Review of existing Red Fox, Feral Cat, Feral Rabbit, Feral Pig and Feral Goat control in Australia. II. Information Gaps
Ben Reddiex, David M. Forsyth.
Department of the Environment and Heritage, 2004
7. Results and Discussion (continued)
The first stage of this review (Reddiex et al. 2004) showed that there was little reliable knowledge about the benefits of feral rabbit control for native species and ecological communities. In contrast, there is some evidence of the impacts of rabbits on native species and ecological communities for rangelands and higher rainfall areas (see Williams et al. 1995). Feral rabbits are believed to impact on native fauna via direct competition for resources and through behavioural interactions such as exclusion of native animals from feeding areas (Williams et al. 1995). However, few studies have experimentally investigated these potential impacts (but see Robley et al. 2002).
There are reliable methods for estimating the relative abundance (i.e., spotlight counts; Caley and Morley 2002) and absolute abundance of feral rabbits (i.e., mark-recapture; Twigg et al. 2000) in most habitat types, and the effectiveness and costs of control are well known (Williams et al. 1995).
There is limited information on the benefits of feral rabbit control for native species and ecological communities. Studies that have investigated the impacts of feral rabbits on pasture composition and biomass have largely focused on modified agricultural landscapes where few threatened native species are present (e.g., Gooding 1955; Myers and Poole 1963; Croft et al. 2002). The impact of feral rabbits on native plant species has largely been inferred from exclosure studies (e.g., Lange and Graham 1983; Leigh et al. 1989; Henzell 1991). The main limitation of such studies when attempting to infer benefits of feral rabbit control is that eradication is not feasible in mainland areas of Australia (i.e., exclosures have feral rabbit densities that are not possible via conventional control).
In rangelands, the current replacement rate of many shrubs and trees is insufficient to prevent their loss in the long-term. Lange and Graham (1983) studied feral rabbit browsing of arid zone acacia (Acacia spp.) seedlings when feral rabbits were at low densities, and found that only seedlings that were protected from feral rabbits and sheep showed good growth. Several other studies have indicated that feral rabbits may prevent regeneration of many shrub and tree species (e.g., Johnson and Baird 1970; Friedel 1985; Auld 1990; Henzell 1991). In the Gammon Range National Park in South Australia, Henzell (1991) reported that feral rabbits were a critical factor in determining mulga regeneration because they killed nearly all of the seedlings, and Foran et al. (1985) found the same response for Acacia kempeana seedlings. In a replicated field experiment, Mutze et al. (1997) reported that feral rabbit control resulted in higher levels of recruitment of the arid zone shrubs of moderate palatability in South Australia. However, it is extremely difficult to undertake field experiments to assess the benefits of feral rabbit control for regeneration in rangelands as germination and establishment of vegetation in rangelands may only occur at time intervals of 5-50 years, mainly as a response to rainfall (Ireland and Andrew 1992; Williams et al. 1995).
In the Coorong National Park in South Australia, Cooke (1987) reported that feral rabbits prevented regeneration of Acacia longifolia and the sheoak Allocasuarina verticilliata. In Kosciusko National Park, where feral rabbits were excluded two new species of forbes were found in seven years, but where feral rabbits were present there was a loss of nine forb species (Leigh et al. 1987). An exclosure study in the mallee in western Victoria found 17 indigenous species of ground layer plants inside feral rabbit exclosures after 2 years that were not present outside (Cochrane and McDonald 1966). However, other herbivores were present in the study area.
The benefits of a reduction in feral rabbit densities resulting from RHD have been monitored at a number of sites (>10) across Australia (Sandell and Start 1999). Despite most of the sites only being monitored for two years post-RHD all but one of the sites found evidence of native vegetation recovery as a result of reduced feral rabbit abundance (Sandell and Start 1999). The structure of vegetation has been reported to have improved due to regeneration of native trees and shrubs, however floristic changes have been variable and dependent on climatic factors (the results for most of these sites are not available). Sandell (2002) found no evidence of widespread germination of woody seedlings, which is not surprising given the episodic nature of such regeneration in many environments.
