Reduced efficacy of baiting programs for invasive species: some mechanisms and management implications

Additional keywords: 1080, bait avoidance, bait-shyness, control, innate behaviour, learned aversion, mammal, personality, pest, physiology, resistance, review, temperament, tolerance, vertebrate

Poison baiting is the method of choice for vertebrate pest control in many parts of the world, being easy to use, cheap, and resulting in rapid knockdown (e.g. Innes et al. 1995; Gentle et al. 2007b; Howald et al. 2007). Baiting is the most cost-effective method for decreasing the population size of introduced pest species on a large scale, as a means of protecting livestock, increasing landscape production, and reducing predation pressure on native species (Armstrong 2004; Wright 2011). However, many pest control operations have shown significant decreases in the success of baiting over time (e.g. Quy et al. 1992; Hickling et al. 1999; Twigg et al. 2002), due to behavioural or physiological responses in the target pest population.

Baiting relies on the target species finding a bait (which, in turn, depends on the attractiveness of the bait) and then consuming it (which depends on the ‘acceptability’ or palatability of bait) and therefore ingesting sufficient quantities of toxin to provide a lethal dose (which depends on the amount of toxin in the bait and the animal’s metabolic capacity to deal with the toxin) (Saunders and McLeod 2007). Assuming that animals find baits, they may still avoid consuming them; they may not recognise the bait as food, they may show innate behavioural avoidance of baits, or show learned aversion behaviour based on previous experience or social learning. If target animals consume sublethal baits, they can show increasing physiological capacity to deal with the toxin present over subsequent generations, leading to populations exposed to baiting showing greater toxin-tolerance over time. We note that although we separately discuss behavioural and physiological mechanisms for bait-resistance, in some instances both bait avoidance and toxin-tolerance have been recorded (e.g. Quy et al. 1992).

Understanding bait-resistance is important for the sustainable control of pests worldwide, to protect wildlife and agricultural production (Hickling et al. 1999; Twigg et al. 2002; Twigg 2014). In this review, we summarise what is currently known about bait-resistance (both behavioural and physiological) in vertebrate pest species of significant economic impact in Australia and New Zealand (Table 1) and provide examples of how management actions can maximise efficacy of baiting programs. We examine the use of sodium fluoroacetate (hereafter ‘1080’; Box 1) baiting in Australia and New Zealand. We address the question of whether behavioural and physiological bait-resistance could develop in the red fox (Vulpes vulpes), an introduced predator of significant economic and environmental importance in Australia (Box 2).

Table 1.  Common vertebrate pest species that are the target of extensive baiting programs in Australia and New Zealand, and examples where evidence of bait avoidance and toxin-tolerance behaviour have been recorded
Accompanying European colonisation, 26 eutherian species have been introduced to the continent and have significantly influenced the ecology of environments, attracting a disproportionate amount of research focus (Fleming and Bateman 2016) and funding (Gong et al. 2009) aimed at minimising their ecological, social and economic impacts. Of the 31 exotic mammal species in New Zealand and its offshore islands, 14 are widespread, warranting 40% of the country’s conservation funding expenditure aiming at their control (Parkes et al. 2006). Introduced ungulates and marsupials are generally defined as pests under the Wild Animal Control Act 1977; the common brushtail possum is controlled under both a national plan under the Act for conservation purposes, and under a National Pest Management Strategy as a vector for bovine tuberculosis (Innes and Barker 1999). n.d., no data

Fig. 1.  Long-term baiting programs running across Australia. These sites tend to be focussed on a threatened species program (e.g. for rock wallabies, mallee fowl, southern brown bandicoot) or a community action group to protect livestock (Anon. 2013). Shaded area shows the distribution of the red fox across the continent (PestSmart Connect 2016b). Relative size of the programs is shown by the size of the dots: size in hectares (ha) or number of landholders (LH) participating. Locations/relative size are approximate.

Innate aversion is non-learned instinctive aversion that is a result of selection pressures on a species and refers to heritable characteristics of individuals within the target population. ‘Temperament’ refers to relatively consistent individual dispositions that underlie and modulate the expression of behaviour, resulting from genetic, epigenetic (i.e. developmental), and environmental effects (McDougall et al. 2006). The rapidly growing behavioural research field of animal personality/temperament recognises six main traits across species and studies: shyness–boldness, emotional reactivity/fearfulness, exploration–avoidance, activity, sociability, and aggressiveness (Boissy 1995; McDougall et al. 2006; Réale et al. 2007; Stamps and Swaisgood 2007). Temperament affects fitness through influences on predation rates, competition for mates and resources, and social interactions (reviewed by Smith and Blumstein 2008; May et al. 2016). For example, variation in neophobia has been linked with reproductive output in farmed silver foxes (V. vulpes) (Korhonen and Niemelä 1996), and variation in boldness has been linked with post-translocation survival in swift foxes (Vulpes velox) (Bremner-Harrison et al. 2004).

Because personality traits are heritable, baiting can result in artificial selection for particular personality traits across the population. Variation in temperament along a bold–shy or neophobia–exploration continuum results in differences in risk-prone behaviour amongst individuals of any given species, based on trade-offs between foraging gains and associated risks (Sloan Wilson et al. 1994; Ioannou et al. 2008). Such variation in behaviour influences latency to approach novel objects, latency to eat novel foods, or responses to threat stimuli (McDougall et al. 2006) and could therefore influence selective survival of individuals targeted through baiting programs. The degree of explorative behaviour could also influence vulnerability to taking a bait, with less exploratory individuals having reduced chance of encountering baits (Travaini et al. 2013).

