to European (Tasmanian pademelon; Gunaikurnai name: bowie) colonisation than previously recognised. Heath rats is now extinct on mainland Australia, but continues have been mistaken for swamp rats in the past to thrive in Tasmania. (Kemper et al. 2010), so the species may also be Pseudomys higginsii (long-tailed mouse) was under-reported from the Mitchell River area. found in the assemblage but is now restricted to The archaeological evidence indicates that the Tasmania. However, fossils have previously been animal bones have accumulated at Wangangarra 1 reported from King and Flinders Islands in Bass in a number of ways. People (see below), owls, and Strait (Driessen 2008), Wombeyan Caves in NSW mammalian carnivores or scavengers appear to have and Pyramids Cave in eastern Victoria (Wakefield 1972). The ages of those fossil assemblages are all brought in animals and their bones. Given that poorly understood, however, and while Wakefield only 1% of the site has been excavated, and that assumed that they had accumulated during the there is excellent preservation of faunal remains , they do not appear to have been reliably over a considerable depth and timeframe, many dated. The Pseudomys higginsii bone from other fauna are likely to be represented in other Wangangarra 1 was found in XU5 at the SU2A/ (unexcavated) parts of the site, and in other sites SU2G interface, at a level dated to between within this region. 80–1,250 cal BP. The discovery of long-tailed mouse remains at Wangangarra 1 extends the fossil range The fish remains of this species almost 100 km west of Pyramids Cave, its previously most westerly occurrence, and A total of 106 fish scales and fish scale fragments indicates that the species was more widely distrib- were excavated (Figure 15), 102 from Square N12 uted on mainland Australia prior to the arrival of and four from N13. All the fish scales came from Europeans than previously recognised. Furthermore, the upper 21 cm of the excavation, from XU12 in the associated radiocarbon ages suggest that the spe- Square N12 and above. Almost all of the scales (92, cies persisted on the mainland for much longer than or 87%) came from 3 to 5 cm depth, the level of the previously thought, and may have survived into SU2B hearth dated to 80–1,250 cal BP (XU3B and recent times. If so, its disappearance from mainland XU4B). The rest of the fish scales came from sedi- Australia may relate to the introduction of foxes, ments surrounding the SU2B hearth. which are absent in Tasmania. One fish otolith was found in Square N12 XU5 The occurrence of Pseudomys cf. auritus (long- (Figure 16), in an XU slightly below but that interfa- eared mouse) at Wangangarra 1 also represents a ces with the base of both the SU2B and SU2C small increase in its geographical range. The species hearths. It was directly AMS radiocarbon dated to is represented by a single lower 1st molar, hence the 730–910 cal BP (Wk-49803), overlapping with the uncertainty associated with its identification. This fire scoop. This is consistent with the observation large rodent was driven to extinction by European that the fish remains are largely associated with the 192 G. L. ROBERTS ET AL.

Figure 15. Excavated fish scales from Wangangarra 1 (photos: Steve Morton).

SU2B and SU2C hearths dated to within the period 80–1,250 cal BP. Based on its size (11.2 mm-long), weight (0.08 g) and morphological characteristics, the otolith most probably came from an Australian bass (Macquaria novemaculeata). The Australian bass is a long-lived species (up to 47 years; Stoessel et al. 2018) found in freshwater coastal rivers and streams along the east coast of Australia, from Wilson’s Promontory in Victoria to Tin Can Bay in Queensland (Harris 1983, 1985; Williams 1970). A medium-length fish such as the one from Wangangarra 1 could weigh up to c. 1 kg (Bray and Thompson 2018). For this region, an otolith of the size found would come from a medium-length Australian bass, typically 26–27 cm long. Australian bass are best caught using a rod or line and lures (Prokop 2006). Early accounts from eastern Victoria describe a range of Aboriginal fishing practices including netting, spear Figure 16. The Wangangarra 1 Australian bass otolith fishing, trapping and opportunistic harvesting, with (photos: Steve Morton). live storage pens also utilised (Clark 2002). In the 1800s and early 