The ancestral homeland of the Gunditjmara people is the Western District of Victoria, taking in the southwest corner of the State. The Gunditjmara were there when the Hentys arrived and imposed their whaling station in the 1830s and the Gunditjmara continue to have a strong social, cultural, and land management presence in the region. As with all Aboriginal Australians, land management and cultural heritage are intertwined and inseparable. A key dimension of land management is conservation of ancestral waterways and wetlands. Freshwater not only nourishes Gunditjmara country, it is the ancestral source of eels which play an important role in Gunditjmara subsistence and identity. The interweaving of Gunditjmara land and waterway management, cultural heritage, eeling and identity is exemplified by the Budj Bim National Heritage Landscape and the Tyrendarra Indigenous Protected Area (Fig. 1). Another Gunditjmara community initiative, the Lake Condah Sustainable Development Project, aims to develop sustainable use and management of cultural and environmental values of the region in the context of World Heritage listing. One aspect of the Project is construction of a concrete weir across a European drainage channel to help restore pre-European water regimes and aquatic habitats to Lake Condah and to reactivate the extensive eel aquaculture system (see Special Box Section).16

Gunditjmara are eel fishers and eel farmers – a tradition passed down from their Elders and enshrined across their wetland landscapes in the form of ancient and elaborate stone-walled fish trapping and aquaculture facilities. This fishery is part of oral history with many families having many stories of fishing from the region's waterways. The fishery fed families and provided an income. Grandfathers always had a bag of eels to sell at pubs in the district. As in the past, today's fishery includes other fish such as Toupon, Blackfish and Mountain Trout.17 Fish are caught in wire mesh traps, either box traps or funnel-shaped traps, and fyke nets similar in form to traditional woven basket traps (Figs. 2 to 4). As in the past, fishing involves detailed knowledge of the seasonal cycle of water flows and fish activities and movements.

Nineteenth century historical records provide tantalising snippets of information on Aboriginal fishing practices of the Western District. For many areas with waterways emptying into the sea, eels were the focus of fishing and a staple protein source. For closed river systems without access to the sea and migrating eels, such as the Corangamite basin, tiny galaxiids were the target fish.19 In addition, James Dawson pointed out that 'there are many varieties of fish in the lakes and rivers, which are eaten by the natives'.20

The two key historical sources on the Gunditjmara fishery are Robinson and Dawson.21 George Augustus Robinson was Chief Protector of Aborigines, Port Phillip Aboriginal Protectorate, between 1839 and 1849. During autumn and winter (March to August) 1841, Robinson travelled around the Western District of Victoria and gained unique insights into Aboriginal culture through first-hand observations and interviews with Aboriginal people and European settlers.22 Robinson's journal recordings are augmented and supported by the subsequent recordings of James Dawson who lived on Kangatong station north of Port Fairy during the 1840s-1860s. Based on close friendships with local Aboriginal people, he became fluent in the Gunditjmara language and his considerable knowledge of Gunditjmara culture was published in 1881 in his Australian Aborigines. Together, Robinson and Dawson detail differing scales of fishing practices employed by the Gunditjmara and neighbouring groups. These practices ranged from small groups spearing fish in lagoons to large-scale engineered facilities with hundreds of metres of constructed embankments, stone walls, and excavated channels. Fundamentally, these records provide archaeologists with glimpses of the potential scale and complexity of the fishery (at least at the time of European contact) and alert researchers to the possible range of archaeological sites that may exist across the landscape today. Equally important, these records invite researchers to investigate the history and complexity of different forms of social organisation required to coordinate operation of large-scale facilities which in some cases involved hundreds and possibly thousands of people. Noel Butlin's estimate of 30 – 35,000 as the immediate pre-contact Aboriginal population of the Western District indicates the potential number of people involved in the fishery.23 On a more basic level, historical references indicate that the Gunditjmara fishery was as much about water management and control as it was about capture of fish.

