A review of recent non-target toxicity testing of vertebrate pesticides: establishing generic guidelines

Abstract

An analysis of the range, extent and importance of New Zealand-based ecotoxicology studies has been completed to better understand the hazards and non-target risks associated with new toxins and baits. This review focuses on compounds that have been recently developed for incorporation into bait for terrestrial vertebrate pest control, namely cholecalciferol, para-aminopropiophenone (PAPP) and zinc phosphide and the effects of these toxins on non-target species. Testing of these compounds has included cage and pen toxicity and bait acceptance studies. Locally conducted acute toxicity studies clarify overseas data and enhance risk assessments. Consistency in approach and selection of surrogate species, and careful selection of native non-target species for acute toxicity testing in cage and pen trials are recommended as important steps before field trials. We propose a systematic approach to cage and pen trials and offer guidelines on species selection.

Keywords:

• Cholecalciferol
• guidelines
• non-target species
• para-aminopropiophenone
• toxicity testing
• vertebrate pesticides
• zinc phosphide

Introduction

In New Zealand, we need to retain and refine the use of terrestrial vertebrate pest control tools, including traps and toxins for conservation, disease control and agricultural protection. Additionally, it is important to develop new alternatives to persistent anticoagulants and sodium fluoroacetate (1080) for field use (Eason et al. 2010a). Registration of new toxins and baits requires a significant commitment by researchers and manufacturers. Approval from the New Zealand Environmental Protection Agency (NZEPA), within the Hazardous Substances and New Organisms legislation, and registration with the Ministry for Primary Industries within the Agricultural Chemistry and Veterinary Medicines (ACVM) Act 1997 are required. Consultation with Māori is a prerequisite, and welfare considerations of target species are a key component of the registration assessment process for vertebrate pesticides.

Ideally, improved vertebrate toxic agents for pest control would combine limited persistence in the environment and humaneness; however, this is a significant challenge. In the 1990s, cholecalciferol was introduced, and a microencapsulated form of zinc phosphide has recently been developed for possum (Trichosurus vulpecula) control (Eason et al. 2010a). These toxins are both low residue risk in the environment and food. In 2011, para-aminopropiophenone (PAPP), a methaemoglobinaemia inducer, was registered for stoat and feral cat control. PAPP is humane in its mode of action and does not bioaccumulate. It has an antidote and is highly toxic to species like stoats (Eason et al. 2010b), but unfortunately has only low toxicity to rodents or possums. Therefore, additional new toxins are being evaluated for possum and rodent control. There is uncertainty with regard to appropriate ‘local’ data requirements in terms of ecotoxicity testing and bait acceptance in cage and pen trials to assist non-target risk assessment of new toxins. In this paper, we outline the research undertaken before registration, and on laboratory, cage or pen trials for active ingredients or products before extensive field use.

Review of New Zealand laboratory-based non-target research

Vertebrate Toxic Agents (VTAs) selected for this exercise are cholecalciferol, zinc phosphide and PAPP, because they have all recently been registered for animal pest control in New Zealand. A summary table listing all studies undertaken within New Zealand on these three compounds has been assembled, and includes publications and reports reviewed here (Table 1).

Table 1  New Zealand-based non-target research undertaken to support the approval of new active ingredients and the registration of new vertebrate toxic agents (VTA) products.