1900s, bone hooks of a kind that Mussel shell could have been used to catch Australian bass such as the one found at Wangangarra 1, were docu- The only type of shell found in the excavation is mented as being used in the nearby Gippsland freshwater mussel (Gunaikurnai name: marlung- Lakes region (Mackaness 1941; Massola 1956; Smyth dun), although the excavated shells are too frag- 1878). The use of bone hooks is, however, suggested mented to allow identification to species. The pres- to have not been widespread, with wooden hooks ence of periostracum on shell fragments throughout recorded from Bay (Massola 1956), their stratigraphic distribution indicates that while and shell hooks, commonly used along the NSW some horizontal or vertical movement may have coast, to the geographical southern extent of Lunella occurred, they did not travel far once on or in the torquata (Turbo torquata in Attenbrow 2010). ground (Figure 17). None of the shell fragments AUSTRALIAN ARCHAEOLOGY 193

Figure 17. Examples of excavated shell fragments with periostracum, Wangangarra 1. Top: N13 XU6. Bottom left: N12 XU4A. Bottom right: N12 XU3B (photos: Steve Morton). shows any of the characteristic signs of animal dam- Tyers Mission (established in 1863), the missionary age, such as fragmentation patterns indicating meat John Bulmer (b. 1833, d. 1913) (Campbell and extraction by birds or other non-human predators. Vanderwal 1999:61) recorded that: Together, nearly three-quarters (35 of the 49, 71%) of Sometimes when a young man was very hungry the shell fragments came from one or other of the (wilke wilkanu) or in Gippsland tongue (ganu three hearths. The remaining 14 fragments all lay ganook) he would take a bag and dive to the very close to these hearths. Many of the shell frag- bottom of the river and get mussels, it would take 3 ments are burnt. Together, these factors indicate that or 4 dives to fill his bag, but as soon as it was full the shellfish, like the fish remains (see above), were he would go quietly ashore and roast his catch. He seemed to enjoy his repast. brought to the site and eaten there by the Old People at various times in the Middle to Late Holocene. The most recent items The closest and most likely source of the shells is the Mitchell River that flows adjacent to the site. A small number of very recently deposited items were Quieter areas of the lower reaches of the Mitchell found in the excavation: aluminium foil, wax from River currently hold populations of freshwater mus- candles, and small pieces of paper and string (Figure sels (Atlas of Living Australia 2019), and this is also 19).Theseallcamefromthetopmost4.8cmofthe likely to have been the case in the past. Four species deposit only (in Square N12, XU1–XU3; in Square are commonly found in the area: Velesunio ambi- N13, XU1), except for one tiny piece of aluminium guus, Hyridella australis, Hyridella depressa and foil weighing less than 0.01 g that came from the Hyridella drapeta (Atlas of Living Australia 2019). looser sediments of animal burrow SU2F in Square All are found in the beds of streams, rivers, lakes N13 XU11. These items of the topmost sediments all and other freshwater bodies such as dams. came from a level associated with a near-surface post- Shell Artefact. One fragment of mussel shell from bomb radiocarbon date calibrating to AD 1961 and Square N12 XU2B, associated with the 80–1,250 cal 1985 (the highest probability [0.847] being on the BP SU2B hearth, shows evidence of use; its margin down-sideofthebombcurveatAD1985).Thevisitor has been modified from having been used as a scra- to the Park who lived in the shelter between 2007 and per or cutting tool (Figure 18). 2019 (see above) was almost certainly the source of Few accounts of the use of freshwater mussels the pieces of aluminium foil, candle wax, paper and have been documented for any part of southeast string recovered in the archaeological excavation. It is Victoria. Angas (1847) described the use of fresh- of interest that there are no traces of such modern water mussel shells as spoons and cutting imple- materials in the undisturbed deposits below 4.8 cm ments, most commonly by women. At the Lake depth. Even the radiocarbon-dated pieces of charcoal 194 G. L. ROBERTS ET AL.

Figure 18. Section of the used margin of mussel shell showing fine chipping, Wangangarra 1 N12 XU2B (photo: Steve Morton).