Small-scale fishing involved individuals and small groups (often families) employing baited lines (without hooks), spears, baskets and weirs. For example, people would proceed to shallow lakes, ponds or marshes to capture eels using special paired eel spears (Fig. 5). During summer eels were dug out of soft mud from the dried beds of swamps.24

Near Terang in April 1841, Robinson noted: 'Saw a dredge for catching small fish. The manufacture same as that employed in making baskets but the shape like a canoe. It was 5 feet long and 18 inches deep the bottom was sharp and worked from a stick and called by the Tjarcote natives, 1. neer.ry.ger, 2. mole' (Fig. 6).25 Dawson also recorded that 'Fishing baskets, about eight or ten feet long, made of rushes in the form of a dragnet, are drawn through the water by two persons'.26 In the 1840s, at a village located south of Caramut, Aboriginal people 'had a superior kind of fishing nets, made like a large round bag with a hoop on the top and a pole attached to it'.27

Dawson describes 'night fishing in deep waterholes' involving the construction of a small 'stage ... formed of limbs of trees, grass, and earth, projecting three or four feet from the bank, and close to the surface of the water'. Using an ignited torch to attract fish, the 'fisher, lying on his face, spies the fish through a hole in the middle of the stage, and either spears or catches them with his hand'.28 These fishing stages appear to be the same as the six 'places the natives made for fishing ... supported by folk sticks' observed by Robinson around the edge of a lagoon along the 'Port Fairy river' (Moyne River) in 1841 (Fig. 7).29

Lourandos points out that spears and nets were used to catch eels during the drier months (especially summer). In contrast, during the wetter months when waterways swell triggering eels to migrate downstream, weir trapping devices were employed.30 Dawson noted that eels were 'prized by the aborigines' and were 'captured in great numbers by building stone barriers across rapid streams, and diverting the current through an opening into a funnel-mouthed basket pipe, three or four feet long, two inches in diameter, and closed at the lower end'.31 It is likely that hundreds of small stone fish traps were used by family groups across the Western District immediately prior to European invasion. In other situations, stone weirs were built at numerous points along waterways. According to Dawson, the celebrated example of such a weir complex was Salt Creek which runs out of Lake Bolac. During autumn rains when many eels exit the lake and swim towards the sea, numerous 'tribes' converge on the creek and at specially allotted locations, individual families set up traps. 'For a month or two the banks of the Salt Creek presented the appearance of a village all the way from Tuureen Tuureen, the outlet of the lake, to its junction with the Hopkins' (a distance of over 40 kilometres).32 Robinson reported that at 'Lake Boloke [Bolac], during the eeling season, from eight hundred to one thousand natives at one time have been seen'.33 With so many people, the total haul of eels must have numbered in the tens of thousands and weighed many tonnes.

Eel fishing at Salt Creek reveals how small-scale, family-sized fishing facilities such as individual weirs across narrow waterways could be brought together into large-scale trapping complexes to take advantage of major runs of fish and to feed numerous groups. Clearly, with so many groups operating weirs, controlling water flows, and removing eels, considerable coordination and negotiations would be required to ensure that participants downstream had access to eels. In this connection, Lake Bolac and Salt Creek were owned by a single group (the 'Boloke tribe') who had regionally recognised 'exclusive rights to the fish' and no doubt to control and manage groups who visited to use the resource.34

Robinson and Dawson also recorded another form of large-scale trapping complex that operated as huge integrated trapping systems. These integrated trapping systems operated at landscape scales and required significant investments of labour to build, maintain and operate. Such investments necessitated negotiated cooperation between numerous people and in many cases numerous clans and possibly tribal groups. Two major types of large-scale fishing facilities were documented by Robinson and Dawson – first, stake and branch or stone weirs over 100m in length and spanning waterways, and second, systems of excavated channels sometimes thousands of metres in length and covering many hectares. For example, on 24 April 1841, near present day Hexham, Robinson noted:

