Commercial harvesting of Kangaroos in Australia

by Tony Pople and Gordon Grigg
Department of Zoology, The University of Queensland
for Environment Australia, August 1999
Chapters 10,11,12 and 13 and Appendix 1 provided by staff at Environment Australia



It has often been remarked that management programs of wildlife (free-ranging birds and mammals; Caughley and Sinclair 1994) have largely been pragmatic as opposed to being guided scientifically through such processes as controlled experimentation (e.g. Downing 1981; Caughley 1985; Macnab 1983). However, there are usually good reasons for this, such as data being too limited to make unequivocal predictions. Conservative trial and error, adjustment and frequent monitoring are therefore to be encouraged.

In contrast to fisheries management, wildlife managers have often been able to monitor animals and their food supply directly, rather than indirectly through catch data. However, large mammals have high adult survival rates, low fecundity rates, late attainment of reproductive maturity and are generally long-lived. Such a life history leads to long time lags in population responses (Goodman 1981). As a result, there is a frustratingly long time series of data required for the dynamics of any large mammal populations to be adequately described and understood. This is particularly pronounced for species in unpredictable environments.

Management of wildlife populations can be divided into three areas: pest control, conservation and sustained-yield harvesting (Caughley and Sinclair 1994). The aim of pest control is to reduce the impact of a population. The aim of conservation is to stop a population's decline and make it increase. The aim of sustained- yield harvesting is to harvest a population without jeopardising future yields. Wildlife management also includes no intervention, where a population is simply monitored. Once the goal for the management of any particular population has been established (essentially a value judgement), the technical decisions on how to reach it can be made. There need to be objective criteria with which to judge the success or failure of management actions. Management goals need to defined precisely. For example, for a pest control operation, the objectives need to be stated in terms of reducing impact (e.g. increasing plant biomass, increasing a competitor's density or condition), rather than simply numbers of pest animals reduced or remaining. Differing management goals may not be incompatible. Small reductions in the density of a pest may reduce damage below economic injury levels, allowing harvesters free reign to manipulate population size to optimise yield. Sustained-yield harvesting, by ecological definition (but not necessarily economic; see below), requires conservation of the resource.

Caughley (1976, 1977) outlined the principles of wildlife harvesting. To harvest a sustained yield from a population at steady density, it must first be manipulated in some way to promote its rate of increase (e.g. reduce it below its ecological carrying capacity or supplement its resources). Four theoretical principles can be identified:

  • Any harvest of a population reduces its abundance and the greater the harvest rate, the smaller the population becomes;.
  • Harvesting theory rests upon populations being regulated by some combination of density-dependent reproduction and mortality. This has been described well for a number of large mammals (Fowler 1987). Harvest mortality is seen as being compensated to some extent by lowered natural mortality rates and increased fecundity rates;
  • Rates of harvest may be raised to levels at which they can cause the extinction of the population;
  • Between no exploitation and over-exploitation to extinction there are many levels of sustained yield, but only one maximum sustained yield (MSY), taken at economic carrying capacity.

While the MSY is not the goal of any State kangaroo management program, it serves a useful and important benchmark for discussing sustained yield harvesting. The sustained yield will be determined by the relationship between the population and its resources. Its calculation was determined for a number of examples of population growth by Caughley and Sinclair (1994). Those relevant to large mammals include:

  • For exponential population growth, the harvest rate appropriate to the MSY will equal the rate of increase of the unharvested population.
  • Where population growth follows a logistic pattern, the harvest rate appropriate to an MSY will be rm/2 from a population size of K/2 (where rm is intrinsic or maximum rate of increase and K is ecological carrying capacity) (Figure 1). The logistic growth model is appropriate for populations whose limiting resource is consumable, but the rate of renewal of that resource is unaffected by the animals. This is rarely appropriate for large mammal populations, although Caughley and Sinclair (1994) provided some exceptions including cases where it at least mimics actual population growth.
  • If growth is the result of an interaction between a population and its food supply, then the MSY and the population from which it is taken will generally be greater than for populations growing logistically. An estimate based upon logistic growth will therefore provide a conservative estimate of the MSY. A large number of parameters will need to be estimated to adequately describe this interactive model of population growth. These will rarely be available (but see below for red kangaroos). However, Caughley (1976) provided an approximation of the MSY using an interactive model.

