Invasive cane toads Rhinella marina have had severe ecological impacts in Australia. The toads brought with them a native-range (New World) parasitic lungworm Rhabdias pseudosphaerocephala that can kill metamorph toads, stunt the growth and reduce the stamina of both metamorph and adult toads. No cases of natural transfer to native frogs have been reported, but experimental studies have shown that metamorphs of one native anuran (the green tree frog Litoria caerulea) are capable of maintaining lungworm infections, without reducing frog survivorship. Thus, we might be able to increase the distribution of the lungworm, and its prevalence in toads, by deliberately infecting green tree frogs. In laboratory studies, we found that the toad lungworm had no detectable effects on growth or survival of adult green tree frogs. Worms that developed in the lungs of the frogs, passed out in the frog's faeces, and were able to infect toads, reducing their stamina. Synthesis and applications: Our results are encouraging for the possibility of using the green tree frog as a Typhoid Mary (to carry parasites to invasive toads without itself suffering any ill effects due to the parasite's presence), but this management approach raises significant issues of statistical power (we can never be sure that lungworms have absolutely no effect on frog viability) and animal welfare (are we justified in infecting the native species to help control the invasive species?).
We urgently need novel ways to control invasive pests. Introduced species have had severe ecological impacts on native fauna and flora in many parts of the world, and increasing globalization likely will result in increasing rates of translocation (Cox, 2004; McKinney, 2005). Classically, biocontrol of introduced pests involves the use of pathogens (or predators) from the native range of the invader to achieve target-specific control (Thompson, 1930; Gnanamanickam et al., 2000). However, given the impact of many of these agents on non-target organisms (Simberloff & Stiling, 1996), an alternative approach is to use species that are native to the invaded area as predators or competitors of the invader (Sheldon & Creed, 2003; Corbin & D'Antonio, 2004; DeRivera et al., 2005; Santos et al., 2009). The strong host specificity of many parasites makes them well suited to invader control, as demonstrated by the use of parasitoid insects (usually wasps and flies) to control commercially harmful flies and aphids (Gnanamanickam et al., 2000). Some nematodes are also reported to be successful in controlling arthropods, being extensively used in the US against mole crickets (Jackson, Alves & Pereira, 2002), and wood wasps in several countries (Hajek, McManys & Júnior, 2007). In the current paper, we explore the possibility that we might be able to use native taxa to enhance the negative effects of an exotic parasite on the alien host species.
The cane toad, Rhinella marina, was introduced to Australia in 1935. The major mechanism of toad impact on native fauna is via poisoning of native predators (Shine, 2010). The spatial extent of toad impact has been increased by the toads’ accelerated rate of invasion through the Australian tropics (Phillips et al., 2006, 2008; Phillips, Brown & Shine, 2010a). In turn, that acceleration of dispersal is due at least partly to greater locomotor endurance in invasion-front toads than in conspecifics from long-colonized areas (Llewelyn et al., 2010), perhaps as a result of escape from locomotor-curtailing lungworm parasites (any pathogen that slows toads down tends to be left behind the increasingly fast-moving front; Phillips et al., 2010b). Rhabdias pseudosphaerocephala is a parasitic nematode of the cane toad; native to toads in Central and South America, the parasite was presumably introduced to Australia within the toads (Dubey & Shine, 2008). This lungworm reduces the viability of cane toads via several pathways (reduced locomotor performance, growth, survivorship), in adult as well as metamorph life stages (Kelehear, Webb & Shine, 2009; Kelehear, Brown & Shine, 2011; Pizzatto & Shine, 2011a,b). These negative effects suggest that the parasite might be used as one component of a biocontrol program for cane toads in Australia (Dubey & Shine, 2008; Kelehear et al., 2009). Clearly, that option depends upon the toad parasite either not spilling-over to native frogs, or not having negative effects on any native frogs that do become infected. Experimental studies suggest that even if young frogs are exposed to high densities of infective larvae in the laboratory, most species do not develop persistent infections (i.e. the nematode larvae dies before reaching the host's lungs), nor show any decrease in either survivorship or locomotor performance (Pizzatto, Shilton & Shine, 2010; Pizzatto & Shine, 2011a). However, two closely related species of tree frogs are capable of developing and maintaining lungworm infections, with worms reaching maturity and breeding in the lungs (Pizzatto & Shine, 2011c). In laboratory experiments, one of these species (Litoria splendida) was badly affected by lungworm infection, whereas the other (L. caerulea) appeared to experience only minor effects, which may not significantly reduce the frog's ability to conduct its usual activities (Pizzatto & Shine, 2011a,c).