Crossed a creek connecting with the Hopkins. Here I observed a large were [weir] at least 100 yards in length and though the first I had seen, I was assured by its structure and its situation before I reached it that it was the work of the Aboriginal natives. I called to Pevay, my V[an] D[ieman's] L[and] attendant, who had passed it with one of the Tcharcote natives and pointed it out to him. He evinced surprise of it and the natives said it was made by black fellows for catching eels when the big water came and was by them called yere.roc. He said they got plenty eels and then showed us how they did it by biting their heads and throwing them on shore. This were [weir] was made of stout sticks, from 2 – 3 inches thick drove in to the ground and vertically fixed, and other sticks interlaced in an horizontal manner. A hole is left in the centre and a long eel pot made of basket or matting is placed before it and into which the eels gather and are thus taken. It is probable that 2 or 3 such pots are set in large weres [weirs]. This were [weir] must have been 100 yards long, at least, and made with wings or corner pieces at the ends thus, or similar to it (Fig. 8).35

A few days later on the 30 April 1841, Robinson recorded another large stake and branch weir in the district of the 'Port Fairy river' (Moyne River) that 'was the property of a family':

From conversations I had with the natives it appears that this was a favourite spot. It was the home of several families. [blank] took me to several spots where he had resided and had worns or huts. He also took me to a very fine and large weir and went through, with several other of the natives, the process of taking eels and the particular spot where he himself stood and took them. I measured this weir with a tape, 200 feet; 5 feet high. It was turned back at each end and 2 or 3 holes in the middle was left for placing the eel pots as also one at each end. The eel pots are placed over the holes and the fisher stands behind the yere.roc or weir and lays hold of the small end of the arrabine or eel pot. And when the eel makes its appearance he bites it on the head and puts it on the lingeer or small stick with a knob at the end, ... or, if near the bank, he throws them out. The fishing is carried on in the rainy season. Arrabine or eel pot made of bark or plaited rushes with a willow round mouth and having a small end to prevent the eel from rapidly getting away (Figs. 9 to 12).36

More generally, Robinson noted that:

These yere.roc or were [wiers] are built with some attention to the principles of mechanics. Those erected on a rocky bottom have the sticks inserted in a grove37

made by removing the small stones so as to form a grove. The were [wiers] is kept in a straightline. The small stones are laid against the bottom of the stick. The upright sticks are supported by transverse sticks, resting on forked sticks as shown above (Fig. 11). These sticks are 3, 4 or 5 inches in diameter. Some of the smaller were [wiers] are in the form of a segment of a circle. The convex side against the current (Fig. 13).38

Elsewhere, Robinson reported that eel baskets were 'from 9 to 12 feet [2.7 – 3.7m] in length'.39 A rare example of an eel basket was obtained by Alfred Kenyon from Lake Condah in 1910 (Fig. 14). Woven eel baskets continue to be made by Gunditjmara women today (see Fig. 1).

In July 1841, Robinson visited the region immediately east of Gariwerd (The Grampians). It was here that he made two extraordinary recordings of eel trapping facilities that differed in form and scale to what he had seen previously. The first hint of these landscape-scale earthworks was observed on 8 July 1841: 'Passed several dieks [dikes] dug by the natives for draining small lagoons into the large ones for the purpose of catching eels, &c. These channels were from a foot to 18 inches deep and from 8 to 300 yards in length'.40 The following day, Robinson approached low-lying country near Mt William and recorded what remains the largest earthen construction of pre-contact Aboriginal people recorded by Europeans for any part of Australia:

At the confluence of this creek with the marsh observed an immense piece of ground trenched and banked, resembling the work of civilised man but which on inspection I found to be the work of the Aboriginal natives, purposely constructed for catching eels. A specimen of art of the same extent I had not before seen and therefore required some time to inspect it, and which the absence of transport enabled me to do. These trenches are hundreds of yards in length. I measured at one place in one continuous trepple [triple] line for the distance of 500 yards. These treble watercourses led to other ramified and extensive trenches of a most

tortuous form. An area of at least 15 acres [6 ha] was thus tracd over. The whole reminded me of the extensive circumvaliations of Chatham Lines, in miniature, at which works, at an early period of my life, I [had] been engaged under that veteran engineer Colonel De Arcy for seven years. These works must have been executed at great cost of labour to these rude people the only means of artificial power being the lever, the application and inventive of which force being necessity. This lever is a stick chisel, sharpened at one end, by which force they threw up clods of soil and thus formed the trenches, smoothing the water channel with their hands. The soil displaced went to form the embankment ... To me it was new and particularly interesting and evinced great perseverance and industry on the part of these Aborigines. This description of work is called by the natives cro.cup.per, i.e. Bennewongham [said so].