The logistic model also assumes that rate of increase declines linearly with density, but Fowler (1981b) described several examples of markedly non-linear density dependence in large mammals. In general, peak productivity will be at population densities greater than half the carrying capacity (Fowler 1981a). While the logistic model frequently does not adequately describe the population growth of large mammals (see McCullough 1992, Hone 1994 and Choquenot et al. 1998, press, for further discussion), it may nevertheless allow a conservative first approximation of the MSY and provide a reference point when examining real population data. It is also instructive, illustrating well the principles of sustained-yield harvesting (Figure 1). However, Crawley (1983) issued several warnings concerning the use of the logistic model. These included such problems as year-to-year changes in rm and K (which will add to the inherent instability of the MSY), potential errors in the estimation of rm and K, and the divergent effects of the timing of density dependence and reproduction in relation to the harvest.

Population growth and yield curve Population size resulting from harvesting at various harvest rates

Figure 1.
Harvesting a population that grows logistically. a, Population growth and yield curve; b, Population size resulting from harvesting at various harvest rates (= quotas that are constant proportions) from population in (a) (after Caughley and Gunn 1996). [When not limited, populations will grow exponentially, such that it is more appropriate and mathematically simpler to express rates of change exponentially rather than as percentage change. A population that doubles, increases at an exponential rate of r = 0.69 (i.e. e to the power of r = 2, where e = 2.718). A population that halves, declines at an exponential rate of r = -0.69. The exponential rate of increase can be calculated by subtracting the natural logarithm of population size at time 1 from the natural logarithm of population size at time 2.]


In markedly fluctuating environments such as arid and semi-arid Australia, the concept of an equilibrium population is clearly inappropriate. An MSY approximated from the logistic model may be wildly inaccurate. Early estimates of sustained yield for kangaroos (Main 1969; Winter 1970) were criticised by Caughley et al. (1977) for, among other things, using static models. Nance's (1985) age-structured model of an eastern grey kangaroo population would be criticised for similar reasons. McCarthy (1996) used logistic-like growth models to simulate the dynamics of a red kangaroo population. Environmental variability was incorporated to simulate the variability seen in real populations. The modelling demonstrated that if environmental variability was ignored, sustainable harvest rates would be overestimated.

For red kangaroos in particular, population density has proved a very poor predictor of rate of increase (Caughley and Gunn 1993); a relationship that is central to single-species models such as the logistic model. However, food supply as measured by pasture biomass or rainfall is correlated strongly with rate of increase (J. Caughley et al. 1984; Bayliss 1985a, 1987; Cairns and Grigg 1993) and, more importantly, causally linked (Shepherd 1987). For example, drought in 1982-3 in eastern Australia resulted in a decline of about 40% in the population size of the three kangaroo species, over an area of more than one million square kilometres (Caughley et al. 1985).

Rainfall is the governing influence on the population dynamics of plants and animals in arid areas (Noy-Meir 1973). In accordance with this, Caughley (1987a) developed an interactive or consonant model for the red kangaroo-pasture system of western New South Wales (Figure 2). The model was rainfall-driven, with random samples from long-term rainfall records stimulating pasture to which the kangaroo population responded. The relationship between rainfall, pasture biomass and pasture growth was described by Robertson (1987). The functional response of kangaroos to pasture was determined by Short (1987). To complete the picture, Bayliss (1987) described the numerical response of red kangaroos as the relationship between kangaroo rate of increase and pasture biomass.

Caughley's (1987a) model had two important feedback loops. The first involved a reduction in pasture growth as pasture biomass increased and the second involved a reduction in pasture biomass as kangaroo density increased. The regulatory effects of these feedback loops were considered relatively weak, as the overwhelming influence on the system was the variability in rainfall, but the feedback loops were nevertheless important. Without them, pasture biomass and kangaroos would increase unchecked. This modelled population was harvested using a number of fixed harvest rates. The MSY was taken at a harvest rate of 10-15% per year from a population that was 30-40% below its unharvested mean density (Figure 2). Caughley (1987a) used an rm of 0.40 in the model. The logistic growth model would have overestimated the MSY and the corresponding harvest rate, and also underestimated the corresponding population size.

