Sensitivity of The Stripe-Faced Dunnart, Sminthopsis Macroura (Gould 1845), To The Phenyl Pyrazole Insecticide, Fipronil, Toxicological Signs And Implications For Pesticide Risk Assessments In Australia

A lack of toxicity data quantifying responses of Australian native mammals to agricultural pesticides prompted an investigation into the sensitivity of the stripe-faced dunnart, Sminthopsis macroura (Gould 1845) to the insecticide, pronil (5-amino-3-cyano-1-(2,6-dichloro-4-triuoromethylphenyl)-4-triuoromethylsulnyl pyrazole, CAS No. 120068-37-3). Using the Up-And-Down method for determining acute oral toxicity in mammals, derived by the Organisation for Economic Cooperation and Development (OECD), median lethal dose estimates of 990 mg kg − 1 (95% condence interval (CI) = 580.7–4770.0 mg kg − 1 ) and 270.4 mg kg − 1 (95% CI = 0.0 - >20000.0 mg kg − 1 ) were resolved for male and female S. macroura respectively. The difference between median lethal dose estimates for males and females may have been inuenced by the increased age of two female dunnarts. Further modelling of female responses to pronil doses used the following assumptions: (a) death at 2000 mg kg − 1 , (b) survival at 500 mg kg − 1 and (c) a differential response (both survival and death) at 990 mg kg − 1 . This modelling revealed median lethal dose estimates for female S. macroura of 669.1 mg kg − 1 (95% CI = 550–990 mg kg − 1 ; assuming death at 990 mg kg − 1 ) and 990 mg kg − 1 (95% CI = 544.7–1470 mg kg − 1 ; assuming survival at 990 mg kg − 1 ). These median lethal dose estimates are 3–10-fold higher than the only available LD50 value for a similarly sized eutherian mammal, Mus musculus (L. 1758; 94 mg kg − 1 ) and that available for Rattus norvegicus (Birkenhout 1769; 97 mg kg − 1 ). Implications for pesticide risk assessments in Australia are discussed.


Introduction
Fipronil (5-amino-3-cyano-1-(2,6-dichloro-4-tri uoromethylphenyl)-4-tri uoromethylsul nyl pyrazole, CAS No. 120068-37-3), a phenyl-pyrazole compound, is a broad spectrum, low dose chemical registered for use in many countries including Russia, South Africa and Australia (Balanca and de Visscher 1997;Bobe et al. 1998)  Although pronil is used throughout the world as a crop protection agent, little information exists concerning either its toxicological impacts on vertebrates, or what the ecological and population-level consequences of exposure might be (Smith et al. 2010). This data gap is problematic for the assessment of environmental risk associated with the use of pronil for locust control where low-volume, oil-based insecticide formulations are used over arid and semi-arid native grasslands to control acridid (grasshopper and locust) populations in several countries including Australia (Story et al. 2005;Walker et al. 2016). The use of pronil for acridid control in Africa has been discontinued largely due to its environmental impacts (Peveling et al. 1999;Peveling et al. 2003;Steinbauer and Peveling 2011).
Mammalian risk assessments undertaken in Australia for pronil cite only two LD 50 values, making the development of species sensitivity distributions for a complete probabilistic risk assessment impossible (Posthuma et al. 2002). Furthermore, both estimates of acute oral toxicity are contained within industry reports listed as "Commercial in Con dence" and so the complete details of the study parameters and results are not available through the established scienti c literature. Rather, summaries by the United States Environmental Protection Agency (USEPA) need to be relied upon to gauge the potential mammalian responses to pronil exposure (Food and Agriculture Organisation of the United Nations 1997). In one study, an LD 50 estimate of 97 mg kg -1 has been reported for an unspeci ed rat species with abnormal gait and posture, piloerection, lethargy tremors and convulsions all reported as signs of intoxication in the study (Environment Australia 1998 Previous research into the impacts of pronil exposure on avian species has shed light on the importance of toxicological testing on a broader range of species than those currently presented in pesticide registration evaluations (Smith et al. 2010). Previously, acute pronil toxicity was only considered of concern in the Galliformes (Tingle et al. 2000). More recently, pronil's avian acute oral toxicity has been shown to group phylogenetically when additional species are tested (Kitulagodage 2011). Moreover, it has been shown that pesticide adjuvants add synergistically to the overall toxicity of formulations (Kitulagodage et al. 2008), the metabolic fate of pronil closely resembles that of organochlorine insecticides OCs (Kitulagodage et al. 2011b) and that it can be maternally transferred resulting in developmental abnormalities in hatchlings (Kitulagodage et al. 2011a).