The most marked examples of baiting influencing the temperament of animal populations are recorded in rodents. Neophobia is wariness expressed by an individual towards unfamiliar objects or foods in their familiar environment, which can result in cautious feeding strategies (Cowan 1977; Sunnucks 1998). After repeated baiting operations using toxic baits, there would be selection for survival of these neophobic individuals, thus decreasing the efficiency of control. Cowan (1977) demonstrated extreme forms of neophobia in commensal Rattus species that are likely to be exposed to traps or food containing poisons, and where avoidance of unfamiliar objects (including bait) obviously has survival value. Repeated poisoning regimes have led to selection for extremely neophobic brown rat (Rattus norvegicus) populations (Quy et al. 1992; Brunton et al. 1993).

Neophobia has also been recorded in Australian populations of the European rabbit (Oryctolagus cuniculus). 1080 baiting was first introduced to control rabbits in the 1950s. There was a major decrease in the number of rabbit kills for 1080 baiting trials between 1958–62 and 1971–75, although there was no significant increase in toxin-tolerance between 1955 and 1977 and bait-aversion was suggested as a possible reason (Oliver et al. 1982). Wariness towards 1080 baits in some Australian rabbit populations was recorded in the early 1960s (Poole 1963; Rowley 1963b), and several studies have directly observed individuals avoiding correctly presented baits (Carrick 1957; Rowley 1958). The selection for more neophobic individuals through repeated baiting was suggested as a likely cause for reduced bait efficacy recorded (Oliver et al. 1982). Similarly, rabbits were identified as showing cautious behaviour towards both non-toxic and toxic baits on initial exposure in New Zealand (Bell 1975).

The common brushtail possum (Trichosurus vulpecula) is regarded as New Zealand’s most detrimental vertebrate pest. Bait aversion due to neophobia has also been identified in New Zealand possums, with ~20% of bait-naïve possums avoiding non-toxic baits (O’Connor and Matthews 1999; Morgan 2004). These animals would therefore have greater chances of surviving exposure to toxic baits.

At an individual level, animals could learn to avoid baits if they survive an initial exposure to toxic baits and create an association of its ill effects with a sensory aspect of toxic bait – such as appearance, taste, or smell (Gustavson 1977). This is termed ‘learned aversion’, and can develop rapidly, after even a single or small number of doses (Morgan et al. 1996), but can persist within populations across multiple baiting events (Cowan 1977; O’Connor and Matthews 1996). Learned aversion threatens the current control programs for rodents (e.g. Quy et al. 1992; Brunton et al. 1993) and the common brushtail possum (e.g. Morgan et al. 1996; Ogilvie et al. 2000). Potential development of learned aversion towards baits has also been raised for foxes (Kay et al. 1999; van Polanen Petel et al. 2001; Carter and Luck 2013) and feral pigs (McIlroy 1983; O’Brien et al. 1986).

Learned aversion to poison baits has long been recognised and studied in rodent populations (see references in Table 1). Aversion can develop rapidly in rodents, with individuals learning to avoid particular flavours after a single flavour or toxicoses experience (Kalat 1974). Many rat populations have developed learned aversion to multiple rodenticide baits, making it increasingly hard to control them, as new baits to present these poisons must be developed (Prakash 1988). Rats can also develop aversion to 1080 baits if they ingest sublethal 1080 doses (Green 1946; Barneet and Spencer 1949; Peacock 1964).

An increase in the frequency and intensity of 1080 baiting to control common brushtail possums in New Zealand has been associated with a marked decrease in the success of possum kills (Hickling 1994; Moss et al. 1998). It is suspected that surviving possums develop learned aversion towards 1080 baits after consuming sublethal doses (e.g. Morgan 1990; Ogilvie et al. 1996), and it was predicted that providing a LD₂₅ dose would rapidly generate aversion in survivors (Morgan et al. 1996). In a study area baited for three consecutive years, 63% of possums avoided baits, and aversion was still evident in individual possums several years after completion of the study (O’Connor and Matthews 1999). Possums ate less non-toxic carrot baits in areas where toxic baiting with carrots was carried out six months previously (compared with unbaited controls) (Hickling et al. 1999).

Social learning influences movement patterns, predator avoidance, mate choice, foraging decisions and food choice across a wide range of taxa (Galef and Laland 2005). Through teaching or observation of conspecifics, social learning increases the likelihood and rate of transmission of behaviour through a group of animals (Galef 1985; Frost et al. 2007). In rodents and European rabbits, feeding behaviour of juveniles is influenced by that of their older conspecifics (Galef 1990). Adults can potentially instil avoidance behaviour in their offspring during foraging though demonstration (e.g. Shier and Owings 2007). Social learning could also occur where conspecifics observe the results of other individuals consuming a bait. Personality traits or temperament also play a part in learned aversion through social learning (Ross 1999): when bold individuals observe their shy conspecifics, they become more cautious, although shy individuals observing bold conspecifics show no change and remain cautious (Frost et al. 2007). This illustrates that even in cases where shyness has not been inherited, parents can teach their young to be more wary of their environment.