The ability of Litoria caerulea to host the cane toad lungworm, but apparently not be severely affected by it, suggests a novel approach to toad management. We might be able to build up parasite densities, and hence help reduce densities of toads by deliberately infecting the native frogs with the toads’ parasite. Three primary critical assumptions of that approach are that (1) the parasite has no detrimental effect on the frogs; (2) the frogs are capable of maintaining a lungworm infection long term; and (3) infective larvae passed in the frog's faeces are capable of infecting toads and of reducing their viability. The present study was designed to evaluate the plausibility of these assumptions.
Material and methods
The invasive cane toad
Cane toads Rhinella marina (Bufo marinus in most previous literature) are highly toxic bufonid anurans native to the New World but introduced to Australia in 1936 in a futile attempt at biocontrol of agricultural pests (Lever, 2001). The toads have now spread throughout most of tropical Australia, killing many native predators that attempt to eat them (Shine, 2010). The high fecundity of toads renders physical removal ineffective, stimulating a need for biologically based control (Shine & Doody, 2011).
The native green tree frog
Green tree frogs Litoria caerulea have a very broad distribution across mainland Australia and are sympatric with cane toads throughout the toads’ Australian range. In addition to this geographic overlap, these frogs resemble cane toads in several other ecological traits (Zug & Zug, 1979): they are common in disturbed habitats (Tyler & Knight, 2009), spawn in ephemeral water bodies (Watson, Davies & Tyler, 1995; tadpoles of both species co-occur in our study area: Cabrera-Guzmán, Crossland & Shine, 2011), are often active on the ground and in low substrates (L. Pizzatto, unpubl. data), and also have many trophically transmitted nematodes (Kelehear & Jones, 2010), suggesting similar diets. Tadpoles of this species strongly suppress growth and survival of cane toad tadpoles in experimental conditions, which has prompted the idea that we might be able to reduce rates of toad recruitment by increasing the abundance of green tree frogs (Cabrera-Guzmán et al., 2011). If we could use this native species to increase toad lungworm densities also, the effectiveness of green tree frogs in toad control might be even greater.
Sources of animals
Lungworm-free cane toad metamorphs were raised in captivity from eggs laid by adult toads collected in Middle Point, Northern Territory of Australia (12°61'S, 131°30'E). Tadpoles were raised in large (650 L) outdoor aquaria (110 × 110 × 90 cm), and fed on pre-frozen lettuce, fish flakes and algae pellets until metamorphosis. Metamorphs were used in the experiments within 3 days after completing tail absorption.
Twenty adult green tree frogs (> 30 g) collected in Middle Point were weighed, individually marked by toe-clipping and treated with the antiparasite drug ivermectin (Ivomec, 0.002 mg g−1 of frog mass, injected subcutaneously) to eliminate parasites, including native lungworms (Rhabdias spp.; Poynton & Whitaker, 2001). Any such parasites might otherwise affect, or mask, the influence of R. pseudosphaerocephala on their hosts. These frogs were maintained in groups of five (sorted such that mean body mass was similar among all groups) in outdoor containers (110 × 110 × 90 cm) with screened lids, under natural illumination and temperature cycles [27.3 ± 0.77°C standard error (SE); range: 10.4–38.4°C]. Each container had a pool of non-chlorinated water, branches with living leaves and sections of plastic guttering for refuge. Ad libitum, cockroaches dusted with vitamins and calcium (Repti-Vite, http://www.masterpet.com/en-AU/; Rep-Cal Calcium with vitamin D3, phosphorus-free ultrafine powder, http://www.repcal.com) were placed in the containers every third day. A second dose of ivermectin was administered 15 days after the first to guarantee disinfestation. Regular checks of the faeces, and dissections of pre-injected and treated individuals in prior studies, confirmed the effectiveness of this de-worming approach. We microscopically inspected faeces to confirm the absence of lungworm larvae prior to the experimental infections. These frogs were used in the experimental infections 45 days after the second dose of ivermectin, when they were again susceptible to new infections (Pizzatto & Shine, 2011b).