The Up-And-Down protocol (UDP), devised and recommended by the Organisation for Economic Cooperation and Development (OECD) for resolving estimates of acute oral toxicity, is a useful alternative to conventional LD50 testing (Bruce 1985;Story et al. 2011). The UDP technique enables a median lethal dose estimate to be quanti ed for a toxicant that is comparable to one achieved from conventional toxicity testing, but requires far fewer animals (Lipnick et al. 1995). Moreover, the LD 50 values derived using the UDP method are comparable to other acute toxicity testing classi cation systems, thus allowing a comparison of pesticide sensitivity of Australian marsupial fauna derived here with non-native eutherian mammals tested elsewhere (Story et al. 2011).
Of the numerous mammal species in Australia, members of the Dasyuridae are the most likely to be affected by pesticide exposure resulting from locust spray operations . A signi cant overlap in habitat preferences between the Australian plague locust (Chortoicetes terminifera Walker 1870), the species most commonly the focus of control operations, and S. macroura, the combination of dunnart's small body mass, some as low as 7 g (van Dyck and Strahan 2008) and high metabolic requirements, their primarily insectivorous diet, and their ability to gorge feed on intoxicated locusts make these species particularly vulnerable to the effects of chemically based locust control (Story 2015).
This study quanti es the acute oral toxicity of pronil to the endemic Australian marsupial, S. macroura and compares the values obtained with the very limited amount of data available for mammals more broadly. Pesticide residue levels of the parent compound, pronil and it's metabolites in plasma, brain, liver, kidney and caudal and subcutaneous adipose tissues were also quanti ed from dunnart tissue to serve as a pilot investigation for a subsequent study into the comparative metabolic fate of pronil in two similar-sized but systematically divergent species, M. musculus (eutherian) and S. macroura (metatherian). Implications for pesticide risk assessments in Australia are discussed.

Materials And Methods
Animal housing.
Dunnarts used in the trial were sourced from a breeding colony kept at Commonwealth Scienti c and Industrial Research Organisation (CSIRO) Black Mountain Laboratories (Acton, Australian Capital Territory, Australia) made up of either eld-collected animals (n = 9) or rst-generation descendants of those individuals (n = 9). All dunnarts were sexually mature at the time of the experiment and were maintained in individual cages on a day:night cycle that re ected ambient Canberra conditions during May-August 2016 and kept at a constant temperature of 23 0 C. Dunnarts were fed low-fat minced beef, supplemented with calcium carbonate (25 g kg -1 ) and 0.015% potassium iodide solution (43 mL per 12 kg lean beef mince) as used by previous authors to maintain S. macroura colonies (Selwood and Cui 2006). Water was available ad libitum. Dunnarts were fasted for 24 h before the administration of pronil doses and then observed using video recording for 48 h after pesticide exposure (see below) and then daily without video recording for the following 12 d. Food was returned to the dunnarts' cages 24 h after dosing.

Page 5/24
Determination of acute oral toxicity.
In total, 18 dunnarts (7 males and 11 females) were used to determine the acute oral toxicity of pronil with doses administered according to the UDP dosing schedule. Each animal was weighed immediately prior to dosing and doses were made up using reference grade pronil (ChemService Inc. West Chester, PA, USA; CAS number: 120068-37-3, Lot number: 3719000), dissolved in 20 μL of acetone made up to 0.2 mL using canola oil. Each dose was given oesophageally using a 23 gauge gavage needle attached to a 1 mL syringe.
We followed OECD Guideline 425 (Organisation for Economic Cooperation and Development 2001) to estimate the acute oral toxicity value, in this case a median lethal dose along with its corresponding con dence interval for each gender. We used the Main Test of this guideline, with an alpha value (α) of 0.25 and a starting dose of 175 mg kg -1 . The UDP protocol stipulates that where no estimate of the substance's lethality is available, dosing should be initiated at 175 mg kg -1 . In most cases, this dose is sublethal and therefore serves to reduce the level of pain and suffering experienced by animals used in the experiment.
