Correspondence Alex Hyatt, CSIRO, Australian Animal Health Laboratory, Private bag 24, Geelong, 3220 Vic., Australia. Email: firstname.lastname@example.org
The marine toad Bufo marinus is native to northern South America, parts of Central America and Southern Texas. It was deliberately introduced into Australia's tropical north-east in 1935 in an unsuccessful attempt to control the cane beetle, a damaging insect pest of sugarcane crops. The toads quickly established in the new environment and began to spread. Today, they inhabit most of the Australian tropics and sub-tropics and have reached Western Australia. Models predict that global warming will enable the toads to extend their range further south. They cause severe environmental impacts, as all life stages of B. marinus contain bufadienolides, alkaloid substances toxic to vertebrates, resulting in death of the predators ingesting it. The continental scale of this biological invasion in combination with the remoteness of the areas affected, poses a specific set of challenges to potential control approaches for cane toads. This review covers different biocontrol strategies pursued over the past 8 years, with particular focus on an immunological approach aiming at the disruption of toad metamorphosis. So far, research efforts have failed to produce a tool for large-scale reduction of toad populations. Considerations of future research priorities and efforts are also discussed.
Introduction and impact of the cane toad in Australia
Native cane beetles have been persistent pests of sugarcane grown in Australia (Mungomery, 1931). In particular, larvae of the grey-back beetle Dermolepida albohirtum and frenchi beetle Lepidiota frenchi presented significant problems during the 1930s (Mungomery, 1935). Control methods applied at local scales (Wilson, 1969) were not effective (Robertson et al., 1995).
Bufo marinus had been widely introduced to islands across the Caribbean and Pacific as an insect pest-control agent. Based on some favourable anecdotal accounts (Mungomery, 1935; Oliver, 1949; Easteal, 1981) and previous data from Puerto Rico on gut contents of B. marinus (Dexter, 1932), 101 toads (51 females and 50 males) were imported into Australia from Hawaii in 1935 by the Meringa Sugar Experimental Station (Easteal, 1981). Two years later, 62 000 juvenile cane toads were released between Bundaberg and Mossman in the far northern Queensland (Easteal, 1981; Easteal & Floyd, 1986). Even though the original common names for B. marinus were ‘giant toad’ or ‘marine toad’, the name ‘cane toad’ is derived from the original purpose of pest control in sugarcane crops.
Since their release in far north Queensland, the toads have rapidly expanded their range westward at an accelerating rate (Urban et al., 2008) and the current distribution of B. marinus is shown in Fig. 1. Whether this increasing rate of spread is due to evolutionary changes in the cane toad population, which selected for better dispersers, or whether it is a result of the environment in the wet dry tropics being more favourable for cane toad dispersal than areas further south and east remains unclear (Phillips et al., 2006, 2007; Urban et al., 2008). Expansion to the south has been considerably slower (van Beurden & Grigg, 1980) because of cooler winter temperatures. However, climate warming is expected to expand the southern range of cane toads in Australia (Kearney et al., 2008).
Impacts of toads on native animals and communities
Characteristics of B. marinus that contribute to its recognition as a pest include its ability to breed in large numbers, short development time, toxicity, predation and it being a vector for at least two diseases.
Cane toads can produce large numbers of eggs (up to 35 000 per female) and can breed multiple times per year. Egg and tadpole development is relatively fast (1–3 days; 15–70 days, respectively; Straughan, 1966; Floyd, 1983) and sexual maturity can be achieved within 1 year (depending upon environmental conditions). These attributes together with a lack of natural predators and the ability of adult toads to live for many years (Zug & Zug, 1979) can result in toad densities in excess of 2000 animals ha−1 (Freeland, 1986). Another characteristic contributing to the toads being successful invaders is their skin, which is resistant to desiccation allowing them to inhabit arid conditions that predominate in tropical Australia during the dry season (Duellman & Trueb, 1986). Toads also have a prodigious non-discriminatory appetite, including a wide variety of invertebrates and small vertebrates including native frogs (Easteal, 1993; Boland, 2004; Crossland et al., 2008).
The toad toxin is a complex mixture of compounds comprised of a family of cardioactive steroids, collectively known as bufadienolides (bufagins) and bufotoxins, the catecholamines epinephrine and nor-epinephrine and indolealkylamines (Chen & Kovarikova, 1967; Meyer & Linde, 1971). The toxicity generally associated with toad poisoning results from the binding of the cardiotonic steroids to the membrane-associated sodium–potassium pump (Na+ K+ ATPase). Disruption of the Na+ K+ ATPase results in a positive inotropic effect or increased force of contraction of the heart, which leads to cardiovascular constriction and failure (Lim et al., 1997; Chi et al., 1998).