An animal’s physiology may protect it from the impacts of baiting. First, animals may have limited exposure to the toxins in baits due to their natural diet. Herbivores are generally assumed to avoid taking meat baits, although there is plenty of evidence that they do (e.g. Dundas et al. 2014). Second, species may also have different toxin tolerance by virtue of the mode of action of the toxin. For example, Tiliqua rugosa skinks have exceptionally high tolerance of 1080 (Twigg and King 1991), possibly because the mode of action (reducing ATP availability) has a minimal effect on ectothermic animals that already have a low metabolic rate. Third, the body mass of animals may confer an advantage, with larger individuals not receiving a lethal dose for their mass. Finally, animals may have the capacity to biochemically metabolise the toxin, removing it from their system. This final mechanism can potentially explain the increased physiological tolerance within populations of a target species over time, and we discuss aspects of biochemical toxin-tolerance below.

Heritable physiologically-based tolerance to toxins can threaten the efficacy of some baiting operations (Hickling 1994; Hickling et al. 1999). The physiological mechanisms for the breakdown and excretion of toxins can be artificially selected for by exposing populations to sublethal doses. Because of the marked selective pressure for tolerance to develop, toxin-tolerance can become established relatively rapidly. For example, anticoagulant rodenticides were developed in the early 1950s, but brown rat and house mouse (Mus musculus) populations resistant to various ‘first generation’ anticoagulant poisons (e.g. Warfarin) were already identified throughout the United Kingdom by the late 1950s to early 1960s (Buckle et al. 2010). More potent anticoagulant rodenticides (e.g. Difenacoum and Bromadiolone) were swiftly brought to the market, but heritable tolerance to the ‘second generation’ anticoagulant rodenticide Diphacinone was already recognised in 1958 in a Scottish routine field trial (Buckle et al. 2010). Toxin-tolerance was identified in some areas within a few years of the introduction of these more potent toxins (Boyle 1960; Jackson 1972; Jackson and Kaukeinen 1972; Buckle et al. 2010; Daniells et al. 2011). After initial exposure to anticoagulant poisons, some groups of individuals showed survival rates of almost 90% in subsequent feeding tests (Buckle et al. 2010).

The 1080-tolerance in Australian native species that coexist with fluoroacetate-bearing plants (Box 1) demonstrates that development of tolerance to this chemical can develop, given sufficient time. Because there is no reversal of this selection pressure (i.e. there is no cost to maintain the capacity to metabolise and excrete these compounds), in many cases these tolerances persist over 70–100 centuries after isolation from the toxic plants (Oliver et al. 1977; Eisler 1995). Comparing data for controlled, formal toxicity trials, the 1080 LD₅₀ dose in European rabbits doubled (from 0.34–0.46 mg to 0.744–1.019 mg pure 1080 kg⁻¹) over a period of 25 years for three Australian field sites that had been regularly baited since the introduction of 1080 baits in the early 1950s, with a non-significant increase in LD₅₀ for a fourth site that had the least exposure to 1080 baiting (Wheeler and Hart 1979; Twigg et al. 2002). There was also a positive correlation between the degree of 1080-tolerance and the length of exposure to the toxin, and Twigg et al. (2002) suggested that continuous ingestion of sublethal doses of 1080 was the most likely mechanism for increased toxin-tolerance. Laboratory studies have also demonstrated rapid development of 1080-tolerance (Tahori 1963; Howard et al. 1973). Laboratory rats (Rattus norvegicus) given sublethal (LD₇₅–LD₉₀) doses of 1080 over successive generations developed a 1.8-fold increase in LD₅₀ levels in five generations (Howard et al. 1973), while house flies (Musca domestica) showed a 7-fold increase over 25 generations (Tahori 1963).

Certain traits of target species are likely to increase the rate of development of toxin-tolerance. Species with r-adapted life strategies tend to be relatively successful in adapting to changes within their environment (Jackson 1972; Corbet and Harris 1991; Lynch and Lande 1993; Kolar and Lodge 2001); the relatively rapid development of 1080-tolerant populations of rodents (Howard et al. 1973) and European rabbits (Twigg et al. 2002) is likely to be facilitated by their high reproductive potential (Silver 1924; Kolar and Lodge 2001). Development of 1080-tolerance may be much slower for animals that show k-adapted life strategies, such as canids and mustelids (Twigg et al. 2002). Furthermore, presumably because reproductive tissues have high energy demands and thus require adequate supply of ATP, sublethal doses of fluoroacetate temporarily suppress fertility of a range of animals, including birds, mammals and reptiles (reviewed by Twigg and King 1991); this effect is also likely to have greater impact on k-adapted species, which have fewer reproductive opportunities.

The widespread use of successive 1080 baiting, particularly in Australia and New Zealand (Box 1), has meant that populations of target species have been repeatedly exposed to the toxin, increasing the risk for developing bait-resistance through either bait avoidance or toxin-tolerance. Bait-resistance could therefore represent a major challenge to on-going control programs, potentially reducing the effectiveness of these costly enterprises. This situation highlights the need for proactive management operations to minimise conditions that contribute to development of bait-resistance.

Increasing the toxin concentration in baits is often the strategy considered first to ensure a high proportion of successful target kills; however, this would increase risk to non-target species. Furthermore, it risks the baits becoming less palatable to target species, increasing development of bait-resistance (Staples 1969; Frampton et al. 1999), particularly if impurities in the toxin decrease palatability as well (e.g. Atzert 1971; Morgan 1982; Morgan et al. 1987). The following sections discuss management actions that may serve to reduce risk of bait-resistance without relying on increasing toxin dose.