The UDP dosing protocol consists of a single-ordered dose progression in which animals are dosed individually and then observed for a minimum of 48 h before a subsequent dose is administered to another animal. If a dunnart survived the dose given to it within this short-term interval, the next animal received a higher dose, but if an animal succumbed to dosing within this time period, the dose progression proceeded with a lower dose (see Tables 1 and 2 ve reversals occurred in any six consecutive animals tested (when a reversal is created by a pair of responses in a situation in which a nonresponse is observed at a particular dose and a response is observed at the next dose tested, or vice versa), or at least four animals have followed the rst reversal and the speci c likelihood ratios exceed the critical value as determined by the AOT software.
After the stopping criteria had been reached, an estimate of the LD 50 value (calculated as the median lethal dose using maximum likelihood statistics) and the associated con dence limits were calculated using the AOT software Statistical Program version 1.0 (Organisation for Economic Cooperation and Development 2001).
The body mass of each dunnart was measured approximately 30 mins before pesticide exposure and then at daily intervals, up to 14 d thereafter for those dunnarts not incurring a lethal dose. Body mass data was analysed using t-tests on data pooled by dose for males and females. Animals which became moribund were euthanased using iso urane under oxygen and tissue samples collected and stored at -80 0 C until subsequent analysis (see below).
Quanti cation of tissue residue levels and determination of the purity of pronil.
Dunnart liver, brain, plasma and fat tissue samples were weighed and homogenised in a Tissuelyser II homogeniser (Qiagen). Samples larger than 0.3 g (liver) were homogenised in a stainless steel 25 ml grinder (Retsch) with a 20 mm stainless steel ball and samples smaller than 0.3 g were homogenised in 2 ml disposable centrifuge tube with a 6 mm diameter stainless steel ball. For every 0.2 g of sample weight, 1 ml of acetonitrile (ACN) with 1% acetic acid (AA) was added. Liver, brain and plasma were were eluted using a ow rate of 0.2 ml min -1 with the following gradient: 1 min at 70% B, 1-10 min 70 to 90% B, 10-11 min 90% B. The volume of injected sample was 1 µl. Fipronil desul nyl (hereafter referred to as p-desul nyl, retention time (RT) 3.2 min), pronil (RT 3.7 min), pronil sul de (hereafter referred to as p-sul de, RT 3.9 min) and pronil sulfone (hereafter referred to as p-sulfone, RT 4.5 min) residues were analysed in negative ion mode and were con rmed by their three most abundant product ions at optimised collision energies.
All pronil and pronil derivatives standards were purchased from Sigma Aldrich. A calibration curve was produced using 0.001, 0.01, 0.1, 1 and 10 µg/ml. Standards were prepared fresh and read before, in the middle and at the end of the sample batch. A positive control containing 0.01 µg/ml of pronil and derivatives and a negative control (ACN +1% AA) was run every three injections to ensure no carry over from previous samples and consistency of quanti cation. Positive controls contained 0.01 µg/ml of pronil and derivatives, and negative controls (ACN +1% AA) were run every 3 samples.

Results
Determination of acute oral toxicity.
Estimates of the median lethal dose values calculated by the AOT (Organisation for Economic Cooperation and Development 2001) software for male and females S. macroura were 990 mg kg -1 (95% CI = 580.7 -4770 mg kg -1 ) and 270.4 mg kg -1 (95% CI = 0 ->20000 mg kg -1 ) respectively. Concern over the difference between median lethal dose estimates for males and females potentially being in uenced by the increased age of two female dunnarts (Table 2) resulted in further modeling of dunnart responses to pronil using the assumptions; assuming survival at 990 mg kg -1 .

Signs of intoxication.
Toxicological signs observed following pesticide exposure included piloerection, withdrawal, eye closure, shivering and, intermittently, a lack of response to disturbance. In dunnarts receiving higher doses (e.g. > 550 mg kg -1 ), it was not until approximately 24 h after exposure that more severe signs typical of pronil toxicity, such as tremors and convulsions were observed. The signs of intoxication displayed by each dunnart were video recorded and a full quantitative analysis will be presented in a subsequent publication.
Changes in dunnart body mass after exposure show high variability but no visually discernable pattern (Fig 2). No statistically signi cant change in body mass was detected for either males (t 0.05(2)3 : p = 0.283) or females (t 0.05(2)8 ; p = 0.035) after pesticide exposure using pooled dose data for those dunnarts not receiving a lethal dose.
Time to death for dunnarts receiving a fatal dose.