It has been reported that the combined effects result in population-level declines (Phillips, Brown & Shine, 2003; Oakwood, 2004). However, robust quantitative data on declines in native predators associated with cane toad arrival are scarce (Griffiths & McKay, 2007; Doody et al., 2009; Letnic, Webb & Shine, 2008). Doody et al. (2009) demonstrated severe declines (∼80–90%) in three species of lizards co-incident with toad arrival at two sites, contributing to the upgraded listing of two of these to ‘vulnerable’ in the Northern Territory (Ward et al., 2006a,b). In contrast, Doody et al. (2009) showed that individual effects did not translate into population-level effects in freshwater crocodiles. However, data presented by Letnic et al. (2008) showed that B. marinus can cause declines of crocodile populations indicating that the impact of B. marinus at population levels is most likely a complex system of interacting ecological factors. The character of ecological communities is also changing in northern Australia as a direct result of cane toads reducing populations of native predators. For example, the reduction in predatory lizards associated with cane toad arrival resulted in increases in prey species of those lizards (Doody et al., 2006; Doody et al., 2009; C. Manolis, pers. comm.).
In summary, the cane toad is recognized by the IUCN (the World Conservation Union) and Invasive Species Specialist Group as one of the world's 100 worst invaders (Lowe, Browne & Boudjelas, 2000). In 2006, the Australian Federal Government officially recognized the cane toad as a ‘key-threatening process’ under the Environment Protection Biodiversity Conservation Act (DEH, 2005). The informal and formal recognition of the toad as an invasive pest in the Australian environment has resulted in virus-based biocontrol strategies and some alternate approaches. Here, we discuss the positive and negative aspects to a number of activities that have been proposed and researched as potential biocontrols of cane toad populations. These include viral-based biocontrol (self-disseminating genetically modified organisms (GMOs), other biological control options including future strategic options.
Viral-based biocontrol strategies for B. marinus
Naturally occurring self-disseminating viruses
Successful biocontrol of vertebrates using pathogens is rare for vertebrate pests. Indeed, only two successful examples exist; the use of two viral biocontrols for the European rabbit Oryctolagus cuniculus, myxoma virus and rabbit haemorrhagic disease virus (RHDV), most notably in Australia (Fenner & Ratcliffe, 1965; Fenner & Fantini, 1999; Cooke, 2002). However, a closer examination of their success suggests caution for extrapolating it to other pest vertebrates such as the cane toad. Firstly, specificity and mortality rates combined with high transmission rates made both viruses very effective biocontrol agents. Myxoma virus initially had a 99% mortality rate in both laboratory and wild rabbits in Australia, although its impact has reduced in the wild as the virus and rabbit co-evolved. Similarly, RHDV also had a mortality rate in inoculated wild or laboratory adult rabbits of around 90%. Secondly, neither virus was discovered by mounting a search. Myxoma virus was discovered by chance in domestic European rabbits held in a laboratory in Uruguay (Fenner & Fantini, 1999) and RHDV arose in China as a new disease in rabbits (Cooke, 2002).
The only other example of an infective vertebrate biocontrol agent has been the use of feline panleucopaenia virus (FPV) in cats on Marion Island (Berthier et al., 2000). In this case, the cats were all immunologically naïve to FPV and were the only species on the island susceptible to the virus. Apart from some ongoing work on Koi-herpes virus against European carp in Australia, no other infectious agents have been discovered for use against vertebrates, despite a long list of problem invasive vertebrate species and substantial efforts expended to find such agents.
From 1990 to 1993, an intensive search was undertaken in Venezuela for the identification and isolation of viruses of native B. marinus. The search was based on the hypothesis that the 101 cane toads introduced into Australia would be immunologically naïve to any such viruses. Seven viruses were identified and isolated, all belonging to the family Iridoviridae and genus Ranavirus. Animal transmission trials at the Australian Animal Health Laboratory demonstrated that while these viruses could kill up to 100% of B. marinus tadpoles, they were also lethal to Australian frogs (Hyatt, Parkes & Zupanovic, 1998). Similar experiments with frog virus 3 (family Iridoviridae, genus Ranavirus) generated similar results. A range of bacteria were also identified but were likewise, deemed unsuitable for the biological control of cane toads. Speare, Freeland & Bolton (1991) have reported the presence of a possible iridovirus in the erythrocytes of B. marinus in Costa Rica; although this has not been followed up, it is highly likely to be an erythrocytic virus common to poikliothermic vertebrates. Whether the agent was species specific remains unknown but it would appear unlikely.