Feral rabbits are a known or perceived threat for 84 species listed under the EPBC Act (Table 1); 13 mammals, 13 birds, 1 fish, 1 amphibian, 2 retiles, and 54 plant species. Few of these species were identified in the above overview. The 54 plant species listed under the EPBC Act for which feral rabbits are a known or perceived threat appear to have that status because feral rabbits either have been observed feeding on those species, or because browse on those species has been attributed to feral rabbits, or species have shown a positive response in areas where feral rabbits are excluded. Hence, there is limited reliable information on the benefits of conventional feral rabbit control for nearly all of the species listed in the EPBC Act for which feral rabbits have been identified as a threat.
We consider that the greatest priority is understanding the benefits of feral rabbit control for native plant species/communities. The next priority would be to determine the indirect impact of feral rabbits on native fauna species.
We advocate an experiment that assesses the functional relationship between feral rabbit density and damage to a combination of native species for which feral rabbits are a known key threatening process (Environment Australia 1999b) and other common native species that feral rabbits may impact upon. Our preferred experimental design is a response surface experiment (Mead 1988), and uses large-scale enclosures to assess the impact of feral rabbit density on native plant species diversity and composition, including seedling survival of planted shrub/tree species. We believe that the alternative approach of comparing vegetation response between feral rabbit control programs and paired non-control areas is less desirable due to potential difficulties in maintaining the desired treatments over extended periods of time and over a large scale, and limited control of other herbivores. The proposed enclosures have the advantage of enabling accurate assessment of feral rabbit densities and therefore relationship to damage, but are also large enough to simulate broad acre conditions (note that enclosures could not be used to simulate broad acre conditions for feral goats and feral pigs).
We suspect that the benefits of differing feral rabbit densities on native vegetation will vary between rangelands and high-rainfall areas (Williams et al. 1995). We therefore suggest conducting the following experiment at sites in each of these two ecosystems. However, we encourage the adoption of this design at as many sites as possible throughout the feral rabbit range. Where possible sites should be selected where published information is available on the dynamics of feral rabbit populations, including changes in abundance of feral rabbits following conventional control, and their associated impacts on native vegetation.
The experimental design would be the same for both ecosystems. The experiment should use a randomised design (see Figure 4), with different feral rabbit densities as the treatments at each site. There should be a minimum of four treatments (i.e., enclosures) at each site.
Figure 4. Experimental design for understanding the relationship between feral rabbit density and damage to native plant species in two ecosystems.
Recommended feral rabbit densities should represent typical feral rabbit densities for the regions studied and for the prevailing environmental conditions, but should include a low density and low-medium density representative of sustained conventional control of feral rabbits, and a medium-high and high density which is representative of uncontrolled feral rabbit populations. Each enclosure should include a number of relatively small exclosures that act as experimental controls. The experiment aims to examine the relationship between feral rabbit control and damage, therefore the densities within treatments should be treated to reflect management. We suggest the following management; low density treatment - remove 90-95% of feral rabbits once per year (small population levels may be prone to extinction in enclosures, and may require intensive management/reintroduction); low-medium density - remove 70-80% of feral rabbits once per year; medium-high density - remove 40-50% of feral rabbits once per year; and high density - no removal.
The reliability of the inferences increases with the number of sites and the number of replicates within each site. However, as long as there are at least five sites in each region there can be a minimum of one experiment in each site (i.e., no replication within sites). Sites should be selected so that they include the plant species predicted to respond to feral rabbit control (either in abundance and/or condition) and where possible include EPBC Act listed species for which feral rabbits are a known or perceived threat (Table 1). All treatments within a site would be undertaken on adjacent areas (see Figure 4), and all treatments should have similar soil types and vegetation composition and structure at the commencement of the study.