Sublethal poisoning has been recognised as the primary cause of both learned aversion towards baits and development of toxin-tolerance (e.g. Morgan et al. 1987, 1996; Warburton and Drew 1994; Hickling 1994; Hickling et al. 1999; Saunders et al. 2000; Morgan 2004). Sublethal doses could ultimately result in target animal populations becoming less sensitive to the toxin as the larger-bodied and less sensitive individuals within the population are likely to survive. As Kinnear et al. (2017) puts it, ‘sub-lethal baits caused by microbial activity (e.g. following bait caching or rainfall) can potentially create bait-shy predators who become intractable killers’. Thus, management strategies should focus on preventing the development of sublethal baits, and also prevent target species encountering them.

Baits being presented in control operations can be of poor quality: unattractive, unpalatable, or with the incorrect toxin levels to deliver a lethal dose (Cook 1999). Dried meat baits are usually cut by hand and consequently can vary markedly in size and consistency. Substandard bait matrix can also easily be broken down into fragments, increasing the likelihood of consumption of inadequate toxin quantities (Nugent et al. 2010). There is also a 10% ‘allowable variation’ in toxin dose (Twigg 2014), with variation between baits due to the mechanical action of withdrawing the needle after each bait is injected.

Because the lethal toxin level for a specific baiting program is usually determined based on the susceptibility of only one target species, sublethal poisoning can occur if another pest species, which is larger-bodied or has a higher 1080-tolerance, is also present (Hickling et al. 1999). For example, feral cats have almost four times greater tolerance (average LD₅₀: 0.40 mg kg⁻¹) of 1080 than canids (LD₅₀ of red fox: 0.15 mg kg⁻¹; LD₅₀ of wild dog: 0.11 mg kg⁻¹) (McIlroy 1981; McIlroy and King 1990), and a 3.0-mg 1080 bait aimed at foxes is hypothetically sufficient to provide a lethal dose in only half of feral cats weighing >7.5 kg (Twigg 2014). In New Zealand farmlands, baiting is carried out for both the common brushtail possum and the European rabbit, where possums have LD₅₀ toxin levels that are at least three times higher than those of rabbits (Hickling et al. 1999). Consequently, possums are exposed to sublethal 1080 doses when they take baits intended for rabbits.

Due to partial consumption (Twigg et al. 2002; Dundas et al. 2014), many smaller animals that cannot consume a whole bait can interfere with the toxin dose present. For example, rodents may gnaw at baits, while invertebrates such as ants can also consume parts of the bait (Merks and Calver 1989). Target animals taking the partial remains of baits are therefore more likely to ingest sublethal doses.

While there are multiple explanations for development of sublethal baits (Fig. 2), in most cases sublethal baits are probably the result of environmental exposure (Twigg et al. 2002). 1080 is highly water-soluble and, over time, leaches out of baits (Twigg et al. 2002); even humid environments can cause the toxicity of uneaten baits to steadily decrease (Staples et al. 1995). Thus rainfall, soil moisture, and temperature play an important role in the longevity of 1080 in baits, both directly and indirectly affecting the activity levels of microorganisms and invertebrates (e.g. blowfly larvae) (Twigg et al. 2002; Saunders and McLeod 2007; APVMA 2008). The bait matrix (especially the ‘skin’ of the bait) influences water penetration and therefore bait longevity. Consequently, bait longevity varies markedly between different bait matrices as well as with the conditions in which they are placed (Fig. 3).

Fig. 2.  Three potential mechanisms for development of bait-resistance in controlled vertebrate species. In this paper we review physiological toxin-tolerance, as well as two avenues of behavioural avoidance of baits: population differences in innate behaviour, and learned aversion in individuals that have been exposed to baits.

Fig. 3.  Longevity of 1080 in baits is strongly influenced by whether they are buried (B; solid lines) or not (N; dashed lines). Longevity is also affected by exposure to rainfall: baits were shelf stored (black dotted line), exposed to prevailing weather under field conditions, or to manipulated rainfall (either protected from rainfall or subjected to twice the usual amount of rain). In some studies, although baits were exposed under field conditions, very little rain was recorded for these sites (‘exp. ARID’). Data are for Foxoff^® (Animal Control Technologies) baits (lines only), eggs (triangles), or dried meat baits (DMB; circles; crosses). Data are from several sources: [1] Fleming and Parker (1991), [2] Kirkpatrick (1999), [3] Saunders et al. (2000), [4] Twigg et al. (2000), [5] Twigg et al. (2001), [6] Gentle (2005). ‡, shelf-stored: 97% of dose retained at six months (APVMA 2008).

