Previous Page Table of Contents Next Page

Management of sea cucumbers in the Northern Territory, Australia, and current research to further improve understanding of the fishery

Colin C. Shelley1 and Philippe Puig2

1 Darwin Aquaculture Centre, Darwin, Australia; 2EWL Sciences, Darwin, Australia


Australian federal, state and territory fishery agencies are committed to the concept of ecologically sustainable (ESD) fisheries and as a result have in place a plan to demonstrate this for all Australian fisheries. As a result the trepang fishery in the Northern Territory (NT) of Australia is being reviewed and a new research program has been initiated to further quantify the current fishery and to develop a suitable monitoring programme to underwrite its sustainability.

Archaeological and historical data from the late 19th and the early 20th century demonstrate that current fishing grounds have been consistently harvested for over 300 years, indicating that long-term sustainability of trepang fishing is possible.

Geographic Information System software (ArcView®) was used to visualise fishing effort by location in the modern fishery after cleaning the initial dataset from fisheries logbooks. The same software was also used to determine relationships between trepang number and weight. Basic analysis of the data using fuzzy logic found several apparent size classes within the data, although their significance was not determined.

A fishery-independent survey of the existing trepang fishery and of potential new grounds is proposed for the next 2 years. This work will combine diver surveys and the use of target specific sampling gear towed by a trawler, utilising a stratified sampling approach to gain information on local habitat preferences. As 12 major fishing grounds account for over 90 % of the total catch of the fishery, these will be targeted in the survey, with biological, physical and habitat data being collected on a relatively fine scale.

Keywords: Trepang, management, fishery, survey, Northern Territory, Australia

Sustainable fishing in Australia

Australian fisheries that export product have had to produce a report that demonstrates their ecological sustainability by December 2004 under the Environment Protection and Biodiversity Conservation Act 1999. If the report is assessed as demonstrating the fishery is managed in an ecologically sustainable manner, the fishery is approved for export for a 5-year period. Otherwise temporary approval for export can be provided if the fishery is moving toward sustainability or, if considered unsustainable, export can be prohibited. The trepang fishery in the Northern Territory (NT) is currently going through this public and expert assessment process. Details of this program, operated by the Department of the Environment and Heritage, can be found at the following web address -

To assist in further understanding the existing trepang fishery in the NT and to ensure its long-term sustainability, the NT Fisheries Group, in collaboration with industry, is to commence a fisheries independent survey of the fishing grounds during 2003/04. The proposed methodology is addressed later in this paper.

History and description of the trepang fishery in the Northern Territory

Macknight (1976) documents the involvement of Macassans in trepang collecting in the NT since 1700. He states that the use of trepang by the Chinese can be dated to the late 16th or early 17th century and its name, hai-sen (sea ginseng) reflects its supposed medicinal properties. Not surprisingly, the distribution of Macassan trepang processing sites, shown in Macknight’s (1976) work, is in general shoreward of contemporary trepang fishing grounds in the NT. In his work, annual estimates of trepang collected by the Macassans were put at over 350 tonnes (dried weight) on average, but up to 600 tonnes in an exceptional year. They collected trepang by diving and by the use of dredges towed behind double out-rigger canoes.

Two partial surveys of the trepang fishery have been undertaken since 1989, first Vail (1989) and then Carter (1995). Like most trepang surveys, population densities were found to vary widely within areas, making attempts to extrapolate to total standing stock very difficult, with extremely high variability. Whilst Carter (1995) found that stock density was not related to the tide or lunar cycle, she did note that trepang were generally absent from habitats adjacent to exposed coastlines.

The trepang fishery in the NT is restricted to 3 of the 12 coastal bioregions found in NT waters (Anon, 1998). All 3 bioregions are characterised by an underlying non-depositional land form, which extends from the coast to the sub-tidal zone; they experience high annual rainfall (1 000-1 400 mm per year); annual surface seawater variations in the range of 5-8 °C; relatively few mangroves fringing the coast and have relatively low tidal ranges from 2-5 m. The sandfish, Holothuria scabra accounts for almost all the catch in the NT, although some other commercial species are collected.