Figure 19. Very recent items from the upper levels of the Wangangarra 1 excavation. Scale in 1 cm units (photos: Steve Morton). A: Candle wax. B: Newspaper fragment. C: Unburnt match-head. D: Cork. E: Aluminium foil. F: Plastic. G: String. from 4 to 5 cm depth, the level that also has the deep- by Aboriginal Traditional Owners, to reveal details est in situ object of recent occupation, proved to be of the Old People in a co-managed national park much older than the surface charcoal, indicating that that is today commonly frequented by visitors. A at least this part of the site remained relatively undis- small, preliminary excavation at the Wangangarra 1 turbed. This is a good indication that in the rockshel- rockshelter demonstrates the great potential for this ters of this region, the gravelly deposits firmly and other rockshelters to contain long archaeo- compacted by fine ash, silt and sand are likely to logical sequences, despite the absence of demon- retain in situ cultural deposits with good chrono- strable Aboriginal artefacts on the site’s surface. stratigraphic resolution despite some degree of more This is a significant find for East Gippsland and recent foot traffic. Gunaikurnai Country, where the only other pub- lished evidence for occupation across the broader Discussion and conclusion region had until now been at Cloggs Cave and New Wangangarra 1 is a rare example in Australia of an Guinea II some 100 km to the northeast (see Flood archaeological excavation initiated and co-ordinated 1980; Ossa et al. 1995). Those two sites had revealed AUSTRALIAN ARCHAEOLOGY 195 long occupational sequences, although in both cases connections with the lakes and coastal land- the original dates and chrono-stratigraphic sequen- scapes downstream. ces obtained in the early 1970s and early 1980s are In conclusion, the partnership research pro- now known to be problematic due to problems of gramme in which the Wangangarra 1 study took sediment mixing and insufficient pre-treatment place has yielded five major results: chemistry. Wangangarra 1 has not only revealed evidence of well stratified deposits and good bone 1. That archaeological research between preservation, but of a long sedimentary sequence Traditional Owners and professional scientists that insinuates the likely presence of people more can not only be scientifically fruitful, but also than 27,970 cal BP, given that the bedrock was welcomed and of great benefit to Traditional extensively burnt in this well-recessed rockshelter Owners who wish to further research the life of and that fires must thus have burnt on it; future the ancestors, if undertaken respectfully and in excavations are planned to further investigate true partnership with mutual recognition of Wangangarra 1’s cultural history. This is of particu- specialist expertise and Aboriginal connections lar interest to the Gunaikurnai Land and Waters with ancestral sites and Country. Aboriginal Corporation, for long archaeological 2. That a number of locally and regionally extinct sequences are particularly rare across Victoria. A animal species survived wider and later in East new series of radiocarbon, OSL and U-series ages Gippsland than previously thought. indicate that Cloggs Cave was occupied from some- 3. That rockshelters with soft deposit but without time before c. 23,000 years ago (David et al. in any surface archaeology may well hold buried press), while at New Guinea II cave nearby occupa- cultural deposits and that these should be inves- tion dates to >25,000 cal BP (non-basal date of tigated as part of management planning. 21,000 + 900/800, SUA 2222) (Ossa et al. 1995:28, 4. That fires over conglomerate bedrock can pro- Table 1). Elsewhere in Victoria, older occupation duce sharp-edged, blade-like L-shaped ‘pseudo- occurs in open sites at Bend Road near Melbourne, artefacts’ of repeated form, but lacking the tell- commencing c. 30,000 years ago (Hewitt and Allen tale marks of flaked stone artefacts (striking 2010:6, 14), and Box Gully at Lake Tyrell in north- platform, bulb of percussion, eraillure scar etc.). west Victoria, where first occupation dates to 5. That during the millennia just preceding the 26,000–32,000 cal BP (Richards et al. 2007:5). The coldest years of the last Ice Age, at a time pre- Drual rockshelter in Gariwerd, with occupation ceding any other known site in East Gippsland, beginning by 25,988–26,840 cal BP (22,140 ± 160 BP), the Old People were probably living in the foot- is the only other rockshelter in Victoria with a pre- hills inland of the more open Mitchell River 20,000 cal BP age (Bird et al. 1998:34, Table 1). floodplains, in a sheltered overhang overlooking Wangangarra 1 and other nearby rockshelters in the the river’s valley. They came back many times Mitchell River National Park have the potential to in subsequent years, using the shelter into provide some of the earliest and longest occupation recent times, lighting campfires, fishing in the sequences for East Gippsland and Victoria more river and wading or diving for shellfish. Initial generally, and with well-preserved faunal remains. results suggest that they repeatedly came to the At Wangangarra 1, initial sedimentation rates Mitchell River valley, sustained by its rich river- were slower in the older layers, increased after ine resources. SU2M following 5,810 cal BP, and increased further after 1,820 cal BP in SU2G and by a series of subse- This information allows the Gunaikurnai to quent interleaved hearth remains. These peak improve understandings of both the resources of the periods of sedimentation are all associated with Park and the cultural heritage that remains within. artefacts and structural evidence of occupation and In doing so, the Gunaikurnai Land and Waters may themselves be interpreted as sedimentary out- Aboriginal Corporation is better positioned to comes of increased occupation. It is of interest that review management measures for both all of the faunal remains definitely deposited by the Wangangarra 1 and other rockshelters and land- Old People came from the nearby riverine environ- scapes within and near the Park. It also gives ment. The implication is that the Mitchell River Gunaikurnai rangers critical new skills and informa- played an important role in the determination of tion to improve management of the Park and its resource scheduling and site location by the Old visitors, which is a major aspiration of Gunaikurnai People; these initial excavations at Wangangarra 1 Traditional Owners. The results of this work give have posed a new question of great interest for the Gunaikurnai scientific details by which to deter- Gunaikurnai Traditional Owners to now further mine, assess and act on community research inter- investigate, especially given strong traditional ests, priorities and aspirations. 196 G. L. ROBERTS ET AL.

