Australia: The Land Where Time Began

A biography of the Australian continent 

Aboriginal Coastal Landscapes at Cape Duquesne, Southwest Victoria, Australia, from the Early Holocene

A substantial contribution to the understanding of palaeoenvironmental change in Australia has been made by Peter Kershaw, in particular in connection with the timing of Aboriginal colonisation and alterations by Aboriginal people of the vegetation communities. The palaeoenvironment of landscapes in southwestern Victoria, with the emphasis on the palaeoecology of lakes and swamps, especially in regard to the appearance of water management by Aboriginal people, as well as systems for trapping fish on the Mt Eccles lava flow, and the relationship of these systems to socioeconomic complexity of Aboriginal groups in the southwest e.g. Kershaw, 2004; Tibby et al., 2006; Builth et al., Kershaw & Lewis, 2011). Though in this chapter Kershaw addresses similar issues of Aboriginal social complexity in southwest Victoria, Kershaw examines them from a nearby coastal landscape perspective.

The complexity of Aboriginal societies in southwestern Victoria has been controversial since the 1880s, the focus of disagreement being on the nature of leadership in the ethnographic period (e.g. Dawson, 1881, 1887; Curr, 1886; Howitt, 1887; 1904; Corris, 1968; Lourandos, 1977, 1980a, b, 1983, 1984, 1987, 1997; Barwick, 1984; Williams, 1985, 1987; Critchett, 1990, 1998; Edwards, 1987; Hiatt, 1996; Keen, 2006; Haydon, 2006), and a century later extending to debates regarding the socioeconomic intensification modelling of Lourandos for the Late Holocene (e.g. Lourandos, 1980a, 1983, 1985a, b, 1988, 1993, 1996, 1997; Beaton, 1983, 1985; McBryde, 1984; Williams, 1985, 1987, 1988; Godfrey, 1989; Bird & Frankel, 1991a, b, 1998, 2005; Lourandos & Ross, 1994; Bird et al., 1998, 1999; McNiven et al., 1999; David et al. 2006; Hiscock, 2008).

Societies that are thought to exhibit characteristics of cultural and social complexity, ‘complex’, ‘affluent’, or ‘transegalitarian’ forgers or hunter-gatherers that contrasts with the idealised view of egalitarian, hunter-gatherers that are highly mobile (Koyama & Thomas, 1981; Price & Brown, 1985; Haydon, 1985; Grier et al., 2006):

Transegalitarian societies are societies that are neither egalitarian nor politically   stratified; they are thus intermediate between generalised hunter-gatherers and chiefdoms in terms of the social and economic inequalities that characterise them. (Owens & Hayden, 1997: 121)

Though much of the debate with regard to transegalitarian features in the archaeological record of southwest Victoria has centred on water control and eel management infrastructure, and earth mounds by Aboriginal people a major source of contention in the region is Aboriginal marine shell middens, with Lourandos (1983, 1993, 1997: 224-227; Lourandos & Ross, 1994: 58-59) documenting the increasing use of, and establishment of, middens, from about 3,500 BP, as evidence that supports his intensification model, though this has been disputed by some other researchers (e.g. Godfrey, 1989; Bird & Frankel, 1991a, b; Hiscock, 2008: 190-191). Comparisons between the structure and contents of middens dating to the Late Holocene and earlier middens, as a result of the absence of data from Early Holocene middens in the region that have been formally excavated.

Crucial evidence for coastal occupation in this region has been provided by excavations on an Aboriginal landscape at Cape Duquesne and in this paper Richards documents the chronology, contents and structure of several middens and patterns of littoral resource exploitation. In this paper the baseline data set for the Early Holocene was compared with data from the Late Holocene as an additional means of evaluating possible changes in the Late Holocene in the use of coastal resources which was related to or reflected increasing complexity in regional societies.

The regional archaeological context and local environmental setting were reviewed briefly prior to the presentation of the Cape Duquesne Aboriginal landscape data.

Previous archaeology

Many researchers have focused on the southwest of Victoria and the adjacent southeast Victoria coastal region over the past 4 decades (e.g. Lourandos, 1976, 1980a, 1983, 1997; Witter, 1977; Clark, 1979; Godfrey, 1980, 1984, 1989, 1994, 1996, 2000; Godwin, 1980; Wesson & Clark, 1980; Simmons & Djekic, 1981; Head, 1985; Frankel, 1986, 1991; Cann et al., 1991; Webb, 1995; Richards & Jordan, 1996; Everett, 1998; Schell 2000a, b: Bird & Frankel, 2001; Debney & Cekalovik, 2001; Richards & Johnston, 2004; Richards & Webber, 2004). Yet there is excavation data for the Early Holocene limited to 2 sites that have been well researched - Bridgewater South Cav and Koongine Cave.