VTA	Birds	Insects/Reptiles	Other studies	Findings from New Zealand studies versus overseas studies	References
Chole-calciferol	Ducks^1 Chickens^1 Canaries^1 Weka^1,2	Wētā^1 Earthworms^2 Snails^2 Honey bees^2 Native skinks^2	Secondary poisoning studies in cats and dogs^1	Confirmed comparatively low toxicity to birds, but low toxicity does not mean no toxicity. Some deaths in canaries and chickens at 2000 mg/kg	^1 Eason et al. 2000 ^2 Booth et al. 2004
				Not toxic to invertebrates
				Skinks would not eat bait
				Low secondary poisoning risk to cats and dogs
PAPP	Australian magpies^3 Blackbirds^3 Ducks^3 Weka^3		Residue analysis in poisoned weka^4 Antidote studies in dogs^5	New Zealand studies confirmed overseas findings Toxicity of PAPP to birds is low compared with other VTAs, e.g. 1080, but there is species variation: ducks were susceptible and weka were affected by doses sufficient to kill cats	^3 Eason et al. 2010d ^4 Hix et al. 2008 ^5 Dall & Spencer 2007
ZP	Chickens^6	Earthworms^7 Honey Bees^8 Wētā^9	Modelling risk to birds, notably weka^10 Bait interference by birds^11 Secondary poisoning of dogs^12,13	Low toxicity to invertebrates Baits not attractive to birds, but birds are susceptible Low secondary poisoning risk to dogs	^6Henderson & Frampton 2007a
					^7O'Halloran & Jones 2003 ^8Henderson & Frampton 2007b
					^9Evans 2003
					^10Henderson & Eason 2006
					^11Ross et al. 2002
					^12Ross & Henderson 2006
					^13Henderson et al. 2003
PAPP, para-aminopropiophenone.

Non-native bird species have conventionally been tested as surrogates to minimise testing in native species. Nevertheless, some consistency has emerged in the approaches used for species selection, with weka (Gallirallus australis) and wētā (Hemideina sp.) chosen as representatives of native bird and insect species respectively. Generally, there has been a greater focus on risks to birds, with less research conducted on invertebrates, reptiles and other phyla. Secondary poisoning studies have frequently been conducted, with the carcasses of poisoned animals fed to cats (Felis catus) or dogs (Canis lupus familiaris). However, inconsistencies in which species to use and the extent of non-target testing are also apparent. Cholecalciferol is the only one of these three compounds that has been tested on New Zealand reptiles (Booth et al. 2004). This was conducted as part of a bait palatability trial rather than an acute toxicity study. Non-target testing for PAPP was conducted on birds only, and no local research has been completed on the toxicity of PAPP to invertebrates or reptiles. Modelling approaches have been undertaken to assess the risk of zinc phosphide to birds (Henderson & Eason 2006) and in a more limited way, to assess the secondary poisoning risks of PAPP (Hix et al. 2008). Locally generated toxicity studies may be needed for different reasons, some of which are hard to predict. For example, from a Māori community viewpoint, it is important that we have an understanding of the potential risks of new VTAs to native species, which may require additional testing (Ogilvie et al. 2010).

In the case of zinc phosphide, 50% lethal dose (LD₅₀) data already existed for 18 bird species from overseas studies before research was conducted in New Zealand. The reason for re-testing the acute toxicity of zinc phosphide in chickens (Gallus gallus domesticus) was in response to a request from the NZEPA, to demonstrate that the starch coating applied to zinc phosphide to reduce taste aversion did not affect toxicity when compared with unencapsulated zinc phosphide (R. Henderson personal communication). In the case of PAPP, additional testing was undertaken in birds on the formulation to be used in the field at the request of the Department of Conservation.

Information gained from New Zealand studies

Cholecalciferol

The risk to non-target species in New Zealand toxicity studies using cholecalciferol confirmed that this compound has comparatively low toxicity to birds. A study on the domestic mallard duck (Anas platyrhynchos) showed no mortality at 2000 mg/kg, but one of four canaries (Serinus canarius) and three of four chickens died at this dose (Ogilvie & Eason 1996) suggesting that care still needs to be taken to limit the access of birds to cholecalciferol. Studies in wētā, earthworms (Aporrectodea caliginosa), honey bees (Apis mellifera) and snails (Cantareus aspersa) (Booth et al. 2004) indicated no toxicity to these species. Skinks (Oligosoma lineoocellatum) were potentially at risk, but would not eat the bait (Booth et al. 2004). Secondary poisoning studies were important as they demonstrated a low risk to cats and dogs from possum carcasses resulting from poisoning with cholecalciferol (Eason et al. 2000).

