Environmental Toxicology and Chemistry # 2012 SETAC Printed in the USA DOI: 10.1002/etc.1757

Hazard/Risk Assessment

RISK ASSESSMENT FOR ADULT EXPOSED TO THE MOSQUITO CONTROL PESTICIDE NALED

TIMOTHY A. BARGAR* Southeast Ecological Science Center, U.S. Geological Survey, Gainesville, Florida

(Submitted 22 June 2011; Returned for Revision 11 October 2011; Accepted 23 November 2011)

Abstract—A prospective risk assessment was conducted for adult butterflies potentially exposed to the mosquito control insecticide naled. Published acute mortality data, exposure data collected during field studies, and morphometric data (total surface area and fresh body weight) for adult butterflies were combined in a probabilistic estimate of the likelihood that adult butterfly exposure to naled following aerial applications would exceed levels associated with acute mortality. Adult butterfly exposure was estimated based on the product of (1) naled residues on samplers and (2) an exposure metric that normalized total surface area for adult butterflies to their fresh weight. The likelihood that the 10th percentile refined effect estimate for adult butterflies exposed to naled would be exceeded following aerial naled applications was 67 to 80%. The greatest risk would be for butterflies in the family , and the lowest risk would be for those in the family Hesperidae, assuming equivalent sensitivity to naled. A range of potential guideline naled deposition levels is presented that, if not exceeded, would reduce the risk of adult butterfly mortality. The results for this risk assessment were compared with other risk estimates for butterflies, and the implications for adult butterflies in areas targeted by aerial naled applications are discussed. Environ. Toxicol. Chem. # 2012 SETAC

Keywords—Butterflies Naled Risk assessment Mosquito control

INTRODUCTION impacts on nontarget have provided mixed indications The mosquito control pesticide Trumpet EC (AMVAC of risk. One investigation reported that survival of late-instar Chemical Corporation), containing the active ingredient naled Cyclargus thomasi bethunebakeri larvae following aerial naled (1,2-dibromo-2,2-dichloroethyl dimethyl phosphate), is aerially applications was reduced 26% compared with a reference applied as an ultralow-volume spray in the Florida Keys, USA, location [7]. Another reported that the mean abundance for to control adult mosquitoes. This has been an issue for resource A. troglodyte floridalis, but not S. acis bartrami, was greater at management agencies responsible for conservation of imperiled locations not treated by mosquito control pesticides relative to butterflies residing within or adjacent to areas of active mos- sprayed locations [1]. In a study not targeting butterflies, aerial quito control. In fact, it has been suggested that the application naled applications reduced abundance for several nontarget of mosquito control pesticides is partially responsible for the families but did not affect diversity [8]. Therefore, it decline of butterfly communities in the Florida Keys [1–3]. appears that naled impacts on insects in the field vary depending Reintroduction efforts for the state-listed and federal candidate on the family of interest. Miami blue butterfly (Cyclargus thomasi bethunebakeri) are The prospective risk assessments that estimate pesticide occurring adjacent to areas of active mosquito control. In impacts on nontarget terrestrial insects might not be applicable addition, aerial naled applications are permitted over the for adult butterflies exposed to an insecticide applied as an Key Deer National Wildlife Refuge on Big Pine Key, Florida, ultralow-volume spray, because they consider exposure only as where two resident butterfly taxa (Florida leafwing butterfly a result of the insect walking across or landing on a contami- [Anaea troglodyta floridalis] and Bartram’s hairstreak nated surface [4,9,10]. Naled is a very labile compound with a butterfly [ acis bartrami]) are candidates for federal half-life ranging from 1.4 to 8.2 h depending on incident light listing under the Endangered Species Act of 1973.) Aerial and relative humidity, with it being least persistent in direct applications of naled may be adversely impacting populations sunlight and most persistent under dark conditions with low of those butterflies. relative humidity [11]. At least in the Florida Keys, aerial Naled is a broad-spectrum organophosphate insecticide that naled applications occur just before sunrise and extend for a is potentially toxic to a variety of insects [1–4]. Based on naled short period after sunrise, meaning that naled residues on toxicity to honey bees, the U.S. Environmental Protection foliage are likely to degrade rapidly, before butterflies (adults Agency’s Office of Pesticide Programs classified naled as or larvae) can accumulate naled as a result of walking across or highly toxic to nontarget terrestrial invertebrates, implying that consuming contaminated foliage. As a result, direct deposition naled impacts on butterflies are possible given sufficient expo- of naled onto the exoskeleton is the most likely route of sure [5]. Available acute toxicity data for several butterfly efficacious exposure for adult butterflies. Therefore, assessing species suggest that honey bee acute toxicity data may not risk to butterflies from ultralow-volume spray applications of be applicable to butterflies [6]. Field studies investigating naled naled requires considering direct deposition in the exposure assessment. The present study presents a prospective risk assessment * To whom correspondence may be addressed ([email protected]). method for adult butterflies exposed to mosquito control insec- Published online 25 January 2012 in Wiley Online Library ticides containing the active ingredient naled that are applied as (wileyonlinelibrary.com). an ultralow-volume spray. Naled is used by several mosquito