Bridgewater South Cave, which is 8 km to the north of Cape Duquesne, was excavated in the mid-1970s (Lourandos, 1976, 1980a, 1983, 1997). This site is comprised of stratified deposits both inside and outside a rockshelter of medium size that had excellent preservation of organic material, though it was disturbed superficially. The evidence of occupation dating to the Late Pleistocene to Early Holocene within stratigraphic Phase A, between 13,250 cal. BP and 9,350 cal. BP is of interest here (Lourandos, 1983: 83; Head, 1985: 5):

There is an over-riding emphasis on land mammals in this phase that consisted of a substantial proportion of macropods (which included the grey kangaroo) and wombat. There was correspondingly very little representation of marine foods. There was evidence of 1 seal and 1 fish, but there were scattered pieces of shell of species from both sandy beach and ocean rock platforms. There was a low frequency of flaked stone (Lourandos, 1980a: 348).

Occupation during Phase A has been characterised (Lourandos, 1980a: 349-350) as likely to have occurred during autumn-winter, evidence suggesting ‘…an ephemeral use of the site as a hunting bivouac...’ (Lourandos, 1997: 201-202).

At 11,000 cal. BP Bridgewater South Cave would have been about 3.75 km from the coastline at Discovery Bay, at 10,000 cal. BP about 3.0 km, and by 9,000 cal. BP only about 2.5 km, therefore within easy reach of the shoreline; it is apparent, however, that for some reason coastal resources were not exploited from this camp, apart from a high value resource such as seals.

Koongine Cave is for several reasons a twin of Bridgewater South Cave, situated at the opposite end of Discovery Bay, and about 85 km to the northwest (Bird & Frankel, 2001: 74). Also, it is a substantial limestone cave set on the edge of the coastal plain in a scarp, and there are significant occupation deposits dating to the Early Holocene, as well as a faunal assemblage that is well-preserved.

The early and middle phases of occupation, from about 11,000-9,000 cal. BP, were interpreted as representing repeated camping events during which the people were hunting a range of medium to small sized land mammals, such as possums, bandicoots, potoroo, wallabies and wombats, and large animals such as grey kangaroos (Bird & Frankel, 2001: 71). In this sense the period of occupation and the species in the faunal assemblage are remarkably similar to that which had been documented at Bridgewater South Cave (Lourandos, 1980a: Table 12:2, 1983: 83; Head, 1985: 5). The middle phase is essentially a version of the early phase though essentially less intensive, that is characterised by episodes of occupation that are shorter and less frequent (Bird & Frankel, 2001: 74).

Though in the early and middle phases the shoreline would have been 10-15 km away, a small amount of marine mollusc shell was recovered from Koongine Cave (Bird & Frankel, 2001: 73), as at Bridgewater South Cave. Based on the presence of emu eggshell it is inferred that occupation occurred in winter, though it is possible there was also additional seasonal usage (Bird & Frankel, 2001: 74) – which is another similarity to Bridgewater South Cave (Lourandos, 1980a: 352).

Spot samples of marine shell from deflated middens were recovered from Discovery Bay and Cape Bridgewater (Godfrey, 1989, 1994; Frankel, 1991) for radiocarbon dating, which produced several age determinations dating to the Early Holocene. This type of uncontrolled sampling, unfortunately, does not provide a reliable basis for characterising the composition of faunal assemblages, as at Cape Duchesne surface shell species proportions are typically not representative of nearby deposits as have been revealed by excavation (see surface vs. excavated shell species proportions for shell midden investigations reported in this chapter). The itinerant dating approach used by Godfrey & Frankel indicates that Aboriginal coastal occupation of this region probably occurred in the Early Holocene, though little is revealed about the nature of this occupation.

Landscape description and environment

The area of the study is along the top of steep cliffs at Cape Duquesne, which is a headland on the southern tip of Cape Bridgewater on the Portland Peninsula. This is a high-energy coast that is dominated by swell waves and winds coming from the west through south-southwest (Short, 1988: 125; Buckley, 1992: 13).