The plan or design of these ramifications were extremely perplexing and I found it difficult to commit it [to] paper, in the way I could have wished, all its varied form and curious curvilinear windings and angles of every size and shape and parallels, &c. At intervals small apertures left and where they placed their arabine or eel pots. These gaps were supported by pieces of the bark of trees and

sticks. In single measurement there must have been some thousands of yards of this trenching and banking. The whole of the water from the mountain rivulet is made to pass through this trenching ere it reaches the marsh; it is hardly possible for a single fish to escape. I observed at short distance higher up, minor trenching was done through which part of the water ran in its course to the more extensive works. Some of these banks were 2 feet in height, the most of them a foot and the hollow a foot deep by 10 or 11 inches wide. The main branches were wider.41

Robinson's journal contains two near identical sketches of the Mt William channels, revealing a repeated attempt to better represent the form of the site (Fig. 15). Ten days later, Aboriginal people made their own drawings of the Mt William channels in Robinson's journal. Robinson stated: 'The natives informed me, in reference to the trenches at Mt William for catching eels, they call them vam. ... The natives made a rude drawing of the vams or trenches for catching eels ... These tracings were first made with a stick on the ground. I then got them to make the same tracings on paper with a pencil' (Fig. 16).42

Dawson provides details on a related form of earthworks used to harvest eels: 'When the streams extend over the marshes in time of flood, clay embankments, two to three feet high, and sometimes three to four hundred yards in length, are built across them, and the current is confined to narrow openings in which the pipe baskets are placed. The eels, proceeding down the stream in the beginning of the winter floods, go headforemost into the pipes, and do not attempt to turn back'.43

To move understanding of Gunditjmara fishing facilities beyond the recordings of Robinson and Dawson requires fieldwork and the fruits of archaeological and environmental research. Without this research, our earliest historical understandings of the Gunditjmara fishery will be forever time-locked to fragmentary European records of the nineteenth century. The recordings of Robinson and Dawson indicate that a fishtrap may represent a single facility owned and used by a family or one small component of a much larger integrated fishing system covering many hectares and operated by numerous clans. As such, it is clear that detailed recording of stone-walled fishing facilities must be combined with a landscape approach to comprehend the potential scale and operation of integrated fishing systems. Detailed archaeological site surveys by the Victorian Archaeological Survey (VAS) and Aboriginal Affairs Victoria (AAV) between 1977 and 1990 recorded five clusters ('systems') of 78 stone-walled fish trapping facilities and associated eel 'holding ponds' at Lake Condah (Figs. 17 and 18).44 Such is the scale of these constructions that archaeologist Peter Coutts and his coresearchers estimated that 'many hundreds of tonnes of basalt boulders have been shifted at Lake Condah to build the intricate network of dams and weirs found there'.45 In the case of the trapping facility mapped in figure 18, the large investment of construction labour reveals creation of a major permanent clan asset intended for re-use over and over again. Although European drainage of Lake Condah rendered the traps inoperable, it was hypothesised by VAS and AAV researchers that many of the trapping facilities operated together depending on different flood water levels. To test this hypothesis, AAV took detailed topographic contour mapping of the lake and using a Geographic Information System (GIS) computer program developed a 'hydrological model' to demonstrate how traps within System 1 in the far southwest corner of the lake linked up at different simulated water levels and flow regimes.46 Similar GIS modelling has been undertaken on trapping facilities along Darlots Creek which connects Lake Condah to the sea.47

One aspect of archaeological mapping that has produced fundamentally new insights beyond that known from historical records is the documentation of what appear to be eel holding ponds associated with eel aquaculture. After heavy winter rains during fieldwork in June 1977, the VAS team noted:

As the lake level dropped, water was trapped in pools, some of which were very deep (3 – 4m) and extensive (50m across). These pools were connected by the now dry or damp stone races and canals or by a series of minor pools. In prehistoric times, fish could have been captured in these ponds and eels could have been taken in traps if they attempted to escape overland along the artificial structures towards the main body of the lake.48