1 2 Environ. Toxicol. Chem. 31, 2012 T.A. Bargar control districts in the state of Florida and elsewhere and is (nearest 0.001 g) were coupled as a metric of surface area per also registered by the U.S. Environmental Protection Agency unit mass (cm2/g, the exposure metric), which was combined for the control of black flies and leaf-eating insects on a variety in Equation 1 with deposition data to estimate a weight- of agricultural crops. Although the focus of this risk assessment normalized exposure for adult butterflies in the field. is on impacts on butterflies in the Florida Keys, it has impli- cations for other areas where pesticides containing naled are Exposure ðmg=gÞ used. ¼ deposition ðmg=cm2Þexposure metric ðcm2=gÞ (1)

MATERIALS AND METHODS The product of the deposition and exposure metric data has the same units as the toxicity data, which facilitates risk Field studies were conducted to measure naled deposition estimation. Toxicity data for insects are often reported as following aerial applications over the Key Deer National Wild- chemical mass per organism rather than per unit body weight. life Refuge, Florida, USA (hereafter, Refuge). Two residue Therefore, use of the term ‘‘exposure’’ in the present study sampler types were used to assess naled deposition at each test refers to weight-normalized exposure to avoid confusion. location immediately following aerial applications, glass fiber 2 Effects were estimated based on published acute toxicity filter papers (452.4 cm , Whatman No. 4 qualitative) pinned to data and on unpublished data determined by the author aluminum foil-covered Styrofoam blocks and placed flat on the (Table 1). The range of toxicity values in this data set (four ground and acrylic yarn (0.47 cm 5.79 m) stretched within orders of magnitude) is not unusually small given the variability polyvinylchloride frames set approximately 1 m above the of acute toxicity data sets typically available for organisms ground. Yarn samplers were also used to assess naled deposition exposed to organophosphate insecticides [12]. Most of the because small droplets typical of ultralow-volume sprays are toxicity data available for butterflies were derived using the unlikely to settle onto the filter papers before the samplers honey bee acute contact toxicity method, in which the chemical are retrieved. During each of two field trials, three yarn and is applied to the dorsal side of the thorax. However, adult three paper samplers were placed at each of eight locations in butterflies are responsive to naled deposition onto wings [6], the Refuge. The samplers were deployed less than 1 h before the indicating that exposure routes other than deposition onto the applications. Applications began shortly before dawn (15– thorax should be considered in the risk assessment. To this end, 20 min), when an airplane flew over the Refuge at an elevation a refined effect estimate was generated for this risk assessment of approximately 30.5 m and sprayed (Micronair AU4000 atom- using Equation 2. The refined effect estimate incorporates izer) Trumpet EC at a rate of 70 g a.i./ha (840.6 g end-use relative surface areas of the body and wings and the toxicity product/ha). Four teams of personnel were responsible for as a result of exposure on the body and wings. residue sampler recovery to hasten sampler retrieval after the sprays. Each team was responsible for separate pairs of loca- Effect ðmg=gÞ tions beginning approximately 50 min after the application had ¼ð þð Þ ceased. The 50-min delay allowed for droplet settling to the Propwing LD50wing Propthorax LD50thorax (2) ground following the sprays. Each residue sampler was indi- vidually placed into separate, critically cleaned, foil-wrapped Prop and Prop represent the proportion of the total 40-ml glass vials, filled with 30-ml pesticide-grade hexane and wing thorax butterfly surface area comprised by the wings and thorax, was kept in a freezer (108C) until transport to Florida A&M respectively. Prop and Prop were held constant at University (Public Health Entomology Research and Education wing thorax 0.93 and 0.07, respectively, because the percentage of the total Center) for analysis. All residue samplers were recovered surface area comprised by the wings was fairly consistent within 90 min after the aerial application. Field spikes were conducted by pipetting 200 ml of a naled spike solution onto two of the three yarn and paper samplers at the reference location and into a vial containing 1 ml hexane (spike check). Recoveries Table 1. Acute toxicity (LD50, mg/g bw [95% CI]) of naled to were 26 and 14% for the paper and yarn samplers, respectively, adult butterflies whereas recoveries were 100% for the spike checks, indicating a a considerable naled dissipation before immersing the samplers in Species LD50thorax LD50wing hexane. Heraclides cresphontes 0.19b Morphometric data (total surface area and weight) were Vanessa cardui 0.541b 2.29 (1.93–2.72)c measured for 22 species of adult butterflies. Aerial applications 30.08 (27.88–32.45)c began shortly before dawn, so butterflies were most likely Eumaeus atala 0.0012d 1.31 (1.06–1.62)c c roosting with their wings folded together on the dorsal side 28.22 (21.70–36.69) Heliconius charitonius 