Caughley (1987a) sought an MSY by harvesting at a constant harvest rate for 100-year simulations. Where populations fluctuate to some extent, the optimal harvest strategy may be to vary harvest rate. Using Caughley's (1976) interactive model of an ungulate-pasture system, Stocker and Walters (1984) found that harvest offtake was increased by doing this. Within limits, the strategy called for a reduction in harvest rate with increasing pasture, and a higher harvest rate at low pasture biomass.

One hundred year similuation of rainfall, subsequent pasture biomass and expected kangaroo density at Menindee, NSW.

Figure 2a. One hundred year simulation of rainfall, subsequent pasture biomass and expected kangaroo density at Menindee, NSW. Modelled mean harvest offtake of kangaroos

Figure 2b.Modelled mean harvest offtake of kangaroos with each point resulting from an independent simulation such as (a).

Simulated mean harvest offtake plotted against mean population size

Figure 2c. Simulated mean harvest offtake plotted against mean population size (after Caughley 1987a).

Selective harvesting

Where the population can be harvested selectively in terms of age or sex, considerable gains can be made in the MSY (Caughley 1977; Fowler 1981a). If the composition of a population becomes weighted towards females (and assuming a polygynous or promiscuous mating system; Wittenberger 1979) or towards the reproductively mature age classes, then its potential rate of increase will be greater. This may result from selective harvesting or environmental perturbations that produce differential survival. The optimal harvesting strategy for an age-structured population has been shown to be the complete removal of one age class followed by the partial removal of a second (Beddington and Taylor 1973). While appropriate for farming and game ranching (e.g. Fairall 1985) where age classes can be readily identified, this strategy is unlikely to be possible for wildlife harvesting. Caughley (1977) argued that trying to maximise yield in terms of numbers, weight or value through age-selective harvesting was unlikely to be worthwhile. The yield from such harvesting was only marginally greater than that from a non-selective harvest. However, this outcome will depend on the relative value of age classes, their ease of identification and the selectivity imposed by either management or the industry.

The marked sexual dimorphism in kangaroos (Jarman 1989) leads to quite strong sex-selective harvesting because value is attached to the size of skins and the weight of carcasses. Dealers in kangaroo products also set minimum carcass weights and skin sizes. Kangaroos exhibit hierarchical promiscuity (Wittenberger 1979; Lee and Cockburn 1985) with males of higher social rank copulating more and with more females than subordinate males (Croft 1989). A reduction in the proportion of males in the population up to some threshold will therefore clearly increase the MSY for kangaroos. P. Timmers, University of Queensland (unpublished data) modelled selective harvests of an age-structured population of red kangaroos as an extension of Caughley's (1987a) interactive model. At a constant harvest rate of 15%, sustained yield was maximised with a harvest comprising 60-70% male. Yield dropped marginally with an increasingly male-biased harvest and mean population density increased. This suggests that yield would be even greater at higher harvest rates and with greater male bias in the harvest.

Where populations are made up of social groups, harvesting may be directed at the whole or part of the group, depending on social organisation and movement patterns (Caughley 1977). For the larger macropods at least, this is of little concern as the only permanent grouping is between a female and her dependent young-at-foot (Jarman and Coulson 1989).

Community interactions

Sustained-yield harvesting relies upon a reduction in intraspecific competition through a lowering of population density. This potentially affects interspecific interactions as well, although presumably to a lesser extent. These interspecific interactions may be positive, facilitative grazing for example, while others, such as predation or competition, will be negative. Reducing the strength of one of these interactions may be the primary management goal, for example where competition with domestic stock is of concern. Exploitative rather than interference competition is usually suggested as occurring among macropods and introduced herbivores in Australia (Dawson and Ellis 1994). Reduction in density of one species should, to some extent, raise the resources available to another. This increase in available resources may be taken up as an increase in a competing species' numbers and can completely alter estimates of the MSY (Caughley 1977). For Caughley's (1987a) interactive model, the presence of sheep would merely reduce red kangaroo density and not influence the optimum harvest rate, but only if sheep numbers were not adjusted upwards following harvesting.