Geomorphology, geology and soils

The Portland Peninsula is a large promontory that juts out southwards into the Southern Ocean, and is tipped by a sequence of protruding headlands and indented bays. The westernmost of these headlands is Cape Duquesne which is bounded by Discovery Bay, which is extensive and is northwest-southwest trending, to the west, and Bridgewater Bay to the east which is much smaller and protected. Descartes Bay is the juncture of Cape Duquesne and Discovery Bay.

There are 3 major geological formations exposed on cliff faces and on the surface that dominate the coastal geomorphology – basalt and tuffs of the Newer Volcanic Formation dating to the Pliocene-Pleistocene, which is overlain by beach and dune calcarenites of the Bridgewater Formation dating to the Pleistocene, and capped by dune sands dating to the terminal Pleistocene-Holocene (Boutakoff, 1963; Bird, 1993: 24). Much of this sequence is exposed in cliffs, which are 25-120 m high, which extend from Cape Duquesne and Cape Bridgewater headlands to the south, and east and northeast into Bridgewater Bay. Mainly sheer basalt cliffs comprise the shoreline of the present, with an occasional sandy beach, such as White’s Beach to the northwest, narrow shingle, boulder talus or sea caves, such as Seal Cave to the east. Also, there are frequent off-shore intertidal platforms of basalt.

Bridgewater Formation was defined (Boutakoff, 1963: 48-51), as consisting of a series of sand dunes that have been lithified (‘limestone dunes’), that were formed on (?of) sand derived from the westering of dunes dating to the Tertiary that were exposed by regressing seas in the Pleistocene glacial periods, also including palaeosols that capped the calcarenites dunes. These Tertiary dunes, classed as rendzina, terra rossa and laterite fossil soils, were considered to have formed after the underlying dunes were calcified during pluvial conditions of interstadial/interglacial periods. They were described as calcareous, sandy and red, reddish-brown or reddish-pink soils (Boutakoff, 1963: 48-51). Boutakoff’s sequence was confirmed by subsequent research in the region (e.g. Kenley, 1976; Douglas, 1979; Land Conservation Council, 1981), until they were upgraded to the Bridgewater Group (Cupper et al., 2003: 343-344) and its distribution was expanded more broadly across southwestern Victoria.


In the terminal Pleistocene a decline in effective moisture, which was largely due to rising temperatures, culminated in a period of maximum aridity from about 17,000-14,000 cal. BP, when woody plants were not common (Kershaw et al., 2004: 158). Increases in temperature and rainfall in the succeeding period, from about 14,000 cal. BP to 11,500 cal. BP resulted in an expansion of the distribution of trees, and this was accompanied by a change from steppe grassland to grassland (Kershaw et al., 2004: 158). Tree cover had reached the levels it was at in the early 19th century prior to clearing by Europeans by the beginning of the Holocene, 11,500 cal. BP (Kershaw et al., 2004: 158). Change continued in the community composition of the vegetation, and there was a sustained increase in Eucalyptus relative to Casuarinacae trees around 8,900-7,800 cal. BP, with the result that there was essentially the vegetation cover that was present in the early 19th century, dry sclerophyll forest/woodland, that was established by the end of this period (Kershaw et al., 2004: 139, 158-159).  

Cape Duquesne investigations

A series of open-air Aboriginal shell midden deposits were discovered on the top of a cliff that was 50 m high by investigations at Cape Duquesne, and at present the waves wash against the base of this cliff. Vast areas of carbonate sands were exposed on the continental shelf along the coast of southern Australia in the Late Pleistocene, as previously noted, and as the high energy sea transgressed and the prevailing winds were strong southwesterlies much of this sand was mobilised (Short, 1088: 121). These sands that were mainly sourced locally from the seafloor that had been exposed piled up against obstructions such as the cliffs at Cape Duquesne, and at times even overtopped these cliffs (Short, 1088: 138). From the terminal Pleistocene to the Early Holocene, the ramp would have extended from the seabed that was exposed near the shoreline up and over the top of the cliffs at Cape Duquesne, which would have provided easy access between the shoreline and the campsite. The base of the ramp would have been eroded later as the sea transgressed, and it would have been removed entirely as the sea level of the present was attained, which would have left the clifftop dunes stranded as the last remnant of the former ramp (Short, 1088: 138-139; Rosengren, 2001a: 215).

The landscape of Cape Duquesne has been shaped significantly by wind erosion, with the result that the surface of the present is a patchwork of landforms and sediments dating to the Pleistocene and Holocene that are exposed differentially. Along the southern edge of the landscape (the clifftop) there are extensive exposures of dune deposits that have been lithified (calcarenites) in the centre and west of the landscape, where all overlying sediment has been scoured away by the action of the wind. These are calcarenites dating to the Pleistocene of the Bridgewater Group.