As there is no known pathogen that is specifically lethal to cane toads, the development of a GMO that would be both specific and lethal to cane toads has been actively explored. Such an agent could have the advantage of being ‘designed’ to deleteriously impact on multiple B. marinus targets (e.g. hormones and enzymes) at different life stages and, if necessary, could also be altered to overcome any evolutionary divergence of the species.
However to release any self-disseminating organism in Australia, GMO or not, requires extensive risk assessment for species specificity to ensure minimal risk to >200 species of native frogs (Tyler & Knight, 2009). Release of an infectious self-disseminating cane toad biocontrol agent also poses significant risk of disease transfer to countries where other Bufo species are endemic and where other susceptible and desirable non-target species may be present. While attenuated pathogens (infectious pathogens that do not cause morbidity or death) might be considered as a means to reduce risk to non-targets, they could still revert to virulence with passage through wild populations. As with live attenuated vaccines released for commercial use, comprehensive testing would be required to ensure that a modified and attenuated virus would not be altered on passage through susceptible species.
After considering all these issues, we sought to determine whether the development of a safe and species-specific GMO biological control agent for B. marinus was feasible. The chosen approach involved the development of a virus capable of lethally infecting cane toad tadpoles to significantly reduce their numbers across Australia. It was intended that the virus itself would be innocuous to toads and other wildlife, but would carry cane toad-specific gene(s). When tadpoles were infected with the recombinant virus, adult proteins encoded by the cane toad-specific gene would be expressed in naïve tadpoles, thereby, inducing an autoimmune response to the protein. This would in turn compromise the survival of the toads. This approach relied on an observation by Maniatis, Steiner & Ingram (1969) that Rana catesbeiana tadpoles injected with adult haemoglobin developed into metamorphs expressing an altered form of haemoglobin compared with control animals.
DNA micro-array analysis was first used to identify genes that are upregulated in the later stages of metamorphosis (Halliday et al., 2008). Many genes, including adult haemoglobin and the gut-associated genes – gastrokine (GKN) and trefoil factor (TFF), were identified using this approach. These three genes in particular were considered to have critical physiological function(s) and were used to produce recombinant viruses to test whether cane toad development can be modified immunologically, as in Rana by either a direct injection of purified proteins, or delivery using a genetically modified viral vector.
Bohle iridovirus (BIV: family Iridoviridae, genus Ranavirus) was chosen as the viral delivery system for the cane toad target genes. The ability of BIV to infect cane toad tadpoles was already established (Hyatt et al., 1998) and in addition, the virus had been isolated in Australia (Speare & Smith, 1992). Initially, a recombinant BIV (rBIV) carrying the adult haemoglobin gene was constructed (Pallister et al., 2007) and used to infect cane toad tadpoles. Although a large number of animals (89 test animals and 51 control animals) were infected with various doses of the rBIV, no difference was detected either in metamorphosis or the expression of adult haemoglobin between treated and control animals (Pallister et al., 2008).
Subsequently, two additional rBIV were constructed, carrying either GKN or TFF. These genes are expressed later in development than adult haemoglobin, potentially allowing more time for the development of an immune response. Antibodies to GKN and TFF, generated in rabbits and chickens, were initially directly injected into cane toad tadpoles. Analysis of morphological data indicated a significant increase in the stomach length of treated metamorphs in both trials, but metamorph survival was not affected (Shanmuganathan et al., 2008). Assessment of viral delivery of GKN or TFF has yet to be attempted.
In summary, targeting the immune system of cane toad tadpoles with rBIV does not appear to be an effective strategy for the biocontrol of cane toads. Possible reasons include inappropriate target selection and/or the biology of cane toad metamorphosis (e.g. short development time). Further research may resolve some of these problems but it is unlikely that the immune strategy will produce a viable cane toad biocontrol. In addition, before or in conjunction with further research involving GMOs, it will be important to conduct a risk assessment involving the release of such agents. Such an assessment would involve determining the risk of transmission of GMOs to other animals, the associated risk of infection and disease and the inadvertent movement of infected animals and GMOs to other countries.
Other biocontrol options
There are three main issues to address when considering alternative biological controls for cane toads: the life-history stage that should be targeted; whether the control agent is required to stop the invasion of toads into a new area; or whether it is intended to reduce local and/or trans-continental (Australia) toad population densities (McCallum, 2006).