Burial of fox baits is mandatory across much of eastern Australia in order to reduce non-target impacts (Saunders and McLeod 2007). With no strict guidelines in place there are many interpretations of the term ‘buried’, ranging from the bait being covered with a thin layer or clod, to shallow depressions, to burial 10–15 cm deep (Saunders and McLeod 2007). However, burial in soil, particularly deep burial (>10 cm), significantly shortens the effective life of 1080 meat baits (APVMA 2008) due to leaching as well as biodegradation of the toxin by bacteria and fungi (Twigg and Socha 2001), increasing likelihood that the bait will contain a sublethal dose. Some baits last only one week before the 1080 has sufficiently degraded that the bait is no longer toxic to the target species (Fig. 3 and references therein). Baits that have been left in the field longer than these average times would therefore contain a sublethal dose; assuming the bait remains attractive and palatable, this can increase the risk of learned aversion developing. Buried meat baits are also taken significantly less often by foxes than are baits placed on the surface (Thomson and Kok 2002). The potential risk of non-target take can be reduced by placing baits on the soil surface and lightly covering them with leaves, sticks, or twigs, or placing them under logs or in a shallow depression. At the very least, any buried 1080 bait should be removed and replaced every seven days; Saunders and McLeod (2007) present a decision tree guiding bait choice in scenarios where replacement is warranted.

The timing of baiting also needs to be considered carefully. While standard operating procedures for the deployment of 1080 baits recommend that baits not be deployed under wet conditions, it would be important to have some idea of how long baits are likely to last in the environment to ensure that they are not deployed before significant rainfall events. Although it is impossible to retrieve grain-based baits used for rabbit and feral pig control, the importance of retrieving larger baits after control operations has been broadly recognised (APVMA 2008) to minimise the time baits remain in the environment to avoid becoming sublethal.

Incorporation of 1080 by an individual (i.e. fluoroacetate intoxication) can initiate a vomit response, most commonly observed in carnivores and feral pigs (e.g. McIlroy 1981; Sinclair and Bird 1984; O’Brien et al. 1986; O’Brien and Lukins 1988; Choquenot et al. 1996). Pen trials investigating the toxicity of 1080 to feral pigs observed high rates of vomiting in individuals that ingested 1080, and some pigs that vomited survived very high 1080 doses (Hone and Kleba 1984), which has raised concerns that sublethal dosing due to vomit responses can possibly lead to subsequent aversion (McIlroy 1983; O’Brien et al. 1986). However, additional trials, more closely simulating field conditions, demonstrated that while vomiting frequently occurred in feral pigs poisoned with 1080, it does not improve their survival as sufficient fluoroacetate had been absorbed before emesis (O’Brien 1988; O’Brien and Lukins 1988; Twigg et al. 2005). Similarly, the mortality of fat-tailed dunnarts (Sminthopsis crassicaudata) offered meat containing an estimated LD₉₀ of 1080 was only one-tenth of that in animals orally administered the same dose in water, presumably due to vomiting (Sinclair and Bird 1984). However, although all dunnarts that ingested meat laced with 1080 vomited, two individuals that vomited 100% of the poisoned meat volume still died, indicating that they had already assimilated a lethal dose (Sinclair and Bird 1984).

Consumption of vomitus by feral pigs can potentially increase the chances of sublethal dosing through secondary poisoning (McIlroy 1983). While the amount of 1080 in vomitus is likely to be similar to that of the bait itself (Gentle et al. 2005), relatively small amounts of this material (~50 g) are likely to occur in the environment (Twigg et al. 2005). There is also the potential for non-target species to ingest vomitus and therefore be exposed to poisoning, a reasonable consideration given the large 1080 doses in feral pig baits. However, it should be noted that vomiting has not been commonly recorded in studies of 1080 baiting of free-ranging feral pigs (Twigg et al. 2005), which suggests that there may be differences between caged and field experiments in the amount of bait consumed while in traps compared with their natural diet.

In the case of feral pigs, management of their vomiting behaviour after poisoning (e.g. by including an antiemetic agent that prevents vomiting: McIlroy 1983) could help to reduce learned aversion developed as a consequence of regular exposure to sublethal doses of a toxin. While Rathore (1985) found such agents to be successful in preventing vomiting, Hone and Kleba (1984) reported that the majority (70%) of pigs still vomited, although kill rate was increased (O’Brien et al. 1986; Choquenot et al. 1996). Twigg and Parker (2010) discusses the potential implications of any bait additives for non-target animals that might ingest baits.

Learned aversion to baits requires that the ill-effects of the toxin have an immediate effect on the target animals, thus allowing them to develop the association with the bait itself. 1080 is a relatively fast-acting toxin. Eastern quolls (Dasyurus viverrinus) show a vomit response within 13–49 min (mean: 26 min) of ingesting 1080, while in Tasmanian devils (Sarcophilus harrisii) it is 18–82 min (mean: 55 min) (McIlroy 1981). In foxes, symptoms are evident 30 min after bait ingestion (Saunders and McLeod 2007). These times may be sufficiently short for animals to associate symptoms of fluoroacetate intoxication with the last food item they consumed.

Learned aversion to baits also requires that the baits are sufficiently distinctive (e.g. through novelty or some distinguishing feature) for the target animal to be able to distinguish and therefore reject them. Several studies indicate that animals can discern fluoroacetate in their meals. Dietary studies indicate that crested pigeons (Ocyphaps lophotes) and grey kangaroos (Macropus fuliginosus) are able to discriminate between highly toxic and less toxic seeds/plant species, consuming less of the former, suggesting that they discern fluoroacetate (Twigg and King 1991). Two native carnivore species distinguish between baits or meat with and without 1080 added (spotted-tailed quoll, Dasyurus maculatus: Körtner and Watson 2005; fat-tailed dunnart, Sinclair and Bird 1984). Four rodent species show reduced intake of foods dosed with 1080 (Calver et al. 1989). Pen trials with feral pigs showed a significant reduction in their daily consumption of wheat when 0.05% 1080 was added, with intake returning to prepoison levels after three days (Hone and Kleba 1984). Similarly, Morgan et al. (1987) reported that while less than 5% of common brushtail possums avoided non-toxic baits, 17–22% avoided baits after 1080 was added. Such studies indicate that differences in the smell and/or taste of non-toxic and toxic baits can influence whether they are eaten or not.