An exposed palaeosols extend approximately east-west across the landscape, which is comprised of greyish sandy loam, most of which is similar to the Bridgewater Sandy Loam, a rendzina soil (Gibbons & Downes, 1964: Appendix 1). The calcarenites is directly overlain by the sandy loam, which varies from patches that are mere remnants a few centimetres thick filling hollows on calcarenites surfaces that are uneven, to deposits that are more than 1 m thick. Calcarenites of the Bridgewater Group commonly have such palaeosols associated with them, and the palaeosols are considered to be part of the unit. Palaeosols are again active soils in the erosional situation of the present, though they are generally vegetated very poorly.

A large dune which for the most part is actively eroding and with little vegetation cover, dominates the western end of the landscape, though in the north it is vegetated and stable. It  is indicated that more than 1 m of sand has been removed from the southern surface of this dune as a result of wind erosion, by remnant sediment pedestals and exposed carbonate root casts. This dune is a remnant of the clifftop dunes that had been deposited since the Last Glacial Maximum (LGM), and much of the greyish-brown sandy surface that is eroding dates to at least the Early Holocene. It overlies the sandy loam palaeosols and the calcarenites of the Bridgewater Group.

All along the landscape to the edge of the cliff the vegetated area 50-150 m to the north is comprised of the remaining stabilised dune field. There are sandsheets to the south of the dune field in patches on the landscape, which represent clifftop dunes that have been destabilised and reworked by the action pf the wind in the Holocene.

Archaeological survey and surface record

A detailed systematic archaeological survey was carried out in an area where much of the area had been deflated to some extent, which extended 1 km along the clifftops and up to 400 m inland. The identification of the Aboriginal landscape was the goal of the survey, which focused on the distribution of flaked and ground stone artefacts, shell middens, hearth features, as well as other evidence of human occupation. The gross extent of the distribution of Aboriginal cultural material that was exposed on the surface, as recovered by the initial pedestrian survey – 58 hearth features, 8 shell middens, and thousands of flaked stone artefacts recovered from an area of about 500 m east-west and 125 m north-south (62,500 m2). As the landscape was mapped, hearth features and shell middens were recorded in detail, 3 of the hearth features that were exposed were excavated and 8 shell midden deposits were tested. Of the large number of lithic artefacts that were observed on the surface were in situ, so they were generally not mapped individually, though samples of stone artefacts, animal bones, marine shell, hearth stones charcoal and sediments were mapped and collected for identification and analysis.

Features such as clifftop dunes and sandy loam palaeosols that capped calcarenites dating to the Late Pleistocene and Early Holocene are the only places where in situ shell midden deposits are located, though it was found that stone artefacts and deflated hearth features are distributed more widely across the landscape, which suggests that they derived mostly from more recent deposits that have largely eroded away. It was indicated by the technology of stone artefacts and radiocarbon age determinations of 2 remnant hearth features that there was substantial occupation across this landscape in the Late Holocene, but the study reported here focussed on shell midden deposits of the Early-Holocene hearth features and shell middens.

Obviously loose shells that were not in their primary context were exposed on the surface were displayed on surface exposures on middens, though it appeared that other shells were eroding out of the surface as a result of the removal of surrounding soil particles by wind, and appeared to be in situ. Shell Midden A (SM A) was the first to be excavated in order to determine if in situ deposits were indeed present, to identify subsurface structure and the characteristics of the midden, as well as to obtain samples for identification, analysis and chronometric dating.

Shell Midden A – excavation

SM A is located at the northwestern corner of the landscape. On the surface it is observed as dense concentration of Turbo undulata and Cellana tramoserica shells and a small amount of charcoal that has been exposed by wind erosion.  Some shells were loose, though there appeared to be an in situ shell deposit that was being eroded out of sediments below the loose shell and sand that had been blown in recently. The midden area that was exposed on the surface of the ground was approximately 3 m in diameter; however, it extended beneath dune deposits that were partially consolidated to the northwest and is believed to possibly have a much larger area than the exposed area that was being eroded.


A very dense deposit of whole and fragmented marine shellfish shell was revealed by excavation that extended across the square, with associated chunks of charcoal that was coated with carbonate as well as flaked stone, to an average depth below the surface of 4 cm. For another 2 cm small amounts of shell and charcoal continued in the unconsolidated sand. A patch of burnt sediment and charcoal with an associated large flat rock (manuport) was uncovered below the base of the midden in the northeast of the square and extended 9 cm below the surface. Stratigraphically, this feature is earlier than SM A.