As only 6 deaths (2 males and 4 females) occurred within the 48 h time limit placed on the determination of acute oral toxicity, across a range of dose levels from 99 mg kg -1 -2000 mg kg -1 (Tables 1 and 2), insu cient data exists for a statistical examination of trends concerning the time to death for dunnarts receiving a lethal dose. From the limited data available, time to death tended to decline with increasing dose greater than 175 mg kg -1 .
Residues of pronil and its metabolites in tissues.
Dunnarts given doses of either 990 mg kg -1 or 2000 mg kg -1 had higher tissue levels of both the parent, pronil, and the oxidative metabolite, p-sulfone, in subcutaneous and caudally stored fat samples, although no discernible pattern associating increased tissue residues with an increasing administered dose was evident (Fig 3). Fipronil and p-sulfone residues were either very low or absent from liver, brain and plasma samples taken from dunnarts across all doses (Fig 3).
Dunnarts not surviving the administered dose had higher levels of the parent compound, pronil, and the oxidative metabolite, p-sulfone, in liver tissue but similar levels in brain tissue. These dunnarts showed higher levels of both pronil and p-sulfone in both the subcutaneous and tail fat, indicating that the psulfone is being produced and rapidly (given the time course of the current study) stored in adipose tissues (Fig 4). Comparatively high levels of the p-sul de metabolite were also seen in the subcutaneous fat sampled from dunnarts not surviving a given dose (Fig 5). Brain, liver and plasma tissues from dunnarts surviving the dose contained very little, if any, p-desul nyl and p-sul de metabolite residues. However, the few that did not survive dosing contained relatively large amounts of these metabolites in subcutaneous fat (range = 5.91 -6354.34 ug kg -1 ), with smaller amounts stored in tail fat (range = 2.00 -85.78 ug kg -1 ) (Fig 5).
Mean pronil and p-sulfone tissue levels were similar in male and female dunnarts with maximal residues being detected in subcutaneous and caudally stored fat (Fig 6). Both male and female dunnarts demonstrated an equal propensity to store both pronil and p-sulfone in subcutaneous and tail fat reserves. Males had comparatively higher levels of pronil in brain tissue than females, although sulfone in the brain and liver tissues sampled were similar (Fig 6). While male dunnarts showed p-sul de and p-desul nyl residues in subcutaneous and caudally stored fat, female residue levels were extremely low (Fig 7).

Discussion
Median lethal dose.
Both genders of S. macroura tested in the current study were signi cantly less sensitive to pronil than the only other mammals tested, M. musculus (L. 1758; 94 mg kg -1 ) and Rattus norvegicus (Birkenhout 1769; 97 mg kg -1 ) (Food and Agriculture Organisation of the United Nations 1997) in the literature to date. This result directly contrasts with a 10 -14 fold difference in acute oral toxicity for both dunnart species (S. crassicaudata = 129 mg kg -1 CI = 74.2 -159.0; S. macroura = 97 mg kg -1 CI = 88.3 -120.0) to the organophosphorous pesticide, fenitrothion, when compared to M. musculus (1100 -1400 mg kg -1 ), using the same technique for the resolution of median lethal dose estimates (Story et al. 2011). Whilst the two chemicals mentioned above exert their in uence on different physiological pathways, the signi cant differences in patterns of acute oral toxicity compound the lack of acute oral vertebrate toxicological data thereby reducing the predictive value of pesticide risk assessments for endemic Australian vertebrates. Current risk assessment frameworks for pesticides generally use, in part, the lowest median lethal dose for mammals to assess hazard of a chemical (Newman 2015). Increasingly, median lethal dose estimates, either LD 50 or LC 50 data, obtained from chemical exposure studies can be incorporated into species sensitivity distributions (SSDs) to comparatively assess toxicity and derive hazard threshold values (Posthuma et al. 2002). However, the generation of a distribution using three data points, while possible with the assistance of extrapolation factors (as outlined in (Posthuma et al. 2002)), is less likely to provide a robust representation of the desired risk thresholds (e.g. HD 05 ) rendering the estimation of safe residue levels problematic. Recent research has highlighted a similar problem in relation to the avian acute oral toxicity pro le of pronil. While previous risk assessments for this pesticide have cited a primarily bimodal toxicological pro le with a highly sensitive species at one end (the northern bobwhite, The assessment of agricultural and veterinary chemicals for registration in Australia is a process that is evolving over time as both the amount of data submitted to support registrations increases and assessment methodologies and detection levels improve (Hyman 1997). If the use of SSDs to assess protection thresholds in relation to Australian endemic species is to continue, then further sensitivity research will be required to circumvent the need to extrapolate from a narrow range of organisms tested under standard laboratory conditions to free-living populations or ecosystems. The results of the present study show the limitations of this approach and highlights the importance of evaluating the effects of pesticides on non-target species that are likely to be exposed, particularly when these species are phylogenetically distinct from those used in studies of pesticide sensitivity originating in North America or the European Union.