Within the broad area of ‘biological control’, a range of potential strategies have been proposed. In addition to conventional viral pathogens, the strategies suggested include sterile male release (Mahony & Clulow, 2006), daughterless male release (Koopman, 2006; Thresher & Bax, 2006) and release of parasitic lungworms (Dubey & Shine, 2008). McCallum (2006) recently evaluated these approaches with the intent of identifying which of them is theoretically feasible as a control strategy. The evaluation was based upon whether the intended goal was prevention of further invasion or reduction of the density of existing populations.
Sterile male release is a species-specific method of population control that relies on the mass rearing, sterilization and release of a large number of individuals with a fitness equivalent to wild-type animals. Released sterile males compete for and mate with wild females, reducing their reproductive output and, ultimately, if enough sterile males are released for a sufficient length of time, eradication of the population is achieved. It is usually applied to prevent the establishment of pest insect populations, with large numbers of sterile males being released to inundate a relatively small invading population so that most matings produce no offspring (Knipling, 1979). Simple models suggest that both sterile male release and daughterless male release strategies would not be applicable to the biocontrol of cane toad invasions, unless the male release was several orders of magnitude greater in size than the existing population (McCallum, 2006).
Daughterless male release is a sex-skewing technology based on limiting the number of females in the wild population. It is achieved via producing a genetically modified strain of animals that are capable of producing only male offspring. Theoretically, this would occur through male tadpoles developing into fertile males, and female tadpoles reversing their sex and also developing into fertile males. If successful, over the time it would become more difficult for males to find mates with which to breed. Daughterless male release is more feasible as a biocontrol strategy to reduce existing populations because the effect of the release continues beyond the lifespan of the original modified individuals (Schliekelman, Ellner & Gould, 2005) and it may even be capable of eliminating local populations (Thresher & Bax, 2006). However, elimination of toad populations requires local populations to be closed, for the daughterless concept to be effective. Clearly, there would be very strong selective pressure for females capable of choosing fully fertile rather than the daughterless males, which might also undermine long-term control.
In principle, control of established populations through a species-specific viral control agent, whether obtained from cane toads or their near relatives or created through genetic manipulation is feasible. Given that most pathogens have a threshold host population below which they are unable to become established, virally vectored control is much less likely to be a suitable strategy for preventing invasions. Because there is evidence of intense intra-specific competition in the tadpole stage (Lampo & De Leo, 1998), a pathogen affecting the adult stage is most likely to reduce population size than one affecting tadpoles. However, given the experience with rabbits and myxomatosis, very strong selective pressure both on the host to develop resistance to the pathogen and on the pathogen itself to reduce pathogenicity is to be expected, which would limit the long-term effectiveness of any such control (Hyatt et al., 2008).
Many approaches have been explored for reducing the numbers of B. marinus in Australia. These include local control measures such as physical removal, physical barriers and the use of sex and tadpole alarm pheromones; transcontinental control measures such as a GMO virus or cane toad-specific lungworms and lastly, local and transcontinental control measures such as, toad-specific poisons, daughterless male and sterile male release. In a review, Shannon & Bayliss (2008) concluded that removals, barriers and sex pheromones would not be effective in large scale. The potential success for a toad-specific poison, disruption of lifecycle by antibiotics/probiotics and sterile male release was thought to be low. Continued research using GMOs was seen as requiring further long-term research, and because of specificity risks, along with the lack of public and political acceptance of a GMO, the potential success for this approach was rated low to medium.
Shannon & Bayliss (2008) recommended an integrated control approach involving (1) the release of a large number of small sterile males (called teacher toads) in front of the invasive front, so that native predators will learn to avoid toads before the arrival of larger, more lethal cane toads at the front; (2) deliberate infection of sterile males with lungworm parasites (Rhabdias sp.) to disseminate the parasites in establishing toad populations more rapidly than would otherwise occur; (3) concomitant use of an alarm pheromone as a broad-scale chemical-control method to increase the mortality of early life stages. These recommendations were based on the work reported by Dubey & Shine (2008), Webb et al. (2008) and Shine (2008). However, lack of epidemiological knowledge and the known long-term presence and distribution of Rhabdias in Australian cane toads, environmental risk due to the release of large number of sterile males, problems associated with generating large numbers of sterile males and the lack of expert review of the feasibility of such an elaborate strategy are just a few of the concerns raised for this approach.