There was unconsolidated sediment in the midden deposit, dark greyish-brown (10YR4/2 wet) to light grey (10YR7/2 dry) calcareous sand that contained clods of carbonate and sediment aggregates, as well as land snails and rootlets that provided a non-cultural minor organic component (Johnston, 1996). The values of soil pH varied from 8 (field), which was a highly favourable environment for the preservation of bone, to (lab), which is less favourable for bone preservation (Reitz & Wing, 2008: 140-141). It was revealed by detailed analysis that the sand was medium to fine grained (Wentworth, 1933), and was composed of 45 % carbonate and 15 % quartz. It was indicated that this was an Aeolian deposit that originated from beach sands, based on the negative skewness in the distribution of the particle size, grain shape and surface texture, as well as the nature of the deposit which was moderate to well-sorted. The sediment below the midden appeared virtually identically in the field, though slightly darker (10YR3/2).

It is suggested by the presence of the loose surface in only a small area that is directly above in situ deposits that the top of the midden was exposed only a short time before the investigations and that nor much had been lost to the action of the wind, i.e. there was no downwind trail of smaller items.

Marine shellfish

The Minimum Number of Individuals (MNI) identified to species in SM A comprised 650 individuals from 8 marine mollusc taxa that have been identified in more than 7.5 kg of shell that was recovered. Turbo undulata is the most common shellfish, at a bit less than 70 % of the total MNI, with 18 % being contributed by Cellana tramoserica, Austrocochlea concemerata 9 %, as well as 5 other taxa each of which represents 1 % or less.

Radiocarbon dating

There are 4 radiocarbon age determinations that are pertinent to the dating of SM A, 3 of which were on Turbo undulata shell from the midden itself, and 1 on wood charcoal that was recovered below the midden. Initially an age determination of about 10,300 cal. BP (Beta-93569) was given by XU 1 and another from XU 2 dated to about 9,850 cal. BP (Beta-96584). Though they overlap at 2 standard deviations, the age determination for XU 1 was slightly older than that from XU 2. A second age determination on a shell was carried out for XU 2, about 10,500 cal. BP (UB-4369), the date that was obtained was older than both dates that were previously determined, and in sequence.

A radiocarbon determination to the terminal Pleistocene at about 11,800 cal. BP (UB-4370), resulted from a further determinations on wood charcoal from the layer beneath the midden deposit that was rich in charcoal, which indicated that there was a distinct occupation layer that pre-dated formation of the midden.

Other excavations and investigations

Test pits were excavated at 6 other shell scatters on the surface, after the excavation of SM A, where it was evident there were in situ midden deposits. These test pits were excavated in order to evaluate whether there were also intact midden deposits, and if there were midden deposits to characterise the nature of any midden layers that were present, and to obtain samples of any shell or other materials so they could be identified and dated. Similar methods of excavation and recording were used for these test pits as were employed at SM A.

Shell Midden B

SM B was located 140 m east of SM A, in a flat expanse of dark-grey sandy loam that was highly consolidated and almost free of vegetation. It appeared as though shells of Turbo undulata and flint flakes were eroding out of the sandy loam palaeosols in an area of 7.0 m east-west by 6.1 m north-south.

A single 25 cm square test pit was excavated at SM B at the approximate centre of the midden that was exposed in an area where there was a high density of shell on the surface. The uppermost 3 cm of the first XU contained almost all the shell, though there was a small amount that extended to a depth of 7 m in the second XU.

In the 70 g of shell that was recovered 2 species were identified. The assemblage was dominated by Turbo undulata, at 54 % of the MNI, and 46 % was contributed by another species, Austromytilus rostratus, inhabiting rock platforms. A sample of fragments of Turbo undulata from XU 1 was radiocarbon dated to about 10,150 cal. BP (Beta-93567) and another in a shell of Turbo undulata from XU 2 was radiocarbon dated to about 10,000 cal. BP (Wk-20816) by AMS. Statistically, the ages of the samples are the same at the 95 5 level (T statistic) (Stuiver at al., 2011).