0.90 (0.40–1.37)c of the body. Therefore, digital photographs were taken of Junonia coenia 6.84 (5.45–8.57)c 13.6 (10.5–17.6)c butterflies with their wings in that position to measure total 5.5 (3.5–7.4)e surface area. The area of a polygon outlining the wings, thorax, Anartia jatrophae 14.68 (9.01–23.99)c 1.58 (1.18–2.13)c and head of the photographed butterfly was measured in ImageJ Urbanus proteus 0.1892d 0.3632d (National Institute of Health). The accuracy of the measurement d 2 Pyrgus oileus 0.0823 method (within 0.01 cm ) was assessed by measuring the area of Ascia monuste 2.0 (1.4–2.5)e individual cells of grid paper with 1-cm2 cells. Estimating butterfly exposure to naled by strictly equating a Median lethal dose (LD50) values based on either thorax or wing-only naled deposition onto filter papers with deposition onto butter- exposures. b From Eliazar et al. [3]. flies does not consider the role of the organism mass in diluting c From Hoang et al. [6]. chemical dose. Plus, mass is typically a component of reported d From Salvato [1]. toxicity data (mg/g). Therefore, surface area and fresh weight e T. Bargar, unpublished data. Toxicity values estimated by Probit analysis. Risk assessment of naled impacts on adult butterflies Environ. Toxicol. Chem. 31, 2012 3 among several species, ranging from 88.2 to 96.1 (avg ¼ 93, unclear if this difference is due to sampler orientation, type, or stdev ¼ 3.08). The median lethal dose values (LD50wing and surface area. LD50thorax) for the refined effect estimate calculation were Morphometric data (surface area and weight) for 22 different generated by Monte Carlo sampling (Oracle Crystal Ball ver butterfly species, encompassing five different families are 11.1) of distributions (assumed log-normal) based on the tox- reported in Table 2. The Papilionidae had the largest surface icity data shown in Table 1. For species with multiple LD50s, areas available for pesticide deposition (average ¼ 23.25 cm2), geometric means of toxicity values were generated for the whereas the Lycaenidae had the smallest surface areas (aver- distributions. age ¼ 3.27 cm2). The Papilionidae were also generally the Risk is a function of exposure and effect, as shown in heaviest butterflies, and the Lycaenidae were the lightest. As Equation 3. might be expected, surface area and weight were positively ¼ ¼ correlated (Pearson correlation coefficient [r] 0.91). Exposure Risk exposure effect (3) estimates for butterflies based solely on total surface area would where exposure and effect values are derived by Equations 1 be 3.9 and 0.55 mg per butterfly for Papilionidae and Lycaeni- and 2, respectively. An organism is considered to be at risk when dae, respectively, when compared with 50th percentile depo- 2 exposure levels equal or exceed toxicity (RQ ¼ 1). The initial sition level onto filter papers (0.168 mg/cm ). This would risk estimate will be based on the conservative comparison of the indicate that risk would be greater for Papilionidae assuming maximal exposure, with the lowest toxicity under the assumption equivalent sensitivities between the families. However, those that risk would be unlikely if the conservative comparison exposure estimates do not take into account body weight. resulted in RQ < 1. If RQ > 1, then a refined procedure was Therefore, morphometric data were used to calculate an expo- employed to estimate risk by using the entire exposure and sure metric for each species that would normalize surface area toxicity data sets to develop a joint probability curve [13]. for body weight to generate a weight-normalized exposure Finally, deposition levels equivalent to refined effect estimates estimate (Table 2). What is evident from the data in Table 2 were estimated using a rearrangement of Equation 1. is that the exposure metric is not positively related to total surface areas (r ¼0.32). For example, despite having the Deposition level ðmg=cm2Þ largest total surface areas, the average exposure metric for Papilionidae (74.75 cm2/g) was not the greatest among the five ¼ ðm = Þ ð 2= Þ effect g g metric cm g (4) butterfly families, whereas the family with the smallest average total surface area (Lycaenidae) had the largest average metric The effect parameter in Equation 4 is the refined effect (209.1 cm2/g). The implication is that the weight-normalized estimate from Equation 2. Monte Carlo simulation was con- exposure would be greater for butterflies with higher exposure ducted on the LD50 parameters in Equation 2 to generate a metrics, given the same deposition level. For example, based on distribution of refined effect estimates and predict the 10th, the 50th percentile deposition level for paper, the exposure 50th, and 90th percentile effect estimates for use in Equation 4. estimates for Papilionidae and Lycaenidae were 12.6 mg/g and The median and maximum exposure metric for each family of butterflies was combined in Equation 4 with each percentile effect estimate to determine the deposition level equivalent to Table 2. Morphometric data (surface area, weight) and derived exposure that effect estimate for each butterfly family. metricsa for adult butterflies. Total Exposure surface Weight metric RESULTS Family Species areab (cm2) (g) (cm2/g)