The quality and suitability (acceptance and palatability) of the bait matrix can influence bait take by the target species (Twigg 2014). Oliver et al. (1982, p. 132) pointed out that ‘neophobia is likely to affect adversely all rabbit control by poison baiting whatever bait or poison is used, unless the poison can safely be incorporated into an existing and familiar food’. For example, baiting programs using dried meat baits use large quantities of meat and therefore often import meat (e.g. horse meat) slaughtered elsewhere to the site where baits are produced. Where this meat is different from the target species’ natural diet, then this will result in the bait matrix being novel in comparison to other potential foods in their environment. By virtue of their production, manufactured baits are particularly likely to trigger neophobic reactions.

While 1080 in its pure form is odourless and tasteless (Atzert 1971), in commercial form it has been identified as sometimes containing an acetic acid impurity, giving it a faint vinegar smell (Morgan 2004).

Many programs have permanent baiting sites marked in some manner to enable recovery of untaken baits at the end of a baiting cycle or redeployment at fixed locations over successive baiting cycles. Animals may learn to avoid these marked positions or animals may be alerted to the presence of baits by regular human presence at those sites. Animals that show aversion to human presence or disturbance will therefore be alerted to the presence of baits and avoidance of the general area could increase. For example, Rattus norvegicus demonstrate latency to pick up food presented in bait stations, thereby reducing effective administration of toxic baits via bait stations (Stryjek and Modlinska 2016). Baits are often also monitored by infrared camera traps, enabling identification of the species taking baits (e.g. Dundas et al. 2014), but also potentially alerting the target animals to their presence (Meek et al. 2014, 2016). In a trial on agricultural sites in south-west Western Australia, Dundas et al. (unpubl. data) found that 30% of 33 dried kangaroo meat baits monitored by camera were still present at 45 days compared with only 16% of 401 baits that were not monitored by camera.

Maximising bait uptake within days of bait deployment, reducing the amount of time that toxic baits are left in the environment and therefore subject to leaching and biodegradation, will therefore reduce the likelihood of sublethal toxin doses. Maximising bait uptake by the target species also serves to minimise risk to non-target species. Several methods have been proposed to increase rapid bait uptake.

Bait additives – in the form of dyes or other cues that can mask signals used by target animals to identify them – can disrupt the development of learned bait aversion (e.g. Thomas et al. 1996). For example, masking the smell and/or taste of the toxin with cinnamon and orange flavours was successful in masking the presence of 1080 in common brushtail possum baits (Morgan 2004). Microencapsulation to disguise the bitter taste of some toxins has been trialled (e.g. Shapiro et al. 2016).

Additives that disrupt the possibility of learned aversion have also been trialled. For example, Cook (1999) tested corticosterone glucocorticoids (which is released in response to stress) and mifepristone (another hormone that inhibits the actions of glucocorticoids) bait additives in rats, and found that high doses of mifepristone increased bait consumption and decreased aversion towards baits. Devine and Cook (1998) showed that the addition of neurotransmitter antagonists reduced development of learned aversion to baits by European rabbits.

There are several considerations in the decision of whether or not to pursue bait additives. With any additives, it is important to first establish the need to make existing baits more attractive and then test the effect of these attractants under field conditions to warrant the additional costs (Saunders and McLeod 2007). There is a risk that attractants may prolong attractiveness of baits, so that they are still attractive when they are sublethal, hence increasing risk of developing aversion. Some bait additives can have adverse effects on the mode of action of the toxin (e.g. the administration of diazepam together with 1080 resulted in decreased sensitivity of foxes to 1080: Marks et al. 2009; Twigg and Parker 2010). The implication of using bait additives also needs to be considered for non-target animals (Twigg and Parker 2010).

In addition to extra costs due to the additive itself, the inclusion of any new bait additive requires regulatory approval (e.g. Australian Pesticides and Veterinary Medicines Authority (APVMA); New Zealand Food Safety Authority (NZFSA)), which ultimately adds to the cost of bait manufacture, and therefore baiting. Although ‘evaluations of variations to available products can be shorter’ than evaluations for completely new products, the processes are acknowledged to ‘be lengthy [with] timeframes prescribed in the legislation’ (Anon 2017) (see also Springer 2011).

A period of adequate free-feeding, preceding or after non-toxic baiting, could minimise development of learned aversion (Choquenot et al. 1996; Ross et al. 2000). With prefeeding, animals are attracted to bait stations and become habituated to the scent associated with a non-lethal reward, reducing neophobia (Ross 1999); toxic baits can then be deployed (using the same scent) once the target species has become acclimated to the feeding site (Choquenot et al. 1996). Prefeeding can also increase the amount of bait consumed by the target species, reducing the likelihood of sublethal doses. Prefeeding is a regulatory prerequisite for baiting feral pigs and rabbits (unless ‘one-shot’ oats are used for rabbit-baiting – as standard practice in Western Australia: Oliver et al. 1982).