Shell midden D

SM D is located in the south-central portion of the landscape, in an area that is flat and sparsely vegetated of exposed dark grey palaeosols about 50 m from the edge of the cliff. Scattered on the surface were flint artefacts and Donax deltoides, Cellana tramoserica, Turbo undulata and Polyplacophora shells, and some Donax deltoides appeared to be in situ and in the process of eroding out of the sediment over an area of 13.6 m east-west by 8.9 m north-south within an area of dark grey sediment that was much larger.

At SM D a 25 cm square was tested in the centre of the exposed deposit in a concentration of surface shells. It was revealed by excavating a single 5 cm XU in dark grey, consolidated dense sandy loam in situ midden deposits extended from the surface to a depth of 2-3 cm, and the termination was sharply defined.

199 g of shell was the total recovered from the SM D assemblage, with a trace amount of charcoal. Of the MNI Donax deltoides comprises 95 %, with 5 % representing Turbo undulata. A date of about 9,100 cal. BP (Beta-93566) was obtained from a sample of whole Donax deltoides shells.

Shell Midden E

Around the periphery of a large blowout situated towards the northeastern part of the landscape ancient sediments were exposed that were up to 1 m deep. Cultural material that comprised SM E was exposed on the eastern side of the blowout that was steep sided, as well as on the surface beyond the blowout. In situ items in section and partially exposed on the eroding surface (not the blowout) included shell, notably Ostrea angasi, flint artefacts that included a flake core, and charcoal, as well as a hearth feature. SM E has an area of 750 m2, which makes it the largest midden on the landscape.

A 25 cm square test pit was excavated on the eastern edge of the eroded area where in situ shell, charcoal and flint artefacts were observed in section as well as on the surface. The excavation of more than 1 XU was not possible because of time constraints, and excavation of this unit revealed a midden that was moderately dense to a depth of at least of 6 cm (which was supported by material that was visible in the eroded section to a similar depth). As the sediments were cemented by calcium carbonate excavation was very difficult within the first few centimetres. A consolidated light grey sandy loam was present beneath this crust.

Turbo undulata, which comprised 70% of the MNI, dominated the excavated 60 g sample. There were also 3 other species, Cellana tramoserica, Austromytilus rostratus and Ostrea angasi, which each were represented by 10 % of the MNI. A radiocarbon age determination of about 10,250 cal. BP (Beta-93568) was provided by a sample of Ostrea and Turbo undulata.

Feature 50

Feature 50 (F 50) is 10 m to the northeast of the blowout on SM E, is a cluster of burnt calcarenite cobbles 4.6 m east-west by 4.2 m north-south, some of which are in situ and some on the surface of the greyish brown compact sandy loam surface, following the removal of the soil by the action of the wind. These are the only hearth stones that have been observed in situ that date to the Early Holocene sediments on the Cape Duquesne landscape. Among the hearth stones were several Ostrea angasi shells that included 2 shells that were in situ, each of which was about 50 % exposed on the surface. A single Ostrea shell was removed to be radiocarbon dated which proved to be about 10,700 cal. BP (Wk-9563).

Shell Midden F

SM F is a surface exposure of eroding shells on the surface that is largely unvegetated that measured 18.8 m east-west by 14.6 m north-south. Densities of Cellana tramoserica, turbo undulata, Thais orbita, and Donax deltoides shell were exposed by a single pass of the plough that the land management authority used to create furrows for replanting to stabilise the area.

A 40 cm square was placed over an average surface expression of Cellana tramoserica and Turbo undulata shell located between the furrows. The loose shells, some of which were nearly whole and others fragmented, were swept up and bagged prior to the excavation of XU 1.

Excavation showed that the shells of Cellana tramoserica and Turbo undulata were very dense, comparable to SM A and SM D. In the northwest corner, and also in the southwest, but deeper, a few small pieces of dubiously cultural charcoal were found near the surface. To remove the sediment, that was tightly packed; from around the shells pointing trowels and dental picks were used to remove them without damage. The first XU was continued to the bottom of the dense concentration of shell, which extended to a depth of 5 cm below the surface.

In order to establish the depth of sterile deposits XU 2 was excavated. Small shell fragments were found which according to Thomas were clearly from the upper 1 cm of the 5 cm deep XU. There were small pieces if charcoal scattered throughout the XU.

The surface consisted of a greyish-brown (10 YR 5/2) fine sand that almost completely lacked vegetation cover. In both excavated XUs the sediment was identical to that of the surface, with the exception of being lightly consolidated in comparison with the highly consolidated surface crust.

SM F, the only midden that was dominated by Cellana tramoserica; out of 1,049 g of sell, Cellana tramoserica comprised 58 % of the total MNI, Turbo undulata 29 %, Thais orbita 7 %, Polyplacophora 3 %, and Austromytilus rostratus <1 %.