PAPP

As with cholecalciferol, PAPP cage studies with various doses applied showed that this new toxin has comparatively low toxicity to birds (Hix et al. 2008; Eason et al. 2010b) when compared with 1080 or brodifacoum. However, a much lower LD₅₀ was produced in mallard duck than that previously reported in other bird species, and there was a protracted sub-lethal effect in weka, including lethargy and loss of appetite, which is consistent with the induction of methaemaglobinaemia (Eason et al. 2010b). Again, the New Zealand data highlight the need to limit bird access to bait containing PAPP.

Zinc phosphide

Studies in chickens illustrated that the microencapsulated formulation of zinc phosphide had the same risk to non-target species as unencapsulated zinc phosphide (Henderson & Frampton 2007a). Studies in wētā, earthworms and honey bees indicated a low risk to these species (Evans 2003; O'Halloran & Jones 2003; Henderson & Frampton 2007b). As with cholecalciferol and 1080, secondary poisoning studies demonstrated a low risk to cats and dogs from consuming possum carcasses poisoned with zinc phosphide (Ross & Henderson 2006).

Criteria for studies required and choice of species for toxicity testing

When using vertebrate pesticides, non-target species of concern include birds, reptiles and invertebrates. Our simple analysis of information from New Zealand research in the above section indicates that there is considerable value to be gained from these studies. Consistency is recommended in the selection and testing of exotic or native non-target species for acute toxicity testing in laboratory-based cage and pen trials. The necessity for New Zealand-based research should be linked to the amount and quality of acute toxicity information from overseas sources for any given toxin in non-target, albeit exotic, species. Studies conducted in New Zealand should follow, as closely as possible, international guidelines in terms of experimental and protocol design, as shown by Booth et al. (2004). Mathematical extrapolations that take into account body size (Mineau et al. 1996) and eating habits, coupled with knowledge of bait types, delivery systems and concentrations of the active ingredient in bait (see Eason & Spurr 1995), can extend the use of acute toxicity information.

Exposure is a key component of the risk equation for non-target toxicity testing. When a new VTA is being developed, the following approach should be followed to help determine which species or family requires more extensive toxicity testing. For toxicity testing of the VTA, attention should be concentrated on non-native, non-target species, e.g. mallard ducks, chickens, canaries. In parallel with toxicity testing, bait palatability trials using non-toxic bait and the proposed method of bait presentation should be undertaken on native species in cage and field trials. This will help to determine the likely exposure of non-target species to the bait being used as the carrier for the VTA. If the majority of native species show no interest in the bait, there should be no need to conduct further toxicity testing, assuming secondary poisoning is not an issue. However, if any native species shows a predisposition for the bait being tested, then further toxicity testing on the taxa or species of concern should be carried out or alternative bait matrices should be developed. This approach will help to concentrate efforts and minimise unnecessary testing.

How much information is required?

A hierarchy, or sliding scale, of additional data requirements relative to the extent of pre-existing overseas data is recommended, based on the extent of the overseas literature (Table 2). For example, where there are extensive acute toxicity data in birds, invertebrates and reptiles, there should be minimal need for additional LD₅₀ testing, and a greater focus on the toxicity of bait to non-target native species.

Table 2  Requirements for additional New Zealand data, based on the extent of overseas research.

Need for additional testing	Extent of existing acute toxicity data	Additional requirements
Low	Extensive acute toxicity data in birds, invertebrates and reptiles	LD50 testing unnecessary; focus on toxicity of bait and observation in a native species
Medium	Moderate amount of acute toxicity data in birds, invertebrates and reptiles	Focus on toxicity of bait and observation in a native species, may require additional LD50 data
High	Limited amount or no acute toxicity data in birds, invertebrates and reptiles	Generation of LD50 data in birds, palatability and toxicity of bait plus observation in native species
LD50, median lethal dose.