Acknowledgements Attenbrow, V. 2010 Aboriginal fishing in Port Jackson, and the introduction of shell fish-hooks to coastal New The authors thank Patrick Mullett (RAP Crew Leader), South Wales, Australia. In D. Lunney, P. Hutchings Paula Martin (RAP Heritage Officer), Nicky Moffatt (RAP and D. Hochuli (eds), The Natural History of Sydney, Heritage Officer), Paul Harrison (RAP crew), Steve pp.16–34. Mosman: Royal Zoological Society of New Hood (RAP Crew), Chris Mongta (Natural Resource South Wales. Management Crew), Keily Ratzmann (Natural Resource Bilney, R.J., R. Cooke and J.G. White 2010 Management Crew), Max Solomon (Natural Resource Underestimated and severe: Small mammal decline Management Crew), Doug Harrison (Natural Resource from the forests of south-eastern Australia since Management Crew) and Bradley Hood (Natural Resource European settlement, as revealed by a top-order preda- Management Crew) for feedback and assistance during the tor. Biological Conservation 143(1):52–59. excavation. The authors also thank Katherine Mullett Bird, C.F.M., D. Frankel and N. van Waarden 1998 New (Aboriginal Heritage Coordinator, East Region Managing radiocarbon determinations from the Grampians- Country Together, Parks Victoria); Matt Holland (Regional Gariwerd region, western Victoria. Archaeology in Project Coordinator (Gunaikurnai), Eastern Region, Parks Oceania 33(1):31–36. Victoria); Aboriginal Victoria Heritage Project Officer Bray, D.J. and V.J. Thompson 2018 Australian bass, Emlyn Collins who collected the sediment samples from Macquaria novemaculeata (Steindachner 1866). Fishes the hill-slope outside the site; the Australian Research of Australia. Retrieved 14 February 2019 from < http:// Council Centre of Excellence for Australian Biodiversity fishesofaustralia.net.au/ >. and Heritage (project CE170100015) for funding; and the Bronk Ramsey, C. 1995 Radiocarbon calibration and ana- Monash Indigenous Studies Centre at Monash University lysis of stratigraphy: The OxCal program. Radiocarbon for support. The process and results of the field and – laboratory work and ranger training associated with this 37(2):425 430. Bronk Ramsey, C. 2009 Bayesian analysis of radiocarbon project were reported at a number of GLaWAC RAP – Elders Board Sub-Committee meetings in Bairnsdale and dates. Radiocarbon 51(1):337 360. Forestec. The authors thank Dr Doris Paton (GLaWAC Bronk Ramsey, C., M. Dee, S. Lee, T. Nakagawa and R.A. Staff 2010 Developments in the calibration and model- Board Member Director), Grattan Mullett (Cultural Hub – Manager), Daniel Miller (On Country General Manager), ling of radiocarbon dates. Radiocarbon 52(3):953 961. Campbell, A. and R.L. Vanderwal (eds) 1999 John Ruth Fitzclarence (Marketing and Communications ’ Manager) and Sophie Harvey for support during the course Bulmer s Recollections of Victorian Aboriginal Life. – of this research. 1855 1908. Melbourne: Museum Victoria. Clark, I.D. 2002 The ebb and flow of tourism at Lal Lal Falls, Victoria: A tourism history of a sacred Aboriginal site. Australian Aboriginal Studies 2002(2):45–53. Disclosure statement David, B., J.-J. Delannoy, R. Gunn, L.M. Brady, F. Petchey, J. Mialanes, E. Chalmin, J.-J. Geneste, I. No potential conflict of interest was reported by Moffat, K. Aplin and M. Katherine 2017 Determining the author(s). the age of paintings at JSARN-113/23, Jawoyn Country, central-western Arnhem Land plateau. In B. David, P.S. Tac¸on, J.