Bait avoidance in common brushtail possums has been largely minimised with prefeeding where the prefed and toxic baits were similar, reducing the ability of possums to use novel cues to form associations between the toxic bait and its ill effects (Moss et al. 1998). Ross et al. (2000) found that while postfeeding was relatively ineffective in reducing aversion development in possums, prefeeding had a significant effect: 97% of non-prefed possums developed aversion compared with only 22% of prefed possums.

Free-feeding is not recommended for all vertebrate pests. Because foxes cache baits (van Polanen Petel et al. 2001), prefeeding can have negative consequences on toxic bait uptake (Gentle 2005). Additionally, free-feeding can also train non-targets to feed on the bait trails or stations and substantially increases the cost of control programs (Saunders and McLeod 2007).

Consideration should also be given to the timing of baiting, because seasonal saturation of an environment with an abundant, palatable and easily acquired food source is likely to make baits less desirable (Saunders and McLeod 2007), and hence increase the length of time that they are left in the environment exposed to leaching and biodegradation.

Taking advantage of peak food demands for the target species may help to improve effectiveness of baiting programs, especially where the target species shows strong seasonal reproductive cycles, such as the fox (Saunders and McLeod 2007). In New South Wales, the greatest decline in fox body condition for both sexes is recorded between August and November, coinciding with the peak birth and cub raising period (Saunders and McLeod 2007). This is also the time of year when grass cover is an impediment to shooting. Late summer (February–May) coincides with the dispersal of naïve individuals (i.e. young) into the population, which are more likely to take baits than wary adults. Twice-yearly baiting has been shown in some studies to keep fox population densities low all year round (Saunders and McLeod 2007), although other studies indicate that bait coverage, even under biannual programs, can still be insufficient for effective fox control (Bengsen 2014).

A general recommendation evident in the literature is the need for variation in the control methods used, avoiding over-reliance on a single control technique (e.g. Hickling et al. 1999; Kinnear et al. 2017).

Using the same bait type repeatedly could increase the likelihood of animals developing bait avoidance associated with the particular appearance of baits. Switching between bait types has been recommended to reduce target species becoming habituated to avoiding baits (APVMA 2008). There may be some capacity for bait switching: although virtually all rabbit control is undertaken using oat products, there are grain and manufactured baits for feral pigs and several manufactured bait types for canids, which offer the opportunity for alternating bait types. Varying the appearance or smell of baits (e.g. using different meat sources for dried meat baits) could increase bait take by individuals that have developed learned aversion to a specific bait. Cage (Morgan et al. 1996) and field-based (Ogilvie et al. 1996) studies show that altering the bait matrix can partially overcome aversive behaviour. Marlow et al. (2015) found that changing from dried meat baits to Probait^® increased uptake by foxes in two Western Australian reserves. Switching between a cereal bait and gel increased efficacy in common brushtail possums (Ross et al. 2000). As well as palatability of the bait, longevity of different bait types under the specific climatic conditions experienced should be considered as part of the decision around which baits to use. For example, where the baits need to be buried to reduce non-target interference, chicken egg baits might offer a better option than meat baits (Fig. 3).

Alternating toxins may slow development of toxin-tolerance. While 1080 is a relatively fast-acting toxin, and thus allows target species to associate it or the bait with its symptoms if they do not die after ingestion, anticoagulants are comparatively slow, minimising the development of any association between consumption and adverse consequences (Henderson et al. 1997; Ross et al. 1997). Ross et al. (1997) demonstrated that while 1080 baits killed only 40% of 1080-avoiding possums, 70% were killed when alternative toxins (Brodifacoum and Cholecalciferol) were used, and 100% when bait type was also varied. Switching toxins could be the most promising technique to overcome aversion problems in common brushtail possums (O’Connor and Matthews 1999), while switching both toxin and bait matrix improved outcomes of baiting California voles (Microtus californicus) (Baldwin et al. 2016).

Several alternative toxins have been used or tested for use in Australia. Para-aminopropiophenone (PAPP) has been approved for use by the APVMA (January 2016) and PAPP baits targeting wild dogs (DOGABAIT^®) and foxes (FOXECUTE^®) are now commercially available in some states (Pestsmart Connect 2016a). This toxin has an antidote, reversing the effect of the compound if administered within 30 min. The toxin has some species selectivity, with carnivores (both native and introduced) more sensitive to PAPP than non-carnivore species (birds and humans) (reviewed by Eason et al. 2014). Considerable progress has been made with developing sodium nitrite baits for controlling feral pigs (Staples 2017). Other bait toxins used in Australia and New Zealand include strychnine, cyanide, and zinc phosphide (Eason et al. 2008).

Using alternative poisons may come with limitations. While 1080 can kill in a single dose, large and multiple doses of some alternative poisons are often needed to accumulate to a lethal level. Anticoagulants provide a promising direction for future research to overcome aversion in introduced pest species (Henderson et al. 1997), although Brodifacoum is associated with increasing primary and secondary non-target risks (Eason and Spurr 1995) and is relatively costly (Henderson et al. 1997). Considerable effort goes into developing products for vertebrate control, ensuring that they are efficacious, safe for the user and the environment, relatively humane and provide cost-effective control before they can be considered for registration. While the number of available population control options needs to be maximised to ensure that our long-term ability to reduce the impacts of vertebrate pests is maintained, bait preparation and presentation standards should remain first priority, preventing sublethal dose baits.