We obtained infective larvae of the lungworm R. pseudosphaerocephala by culturing eggs of worms recovered from the lungs of adult toads collected in Middle Point (local prevalence 15.1% of infected toads, and intensity 24.0 worms ± 7.58 SE, n = 192; C. Kelehear, unpubl. data) and euthanized by an overdose of sodium pentobarbital (50% Lethabarb diluted in saline, 1 mL kg−1) injected into the body cavity. The worms were lacerated in a Petri dish to release eggs, which were then collected and smeared in moist toad faeces, and kept at 28°C. As soon as infective larvae developed, they were washed from the cultures to a clean dish, and used in the experimental infections of adult green tree frogs and toad metamorphs (for methodological details, see Kelehear et al., 2009). Pilot studies suggest that larvae obtained by this method were equivalent in their infectivity and impacts to larvae obtained from toad faeces.
Infections of adult green tree frogs
Two containers of green tree frogs were used as controls and two as the worm-exposed treatment (n = 10 animals in each treatment group, five per container). We gently opened the mouths of the frogs with a guitar pick, and used a syringe to discharge 0.4 mL of water containing lungworm infective larvae. Frogs in the control group were given 0.4 mL of clean water, using the same protocol. This procedure was repeated 2 weeks later using newly developed infective larvae to ensure successful infections. We collected and cultured (as previously described) the faeces of frogs from both groups to search for the presence of worms. All frogs were reweighed 120 days post-exposure (DPE).
Infections of toad metamorphs
Ninety captive-raised toad metamorphs were weighed (M0; to the nearest 0.001 g), and we assessed their sprint speeds and endurance [time to exhaustion (TE)] in locomotor trials, to quantify a component of anti-predator behaviour. For this test, we placed each toad in a 1-m long wooden gutter track, lined with moist paper towel. Mimicking a predator attack, we stimulated the toads to continue hopping by touching their posterior body with a stick. We scored the time each individual took to hop each of the three 25-cm long sections, and then the total duration of activity prior to exhaustion (TE0; defined as when the animal refused to hop after three touches). All locomotor trials were conducted between 1000 and 1530 h, air temperatures averaged 27.3°C ± 0.11 SE and air humidity 80.0°C ± 1.68 SE.
The toads were then individually marked (by toe-clipping) and housed in groups of five in 33 × 21 × 11 cm plastic boxes fitted with well-ventilated lids. The boxes were placed over a plastic tray containing iodine scrub (Betadine Surgical Scrub; 7.5% weight/volume povidone-iodine) to avoid contamination, and kept indoors under a UV light tube (natural light/dark cycles), at 26.5 ± 0.05°C SE. Each container had a substrate of pre-boiled soil and tree bark (for refuge), and a pool of non-chlorinated water. Each box was randomly allocated to one of three treatments (six boxes per treatment, five toads per box), which were added to the soil: (1) 3 mL of non-chlorinated water (controls); (2) 3 mL of non-chlorinated water containing about 1500 R. pseudosphaerocephala infective larvae from worms developed in the lungs of adult green tree frogs; and (3) 3 mL of non-chlorinated water containing about 1500 R. pseudosphaerocephala infective larvae from worms developed in the lungs of adult cane toads.
Boxes were checked daily for mortality, and toads were fed ad libitum on termites every third day throughout the experiment. Because the effects of the parasite may change during the infection process, we re-assessed the hosts’ body masses and locomotor performance at 5 days post-lungworm exposure (DPE; at this early stage, larvae are migrating through the hosts’ tissues), at 10 DPE (when some immature worms may have reached the lungs, and hosts may have developed an immune response), and at 43 DPE (when adult worms are already feeding and breeding in the lungs; Pizzatto & Shine, 2011b,c). At 43 DPE, toads were euthanized by immersion in 4 g L−1 buffered tricaine methanesulfonate (MS222; Sigma-Aldrich, St. Louis, MO, USA), and their lungs inspected for the presence of worms.