-J. Delannoy and J.-J. Geneste (eds), The Funding Archaeology of Rock Art in Western Arnhem Land, – This work was supported by Centre of Excellence for Australia, pp.371 422. Terra Australis 47. Canberra: Biodiversity and Heritage, Australian Research Council ANU Press. (AU) [CE170100015]. David, B., J. Fresløv, R. Mullett, J.-J. Delannoy, F. Petchey, Gunaikurnai Land and Waters Aboriginal Corporation, J. Mialanes, L. Russell, R. Wood, L. Arnold, M. McDowell, J. Berthet, R. Fullagar, V.N.L. ORCID Wong, H. Green, C. Urwin, L. Metz, J. Crouch and J. Ash in press Paradigm shifts and ontological turns at Georgia L. Roberts http://orcid.org/0000-0003- Cloggs Cave, Gunaikurnai Country, Australia. In O. 2369-1118 Moro Abadıa and M. Porr (eds), Ontologies of Rock Bruno David http://orcid.org/0000-0002-8567-6135 Art: Images, Relational Approaches and Indigenous Jerome Mialanes http://orcid.org/0000-0002-6790-7542 Knowledge. London: Routledge. Fiona Petchey https://orcid.org/0000-0002-3308-9983 Driessen, M.M. 2008 Long-tailed mouse Pseudomys higginsi. Matthew McDowell https://orcid.org/0000-0001- In S. Van Dyck and R. Strahan (eds), The Mammals of 8009-4171 Australia, pp.638–639. Sydney: Reed New Holland. Vanessa N. L. Wong https://orcid.org/0000-0001- Easton, J.G. 1938 The geology of the Mitchell River 9490-3187 –  Gorge. Mining and Geological Journal 1:73 75. Katherine Szabo https://orcid.org/0000-0002-9542-6535 Fergusson, C.L. 2003 Ordovician–Silurian accretion tec- tonics of the Lachlan Fold belt, southeastern Australia. – References Australian Journal of Earth Sciences 50(4):475 490. Fernandez-Jalvo, Y., L. Tormo, P. Andrews and M.D. Angas, G.F. 1847 Savage Life and Scenes in Australia and Marin-Monfort 2018 Taphonomy of burnt bones from New Zealand: Being an Artist’s Impressions of Countries (). Quaternary and People at the Antipodes. New York: Reed. International 495:19–29. Atlas of Living Australia 2019. Retrieved 27 February Flood, J. 1974 Pleistocene man at Cloggs Cave: His tool from < https://www.ala.org.au/ >. kit and environment. Mankind 9(3):175–188. AUSTRALIAN ARCHAEOLOGY 197

Flood, J. 1980 The Moth Hunters: Aboriginal Prehistory of Ossa, P., B. Marshall and C. Webb 1995 New Guinea II the Australian Alps. Canberra: Australian Institute of cave: A Pleistocene site on the Snowy River, Victoria. Aboriginal Studies, Archaeology in Oceania 30(1):22–35. Fresløv, J. 2005 An Aboriginal Archaeological Heritage Pavey, C.R., J.R. Cole, P.J. McDonald and C.E. Nano 2014 Management Plan for the Dinner Plain Site Complex, Population dynamics and spatial ecology of a declining Victorian Alps. Unpublished report to Parks Victoria desert rodent, Pseudomys australis: The importance of and the Department of Sustainability and Environment. refuges for persistence. Journal of Mammalogy 95(3): Fresløv, J., P. Hughes and R. Mullett 2004 Post Wildfire 615–625. Indigenous Heritage Survey. Volume 1: Background, Prokop, F.B. 2006 Australian Fish Guide. Melbourne: Survey, Results and Recommended Management Options. Unpublished report to Parks Victoria, the Australian Fishing Network. Department of Sustainability and Environment, and Reimer, P.J., E. Bard, A. Bayliss, J.W. Beck, P.G. Aboriginal Affairs