Naled deposition onto either yarn or paper samplers did not Pieridae Ascia monuste 14.5 0.15 96.7 differ between the field trials (p ¼ 0.24 for paper, p ¼ 0.96 for Phoebis sennae 19.9 0.25 79.6 yarn; two-sample t test), indicating relative uniformity of Eurema nicippe 6.9 0.05 138.0 deposition between spray missions. Cumulative distributions Papilionidae Papilio polyxenes 19.3 0.22 87.7 Papilio troilus 27.2 0.44 61.8 for both samplers give different indications of naled deposition Nymphalidae Vanessa cardui 8.8 0.15 58.7 at the deployment locations (Fig. 1). Yarn samplers gave a Junonia coenia 9.25 0.11 84.1 lower indication of naled deposition than paper samplers. It is Agraulis vanilla 15.4 0.21 73.3 Anartia jatrophae 11.75 0.12 97.9 1 Heliconius charitonius [6] 19 0.19 100.0 Phyciodes phaon 3.7 0.03 123.3 0.9 Hermeuptychia sosybius 5.75 0.05 115.0 0.8 Lycaenidae Leptotes cassius 1.3 0.003 433.3 0.7 Strymon istapa 2.1 0.01 210.0 0.6 Eumaeus atala [6] 7.3 0.1 73.0 Calycopis cecrops 2.4 0.02 120.0 0.5 Hesperidae Urbanus proteus c 6.7 0.2 33.5 0.4 Percentile Percentile Erynnis spp. 4.4 0.07 62.9 0.3 Erynnis spp. 2 5 0.07 71.4 0.2 Thorybes spp. 5.05 0.15 33.7 Pyrgus sp. 1.9 0.03 63.3 0.1 Pyrgus oileusc 2.8 0.05 96.7 0 0 0.1 0.2 0.3 0.4 0.5 0.6 a Exposure metric ¼ total surface area weight. Naled Deposition (μg/cm2) b Total surface area is equal to twice the result of the area determined by ImageJ, which calculates area for one side of the butterfly. Fig. 1. Cumulative distributions for the average naled deposition onto filter c Surface area measurements for mounted butterflies (wings spread) deter- paper (solid circles) and yarn (open circles) samplers at each sampling mined by ImageJ were not doubled as for the other species. The weights are location following aerial applications for adult mosquito control. from Salvato [1]. 4 Environ. Toxicol. Chem. 31, 2012 T.A. Bargar

35.1 mg/g, respectively. A cumulative distribution of exposure AB metrics for individual species indicated that the exposure would 100 100 be greatest for species within the Lycaenidae family and that the exposure would be least for species within the Hesperidae 80 80 family (Fig. 2). The highest exposure estimate (0.5817 mg/cm2 433.3 cm2/ 60 60 g ¼ 252.0 mg/g) and the lowest refined effects estimate ([0.97 m þ m ¼ m 40 40 1.31 g/g] ]0.03 0.19 g/g] 1.3 g/g) resulted in an RQ Frequency of 193.9, indicating considerable risk for adult butterflies. 20 20 Because maximal exposure and sensitivity for adult butterflies is atypical, a refined risk analysis was conducted. The initial 0 0 indication of risk likelihood is evident by a comparison of the 0.0 1.0 100.0 0.0 1.0 100.0 cumulative distributions of exposures for L. cassius and Exposure (µg/g) Exposure (µg/g) U. proteus to the distribution for the refined effect estimates (Fig. 3). Because it had the smallest exposure metric, the Fig. 3. Cumulative distributions for refined effect estimates (þ) and the exposure distribution for U. proteus represents the lowest weight-normalized exposureestimatesforLeptotescassius(open circles)and Urbanus proteus (solid circles) based on their respective exposure metrics potential exposure for adult butterflies. The exposure distribu- and naled deposition onto filter paper (A) and yarn (B) samplers. The area tion for L. cassius represents the highest potential exposure bracketed by the exposure estimate distributions for these two species because L. cassius had the largest exposure metric. Most of the represents the area of potential exposure for adult butterflies. refined effect estimate distribution for either species was exceeded by the estimated exposure levels based on naled equivalent to the 10th percentile refined effect estimate would deposition onto filter paper samplers (Fig. 3A), while it was be the least for Lycaenidae butterflies and the most for between the estimated exposure distributions based on deposi- Hesperidae butterflies, presuming equivalent sensitivities tion onto yarn samplers (Fig. 3B). The risk indicated by the yarn among adult butterflies. That is, risk would be greatest for samplers was generally lower compared with the risk indicated the Lycaenidae and least for the Hesperidae. The equivalent by the paper samplers. Data in Figure 3 were used to develop deposition levels shown in Table 3 represent potential guideline joint probability curves to show risk likelihood based on naled levels for protection of adult butterflies from adverse impacts deposition onto the filter paper (Fig. 4A) and yarn (Fig. 4B) from exposure to naled. The potential suitability of the equiv- samplers. One joint probability curve represents butterflies with alent deposition levels as guidelines would be evident from an exposure metric similar to that of L. cassius, whereas the comparison of equivalent deposition levels with effects data other represents butterflies with an exposure metric similar to collected during the field trials. Total cholinesterase (TChE) U. proteus. For the application conditions during the field trials, activity in two adult butterfly species (Ascia monuste and the likelihood that exposure will exceed the 50th percentile Agraulis vanillaea) was measured following their exposure refined effect estimate for adult butterflies ranges from 70 to to naled after aerial applications (author’s unpublished data). 95% based on the paper samplers and from 33 to 87% based on The average TChE activity for those species was depressed the yarn samplers. The range in exposure probability between between 22 and 37% at a deposition level of 0.05 mg/cm2. That the two curves in Figure 4A and B reflects risk uncertainty for deposition level is greater than the levels equivalent to the 10th adult butterflies. What is apparent from Figure 4 is the high and 50th percentile refined effect levels but less than most of probability that exposure during typical aerial applications will those equivalent to the 90th percentile refined effect level. exceed the refined effect estimate for butterflies with an expo- Because of TChE depression in butterflies at 0.05 mg/cm2,a sure metric similar to that for L. cassius. lower deposition level would be necessary to reduce impact on Table 3 shows naled deposition levels that would result cholinesterase enzymes in adult butterflies. The two species in butterfly exposures equivalent to the 10th, 50th, and used in the field study were in the Pieridae (A. monuste) and 90th percentile refined effect estimates. Generally, deposition Nymphalidae (A. vanillae) families. Perhaps a deposition level