Future control options
Classical biological control
The standard process for the selection of classical biological control agents is to define the level of specificity required for release in the target country, and to use this as the basis of conducting searches for potential control agents. This approach has never been explicitly applied to the search for biocontrol agents of cane toads and merits consideration. The specificity required for cane toads in Australia would be at the level of the genus Bufo, as Australia has no native toads. Therefore, searches could target other Bufo species in off-shore geographical locations where, preferably, multiple Bufo species are present and B. marinus is absent or has recently disappeared (Peacock, 2006). The premise for such a search is that if Australian cane toads have not been exposed to the microbiological fauna associated with such off-shore Bufo species, then they may be highly vulnerable as demonstrated with the myxoma virus in rabbits.
One potential solution for the problems associated with the use of an infectious disseminating agent is to use a non-disseminating, or replication-incompetent biocontrol agent. A non-disseminating (or replication-incompetent) virus is one that can enter a cell but cannot produce infectious progeny capable of infecting new cells. A non-disseminating virus, therefore, retains the ability of a disseminating, or replication competent, virus to infect cells and stimulate the immune response while ameliorating some of the safety concerns associated with the uncontrolled spread of a replication competent virus.
Non-disseminating viruses are basically of two different types: (1) gene-deleted viruses – have one or more genes deleted from the genome, at least one of which is required for virus replication; (2) virus-like particles (VLP) – these are artificially constructed particles, where an expression system, often baculovirus, is used to express the capsid protein of a virus. The capsid proteins expressed then spontaneously aggregate into structures that are similar to the original virus but lack any genetic material. The VLP is capable of infecting a cell, but cannot undergo any part of the replication process.
One of the primary questions relating to this approach is that of dose. In general, an effective dose of a non-disseminating agent is higher than that for a self-disseminating agent. What dose of a non-disseminating agent would be required for effective immunization and would this dose be achievable in the wild? Such viruses can be administered via baits in a predetermined amount known to stimulate a response, for example, the recombinant vaccinia vaccine used to vaccinate foxes against rabies was delivered in a bait that carried 108 tissue culture infectious doses of the recombinant vaccinia virus (Brochier et al., 1990). Obviously, this type of approach would be restricted to local control strategies.
Interfering small RNA (RNAi)
RNAi also known as gene silencing is a technology that can theoretically lead to ‘switching off’ targeted genes in the cells of plants and animals. ‘Switching off’ is achieved via the functional inactivation of a specific gene by a corresponding double-stranded RNA that causes either inhibition of translation or breakdown of the complimentary mRNA (encoded by the gene).
This technology could be used to target critical physiological functions (e.g. specific hormones controlling water regulation and reproduction, enzymes involved in metabolic and ion transport and proteins involved in the maintenance of the physiology such as gut maintenance proteins) in a range of toad life stages and potentially, could be delivered in non-infectious baits. Another advantage of this technology is that the RNAi products could be designed with species specificity and, if successful, could be incorporated into an infectious agent (Hyatt et al., 2008).
Cane toad Na–K–ATPase ion pump
All animals need to move ions in and out of their cells. In amphibians, such as cane toads, the movement of sodium and potassium ions is controlled by the Na–K–ATPase pump. Investigation of cane toad genetics has revealed that one section of the DNA coding for the pump has a different sequence in cane toads compared with other amphibians, leading to a different amino acid sequence in that section. It is thought that this sequence difference allows toads to avoid the harmful effects of their own toxin that all life stages of the toad produce. It may be possible to develop a designer drug to target the cane toad-specific section of the ion pump; this would theoretically lead to its inactivation and to the rapid death of the toads (Hyatt et al., 2008).
Because of the remoteness of areas affected by the cane toad, and the continental scale of the problem, it appears that the ‘holy grail’ for B. marinus biocontrol would be a self-disseminating agent that is highly effective and species specific, ideally targeting adult animals. To address this challenge, several innovative approaches have been investigated, using cutting-edge molecular biology and genetics to target, for example, metamorphosis or population sex ratios, and more have been suggested. So far these approaches have not resulted in a tool for a continental-scale reduction of cane toads. However, they represent pioneering work in the field of biocontrol research and have huge potential to open up new methods for successfully reducing the impact of any invasive species. At this point in time, there are no effective tools available to significantly reduce Australian toad populations on a landscape scale. Without continued investment in innovative research, cane toads will continue to advance across the Australian landscape and arguably continue to detrimentally impact on the Australian environment for decades to come.
The authors wish to thank Dr Peter Kerr, CSIRO Entomology, Black Mountain Laboratories, ACT, Australia for his constructive comments on this paper.