An age estimate by radiocarbon dating for whole Cellana tramoserica from XU 1 of about 10,150 cal. BP (WK-9604) was obtained. A  Turbo undulata shell from XU 2 gave a date of about 10,400 cal. BP (Wk-29818). At the 95 % level (T statistic) the samples are significantly different (Stuiver et al., 2011).

Shell Midden G

SM G in the northeast of the landscape, was eroded by wind action and was largely free of vegetation. Over an area of about 16 m in diameter there were flint artefacts, Cellana tramoserica, Turbo undulata and Donax deltoides shell, as well as animal bones that were fragmented exposed on the surface. A 40 cm square test pit was used to investigate SM G in an area of average shell density. The surface sediments were loose, light grey (10YR 5/2) sand. The sediments in XUs 1 and 2 were lightly consolidated, light grey (10YR 5/2) fine sand with a few rootlets. The addition of small calcarenite pebbles in the sand was the only change in the sediment of XU 3.

The upper few centimetres of the first XU contained a concentration of Turbo undulata shell, though very little was found in the 2 XUs below that. Turbo undulata shell comprised 100 % of the shellfish MNI in a sample of 132 g of shell that was recovered.

A radiocarbon determination of about 10,050 cal. BP (WK-9564) was provided by large Turbo undulata fragments from XU 1.  An AMS age of determination of about 9,800 cal. BP (Wk-29817) was yielded by a single Turbo operculum from XU 2. At the 95 % level (T statistic) the samples differ significantly (Stuiver et al., 2011), though they overlap at 2 s.d.

Shell Midden I

SM I. that is located at the eastern end of the landscape, has a diameter of 3.2 m and is comprised of a cluster of Donax deltoides shell and a few flint artefacts lying on the surface. This midden was not excavated, though a sample of Donax deltoides which consisted of some in situ shells, though mostly surface shells, from an area of about 1 m in diameter. Radiocarbon ageing estimated an age of about 8,750 cal. BP (Wk-9,562) on whole Donax deltoides shell.

Shell Midden J

SM J is located in the south-central part of the landscape, and measures 13.5 m north-south, by 9.1 m east-west on an exposure of sandy loam sediment that was much larger and almost unvegetated on which there were many surface cracks. SM J had 1 dense in situ exposure of mainly Turbo undulata shell that covered an area of about 5 m2, and scattered Turbo undulata, Cellana tramoserica, Donax deltoides and Austromytilus rostratus shell that were evident over a larger area. It was noted that there were 2 Austromytilus rostratus shells on the surface of the southern edge of the midden. Some shells of this species were recovered from this area, as there had been only a small amount shell of this species in other middens, for basic subsistence information as well as for radiocarbon dating. On the western edge of the midden a 40 x 50 cm square was set out so that it included the Austromytilus rostratus shells that were partially exposed.

The Austromytilus rostratus shells were in a crumbly condition and it was only the highly consolidated sediment that held them together, so as they were exposed gently with pointing trowels they fell apart. Excavation proved to be generally very difficult, progressing very slowly in spite of the tools which were highly sharpened. In the first XU very little additional shell was found, and the small amount that was found was fragmentary and friable. Near the bottom of the level there was almost no shell apparent, and the last 1 cm of which was carefully completed by shovel shaving. The second 5 cm-thick XU was excavated by the cautions use of hand picks. There was very little shell found, and what there was there, was fragmented. Excavations were carried out on a third XU, but only in the southeast quadrant (a 25 cm square). There was only a tiny shell fragment and a few pieces of charcoal present.

The surface of the square, which was a greyish-brown (10 YR5/2) sandy loam that was highly consolidated, had no vegetation cover. Cracks that were visible continued down to XU 2. In XU 3 the sediment is a dark brown (10 YR 3/3), consolidated sandy loam with small, and rounded calcarenite pebbles.

Only a small amount of shell was recovered from the test excavation in SM J, with a total of 13 g being collected. Austromytilus rostratus was the only species that was represented, and the MNI was 5. In each of the excavated units there was a trace of charcoal, as well as a few grams of red ochre in XU 1 and in XU 2 there was a tiny amount.

An AMS radiocarbon age determination was obtained on Austromytilus rostratus shell from XU 1 of about 10,950 cal. BP (Wk-9532). This is the earliest age that has been determined on shell from this landscape. 