100 90 Nymphalidae AB100 100 80 90 90 70 Lycaenidae 80 80 Pieridae 70 70 60 60 60 50 Papilionidae 50 50 40 40 40 Percentile Percentile Hesperiidae 30 30 30 20 20 Exposure Probability Exposure Probability 20 10 10 10 0 0 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 0 Refined Effect Estimate Percentile Refined Effect Estimate Percentile 1 10 100 1000 Exposure Metric (cm2/g) Fig. 4. Joint probability curves showing the likelihood that adult butterfly exposure to naled will exceed the refined effect estimates when exposure is Fig. 2. Cumulative distribution of exposure metrics for 69 adult butterflies, estimated based on filter paper (A) and yarn (B) samplers. One curve (solid comprising 22 different species from five different families. Based on this line) shows risk likelihood for Urbanus proteus, whereas the other shows distribution, exposure would be greatest for butterflies within the Lycaenidae risk likelihood for Leptotes cassius (dashed line). Risk likelihood is expected familyand least for butterflieswithin the Hesperidae family. [Color figurecan to be greatest for L. cassius because it had the greatest exposure metric be seen in the online version of this article, available at wileyonlinelibrary. (433.3 cm2/g), and risk likelihood is expected to be least for U. proteus com] because it had the lowest exposure metric (33.5 cm2/g). Risk assessment of naled impacts on adult butterflies Environ. Toxicol. Chem. 31, 2012 5