Trepang management in the Northern Territory

Currently trepang fishing in the NT is controlled by regulation. It is predominantly input controlled. There are only six licences in the fishery which are limited by area, species, minimum size, and the number of divers and assistants on each vessel. The fishery extends three nautical miles seaward from the high water mark. Trepang can only be collected by hand or hand held instruments.

In the NT, a special unit of the police undertakes all fisheries enforcement. In the case of the trepang fishery, vessels leaving or arriving in port can be inspected and catch returns may be verified against processing or wholesaler records. At sea, the police can undertake inspections of vessels and their operations if they so wish. Daily records on fishing activity and sales are submitted to the Fisheries Group each month and that data is saved in a database.

Visualisation of trepang catch data using a Geographic Information System (GIS)

Trepang catch data recorded by fishermen for the NT Government prior to this study has only been analysed using standard statistics. It is difficult to visualise possible patterns in trepang harvesting practices along the several thousand kilometres of the NT coastline from a spreadsheet of 1 700 data points. The purpose of undertaking a GIS study was simply to visualise the data in their geographic context and, if possible, identify patterns over time, which would cast some light on the dynamics of the trepang harvesting industry in the NT.

The GIS software used throughout this study was Arc View 3.2 with the Spatial Analyst and free extensions downloaded from the ESRI website (

Dataset description

The original dataset was extracted from the NT Government Oracle database and provided as an Excel spreadsheet. A table was derived from the original dataset after eliminating all empty records, assigning a consistent format to all data in the same field and checking site names against the Australian Gazetteer[36] database. The fields of the table were:

- ID: unique identifier of each record

- Year: derived from the collection date

- Licence_no: the NT Government issued six licences for trepang harvesting

- Hours: number of hours spent harvesting

- No_of_peo: number of people involved

- Collection: four digit code number of the one degree fishing grid cell where trepang were harvested

- No_collect: number of trepang collected

- Weight: wet weight of trepang collected

- Location: name of the location within the grid cell defined in Collection where harvesting took place

- Prop_nam: Proper name (by reference to the Australian Gazetteer) of Location

- Dlong: Official longitude, as a decimal number, listed in the Australian Gazetteer under Prop_nam

- Dlat: Official latitude, as a decimal number, listed in the Australian Gazetteer under Prop_nam

Once created this initial dataset was imported into Arc View GIS as an “Event theme” using the coordinates Dlong and Dlat available for all but about twenty records. The latter were eliminated as no Prop_nam, extracted from the Australian Gazetteer, could match the entry recorded by the fishermen under Location. For all other records Collection, Location, Prop_nam, Dlong and Dlat were consistent.

Figure 1 displays all locations where trepang were harvested between 1996 and 2002.

The initial intention was to display yearly catches expressed as wet weight, in kilograms, of trepang collected in each location. The first discovery was the near complete absence of weight data in the 1996 records, only numbers of trepang harvested in 1996 were available. The first task therefore consisted in extrapolating weights missing in 1996 from records, in subsequent years, on the basis of the correlation between number and weights. In the past, total wet weight of trepang caught had been divided by total numbers of specimens caught to derive an average wet weight for missing records. This approach however was clearly unsuitable when dealing with specific locations.

Deriving weights from numbers of specimens harvested

The relationship between wet weight and number of trepang harvested was explored using Gstats, a free (in its beta version) extension created by SDS Data Services ([email protected]).

The first attempt at regressing weights against numbers of trepang harvested for all records where these two fields were complete produced the plot displayed in Figure 2.

Figure 1. A map of the Northern Territory, illustrating the thirteen cells of the fishing grid in which catches were recorded and individual harvesting locations identified. (Each cell of the fishing grid is one degree of longitude wide and one degree of latitude high (60 nm x 60 nm).