Fiprole ( pronil and metabolite) residues in tissues and body mass.
The use of the UDP methodology to quantify a median lethal dose unavoidably results in very small experimental groups, sometimes n = 1, thereby resulting in secondary data sets, such as residue loads from tissue samples, that are unable to be subjected to appropriate statistical analyses. Despite this limitation, the current study quanti ed prole residue levels in kidney, liver, plasma, brain and caudal and subcutaneous adipose tissue samples taken from individual dunnarts at either the time of death or at the end of the 14 day post-dose observation period. Obviously, these results need to be viewed with a great deal of circumspection. However, we report these results from the current study as a precursory dataset to maximise the amount of information derived and to better inform a subsequent study into the comparative metabolic fate of pronil in two similar-sized, but systematically divergent species, M. musculus (eutherian) and S. macroura (metatherian) accepting the abovementioned limitations.
Studies investigating the biotransformation of pronil in rats (Food and Agriculture Organisation of the United Nations 1997) have quanti ed 3 primary metabolites after hepatic transformation of the parent compound pronil (Fig 1.). Of these metabolites, the p-sulfone and p-desul nyl have been shown to be of toxicological concern in previous studies. The oxidative p-sulfone metabolite has a six-fold higher binding a nity for the postsynaptic GABA receptor (Hainzl et al. 1998) and metabolism of the parent compound to this derivative has been shown to add synergistically to the overall toxicity of a pronilbased formulation in pesticide-exposed birds (Kitulagodage et al. 2011b). Moreover, avian studies have demonstrated that inclusion of p-sulfone residues in a regression analysis of post-exposure body mass loss provided a much better t than regressions comparing loss of body mass with the parent compound, pronil, alone in brain, liver and adipose tissues (Kitulagodage et al. 2011b). The overlap between symptoms of intoxication, the time course of p-sulfone residues in brain, liver and adipose tissue, lack of post-dose feeding activity and subsequent weight loss in dosed birds provided insight into an observed increased selective toxicity to the three galliform species tested (Kitulagodage 2011; Kitulagodage et al. 2011b).
In the current study, pronil and p-sulfone residues were more prominent at the higher doses administered (e. g. 990 and 2000 mg kg -1 ) with the residue load occurring in subcutaneous and caudally stored fat, liver and brain, in descending order of magnitude. Slightly higher levels of pronil were present in male (versus female) brains at the time of analysis, but very little difference existed between either pronil or p-sulfone levels in either subcutaneous or tail fat and plasma. Dunnarts not surviving the administered dose showed higher pronil and p-sulfone levels across adipose tissues, liver and brain. However, as was the case with dunnarts surviving a given dose, very little, if any, plasma-bound residue bringing into question whether the use of pronil residue in plasma is suitable as a biomarker of pesticide exposure in wildlife monitoring studies. While the detection of p-sulfone in the liver and adipose tissues of males and females across the various administered doses indicates the metabolism of pronil to the p-sulfone metabolite, the levels detected, in addition to low levels of this metabolite nding its way to brain tissue and an absence of weight loss in dunnarts surviving the administered dose, is contrary to the ndings in the abovementioned avian studies. Further research into the metabolic fate of this pesticide in marsupials is required to better elucidate the role of the p-sulfone metabolite in determining the overall toxicity of pronil-based pesticide formulations, as seems to be the case in more sensitive avian orders.
Fip-desul nyl is generally considered to be a photolytic breakdown product and not a metabolite as such.