Table 3. Naled deposition levels resulting in adult butterfly exposure Environmental managers interested in conserving butterflies equivalent to the 10th, 50th, and 90th percentile refined effect estimates. need a guideline deposition level to help them reduce the risk to Equivalent naled deposition butterflies from naled applications. The present study suggests a 2 b Exposure levels (mg/cm ) range of naled deposition levels that will minimize the like- metric lihood of adult butterfly mortality. An environmental manager Butterfly family (cm2/g)a ABCcould select the deposition level associated with the appropriate butterfly family to reduce risk for an imperiled species, or they Hesperidae 96.7 0.011c 0.029 0.094 Papilionidae 87.7 0.012 0.032 0.104 could select of one of the lower equivalent deposition levels to Nymphalidae 123.3 0.008 0.023 0.074 reduce risk for the entire butterfly community. Guideline dep- Pieridae 138 0.007 0.021 0.066 osition levels have been suggested to reduce risk for butterflies. Lycaenidae 433.3 0.002 0.007 0.021 Naled deposition in excess of 0.1 mg/cm2 (reported as 1,000 mg/ 2 a The greatest exposure metrics for each family are shown to represent the m ) was expected to reduce survival of larval C. thomasi maximum potential exposure. bethunebakeri [7]. That deposition level is greater than the b The values shown in the three columns represent naled deposition levels majority of the equivalent deposition levels shown in Table 3 that would result in butterfly exposure equivalent to the 10th (column A), indicating that a guideline level of 0.1 mg/cm2 would not 50th (B), and 90th (C) percentile refined effect estimates. The 10th, 50th, m appreciably reduce risk for adult butterflies. Others have sug- and 90th percentile refined effects estimates are 1.02, 2.85, and 9.1 g/g, m 2 respectively. gested a more conservative guideline level (0.0085 g/cm ) c Equivalent deposition level ¼ refined effect estimate exposure metric; based on the 10th percentile LD50 [6]. Although that level of for example, 1.02 mg/g 96.7 cm2/g ¼ 0.011 mg/cm2. deposition may be protective, it is unclear whether adequate mosquito control would be achieved if naled applications were equivalent to the 50th percentile effect level for those two reduced to ensure that that level of deposition was not exceeded. families (0.021–0.023 mg/cm2) would be protective of those Zhong et al. [7] measured mosquito control efficacy following butterflies as well as Papilionidae and Hesperidae. They may aerial applications but the deposition level associated with an not be protective of butterflies in the family Lycaenidae. adequate level of mosquito control was not determined. When trying to allow for mosquito control and reduce risk for butter- flies, the choice of a threshold deposition level should consider DISCUSSION not only protecting butterflies but also the level of deposition This risk assessment has predicted considerable impacts for necessary to attain a desired level of adult mosquito control. adult butterflies following aerial naled applications under the The exposure analysis in this risk assessment used morpho- rates used during field trials. Risk for butterflies with an metric data for individual butterflies, data that appear to be exposure metric similar to that of L. cassius was similar whether lacking in the open literature. The morphometric data were used the estimate was based on naled deposition onto yarn or paper to derive an exposure metric for estimating weight-normalized samplers, indicating that sampler type has an insignificant exposure that can be directly related to weight normalized effect impact on risk estimation for butterflies with a large exposure levels in risk assessment. The exposure estimates in the present metric. On the other hand, there was an apparent difference for study utilized a metric assuming that butterflies roost at night butterfly species with an exposure metric similar to that for with their wings folded together in a nonhorizontal orientation. U. proteus, so sampler type may be significant for risk estima- Some butterfly species do spread their wings when roosting tion for butterflies with a small exposure metric. Although it was (personal observations). Such an orientation could double the clear that the two sampling methods provided different indi- surface area available for naled deposition, potentially increas- cations of naled exposure, it was not clear which method would ing exposure and significantly impacting the risk estimation. be more appropriate. Additional work comparing naled depo- This is the first effort to estimate insect exposure as chemical sition data provided by the two samplers with biological mass per unit body weight for pesticides applied by ultralow- response would give insight into the more appropriate sampling volume sprays for the purpose of evaluating risk to adult method. butterflies. Other risk assessments for insects have typically The Refuge currently permits up to nine aerial naled appli- based exposure estimates on chemical mass per unit surface cations each year, depending on mosquito abundance. As noted, area or per organism [4,6,7,9,10]. The exposure metric used in two butterfly species that are candidates for listing under the the present study may be more appropriate for estimating risk Endangered Species Act exist on the Refuge and likely are at given that an exposure estimate based on body weight has a risk from the applications. Naled is one of several chemicals greater relationship with effects [14,15]. Certain inferences for (sumethrin, etofenprox, prallethrin, malathion) aerially applied risk to butterflies could be made based on the exposure metric. for adult mosquito control in the state of Florida, and it is Efficacious exposure (exposure most related to biological the primary pesticide used in aerial applications (http://www. response) would be positively correlated with the metric and flaes.org/aes-ent/mosquito/reports.html). Within the United would not necessarily be related to total surface area. Also, if the States, 70% of registered naled use is for adult mosquito control relative sensitivity to a chemical is equal among species, the risk [5]. The risk estimated in the present study was based on the would be less for species with a lower metric. However, given likelihood that the exposure exceeds an acute effect level the current paucity of morphometric data for adult butterflies, resulting in mortality. The risk estimate indicates a high like- inferences about the relative metrics and therefore risk among lihood that exposure will result in mortality, but it does not butterfly families would be associated with considerable uncer- indicate the magnitude of impact on butterfly populations or tainty. It should be noted that the role of the metric in risk communities. The ability of butterfly populations or commun- estimation would be greatest for highly labile chemicals that ities to withstand mortality events as a result of aerial naled impact butterflies through direct deposition onto the organism. applications is unknown. Incorporation of life history data for Butterfly exposure for more persistent chemicals (e.g., perme- the species of interest within risk assessments would allow thrin) will increasingly be related to butterflies coming into prediction of longer term impacts on populations. contact with contaminated surfaces. 6 Environ. Toxicol. Chem. 31, 2012 T.A. Bargar