Rationale for species selection

Mallard ducks are commonly used in Organisation for Economic Co-operation and Development (OECD) toxicology studies. The northern bobwhite quail (Colinus virginianus), Japanese quail (Coturnix japonica), zebra finch (Poephila guttata), rock pigeon (Columba livia) and budgerigar (Melopsittacus undulatus) are also species recommended by OECD (2010) guidelines for the testing of chemicals. This list is intended to act as a guideline rather than limit the species used. Chickens, Australian magpies (Gymnorhina tibicen), blackbirds (Turdus merula) and weka (Eason et al. 2010c) have all been used when testing vertebrate pesticides for field use in New Zealand. Chickens have been tested in primary and secondary poisoning trials as initial surrogates for weka. Exotic species should always be tested before native species because a considerable amount of toxicology data exist on these species, and the relative toxicity of toxins with unknown toxicity to birds can be compared with other toxins. In addition, native species may be endangered, legally protected or have special cultural significance attached to them, which makes their use less appropriate. When native species are used, their use is likely to be linked to specific questions.

An example of such specific questions being raised are the effects of new VTAs on species of cultural importance. Māori have strong views and concerns regarding native species, particularly those species that are likely to attempt to access bait, even when it is delivered in bait stations. This is one reason for the inclusion of weka in toxicity trials. Also, weka are not legally protected on the Chatham Islands, which enables their selection as a relevant species for toxicity testing, particularly as they are a naturally inquisitive species and so might be considered at higher risk of taking bait. They are relatively easy to hold in captivity, making them a useful non-target species for testing poisoning. In this way, testing can take into account cultural views of Māori.

Acute toxicity is usually assessed by administration of the toxin by gastric intubation. This provides the measure of toxicity or hazard. However, the bait matrix, its palatability and the eating habits of non-target species, can influence the toxicity of the actual bait. Therefore, when testing the susceptibility of birds and other non-target species, some toxicity data derived from eating baits at the same toxic loading as used in the field, as well as traditional LD₅₀ data, are desirable (Table 3).

Table 3  Rationale for selection of species for cage and pen trials.

				When testing should occur based on need for additional testing (Table 2)
Taxa	Species	Type of test	Rationale for testing	High	Medium	Low
Birds
	Chicken	LD50	Generation of acute toxicity data	✓
	Chicken	Bait	Palatability and toxicity information on bait	✓	✓	✓
	Duck	LD50	Confirmation of susceptibility of a second bird species	✓
	Duck	Bait	Palatability and toxicity information on bait and enhanced understanding of susceptibility of birds	✓	✓
	Rock pigeon	Bait	Palatability and toxicity information on bait and enhanced understanding of susceptibility of birds	✓	✓
	Blackbird	Bait	Palatability and toxicity information on bait and enhanced understanding of susceptibility of birds	✓	✓
	Canaries	Bait	Palatability and toxicity information on bait and enhanced understanding of susceptibility of birds	✓	✓
	Weka	Bait	Palatability and toxicity information on bait and enhanced understanding of susceptibility of birds	✓	✓	✓
Reptiles
	Introduced skink	LD50	Generation of acute toxicity data	✓	✓^1
	Introduced skink	Bait	Palatability and toxicity information on bait	✓	✓	✓
Invertebrate
	Wētā	LD50	Generation of acute toxicity data	✓	✓^1
	Wētā	Bait	Palatability and toxicity on a native insect species likely to eat bait	✓	✓	✓
	Earthworm	LC50	Generation of acute toxicity data	✓^2
	Honey bees	LD50	Additional information on toxicity to invertebrates	✓^3
LD50, median lethal dose.
^1Dependent on amount of existing acute toxicity data for taxa.
^2Where vertebrate toxic agents are to be applied directly on to the ground.
^3If bait is shown to be palatable to honey bees.

Toxicity testing approach and methodology

The guidelines of the OECD should be followed in toxicity studies as far as possible, with some adaptation to take into account New Zealand-specific questions relating to bait palatability, and constraints associated with testing native species. OECD (2008) guidelines recommend the use of the ‘up-down’ procedure (UDP) as the most useful for assessing LD₅₀, as opposed to other approaches to acute toxicity tests. UDP uses a single ordered dose progression, with a recommended starting dose of 175 mg/kg if information about the LD₅₀ is unknown (Rispin et al. 2002). Doses are then either increased or decreased depending on the result from the first group of animals/birds tested (Rispin et al. 2002).