The size of each treatment enclosure should be at least 4 ha, but if resources permit we encourage the size of each enclosure to be increased (we have costed this experiment based on 4 ha enclosures). Feral rabbits generally do not forage far from their warrens. Wood et al. (1987) reported an inverse relationship between distance from warrens and the intensity of feeding, with 800kg/ha of forage removed <12m from the warren, 220kg/ha 25 m from the warren and 150kg/ha at 100 m from the warren. Feral rabbit home range differs markedly from one environment to another (range 0.05-4.70 ha; Myers et al. 1994). Each enclosure will be fenced in a manner that prevents feral rabbits from moving outside their intended enclosure, and they will be fenced to a height (c. 1.8 m) that prevents entry of other herbivores that are likely to affect the species of interest (e.g., Henzell 1991; Grice and Barchia 1992). Several fence designs achieve these requirements (review in Long and Robley 2004). Predator control will need to be undertaken around all enclosures throughout the duration of the experiment (predator control has been included in the experiment costing).
Prior to the treatments being imposed, the vegetation within each enclosure should be sampled. Monitoring protocols for assessing grassland species composition and biomass are widely available (e.g., dry-weight-rank technique; Mannetje and Haydock 1963; modified step-point sampling technique; Cunningham 1975). Monitoring of plant composition, condition and biomass should be undertaken quarterly as there are pronounced seasonal variation in many grassland systems.
The impact of feral rabbit densities on the survival of shrubs/tree species would be assessed through monitoring the survival of planted seedlings in all enclosure treatments. As mentioned above, germination and establishment of vegetation in rangelands may only occur at time intervals of 5-50 years, mainly as a response to rainfall (Ireland and Andrew 1992; Williams et al. 1995). Therefore, it is unlikely that establishment of shrub/tree species will occur naturally in the enclosures during the timeframe of the study. The shrub/tree species selected will act as a proxy for species that feral rabbits are believed to prevent regeneration of (e.g., Acacia spp.; Williams et al. 1995) and therefore be similar in palatability and structure. However, we have also costed the addition of a simulated rainfall treatment to this experiment that may enable natural regeneration to occur (i.e., doubling the number of enclosures at each site). This would involve replicating the above enclosures at each site and randomly selecting one block to be irrigated.
The abundance of feral rabbits should be monitored throughout the study (quarterly) using mark-recapture methods. Enclosures should also be inspected at least every fortnight, to ensure the fence has not been breached by feral rabbits or other herbivores and on the potential for incursion (e.g., overhanging branches that may fall on fences, or proximity to creeks that might erode the fence).
Other covariates should be monitored at each site. For example, rainfall is thought to be important for the germination of some seeds. Hence, a response of feral rabbit control might not occur until a threshold soil moisture has been exceeded. Other covariates might include the abundance of small native herbivores that may enter the enclosures, presence of disease (e.g., myxomatosis and RHD), and temperature.
This design will enable benefit-cost analyses to be undertaken as it will provide a relationship between incremental pest density and incremental damage, without which cost-benefit analyses are tenuous (Fleming et al. 2001).
How long should the experiments run for? The answer will depend on the plant species monitored and the environmental conditions that occur during the experiment. And there is always the possibility of 'demonic intrusion' (e.g., destruction of enclosure fences) ruining even the best design. However, we believe that there should be at least 1 sample in all treatments prior to the commencement of the study to gather accurate baseline information on the response variables that are to be assessed and at least four years of monitoring before the experiment is reviewed to enable sampling of different seasonal conditions. This design also enables the treatments to be reversed (i.e., feral rabbit densities changed between enclosures). The key relationship is that between feral rabbit density and damage. We expect that at least five sites are needed to provide a reasonable confidence interval around this relationship in each ecosystem.
Until study sites are identified, the cost of the experiment can only be considered indicative (Table 4). We estimate that the start-up costs of the experiment for one ecosystem with five sites will be (including overheads) $490K (excludes the simulated rainfall treatment). The annual ongoing cost will be $320K.
|Item||Start-up (year 1) costs ($000)||Ongoing (year 2and beyond) costs ($000)||Final year costs ($000)|
|a) Excludes simulated rainfall treatment|
|b) Includes simulated rainfall treatment|
1 Assumes 100% overheads, but not all organisations charge overheads.
2 Irrigation costs will depend upon the location of sites.