This risk assessment did not incorporate the potential role of butterfly abundance and diversity in ocean-bordering habitats vegetation in moderating exposure. Vegetation can significantly immediately following hurricanes [23]. Increased hurricane reduce pesticide drift [16] and reduce effects to nontarget frequency and intensity over the Atlantic Ocean in response insects [17]. Interception of naled droplets by vegetation fol- to global warming [24] could adversely impact butterfly com- lowing aerial applications may have resulted in reported munities in coastal habitats or low-elevation areas such as the effect levels (26% mortality) that were apparently lower than Florida Keys. However, the long-term impact of global climate estimated by this risk assessment [7]. Larvae were placed change on resident butterflies of the Florida Keys has not onto vegetation during the study by Zhong et al. [7] and received much attention. might not have been exposed fully to naled. An additional The role of mosquito control operations in the decline of factor might have been the relative surface areas of the larvae butterfly populations in the Florida Keys has been debated. and adult butterflies. The larvae in the Zhong et al. study Habitat loss might have played a role in the overall decline of were approximately 1 cm long, presenting a much smaller butterflies in the Florida Keys. Butterfly populations that surface area per unit body weight for naled deposition. Differ- depend on the remaining native habitat are susceptible to ential sensitivity between larvae and adults is an unlikely factor multiple pressures that can cause a further decline in their given that larvae may be more sensitive to naled than adults [6]. populations. This risk assessment indicates that aerial applica- A refinement of naled risk to adult butterflies should include an tions of naled could present a significant risk to butterflies. evaluation of exposure reduction resulting from vegetative Applications targeting the remaining native habitat could cover. adversely impact the butterfly populations that depend on them This risk assessment also did not address the potential impact for their continued existence. of adult butterfly behavior on their exposure. Although most adult butterflies rest with their wings folded, a few species (e.g., Acknowledgement—C. Weiser of the U.S. Geological Survey provided a Heraclides cresphontes, H. aristodemus ponceanus) rest with considerable amount of assistance in the field studies. The Florida Keys their wings spread. Such a behavioral difference would effec- Mosquito Control district conducted aerial spray missions that allowed the tively increase the surface area potentially available for expo- present study to occur and provided personnel to assist in the deployment and retrieval of residue samplers. The Key Deer National Wildlife Refuge sure and increase the risk for such species. In addition, the permitted the present study to take place on their property, provided space for potential exists for interspecific differences in preferred roost- coordination of the field studies, and also provided personnel to assist in the ing habitat. Such information is not widely available in the open deployment and retrieval of residue samplers. H. (He) Zhong and C. Brock literature, but the potential preference for certain types of provided analytical support for determination of naled residues on samplers roosting locations by particular species could alter their expo- collected during the field trials. Funding for the present study was provided by the U.S. Fish and Wildlife Service’s National Wildlife Refuge System and sure and risk relative to other species. the U.S. Geological Survey. Any use of trade, product, or firm names is for Concerns have been raised about the decline of butterfly descriptive purposes only and does not imply endorsement by the U.S. populations worldwide [18]. Chief causative factors include Government. habitat loss and possibly climate change. Habitat loss or frag- mentation resulting from altered land use practices or indus- REFERENCES trialization has reduced species diversity in Japan [19] 1. Salvato MH. 2001. Influence of mosquito control chemicals on and reduced abundance for a protected species in the United butterflies (Nymphalidae, Lycaenidae, Hersperiidae) of the lower Kingdom [20]. The Florida Keys are a popular tourist destina- Florida Keys. J Lepidop Soc 55:8–14. tion and have experienced considerable development pressures 2. Calhoun JV, Slotten JR, Salvato MH. 2000. The rise and fall of tropical over the past several decades, increasingly fragmenting native blues in Florida: Cyclargus ammon and Cyclargus thomasi bethune- bakeri (: Lycaenidae). Holarct Lepidop 77:13–20. habitats. The two areas in the Florida Keys with the largest 3. Eliazar PJ, Emmel TC. 1991. Adverse impacts on non-target insects. intact native habitats are upper Key Largo and Big Pine Key, In Emmel TC, Tucker JC, eds, Mosquito Control Pesticides: which both have significant areas protected by either state parks Ecological Impactsand Management Alternatives. Scientific Publishers, (Dagny Johnson Key Largo Hammock Botanical State Park) or Gainesville, FL, USA, pp 17–20. 4. Schleier JJ III, Peterson RKD. 2010. Toxicity and risk of permethrin and federal refuges (Florida Keys National Wildlife Refuge Com- naled to non-target insects after adult mosquito management. plex). Those are also the areas with remaining populations of Ecotoxicology 19:1140–1146. federally protected butterfly species (H. aristodemus poncea- 5. U.S. Environmental Protection Agency. 2002. January 2002. Interim nus, A. troglodyte floridalis, and S. acis bartrami). One addi- reregistration eligibility decision for naled. EPA-738-R-02-008. tional protected taxon (C. thomasi bethunebakeri) is restricted Prevention, Pesticides, and Toxic Substances, Washington, DC. 6. Hoang TC, Pryor RL, Rand GM, Frakes RA. 2011. Use of butterflies to a few small cays protected within the Key West National as nontarget insect test species and the acute toxicity and hazard Wildlife Refuge. In addition, wildfires important for mainte- of mosquito control insecticides. Environ Toxicol Chem 30:997– nance of healthy host plant populations for A. troglodyte 1005. floridalis and S. acis bartrami have been suppressed on Big 7. Zhong H, Hribar LJ, Daniels JC, Feken MA, Brock C, Trager MD. 2010. Aerial ultra-low volume application of naled: Impact on nontarget Pine Key. However, recent controlled burns have been con- imperiled butterfly larvae (Cyclargus thomasi bethunebakeri) and ducted on Big Pine Key to restore the pineland scrub habitat on efficacy against adult mosquitoes (Aedes taeniorhynchus). Environ which those species depend. Therefore, habitat loss and frag- Entomol 39:1961–1972. mentation likely have played significant roles in the decline of 8. Breindebaugh MS, De Szalay FA. 2010. Effects of aerial applications of butterfly species in the Florida Keys. naled on nontarget insects at Parris Island, South Carolina. Environ Entomol 39:591–599. Global climate change in Central Europe is believed to have 9. Tuell JK, Isaacs R. 2010. Community and species-specific responses of resulted in the retreat of a cold-adapted species (Lycaena helle) wild bees to insect control programs applied to a pollinator-dependent to mountain habitats more favorable for their survival [21]. The crop. J Econ Entomol 103:668–675. decline of habitat specialist species in the Mediterranean Basin 10. Vaal M, Kuijer E, Van Mannekes M, Groenendijk D, Van den Berg M. 2002. Effects and risk analysis of insecticides used in Dutch agricultural relative to habitat generalists was attributed to global warming systems on non-target insects. Environ Eng Manag J 1:221–229. [22]. The only study potentially alluding to the role of climate 11. Tietze NS, Shaffer KR, Hester PG. 1996. Half-life of naled under three change in butterfly decline in the Florida Keys noted reduced test scenarios. J Am Mosq Control Assoc 12:251–254. Risk assessment of naled impacts on adult butterflies Environ. Toxicol. Chem. 31, 2012 7