Figure 2. A scattergram of wet weight against number of trepang, for all logbook records, 1996-2002.

Examination of Figure 2 suggested that an average wet weight calculation was inappropriate for extrapolation purposes. Points appeared to line up along a number of well defined slopes.

As a result, a systematic exploration of regression patterns at all harvesting locations with substantial catches was undertaken. Figure 3 demonstrates the disparity of regression lines observed at different sites.

Regression lines of the data in Figure 3 reflect the variability of the relationship between wet weight and number of specimens. Patterns of regression lines observed did not appear to be the result of random errors or processes.

Figure 3. Plots of wet weight (y-axis, kg) against number of trepang (x-axis) for Trepang Bay, Maningrida and North West Bay, which are respectively in the west, centre and east of the study area. The different symbols on the plots represent different fuzzy logic classes.

They showed, across all harvesting sites, from east to west, an increasing number of well defined slopes. Individual regression slopes were calculated (Table 1) for the eleven most productive locations (Figure 4). The twelfth location, Melville Bay, was not included as many records lacked wet weight measurements.

A fuzzy classification algorithm designed by Hong and Lee (1996) was selected to derive representative wet weights for each class (age group, species or other common characteristic). This automatic classification is the first step in the design of fuzzy rules. In this study, however, it only provides an objective categorisation of trepang based on their weight derived from regression lines, and assigns to each category a weight, which is not an average weight but the most representative weight.

Table 1. Individual slopes of regression lines for the eleven most productive trepang harvesting sites, sorted by slope. Classes of trepang 1, 2, 3 and 4 were derived using a fuzzy classification algorithm published by Hong and Lee (1996). Each class corresponds to a triangular membership function defined by the three vertices V1, V2 and V3. V2, the apex, is the most representative value of each class.







Northwest 1



Bartalumba 1









Bartalumba 2



Maningrida 1



Bartalumba 3



Bowen 1



South Goulburn









Maningrida 2



Northwest 2



Port Bremer



Bowen 2



Rolling Bay












Northwest 3






Bartalumba 4



Table 1 suggested that, on the basis of their weight, trepang harvested in NT waters between 1996 and 2002 could be grouped into four classes. The most representative specimens in each class had the following wet weights:

Class 1


most representative weight = 0.451 kg

Class 2


most representative weight = 1.460 kg

Class 3


most representative weight = 2.265 kg

Class 4


most representative weight = 3.116 kg

What is the significance of these classes? Perhaps they reflect different size classes. Results from the proposed survey will hopefully shed more light on the value of this classification.

Figure 4. Annual harvest from the twelve most productive sites (the same sites used in the previous fuzzy classification) displayed every second year since 1996.

The visualization of the trepang catch on an annual basis demonstrates that areas harvested vary from year to year, as do catches from different sites. Whilst an interesting approach, the value of visualization of harvest data should be carefully assessed, in that fishing in different areas may well have been controlled by factors such as weather conditions, economics (distance from port to fishing site), crew availability and serviceability of boats, in addition to on-the-ground stock densities.

Trepang survey methodology and design

Analysis of trepang and other sedentary benthic surveys has identified the use of transects as the main technique utilised to generate population estimates. The size of transects used has varied from 40 m x 2 m diver transects (Friedman, pers. comm.) to 1 500 m tows of trawls (Joll, 1995).

The use of specialised towed gear to collect H. scabra in the forthcoming survey, recognises Vail’s (1989) assertions that a trawl could collect from 80-85 % of the catch, and that because of the soft, silty substrate on which trepang were found in the NT, there was little damage to the environment. He found that H. scabra was found between depths of 0-10 m, with most being found down to 4 m, although Carter (1995) found little difference in inter-tidal and sub-tidal population densities.

Skewes et al. (2000) stated that the burrowing habit of H. scabra could result in underestimates of abundance, of up to 60 % in seagrass beds at high tide. The survey to be undertaken in the NT will use the same stratification by substrate approach as used by Skewes et al. (2000) to estimate stock.