According to Richards there are 3 natural phenomena that allowed the formation and subsequently the preservation of the Aboriginal landscape at Cape Duquesne:

a)      A cliff; 

b)      A sea bed that slopes steeply; and

c)      A sand ramp.

The sand ramp joined the top of the cliff to the shore line of the Early Holocene, thereby providing an environmental setting that the Aboriginal people turned into their own landscape, which allowed them to collect marine littoral resources from the Southern Ocean and transport them to their camping places at the top of Cape Duquesne where they were processed and eaten as a result of the steep slope of the seabed, At 10,000 cal. BP the clifftop location would have been a bit over 1 km to the north of the shoreline and about 75 m above it, which provided a convenient camping location that had a commanding view over an extensive coastal plain.

This is not an unusual occurrence of a single early midden; rather, it is indicated by the evidence from the landscape of Cape Duquesne that coastal occupation was a regular feature of the pattern of settlement-subsistence in southwest Victoria from the onset of the Holocene, at least.

The archaeological research at Cape Duquesne has contributed to the study of socio-complexity of southwest Victoria as it has provided a baseline dataset for comparison with Aboriginal use of the landscape in the Late Holocene, especially:

1)      The middens‘ physical nature, such as size, composition and structure, which indicates that individual  patches of midden ranged up to 750 m2 in area, and contained 11 taxa of shellfish, stone artefacts and hearth features.

2)      A firm chronology, which is based on 15 radiocarbon dates, demonstrates repeated occupations of marine shell middens of the area over a maximum span of 2,500 years (about 11,100 cal. BP to about 8,600 cal. BP) with a focus on the period about 10,500 cal. BP.

3)      Statistics that have been derived from measurements of middens and chronology, such as annualised rate of deposition, estimates up to 0.06 m3 per year.

4)      The character of the exploitation of littoral marine resources had a focus that was highly patterned, on 3 species that were available on rock platforms from the high tidal zone to the low tidal zone. There were other species of shellfish that were taken opportunistically as they were found at these same locations. The people focused on shellfish that were relatively K-selected that were collected individually and processed by hand prior to eating.

5)      The character of landscape by humans; this was on a small scale and was probably seasonal, which indicated that reliance of shellfish as a food source by small groups of people was short-term.

A nearby Aboriginal landscape at Cape Bridgwater dating to the Late Holocene, in marked contrast, was occupied about 5,000 years after Cape Duquesne, contained a very large midden deposit that showed evidence of a concentration on mass collection and processing of a very small, sessile shellfish species that was present in very large beds in the high-tidal zone. There was a Late Holocene annualised midden with a deposition rate estimated to 110 times greater than at Cape Duquesne in the Early Holocene. The midden at Cape Bridgewater is structurally different, as it was much larger and thicker and had huge areas of continuous midden deposit, while the midden deposit at Cape Duquesne is discontinuous, patchy and thinner.

It is suggested by Richards that it must be concluded that there is no indication of complexity in the archaeological record at Cape Duquesne dating to the Early Holocene; the evidence provides a textbook signature for generalised, hunter-gatherers that are highly mobile. This finding is consistent with other Early Holocene occupations in southwest Victoria that have been documented, such as Bridgewater South Cave, Koongine Cave, Blackfellow’s Waterhole, Billimina Rockshelter and Drual Rockshelter. It is also very clear that in terms of coastal occupation and use of a littoral resource by Aboriginal people in this area in the Late Holocene compared to the Early Holocene, something that was very different was happening. The occupations were more highly organised in the Late Holocene, and were probably scheduled more tightly, by larger groups that remained on the coast for longer periods, as well as utilising a broader range of species while focusing increasingly over time on the mass collection and processing of a single r-selected species. According to Richards this pattern is entirely consistent with the appearance in the Late Holocene of semi-sedentary, high population density, complex, transegalitarian societies, as had been previously proposed (Lourandos, 1980a, b, 1983, 1987) and (Williams, 1985, 1987, 1088).

It is also consistent with the results of Kershaw et al. on the lava flow at Mt Eccles.

Sources & Further reading

Richards, Thomas, in Haberle & David eds., 2012, Peopled Landscapes: Aboriginal and Biographic Approaches to Landscapes, ANU E Press, The Australian National University


Author: M. H. Monroe
Last updated:
Journey Back Through Time
Experience Australia
Aboriginal Australia
National Parks
Photo Galleries
Site Map
                                                                                           Author: M.H.Monroe  Email:     Sources & Further reading