Ducks and chickens

When additional LD₅₀ data are required, chickens and mallard ducks can be obtained from a commercial breeder and housed individually. Because of the ethical costs of undertaking LD₅₀ tests, use of the minimum number of any individual species is appropriate to obtain an approximate LD₅₀. Cohorts of five birds of each species are the maximum number that should be presented with the toxin by oral gavage to establish an LD₅₀. Additional individually housed chickens and ducks can also be presented with toxic baits to compare the LD₅₀ with susceptibility to bait. All surviving birds should be observed for 14 days following treatment. Post-mortem inspections should be conducted on birds that die and on survivors after 14 days. The benefit of using these non-native species for toxicity testing is that they are internationally recognised, whereas the disadvantage is that there can be significant variation in the toxicity of a VTA between families within the same taxa. However, it is impractical to conduct toxicity testing for a new VTA on all bird families in New Zealand.

Where a precise LD₅₀ is not required, UDP is recommended. There are three tests that can be conducted—the limit dose test, preferred when toxicity is expected to be low; the LD₅₀ slope test, used when a dose–response curve or confidence interval is required; and the LD₅₀-only test, preferred when only the median lethal dose is required. Following the UDP minimises the number of animals required for toxicity testing (OECD 2010). As with the procedure above, cohorts of five birds of each species are dosed with the specified amount of toxin, excluding control birds. The procedures to follow vary depending on which of the three tests are being conducted. Details can be found in OECD (2010).

Additional bird species

If additional LD₅₀ data are deemed desirable, rock pigeons, blackbirds, canaries and weka should be presented with toxic bait and their susceptibility should be assessed. Five birds should be used to establish an LD₅₀, held in individual cages placed next to each other so that birds can have visual and audible social interactions (OECD 2010). It is recommended that the use of wild-caught birds is limited because of the risk of disease and animal welfare considerations (OECD 2010). The UDP approach to gathering LD₅₀ data for additional species should always be used to minimise the ethical cost of these studies. As with ducks and chickens, all surviving birds should be observed for 14 days following treatment. Post-mortem inspections should be conducted on birds that die, and on survivors after 14 days.

Special requirements for consultation regarding the conduct of toxicity studies in weka may be problematic. The study design for weka may be altered in consultation with Māori, the NZEPA and other stakeholders such as the Department of Conservation. It may be preferable not to conduct an acute toxicity study in this species but to simply measure the amount of non-toxic bait eaten, and extrapolate likely effects based on acute toxicity data in non-native species. This approach has been used in the past (Booth et al. 2004), and weka were returned to the wild after the experiment.

Reptiles and invertebrates

Introduced skinks (e.g. the rainbow skink, Lampropholis delicata) should be used as a surrogate for native reptiles where LD₅₀ data are required. The UDP approach to toxicity testing is recommended to minimise the number of animals used. Cage trials using native skinks and geckos to monitor their behavioural responses to non-toxic baits and to measure the palatability of the baits are recommended as this will help to determine exposure risk (Booth et al. 2004).

For toxicity testing on native invertebrates, Wellington tree wētā (Hemideina crassidens) should be used because they have been used in previous toxicity studies (e.g. Booth & Wickstrom 1999; Ogilvie & Eason 1996; Evans 2003) and are easy to maintain (Barrett 1991). The recommended methodology when undertaking LD₅₀ tests on wētā is to oral dose the VTA in a vegetable oil carrier using a micro-syringe (Ogilvie & Eason 1996). Cage trials should also be undertaken using non-toxic bait to determine whether wētā will eat bait.

Toxicity testing on earthworms is most important where a VTA is to be applied directly onto the ground (e.g. aerially applied or hand laid). Where toxicity testing is carried out it should follow OECD guideline 207 (OECD 1984). Toxicity testing on honey bees should only be required if baits are shown to be attractive to bees. OECD guideline 213 (OECD 1998) should be followed for toxicity tests on bees.