In many parts of Australia, several species of macropods are harvested together for essentially the same markets. This raises the possibility of an MSY based upon combinations of species, although this has not been used in kangaroo management. Viewed another way, the harvest offtake of one species may be influenced greatly by the presence of another. In the light of predator-prey theory, such species may be acting as alternative prey, allowing another species to be harvested in areas when it would otherwise be at too low a density to be taken. This has been observed in multi-species fisheries, including whaling where harvesting has been directed away from the larger to the smaller species following depletion in stocks of the former (Beddington 1979). Species interactions will be influenced by habitat use and diet, while interspecific differences in weight, skin quality and ease of shooting (habitat use and behaviour) will affect shooter selectivity.

Side effects of harvesting

A reduction in population size and an increase in its resource base are obvious consequences of harvesting. The magnitude of these effects have been discussed above. By contrast, competitors and predators of the target species may be controlled deliberately to increase the size of its population. A population may also be moulded through judicious culling to produce animals with certain characteristics, such as a particular body size or antler breadth as in the case of some ungulates. This may occur for a population managed for sport hunting. Game ranching and farming of domestic animals are extreme examples of this. However, there are a number of side effects of harvesting that are seldom desirable or at least at odds with the stated management aims. Furthermore, many of these culling-induced problems will lock management into a program of continued intervention (Caughley 1983).

Caughley (1983) described a number of problems associated with culling large mammals. Firstly, a reduction in grazing pressure may result in the exclusion of plant species that either increase under grazing or invade grazed areas. Dampening the fluctuations in herbivore numbers through culling may also reduce ecological resilience (Walker and Goodman 1983). So, while harvesting may result in herbivores in better condition and a greater plant biomass, this does not necessarily equal a healthier system (Shepherd and Caughley 1987). Conversely, culling may have a destabilising effect on the system when its rate increases with declining population density. The troughs in population fluctuations will be deepened, allowing vegetation biomass to increase with fewer impediments.

Herbivore populations appear to exist in a number of possible stable states with predators (Sinclair 1989) and food supply (Noy- Meir 1975). Culling of either predators or herbivores may push the system from one stable state to another. A reduction in predator numbers would result in lower equilibrium vegetation levels while reducing herbivore numbers would increase the mean biomass of vegetation.

Harvesting will usually lead to an increase in fecundity and an increase in juvenile survival rate. A greater proportion of the population will therefore be juveniles; more so if the harvest targets adults. Juvenile survival rate is far more variable than that of adults. A harvested population will therefore be more reactive (Caughley 1983). For long-lived animals under any harvesting regime, the population's life table and therefore its dynamics will be altered to the extent that there will be higher turnover rates, more frequent fluctuations and the population's age distribution will never reach a stable form. The overall impact will depend upon the 'ecological problems to which the original life table was an adaptive solution' (Goodman 1981).

Harvesting may also alter the genetic structure of a population. Genetic diversity may change in response to changes in absolute population size and the degree and frequency of temporal population fluctuations. Selective harvesting, by definition, introduces a further selection pressure, which is most pronounced in game ranching operations and breeding programs for domestic animals.

Management options

Crucial to any management program are its ultimate goals. The actions taken to meet these goals will vary greatly between an eradication program, a program aimed at population control or harvesting for a sustained yield. Alternatively, the objective of sport hunting may be to harvest animals of a certain size or worthy as trophies. The goal may also be one of no intervention, with the population or system being left to its own devices. Culling operations have been proposed for both agricultural lands and national parks. Beneficiaries have been identified as the harvester, the harvested population, a competitor's population, the prey or plant population, or the ecological community in general (Shepherd and Caughley 1987; Owen-Smith 1983). However, kangaroo management in Australia has been made difficult by conflicting objectives in addition to divergent community attitudes to kangaroos (see above). On both national parks and grazing lands, kangaroos are believed to be degrading the environment, hampering vegetation regeneration programs (Cheal 1986; Gardiner 1986a, 1986b; Grice and Barchia 1992) or in balance with their food source (Shepherd and Caughley 1987). Kangaroos are harvested throughout Australia as both a renewable resource and a pest. While the aims for any management program are ultimately value judgements, if they are ambiguous or at least unclear then the outcome or appropriateness of the subsequent management action cannot be judged.