Our experimental design (keeping frogs in groups rather than separately) introduces analytical complexities because the data for one anuran are not independent of data for other animals in the same enclosure. We kept the animals in groups to expose them to infective larvae multiple times (from their own faeces and from conspecifics that were kept in the same box), maximizing both prevalence and intensity of infection. In pilot studies, anurans kept singly were rarely infected, precluding us from distinguishing whether the infection method was unsuccessful or if the frogs were not suitable hosts (L. Pizzatto, unpubl. data). Increasing the intensity of infection also maximized our ability to detect negative effects of the parasites (such effects likely would be small or non-detectable if the animals hosted few worms; see Goater, 1994; Goater & Vandenbos, 1997). Our aim was to identify possible negative impacts of infection overall, and not to differentiate between the effects of primary and secondary infection, or of differing levels of infection. Maintaining the animals in groups also mimics common conditions in the field, where high densities of metamorph toads are concentrated around the moist edges of natal water bodies (Freeland & Kerin, 1988; Child et al., 2008), and where green tree frogs sometimes aggregate in shelter sites during dry conditions (L. Pizzatto, pers. obs.). To compensate for the non-independence of our data, we included the potential effects of containers within all of our analyses (see details in the analyses section).
Infections of green tree frogs
We compared the body masses of control versus parasite-exposed frogs at 120 DPE (M120) by three-factor analyses of variance, including the frogs’ initial body mass (M0) as a covariate and enclosure (nested with treatment) as a random factor. Data on body mass were log-transformed prior to analyses. Average prevalence (average % of infected individuals) and intensity (average number of parasites in infected individuals; see Margolis et al., 1982) were quantified by inspecting the lungs of euthanized frogs (euthanasia performed as described for the adult toads).
Infections of metamorph toads
We quantified parasite prevalence (percentage of individuals carrying worms in the lungs at the end of the experiments) and intensity of parasitism (mean number of parasites per infected host; Margolis et al., 1982). We used the Kruskal–Wallis test to compare the number of worms in infected individuals among treatments, and to compare the average frequency of infected versus non-infected individuals among treatments. We used repeated-measures multivariate analyses of covariance to test the effects of the treatments (controls vs. lungworms from L. caerulea vs. R. marina) on the toads’ body mass (M), sprint time (ST = shortest time to hop one of the three 25 cm sections of the gutter) and TE. In the ST and TE analyses, we initially added body mass as a covariate, but if this variable did not have a significant effect, it was removed from the final model. Interactions between the variables and DPE were tested by the Hotelling–Lawley test. We used the Kaplan–Meier method and Wilcoxon test to compare survivorship of toads among treatments. In all analyses, we used the average values per box for all the variables we measured to correct for non-independence of data within groups, instead of treating each individual anuran as an independent unit of replication. All statistical analyses were performed in JMP 7.0 (SAS, 2007). Averages reported in the text are followed by SEs.
Infections of green tree frogs
Survivorship was 100% in both controls and parasite-exposed frogs. The frogs’ mean body mass at 120 DPE was not affected by treatment (F1,2 = 0.23, P = 0.680), but frogs that were initially larger tended to grow faster (effect of logM0: F1,15 = 35.24, P <0.0001). Parasite prevalence at the end of the experiments (i.e. 120 DPE) averaged 50 ± 10%, and intensity was 1.67 ± 0.67 parasites (range 1–5), presumably reflecting the low doses administered. As the reproductive season approached, male frogs began calling, and at the time of euthanasia, all females had well-developed eggs in their ovaries.
Infections of metamorph toads
There was no significant difference in the mean intensity of parasitism between metamorph toads infected with lungworms from green tree frogs (average ± SE: 3.1 ± 0.55; range 1–8 worms) or cane toads [3.5 ± 0.88, range 2–6 worms; X2 = 0.35, degrees of freedom (d.f.) = 1, P = 0.554]. The two groups also did not differ significantly in lungworm prevalence (toads exposed to worms developed in green tree frogs: 54.4 ± 10%, range 40–100% infected; toads exposed to worms developed in conspecifics: 55 ± 16.7%, range 0–100% infected; X2 = 0.03, d.f. = 1, P = 0.870) or anuran survivorship (including comparison with controls: X2 = 2.22, d.f. = 2, P = 0.329, Fig. 1).