Related inferences concerning holding ponds along Darlots Creek were made by archaeologist Heather Builth.49 Builth found that channels were constructed to divert water from Darlots Creek into adjacent complexes of depressions/swamps augmented by dams and linked through further channels. Based on GIS hydrological modelling, Builth hypothesised further that these 'pens' were used to hold young eels (elvers) swimming upstream during spring and to provide favourable long-term conditions for subsequent growth into adulthood. Following maturation, adult eels would after their normal 7 – 20 year 'terrestrial' life cycle, begin their autumn migration downstream where many were trapped and killed for consumption.50 Based on ponds constructed artificially to 'hold' and 'grow' eels (and most likely other fish), Builth argued that the Gunditjmara practiced 'eel aquaculture'.51

The suggestion that Aboriginal people of western Victoria practiced eel farming was first proposed in 1980 by Harry Lourandos based on archaeological research at Toolondo on the northwest side of Gariwerd.52 Here Aboriginal people excavated a 3km long channel system to connect Budgeongutte Swamp (within the natural range of eels) with Clear Swamp (located to the northwest and outside the natural range of eels). According to Lourandos, 'the size and construction of these drains points to their operation as more than mere eel harvesting devices'.53 The main channel is 2.5m wide and 1m deep but would have been deeper except for erosion infill. It most likely functioned to allow migrating eels to extend their range into a new and previously inaccessible swamp. Lourandos argued that engineered channel systems recorded archaeologically at Toolondo, and historically at Mt William by Robinson, increased the habitat and 'annual production of eels', and 'helped to regulate the size of the local eel population at all seasons'.54 This situation led Lourandos to conclude that Toolondo and Mt William 'could be viewed, in effect, as eel 'farms', or at least managed eel habitats'.55 Historian Noel Butlin also thought it appropriate to refer to the Aboriginal eel fishery of the Western District as 'eel farming'.56

Despite considerable insights into the scale and complexity of the Gunditjmara fishery, few inroads have been made into its long-term history and antiquity. Previous researchers generally agree that as most other archaeological sites in the Western District have been radiocarbon dated to the past 3000 – 4000 years, so too it is likely most fish trapping facilities have a similar antiquity.57 However, archaeologists Caroline Bird and David Frankel rightly point out 'there is no direct dating for the stone systems'.58 To overcome this chronological void, we have been developing a methodology to date trapping facilities based on radiocarbon dating of charcoal fragments within sediments burying the lower sections of stone-walled trapping structures. The method is based on the simple proposition that stone walls must be older than the age of sediments burying them and more recent than the age of underlying sediments. Our research began in 2005 and represents a partnership project between Monash University and the Gunditjmara community (through the Winda-Mara Aboriginal Corporation). To date, our excavations have focused on Muldoons Trap Complex located within the flood zone of the southwest margin of Lake Condah. This trap complex, recorded originally in the late nineteenth century, includes 350m of channels constructed variously by addition of basalt block walls and excavation of sediments and removal of basalt blocks from lava bedrock.59 Some excavated basalt blocks weigh more than 50kg and would have required the assistance of strong wooden levers and the lifting powers of two people. Two locations have been targeted for archaeological excavation – sediments partly burying the lower sections of a barrier wall feature and sediments partly filling a channel (Fig. 19, 20 and 22). Excavations reveal that the lower sections of trap features are buried by clay-rich flood sediments, indicating that siltation would have been a maintenance issue for past users of the trap complex. Preliminary radiocarbon dating suggests the trap complex was constructed within the past 500 years. Future excavations aim to build up a more complete picture of the age of stone-walled fishing facilities around Lake Condah and to determine to what extent the intensity of trap construction and use changed through time.