Victoria. Blackwell, C.B. Ramsey, C.E. Buck, H. Cheng, R.L. Harris, J.H. 1983 The Australian Bass Macquaria novema- Edwards, M. Friedrich, P.M. Grootes, T.P. Guilderson, culeata. Unpublished PhD Thesis., University of New H. Haflidason, I. Hajdas, C. Hatte, T.J. Heaton, D.L. South Wales, Sydney. Hoffmann, A.G. Hogg, K.A. Hughen, K.F. Kaiser, B. Harris, J.H. 1985 Age of Australian bass, Macquaria nove- Kromer, S.W. Manning, M. Niu, R.W. Reimer, D.A. maculeata (Perciformes: Percichthyidae), in the Sydney Richards, E.M. Scott, J.R. Southon, R.A. Staff, C.S.M. Basin. Marine and Freshwater Research 36(2):235. Turney and J. van der Plicht 2013 IntCal13 and mar- Hewitt, G. and J. Allen 2010 Site disturbance and archaeo- ine13 radiocarbon age calibration curves, 0–50,000 logicalintegrity:ThecaseofBendRoad,anopensitein years cal BP. Radiocarbon 55(4):1869–1887. Melbourne spanning pre-LGM Pleistocene to Late Richards, T., C. Pavlides, K. Walshe, H. Webber and R. – Holocene periods. Australian Archaeology 70(1):1 16. Johnston 2007 Box Gully: New evidence for Aboriginal Hua, Q., M. Barbetti and A.Z. Rakowski 2013 occupation of Australia south of the Atmospheric radiocarbon for the period 1950–2010. Radiocarbon 55(4):2059–2072. prior to the last glacial maximum. Archaeology in – Kemper, C., G. Medlin and M. Bachmann 2010 The dis- Oceania 42(1):1 11. covery and history of the heath mouse Pseudomys Smyth, R.B. 1878 The Aborigines of Victoria, with Notes shortridgei (Thomas, 1907) in South Australia. Relating to the Habits of the Natives of Other Parts of Transactions of the Royal Society of South Australia Australia and Tasmania. Melbourne: John Ferres, 134(1):125–138. Government Printer. Lyman, R.L. 2008 Quantitative Paleozoology. Cambridge: Stoessel, D.J., J.R. Morrongiello, T.A. Raadik, J. Lyon and Cambridge University Press. P. Fairbrother 2018 Is climate change driving recruit- Mackaness, G. 1941 George Augustus Robinson’s journey ment failure in Australian bass Macquaria novemacu- into south-eastern Australia, 1844. Journal of the Royal leata in southern latitudes of the species range? Marine – Australian Historical Society 27(5):318 349. and Freshwater Research 69(1):24–36. Massola, A. 1956 Australian fish hooks and their distribu- Wakefield, N.A. 1972 Palaeoecology of fossil mammal tion. Memoirs of the National Museum of Victoria assemblages from some Australian caves. Proceedings of – 22(3):1 17. the Royal Society of Victoria 85(1):1–26. McDowell, M.C. 2014 Holocene vertebrate fossils aid the Williams, N.J. 1970 A comparison of the two species of management and restoration of Australian ecosystems. the genus Percalates Ramsay and Ogilby (Percomorphi: Ecological Management & Restoration 15(1):58–63. McRae-Williams, M.S., N.J. Rosengren and S.M. Kraemers Macquariidae) and their taxonomy. NSW State 1981 Sites of Geological and Geomorphological Significance Fisheries Research Bulletin 11:60. in East Gippsland, Victoria.Melbourne:Ministryfor Xiao, X., Z. Chen and B. Chen 2016 H/C atomic ratio as Conservation. a smart linkage between pyrolytic temperatures, aro- Medlin, G.C. 2008 Long-eared mouse Pseudomys auritus. matic clusters and sorption properties of biochars In S. Van Dyck and R. Strahan (eds), The Mammals of derived from diverse precursory materials. Scientific Australia. pp. 615–616. Sydney: Reed New Holland. Reports 6(1):22644.