Body mass of the metamorphs varied only with DPE (F3,10 = 25.44, P <0.0001) and was not affected by treatment (F2,12 = 2.25, P = 0.148), or the interaction between DPE and treatment (F6,18 = 1.05, P = 0.428, Fig. 2a). Sprint speed depended on the toad's initial body mass (F1,11 = 5.90, P = 0.033) but was not affected by treatment (F2,11 = 0.42, P = 0.666), DPE (F3,9 = 0.33, P = 0.800) or their interaction (F6,16 = 1.67, P = 0.192, Fig. 2b). Controls exhibited higher stamina than toads exposed to lungworms sourced from either green tree frogs or toads, especially at 5 and 43 DPE (treatment: F2,12 = 7.58, P = 0.008; interaction: F6,18 = 6.07, P = 0.0013; Tukey: controls > worms developed in green tree frogs = worms developed in cane toads), but TE did not change with DPE (F6,10 = 2.46, P = 0.123, Fig. 2c).
In our experiments, the toad lungworm did not affect body mass and survivorship of infected adult green tree frogs, and had no apparent impact on the anurans’ breeding behaviour and egg production. Very high intensity of infection by toad lungworms can reduce the endurance of green tree frog metamorphs, but may have little effect on the frogs’ normal activities (see discussion in Pizzatto & Shine, 2011c). Low-intensity infections, as we induced in this study, are unlikely to affect the locomotor performance of adult frogs (L. Pizzatto, pers. obs.). Our results reinforce earlier findings on the lack of major deleterious effects of lungworm infection on the metamorph stage of green tree frogs (Pizzatto & Shine, 2011a,c). Transfer of a parasite to a novel host can have variable effects on both hosts and parasites (Perlman & Jaenike, 2003). In a novel host, parasites can evolve reduced virulence and density through the generations, as seen in Drosophila–Wolbachia interactions (McGraw et al., 2002). However, this is not always the case. For example, in interactions between Drosophila and their nematodes, the parasites may have similar or more severe impacts on original hosts than on novel hosts (Perlman & Jaenike, 2003). Many studies have reported catastrophic effects of parasites in novel hosts, however, often involving introduced parasites. For example, apparently, healthy Asian fish carry a microparasite that causes spawning failure, loss of body condition and mortality of one already-threatened species of European cyprinid fish (Gozlan et al., 2005). Several severe human diseases, such as ebola, influenza and HIV, also resulted from switches from original hosts (Parrish et al., 2008).
The lungworm larvae's infectivity and effects on toads were similar regardless of whether the adult lungworms producing those larvae developed inside the natural host (a cane toad) or a novel host (a green tree frog). Worms that developed in the lungs of the new host in the current study were fully viable, producing larvae that were able to infect cane toads (see also Pizzatto & Shine, 2011c), apparently without losing virulence (i.e. they negatively impacted the host's endurance). Regardless of whether they were derived from adult worms in the lungs of toads or frogs, the lungworms reduced the endurance (stamina) of infected toad metamorphs. In combination with our earlier results, the current study suggests that it would be worthwhile to conduct additional investigations on the feasibility of experimentally infecting green tree frogs to serve as a lungworm reservoir in the field.
Many issues remain to be addressed; for example, although toads and green tree frogs are common in the same habitats, especially around houses (Tyler & Knight, 2009), whether or not lungworms in the faeces of green tree frogs would encounter and infect cane toads depends upon the spatial ecology and specific habitat use of the two species, as well as the duration of infectivity of the larvae. Second, the toad lungworm has not been found naturally infecting Australian frogs (Dubey & Shine, 2008; L. Pizzatto, unpubl. data), perhaps suggesting limited opportunities for spill-over and spill-back mechanisms. However, long-term field-based studies would be necessary to verify if host switching and/or changes in virulence occur through time, and if green tree frogs can maintain their lungworm infections under natural conditions. If infections cannot be maintained in free-ranging frogs, the parasites lose virulence after prolonged colonization of the new host, or the parasites do not spillback to the toads in natural conditions, the proposed strategy would be unlikely to work. Also, it is important to consider the possible overlap in the occurrence of green tree frogs and other taxa, such as the magnificent tree frog Litoria splendida, which is very sensitive to the lungworms (based on laboratory trials; Pizzatto & Shine, 2011c). The area of occurrence of Litoria caerulea encompasses that of L. splendida, and the two species are likely to overlap, especially in urban habitats (Tyler & Knight, 2009).