Coring of sediments from the bottom of Lake Condah provide a history of sediment build-up and environmental change linked to changing water levels and vegetation. Initial coring by palaeogeographer Lesley Head suggested that while a date of 9000 years ago for the beginnings of the lake provides a maximum possible date for fishtraps, most fishtraps would have only operated with water levels of the past 2000 years or so.60 More recent coring by Peter Kershaw and his team of palaeoenvironmental researchers from Monash University suggests the lake began around 11,000 years ago with increased water levels commencing 4000 – 5000 years ago (Fig. 21).61 As other lakes and swamps in western Victoria tend to show decreasing water levels at this time, indicative of reduced rainfall, the anomalous situation at Lake Condah may reflect damming of the Darlots Creek outlet of the lake by Aboriginal people.62 It is argued that artificially increasing water levels in Lake Condah extended the period over which the traps could operate. This hypothesis is startling for it reveals a degree of innovative and purposive environmental manipulation and control by Aboriginal people never before documented for pre-European Australia. Indeed, if correct, the Lake Condah dam would equal in age the oldest known dam in the world, documented in Egypt.63 Whatever the case, the new evidence from Lake Condah supports Lourandos' hypothesis that development of the Western District eel enhancement systems may have been 'stimulated' by the onset of drier climatic conditions of 'the last 3000 years or so', which 'would have increasingly endangered aquatic resources'.64 In this respect, current Gunditjmara plans to construct a weir to artificially restore water levels to Lake Condah can be seen as an extension of customary water management practices. Whereas 4000 – 5000 years ago a dam may have been built to offset the drying effects of climate change, currently a weir is being built to offset the drying effects of European drainage.

The concept of 'firestick farming' spelt the end of stereotypical representations of Indigenous Australians as hunter-gatherers living passively off the natural productivity of the land.65 More recently, the role of Indigenous Australians in actively manipulating aquatic resources in marine and riverine environments has been explored.66 Understanding the long-term effects of active Indigenous environmental engagement, manipulation, transformation and management is fundamental to informed management of Australia's aquatic biodiversity.67 In this respect, recent research into understanding the long-term history of the Gunditjmara fishery and associated aquaculture has fundamentally challenged notions of Aboriginal people as huntergatherers and natural resource availability. The key to understanding the history of this fishery is multidisciplinary research. It is clear that archaeological excavation of sediments burying traps and related facilities and palaeoenvironmental coring of swamps and potential fish holding ponds will shed fundamentally new insights into the form, complexity and antiquity of this fishery. In both situations, the challenge will be to obtain a large sample of study sites across a range of wetland landscapes to identify reliable patterns of chronological changes in trap and aquaculture development.

It would be naïve to assume a fishing facility used over hundreds or even thousands of years by numerous generations played the same role in society throughout its long history. This, in essence, is the key to historicising the fishery – to understand how the use of fishing facilities, both individually and collectively, changed through time as a dimension of changing social and environmental conditions. We cannot assume that all fishing facilities seen across the Gunditjmara landscape today were built and used during the same period in history. In this sense, determining when a fishing facility stopped being used and was abandoned is equally as important as ascertaining when it was built. Understanding the degree of contemporaneity of faculties will be a major challenge for future archaeological research.

It is important to always remind ourselves that fishing facilities represent only one aspect of the complex lifeway of the Gunditjmara and their ancestors. To understand and explain long-term patterns in the eel fishery will require equally detailed contextual information on long-term changes in use of other sites in the region such as habitation structures (e.g. stone houses), campsites (e.g. stone artefact scatters), oven mounds, and shell middens along the adjacent coast. Research is also required to further test hypotheses relating to the occurrence of eel smoking trees and eel caching/storage structures.68 Where archaeological research is most lacking is in understanding the spiritual dimensions of the Gunditjmara landscape and in particular ritual sites associated with the fishery.69 Equally challenging will be continuing theoretical debates on the extent to which long-term cultural changes in Western District Aboriginal society reflected largely environmental change or an interplay of environmental and social factors.7071 In all cases, the aim is to construct a long-term history of the Gunditjmara fishery that not only documents its technological dimensions, but also elucidates the social dimensions of the fishery and the complex organisational arrangements necessary to operate small and large-scale fishing facilities. To downplay these organisational dimensions is to miss the key point that operation of the Gunditjmara fishery was and remains quintessentially a social act. In this sense, if archaeologists are to write a history of the fishery they need to continually explore not only the broader social conditions and relationships governing the manufacture, use, and maintenance of fishing facilities through time, but also identify the intentional actions of people, both individually and collectively, embedded in every ancient fishing facility across the Gunditjmara landscape.