12. Mayer FL Jr, Ellersieck MR. 1986. Manual of acute toxicity: 18. Warren MS, Bourn NAD. 2011. Ten challenges for 2010 and beyond to Interpretation and data base for 410 chemicals and 66 species of conserve Lepidoptera in Europe. J Insect Conserv 15:321–326. freshwater . Fish and Wildlife Service Resource Publication 19. Nakamura Y. 2011. Conservation of butterflies in Japan: Status, actions 160. U.S. Fish and Wildlife Service, Washington, DC. and strategy. J Insect Conserv 15:5–22. 13. Ecological Committee on FIFRA Risk Assessment Methods (ECO- 20. Smee M, Smyth W, Tunmore M, ffrench-Constant R, Hodgson D. 2011. FRAM). 1999. Ecological committee on FIFRA risk assessment Butterflies on the brink: Habitat requirements for declining populations methods: Report of the aquatic workgroup. U.S. Environmental of the marsh fritillary (Euphydryas aurinia) in SW England. J Insect Protection Agency, Office of Pesticide Programs, Washington DC. Conserv 15:153–163. 14. Van Wezel AP, Sijm DTHM, Seinen W, Opperhuiezen A. 1995. Use of 21. Habel JC, Rodder D, Schmitt T, Neve G. 2011. Global warming will lethal body burdens to indicate species differences in susceptibility to affect the genetic diversity and uniqueness of Lycaena helle populations. narcotic toxicants. Chemosphere 31:3201–3209. Glob Change Biol 17:194–205. 15. McCarty LS, Mackay D. 1993. Enhancing ecotoxicological modeling 22. Stefanescu C, Torre I, Jubany J, Paramo F. 2011. Recent trends in and assessment: Body residues and modes of action. Environ Sci Technol butterfly populations from north-east Spain and Andorra in the light of 27:1719–1728. habitat and climate change. J Insect Conserv 15:83–93. 16. Lazzaro L, Otto S, Zanin G. 2008. Role of hedgerows in intercepting 23. Salvato MH, Salvato HL. 2007. The influence of hurricane and tropical spray drift: Evaluation and modeling of the effects. Agric Ecosyst storm activity on resident butterflies in the lower Florida Keys. J Lepidop Environ 123:317–327. Soc 61:154–164. 17. Otto S,LazzaroL, FinizioA, ZaninG.2009.Estimatingecotoxicological 24. Bender MA, Knutson TR, Tuleya RE, Sirutis JJ, Vecchi GA, Garner ST, effects of pesticide drift on nontarget in field hedgerows. Held IM. 2010. Modeled impact of anthropogenic warming on the Environ Toxicol Chem 28:853–863. frequency of intense Atlantic hurricanes. Science 22:454–458.