The proposed survey will include several sites previously surveyed (Vail, 1989; Carter, 1995), so enabling an immediate comparison over time. When combined with fishing effort data from those sites, the results may increase our understanding of the dynamics of the fishery.

An initial pilot study will be used to refine the methodology for the main survey. The pilot and then the main survey will take a stratified approach examining trepang densities with depth, and in relation to seagrass cover and sediment type. A few depletion trials will be undertaken on selected transects to determine if underestimates of stock density are as significant as described by Skewes et al. (2000), using techniques as described by Rago et al. (1999). Sediment samples will be routinely collected and analysed for size distribution and organic content. In waters less than 5 m depth (m.l.w.s - mean low water spring tides), transects will be undertaken by divers. In deeper water, target-specific sampling gear, designed specially for this work, will be utilised. Habitat strata used in the survey will be estimated using GIS tools, so that gross estimates of stock from different habitats at the different sites can be made, as was undertaken by Long et al. (1993) for trochus and Skewes et al. (2000) for H. scabra. There are to be two surveys, one of the existing fishery and one of areas in NT waters outside of the existing fishery to establish what other resources might be available. For all trepang collected during the surveys, the species, length, wet mass, gonad mass, sex, gutted weight and weight of internal organs will be recorded.


In the Torres Strait, Skewes et al. (2000) carried out 165 transects for H. scabra on Warrior Reef and found maximum densities in the seagrass zones. That survey was the second of its kind and it was suggested that such surveys repeated every year would be a useful monitoring tool for the fishery. Whether or not annual stock surveys for most trepang fisheries can be considered an economically useful option for most fisheries has yet to be determined, however the fact that most trepang species occur in major aggregations at least makes it more feasible than widely dispersed populations. In Western Australia, 60 % of the catch is found in just 2 grids (each 60 nm x 60 nm) and in the NT over 90 % of the catch is collected from just 12 sites. Whilst detailed survey work as carried out by Skewes et al. (2000) is very useful for examination of a relatively small area (Warrior Reef), such a detailed approach would not be economically feasible for a fishery distributed over a thousand nautical miles and numerous locations, as found in the NT.

Whilst monitoring effort in a trepang fishery is a useful tool, care should be taken in comparing values between sites, as fishing conditions (e.g. water visibility, currents), gear used (e.g. hookah) and a range of economic factors (e.g. distance from nearest port) can have a major impact on the productivity of divers. When the time spent fishing for trepang in one area can vary from less than a day to perhaps a few weeks, how should catch per unit effort be analysed in that location and what, if anything, does catch per unit effort mean averaged across a geographically fragmented fishery?

Trepang fishing records in the NT are currently recorded using a grid, with each square of the grid having sides of 60 nm in length. More precise information could be obtained using a 6 nm grid, as has been introduced for recording prawn trawler catches across northern Australia. The approach of using GIS statistics combined with fuzzy logic to tease out further information from catch data has proved useful in reviewing the NT trepang fishery. It has enabled more information to be extracted from data recorded than has previously been the case. Fuzzy logic, a critical component of modern engineering and expert medical systems, has yet to be used to any great extent in either natural resource management or fisheries science. Fuzzy logic, as its name suggests, is valuable in making sense of what-ifs, maybes, ands, and ors, compared to most fisheries tools which seek to explain a fishery in black and white terms, too often using data of questionable value and limited scope. Such tools as fuzzy logic and GIS can theoretically be of enormous value in making sense of extremely complicated ecosystems and assist in the development of inter-active ‘expert (management) systems’ for many fisheries.

To date the management of trepang in the NT has been at a gross level. Uthicke and Benzie (2001) detected separate stocks of sandfish along the Queensland coast, which indicated restricted gene flow between populations, and inferred that, as a result, consideration should be given to managing that species on a local scale. With information to be gathered from the 12 most significant trepang sites in the NT from the proposed survey, information at a finer scale will be available to assist in refining future management plans. Visualisation of trepang catches over time can assist in understanding the dynamics of a trepang fishery. Such visualisation can be considered part of a more stratified approach to fishery assessment. Production of average catches or average weights have limited use when the fishery is broken down into sites or areas, with each potentially having different factors affecting their productivity. For example in discussions with fishermen, some areas are known to have regular recruitment every year, whereas others appear to have infrequent recruitment.