A fourth concern is that we understand too little of the dynamics of host–parasite interactions to estimate how a given increase in parasite abundance would translate into increased infection rates of toads, or how an increased infection rate would affect the viability of cane toads in the field. Although the decreased stamina of infected toads suggests that lungworm infection may reduce toad viability, we did not see the more severe effects (reduced growth and survival) documented in an earlier laboratory study (Kelehear et al., 2009). Such differences in magnitude of impact are common in studies that experimentally infect anurans (e.g. Goater, 1994; Goater & Vandenbos, 1997). Given that R. pseudosphaerocephala has been shown to reduce growth rates of cane toads both in the metamorph and adult stages, and in the field as well as the laboratory (Kelehear et al., 2009, 2011), the prospect of enhancing such negative effects by increasing parasite densities warrants further investigation. The abundance of cane toads in many areas with high rates of lungworm infection (82.8% prevalence, 16.1 worms mean intensity, ranging from 1 to 230; Barton, 1998) demonstrates that lungworms are not a solution to toad control; but in conjunction with other methods (such as pheromones and tadpole competition; Hagman & Shine, 2008; Cabrera-Guzmán et al., 2011), increasing lungworm densities (especially at the invasion front where they tend to be absent; Phillips et al., 2010b) may be a useful component of an integrated approach to managing toads and reducing their rates of dispersal.
In this study, we tested three basic critical assumptions about the conditions that must be met for the feasibility of a new management approach, without which further testing would be futile. Our results are broadly encouraging for the prospect of using green tree frogs to increase rates of lungworm infection in invasive cane toads. Adult green tree frogs maintained lungworm infections without experiencing overt ill effects on their survival or growth. The lungworm larvae passed out in frog faeces were capable of infecting toads, and thereby reducing their stamina, a key trait for the toads’ rate of dispersal. We conclude, however, with two issues that are significant impediments to using native frogs to increase infection rates of invasive toads. The first issue is the difficulty of demonstrating a lack of negative impact of lungworm infection on native anurans. We did not detect any evidence that lungworms reduce survival or growth rates of green tree frogs (and the same appears to be true of metamorphs of this species; Pizzatto & Shine, 2011a,c), but a lack of evidence does not mean a total lack of effect. Intuition suggests that high lungworm numbers would affect native frogs to some degree. Other traits that require high metabolic rates (such as calling in males; Taigen & Wells, 1985) may be negatively influenced by parasitism. Although our observations suggest little (if any) effect of the lungworms on frog breeding behaviour, this issue requires quantitative investigation. Additional studies are also required to investigate whether the lack of effects of the parasites on the frogs observed in our laboratory trials are also true under natural conditions. For example, co-occurrence of different parasite species within a host, as commonly occurs in natural situations, could result in synergistic, exacerbated negative effects on the hosts (Ezeamama et al., 2008).
This situation raises a second issue: the ethics of infecting native species in order to manage an invasive species. Numbers of native frogs do not decrease after cane toads invade (see review by Shine, 2010), and thus we might inflict significant suffering on native species that otherwise would experience little if any negative impact from the invader. Is it ethically acceptable to impose suffering on a native species in order to reduce the (catastrophic) impact of the invasive cane toad on other native taxa (such as large lizards, snakes, crocodiles and marsupial predators; Shine, 2010)? This issue needs to be canvassed with the scientific and general community before we contemplate using native frogs as Typhoid Marys to manage abundance of the invasive toad; at the moment, we would not advocate the approach except in areas of high conservation value where toads are likely to have major impacts. The greatest advantage of exploiting native frogs as parasite reservoirs likely is at (or shortly in advance of) the toad invasion front: reduced stamina from lungworm infection might slow the rate of toad advance (Llewelyn et al., 2010). In long-established toad populations, infecting toads rather than frogs might be easier, more effective and less ethically troubling than infecting native frogs. As Rhabdias spp. are widespread nematodes of anurans, consistently affecting locomotor performance (e.g. Goater, Semlitsch & Bernasconi, 1993; Marr et al., 2010), similar approaches may be useful to manage other invasive anurans.
We thank Team Bufo members, M. Greenlees and E. González-Bernal, for helping collect the tree frogs, M. Crossland for providing the tadpoles, N. Somaweera and M. Franklin for help with the experiments and animal husbandry, and the anonymous referees for comments. P. Hopkinson and his crew from the Coastal Plains Research Station (DPFPI, NT) provided logistical support. This project was supported by a grant from the Australian Research Council to L. P. (DP0984888), with approval from the University of Sydney Animal Care and Ethics Committee (L04/5–2010/2/5334), and Northern Territory Parks and Wildlife Commission (9857).