Shepherd and Caughley (1987) described several harvesting systems relevant to kangaroos. A fixed or variable number of animals could be taken each year. However, for a population that fluctuates dramatically in response to the environment, a harvest that does not in some way track the population would be inefficient and potentially dangerous at low densities. If the relationship between harvesting effort (e.g. number of shooters or length of season) and harvest offtake or rate is known, then the harvest could be regulated by effort. This may need to be adjusted if the relationship between effort and offtake varies seasonally or with some other factor unrelated to population density.

The advantages of a constant offtake is in the stability of market supply and reduced monitoring. A slight variant on this, basing quotas on 3-5 year, rolling population means has been considered as a way of dampening the large fluctuations in quotas based on fixed proportions. However, modelling by P. Timmers, University of Queensland (unpublished data), based on Caughley's (1987a) interactive model for red kangaroos, suggests that such a strategy would reduce average yield and increase the variability in both the yield and population size compared with harvesting fixed proportions.

Regulating effort would also require less direct monitoring of populations, but the relationship between effort and offtake needs to be determined. Caughley and Sinclair (1994) argued that it is generally a safer option than a quota based system. It allows fine tuning without the destabilising effects that fine tuning a quota can bring. Harvesting a fixed number near the MSY is particularly dangerous as slight errors in population estimation or monitoring of harvest offtake, or small environmental perturbations, can lead to a density slide. The danger is lessened if the quota is a proportion of the population (see Figure 1). Kangaroos are currently harvested under a variable quota (i.e. a fixed harvest rate) system, probably due to the attractions of administrative control as suggested by Shepherd and Caughley (1987).

An alternative system of harvesting, particularly appropriate to fledgling harvesting operations, is to divide a region up into a patchwork of blocks and only harvest certain blocks (McCullough 1996). Harvested blocks will be replenished by recruitment from within the block and immigration from neighbouring unharvested blocks. The attraction of the system is that it allows a wide margin of error. Additional blocks can be harvested depending upon rates of replenishment.

Harvesting with solely economic goals may lead to the collapse of the stock and possibly extinction (for discussion and examples see Caughley and Gunn 1996). Here, maximising the sustained yield is replaced with maximising economic gain. A population whose rate of increase is low, but value is high, is in danger of being overharvested without close management because of the discount rate. Basically, future earnings must be weighed up against the growth of current earnings when invested; essentially 'discounted by the time it takes to get the money' (Caughley and Sinclair 1994). This argument was outlined in general by Clark (1976) and has been shown to underpin the potential or actual overexploitation of elephants (Caughley 1993) and whales (May 1976).

A further potential danger is associated with a common resource as opposed to one that has proprietary rights. Overexploitation, described by Hardin (1968) as the 'Tragedy of the Commons', can result from individuals having no incentive in conserving a shared resource. Rawlinson (1988) raised this concern for kangaroos, particularly if their value was to increase as argued by Grigg (1991). McCallum (1995) described the possible pitfalls of the current open access harvest of kangaroos, equating it with an infinite discount rate. Even if kangaroos were privately owned, the uncertainty in future population size will increase the discount rate. This may result from a fluctuating environment and from high levels of immigration and emigration which will be particularly pronounced on small properties.

Where there is a high potential for overexploitation of the resource, there must be both regulation and enforcement to ensure its conservation. Furthermore, management should be independent of the industry as their objectives may be diametrically opposed (Caughley and Sinclair (1994). Contrary to Rasker et al. (1992), Caughley (1993) argued that the secure status of elephants in southern Africa was the result of government law enforcement rather than private ownership. However, such arguments will be confounded when there are non-consumptive values attached to the resource such as for tourism. Here, the management aims are the maintenance of certain population levels or a certain mix of species rather than the consumptive use of a sustained yield.