Macknight’s (1976) examination of the Macassan trepang industry in the NT stated that ‘the very continuation of the industry (trepang) at a more or less consistent level for such a long period confirms that overfishing cannot have been a serious problem. Perhaps a limited area might be cleaned out for a season by several weeks work, but by the following year it would repay another visit...’. Examination of the way in which catches vary from different sites in the contemporary fishery (Figure 4) from year to year, may well reflect the way in which Macassan fishermen sustained a fishery over several hundred years. In the NT, a sustainable fishery is being maintained by a combination of regulation, enforcement and economic pragmatism. Its future sustainability should be on an even firmer footing when a more detailed understanding of the fishery is gained from the proposed trepang surveys and whatever monitoring programs may be put in place in the future. It is the understanding of the authors that under Australian economic conditions, where catch per unit effort for trepang is too low, fishermen cannot afford to continue to fish, and so move on to try other sites. In the NT where one company currently owns all licences, overfishing would be against the long-term best interests of the licensee. In a more competitive fishery, the desire to fish as much as you can before your competitor does is more likely to lead to overfishing. Where catches vary so much from one year to the next and from site to site within a fishery, as they appear to do in the NT trepang fishery, management strategies sensitive to the overall dynamics and economics of the fishery need to be carefully considered.


We would like to acknowledge input and editorial advice from a range of Fisheries Group personnel and the support of Tasmanian Seafoods Pty Ltd in developing the trepang survey proposal.


Anon. 1998 Interim marine and coastal regionalisation for Australia: an ecosystem-based classification for marine and coastal environments. Version 3.3. Environment Australia, Commonwealth Department of the Environment, Canberra. 104pp.

Carter, J.L. 1995. Sustainable harvests of trepang by traditional owners of Gurig National Park and the Maningrida area of northern Arnhem Land. Report to Environment Australia. Department of the Environment and Heritage, Canberra. 166pp.

Hong, T.P. & Lee, C.Y. 1996. Induction of fuzzy rules and membership functions from training examples. Fuzzy Sets and Systems, 84:33-47.

Joll, L.M. 1995. Stock evaluation and recruitment measurement in the W.A. pearl oyster fishery. Fisheries Research and Development Corporation Report, Project 92/147. 63pp.

Long, B.G., Poiner, I.R. & Harris, A.N.M. 1993. Method of estimating the standing stock of Trochus niloticus incorporating Landsat satellite data, with application to the trochus resources of the Bourke Isles, Torres Strait, Australia. Marine Biology, 115:587-593.

Macknight, C.C. 1976. The voyage to Marege: Macassan trepangers in northern Australia. Melbourne University Press. 175pp.

Skewes, T., Dennis D. & Burridge, C. 2000. Survey of Holothuria scabra (sandfish) on Warrior Reef, Torres Strait, January 2000. CSIRO report. 28pp.

Rago, P., Weidman, C. & Weinberg, J. 1999. Depletion estimators of survey catchability: theory and field experiments. Proceedings of the Fifth National NMFS Stock Assessment Workshop. NOAA Technical Memo NMFS-F/SPO-40. 101pp.

Uthicke, S. & Benzie, J.A.H. 2001. Restricted gene flow between Holothuria scabra (Echinodermata: Holothuroidea) populations along the north-east coast of Australia and the Solomon Islands. Marine Ecology Progress Series, 216:109-117.

Vail, L.L. 1989. Trepang resource surveys Melville Island, Gove Harbour, Croker Island. Northern Territory Museum of Arts and Sciences, Darwin, NT. 111pp.


Previous Page Top of Page Next Page