In the current study, analysis detected generally low levels of this compound (range = 0 -46.07 ng g -1 with one male dunnart (dose = 99 mg kg -1 ) returning an outlier value of 281.89 ng g -1 in adipose tissue) and due to its toxicological signi cance, we have reported these results. Fip-desul nyl is considered of high toxicity with an acute oral LD50 of 15 (males) -18 (females) mg kg -1 for M. musculus (Food and Agriculture Organisation of the United Nations 1997). When administered orally to mice, the pdesul nyl metabolite has been shown to decrease body weight at doses of 30 and 60 ppm, whereas a lower dose of 3 ppm was seen to increased motor activity, irritability and aggression with convulsions also observed (Food and Agriculture Organisation of the United Nations 1997). Although present in small quantities, presumably as a result of photolytic breakdown of the dosing formulation immediately after preparation, it's acute toxicity would necessitate its inclusion in residue analysis for any future eld based trial investigating in situ wildlife impacts. Higher levels of the p-sul de metabolite (range = 0 -85.78 ng g -1 with the same male dunnart as above (dose = 99 mg kg -1 ) returning an outlier value of 6345.34 ng g -1 in adipose tissue) were also found in adipose tissues of pesticide-exposed dunnarts. The higher LD50 values for this compound reported for mice (69 (males) and 100 (females) mg kg -1 (Food and Agriculture Organisation of the United Nations 1997)) indicates a moderate toxicity for this species, with similar toxicological signs as those reported for the other breakdown products ( p-sulfone and pdesul nyl) as well as the parent ( pronil).
The Australian arid zone is characterised by low productivity and highly variable rainfall (Stafford-Smith and Morton 1990). Species inhabiting these environments have evolved a range of adaptations which assist them in coping with the inconsistent, and often sparsely distributed resources -such as the ability for rapid, long-range movement enabling animals to access areas of recent rainfall and capitalize on the increase food resources (Dickman et al. 1995;Letnic and Dickman 2005). The Dasyuridae caudally store fat to provide an energy reserve that can be utilised during times of resource limitation (Morton and Dickman 2008a; Morton and Dickman 2008b). The ability for lipophilic xenobiotic compounds, such as agricultural pesticides and their toxic metabolites, to be stored along with these fat reserves has the potential to ensure that pesticide residues remain biologically available by being constantly metabolized as dunnarts utilise caudally stored fat to maintain the energetic resources necessary for sustaining daily life during times of drought. Conventional toxicity testing used for chemical risk assessments generally de nes exposure times for the determination of median lethal dose values to quantify mortality (Newman 2015). The tendency for toxic substances to be stored in adipose tissue and later metabolized when animals are facing resource limitations, extends the exposure period for chemicals signi cantly beyond, for example, either the 48 hr acute oral toxicity test limit or the 30 d reproductive test limit more commonly used in pesticide risk assessments Story et al. 2016).
The scarcity of information quantifying the responses of evolutionarily unique Australian endemic species to pesticides impedes the development of biologically relevant risk assessments for the registration of chemicals in Australia. The lack of sensitivity to pronil displayed by S. macroura, as measured by acute oral toxicity, directly contrasts with the increased sensitivity (10 -14 fold) of the same species to another locusticide, fenitrothion (Story et al. 2011), highlighting the need for a better understanding of the biochemical pathways responsible for any species susceptibility to xenobiotics and thereby increasing the predictive value of risk assessments. Additional studies are now required to better understand the metabolic fate and biochemical parameters responsible for pesticide metabolism in mammals, particularly when the active ingredient of pesticide formulations can produce toxic metabolites. Finally, while the relatively high median lethal dose values quanti ed here would suggest a minimal impact of pesticide exposure on the species tested, no information quantifying the pesticide exposure of S. macrourain situ exists. Clearly, more research into dietary and non-dietary pesticide exposure pathways and residue loads are required to better inform impacts assessments.

Declarations
Funding and con ict of interest. This study was funded by the Australian Plague Locust Commission and the Commonwealth Scienti c and Industrial Research Organisation (CSIRO). We acknowledge that the listed authors are employees of the funding organisations but that this relationship had no in uence the outcomes of the work reported in this paper. None of the authors are associated with the company responsible for manufacturing the chemical under investigation in the current study. Tables   Table 1. Dose progression for Up-And-Down protocol given with short-term (48 h) and long-term (14 d) fates of individual male Sminthopsis macroura dosed orally with pronil and time to death for those dunnarts encountering a lethal dose.  Mean percentage change in dunnart body mass after exposure to pronil by gavage for all dose levels in those dunnarts not receiving a lethal dose. Error bars represent ± 1 standard error and are offset for clarity.