Alternative delivery mechanisms for toxins, delivering lethal doses that would reduce opportunity for learned aversion, still require further investigation (Saunders and McLeod 2007), particularly in the field. Spring-loaded mechanical ejectors (known as M-44 ejectors or canid pest ejectors) were registered for use in Australia in 2016 (ACTA, 2016a); the device is triggered by a minimum force required to release the toxin, i.e. by an animal pulling on the trigger with its teeth (Osborne et al. 2014). Feral cat grooming traps (e.g. ‘Spitfire’, ‘Felixer’) use a combination of criteria based around body size and habits (e.g. scent marking) to trigger a lethal dose of toxin squirted onto the animal’s pelt which it then ingests when cleaning itself (Murphy et al. 2014; Read et al. 2014; Sjoberg et al. 2014). Similarly, a range of novel devices have been designed for rats (e.g. Murphy et al. 2014), common brushtail possums (e.g. Blackie et al. 2014), and mustelids (e.g. Morriss and Warburton 2014). The efficacy of these alternative control tools relies on the behavioural patterns of the target species (i.e. biting and pulling for the canid pest ejectors, lure investigation and scent marking, etc.), and, to varying degrees, such devices could also impose selection on the populations being targeted. Delivery systems specific to the target species (e.g. feral pig feeder: ACTA 2016b) will reduce non-target exposure to toxin that could contribute development of bait-resistance, although such devices may accelerate selection for neophobic individuals in the target species population.

The key to targeted baiting without developing bait-resistance is likely to lie in maintaining a multimodal approach to reduce development of aversion responses. A combination of pest control techniques such as baiting, shooting and trapping is recommended in rotation with baiting practises (Warburton and Drew 1994; Morgan et al. 1996). For example, the Red Card for Rabbits and Foxes program in the Western Australian wheatbelt (http://www.redcard.org.au/; last accessed 14 August 2017) has been successfully coordinating baiting and shooting across private land (Fig. 4). Such efforts can show success in terms of perceived knockdown and livestock protection (McLeod et al. 2011), although long-term success of such programs relies on continued funding and community engagement.

Fig. 4.  The number of red foxes killed on a single weekend in a local district in Western Australia under the Red Card for Rabbits and Foxes program (http://www.redcard.org.au/; last accessed 14 August 2017) signifies the abundance of this introduced pest across agricultural landscapes. Photograph by P.A. Fleming.

1. Development of bait avoidance and toxin-tolerance are inherent in any long-term baiting program, threatening current and future control of introduced pest species, and making strategic choices around the most effective management options a pressing issue.

2. Increasing toxin-tolerance over time has been recorded in one field study (European rabbits: Twigg et al. 2002); by contrast, there have been many cases demonstrating innate and learned aversion towards baits (Table 1). Learned aversion to baits can take place after even a single bait dose, while innate avoidance responses at the population level, and the build-up of toxin-tolerance, take place over longer, evolutionary time scales, and therefore require long-term monitoring data to fully understand.

3. There is limited evidence for bait-resistance in many introduced pest species. However, a potential for bait-resistance to develop has been identified through learned aversion behaviour in the feral pig (McIlroy 1983; O’Brien et al. 1986) and through both behavioural and physiological resistance in the red fox (e.g. van Polanen Petel et al. 2001; Gentle et al. 2007a). In both of these species, the likelihood of sublethal dosing is enhanced due to baiting practices and caching behaviour in foxes.

4. It has long been recognised that baiting is the most cost-effective method of controlling many vertebrate pest species. In addition to the constraints of funding and non-target interference (Staples 1969; Frampton et al. 1999), the development of bait-resistance also needs to be considered as part of designing pest control actions. If appropriate action is not taken, it is possible for bait-resistance to increase, resulting in bait-shy populations that we would be unable to control (O’Connor and Matthews 1996). There is clearly a need for more research if we are to maintain the efficacy of current pest control methods into the future.

• Bait avoidance: reduced interactions with bait over time, either due to innate characteristics or learned aversion.

• Bait-efficacy: achieving the desired outcome of progressive decreases in pest species population being controlled.

• Bait-resistance: the measure of difficulty that a chemical or strategy has at controlling a specific pest population under practical conditions (Blaszkowicz 2016).

• Behavioural resistance: where the animal alters its behaviour and therefore increases its likelihood of survival (Blaszkowicz 2016).

• Innate bait avoidance: non-learned, instinctive avoidance to approaching or consuming baits, due to natural differences in behavioural responses, e.g. neophobia, wariness.

• LD dose: toxin dose required to kill a portion of the population; e.g. LD₅₀ is the amount of poison that will kill 50% of animals in a sample, LD₂₅ the dose that will kill 25% of animals in a sample.

• Learned bait aversion/bait-shyness: acquired aversion behaviour to approaching or consuming baits due to association of a sensory aspect of toxic bait – such as appearance, taste, or smell – with its ill effects. Thus, if poisons are used for control they must provide no sensation of illness following ingestion.

• Physiological resistance: where aspects of an animal’s physiology enable it to survive what would normally be considered a lethal dose, due to its body mass or its natural diet precluding consumption of a lethal dose, the mode of action of the toxin, or the animal’s ability to break down the toxin.

• Sublethal dose of toxin: a dose that does not kill the animal.

• Toxin-tolerance: physiological (biochemical) capacity to metabolise or excrete the toxin.

The authors have no conflict of interest to declare.