Discussion

At present, the NZEPA does not have a specific checklist of tests required for testing vertebrate pesticides, relative to their modes of application. However, they do provide guidance and encourage applicants to discuss data packages early in the application development process to try and tailor data needs to the specific product and use pattern (R. Toy personal communication). The aim is to always have enough data to assess risks to non-target organisms from use in New Zealand.

Recent ecotoxicology research and secondary poisoning trials linked to the development of cholecalciferol, PAPP and zinc phosphide, shows that these studies have provided useful information. However, laboratory trials are only a first step in the risk assessment of non-target species. The principles that underpin the development of testing vertebrate pesticides to assess the risk to non-target species are as follows.

•	Adverse reactions in non-targets can be predicted from toxic effects observed in surrogate species exposed to chemicals in laboratory conditions, when coupled with field observations.
•	Administration of high doses improves predictability.
•	Ecotoxicology, when combined with field monitoring of individual non-target birds and other animals, and then populations, forms the basis for risk assessment and environmental protection (Eason et al. 2010c).

In the past, acute toxicity testing was undertaken in a large number of non-target species for some vertebrate pesticides. LD₅₀ data for brodifacoum were collected for 13 bird species, including pūkeko (Porphyrio porphyrio melanotus) and silvereye (Zosterops lateralis) (Godfrey 1985) before it was registered. 1080 LD₅₀ data have been generated in hundreds of species (McIlroy 1986). Conducting extensive LD₅₀ testing would now be considered ethically unacceptable. There are considerable ethical and compliance obligations associated with undertaking acute toxicity studies in birds and other species that increase when testing toxins in protected native species. Ideally, toxicity testing should be undertaken on the full range of native non-target species likely to encounter bait. However, as many native species are protected and many are threatened species, it is now a more acceptable practice to use surrogate species as far as possible in cage trials. Data from surrogates and one native species (weka) are then extrapolated to other species. This was the case for PAPP during the course of its research and registration (Eason et al. 2010b). Acute toxicity data are complemented by field trials, including monitoring of both target and non-target species.

Analysis of previous research has provided some guidance in terms of types of tests and selection of species for these tests, and gives some guidance for applications to register new toxins in the future. Consistency in the approach and selection of surrogate species, and careful selection of native non-target species for acute toxicity testing in cage and pen trials is recommended. Conventional (exotic) bird species such as chickens, mallard ducks and blackbirds should be tested before native species. Māori have strong views and concerns regarding native species, particularly those likely to try and access bait even when it is in bait stations, hence the inclusion of weka. It is further recommended that, while some systematic bird and invertebrate testing is needed, modelling approaches can reduce the financial and ethical costs of testing the effect of toxins in non-target species. Optimising the modelling approaches, coupled with acute toxicity data, careful observation of the behaviour of non-target animals to baits in pen and field trials, and population monitoring, will improve non-target risk assessment.

Any new toxin or bait is likely to be registered for ground control in bait stations in the first instance. Registration for aerial application is unlikely to be approved until several years of experience has been gained in product usage. Requirements for non-target monitoring after aerial application have recently been reviewed (Veltman & Westbrooke 2011), but are beyond the scope of this review. Field monitoring of individual non-target birds and animals, and population monitoring pre- and post-registration of new toxins, baits, baiting strategies and delivery systems will remain critically important, as will continued research to improve target specificity of animal pest control technologies. Veltman and Westbrooke (2011) recommend continued diligence and vigilance to improve monitoring, and suggest that more studies are needed when baiting strategies are changed, such as greater use of pre-feeding with non-toxic bait to lure animals to the toxin. Similarly, more forest bird population monitoring will be required at sites where new toxic baits are used. Continued research to improve target specificity of animal pest control technologies remains critically important for conservation, as well as for tuberculosis vector control.

Acknowledgements

This review was funded by the New Zealand Animal Health Board.